Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Pulsed corona-induced degradation of organic materials in water Hoeben, W.F.L.M. DOI: 10.6100/IR535691 Published: 01/01/2000 Document Version Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers) Please check the document version of this publication: • A submitted manuscript is the author's version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website. • The final author version and the galley proof are versions of the publication after peer review. • The final published version features the final layout of the paper including the volume, issue and page numbers. Link to publication Citation for published version (APA): Hoeben, W. F. L. M. (2000). Pulsed corona-induced degradation of organic materials in water Eindhoven: Technische Universiteit Eindhoven DOI: 10.6100/IR535691 General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal ? Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Download date: 02. Aug. 2017 Pulsed corona-induced degradation of organic materials in water PROEFSCHRIFT ter verkrijging van de graad van doctor aan de Technische Universiteit Eindhoven, op gezag van de Rector Magnificus, prof.dr. M. Rem, voor een commissie aangewezen door het College voor Promoties in het openbaar te verdedigen op donderdag 15 juni 2000 om 16.00 uur door Wilhelmus Frederik Laurens Maria Hoeben geboren te Geldrop Dit proefschrift is goedgekeurd door de promotoren: prof.dr. W.R. Rutgers en prof.dr.ir. C.A.M.G. Cramers Copromotor: dr.ir. E.M. van Veldhuizen CIP-DATA LIBRARY TECHNISCHE UNIVERSITEIT EINDHOVEN Hoeben, Wilhelmus Frederik Laurens Maria Pulsed corona-induced degradation of organic materials in water / by Wilhelmus Frederik Laurens Maria Hoeben Eindhoven: Technische Universiteit Eindhoven, 2000. -Proefschrift.ISBN 90-386-1549-3 NUGI 812 Trefw: gepulste corona / elektrische ontladingen / AOP / fenol / afbraak / oxidatie / conversie / efficiëntie / vloeistofchromatografie Subject headings: pulsed corona / electrical discharges / AOP / phenol / degradation / oxidation / conversion / efficiency / liquid chromatography This project has been financially supported by “Technologie voor Duurzame Ontwikkeling (TDO)”, Technische Universiteit Eindhoven. Ontwerp omslag: B. Mobach / TUE Drukwerk: Universiteitsdrukkerij Technische Universiteit Eindhoven “A company has control of its production only if it also knows the make-up of its waste water” [Ullmann, Encyclopedia of Industrial Chemistry 1994, 5th edn. Vol. B6, Weinheim: Verlag Chemie, ISBN 3-527-20136-X, 474] Aan mijn ouders, broer en zus Contents 1. Introduction .......................................................................................... 1 2. Theory ..................................................................................................11 3. Experimental setup .............................................................................37 4. Results .................................................................................................49 1.1 1.2 1.3 1.4 2.1 2.2 2.3 2.4 Advanced oxidation processes ..................................................................... 2 Electrical discharges ................................................................................... 6 Model compounds ...................................................................................... 7 Thesis scope ............................................................................................. 9 Corona discharges.....................................................................................11 Oxidizers..................................................................................................13 Degradation of organic compounds ..............................................................15 Oxidation of model compounds ...................................................................20 2.4.1 Phenol ..........................................................................................20 2.4.2 Atrazine ........................................................................................23 2.4.3 Malachite green..............................................................................23 2.4.4 Dimethyl sulfide .............................................................................24 2.5 Diagnostics ..............................................................................................24 2.5.1 Chemical diagnostics ......................................................................25 2.5.2 Electrical diagnostics.......................................................................31 2.5.3 Optical diagnostics .........................................................................32 3.1 3.2 3.3 3.4 Reagents and reactors ...............................................................................37 Chemical diagnostics .................................................................................40 Electrical diagnostics .................................................................................45 Optical diagnostics ....................................................................................46 4.1 Pulsed corona discharges ...........................................................................49 4.1.1 Hydroxyl radicals............................................................................49 4.1.2 Ozone...........................................................................................55 4.1.3 Corona pulse energy .......................................................................59 4.1.4 Corona treatment of deionized water.................................................63 4.2 Oxidation of phenol ...................................................................................65 4.2.1 Chromatography.............................................................................65 4.2.2 Mass spectrometry .........................................................................87 4.2.3 Spectroscopy.................................................................................92 4.2.4 Electrical conductometry ...............................................................102 4.2.5 Microtox ecotoxicity .....................................................................105 4.2.6 Total organic carbon .....................................................................108 4.3 Oxidation of other model compounds .........................................................109 4.3.1 Atrazine ......................................................................................109 4.3.2 Malachite green............................................................................110 4.3.3 Dimethyl sulfide ...........................................................................112 5. Discussion ............................................................................................113 5.1 5.2 5.3 5.4 5.5 Pulsed corona discharges .........................................................................113 Corona-induced phenol oxidation ...............................................................116 Phenol oxidation pathways .......................................................................122 Analysis techniques.................................................................................137 AOP comparison .....................................................................................141 6. Conclusions ..........................................................................................145 6.1 6.2 6.3 6.4 Pulsed corona discharges .........................................................................145 Oxidation of model compounds .................................................................146 Analytical techniques...............................................................................147 Outlook .................................................................................................148 7. References ...........................................................................................149 Summary ...................................................................................................159 Samenvatting ............................................................................................161 Dankwoord / Acknowledgements .........................................................163 Curriculum Vitae ......................................................................................164 1. Introduction Since a long time, natural processes have not been able anymore, to rectify the environmental load caused by the ever-increasing world population. Our water reserves are a main issue of interest, because pollution from both the atmosphere and soil will eventually enter the aqueous phase by deposition and percolation respectively. Sources of pollution are both nature and mankind. Examples of natural pollution are volcanic activity, forest fires and decomposition of vegetation. Pollution by mankind is caused by e.g. nutrition, transportation, accommodation, synthesis and energy exploitation. Although probably not always acknowledged, chemical activity is indispensable to sustain life; also it is needed to comply with a high standard of living. Examples are medicaments, cleaning and disinfecting products, cosmetics, stabilizers, artificial fertilizers, pesticides, fuel, batteries, polymers (thermoplastics, thermosetting resins, elastomers, fibers), paint and dyes. Both synthesis and application of these product classes inevitably yield pollution. In addition to biological waste like carbohydrates, proteins, urea, fats, food & vegetation residues and carbon dioxide, we also encounter priority compounds. These materials exhibit carcinogenic, mutagenic and/or teratogenic properties, which implies that a noeffect-level in fact does not apply for these compounds. In addition, priority compounds can be highly persistent. Some organic priority compounds are for instance [1]: halogenated dioxins/ benzofurans/xanthenes from the incineration of halogenated phenols, polychlorinated biphenyls (PCB’s) used as dielectric media, fire retardants; polycyclic aromatic hydrocarbons (benzo[a]pyrene, dibenzo[a:h]anthracene) in soot and coal tar/pitch from the incomplete combustion of hydrocarbons and from coal gasification; simple aromatic hydrocarbons (benzene, nitrobenzene, p-dichlorobenzene, o-phenylenediamine) used as precursors in organic chemical synthesis; chlorinated aliphatics (chloroform, tetrachloromethane, trichloroethylene) applied as solvent and/or stain remover; pesticides (DDT, kepone, lindane) for crop protection and pest control; ammunition (TNT, picric acid, nitroanilines); monomers (acrylonitrile, vinylchloride, urethane) from the synthesis, processing and incomplete combustion of polymers, dyes (benzidinebased) for the colorization of e.g. textile, leather and polymers. Inorganic priority compounds are for instance heavy metals & salts (Cd, Ni, Cr), asbestos, arsenic/compounds, beryllium/compounds and radioactive materials. Although we left the ages of unscrupulous operation long ago, we inherit innumerous highly polluted waste sites of former gasworks, ammunition and pesticide plants, oil/gas drill and refinery locations, mining sites, fuel stations, dry-cleaning facilities, waste dump and incineration sites. Conventional microbiological degradation desperately needs the assistance of new technologies, like for instance advanced oxidation processes, to degrade hazardous persistent materials by chemical oxidation. 2 Chapter 1. 1.1. Advanced oxidation processes Advanced Oxidation Processes (AOP’s) aim at the in-situ production of strong oxidizers. The oxidizing power is reflected by the standard reduction potential E0. Table 1.1 shows some oxidizers in decreasing power order and E0 values, expressed for reduction halfcell reactions [2,3]. The potential is defined relative to the standard hydrogen electrode potential [4]. The Gibbs free energy change ∆G of the redox-reaction is calculated from the resulting electromotive force of both half-cell reactions corrected for activity dependence (E), the number of electrons involved (n) and the Faraday constant (F=96485 C/mol), see Equation 1.1. Table 1.1 Standard reduction potential values for some oxidizers at T=298.15 K, for acidic conditions pH=0 applies. Reduction half-cell reaction XeF+ e- → Xe + F2OF2 (g) + 4H+ + 4e- → O2 (g) + 4HF OH + H+ + e- → H2O O (g) + 2H+ + 2e- → H2O O3 + 2H+ + 2e- → O2 + H2O H2O2 + 2H+ + 2e- → 2H2O HClO2 + 2H+ + 2e- → HClO+H2O HO2 + H+ + e- → H2O2 Cl2 + 2e- → 2 Cl∆G = − n ⋅ F ⋅E E0 (V) 3.4 3.29 2.56 2.43 2.08 1.76 1.67 1.44 1.40 (1.1) The strongest oxidizers known are xenonfluoride (XeF) and possibly H4RnO6, but these oxidizers are not commercially attractive for water treatment because of both extreme reactivity and remaining toxicity in reduced form. Also, halogen-based oxidizers are not acceptable as oxidizer, because they halogenate organic materials to e.g. trihalomethanes [5] which are very harmful compounds; in addition their reaction leads to salt formation. It is obvious, that metal-based oxidizers like permanganate (MnO4-) and dichromate (Cr2O72-) also are not desirable. Of interest are thus oxygen-based halogen/metal-free oxidizers like the hydroxyl radical (OH), atomic oxygen (O), ozone (O3) and hydrogen peroxide (H2O2). Next, a concise description is presented for major AOP’s with regard to the generation of oxygen-based halogen-free oxidizers, particularly hydroxyl radicals. A comparison of AOP’s is discussed in section 5.5. Introduction 3 Ozone-UV oxidation In the ozone-UV technology [6,7], hydroxyl radicals are produced from ozone, water and UV photons; high-pressure mercury or xenon lamps deliver the photons, see Equation 1.2. O3 + H2O + hν → 2OH + O2 λ≤310 nm (1.2) Ozone is produced on location by an ozonizer, which converts atmospheric or pure oxygen into ozone by corona discharges [8,9]. These electrical discharges are produced in a barrier discharge electrode setup, where the electrodes are separated by a dielectric material e.g. glass or ceramic at a thickness of about 0.5-3 mm. The applied voltage is 8-30 kV and the frequency range is 60-2000 Hz. The energy efficiency is about 60 g/kWh for air or 120 g/kWh for oxygen [10]. The theoretical maximum efficiency is calculated from the standard formation enthalpy change ∆Hf0=144.8 kJ/mol for the reaction 3O2→2O3 and is about G=1193 g O3/kWh. Commercial ozone generators are based on different electrode configurations, e.g. fluid-cooled shell & tube type generators for generation of large ozone amounts and air-cooled plate type generators for small amounts. Cooling is very important, to prevent decomposition of ozone. Hydrogen peroxide-UV and Fenton oxidation Hydrogen peroxide is decomposed by UV photons into hydroxyl radicals [11], see Eq.1.3a. Also, the reaction of hydrogen peroxide with iron (II) ions produces hydroxyl radicals; this reaction is known as the Fenton reaction (Eq.1.3b) [12]. In addition, Fe(III) ions contribute to hydroxyl radical formation by Eq.1.3c/d (Fenton like reaction) and indirectly by regeneration of Fe(II). H2O2 + hν → 2OH Fe2+ + H2O2 → OH + OH- + Fe3+ Fe3+ + OH- Fe(OH)2+ Fe(OH)2+ + hν → OH + Fe2+ 250 nm<λ<300 nm λ=350 nm (1.3a) (1.3b) (1.3c) (1.3d) The advantage of photo-Fenton/Fenton like reactions over hydrogen peroxide-UV is mainly explained by the efficient use of light quanta, because the absorption of Fe(III) chelates (hydroxo, carboxyl) extends to the near UV-visible region and their molar absorption coefficient is relatively high compared to the molar absorption coefficient of hydrogen peroxide. Synthesis of hydrogen peroxide is mainly performed according to the following processes [13,14]: *Anthraquinone (AO) process: Reduction of a 2-alkyl-9,10-anthraquinone to the corresponding hydroquinone by hydrogen, followed by the oxidation of the hydroquinone by oxygen to hydrogen peroxide and the anthraquinone. *2-Propanol process: Oxidation of 2-propanol by oxygen produces 2-propanol-2hydroperoxide, which decomposes into hydrogen peroxide and acetone. *Electrochemical processes: Anodic oxidative coupling of sulfate ions to persulfate ions, followed by hydrolysis of the persulfate via the peroxomonosulfate to hydrogen peroxide and bisulfate ions. The theoretical maximum efficiency, calculated from the standard formation enthalpy change ∆Hf0=98.3 kJ/mol for the reaction H2O (l) +½O2 (g) → H2O2 (l) is about G=1246 g H2O2/kWh. 4 Chapter 1. Photocatalytic oxidation Photocatalytic oxidation produces hydroxyl and hydroperoxyl radicals at an irradiated semiconductor surface in contact with water [15,16]. Excitation of electrons in the semiconductor surface layer by UV photons will promote electrons from the valence band to the conductivity band. In this way electron-deficient holes (h+) are created in the valence band and free electrons (e-) will be available in the conductivity band. Equations 1.4a-f are the main reactions, that take place at the irradiated semiconductor surface. Water is absorbed onto the surface, resulting in the formation of H+ and OHions, see Eq.1.4a/b. Hydroxyl radicals are produced by oxidation of water (Eq.1.4c) or oxidation of hydroxyl ions (Eq.1.4d), while hydroperoxyl radicals are obtained from the superoxide anion (O2-), see Eq.1.4e/f. 2H2O + 4h+ → 4H+ + O2 2H2O + 2e- → 2OH- + H2 H2O + h+ → OH + H+ OH- + h+ → OH O2 + e- → O2O2- + H+ → HO2 (1.4a) (1.4b) (1.4c) (1.4d) (1.4e) (1.4f) Some applied semiconductors are titanium oxide (TiO2), zinc oxide (ZnO) and cadmium sulfide (CdS). The most well-known is the anatase crystal structure of TiO2. Its bandgap energy is 3.2 eV; the irradiation wavelength λ<385 nm. TiO2 has favourable photochemical stability and photocatalytic activity. Wet oxidation In wet oxidation, water with dissolved oxygen is used to oxidize the target compound [17,18]. The process can be performed at e.g. subcritical (4 MPa<p<20 MPa, 513K<T<593K) or supercritical conditions (p>22.1 MPa, T>647K). These conditions enable optimal solubility of oxygen and organic compounds in water. Metal ions can be added to catalyze the oxidation. Equations 1.5a-h are the main reactions. Hydroxyl radicals are produced from the dissociation and oxidation of water (Eq.1.5a/b). Hydroperoxyl radicals are formed from the oxidation of water (Eq.1.5b) and the target compound RH (Eq.1.5f). Hydroxyl radicals are also produced from hydrogen peroxide (Eq.1.5d) and from the reaction of atomic oxygen with the target compound (Eq.1.5h). Hydrogen peroxide is produced by recombination of hydroperoxyl radicals (Eq.1.5c) or by reaction of hydroperoxyl radicals with the target compound (Eq.1.5g). Atomic oxygen is produced from the dissociation of oxygen (Eq.1.5e). Although the hydroperoxyl radical is less reactive than the hydroxyl radical, it plays an important role because of its relative abundance. H2O → OH + H H2O + O2 → OH + HO2 2HO2 → H2O2 + O2 H2O2 → 2OH O2 → 2O RH+O2 → R + HO2 RH + HO2 → R + H2O2 RH + O → R + OH (1.5a) (1.5b) (1.5c) (1.5d) (1.5e) (1.5f) (1.5g) (1.5h) Introduction 5 Radiolysis Irradiation of water by high-energy photons or electrons dissociates water molecules into hydroxyl radicals and hydrogen atoms or ionizes water molecules, see Eq.1.6a/b [19,20]. Ionized water molecules react with water to produce hydroxyl radicals, see Eq.1.6c. By saturation of the water with nitrous oxide (N2O), solvated electrons (Eq.1.6d) are converted into hydroxyl radicals (Eq.1.6e). Also the target compound is dissociated or ionized. Halogenated target compounds RXn react rapidly with solvated electrons, see Eq.1.6f. H2O → OH + H H2O → H2O+ + eH2O+ + H2O → H3O+ + OH e- + H2O → eaqN2O + eaq- + H2O → N2 + OH + OHRXn + eaq- → RXn-1 + X- (1.6a) (1.6b) (1.6c) (1.6d) (1.6e) (1.6f) High-energy photons are obtained from a radioactive source (60Co-γ) and electrons are produced by an electron beam accelerator or a Van de Graaff generator. Ultrasonic irradiation The introduction of ultrasonic energy into a liquid causes electrohydraulic cavitation [21,22]. The applied frequency range is from 15 kHz up to 1 MHz. The generation of ultrasound energy can be performed by electromechanical (piezoelectric or magnetostrictive) or liquid-driven (liquid whistle = low intensity) transducers. The cavitation process involves the oscillation of the radii of pre-existing gas cavities by the periodically changing pressure field of the ultrasonic waves. The rapid implosion of the eventually instable gas bubbles causes adiabatic heating of the bubble vapour phase. In this way, localized and transient high temperatures and pressures are reached, e.g. p>300 bar and T>3300 K in aqueous solution. These vigorous conditions invoke dissociation and pyrolysis of the liquid phase molecules and present target compounds. Water will be dissociated into hydroxyl radicals and hydrogen atoms, see Eq.1.7a. Organic compounds are dissociated into radicals (Eq.1.7b/c) and functional groups like carboxyl and nitro groups are removed, see Eq.1.7d/e. H2O → OH + H AB → A + B RXn → RXn-1 + X RCOOH → RH + CO2 RNO2 → RO + NO (1.7a) (1.7b) (1.7c) (1.7d) (1.7e) 6 Chapter 1. 1.2. Electrical discharges The discharge of electric energy into a dielectric medium may cause dissociation, ionization and excitation of the dielectric molecules or atoms [23]. Depending on the energy input, the produced plasma is non-thermal or thermal. In thermal plasmas the ionization level is high, about 10-2. Examples of thermal electrical discharges are lightning and arc discharges. Typical numbers of electron density (ne) and electron energy (Te) for lightning discharges are about ne=1⋅1017-5⋅1017 cm-3 and Te=2.2 eV (corresponding to 25000 K). Corona and glow discharges are non-thermal plasmas. Their ionization level is very low, about 10-6. The electron density of a corona plasma is about ne=1013 cm-3. The chemical reactivity of corona discharges is based on the fact, that the electric field strength at the discharge streamer heads is extremely high viz. about 200 kV/cm, corresponding to 1000 Td. This implies an average electron energy of about Te=10 eV, which reaches beyond the dissociation energy of water (5.16 eV), oxygen (5.17 eV) and nitrogen (9.80) [24]. Within the energy distribution of electrons in the streamer head, even higher energetic electrons exist that cause ionization [25] of oxygen (12.07 eV), water (12.62 eV) and nitrogen (15.58 eV). A very particular advantage of pulsed corona discharges is the fact, that a highly reactive streamer discharge medium is created, while the bulk gas is at ambient temperature and pressure [26,27]. Therefore, pulsed corona promises higher efficiency than other advanced oxidation processes. Corona discharges in water produce hydroxyl radicals and hydrogen atoms from the dissociation and ionization of water molecules, see Eq. 1.8a-c. In a humid gas phase, corona discharges additionally create radicals, ions and metastables from the dissociation and ionization of the gas phase molecules or atoms. In humid air, the following main oxidizer species are produced: hydroxyl radicals, ozone, atomic oxygen, singlet oxygen and hydroperoxyl radicals, see Eq. 1.8a-n. Also, small amounts of nitrogen oxides like NOx and N2O are formed according to Eq. 1.8o-r. H2O + e- → OH + H + eH2O + e- → H2O+ + 2eH2O+ + H2O → H3O+ + OH N2 + e- → N2* + eO2 + e- → O2* + eN2 + e- → 2N + eO2 + e- → 2O + eN2 + e- → N2+ + 2eO2 + e- → O2+ + 2eO2 + e- → O2O2 + e- → O- + O O2 + O → O3 H + O2 → HO2 H + O3 → HO3 N + O → NO NO + O → NO2 N2+ + O2- → 2NO N2 + O → N2O dissociation ionization dissociation excitation excitation dissociation dissociation ionization ionization attachment dissociative attachment association association association association association recombination association Next to these reactions, many others exist [28]. (1.8a) (1.8b) (1.8c) (1.8d) (1.8e) (1.8f) (1.8g) (1.8h) (1.8i) (1.8j) (1.8k) (1.8l) (1.8m) (1.8n) (1.8o) (1.8p) (1.8q) (1.8r) Introduction 7 Figure 1.1 shows a typical CCD image of pulsed positive corona discharges in air over a water surface. The bright areas are a superposition of about 100 streamer discharge channels. The streamer channels are thin; their diameter is 1 mm or less. The streamer discharges start from a 30 pins anode and then propagate towards the gas-liquid interface. The discharges do not enter the aqueous phase due to the high relative permittivity of water. The cathode plate is situated directly outside and underneath the glass reactor vessel. Figure 1.1 CCD image of pulsed positive corona discharges in air over a water surface. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The CCD camera settings are: diaphragm f/5.6, exposure time 1 s. The image has been taken by A.H.F.M. Baede. The first applications of electrical discharges in water date from Russian experiments in the seventies on pre-breakdown phenomena in liquid dielectrics by Klimkin [29] and Alkhimov [30] and also for the destruction of bacteria in water. In the early eighties water ozonation was introduced for waste water treatment, while from about 1987 the pulsed corona discharge technology was introduced for the degradation of water pollutants by Clements and Sato [31]. Till now, pulsed corona discharge technology has been in an experimental stage, mainly for the removal of nitrogen oxides and sulfur dioxide from flue gas and the destruction of hydrocarbons and odor components in waste gas [27]. 1.3. Model compounds In this thesis, the oxidizing power of pulsed corona discharges is investigated with regard to organic compounds. The choice has been made to study some well-known compound classes viz. the bulk chemical phenol, the herbicide atrazine, the dye malachite green and the odor component dimethyl sulfide, see Figure 1.2. Phenol, atrazine and malachite green have been degraded in aqueous solution, dimethyl sulfide has been degraded in the gas phase. A summary of particular compound properties and applications is discussed now. 8 Chapter 1. Cl OH N HN N N + N N S NH X- phenol Figure 1.2 atrazine malachite green dimethyl sulfide Model compounds applied for corona-induced oxidation. Phenol (hydroxybenzene) [32,33] is a moderately toxic crystalline solid, readily absorbed by the skin. Investigations and carcinogenic or co-carcinogenic properties give ambiguous indications. The human lethal dose value is about 140 mg/kg. Phenol is a major precursor in synthesis of e.g. polymers (phenol-formaldehyde, polycarbonate and epoxy resins) including light stabilizers, pharmaceuticals (acetylsalicylic acid, vitamin E, antioxidants), microbicides/fungicides (chlorinated phenols, hydroxybiphenyls), dyes (azo, nitro, triarylmethane, from aniline raw material), photochemicals (diazocoupling, developers) surfactants (alkylphenols), fragances (phenolic ethers); therefore phenol occurs in waste flows released from the synthesis of these compounds. Also during oil refinery and cokes production phenol is set free. Lignins, the biopolymeric construction materials of plant cell walls and wood tissue, consist of e.g. hydroxybenzene-functional monomeric units [34]. Lignins occur in waste flows from paper mills. The proper water solubility of phenol makes it ideally suitable for coronainduced aqueous phase oxidation. Atrazine (6-chloro-N-ethyl-N’-(1-methylethyl)-1,3,5-triazine-2,4-diamine) is a harmful and persistent herbicide [35,36]. It is widely used for selective weed control (broadleaf and grassy weeds) in corn, sugar cane and asparagus but is also used for nonselective weed control on noncropped land. Atrazine is mobile in sand and loam and low to intermediately mobile in clay loam; irrigation induced leaching and dilution has led to atrazine residues persisting for 3 years in soil of irrigation ditches at depths certainly up to 90 cm. Atrazine is a possible human carcinogen [1]. The maximum permissible concentration of atrazine in drinking water is 0.1 ppb. Malachite green [37,38] is a triphenylmethane cationic dye. These dye types favour a high color strength and brilliance, but their light fastness is generally poor. Its green color is due to the absorption of the wavelengths λ=427.5 nm (FWHM=40 nm, log (ε)=4.30) and λ=621 nm (FWHM=60 nm log (ε)=5.02). Malachite green is applied for coloring paper, leather, inks, waxes and polyacrylonitrile fibers. Malachite green exhibits acute oral toxicity and is a suspected human carcinogen due to its photosensitizing properties. It causes acute lethal toxicity in fish, inhibits algae growth and also inhibits activated sludge. Introduction 9 Dimethyl sulfide [39,40] is a highly volatile liquid spreading a very unpleasant odor. The odor detection limit is about 0.75 mg/m3. The boiling point is 37°C and partial pressure is p293K=55.5 kPa. On industrial scale, dimethyl sulfide is emitted from e.g. Kraft pulping paper mills, fish processing and from incineration of cattle cadavers. Dimethyl sulfide is used as marker for odourless gases. Organic sulfides also occur in abundance from natural sources. Inhalation of dimethyl sulfide may cause narcosis and paralysis of the nerve system controlling the respiration and circulation. The maximum allowable concentration is suggested to equal the odor detection limit, about 0.75 mg/m3 equally to 0.29 ppm. 1.4. Thesis scope The scope of this thesis is an investigation of the applicability and technical feasibility of pulsed positive corona discharges for the degradation of organic materials at low concentration in aqueous solution. Although research on aqueous phase remediation by AOP’s is a topic of growing interest since about 20 years, corona research for these purposes is still rather exotic. For this reason a multidisciplinary project has been performed in which the following knowledge sources have participated: the physics of electrical discharges, physical-chemical analysis and organic chemistry. In this way an adequate fundamental comparison can be made between corona discharge technology and other more known AOP’s; a financial consideration of corona with regard to other AOP’s and conventional waste water treatment has not been part of this project. The study is based on the model compounds phenol, atrazine, malachite green and dimethyl sulfide; the corona-induced degradation of phenol has been studied in detail. Key parts are the conversion efficiency and identity of the oxidation products. In chapter 2, a concise description is presented with regard to corona discharges, oxidizer properties, degradation of organic compounds and applied chemical/electrical/ optical diagnostics. Chapter 3 describes the applied reagents, reactors, diagnostics and configurational settings. Chapter 4 represents the experimental results of corona-induced oxidizer production and model compound degradation. The corona-induced production of the oxidizers hydroxyl radicals and ozone is discussed in section 4.1. In-situ electron spin resonance and exsitu molecular probe fluorescence spectrometry have been applied for the detection of hydroxyl radicals, while the formation of ozone has been quantified using in-situ absorption spectrometry. Section 4.2 describes the corona-induced degradation of phenol. The conversion efficiency has been determined from the conversion and required energy input. The conversion has been determined by liquid chromatography, namely ion-exclusion chromatography and reversed-phase high performance liquid chromatography. The energy input has been determined from corona pulse voltage and current waveform measurements. In addition to ex-situ conversion determination, oxidation progress has also been monitored by application of in-situ laser-induced fluorescence spectroscopy. The obtained phenol oxidation product mixtures have been analyzed by several analytical techniques. Primary product identification has been performed by IonSpray and electron-impact mass spectrometry. The environmental impact has been studied by Microtox ecotoxicity tests and total organic carbon measurements. 10 Chapter 1. Electrical conductometry has been applied to monitor phenol oxidation progress by the formation of carboxylic acids. The analysis of gaseous phenol oxidation products has been studied by Fourier transform infrared spectroscopy and an aldehyde screening test. Oxidation pathway models have been constructed in order to account for the composition of the phenol oxidation product mixture. In section 4.3 the corona-induced degradation of atrazine, malachite green and dimethyl sulfide is described. Atrazine conversion has been determined by reversed-phase HPLC, malachite green degradation by several reactor geometries has been measured by decolorization using absorption spectrometry. The oxidation of dimethyl sulfide has been measured by gas chromatography. The influence of different corona parameters and reactor configurations has been determined from phenol conversion and malachite green decolorization measurements; this has resulted in a preferred corona configuration. In chapter 5, a discussion is presented about pulsed corona discharges, corona-induced phenol degradation, phenol oxidation pathways, analysis techniques and a fundamental comparison of different AOP’s. The conclusions are summarized in chapter 6. 2. Theory This chapter starts with a fundamental description of corona discharges in air. Oxidizer properties and general degradation pathways of organic compounds are discussed, followed by specific model compound oxidation products reported from literature. Finally the applied chemical, electrical and optical diagnostics are concisely summarized. 2.1. Corona discharges The requirements for the formation of corona discharges in air at atmospheric conditions are a sharply non-uniform electrical field and a starting condition [23]. The sharply non-uniform E-field is achieved by applying a high voltage to e.g. a point-toplane electrode configuration in air. Corona triggering is enabled, if an ion-electron pair is produced within the inception region of the high voltage electrode, where the electron can gain enough energy to ionize molecules of the dielectric. The ion-electron pair is usually produced by cosmic rays or natural radioactivity, which cause ionization of air at a rate of about 109 m-3s-1. The polarity of a corona can be either positive or negative. The positive corona consists of cathode-directed streamer discharge channels, while the negative polarity corona is anode-directed. Positive polarity corona Once the ion-electron pair is formed within the inception area, the electron is accelerated in the electric field, see Figure 2.1. On its way to the anode the electron collides with other molecules from the dielectric; if the applied field is high enough, primary avalanche electrons are produced. The molecules excited in this primary avalanche cause photoionization by emitting high energy photons and secondary avalanche electrons are produced. The primary avalanche electrons sink in the anode and leave behind a primary avalanche of positive ions having low mobility. The secondary avalanche of electrons runs into the primary avalanche positive ions and a quasineutral plasma channel is produced. The remaining secondary avalanche positive ions form a positive space charge at the head of this plasma channel. If the electical field induced by the space charge reaches a value of the order of the external field, a streamer can be produced. Then, the number of positive ions in the head should reach a value of at least 108, according to Meek [41]. By recurrence of the described processes, the streamer channel grows from the anode tip into the direction of the cathode (cathode directed streamer), see Figure 2.2. At the streamer head, an intense electrical field strength of about E=200 kV/cm is reached that accounts for chemical reactivity viz. radical formation. The streamer grows from the anode at a speed of about 108 cm/s, but stops where the E-field drops below the critical value. Corona discharges have been applied in pulsed form, to prevent transport of ions, that is Ohmic dissipation. 12 Chapter 2. A +- A - + A eIiI+ A - eI +i + I - + eIIiII+ + + - + A iI+/eIIiII+ + + - + N ≥108 i C C C C C 1 2 3 4 5 Figure 2.1 The formation of a cathode-directed streamer in air; A = anode tip, C = cathode plate. (1) The starting condition: an ion-electron pair within the inception area. (2) The production of a primary avalanche of electrons eIleaving behind a primary avalanche of ions iI+; the electron avalanche causes photo-ionization. (3) The primary avalanche electrons sink into the anode, while secondary avalanche electrons eII- produced by photoionization, run into the primary avalanche ions. (4) Recombination of secondary avalanche electrons eII- and primary avalanche ions iI+ produces a quasineutral plasma channel; at the end of this channel a positive space charge is formed by the secondary avalanche ions iII+. (5) If the number of positive ions in the channel head is at least 108, then a streamer can be produced. Figure 2.2 The E-field of a cathode-directed streamer in air. At the streamer head the field strength is about 200 kV/cm. The breakdown field strength of air at ambient conditions is Ec≈30 kV/cm. Negative polarity corona Positive ions produced by the primary avalanche of electrons decrease the field strength at the negatively charged point-shaped electrode. Therefore, the electrons lose energy on their way to the anode and get attached to electronegative oxygen. These negative ions drift slowly towards the anode, while the primary avalanche positive ions sink into the cathode. Then the field at the cathode recovers and the processes restart at the formation of a new ion-electron pair within the inception area. For the case of nonpulsed DC operation, this recurrent process is known as the Trichel pulse regime. Opposite to positive polarity corona, the electrons can only gain kinetic energy within the inception area and therefore the formation of radicals is limited to this region. Theory 13 2.2. Oxidizers Some characteristic properties are described for the following oxidizers produced by advanced oxidation processes: the hydroxyl radical, the ozone radical ion, ozone, atomic oxygen, hydrogen peroxide and the hydroperoxyl radical. Hydroxyl radical The hydroxyl radical (OH) is one of the strongest oxidizers among the oxygen-based oxidizers [42]; its standard reduction potential is E0=2.56 V in acidic environment, see also Table 1.1. The hydroxyl radical is extremely reactive: its life in water is about 2 ns and the radius of diffusion is about 20 Å [43]. In its reaction with inorganic ions, electrons are transferred from the ion to the hydroxyl radical, via an intermediate adduct consisting of the ion, the hydroxyl radical and -depending on the coordinating properties of the ion- a solvent molecule. With regard to organic molecules, the hydroxyl radical reacts electrophilic and adds to unsaturated bonds of e.g. alkenes and aromatic rings. The hydroxyl radical also abstracts hydrogen atoms from organic molecules. In a strongly alkaline environment, the hydroxyl radical exists in its conjugated form: the oxygen radical ion O-, according to Equation 2.1. The acid dissociation constant of the hydroxyl radical is about pKa=11.9. OH + OH- O- + H2O (2.1) The oxygen radical ion is a nucleophilic particle that preferentially abstracts hydrogen atoms from organic molecules. It reacts much more slowly than the hydroxyl radical. Ozone radical ion Although the hydroxyl radical is generally assumed to be the strongest oxygen-based halogen-free oxidizer, the ozone radical ion (O3-) is reported to be an even more powerful oxidizer in acidic solution [2]. It has a standard reduction potential E0=3.3 V. The reduction half-cell reaction is given by Equation 2.2a. The O3- ion is produced from the reaction of the oxygen radical ion and oxygen, according to [42], see Equation 2.2b. In aqueous solution, the O3- ion will oxidize water by which a hydroxyl radical, a hydroxide ion and oxygen are produced, see Equation 2.2c. O3- + 2H+ + e- → O2 (g) + H2O O - + O 2 → O 3O3- + H2O → OH + OH- + O2 E0=3.3V at T=298.15 K and pH=0 (2.2a) (2.2b) (2.2c) 14 Chapter 2. Ozone Ozone (O3) [8,9] is a strong oxidizer, as is indicated by E0 = 2.08 V. It oxidizes water to hydrogen peroxide. Therefore the bulk solubility of ozone in water is rather low viz. about 0.1 mM at T= 293 K [44]. By irradiation with photons of wavelength λ≤310 nm, ozone is decomposed into a singlet oxygen atom and a singlet oxygen molecule (Eq.2.3a) [6]. In humid air, the singlet oxygen atom reacts with water to hydroxyl radicals (Eq.2.3b); in the aqueous phase initially hydrogen peroxide can be produced due to recombination of hydroxyl radicals that cannot escape from the solvent cage, see Eq.2.3.c. The singlet oxygen molecule is also very reactive, its life in water is about 4.4 µs [43]. O3 + hν → O (1D) + O2 (1∆g) O (1D) + H2O(g) → 2OH O (1D) + H2O(l) → H2O2 λ≤310 nm (2.3a) (2.3b) (2.3c) In acidic environment at normal temperatures, ozone reacts selectively with organic compounds as an electrophilic molecule [45]. The electrophilic behaviour of ozone is explained by the positively charged oxygen atom in the possible resonance structures, which are mainly represented by (A) and to a small extent by (B), see Figure 2.3. O O + O O O O (A) Figure 2.3 + O O O O + O + (B) O Ozone resonance structures; the positively charged oxygen is electrophile. Ozone is destroyed by hydroxyl radicals, according to the Equations 2.4ab. The net reaction is the conversion of ozone into oxygen. O3 + OH → O2 + HO2 HO2 + O3 → OH + 2O2 (2.4a) (2.4b) Ozone mass transfer from the gas phase into water is diffusion controlled; the Henry coefficient KH, expressing the equilibrium partitioning of a compound between the gas and liquid phase, is large viz. KH≈3.76⋅103 at 20°C [46] which implies a negligible resistance to mass transfer in the gas film compared to the liquid film. The mass transfer rate is influenced by e.g. the gas phase ozone concentration, temperature, pressure, gas dispersion, solution ionic strength, solution acidity and presence of reactive compounds in the liquid phase. Atomic oxygen Atomic oxygen (O) is produced by dissociation of molecular oxygen, which requires an energy of about 498.4 kJ/mol [24] corresponding to 5.2 eV. In acidic environment the oxygen atom is a stronger oxidizer than ozone, E0 =2.43 V. Its stability is however very limited. In the gas phase, atomic oxygen directly reacts with molecular oxygen to ozone, where the activation energy of this reaction is only Ea=16.7 kJ/mol [47]. Atomic oxygen oxidizes water to hydrogen peroxide. Theory 15 Hydrogen peroxide Hydrogen peroxide (H2O2) [13,14], the dimerization product of hydroxyl radicals, is less reactive than the hydroxyl radical. Its standard reduction potential is E0=1.76 V in acidic environment. By photolysis, hydrogen peroxide decomposes into hydroxyl radicals; the HO-OH bond strength is only 213 ±4 kJ/mol [24], which corresponds to 2.2 eV. Concentrated hydrogen peroxide (>90%) is extremely instable; the decomposition into water and oxygen is strongly exothermic viz. 98.3 kJ/mol. It is due to the ability of hydrogen peroxide to simultaneously oxidize and reduce itself. Hydrogen peroxide is a weak acid: its acid dissociation constant is about pKa=11.75 at T=293 K; however in 50% aqueous solution pKa~9 [47], see Equation 2.5. H2O2 + H2O H3O+ + HO2- (2.5) Hydroperoxyl radical The hydroperoxyl radical (HO2) is a much less strong oxidizer than the hydroxyl radical, ozone or hydrogen peroxide; its standard reduction potential is E0=1.44 V in acidic environment and thus just excels chlorine as oxidizer, see Table 1.1. The HO2 radical is produced in oxygen enriched water from hydrogen atoms that are formed by dissociation of water molecules, see Equation 2.6ab. In alkaline environment, the hydroperoxyl radical exists as the superoxide radical ion O2-, see Equation 2.6c. The acid dissociation constant of the hydroperoxyl radical is about pKa=4.4. Hydroperoxyl radicals often react with each other to hydrogen peroxide and oxygen, see Equation 2.6d [48]. H2O → H + OH H + O2 → HO2 HO2 + OH- O2- + H2O 2HO2 → H2O2 + O2 (2.6a) (2.6b) (2.6c) (2.6d) 2.3. Degradation of organic compounds Chemical oxidation is an important method to degrade organic compounds. The objective of degradation is mineralization i.e. conversion of the target compound to carbon dioxide, water and -depending on the nature of the compound- inorganic ions like e.g. chloride, nitrate, phosphate and sulfate; the inherent toxicity of possibly obtained fluoride and bromate ions cannot be overcome by the oxidation process. In practice, complete mineralization is normally not requested, except for extremely dangerous materials. In many cases it is both justified and efficient to partially degrade the target compound in order to enable further degradation by microbiological treatment. For that case, the chemical oxidation step is needed to destroy persistent molecular structures, to remove high ecotoxicity and enhance water solubility. Examples of degradation pathways of organic compounds are discussed now. The degradation of unsaturated bonds by ozone, hydroxyl radicals and oxygen is discussed. In addition to chemical oxidation, reduction and pyrolysis are briefly mentioned. 16 Chapter 2. Chemical oxidation Ozone can react both directly and indirectly. The indirect way takes place under neutral or alkaline conditions via hydroxyl radicals. The direct way in acidic environment is the electrophilic addition of ozone to unsaturated bonds of alkenes and aromatic compounds. This addition reaction initially produces a molozonide, which rearranges immediately to an ozonide, see Figure 2.4. The ozonide decomposes by ring- cleavage and a zwitterion and an aldehyde or a ketone are produced. In water, the zwitterion hydrolyzes to a hydroxyalkyl hydroperoxide. Depending on the substituent groups, the hydroxyalkyl hydroperoxide decomposes into an aldehyde or a ketone by elimination of hydrogen peroxide or a rearrangement to carboxylic acids occurs [49]. R4 R3 R1 + R2 O O R43 + C OO R3 O R2 R4 O R3 O molozonide + R1 O O R2 R12 R43 ozonide R12 R43 C OOH OH H2 O zwitterion Figure 2.4 R1 O O alkene R12 R4 + + C OO zwitterion R21 R34 O aldehyde/ketone RCOOH, RCHO, R2CO hydroxyalkyl hydroperoxide The reaction of ozone with an unsaturated bond of an alkene or an aromatic compound yields bond cleavage. Products are carboxylic acids, aldehydes or ketones. R is a substituent group. Hydroxyl radicals attack regions of high electron density and therefore add to unsaturated bonds of aromatic compounds and alkenes. Attack of a hydroxyl radical on an aromatic compound produces hydroxycyclohexadienyl radicals; attack of oxygen on these radicals yields endoperoxyalkyl and endoperoxyl radicals; the endoperoxyl radicals yield endoperoxides [19,50], see Figure 2.5. The very instable endoperoxides decompose by ring-cleavage to unsaturated aliphatic hydrocarbons with polyfunctional groups like carboxyl, aldehyde, carbonyl or alkanol groups. Also carbon monoxide may be eliminated. X X HO HO aromatic compound Figure 2.5 ⋅ O2 O O HO X X X ⋅ O2 hydroxy endoperoxy cyclohexadienyl alkyl radical radical O O HO ⋅ HO OO endoperoxyl radical O O -OH,=O polyfunctional aliphatic hydrocarbons endoperoxides The attack of a hydroxyl radical and oxygen on an aromatic compound produces endoperoxides, which decompose to unsaturated aliphatic hydrocarbons with polyfunctional groups. Theory 17 Attack of the hydroxyl radical and oxygen on an alkene produces hydroxyalkylperoxyl radicals. These radicals dimerize to a tetraoxide intermediate. The tetraoxide may decompose in many ways. However, an important pathway is a fragmentation reaction that yields α-hydroxyalkyl radicals, aldehydes/ketones and oxygen. The α-hydroxyalkyl radical scavenges oxygen and produces an α-hydroxyalkylperoxyl radical that yields an aldehyde or a ketone by elimination of hydroperoxyl radicals, see Figure 2.6 [50,51]. R4 R3 R1 R43 HO R2 R34 alkene R43 OH ⋅ OH ⋅ R12 O2 R34 R21 hydroxyalkyl radical R12 + α-hydroxyalkyl radical Figure 2.6 OH R12 ⋅ OO 2x OH R21 ⋅ R43 OH ⋅ R12 + 2 R12 OO O α-hydroxyalkyl peroxyl radical aldehyde or ketone + R34 R21 + O2 O α-hydroxyalkyl radical R43 R12 2 tetraoxide hydroxyalkyl peroxyl radical R43 O2 R43 aldehyde or ketone ⋅ HO2 hydroperoxyl radical The attack of a hydroxyl radical and oxygen on an alkene produces a hydroxyalkylperoxyl radical; dimerization of hydroxyalkylperoxyl radicals yields a tetraoxide intermediate. The tetraoxide decomposes into αhydroxyalkyl radicals, aldehydes/ketones and oxygen. The attack of oxygen on an α-hydroxyalkyl radical yields an aldehyde or a ketone by elimination of a hydroperoxyl radical. Hydroxyl radicals also abstract hydrogen atoms from a saturated hydrocarbon chain, by which radical sites are created on the hydrocarbon chain where oxygen can attack. This results in the formation of unsaturated bonds and hydroperoxyl radicals, see Figure 2.7. The produced unsaturated hydrocarbon will be cleaved by ozone attack, see Figure 2.4. H HO H -H2O saturated hydrocarbon Figure 2.7 H • radical hydrocarbon O2 H H H O O peroxy radical hydrocarbon + HO2• unsaturated hydroperoxyl hydrocarbon radical Hydrogen abstraction from a saturated hydrocarbon chain by a hydroxyl radical, followed by oxygen attack produces an unsaturated hydrocarbon and a hydroperoxyl radical. During oxidation, covalently bonded halogens, nitrogen, phosphorous and sulfur -if present- are removed from the target molecule and converted to inorganic ions like halides, nitrates, phosphates and sulphates. Figure 2.8 shows the oxidation of dichloromethane by hydroxyl radicals and oxygen, eventually yielding carbon monoxide, carbon dioxide and hydrogen chloride; the intermediate phosgene is highly toxic, but it rapidly hydrolyzes to carbon dioxide and hydrogen chloride [50,51]. 18 Chapter 2. Cl H C H Cl dichloromethane Cl H C Cl O2 H Cl C H O2 -H2O + Cl H C OO• - HO Cl HO -ClOH H Cl C OO• H O H2O CO2 + 2H+ + 2Cl- Cl Cl phosgene CO + HO + H+ + Cl- The oxidation of dichloromethane by hydroxyl radicals and oxygen eventually yields carbon monoxide, carbon dioxide and hydrogen chloride; phosgene is a highly toxic intermediate, which rapidly hydrolyzes to carbon dioxide and hydrogen chloride. Figure 2.8 Reduction Reduction of unsaturated hydrocarbons by hydrogenation does not invoke bond/ringcleavage but only saturation takes place [52]. Nevertheless aromaticity is destroyed in this way. Figure 2.9 shows the hydrogenation of benzene to cyclohexane. In contrast to benzene, cyclohexane is less harmful [1]. H benzene H ⋅ H hyd H hydroxycyclo cyclohexadienes hexadienyl radicals hyd cyclohexene cyclohexane Reduction of benzene to cyclohexane by hydrogenation (hyd). Figure 2.9 Reduction of azo dyes invokes cleavage of the azo bond (-N=N-), which implies fragmentation of the azo dye molecule into two amino (RNH2) compounds, see Figure 2.10. This reaction explains reductive fading (decolorization) of the dye, induced by ketyl or carboxy radicals [53]. However, this degradation reaction produces e.g. highly harmful aniline. O O NH OH NH OH N SO3Na N SO3Na NH2 reduction 4H+ + 4e- SO3Na NH2 + SO3Na aniline Acid Red 1 Figure 2.10 The reduction of the azo dye Acid Red 1; this reaction is an example of reductive fading (dye decolorization). Theory 19 Pyrolysis Pyrolysis is thermal decomposition of an organic compound in the absence of oxygen. By pyrolysis, molecules are dissociated into radicals and elimination of functional groups may take place like e.g. decarboxylation of carboxylic acids, dehydration of alkanols and esters, dehalogenation, loss of nitro and sulfone groups, nitrogen, carbon monoxide, see Figure 2.11. Thermal cracking of alkanes yields lower molecular weight alkenes, but pyrolysis of simple aromatics leads to polymerization viz. polycyclic aromatic hydrocarbons [54]. 1. CCl4 CCl3 + • NO2 Cl • O + 2. OH 3. O OH OH OH O 4. Figure 2.11 NO O OH + CO2 + Pyrolysis of organic compounds: 1. Dissociation of tetrachloromethane; 2. Nitric monoxide release from p-nitrophenol; 3. Decarboxylation of malonic acid yields acetic acid; 4. Cracking of n-butane to ethane and ethylene. Knowledge of the target compound conversion level and/or conversion efficiency is not sufficient to qualify an advanced oxidation process. It is also very important to identify the intermediate and final oxidation products. During the oxidation of organic compounds intermediates may be produced, which exhibit higher toxicity than the target compound. Examples are dibenzofurans and dioxins, produced from supercritical water oxidation of phenol, as will be discussed in section 2.4.1. The oxidation of halogenated hydrocarbons yields highly harmful halogenated aldehydes/carboxylic acids [51,55]. Therefore oxidation progress must have proceeded, until these harmful intermediate products have been converted. 20 Chapter 2. 2.4. Oxidation of model compounds A survey from literature is presented about the general oxidation pathways of the applied model compounds phenol, atrazine, malachite green and dimethyl sulfide. 2.4.1. Phenol The oxidation of phenol produces a wide oxidation product range, consisting of polyhydroxybenzenes/quinones, ring-cleavage products and polymerization products. Literature data originate from radiolysis [50,56,20], oxidation by hydrogen peroxide [11,57,58], ozone [7] and other chemical oxidizers [34,52], photocatalytic oxidation [15], photolysis [59] and oxidation by supercritical water [18,60,17]. Polyhydroxybenzenes and quinones Among the polyhydroxybenzenes [61] are the dihydroxybenzenes (DHB’s): catechol (1,2-DHB), resorcinol (1,3-DHB), hydroquinone (1,4-DHB) and the trihydroxybenzenes (THB’s): pyrogallol (1,2,3-THB), hydroxyhydroquinone (1,2,4-THB) and phloroglucinol (1,3,5-THB), see Figure 2.12. These compounds are produced by attack of the hydroxyl radical on the benzene ring. With increasing amount of hydroxyl groups attached to the benzene ring, the stability of the hydroxybenzene towards oxidation strongly decreases [52]. Therefore higher hydroxylated benzenes are not likely to be found during vigorous oxidizing conditions. Quinones are produced by oxidation of polyhydroxybenzenes. The following quinones are reported: 1,4-benzoquinone, 1,2-benzoquinone and hydroxybenzoquinone. 1,4benzoquinone is produced by oxidation of hydroquinone. 1,2-benzoquinone is a very unstable oxidation product of catechol. Hydroxybenzoquinone is produced by hydroxylation of 1,4-benzoquinone or partial oxidation of hydroxyhydroquinone; it is reported to undergo polycondensation in aqueous solutions. 1,3-benzoquinone does not exist, because the structure would be nonplanar and highly strained [62]. OH OH OH OH OH OH OH OH OH catechol resorcinol OH hydroquinone O O O O pyrogallol OH OH OH HO OH hydroxy- phloroglucinol hydroquinone OH O 1,4-benzoquinone 1,2-benzoquinone Figure 2.12 Polyhydroxybenzenes and quinones produced by the oxidation of phenol. O hydroxybenzoquinone Theory 21 Ring-cleavage products The oxidation of polyhydroxybenzenes and quinones produces ring-cleavage products. Observed products are unsaturated and saturated C1-C6 hydrocarbons with polyfunctional groups like carboxyl-, aldehyde-, ketone- or alkanol- groups, see Figure 2.13. Alkanol-functional groups are oxidized to aldehyde groups, while aldehydes are oxidized to carboxylic acids. The following classes can be mentioned: Saturated monocarboxylic acids: formic, acetic, propionic and glyoxylic acid. Saturated dicarboxylic acids: oxalic, malonic, ketomalonic, D,L-malic, succinic, glutaric and adipic acid. Unsaturated monocarboxylic acids: acrylic acid. Unsaturated dicarboxylic acids: maleic, fumaric and cis,cis-muconic acid. Saturated aldehydes: formaldehyde, acetaldehyde and glyoxal. Unsaturated hydrocarbons: acetylene and butadiene. Acetylene is reported to be produced under supercritical conditions by addition of oxygen to catechol [17]. Butadiene is reported as a decomposition product of hydroxyhydroquinone [16]. O O H O H H H H O formaldehyde acetaldehyde O O H glyoxyal OH OH formic acid O O oxalic acid O malonic acid OH O adipic acid OH O ketomalonic acid OH D,L-malic acid O O O Figure 2.13 O HO O HO OH acrylic acid O OH O OH O succinic acid glutaric acid OH O OH OH HO OH O O OH O propionic acid glyoxylic acid OH O OH H OH acetic acid O HO O O HO O OH maleic acid HO OH O fumaric acid O OH O OH cis,cis-muconic acid Aldehydes and carboxylic acids produced by the oxidation of phenol. 22 Chapter 2. Polymerization products The radical-induced oxidation of phenol also invokes molecular coupling, see Figure 2.14. Reported dimerization products are 4,4’/2,4’/2,2’-dihydroxybiphenyl and 4/2hydroxydiphenylether. These products are formed by dimerization of phenoxy radicals. Purpurogallin is produced from the dipolar dimerization of the ortho-quinone of pyrogallol. Supercritical water oxidation of phenol yields the following multiring condensation products: dibenzofuran, dibenzofuranol, dibenzo-p-dioxin, 9H-xanthene-9one, 2,3-dihydro-1H-indene-1-one. Polymerization products are the so-called “synthetic humic acids” consisting of hydroquinone and (hydroxy)benzoquinone monomeric units; these amorphous products exhibit a dark-brown color. The toxicity of the coupling products is higher to much higher than the toxicity of phenol. Especially the benzofurans and dioxins are highly unwanted, but these compounds are avoided or destroyed by supercritical conditions over T=600°C. OH OH OH O OH OH OH OH O OH 4,4'-DHBP 2,4'-DHBP 2,2'-DHBP dihydroxybiphenyl 4-HDE 2-HDE hydroxydiphenylether O O O O dibenzofuran dibenzofuranol OH O O O dibenzo-p-dioxin 9H-xanthene-9-one OH OH HO HO purpurogallin O OH OH O O O O OH n OH O n synthetic humic acids Figure 2.14 O Polymerization products formed during the oxidation of phenol. 2,3-dihydro-1HIndene-1-one Theory 23 2.4.2. Atrazine The oxidation of atrazine involves deaminoalkylation, dechlorination and hydroxylation of the s-triazine ring. Ring-cleavage has not been reported. Literature data have been obtained from photocatalytic oxidation [36] and photo-Fenton oxidation [63,64]. Some important oxidation products are shown by Figure 2.15. By oxidation of the aminoalkyl groups the following products are formed: 4-acetamido2-chloro-6-(isopropylamino/ethylamino)-s-triazine. Partial dealkylation products are deethylatrazine and deisopropylatrazine. Complete oxidation of the alkyl groups yields diaminoatrazine. The dechlorination preferentially occurs after considerable degradation. Due to deaminoalkylation and dechlorination nitrate and chloride ions are produced. Also ethane has been detected. The final oxidation product is cyanuric acid (2,4,6trihydroxy-1,3,5-triazine). Cl N Cl N N HN N H 2N N NH N NH deethylatrazine O Cl N HN N N Cl 4-acetamido-2-chloro6-(isopropylamino)-s-triazine NH HN N N NH NH2 deisopropylatrazine HO N N OH cyanuric acid Cl O 4-acetamido-2-chloro6-(ethylamino)-s-triazine N N N HN Cl N atrazine N OH N H2 N N N NH2 diaminoatrazine Figure 2.15 Atrazine and some major oxidation products. 2.4.3. Malachite green The oxidation of malachite green is described in literature by the lowering of light fastness [53]. Two degradation mechanisms may apply viz. dealkylation (methyl groups attached to nitrogen) and molecular fragmentation, starting from the carbinol base. Reported products are N,N-di- and N-monomethyl-4-aminobenzophenone and N,Ndimethyl-4-aminophenol, see Figure 2.16. 24 Chapter 2. N N + N N OH [ox] OH H+ malachite green MG carbinol base O O N N N N,N-di- and N-monomethyl4-aminobenzophenone Figure 2.16 OH N,N-dimethyl4-aminophenol Malachite green and some of its oxidation products. 2.4.4. Dimethyl sulfide Literature data on the oxidation of dimethyl sulfide have been obtained from [39,40,65]. Dimethyl sulfide is initially oxidized to dimethyl sulfoxide, which can be further oxidized to dimethyl sulfone, methanesulfonic acid and finally sulfuric acid, see Figure 2.17. S S O O S O dimethyl sulfide dimethyl sulfoxide dimethyl sulfone [ox] Figure 2.17 [ox] [ox] O S OH O methanesulfonic acid [ox] O HO S OH O sulfuric acid Dimethyl sulfide and some of its oxidation products. 2.5. Diagnostics An overview is presented of major chemical, electrical and optical diagnostics, which have been applied to study the formation of oxidizers and the oxidation of model compounds by pulsed corona discharges. Applied chemical diagnostics are liquid chromatography, mass spectrometry, aldehyde screening, electron spin resonance, Microtox ecotoxicity, total organic carbon content and acidity. Electrical diagnostics are corona pulse voltage & current measurements and conductometry. Applied optical diagnostics are UV absorbance spectrometry and fluorescence & infrared spectroscopy. Theory 25 2.5.1. Chemical diagnostics Liquid chromatography Separation of the liquid-phase oxidation product mixture into its components is necessary for determination of the conversion of the target compound and identification of the oxidation products. In this way every product can be detected separately with maximum sensitivity, because mutual influence is not possible. In this thesis the liquidphase oxidation product mixture has been separated by reversed-phase high performance liquid chromatography (rp-HPLC) and ion-exclusion chromatography (ICE) [66,67,68]. In liquid chromatography, a sample in a carrier flow i.e. eluent or mobile phase is introduced into the separation column containing a stationary phase. The sample components will partition between the stationary phase and the mobile phase due to component-specific physical interaction mechanisms. In this way component-specific retention thus separation is established. Detection of the eluting components is commonly performed by a UV absorbance detector but the detection by e.g. mass spectrometry, fluorescence, electrical conductivity or refractive index are also optional. Rp-HPLC has been initially applied for exploration of the complex oxidation product mixture. It is the most applied and versatile liquid chromatography technique, suitable for separation of a wide group of organic compound classes including phenols, polycyclic aromatic hydrocarbons, alkanols and alkanes. The retention mechanism is based on non-specific hydrophobic interaction (dispersive forces) but also on dipoledipole and proton donor/acceptor interaction. In rp-HPLC the non-polar stationary phase is commonly an alkyl-bonded silica packing e.g. a C8- or C18-alkane grafted on silica; the polar mobile phase is a mixture of water and an organic modifier. The retention behaviour of the components, thus the separation, can be adjusted by changing the eluent composition i.e. eluent strength. The eluent strength is the power of the eluent to displace components interacting with the stationary phase. The used eluents often consist of mixtures of water and acetonitrile or methanol, which do not absorb the UV light of the absorbance detector at analytically-important wavelengths (cutoff wavelength: acetonitrile: 190 nm, methanol: 210 nm, water: 191 nm [69]). The eluent dosage is performed at constant composition (isocratic conditions) or by means of a gradient (increasing eluent strength). The gradient dosage is applied to force the elution of compounds that strongly interact with the stationary phase. Unfortunately rp-HPLC is not suitable for the separation of carboxylic acids, which are important oxidation products. In order to resolve the carboxylic acids, ion-exclusion chromatography (ICE) has been applied. The retention mechanism is mainly based on the exclusion of anions from a cationic-exchange resin. The anions cannot penetrate the resin, because they encounter the Donnan potential, which guarantees the electrical neutrality within the resin. The applied ICE column has a polystyrene-divinylbenzene partially crosslinked resin with sulfonate (RSO3-H+)-functional groups. An aqueous solution of a strong acid, here trifluoroacetic acid, is used as eluent. Strong carboxylic acids exist in ionic form (H+A-) and will be excluded from the stationary phase. On the contrary, weak carboxylic acids in molecular form will be able to diffuse into the resin pores. 26 Chapter 2. The carboxylic acids are thus separated by their acidic strength, which is reflected by the acid dissociation constant pKa. Other ICE retention mechanisms are hydrophobic interaction and molecular size, both originating from the partially cross-linked resin. The separation can be influenced by the eluent acidity, because the eluent acidity determines the dissociation behaviour of the carboxylic acids according to the Henderson-Hasselbalch relationship, see Equation 2.7. α Equals the dissociation fraction. α pK a = pH − log 1− α and α= [ A− ] [HA] + [ A− ] HA + H2O H3O + + A− (2.7) Although the eluent strength will also influence the retention behaviour of ion-exclusion chromatography, the addition of organic modifiers to the acidic aqueous mobile phase has not been applied for the used column. The partially cross-linked polymeric resin is expected to swell due to the absorption of organic modifier, which may result in cracking of the resin. Next to UV absorbance detection, the carboxylic acids have been detected using a conductivity detector. However, the acidic eluent necessary for ICE separation causes a high background conductivity, which makes the detection of separated carboxylic acids impossible. Therefore the eluent conductivity has to be decreased by removal of the highly conductive hydronium ions, which is accomplished by a suppressor. The applied suppressor is a micromembrane suppressor, which consists of cation-exchange membranes and is supplied with an aqueous ammonia solution. According to the Donnan equilibrium, these membranes exclude the trifluoroacetate anions of the trifluoroacetic acid eluent but allow hydrogen ions to pass by exchange with ammonium ions, while electrical neutrality is maintained. The highly conductive hydronium ion is thus replaced by the less conductive ammonium ion. Liquid chromatography / mass spectrometry In order to identify the oxidation products after separation by the liquid chromatograph, a mass spectrometer is connected to the LC system by means of a special interface. This LC-MS coupling has to introduce a huge eluent flow (typically about F=1 ml/min) containing tiny amounts of separated components, into the high vacuum of the quadrupole mass spectrometer. In this work an IonSpray interface has been utilized [70,71]. The IonSpray interface is a pneumatical and electrostatical nebulizer. A splitted fraction of the eluent carrying the separated components from the LC system is introduced into a hollow needle, which is energized at high voltage. Together with the eluent a nebulizer gas flow is introduced; if necessary in combination with an organic solventbased make-up liquid to improve the sprayability of high water content eluents. A mist of highly charged droplets is produced in the direction of the mass spectrometer. Due to evaporation the droplets decrease in size and the electrical field at the surface of the droplet increases. When a critical field has been reached, ions are emitted from the surface of the droplets. These ionization conditions are very mild: no fragmentation takes place and thus molecular ions are produced. These ions are transferred to the orifice, where they can enter the quadrupole mass-spectrometer. Eluent molecules are prevented from entering the mass spectrometer by a gas curtain interface. Theory 27 IonSpray is particularly suitable for the analysis of thermolabile and ionic components. For the identification of carboxylic acids and hydroxybenzenes, the IonSpray interface has been operated in the negative ion mode: the applied needle voltage is negative with respect to the grounded wall. Also ammonia has been added to the aqueous oxidation product mixture, to facilitate the production of anions (e.g. phenolate, carboxylates). Next to the IonSpray interface, the Atmospheric Pressure Chemical Ionization (APCI) interface has been tested for applicability. Here, the splitted flow from the LC system is nebulized and evaporated by heating. The evaporated flow is directed to a corona discharge needle, where chemical ionization of sample and solvent takes place. APCI is particularly suitable for the processing of high eluent flow rates containing high concentration of electrolytes. However, the sample components have to be thermostable. Electron-impact mass spectrometry In addition to IonSpray-MS, electron-impact mass spectrometry (EI-MS) has been performed for the identification of phenol oxidation products. After evaporation in a prevacuum compartment, the mixture components are introduced into the main vacuum compartment by differential pumping. Here the components are ionized by an electron beam. Produced are molecular fragments but also molecular ions. Solid phase extraction Solid phase extraction (SPE) [66] is a sample preparation technique implying a trace enrichment procedure and solvent transfer step, in order to obtain samples with detectable amounts of components in a suitable solvent. The SPE column contains a sorbent bed material that is similar to the stationary phases used high performance liquid chromatography. The procedure involves four steps. First the SPE sorbent bed is conditioned, where an organic solvent is used to increase the interaction surface area and displace contaminants; excess solvent is then removed by rinsing with a liquid similar to the sample solvent. The second step is the sorption of the mixture components onto the sorbent bed. The third step is the removal of undesired sample matrix with a weak solvent. In the last step, the components are desorbed into a small volume by means of a solvent with sufficient strength. Adsorption and desorption are performed by vacuum suction. Aldehyde screening test The production of volatile aldehydes during the oxidation of phenol has been verified by a gas sampling aldehyde screening test according to NIOSH [72]. This test is suitable to identify C1-C7 aliphatic saturated aldehydes (e.g. formaldehyde, acetaldehyde) but also unsaturated aldehydes like acrolein and crotonaldehyde. The test involves the collection of gas phase components in a sampling tube, filled with the derivatization agent 2(hydroxymethyl)-piperidine (HMP) on a carrier phase. HMP specifically reacts with aldehydes to an oxazolidine compound, see Figure 2.18. 28 Chapter 2. + N H R O N H OH O + H2 O R HMP aldehyde Figure 2.18 oxazolidine The chemical derivatization of an aldehyde with HMP. The oxazolidine can be identified by gas chromatography-mass spectrometry. The molecular ion of a specific aldehyde equals the molecular weight of the original aldehyde plus 97. For example formaldehyde (HCHO) is recognizable by a base peak at m/z=97 and other characteristic ions have m/z values 126 and 127 (molecular ion C7H13NO). Electron spin resonance Electron spin resonance (ESR) has been applied in-situ, to identify oxidizer radicals produced by the corona discharges and radical intermediate phenol oxidation products. Of interest are inorganic oxidizer radicals e.g. the hydroxyl radical (HO•) / oxygen anion (O-•) and hydroperoxyl radical (HO2•) / superoxide anion (O2-•); organic radical intermediates that are produced by oxidation of phenol are e.g. the dihydroxycyclohexadienyl[peroxyl] radicals C6H5(OH)2[OO]•. Next to the mentioned radicals, hydrogen atoms may be detectable. A spin trap has been applied to trap the short living radicals and form a stable measurable adduct. Literature references originate from in-situ glow discharge electrolysis [73], in-situ radiolysis [74] and ex-situ sonolysis [22] of aqueous spin trap solutions. 5,5-dimethyl-1-pyrroline N-oxide (DMPO) spin trap has been used. The formation of the DMPO-OH adduct from DMPO and the hydroxyl radical is shown by Figure 2.19. The DMPO-OH adduct exhibits a half-life of about 2 hours [75]. The DMPO-OOH adduct, however, is reported to be highly unstable [76]. N O H + DMPO Figure 2.19 HO N O H OH DMPO-OH The trapping of a hydroxyl radical by the spin trap DMPO. The ESR spectrum of the DMPO-OH adduct consists of a 1:2:2:1 quartet, due to the equivalence of the 14N and 1H hyperfine coupling constants viz. aN=aH=1.49 mT. DMPO-H has a nine-line spectrum due to coupling with one 14N nucleus (aN=1.66 mT) and 2 identical protons (aH=2.25 mT). However, no hyperfine coupling constants are known for the adduct of DMPO and the organic DHCHD(P) radicals. Theory 29 Microtox ecotoxicity test In order to investigate the detoxification progress during the degradation of phenol, Microtox ecotoxicity tests have been performed [77]. In the Microtox test, the ecotoxicological effect of a chemical substance or a mixture is determined from the bioluminescence behaviour of the marine bacterium Vibrio fischeri before and after exposure to the medium. The bioluminescence output decreases with increasing toxicity of the chemical compound. The ecotoxicity is expressed as an effect concentration (EC) at a particular effect level, in this thesis as a 20 % bioluminescence decrease (E20). Low EC values imply high ecotoxicity. The EC value is determined by interpolation of the dose-effect relationship, see Equation 2.8. The effect is expressed by the gamma value (Γt), that is related to the inhibitory effect (Ht) according to Equation 2.9. The inhibitory effect is calculated from the bioluminescence intensity after the exposure (ITt) and a corrected background intensity (Ict), according to Equation 2.10. log(ct ) = b ⋅ log(Γt ) + log(a) Γt = Ht 100 − Ht Ht = (2.8) (2.9) Ict − ITt ⋅100 Ict (2.10) Evidence for acute ecotoxicity is obtained, by measuring the effect level at different exposure times. Table 2.1 shows EC505min values for phenol and some of its oxidation products [78]. Although 50% effect levels are more commonly reported in literature, the measurement of 20% effect levels implies higher sensitivity of the Microtox test. It should be noted, that a comparison of effect concentrations determined at different levels i.e. 20% or 50% and different exposure times like 5 min, 15 min or 30 minutes is not possible. The deviation of the reported EC50 values is probably caused by the origin of the applied bacteria cultures viz. the supplier and supplied state (freeze-dried, liquiddried or fresh). Table 2.1 EC505min effect concentrations of phenol and some of its oxidation products. The phenol EC505min value is an average value determined from a set of 22 literature references [78]. Compound Phenol Catechol Resorcinol Hydroquinone 1,4-Benzoquinone Formaldehyde Glyoxal Glyoxylic acid Oxalic acid Formic acid Acetic acid EC505min (mg/l) 28.6 ±7.9 32.0 310; 375 0.042; 0.079 0.0085; 0.020; 0.08; 1.4 3.0; 8.7; 9.0; 10; 10.1; 904.17 754 ±55.1 11.2 ±0.16 11.3 ±0.22 7.91 ±0.22 9.24 ±0.38 30 Chapter 2. Total organic carbon A different way to measure oxidation progress of an organic compound is the determination of the carbon content of the oxidation product mixture. The total carbon content TC is defined as the sum of the total organic carbon TOC (hydrocarbons) and total inorganic carbon TIC (carbonate, bicarbonate, carbon dioxide) [79]. Due to oxidation, the carbon skeleton of an organic compound is gradually chopped into shorter carbon chain molecules containing oxygen-based functional groups viz. aliphatic aldehydes and carboxylic acids. The last member in the oxidation sequence is formic acid, which upon oxidation yields the unstable carbonic acid. The TOC level of the oxidation product mixture decreases by release of carbon dioxide (mineralization) and volatile or gaseous intermediates. The measurement of total organic carbon does not reveal information about the chemical or toxicological properties of the sample. However, if the total carbon content equals the total inorganic carbon content, the organic carbon has been completely mineralized and the remaining toxicity is only due to inorganic ions originating from covalently bonded elements. Acidity The acidity of the oxidation product mixture is an indication for oxidation progress, because the oxidation of an organic compound yields carboxylic acids. Unfortunately, it is impossible to deconvolute the overall acidity into concentrations of all carboxylic acids present in the oxidation product mixture. The produced carboxylic acids are generally weak acids, compared to mineral acids like hydrochloric acid and nitric acid. The strongest acid, produced by oxidation of phenol is oxalic acid. The acidic strength is expressed by the acid equilibrium constant Ka, according to Equation 2.11. The acid dissociation constant pKa=-10log(Ka) of phenol and some of its acidic oxidation products is shown by Table 2.2 [80]. Ka = [H3O + ] ⋅ [ A− ] [HA] Table 2.2 HA + H2O H3O + + A− 2.11 Acid dissociation constants of phenol and some of its acidic oxidation products; dibasic acids (H2A) show a two-step dissociation. Compound Oxalic acid Maleic acid Glyoxylic acid Formic acid Phenol pKa,I 1.23 1.83 3.18 3.75 9.89 pKa,II 4.19 6.07 - Theory 31 2.5.2. Electrical diagnostics Corona pulse energy & conversion efficiency The energy of a single corona discharge pulse (Ep) is determined from the pulse voltage V(t) and corona current Icor(t). The corona discharge current Icor(t) is calculated from the difference of total current I(t) and capacitive current Icap(t), see Equation 2.12. The capacitive current is determined by measuring the capacitance of the reactor geometry Cg at voltages below the corona onset, where I(t)=Icap(t), see Equation 2.13. The corona pulse energy Ep is then calculated by integration of pulse voltage times corona current over the pulse time, see Equation 2.14 [81]. Icor (t) = I(t) − Icap (t) (2.12) dV (t) ∧ dt E p = ∫ V (t) ⋅ Icor (t) dt Icap (t) = Cg ⋅ V < Vonset (2.13) (2.14) pulse The conversion efficiency is defined according to the G yield value. G expresses the number of target compound molecules converted with regard to the required energy input, illustrated by Equation 2.15. X is the conversion of the target compound, C0 is the initial target compound concentration (mol/l), Vol is the solution volume (l), Ep is the pulse energy (J), f is the pulse repetition rate (Hz) and t is the oxidation time (s). In literature G units are expressed as mol/J, “molecules per 100 eV” or g/kWh. Table 2.3 shows the interconversion of the efficiency units. G= X ⋅ C0 ⋅ Vol Ep ⋅ f ⋅ t Table 2.3 (2.15) The interconversion of G yield efficiency units. mol/J (100eV)-1 g/kWh mol/J= 1 1 100N Ae 1 3.6 ⋅ 106 FW (100eV)-1= 100NAe 1 100N Ae 3.6 ⋅ 106 FW 3.6 ⋅ 106 FW 100N Ae 1 g/kWh = NA e FW 6 3.6⋅10 FW = Avogadro’s constant: 6.022⋅1023 mol-1 = 1.6022 ⋅10-19 J/eV = target compound molecular weight (g/mol) 32 Chapter 2. Electrical conductometry Electrical conductometry has been applied to measure the conductivity of the oxidation product mixture in order to monitor oxidation progress. The oxidation of an organic compound yields carboxylic acids, which are partially dissociated in aqueous solution depending on their acid strength and therefore contribute to the electrical conductivity. The conductance G (unit Siemens) is the reciprocal value of the electrical resistance R according to Ohm’s law, see Equation 2.16. ρ Equals the resistivity, A and l are the conductor’s area and length. G= 1 = R 1 l (ρ ⋅ ) A = 1 A 1 ⋅ =σ ⋅ ρ l K (2.16) The conductivity σ equals the product of the conductance G and the cell constant K. With regard to the phenol oxidation product mixture, conductivity is nearly exclusively due to protons. The carboxylic acid anions and other anions like chloride and nitrate contribute to a limited extent to conductivity, as is illustrated by the molar conductivity at infinite dilution λ0=(σ/c)c→0, see Table 2.4 [82]. Table 2.4 Molar conductivity at infinite dilution for some ions observed during oxidation of an organic compound. Ion Proton Chloride Oxalate Nitrate Formate Acetate H-Oxalate *) Formula H+ Cl(½)C2O42- *) NO3HCOOCH3COOHC2O4- λ0 (m2Smol-1)⋅104 349.7 76.3 74.1 71.4 54.6 40.9 40.2 Double charge 2.5.3. Optical diagnostics UV absorbance spectrometry Quantitative ozone measurements have been performed by UV absorbance spectrometry. By absorption of energy, a molecule is transferred from its ground state to an excited state. The relaxation of the molecule to the ground state may involve radiationless transfer, fluorescence or phosphorescence, depending on the electronic structure of the molecule. According to Lambert-Beer’s law, the intensity of the absorbed radiation (I) is related to the number density of the absorbing molecule (N), the absorption cross section (σ(λ)) and the optical path length (x) see Equation 2.17. This relation is valid for monochromatic light and diluted solutions viz. concentration ≤10-2 mol/l [83]. The cross section for ozone at λ=260 nm is σ260=1.14⋅10-21 m2 [9,84]. I = I0 exp(− Nσ x) (2.17) Theory 33 Laser-induced fluorescence spectroscopy Laser-induced fluorescence (LIF) spectroscopy has been applied for in-situ conversion measurements of phenol in aqueous solution. The fluorescent properties of a molecule are determined by both electronic and structural requirements [85]. Electronic requirements are that the molecule absorbs energy frequencies lower than the strength of the weakest bond, the molecule’s first excited singlet state S1 has a life-time of about 10-8 s and the first excited singlet state (S1) and lowest triplet state (T1) are well separated. In all media, except for lowpressure gasses, organic molecules emit fluorescence from the lowest vibrational level of the first excited singlet state to the singlet ground state (S0), see Equation 2.18. S1 → S0 + hνfl (2.18) Structural requirements are that the molecule contains a planar rigid conjugated system of double bonds, preferentially in a cyclic structure and especially in linear polycyclic molecules where the π electrons can readily circulate inside the molecule; groups substituted to the conjugated system shall be electron donating like e.g. hydroxyl and (dimethyl)amino groups and polysubstitution shall not influence the π electron mobility. The application of LIF spectroscopy for monitoring the phenol conversion during oxidation in aqueous solution is not straightforward, because of following reasons. The fluorescence radiation from phenol excited states is quenched by other phenol molecules, oxidation products and water. The excitation laser beam is absorbed by phenol and the oxidation products in aqueous solution. Some oxidation products also show fluorescence by laser excitation. The different processes that occur after excitation of a phenol molecule by the laser beam are described by Figure 2.20. Four energy levels are described viz. the ground state (0), the initial phenol excited state (2), a lower excited phenol state (1) and the electronic state of a quenching molecule Q (3). 2 n2/τ 1 B02 n0 ρ kQnQn2 3 kQnQn1 A20n2 A10n1 0 Phenol Figure 2.20 Q A Jablonski diagram for different energy states of phenol and a quenching compound Q. 34 Chapter 2. ni equals the population density of state i. A20 and A10 are the Einstein coefficients for spontaneous emission from state 2 to 0 and state 1 to 0 respectively. kQ is the quenching coefficient. τ is the characteristic time constant for decay from state 2 to state 1. B02 is the Einstein coefficient for induced absorption for the transition from state 0 to 2. ρ equals the laser photon density, which can be expressed according to Lambert-Beer’s law by Equation 2.19. ρ0 equals the laser intensity before entering the aqueous solution, σ02 is the absorption cross section of phenol in aqueous solution and n0 is the population density of the phenol ground state 0. ρ = ρ 0 exp(−σ 02Ln0 ) (2.19) The population density of excited state 1 grows by decay of excited state 2 to the excited state 1 according to n2/τ. The population density of excited state 1 decreases by decay to the groundstate 0 by A10n1 and by quenching due to molecule Q according to an amount kQnQn1, see Equation 2.20. The population density of excited state 2 grows by laser excitation by an amount B02n0ρ, The population density of the excited state 2 decreases by decay to the excited state 1 by n2/τ, by decay to the ground state 0 by A20n2 and by quenching due to molecule Q according to an amount kQnQn2, see Equation 2.21. dn1 n2 = − A10 n1 − kQ nQ n1 dt τ (2.20) dn2 n = B02 n0 ρ − 2 − A20 n2 − kQ nQ n2 τ dt (2.21) The assumption of steady-state conditions is valid for collision-dominated conditions, which is justified for the liquid phase. Then, the derivates of population density with respect to time are equal to zero and the fluorescence intensity can be expressed by Equation 2.22. ILIF = ηA10 n1 = η B 02 ρ 0 n0 exp (− σ 02 Ln0 ) (kQ nQ + A10 )(kQ nQτ + 1 + A20τ ) (2.22) η is a proportionality constant. From Equation 2.22 it can be derived that the fluorescence intensity versus the population density n0 reaches a maximum value at n0=(σ0L)-1. At higher phenol concentrations absorption and quenching overrule the fluorescence. Phenol complies with the requirements for fluorescence. When excited at a wavelength of about λex=270 nm in aqueous solution, phenol emits a maximum fluorescence near 300 nm<λfl<310 nm [85-90]. The excitation wavelength of the dihydroxybenzenes is in same range i.e. 265 nm<λex<285 nm and these compounds show a somewhat weaker fluorescence at the wavelengths 315 nm<λfl<340 nm. Theory 35 The trihydroxybenzenes are weakly to non-fluorescent. Phloroglucinol shows the weakest fluorescence, possibly because it also reacts in a tautomeric keto-form [61] which is non-aromatic, see Figure 2.21. O OH O OH HO phloroglucinol O 1,3,5-cyclohexanetrione Keto-enol tautomerism of phloroglucinol (1,3,5-trihydroxybenzene). Figure 2.21 1,4-Benzoquinone is not a hydroxybenzene but forms a redox couple with hydroquinone; the compound is non-aromatic viz. a cyclic diene and also shows weak fluorescent properties. Saturated hydrocarbons do not comply with the requirements for fluorescence and therefore major literature references on fluorescence involve solely aromatic compounds. Fluorescent molecular probe A fluorescent molecular probe has been applied to identify the hydroxyl radical in corona-exposed aqueous solution. This molecule is a non-fluorescent compound, that specifically reacts with hydroxyl radicals in a very sensitive way to produce a strongly fluorescent molecule. In this thesis the molecular probe coumarin-3-carboxylic acid, abbreviated as CCA, has been utilized [91,92]. CCA reacts with the hydroxyl radical to 7-hydroxycoumarin-3-carboxylic acid, abbreviated as 7-OHCCA, see Figure 2.22. O O HO O O OH OH O CCA Figure 2.22 O 7-OHCCA The non-fluorescent CCA and the strongly fluorescent product of OH and CCA: 7-OHCCA. The excitation wavelength of 7-OHCCA is λex=396 nm. The fluorescence maximum is near λfl=450 nm. Using CCA, time resolved measurement of the hydroxyl radical production rate kinetics is possible. However, a problem is the very limited solubility of CCA in water. The choice for a better solvent or solvent addition are no option: the hydroxyl radical will react with the solvent molecule and quantitative measurements of the hydroxyl radical concentration are not accurate anymore. 36 Chapter 2. Infrared spectroscopy For identification of gaseous phenol oxidation products, infrared spectroscopy has been applied [93-95]. The gross selection rule for IR spectroscopic activity is that a molecule’s dipole moment shall change during the normal mode of vibration. The change of the dipole moment establishes an electrical field that can interact with the electrical vector of the radiation. Normal modes of vibration are mutually-independent synchronous motions of atom groups. Some typical molecular vibrations are stretch vibrations (symmetric and anti-symmetric) and deformations (scissoring, rocking, wagging, twisting). Linear molecules consisting of N atoms vibrate in 3N-5 different ways while non-linear molecules show 3N-6 different vibrations. The wavelength range of interest to organic compounds is 4000-600 cm-1 corresponding to 2.5-17 µm. This region can be divided into two parts viz. the absorption range due to characteristic functional groups (4000-1200 cm-1) and the finger-print area, where absorptions provide information about the overall constitution of the molecule (1200-600 cm-1). With regard to the oxidation of phenol, the following compound classes are of major interest: carbon oxides, aldehydes and carboxylic acids. Indirectly involved are ozone and nitrogen oxides produced by corona in air. Carbon dioxide shows two stretch vibrations and two perpendicular bending motions. The stretch vibrations are located at 2349 cm-1 (anti-symmetric) and 1333 cm-1 (symmetric). The bending modes occur at 667 cm-1. Carbon monoxide shows one stretch vibration at 2143 cm-1. Aldehydes, ketones and carboxylic acids are recognizable by carbonyl stretch vibrations in the range 1760-1690 cm-1. Carboxylic acids show a broad absorption at 3000-2500 cm-1 due to O-H stretch vibrations. At 1300-1080 cm-1 C-O stretch vibrations occur. Alkanols show O-H stretch vibrations at 3640-3610 cm-1 and vibrations of hydrogen bonds at 3600-3200 cm-1. C-O stretch vibrations occur at 1300-1080 cm-1. Nitrogen dioxide has two stretch vibrations viz. at 1618 cm-1 (anti-symmetric) and at 1318 cm-1 (symmetric) and a bending mode at 750 cm-1. Nitrous oxide has two stretch vibrations viz. an XY stretch at 2224 cm-1 and YZ stretch at 1285 cm-1 and two bending modes at 589 cm-1. Ozone has two stretch vibrations viz. at 1103 cm-1 (symmetric) and 1042 cm-1 (antisymmetric) and a bending mode at 701 cm-1. Finally water vapour, produced by application of pulsed corona discharges over the aqueous solution, causes absorptions due to stretch vibrations at 3657 cm-1 (symmetric) and 3756 cm-1 (anti-symmetric) and a bending mode at 1595 cm-1. 3. Experimental setup A description of the applied reagents, experimental configurations and chemical, electrical and optical analysis techniques is presented. 3.1. Reagents & reactors Table 3.1 shows a list of applied model compounds and oxidation products, chromatographic eluents and other reagents. Table 3.1 Applied reagents: name, Chemical Abstract Service registry number, molecular weight, density of liquids, purity and origin. Compound Acetic acid Acetonitrile Acrylic acid Ammonia Argon Atrazine 1,4-Benzoquinone Catechol CCA Dimethyl sulfide DMPO Formic acid Glyoxal Glyoxylic acid Helium Hydrogen peroxide Hydroquinone Hydroxyhydroquinone Maleic acid Malonic acid Methanol Nitrogen Oxalic acid Phenol Phloroglucinol Propionic acid Pyrogallol Resorcinol Succinic acid Toluene Trifluoroacetic acid 1) sg=HPLC supra gradient CAS no. 64-19-7 75-05-8 79-10-7 7664-41-7 7440-37-1 1912-24-9 106-51-4 120-80-9 531-81-7 75-18-3 3317-61-1 64-18-6 107-22-2 298-12-4 7440-59-7 7722-84-1 123-31-9 533-73-3 110-16-7 141-82-2 67-56-1 7727-37-9 144-62-7 108-95-2 6099-90-7 79-09-4 87-66-1 108-46-3 110-15-6 108-88-3 76-05-1 FW (g/mol) 60.05 41.05 72.06 17.03 39.95 215.69 108.10 110.11 190.15 62.13 113.16 46.03 58.04 74.04 4.00 34.01 110.11 126.11 116.07 104.06 32.04 28.01 90.04 94.11 126.11 74.08 126.11 110.11 118.09 92.14 114.02 ρ (g/cm3) 1.05 0.782 1.051 0.91 0.846 1.015 1.22 1.265 1.342 1.11 0.79 0.992 0.865 1.490 Grade(w%)1) 100 sg 99 25 99.999 99.9 98 99+ 99 97 98-100 40 50 99.995 30 99+ 99 99 99 sg 99.9 99+ 99+ >99 >99 99 99+ 99+ 99.7 >98 Manufacturer Merck Biosolve Aldrich Merck Messer Riedel-de Haën Aldrich Aldrich Aldrich PolyScience Aldrich Merck Aldrich Aldrich Hoekloos Merck Aldrich Aldrich Aldrich Aldrich Biosolve Hoekloos Aldrich Aldrich Merck Fluka Aldrich Aldrich Aldrich Merck Merck 38 Chapter 3. Malachite green has been obtained as a gift from the faculty of Chemistry/department of Instrumental Analysis/TUE. No accurate information about the malachite green anion is known, therefore data have not been further specified. It may regard e.g. malachite green carbinol hydrochloride CAS no. [123333-61-9] FW=382.94 g/mol or malachite green oxalate CAS no. [2437-29-8] FW=927.03 g/mol. General remarks All corona experiments have been performed at ambient conditions. The compound solutions have been prepared from deionized water (Millipore, resistivity 10 MΩcm), except for a few exploratory experiments for which tap water has been applied (as indicated). The reactor contents is continuously homogenized by a magnetic stirring bar. The sample volume is 1 ml or less. Due to the different analytical approaches, several reactor types have been applied. Table 3.2 shows the used reactor types and dimensions. Table 3.2 R. no. Applied reactor+anode types, listed by reactor configuration number R. no.; VL=liquid phase volume, Vtot=total reactor contents, ∅=internal diameter, L=length, h=height. Experiment 1a DMPO ESR 1b DMPO oxidation 2a CCA probe 2b CCA oxidation 3a Ozone 3b Phenol /EI-MS 4 5a 5b 5c 6a 6b 6c 7 Calorimetry Conducto-water Conducto-HB Microtox test Phenol Atrazine Phenol/LC-MS Aldehyde screening 8a LIF 8b 9 FTIR 10 TOC 11 Malachite green 12 Dimethyl sulfide configuration Anode1) VL (ml) Cyl+cir2) Beaker Cube Vessel Vessel Cube Reactor Dimensions (mm) ∅=45, h=305 ∅=85, h=122 70x80x100 ∅≤95, h~200 ∅≤95, h~200 57x62x100 Fe Fe Fe Fe Fe Fe 500 249 242 500 500 100 640 500 560 1000 1000 353 Cube3) Beaker 75.5x80x99 ∅=85, h=122 variable 250 598 500 Vessel ∅≤95, h~200 500 1000 Cylinder ∅=45, h=305 1p W 1pFe/W/Pt 1p W 30p Fe 31p Fe 31p Fe 30p Fe 30p Fe 500 570 Vessel Cube Vessel Beaker Bar Cylinder ∅=90, h~100 57x62x100 ∅=70, h=121 ∅=100, h=142 30x80x100 ∅=37, L=310 30p Fe 30p Fe 30p Fe 30p Fe various wire 300 100 250 498 125 gas phase 600 353 500 1000 240 330 Type 30p 30p 30p 30p 30p 30p Vtot (ml) Experimental setup 1) 39 Anode •30,31p Fe The generally applied anode consists of an aluminum plate: diameter 30 mm, thickness 1.5 mm. 30 or 31 steel pins are separated at 5 mm mutual distance; the global pin dimensions are: length=14-15 mm, thickness=0.6 mm, 60 µm tip. •various Different electrode configurations have been tested: single pin/multipin/wire anode in the liquid phase or gas phase at several distances from the liquid/gas interface, the cathode is situated inside/outside the reactor. Details are given in the appropriate section. •wire The applied reactor is a cylindrical gas phase reactor, equipped with a 128 mm effective length wire anode situated in the centre. A cathode wire mesh surrounds the glass tube on the outside. •1p Fe/W/Pt The steel tip equals the generally used anode tips. The tungsten tip has been obtained from a welding rod by grinding and polishing: the material is an alloy of tungsten and a small amount of thorium; global dimensions are: length=15 mm, thickness=1 mm, 30 µm tip. The platinum tip has been obtained from platinum wire; global dimensions: length=15 mm, thickness=0.5 mm, 70-100 µm tip. In all configurations except for reactor configuration 11 and some setups of reactor configuration 4, the cathode is situated directly outside and underneath the reactor glass vessel bottom and therefore is dielectrically separated from the anode. 2) In-situ ESR The reactor volume is about 570 ml; teflon tubing (L~5 m, ∅=4 mm) and viton pump tubing (L~20 cm, ∅~5 mm) have been applied, the circuit volume is about 70 ml. Due to the pulsating pump flow, the anode-to-liquid distance d is somewhat fluctuating. 3) Calorimetry The reactor heat capacity Creactor has been determined from the construction materials (glass and quartz) and the contents (water) [96]. Quartz: (2.3x80x99+2.3x75.5x99)mm3 ≡ 34.6 cm3; cpquartz=0.17 Jg-1K-1; Cquartz=13.0 J/K. Glass: (3.0x80x99+ 3.0x75.5x99 +3.0x85.6x75.6)mm3 ≡ 65.6 cm3; cpglass=0.48 Jg-1K-1; Cglass=78.1 J/K. Water: cpwater=4.184Jg-1K-1; CH2O=4.18⋅volume; therefore Creactor=(91.1+4.18⋅volume) J/K. The heat loss due to imperfect insulation is small, about 0.2 Kelvin per 15 minutes. 40 Chapter 3. 3.2. Chemical diagnostics Liquid chromatography The applied column types, liquid chromatography systems and conditions are summarized by Table 3.3. Table 3.3 LC no. Applied LC setups, listed by the LC configuration number; Col=column no., Sys=system no., F=eluent flow, T=column temperature, con=conductivity, ACN=acetonitrile, TFA=trifluoroacetic acid. Experiment1) 1 DMPO 1.0 mM 2 CCA 1.0 mM 3 Water 1) 2) 3) 4) Col Sys Eluent (v/v%) 2b 1 1 3 H2O/ACN=90/10 3 TFA 1 mM 4 TFA 1 mM 4 Phenol 0.05 mM 5a Phenol 1.0 mM 5b 6 Phenol 1.0 mM air/Ar 3 1b H2O/ACN=80/20 2a 2 H2O/ACN grad3) 12) 1a TFA 1 mM 1 4 TFA 1 mM 7 TOC 1.0 mM phenol 8 CC 1.0 mM HB 9 MT phenol ≤0.4 mM 10 Atrazine 0.12 mM 1 3 1 3 2b 3 3 1b TFA 1 mM TFA 1 mM H2O/ACN=70/30 H2O/ACN=55/45 Settings F Detector λ (nm) T (°C) (ml/min) 1.0 230 20 0.8 210,270 50 0.8 200,210,220, 35 255,270,con 1.0 254, 2804) 20 1.0 210 20 0.8 210 35 0.8 200,210,220, 35 255,270,con 0.8 270 35 0.8 225, 270, 278 35 1.0 270 35 1.0 216 20 TOC=total organic carbon, CC=conductometry & conversion, HB=hydroxybenzenes, MT=Microtox Including OA-HY guard column L=20 mm, ∅=3 mm Grad= gradient: 4 min 0→5%, 3 min 5→20 min, 8 min 20→50%, 1 min 50→75% ACN 0-5 min: 254 nm, 5-10 min: 280 nm Columns: 1 Merck Polyspher OA-HY (ICE) Dimensions: column L=300 mm, ∅=6.5 mm; Stationary phase: Polystyrene-divinylbenzene with -SO3-H+ functional groups, dp=8 µm, crosslink ratio 8%; Column no. 150095 2 Zorbax Rx-C18 (rp-HPLC) Dimensions L=150 mm, ∅=4.6 mm, dp=5µm, Stationary phase: octadecyl grafted silica a Column no. PN883967.902 - DU5023 b Column no. PN883967.902 - DU4041 3 Zorbax SB-C18 (rp-HPLC) Dimensions L= 150 mm, ∅= 4.6 mm, dp=5 µm Stationary phase: octadecyl-grafted silica; Column no. - Experimental setup 41 LC-systems: The injection volume is 20 µl for all analyses. 1a. Philips PU4100 Liquid Chromatograph and Philips PU4110 UV/VIS detector 1b. System 1 including autosampler: Marathon ser.no. 0106, Spark Holland 2. Merck Hitachi L-6200A Liquid Chromatograph and Merck Hitachi L-4250 UV/VIS detector 3. Hewlett-Packard HP1100 series Liquid Chromatograph, HP-G1315A diode array detector, G1311A quaternary pump, G1322A solvent degasser, G1313A autosampler, G1316A column compartment. 4. System 3 in series with a micromembrane suppressor (Dionex AMMS-ICE II) and conductivity detector (Dionex CDM-2). The conductivity detector has been set to output range=300 µS and 1 Volt full scale; software: 106 units per Volt. The suppressor is supplied with a 5 mM aqueous ammonia solution at a flow rate F=2.0 ml/min. Mass spectrometry Configuration 1: Perkin Elmer Sciex API 300 The oxidation product sample is introduced into the IonSpray compartment by means of a liquid chromatograph (Shimadzu LC-10AT); the acetonitrile flow is F=1.0 ml/min; the flow split ratio is 1:10 because the maximum allowable IonSpray flow is 0.2 ml/min. The settings are: NC=-3.5 kV, step=0.3 amu, dwell time=3.0 ms, pause=2.0 ms. The scanned mass-range=40-300 amu. 0.1 v/v% ammonia has been added to the oxidation product mixture to promote the ionizability. Configuration 2: Balzers PPM 421 By means of a PEEK capillary (L=3 m, ∅=0.17 mm), the oxidized solution is introduced into the prevacuum compartment where p<10-2 mbar; this compartment is separated from the mass spectrometer by a 100 µm orifice. The pressure at the detector is p<3⋅10-5 mbar. Ionization is performed by an electron beam of 100 eV. The ions are detected in a count mode. The scanned mass-range is 1-500 amu. Solid phase extraction The following SPE column types have been used: 500 mg C18 and 500 mg C6H5 (Baker). 50 ml of the oxidized phenol solution has been extracted. Desorption has been performed using 0.5 ml methanol. 42 Chapter 3. Gas chromatography Configuration 1: Aldehyde screening The toluene extract from the sorbent tubes has been analyzed by gas chromatographymass spectrometry (GC-MS), using following configurations: Instrument: Column: Detector: Temperature: Shimadzu QP5000 CP-Sil5; L=25 m, dc=220 µm Quadrupole T0=50°C, dT/dt=10°C/min, T1=275°C, F=2 ml/min Instrument: Column: Detector: Temperature: Hewlett-Packard 6890+Leco Pegasus IIA CP-Sil5; L=8 m, dc=50 µm Time of flight T0=60°C, dT/dt=20°C/min, T1=275°C, F=4 ml/min Configuration 2: Conversion of dimethyl sulfide Instrument: Column: Detector: Temperature: Injection: Hewlett-Packard 5890 HP Ultra-1; WCOT MeSil.; L=20m, dc=320 µm, df=0.52 µm, He 100 bar Sulfur Chemiluminescence, T=800°C, pcell=234 mBar Isotherm T=40°C PTV: cold introduction, splitless mode, injection volume=100 µl T0=40°C, dT/dt=12.5°C/s, T1=320 °C, flow 250 ml/min 10 ppm DMS standards have been prepared in methanol, because DMS is insoluble in water. The retention time of DMS for the specified conditions is tR= 0.492 ± 0.009 minutes. Aldehyde screening test The presence of volatile aldehydes during the oxidation of phenol has been investigated using aldehyde-specific gas sampling tubes (Supelco ORBO-23, no. 2-0257-U). The solid sorbent tubes contain 10% 2-(hydroxymethyl)piperidine (HMP) on Supelpak 20N (20/40) carrier divided over two sections: the main section contains 120 mg and the backup section contains 60 mg. The required gas sampling volume is 5 liter. By means of a flow controller (Brooks 5850) and controller unit (Gossen 5875) the reactor has been purged with an argon 5.0 flow at a rate F=200 ml/min, see reactor configuration 7. Figure 3.1 shows the applied setup. After sampling, HMP and possible oxazolidines are extracted from the sorbent tube contents using 1 ml toluene by ultrasonic agitation during 60 minutes. The extract is analyzed by GC-MS. Experimental setup 43 Ar ST FC R Figure 3.1 Setup for sampling of volatile aldehydes from an oxidized phenol solution; ST=sampling tube; Ar=argon 5.0 flow, FC= flow controller, R=corona reactor. Electron spin resonance The identification of oxidizer radicals and phenol oxidation intermediate radicals has been studied by in-situ Electron Spin Resonance (ESR). The applied reactor is constructed from a glass tube (QVF) and two teflon covers with o-rings, see reactor configuration 1a. By means of chemically-inert tubing and a peristaltic pump, a continuous flow system has been created between the corona reactor and an ESR quartz flat cell (flow area: 1x10 mm), see Figure 3.2. The flow rate is about 8 cm3/s (N~2 s-1). The applied spin trap is 5,5-dimethyl-1-pyrroline N-oxide. Measurements have been performed using a Bruker ESP300 ESR spectrometer. The settings are microwave frequency: 9.75 GHz; modulation frequency: 100 kHz, modulation amplitude: 5 G. The centre field has been set to 3355 G/width 100 G and to 3450 G/width 500 G. The temperature is T=295 K. P R Figure 3.2 ESR In-situ ESR setup for the detection of radicals; R=reactor, P=peristaltic pump. 44 Chapter 3. Microtox test For determination of the reference bioluminescence intensity, two Vibrio fischeri solutions have been prepared using freeze-dried bacteria: a commercial Microtox seawater diluent and the Millipore water used to prepare the standard phenol solutions with NaCl added (20g/l). Before the Microtox test is performed, the degree of acidity of every solution has been adjusted to 5.0<pH<5.5 using a diluted NaOH solution. Also the oxygen content of the solutions has been verified. A series of 250 ml phenol solutions at concentrations 0.02 mM, 0.05 mM, 0.1 mM, 0.2 mM and 0.4 mM have been exposed to pulsed corona discharges during 30 minutes, see reactor configuration 5c. To investigate the possible ecotoxicity increase, caused by corona treatment of Millipore water i.e. the formation of hydrogen peroxide and nitric acid, the 0 mM sample has been included in the test. The 0.4 mM concentration is the upper limit for observation of a 20% effect. All samples have been stored away under argon to bridge the time before analysis. The bioluminescence intensity of untreated and oxidized phenol solutions has been determined in duplicate, with regard to the reference bioluminescence intensity. The EC20 effect value has been reported at the Vibrio fischeri exposure times tVF=5 min, t=15 min and t=30 minutes. Total Organic Carbon The carbon content of phenol solutions has been determined versus the oxidation time (reactor configuration 10) using a Dohrmann TC190 analyzer. Prior to analysis the samples have been diluted by a factor 4 with deionized water. The total organic carbon (TOC) content is determined from the difference of total carbon content (TC) and total inorganic carbon content (TIC). The TC content is calculated from the amount of carbon dioxide that is released from catalytic combustion of the sample. The carbon dioxide concentration is measured by infrared spectroscopy. The TIC content is determined from the carbon dioxide release that occurs after acidification of the sample with phosphoric acid. Acidity A pH measuring device (Metrohm 691) has been used to measure the solution acidity of untreated and oxidized phenol solutions. Before the measurements, the instrument is calibrated using calibration buffer solutions (Merck). Experimental setup 45 3.3. Electrical diagnostics Electrical circuit The electrical circuit for generation of pulsed corona discharges is shown by Figure 3.3. By means of a high voltage power supply of negative polarity (Wallis), a capacitor (Thomson CSF-LCC) is charged through a 10 MΩ load resistor (Metallux). By triggering the spark gap, the energy stored in the capacitor is discharged into a glass reactor vessel, by means of a capacitive electrode configuration. In this way, conductive currents are prevented. Positive corona is produced at the anode, situated at a distance d over the aqueous solution. The anode is single-pin or multipin consisting of 30 or 31 steel pins at 5 mm mutual distance. The cathode plate is located outside and underneath the reactor vessel. A 50 MΩ tail resistor (Philips) is installed parallel to the reactor, to rezero the voltage after every discharge pulse. The spark gap is triggered by frequency-adjustable 9 kV pulses. The reactor contents is stirred by a 1 cm magnetic stirring rod. The circuit is settled in welded-alumina EMC compartments. C =1 nF R =10 MΩ L + anode high voltage source triggered sparkgap reactor I-probe Figure 3.3 R =50 MΩ V-probe cathode The high voltage circuit to generate pulsed corona discharges. The pulse voltage is measured at the anode by means of a high voltage probe (Tektronix P6015A: 1000x, 3.0 pF, 100 MΩ; compensation box 015-049). The current is determined at the cathode by means of a coil probe (Pearson 2877: 1 V/A). Data acquisition is performed by a 400 MHz 2 Gs/s digital oscilloscope (Tektronix TDS380). Data signals are obtained in eightfold averaging mode and 20 MHz filtering enabled. The probe cables are shielded with copper non-woven jackets to avoid interception of high frequency noise, emitted by the spark gap. A coil (L) is inserted in series with the capacitor to damp parasitic high frequency oscillations. For reasons of discrete sampling times, the assumption has been made that the pulse energy is constant in between two measurements: Ep(ti-1<t≤ti) = Ep(ti). With regard to the observed small changes in the pulse energy during oxidation runs of several hours, this assumption is justified. 46 Chapter 3. Probes calibration For calibration of the voltage probe, a 20 V 100 kHz square waveform from a function generator (Textronics CFG250) is presented to the probe. The voltage probe calibration factor is determined from a set of successive acquisitions of function generator voltage and probe-indicated voltage. The observed calibration factor is about 909x. The current probe is calibrated by application of the 20 V 100 kHz square waveform across a 1 kΩ resistor. The current probe calibration factor is determined from a set of successive acquisitions of voltage across the resistor and current indicated by the probe in series with the resistor. The observed calibration factor is about 1.0 V/A. The current probe cable is terminated with a 50 Ω impedance to avoid signal reflection; then 1 volt corresponds to 2 A. Due to a difference in probes cable length, a time lag exists between the voltage and current signal. Assuming that electric charge travels through the probe cable with the speed of light, the time lag is about 3.3 ns/m. The time lag correction is very important for accurate pulse energy determination. The actual time lag has been determined from the voltage and current signals observed during the current probe calibration. The observed time lag is about 8 ns. Electrical conductometry The electrical conductivity of corona-exposed deionized water and aqueous hydroxybenzene solutions has been measured using a conductivity meter (Cole Parmer 01481-92) equipped with a gold-plated electrodes dip-cell (CP 01481-93). The cell constant of the used dip cell equals K=10 cm-1. This conductivity meter is temperature compensated: dG/dT=2%/°C. Before every experiment, the device is calibrated using a 445 µS/cm NaCl standard solution (CP 01489-93) at 25°C. 3.4. Optical diagnostics UV absorption spectrometry The ozone production by pulsed corona discharges in air has been measured by UV absorption spectrometry. The applied setup is shown by Figure 3.4, see also reactor configuration 3a. A high-pressure mercury lamp (Philips 3110E) has been used as source. By means of quartz optics, the light beam is transmitted through a quartz wall reactor (57x62x100 mm, optical path=57.6 mm) and then imaged on the slit of a 0.5 meter monochromator (Jarrell Ash 82025), equipped with a 1200 mm-1 grating and photomultiplier (Hamamatsu R636). The absorption at 260 nm is measured. After amplification (Textronics AM502) of the photomultiplier signal, data acquisition is performed by and A/D converter (TSC500). Experimental setup 47 + + R PM M-JA AM A/D Hg Figure 3.4 UV absorbance setup for quantitative ozone measurements. Hg=highpressure Hg lamp, +=positive quartz lens, R=reactor, M-JA= monochromator, PM=photomultiplier, AM=amplifier, A/D=analog-todigital converter. Laser-induced fluorescence A Nd:YAG solid-state pulsed laser (Continuum 9030) has been applied as excitation source for LIF measurements on aqueous phenol solutions. The applied excitation wavelength λ=266 nm is produced from a twofold frequency doubling of the fundamental wavelength λ=1064 nm. The laser pulse width is about 6 ns and the linewidth is 1.0 cm-1. The beam repetition frequency is set to 10 Hz. The measured laser pulse energy is about 0.8-1.1 mJ. The beam waist at the detection volume is about 3 mm. Figure 3.5 shows a schematic drawing of the setup. By means of a Pellin-Broca prism, the excitation wavelength is separated from the other harmonics (355 nm and 532 nm). The applied reactors are equipped with quartz observation windows, see reactor configuration 8. The fluorescence intensity is measured at a 90° angle relative to the laser beam. By means of a quartz lens and quartz fibre the fluorescence radiation is introduced into a monochromator (Jobin-Yvon H25) equipped with a 150 mm-1 grating. Detection is performed by an ICCD camera (Andor ICCD-452/DH534-18) operated in the gated mode: the ICCD is externally triggered by the Nd:YAG laser. The exposure time is 5 seconds, which yields the fluorescence intensity due to 50 laser pulses. The laser triggers the ICCD camera via a pulse delay generator (Stanford DG535). A low-pressure mercury lamp (Cathodeon 93109) has been used for wavelength calibration. D R + Nd:YAG PB ex.tr. QF M-JY Figure 3.5 ICCD PC DG LIF setup for measuring phenol degradation by pulsed corona. Nd:YAG=laser, PB=Pellin-Broca prism, R=reactor, +=positive quartz lens, QF=quartz fibre, M-JY=monochromator, ICCD=detector, DG=delay generator; PC=computer, D=laser dump. 48 Chapter 3. Fluorescent molecular probe The detection of hydroxyl radicals has been performed by means of the molecular probe coumarin-3-carboxylic acid (CCA). 1 mM aqueous solutions have been prepared. Before oxidation, a sample is taken from the corona reactor, see reactor configuration 2a. This non-fluorescent CCA sample is used for background determination. The fluorescence intensity is measured by a Perkin Elmer LS50B fluorescence spectrometer. The excitation wavelength is λex=396 nm. The applied quartz containers (Hellma QS111) have an optical path L=10 mm and the volume is 3500 µl. The scanspeed is 100 nm/min. The excitation and emission slitwidth are 5 nm. For the hydroxyl radical concentration determination, 100 ml standards have been prepared consisting of CCA and unstabilized hydrogen peroxide. In order to have equal initial CCA concentrations, all standards have been prepared from the same 1.0 mM CCA stock solution by a 10-fold dilution. The standard solutions contain 0.1 mM CCA and increasing amounts of hydrogen peroxide i.e. 4.7⋅10-5 M, 6.6⋅10-5 M, 7.9⋅10-5 M, 2.0⋅10-4 M, 3.6⋅10-4 M and 9.7⋅10-4 M. Infrared spectroscopy The gas phase over oxidized phenol solutions has been analyzed by Fourier-transform infrared spectroscopy (FTIR, Bruker IFS 66). A DTGS detector has been applied. The scanned wavelength range is 3000-1500 cm-1. The resolution is 4 cm-1. The volume of the applied reactor is 500 ml and the optical path is 120 mm, see reactor configuration 9; the reactor is equipped with sapphire windows. The spectrometer is purged with nitrogen at a rate F=500 l/h to exclude carbon dioxide and water vapour. After corona-exposure, the sealed reactor is immediately transferred from the high voltage setup to the spectrometer. Spectra have been recorded 15 minutes after installation of the reactor into the spectrometer, in order to recover a carbon dioxide-free measurement compartment. Background spectra have been recorded from the empty reactor, the reactor filled with the phenol solution at t=0 and the empty spectrometer compartment in between the measurements. 4. Results In this chapter the results will be presented concerning the production of oxidizers by pulsed corona discharges in humid air. A detailed analysis of the oxidation of the model compound phenol is given regarding conversion efficiency, oxidation products and analysis techniques. In addition, the oxidation of the model compounds atrazine, malachite green and dimethyl sulfide is described. 4.1. Pulsed corona discharges This section describes the results of experiments on the effects of pulsed corona discharges in air over deionized water. The production of hydroxyl radicals and ozone is described, reactor geometrical capacitances and pulse energies are reported and the analysis of corona-exposed deionized water is discussed. 4.1.1. Hydroxyl radicals In order to detect hydroxyl radicals in corona-treated water two approaches have been applied. The first approach is trapping of the hydroxyl radical by the spin trap DMPO, followed by in-situ ESR detection of the produced DMPO-OH adduct. The second approach is the reaction of the hydroxyl radical with the fluorescent molecular probe coumarin-3-carboxylic acid (CCA), followed by ex-situ fluorescence spectrometry on produced 7-hydroxy CCA (7OHCCA). The spin trap approach A 500 ml 1.0 mM aqueous DMPO solution has been exposed to pulsed corona discharges in a continuous flow system, see reactor configuration 1a. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. During the oxidation for 70 minutes, several ESR scans have been made in order to detect the DMPO-OH adduct. No ESR-absorption signals have been observed. Hereafter, an extra 2.5 mmol DMPO amount has been added to the continuous flow system in order to be sure, that enough fresh spin trap is available for the trapping of radicals. Also this extra addition has not resulted in any absorption signals after 50 minutes of continuation. A second experiment has been performed, in which a solution containing 1.0 mM phenol and 1.0 mM DMPO spin trap has been oxidized by pulsed corona discharges. Now the spin trap is intended to trap radical intermediates produced by the attack of hydroxyl radicals on phenol i.e. dihydroxycyclohexadienyl(peroxyl) radicals. These organic radicals are more stable in aqueous solution than hydroxyl radicals. Again, no ESR absorption signals have been detected within 75 minutes of corona-exposure time. The possibility may exist that DMPO is destroyed by the pulsed corona discharges. This has been verified by application of liquid chromatography to untreated and oxidized DMPO solutions. A reversed-phase HPLC column and UV absorbance detector have been used according to LC configuration 1. Figure 4.1 shows the conversion of DMPO as a function of time. 50 Chapter 4. Conversion (%) 40 30 20 10 0 0 Figure 4.1 15 30 45 60 Time (min) 75 90 The conversion of a 250 ml 1 mM DMPO solution by pulsed corona discharges. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The observed conversion course will mainly imply the degradation of DMPO by oxidizers, but the production of the DMPO-OH adduct will also be included. Figure 4.2 shows a possible degradation pathway of DMPO by ozone. - H2O O3 N O H DMPO Figure 4.2 H O N O O O molozonide O N O H O O ozonide + N O H2O C O H O O zwitterion H NO2 OH OOH hydroperoxy alkanol - H2O2 NO2 NO2 O OH O H Possible degradation pathway of the spin trap DMPO by ozone. Degradation of DMPO by ozone causes ring-cleavage of the 5-membered ring and 4nitro-4-methyl valeric acid or -valeraldehyde may be produced. Also singlet oxygen in known to degrade DMPO [97]. Identification of oxidation products has not been performed. Although conversion increases significantly with the exposure time, within short times the conversion is still low (X10min<5%) and enough spin trap is available to trap hydroxyl radicals. In addition, the half-life of the DMPO-OH adduct is reported to be about 2 hours [75]. Results 51 The fluorescent molecular probe approach Intensity (arb.u.) A 242 ml 1.0 mM CCA solution has been exposed to pulsed corona discharges for 40 minutes, see reactor configuration 2a. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The fluorescence intensity of the non-treated and oxidized solutions as a function of time is illustrated by Figure 4.3. 10 9 8 7 6 5 4 3 2 1 0 400 Figure 4.3 40 min 30 min 20 min 10 min 0 min 450 500 550 Wavelength (nm) 600 Fluorescence spectra of a 1 mM CCA solution after different coronaexposure times (0-40 min). The excitation wavelength λex=396 nm. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Intensity (arb.u.) The fluorescence from non-treated CCA is marked by t=0 min and is very weak compared to the fluorescence from the reaction product of CCA and the hydroxyl radical: 7-OHCCA. The fluorescence intensity of the oxidized solutions with regard to the non-treated CCA solution (background) is illustrated by Figure 4.4. 10 9 8 7 6 5 4 3 2 1 0 400 Figure 4.4 40 min 30 min 20 min 10 min 450 500 550 Wavelength (nm) 600 Fluorescence spectra of a 1 mM CCA solution after different coronaexposure times (10-40 min). Subtracted background is the untreated CCA solution. The excitation wavelength λex=396 nm. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. 52 Chapter 4. The increase of the fluorescence intensity with the exposure time is almost linear. This implies that the production rate of the hydroxyl radicals is constant. This experiment has been repeated applying a longer exposure time, see Figure 4.5. Intensity (arb.u.) 25 90 min 20 15 60 min 45 min 10 30 min 5 15 min 0 400 450 500 550 600 Wavelength (nm) Figure 4.5 Fluorescence spectra of a 1 mM CCA solution after different coronaexposure times (15-90 min). Subtracted background is the untreated CCA solution. The excitation wavelength λex=396 nm. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Here, the increase of the fluorescence intensity is still reasonably proportional to the exposure time, although it may be expected, that CCA will be degraded after longer exposure times. The stability of CCA towards pulsed corona discharges has been determined by ion-exclusion chromatography using the LC configuration no. 2b. A 500 ml 1.0 mM CCA solution has been oxidized by pulsed corona discharges for 3 hours. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The conversion of CCA versus the oxidation time is shown by Figure 4.6. Conversion (%) 25 210 nm 20 270 nm 15 10 5 0 0 30 60 90 120 150 180 Time (min) Figure 4.6 The conversion of a 500 ml 1 mM CCA solution by pulsed corona discharges. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Results 53 CCA appears to be fairly stable towards pulsed corona discharges. After 3 hours only 20% conversion occurs. Possible degradation of CCA may involve multifold hydroxylation and ring-cleavage of the benzene ring, hydroxylation of or ozone attack on the double bond present in the pyran ring followed by ring-cleavage, see Figure 4.7. Identification of oxidation products has not been applied. O a) O O O OH O OH CCA O HO O b) OH O O d) O O O O OH O Figure 4.7 O O OH O O O O c) O OH Possible degradation of CCA: a) hydroxylation of the benzene ring, b) ringcleavage of the benzene ring by ozone, c) hydroxylation of the pyran ring double bond, d) ring-cleavage of the pyran ring by ozone. Attempts have been made to quantify the production of hydroxyl radicals. The fluorescence intensity of 7-OHCCA has been calibrated versus the amount of hydroxyl radicals produced per solution volume. Fluorescence standard solutions have been prepared from CCA and hydrogen peroxide, where hydrogen peroxide has been used as source for hydroxyl radicals. With regard to the quantitative approach, unstabilized hydrogen peroxide has been utilized, to avoid scavenging of hydroxyl radicals by the stabilizer normally present in commercial hydrogen peroxide solutions. Different amounts of hydrogen peroxide i.e. known amounts of hydroxyl radicals have been added to 100 ml 0.1 mM CCA solutions. A pure 0.1 mM CCA solution has been exposed to corona discharges. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The fluorescence intensity of both the oxidized CCA solution and the standards is shown by Figure 4.8. Intensity (arb.u.) 5 30 min 20 min 10 min 0 min 9.7E-4 M 3.6E-4 M 2.0E-4 M 7.9E-5 M 6.6E-5 M 4.7E-5 M 4 3 2 1 0 400 450 500 550 OH OH OH OH OH OH 600 Wavelength (nm) Figure 4.8 Fluorescence spectra of an 0.1 mM CCA solution after different coronaexposure times (0-40 min) The corona parameters are V=25 kV, f=100 Hz, C=1 nF, d=1.0 cm. Also shown are the fluorescence spectra of the CCA/H2O2 fluorescence standards, indicated by a virtual hydroxyl radical concentration. The excitation wavelength λex=396 nm. 54 Chapter 4. Unfortunately, the fluorescence standards appear to show no correlation between the fluorescence intensity and the amounts of hydroxyl radicals added (as hydrogen peroxide). Also, the fluorescence intensity of the untreated 0.1 mM CCA solution i.e. background is even higher than the intensity of several fluorescence standards. These contradictory results are explained by the fact that the preparation of the fluorescence standards is critical. The addition of hydrogen peroxide to the CCA solution should be in stoichiometric proportions, here equal molarities. However, the solubility of CCA in water is poor: higher concentrations than 1 mM are not feasible. This implies, that hydrogen peroxide should be added in very diluted form to the CCA solution. Very dilute hydrogen peroxide solutions are not stable, because hydrogen peroxide will oxidize the water, according to the redox reactions shown by Equations 4.1a-c. The addition of hydrogen peroxide to the 0.1 mM CCA solutions is thus unverifiable. H2O2 + 2e- → 2 OHH2O2 + 2H+ + 2e- → 2 H2O 2H2O → O2 + 4H+ + 4e- reduction of hydrogen peroxide reduction of hydrogen peroxide oxidation of water (4.1a) (4.1b) (4.1c) Summary The formation of hydroxyl radicals during exposure of water by pulsed corona discharges has been demonstrated by fluorescence spectroscopy using the OH-specific fluorescent molecular probe CCA. The production of hydroxyl radicals appears to be rather constant during 90 minutes of corona exposure. It has not been possible to attribute the observed fluorescence intensity to certain amounts of hydroxyl radicals. Insitu Electron Spin Resonance using the spin trap DMPO has not been able to identify the hydroxyl radical under the same conditions. Results 55 4.1.2. Ozone The production of ozone by pulsed corona discharges in air has been determined for several reactor conditions using UV absorbance spectrometry, see reactor configuration 3a. First, ozone measurements have been performed for corona discharges in an ambient air-filled reactor at different load voltages viz. -25 kV, -20 kV, 15 kV, 20 kV, 25 kV and 30 kV. The ozone concentration versus time is shown by Figure 4.9. It has been observed, that the maximum ozone concentration increases with increasing absolute load voltage. At a given voltage, the negative polarity corona produces less ozone than the positive polarity corona. After about 16 minutes the corona has been switched off: a decrease in ozone concentration follows as a result of the backward reaction of ozone into oxygen by wall recombination. -3 [Ozone] (m ) 6.0E+22 30 kV 25 kV 4.0E+22 -25 kV 20 kV 2.0E+22 -20 kV 15 kV 0.0E+00 0 5 10 15 20 Time (min) Figure 4.9 Ozone concentration versus time in a reactor filled with ambient air, energized at different load voltages. The corona parameters are V (indicated), C=1 nF, f=100 Hz, d=1.0 cm. After about 16 min the corona discharges have been stopped. Hereafter, ozone concentrations have been measured for the case of a reactor filled with ambient air and 100 ml deionized water. The load voltages have been 20 kV, 25 kV and 30 kV, see Figure 4.10. The ozone concentration over the deionized water increases with the load voltage just as for the case of the ambient air-filled reactor, but the maximum ozone concentration over water is lower than the ozone concentration in air. Also it has been observed, that the initial ozone production rate i.e. initial curve slope is higher for the water-filled reactor compared to the ambient air-filled reactor. The difference in maximum ozone concentration can be explained by the fact that ozone is destroyed by hydroxyl and hydroperoxyl radicals produced in humid air over water, according to Equations 2.4ab. Also ozone reacts with water, see Equations 2.3a-c. The water-filled reactor has a smaller gas phase volume (253 ml) than the ambient air-filled reactor (353 ml), therefore initially the ozone production rate is higher. 56 Chapter 4. 6.0E+22 30 kV air -3 [Ozone] (m ) 30 kV water 4.0E+22 25 kV air 25 kV water 20 kV air 2.0E+22 20 kV water 0.0E+00 0 5 10 15 Time (min) Figure 4.10 Ozone concentration versus time in a reactor filled with ambient air and 100 ml deionized water or solely ambient air, energized at different load voltages. The corona parameters are V (indicated), C=1 nF, f=100 Hz, d=1.0 cm. Next, the influence of phenol in water on the gas phase ozone concentration has been measured. 100 ml 1 mM phenol solutions have been oxidized by corona in ambient air at voltages 25 kV and 30 kV. The ozone concentration versus time for deionized water and phenol solutions is shown by Figure 4.11. The presence of phenol in the deionized water additionally decreases the ozone concentration, so it is clear that phenol in aqueous solution consumes ozone produced over the aqueous solution. The reaction products of phenol oxidation will be discussed in sections 4.2 and 5.3. -3 [Ozone] (m ) 6.0E+22 30 kV water 30 kV 1 mM phenol 4.0E+22 25 kV water 2.0E+22 25 kV 1 mM phenol 0.0E+00 0 5 10 15 Time (min) Figure 4.11 Ozone concentration versus time in a reactor filled with ambient air and 100 ml deionized water or 100 ml 1 mM phenol solution, energized at different load voltages. The corona parameters are V (indicated), C=1 nF, f=100 Hz, d=1.0 cm. Results 57 Application of pulsed corona discharges in pure oxygen may produce higher ozone levels. This has been verified by purging the reactor with oxygen at the flow rates F=100 ml/min and F=200 ml/min. The load voltages are 20 kV, 25 kV and 30 kV, see Figure 4.12. With increasing load voltage and constant oxygen flow rate the ozone concentration increases. Increasing the oxygen flow rate appears to have a minor positive effect on the ozone production at 20 kV, but has a strongly negative effect on the ozone production at the load voltages 25 kV and 30 kV. This is explained by the fact, that at higher flow rates the produced ozone is removed from the reactor by the oxygen purge. This effect can also be illustrated by comparison of the ozone production at V=30 kV & F=200 ml/min and the ozone production at V=25 kV in an oxygen saturated reactor (F=0 ml/min). The positive effect of the higher voltage is overruled by the negative effect of a high flow rate. 1.0E+23 30 kV 100 ml/min 25 kV 0 ml/min -3 [Ozone] (m ) 8.0E+22 6.0E+22 30 kV 200 ml/min 25 kV 100 ml/min 4.0E+22 25 kV 200 ml/min 2.0E+22 20 kV 200 ml/min 20 kV 100 ml/min 0.0E+00 0 Figure 4.12 2 4 6 Time (min) 8 10 Ozone concentration versus time in a reactor purged with oxygen at different flow rates, energized at different load voltages. The corona parameters are V (indicated), C=1 nF, f=100 Hz, d=1.0 cm. Finally the ozone production has been monitored versus the integrated corona pulse energy for the reactor filled with ambient air and 100 ml deionized water. The load voltages are 15 kV, 20 kV, 25 kV and 30 kV, see Figure 4.13. From the initial slope of the ozone concentration versus corona energy curves, the following ozone production efficiency has been calculated: 5.45⋅1020 ±0.05 m-3J-1. This value equals about 40 g/kWh for the applied reactor and is quite high, because this efficiency has been obtained in humid air. Nowadays commercial ozone generators reach efficiencies of 50-60 g/kWh in dry air [98]. 58 Chapter 4. -3 [Ozone] (m ) 6.0E+22 30 kV 4.0E+22 25 kV 2.0E+22 20 kV 15 kV 0.0E+00 0 50 100 150 200 E (J) Figure 4.13 Ozone concentration versus the integrated corona pulse energy in a reactor filled with ambient air and 100 ml deionized water, energized at different load voltages. The corona parameters are V (indicated), C=1 nF, f=100 Hz, d=1.0 cm. Summary The production of ozone by pulsed corona discharges in air and oxygen has been demonstrated. The ozone concentration increases with the corona load voltage. Negative corona produces less ozone than positive corona at equal absolute load voltage. Ozone produced by corona over water is destroyed by hydroxyl and hydroperoxyl radicals, which are also formed by the corona in humid air. Ozone produced in the gas phase over an aqueous phenol solution is consumed by phenol. Pulsed corona discharges in oxygen produce higher ozone concentrations than discharges in air, but high oxygen purge flow rates remove ozone from the reactor. The achieved ozone production efficiency for humid air is about 40 g/kWh. Results 59 4.1.3. Corona pulse energy In order to calculate the pulse energy of corona discharges applied in the gas phase, first the capacitance of the applied setup has been determined. The several configurations mainly differ by liquid/gas phase volume and anode type. Figure 4.14 shows a selection of best fits of I(t) and Cg⋅dV(t)/dt for estimation of the geometry capacitances. Table 4.1 shows the measured average geometry capacitances for all applied setup configurations. Table 4.1 Reactor no. 6ab 6c 6c 3b 3b 5ab Average geometry capacitances (Cg) and 95% confidence intervals. Anode (pins) 31 30 30 30 30 1 liquid vol (ml) 500 500 500 100 100 250 Reactor liquid and gas type tap water, air deionized water, air phenol 1 mM, air phenol 1 mM, air phenol 1 mM, Ar purge deionized water, air Cg (pF) 4.3 ±0.1 2.5 ±0.1 2.6 ±0.1 2.3 ±0.2 2.4 ±0.0 1.4 ±0.1 All configurations have the anode situated d=1.0 cm over the aqueous solution, while the cathode plate is situated outside and directly underneath the glass reactor vessel. Although the determined capacitances are rather small, the capacitive current for these configurations is not negligible, because the slope of the voltage is very steep, that is dV(t)/dt≈1011-1012 V/s. The observed oscillations are due to parasitic impedances from the circuit in combination with fast voltage rise times created by the triggered spark gap. The presence of phenol in the water does not influence the capacitance, see reactor configuration 6c; the relative permittivity of phenol is about εR,phenol≈12.4 while water has a relative permittivity εR,H2O≈80.1. In addition, the applied molar fraction of phenol to water is negligible viz. 10-3 M:55.4 M=1.82⋅10-5. With regard to the oxidation of aqueous phenol solutions, neither formic acid (εR≈51.1) nor hydrogen peroxide (εR≈74.6) will contribute to the permittivity of water, because of their low concentration. Data have been obtained from [99]. From the equal capacitances of reactor configuration 3b with air or argon, it can be derived that the capacitance is not affected by vigorously purging of the reactor. Although the relative permittivities of argon and air are nearly equal, the argon purge at a flow rate F=100 ml/min. results in a gas-dispersed phenol solution that may have a different capacitance than the immobile phenol solution in the air-filled reactor. Finally, the capacitances cannot be related to the indicated liquid phase volumes, because the reactors have different dimensions. From the measured capacitances, the capacitive current has been calculated that is part of the total current recorded during the corona experiments. The pulse energy has been estimated from the pulse voltage and corona current. Table 4.2 shows pulse energies, averaged over the different observation times, during the applied pulsed corona-induced oxidation experiments discussed in chapter 4. Characteristic voltage and current waveforms are presented together with the efficiency values in the appropriate sections. 60 Chapter 4. 0.8 0.20 Current (A) 0.4 Cg=4.2 pF Vload=9 kV 0.2 Config. 5ab I(t) Cg⋅dV(t)/dt 0.15 Current (A) I(t) Cg⋅dV(t)/dt 0.6 Config. 6ab 0.10 Cg=1.6 pF Vload=7 kV 0.05 0.0 0.00 -200 -0.2 0 200 400 -200 -0.05 0 Time (ns) Time (ns) Config. 6c water Cg=2.5 pF Vload=7 kV 0.1 0.0 -200 0 200 Config. 6c I(t) Cg⋅dV(t)/dt 1 mM phenol 0.2 Current (A) I(t) Cg⋅dV(t)/dt 0.2 Current (A) 400 0.3 0.3 Cg=2.5 pF Vload=7 kV 0.1 0.0 -200 400 0 200 400 -0.1 -0.1 Time (ns) Time (ns) 0.40 0.40 I(t) Cg⋅dV(t)/dt 0.20 Config. 3b air Cg=2.3 pF Vload=7 kV 0.10 0.30 Current (A) 0.30 Current (A) 200 I(t) Cg⋅dV(t)/dt 0.20 Config. 3b Ar purge Cg=2.4 pF Vload=7 kV 0.10 0.00 0.00 -200 -0.10 0 200 400 -200 -0.10 Time (ns) 0 200 400 Time (ns) Figure 4.14 A selection of best fits of I(t) and Cg⋅dV(t)/dt; the result is Cg. Indicated are reactor configuration numbers. Table 4.2 Time-averaged pulse energies and 95% confidence intervals for the performed experiments. reactor Experiment no. corona in gas phase over aqueous solution 3a Ozone measurements in a reactor, filled with air and 100 ml deionized water 6a 6b 6c 3b 3b 5a 5b Phenol oxidation, 500 ml 0.05 mM; air Atrazine oxidation, 500 ml 0.12 mM; air Phenol oxidation, 500 ml 1 mM; air Phenol oxidation, 100 ml 1 mM; air Phenol oxidation, 100 ml 1 mM; argon flow Deionized water oxidation, 250 ml; air Hydroxybenzenes oxidation, 250 ml 1 mM; air Corona parameters V,C,f 15 kV, 1 nF, 100 Hz 20 kV, 1 nF, 100 Hz 25 kV, 1 nF, 100 Hz 30 kV, 1 nF, 100 Hz 30 kV, 100 pF, 50 Hz 30 kV, 100 pF, 50 Hz 25 kV, 1 nF, 100 Hz 25 kV, 1 nF, 100 Hz 25 kV, 1 nF, 100 Hz 25 kV, 1 nF, 100 Hz 25 kV, 1 nF, 100 Hz Ep (mJ) 0.6 ±0.2 1.4 ±0.4 3.5 ±0.3 5.9 ±1.1 13.4 ±1.4 10.7 ±0.5 10.0 ±0.1 5.8 ±0.2 10.6 ±0.3 4.3 ±0.2 5.6 ±0.3 Results 61 Although this thesis deals with the application of pulsed corona discharges in the gas phase over aqueous solutions of the target compound, also some pulse energy calculations are shown related to the application of corona in the liquid phase. The reason for this approach is to compare pulse energies of both applications. Much environmental corona research is performed as liquid phase corona [100-102]. Both a deionized water and tap water liquid phase have been regarded. Reactor configuration 4 has been used. The anode is made from a W/Th alloy (welding rod) and has a 30 µm tip. Both cathode plate and anode tip are situated in the liquid phase at a mutual distance ∆=2.0 cm. It has not been possible to calculate the pulse energy according to the procedure described by section 2.5.2, because the capacitive current is high as a result of the high capacitance of the water-filled reactor with immersed electrodes. Therefore the pulse energy has been estimated by two different methods viz. a calorimetric determination and a calculation based on the average direct current delivered by the high voltage power supply. At a voltage V=19 kV, a pulse repetition rate f=100 Hz and electrode distance ∆=2.0 cm the pulse energy has been calculated for the reactor, filled with different amounts of deionized water i.e. Vol=200 ml, 250 ml, 300 ml, 400 ml. Table 4.3 shows the energy per pulse calculated by the two methods. Table 4.3 Volume (ml) 200 250 300 400 Pulse energy measured in deionized water at different volumes. The corona parameters are V=19 kV, C=1 nF, f=100 Hz, ∆=2.0 cm. Pulse energy (mJ) Calorimetry Power supply 61 ±9 90 ±8 56 ±8 88 ±7 55 ±8 85 ±7 58 ±9 87 ±7 With regard to both methods the following remarks can be made. A possible error made by the calorimetric method is due to imperfect thermal insulation. However, the observed temperature decrease rate after stopping the experiment is only about 0.2 K per 15 minutes, while the duration of the experiment is comparable, viz. 10-20 minutes. The error involved in the pulse energy obtained from the power source average direct current is due to the assumption, that power dissipation only takes place in the external circuit by the 10 MΩ load resistor. Calorimetric- and average direct current-based measurements on 250 ml tap water under equal corona conditions have yielded pulse energies of 130 ±30 mJ and 260 ±30 mJ respectively. The pulse energy in tap water is much higher than the pulse energy in deionized water, because of the higher electrical conductivity of tap water. The applied deionized water exhibits a conductivity of about 30-100 µS/cm, while the used tap water has a conductivity of about 2580 µS/cm. The reported pulse energies for deionized water and tap water have been determined at a voltage V=19 kV and are yet much higher than the pulse energies in air, determined at V=25-30 kV. 62 Chapter 4. The formation of corona discharges in water requires evaporation of water at the anode tip, thus a high pulse energy. Figure 4.15 shows a CCD image of corona discharges at an anode tip immersed in deionized water. Stroboscopic illumination has been applied using a He/Ne laser chopped at 10 Hz. The discharges appear as bright irregular-shaped channels. Vapour bubbles appear as rows of dots. The size of the vapour bubbles is approximately several tenths of a millimeter. Figure 4.15 CCD image of corona discharges at an anode tip (top) immersed in deionized water. Stroboscopic illumination has been applied. The discharges appear as bright irregular-shaped channels. The vapour bubbles appear as rows of dots. The corona settings are V=20 kV, f=0.1 Hz, d=2.6 cm. The exposure time is 1 s. The image has been taken by A.H.F.M. Baede. The reason for the application of corona in the liquid phase is to produce the oxidizers i.e. hydroxyl radicals directly at the location where they are needed to avoid loss due to recombination. However, in section 4.1.1 it has been shown that by application of pulsed corona discharges over water, yet the action of hydroxyl radicals in aqueous solution can be demonstrated, by means of the hydroxyl radical specific molecular probe CCA. The hydroxyl radicals are produced directly by dissociation of water molecules and indirectly from ozone, water and UV photons. Summary Two different methods have been applied to estimate the pulse energy for application of corona in water viz. a calorimetric determination and a method based on the average direct current from the power supply. By comparison of the pulse energies of corona discharges applied over and in water, it has been shown that the application of corona over water is much more favourable. This is caused by the fact that the production of corona in water requires evaporation of water at the anode tip. The pulse energy measured for corona in tap water is higher than the pulse energy for corona in deionized water, due to a higher electrical conductivity of tap water compared to deionized water. Results 63 4.1.4. Corona treatment of deionized water The application of pulsed corona discharges in an air gas phase over deionized water changes the water in several ways. Corona discharges in air produce ozone and small amounts of nitrogen oxides. In water, the ozone is converted into hydrogen peroxide while the nitrogen oxides are converted into nitric acid. The corona discharges strike the water surface and dissociate water molecules into hydroxyl radicals and hydrogen atoms. Hydroxyl radicals recombine to hydrogen peroxide. Hydrogen atoms react with dissolved oxygen to hydroperoxyl radicals, which recombine to yield hydrogen peroxide and oxygen. Recombination of hydrogen atoms produces hydrogen. Also, the metal of the anode tip may be sputtered due to the high electric field strength, so elementary metal or metal oxides may be found in the water. The chemical change of coronaexposed deionized water has been investigated by application of electrical conductometry, spectrochemical ICP analysis and ion-exclusion chromatography. 250 ml deionized water samples have been exposed to pulsed corona discharges, see reactor configuration 5a. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Three different anode tips have been tested: steel, tungsten and platinum. The following relative temperature increase (∆T/T0) values have been measured after 60 minutes of corona exposure: Fe: 0.14, W: 0.15, Pt: 0.08. Figure 4.16 shows the electrical conductivity of the deionized water as a function of the corona-exposure time. Conductivity (µS/cm) 500 Fe 400 W 300 Pt 200 100 0 0 Figure 4.16 20 40 Time (minutes) 60 The electrical conductivity of 250 ml deionized water samples versus the oxidation time. Used anode tip materials are steel, tungsten and platinum. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Measured pulse energies are: EpFe=4.5 ±0.3 mJ, EpW=4.1 ±0.2 mJ, EpPt=4.2 ±0.3 mJ. The conductivity of the deionized water samples increases with the corona-exposure time. The increase is highest when corona is produced using the steel anode tip, while the platinum tip brings about the lowest conductivity increase. The conductivity differences using different anode tip materials may imply the effect of anode material thus the sputtering of anode material into the water. However, the anode tip geometries are far from identical, because this is very difficult to achieve. Therefore the amounts of sputtered material may be different. Spectrochemical ICP analysis has been applied to the deionized water samples oxidized by the steel and tungsten tips, but no metals have been identified. Therefore it is likely, that the conductivity increase arises from nitric acid. The mutually different anode tips may have produced different amounts of nitrogen oxides versus the oxidation time. 64 Chapter 4. The 30 minutes corona-exposed deionized water samples have been analyzed by ionexclusion chromatography using a diode array UV absorbance detector and conductivity detector in series, see LC configuration 3. Figure 4.17 shows the chromatograms of oxidized and untreated deionized water obtained by using steel, tungsten and platinum anode tips. 4.40E+05 Conductivity (arb.u.) UV absorbance (mAU) 100 80 60 Fe: 3.06 min W: 3.05 min Pt: 3.06 min 40 20 Fe: 8.72 min W: 8.74 min Pt: 8.75 min 4.35E+05 Fe W Pt 4.30E+05 4.25E+05 4.20E+05 0 0 5 10 Retention time (min) Figure 4.17 15 20 0 5 10 15 Retention time (min) 20 UV absorbance (λ=210 nm, left) and conductivity (right) chromatograms of 250 ml 30 minutes exposed deionized water samples. Used anode tip materials are steel, tungsten and platinum. The parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The UV absorbance chromatogram shows three similar strong absorptions at about tR=3.06 minutes. The UV absorbance intensity decreases according to the order 200 nm>210 nm>220 nm>>255 nm>270 nm. This first peak represents excluded anions and is only present in corona-exposed aqueous solutions. Nitrate ions are likely to account for this peak. Nitrite is less probable, because it is oxidized to nitrate by the pulsed corona discharges. The conductivity chromatograms reveal a very strong negative peak at about tR=3.5 min and a weak conductivity signal at tR=8.7 minutes, for all solutions. The strong negative peak represents the elution of water from the sample. This peak overrules a possible conductivity signal due to the components eluting at tR=3.06 min, considering the time difference of about 0.24 minutes between the signals from the UV absorbance and conductivity detector. The components eluting at tR=8.7 min are retained by the ICE column, thus their identity equals either an organic molecule or a cation. However, the fresh deionized water is, except for traces, free from any organic or inorganic compounds. Metal ions are likely to be retarded by the ICE mechanism but have not been identified by spectrochemical ICP analysis. Summary By exposure of deionized water to pulsed corona discharges, the electrical conductivity significantly increases. This is likely due to nitrate ions, originating from nitrogen oxides, which are produced by corona in air. Corona-induced anode metal sputtering is theoretically possible, but metals have not been identified. Results 65 4.2. Oxidation of phenol This section gives a detailed analysis of the oxidation of the model compound phenol regarding conversion, energy efficiency, oxidation products and analysis techniques. 4.2.1. Chromatography Reversed-phase HPLC 100 100 80 80 Conversion (%) Conversion (%) The initial measurements of phenol conversion have been performed using a standard reversed-phase HPLC column according to LC configuration 4. This column is suitable for the separation of phenol from its oxidation product components e.g. polyhydroxybenzenes and carboxylic acids. Phenol conversion is calculated from the UV absorbance detector peak area as a function of time. 500 ml 5 mg/l (0.05 mM) phenol solutions have been prepared using tap water. The pulsed corona discharges take place in air over the phenol solution, see reactor configuration 6a. The influence of the parameters voltage (V), pulse repetition rate (f), anode-tip-to-water distance (d) and solution acidity (pH) on the phenol conversion are shown by Figure 4.18. During these experiments, only one parameter is varied at a time, while the others are kept constant at the standard values V=30 kV, f=50 Hz, d=1.0 cm, pH=n.a. (not adjusted), t=30 min (except V series: t=45 min). The plots show individual trends. The absolute values cannot be compared mutually, because singular data acquisition has been applied. 60 t=45 min f=50 Hz d=1.0 cm pH=n.a. 40 20 t=30 min V=30 kV d=1.0 cm pH=n.a. 40 20 0 0 15 20 25 30 Voltage (kV) 35 0 40 100 100 80 80 Conversion (%) Conversion (%) 60 60 t=30 min V=30 kV f=50 Hz pH=n.a. 40 20 0 50 100 150 Frequency (Hz) 200 250 60 t=30 min V=30 kV f=50 Hz d=1.0 cm 40 20 0 0.0 Figure 4.18 0.5 1.0 1.5 Distance (cm) 2.0 2.5 2 4 6 pH (-) 8 The conversion of phenol as a function of the indicated parameters. 500 ml 5 mg/l (0.05 mM) phenol solutions have been oxidized. 10 66 Chapter 4. The conversion of phenol appears to increase non-linear with the applied corona load voltage. This is likely due to the radical formation processes by the corona plasma. By increasing the voltage the radical production indeed grows, but the probability of recombination also increases. Therefore the radical production at higher voltages is likely to be less efficient. The conversion increases with the pulse repetition rate for the range 0 Hz<f<100 Hz. The unexpected decrease of the conversion at 200 Hz might be related to corona instability at higher repetition frequencies as a result of the application of a pressurized triggered spark gap. Conversion is considerably higher in an alkaline solution than in an acidic solution. In alkaline solution ozone reacts by hydroxyl radicals which are far more reactive than ozone. In addition, at high pH the existence of the phenolate anion (C6H5O-) is extra favourable with regard to the electrophilic nature of hydroxyl radicals [7]. Adjusting the anode-tip-to-water distance at about d=1.0 cm yields the highest conversion. Application of pulsed corona discharges at the gas-liquid interface limits the production of oxidizers e.g. ozone, therefore a certain gas phase volume i.e. distance is favourable. This effect has also been observed by experiments on the influence of electrode configurations on the decolorization of malachite green dye, which will be described in section 4.3.2. The application of pulsed corona discharges in an aqueous solution is energetically unfavourable, because this causes evaporation of water at the anode tip [103]. In addition to this experiment, the influence of the solution volume on the conversion of phenol has been determined in threefold, according to the parameters V=30 kV, C=100 pF, f=50 Hz, d=1.0 cm, t=30 min. The conversion of the 500 ml 5 mg/l solution is 60.7% ±7.0% while the conversion of the 250 ml 10 mg/l solution is 73.5% ±4.0%. The oxidizers are more efficiently consumed for the case of the 250 ml solution, compared to the 500 ml solution. Application of pulsed corona discharges to thin liquid films will be favourable to conversion. In the subsequent experiments, pulsed corona discharges have been applied in the gas phase over the aqueous solution of the target compound [104], considering an anodetip-to-water distance d=1.0 cm. Although alkaline conditions favour the conversion of phenol, no such pH adjustments have been made, because the applicability of the pulsed corona technology should be qualified in an intrinsic way. The oxidation of a 5 mg/l (0.05 mM) phenol solution in tap water has been performed in threefold. The corona parameters are V=30 kV, C=100 pF, f=50 Hz, d=1.0 cm. The conversion and 95% confidence interval are plotted as a function of time, see Figure 4.19. After an oxidation time of 1 hour, a conversion X=92% ±7.9% has been reached. The relationship between ln(C/C0) and the oxidation time t is linear, implying first order reaction kinetics with a rate constant k1≈4.1⋅10-2 min-1. Characteristic pulse voltage, current and power are shown by Figure 4.20. The efficiency of phenol conversion, expressed by the G yield value is shown by Table 4.4. Results 67 Conversion (%) 100 80 60 40 20 0 0 Figure 4.19 20 40 Time (min) 60 80 The conversion of a 500 ml 5 mg/l (0.05 mM) phenol solution (tap waterbased). The corona parameters are V=30 kV, C=100 pF, f=50 Hz, d=1.0 cm. 8.0E+05 30 2.0E+04 20 I(t), Icor(t) 1.0E+04 10 0.0E+00 -100 6.0E+05 0 0 100 200 -1.0E+04 300 400 -10 Power (W) V(t) Current (A) Voltage (V) 3.0E+04 4.0E+05 2.0E+05 0.0E+00 -100 -2.0E+05 Time (ns) 0 100 200 300 400 Time (ns) Figure 4.20 Typical pulse voltage, current and power waveforms for corona in air, recorded after 1 hour of oxidation. The corona parameters are V=30 kV, C=100 pF, f=50 Hz, d=1.0 cm. Table 4.4 Conversion (X), pulse energy (Ep) and efficiency (G) during oxidation of a 500 ml 5 mg/l (0.05 mM) phenol solution (tap water-based); average values and 95% confidence intervals are presented. The corona parameters are V=30 kV, C=100 pF, f=50 Hz, d=1.0 cm. t (min) 15 30 45 60 X (%) 24.5 ±18.3 60.7 ±7.0 85.8 ±35.6 92.3 ±7.9 Ep (mJ) 13.7 ±4.2 13.1 ±5.1 14.3 ±25.4 12.4 ±7.4 G (mol/J)⋅108 1.08 ±1.14 1.39 ±0.54 1.20 ±2.63 1.10 ±2.08 G (100eV)-1 0.10 ±0.11 0.13 ±0.05 0.12 ±0.25 0.11 ±0.20 G (g/kWh) 3.7 ±3.9 4.7 ±1.8 4.1 ±8.9 2.5 ±5.5 The waveforms show rather much parasitic high frequency oscillations. These have been largely suppressed in the subsequent experiments. The 95% confidence levels of conversion and efficiency are rather large. The reason for the poor reproducibility is likely due to corona instability during these introductory experiments. 68 Chapter 4. The first quantitative measurements have been applied to phenol, hydroquinone, 1.4benzoquinone and resorcinol. 500 ml 25 mg/l (0.27 mM) phenol solutions (tap waterbased) have been exposed to pulsed corona discharges in air for 5 hours. The corona parameters are V=30 kV, C=100 pF, f=50 Hz, d=1.0 cm. The concentration determination has been performed in threefold, every 30 minutes. The first analysis showed the production of hydroquinone and resorcinol. The duplicate and triplicate measurements only showed hydroquinone and the conversion rate of phenol and hydroquinone was somewhat lower than for the case of the first measurement, see Figure 4.21. It has not been possible to distinguish hydroquinone from 1.4-benzoquinone, because these compounds show co-elution using the reversedphase HPLC column according to LC configuration 4. Hydroquinone is initially produced and may be oxidized to 1.4-benzoquinone. The carboxylic acids formic acid and acetic acid have been qualitatively identified by Capillary Zone Electrophoresis. Concentration (mol/l) 3.0E-04 phenol hydroquinone 2.0E-04 resorcinol 1.0E-04 0.0E+00 0 60 120 180 240 300 Time (min) Figure 4.21 The oxidation of a 500 ml 25 mg/l (0.27 mM) phenol solution (tap waterbased). The corona parameters are V=30 kV, C=100 pF, f=50 Hz, d=1.0 cm. Results 69 Ion-exclusion chromatography The applicability of a reversed-phase HPLC column for separation of the phenol oxidation product mixture has appeared to be limited. Although polyhydroxybenzenes can be properly separated, the carboxylic acid-functional ring-cleavage products cannot be retained. In order to separate both of these mutually different product groups, an ion-exclusion column has been applied. The conversion of phenol has been measured simultaneously using two different LC configurations with a reversed-phase HPLC column (5a) and an ion-exclusion column (5b). 500 ml 1.0 mM (94 mg/l) phenol solutions have been oxidized by pulsed corona discharges in air for 3 hours, see reactor configuration 6c. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The conversion determination has been performed in threefold, every 15 minutes during 3 hours. The samples have been directly analyzed after sampling. At startup and at the end of the experiment the solution acidity, conductivity and temperature have been recorded. Figure 4.22 shows the observed conversion as a function of the oxidation time measured by both columns. Characteristic pulse voltage and current waveforms are shown by Figure 4.23. The efficiency, conversion and pulse energy are listed by Table 4.5. There appear to be no significant differences between the conversion measurements using a reversed-phase HPLC column or an ion-exclusion column. Therefore the more powerful ICE column has been used for subsequent separations. The conversion time relationship seems to obey first order kinetics with a rate constant k1≈2.9⋅10-3 min-1. After three hours of oxidation, the conversion is about 39%. The efficiency is in the range of 1.9⋅10-8-2.7⋅10-8 mol/J and slowly decreases as a function of the oxidation time, because during oxidation progress less phenol molecules are available. 50 rp-HPLC Conversion (%) 40 ICE 30 20 10 0 0 60 120 180 Time (min) Figure 4.22 Phenol conversion measured simultaneously by rp-HPLC and ICE chromatography. Oxidation of a 500 ml 1.0 mM (94 mg/l) phenol solution. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. 70 Chapter 4. 3.0E+05 30 V(t) 2.0E+04 1.0E+04 20 10 I(t) Icor(t) 0.0E+00 0 -100 0 100 200 Time (ns) 300 Current (A) Power (W) Voltage (V) 3.0E+04 2.0E+05 1.0E+05 0.0E+00 400 -100 0 100 200 Time (ns) 300 400 Figure 4.23 Typical pulse voltage, current and power waveforms for corona in air, recorded after 1 hour of oxidation. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Table 4.5 Phenol conversion (X), pulse energy (Ep) and efficiency (G) during the oxidation of a 500 ml 1.0 mM (94 mg/l) phenol solution; average values and 95% confidence intervals are presented. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The conversion has been determined by ICE chromatography. t (min) 0 15 30 45 60 75 90 120 150 180 X (%) 0.0 ±0.0 4.6 ±1.1 9.5 ±4.4 13.3 ±0.5 17.5 ±3.3 20.7 ±2.3 24.1 ±2.8 30.0 ±3.2 34.3 ±1.9 39.3 ±1.9 Ep (mJ) 9.9 ±0.5 10.2 ±0.1 10.1 ±0.8 10.0 ±0.3 10.0 ±0.7 9.9 ±0.3 G (mol/J)⋅108 2.67 ±1.10 2.41 ±0.44 2.21 ±0.20 2.10 ±0.22 1.92 ±0.07 1.85 ±0.04 G (100eV)-1 0.26 ±0.11 0.23 ±0.04 0.21 ±0.02 0.20 ±0.02 0.19 ±0.01 0.18 ±0.01 G (g/kWh) 9.06 ±3.74 8.15 ±1.50 7.50 ±0.67 7.10 ±0.75 6.50 ±0.24 6.27 ±0.13 The change in solution acidity, conductivity and temperature are respectively ∆pH=-3.1 ±0.1, ∆σ=+1773 ±106 µS/cm and ∆T/T0=+0.1 ±0.1. The production of carboxylic acids is evidently shown by both the drastic pH decrease and conductivity increase. The energy dissipation, illustrated by the small solution temperature increase is very favourable, compared to the necessarily forced cooling that is applied for pulsed corona discharges in water. An approach has been made to identify a number of important oxidation products of phenol, by comparison of the retention times of unknown components in the ionexclusion chromatogram by the retention times of pure possible candidate oxidation products. The following components have been verified: the polyhydroxybenzenes: catechol, resorcinol, hydroquinone, pyrogallol and hydroxyhydroquinone; the quinone: 1,4-benzoquinone; the carboxylic acids: succinic acid, maleic acid, malonic acid, propionic acid, oxalic acid, acrylic acid, acetic acid and formic acid. Figure 4.24 shows a representative ICE chromatogram, recorded after 3 hours of oxidation time. 71 0 Figure 4.24 20.48 21.27 14.99 15.67 3.43 4.19 4.81 5.47 5.94 6.88 7.14 8.50 9.33 9.72 10.23 31.86 Absorbance (arb.u.) 3.24 Results 10 20 Retention time (min) 30 A representative ion-exclusion chromatogram of a 500 ml 1.0 mM (94 mg/l) phenol solution, oxidized for three hours. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The UV absorbance detector is set at 210 nm. Table 4.6 shows a survey of retention times. The retention times observed from the oxidation product mixture are average values, recorded in threefold after 2 h, 2.5 h and 3 hours of oxidation time. Also shown are pure candidate oxidation products. Table 4.6 Retention times observed in the chromatograms of oxidized phenol solutions and retention times of pure candidate oxidation products; average values and 95% confidence intervals are presented. Unidentified peaks are labelled with a question mark. Oxidation product mixture peak no tR (min) 1 3.24 ±0.00 2 3.45 ±0.01 3 4.20 ±0.00 4 4.81 ±0.00 5 5.47 ±0.01 6 5.94 ±0.00 7 6.88 ±0.01 8 7.15 ±0.01 9 8.49 ±0.01 10 9.34 ±0.01 11 9.72 ±0.01 12 10.23 ±0.01 13 14.94 ±0.03 14 15.72 ±0.03 15 20.53 ±0.04 16 21.31 ±0.04 17 31.88 ±0.03 tR (min) 3.61 4.75 5.99 7.21 8.04, 8.69 10.25 14.36 16.22 20.49 21.26 31.86 Pure compound name nitrate, hydrogen peroxide (as peak 1), oxalic acid ? maleic acid ? malonic acid ? succinic acid formic acid, acetic acid ? ? propionic acid pyrogallol hydroxyhydroquinone hydroquinone / 1,4-benzoquinone catechol / resorcinol phenol 72 Chapter 4. The first peak in the ion-exclusion chromatogram represents a component, which leaves the column without retention. The peak may also represent different co-eluting unretarded components. It may be assumed that this peak represents negative ions, because the ion-exclusion column excludes these ions, see section 4.1.4. An asymmetric second peak directly follows the first peak. The candidate anion is nitrate (NO3-) that is formed in aqueous solution by nitrogen oxides (NOx), produced by corona discharges in air. It may also be possible, that this peak is caused by hydrogen peroxide or dissolved ozone. An experiment has been performed, in which an oxidized phenol solution has been purged by helium. The peak area of the first peak remains unchanged by the helium purge; thus ozone cannot explain this peak, because it has been removed by the purge. However, by continuation of the corona discharges in helium, the peak area of the first peak increases. Hydrogen peroxide might thus also explain the existence of this peak, because it is produced from hydroxyl radicals, which are formed by the dissociation of water molecules. The ionexclusion chromatogram of a diluted hydrogen peroxide solution reveals the same typical set of two adjacent peaks, as observed in the chromatogram of oxidized phenol solutions. The asymmetric second peak may also represent oxalic acid, the strongest carboxylic acid present in the phenol oxidation product mixture: its first dissociation constant is pKa,I=1.23 [80]. It has been observed that the elution regions of the carboxylic acids and the different polyhydroxybenzenes are distinct. With regard to the used setup and conditions, the carboxylic acids elute between 3-10 minutes, the trihydroxybenzenes between 14-17 minutes, the dihydroxybenzenes between 20-22 minutes and monohydroxybenzene phenol at about 32 minutes. However, the identity of every single peak cannot be completely guaranteed without the definite proof of mass spectrometry. As yet, the production and conversion of the oxidation products have been reported by plotting the total peak area of the mentioned product classes as a function of the oxidation time, see Figures 4.25 and 4.26. 1.5E+06 UV absorbance (counts·s) UV absorbance (counts·s) 3.0E+06 2.0E+06 1.0E+06 1.0E+06 5.0E+05 0.0E+00 0.0E+00 0 30 60 90 120 Time (min) Figure 4.25 150 180 0 30 60 90 120 150 180 Time (min) The production of dihydroxybenzenes (left) and trihydroxybenzenes (right) as a function of time during the oxidation of a 500 ml 1.0 mM (94 mg/l) phenol solution. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Results 73 The dihydroxybenzenes catechol, resorcinol and hydroquinone seem to reach a maximum concentration after about 75 minutes. They may be transformed into trihydroxybenzenes, quinones or may undergo ring-cleavage by oxygen or ozone attack. The trihydroxybenzenes pyrogallol/hydroxyhydroquinone will definitely undergo ringcleavage, because they are stronger reducing agents than the dihydroxybenzenes. The trihydroxybenzene phloroglucinol (1,3,5-THB) is an exception, because it also reacts in a tautomeric keto-form, see section 2.5.3 Figure 2.21. By ring-cleavage of the hydroxybenzenes, a large variety of carboxylic acids is produced. UV absorbance (counts·s) 1.5E+07 1.0E+07 5.0E+06 0.0E+00 0 30 60 90 120 150 180 Time (min) Figure 4.26 The production of carboxylic acids as a function of the time during the oxidation of a 500 ml 1.0 mM (94 mg/l) phenol solution. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. From corona experiments in helium, also phenol conversion has been reported. A comparison of conversion values by corona in helium and air is reported in threefold after 30 minutes and 60 minutes of oxidation time, see Table 4.7. Table 4.7 Phenol conversion (X) by corona in helium and air. Oxidation of a 500 ml 1.0 mM (94 mg/l) phenol solution; presented are average values and 95% confidence intervals. The parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Time (min) 0 30 60 X (%) helium 0 9.6 ±6.4 13.4 ±3.4 X (%) air 0 9.0 ±2.1 17.5 ±2.5 The conversion by corona discharges in helium cannot be caused by ozone, because oxygen has been removed from the reactor by the helium purge. Therefore it is likely, that hydroxyl radicals are produced by helium ions/metastables bombardment of water molecules. 74 Chapter 4. Quantitative ion-exclusion chromatography Six major phenol oxidation products have been selected for quantitative analysis i.e. hydroquinone, hydroxyhydroquinone, formic acid, oxalic acid, glyoxylic acid and glyoxal. The analysis has been performed using the ion-exclusion column and a series connection of a UV absorbance and conductivity detector, according to LC configuration 6. The UV absorbance detector has been applied to detect the hydroxybenzenes, while the conductivity detector has been utilized for detection of the carboxylic acids. Pulsed corona discharges have been applied in both an air and argon gas phase over 100 ml 1.0⋅10-3 M phenol solutions, see reactor configuration 3b. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. By application of the corona in both air and argon, important aspects of the degradation mechanism can be resolved. The question is, whether either the hydroxyl radical or ozone accounts for degradation, or both oxidizer species contribute together to the chemical conversion of phenol in aqueous solution. The concentration of phenol and the selected oxidation products has been measured as a function of the oxidation time [105]. The calibration lines of the standard solutions are shown by Figures 4.27 and 4.28. It has been observed that the calibration lines of phenol, hydroquinone and formic acid have high correlation coefficients (r2). The standard series of oxalic acid and glyoxylic acid show moderate correlation. The correlation of the hydroxyhydroquinone standards is poor and the glyoxal data show no correlation. 20000 200 nm 8000 270 nm y = 1E+07x + 58.69 R2 = 0.9996 4000 y = 2E+06x + 15.101 R2 = 0.9994 Absorbance (mAUs) Absorbance (mAUs) 12000 15000 5000 8.0E-04 0.0E+00 1.2E-03 200 nm Absorbance (mAUs) Absorbance (mAUs) 8.0E-04 1.2E-03 25 270 nm 3000 2000 1000 200 nm 20 210 nm 15 10 5 0 0 4.0E-04 8.0E-04 [Hydroxyhydroquinone] (mol/l) Figure 4.27 4.0E-04 [Hydroquinone] (mol/l) 5000 0.0E+00 y = 1E+06x + 11.995 R2 = 0.9984 0 4.0E-04 [phenol] (mol/l) 4000 270 nm 10000 0 0.0E+00 y = 2E+07x + 244.39 R2 = 0.9978 200 nm 1.2E-03 0.0E+00 4.0E-04 8.0E-04 1.2E-03 [Glyoxal] (mol/l) UV absorbance detector calibration lines of phenol, hydroquinone, hydroxyhydroquinone and glyoxal. Indicated are UV absorbance detector wavelengths. Results 75 6.0E+05 4.0E+05 y = 6E+08x - 3180.4 2 R = 0.9983 2.0E+05 8.0E+05 0.0E+00 0.0E+00 4.0E-04 8.0E-04 1.2E-03 [Formic acid] (mol/l) Conductivity (arb.u.) 6.0E+05 y = 5E+08x - 13546 R2 = 0.9625 Conductivity (arb.u.) Conductivity (arb.u.) 8.0E+05 6.0E+05 y = 8E+08x - 91872 R2 = 0.8966 4.0E+05 2.0E+05 0.0E+00 0.0E+00 4.0E+05 4.0E-04 8.0E-04 1.2E-03 [Oxalic acid] (mol/l) 2.0E+05 0.0E+00 0.0E+00 4.0E-04 8.0E-04 1.2E-03 Glyoxylic acid (mol/l) Figure 4.28 Conductivity detector calibration lines of formic acid, oxalic acid and glyoxylic acid. The correlation quality of the standard is dependent on the chemical stability of the standard compound. Hydroxyhydroquinone is susceptible to oxidation, even by oxygen present in water. Glyoxal hydrolyzes and consequently exists in several oligomerized forms in aqueous solution [106]. Due to the large number of standards and samples in combination with long oxidation and analysis times, ageing of the samples is unavoidable. Nevertheless, ageing has been minimized by storage of both the standards and samples before analysis in dark vials under nitrogen at 0°C. Characteristic UV absorbance and conductivity chromatograms, obtained from phenol solutions oxidized by corona in air and argon during 2 hours, are shown by Figures 4.29 and 4.30. It has been observed, that the oxidized solutions have different colors: the solution oxidized by corona in air is pale yellow, while the solution oxidized by corona in argon is pale beige. The observed retention times of the standards are shown by Table 4.8. Both the hydroxybenzenes and glyoxal do not exist in ionic form in aqueous solution, therefore these compounds are not detectable by the conductivity detector. The time lag between the signals of conductivity and UV absorbance detector is caused by the intermediate suppressor. 76 Chapter 4. Figure 4.29 UV absorbance chromatograms of 100 ml 1.0 mM phenol solutions, oxidized by corona in air (left) and argon (right) during 2 hours. The corona parameters are: V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Indicated are UV absorption wavelengths. Figure 4.30 Conductivity chromatograms of 100 ml 1.0 mM phenol solutions, oxidized by corona in air and argon during 2 hours. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Table 4.8 Absorbance (UV) and conductivity (CON) detector average retention times of the standard compounds. Also shown are 95% confidence intervals. UV tR 3.18 4.84 7.26 10.03 17.25 20.83 32.30 (min) ±0.09 ±0.01 ±0.01 ±0.01 ±0.01 ±0.03 ±0.07 CON 3.44 5.07 7.49 tR (min) ±0.06 ±0.02 ±0.01 - Standard compound Oxalic acid Glyoxylic acid Formic acid Glyoxal Hydroxyhydroquinone Hydroquinone Phenol Results 77 The conversion of phenol in aqueous solution by pulsed corona discharges in both an air and argon atmosphere is shown by Figure 4.31. The conversion by corona in argon appears to be higher than the conversion in air. After 2 hours, the corona in air has converted 59% of the initial present phenol amount, while the corona in argon has converted 88%. The conversion seems to obey first order kinetics: the rate contants are k1air ≈8.2⋅10-3 min-1 and k1argon ≈1.8⋅10-2 min-1. Conversion (%) 100 200 270 200 270 80 60 nm nm nm nm air air Ar Ar 40 20 0 0 30 60 90 120 Time (min) Phenol conversion during oxidation of 100 ml 1.0 mM phenol solutions by corona in air and argon. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Indicated are UV absorbance wavelengths. Figure 4.31 The DHB hydroquinone elutes just before its isomers resorcinol (1,3-DHB) and catechol (1,2-DHB), which cannot be separated by the ion-exclusion column, see Figure 4.29. An estimation of the order of magnitude of the total DHB concentration has been performed by relating the total DHB peak area to the calibration line of hydroquinone, assuming similar extinction coefficients for the three DHB isomers. Figures 4.32 and 4.33 show the hydroquinone concentration and the total DHB concentration versus the oxidation time, respectively. During the oxidation the maximum hydroquinone concentration is 8.4⋅10-6 M in air and 3.1⋅10-5 M in argon, after about 45 minutes. [Hydroquinone] (mol/l) 5.0E-05 270 nm air 270 nm argon 4.0E-05 3.0E-05 2.0E-05 1.0E-05 0.0E+00 0 30 60 90 120 Time (min) Figure 4.32 The production of hydroquinone during the oxidation of 100 ml 1.0 mM phenol solutions by corona in air and argon. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. 78 Chapter 4. The total DHB concentration reaches a maximum value of 1.3⋅10-4 M in air and 3.4 ⋅10-4 M in argon after about 60 minutes. It is remarkable, that the DHB amounts produced by corona in argon are considerably higher than the DHB amounts produced by corona in air. 5.0E-04 270 nm air [DHB] (mol/l) 4.0E-04 270 nm argon 3.0E-04 2.0E-04 1.0E-04 0.0E+00 0 30 60 90 120 Time (min) An estimation of the production of total dihydroxybenzenes during the oxidation of 100 ml 1.0 mM phenol solutions by corona in air and argon. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Figure 4.33 The THB hydroxyhydroquinone has not been identified in the phenol oxidation product mixture. A possible explanation for the absence of this compound is the limited stability of this strong reducing agent. Nevertheless, it has been observed by ICE analysis immediately after corona treatment in air, see Figure 4.24. With regard to the ringcleavage products of phenol, the following observations have been made. The production of formic acid by corona in air is much higher than the production by corona in argon. After 2 hours of corona discharges, a concentration of 2.3⋅10-4 M is reached by corona in air and a concentration of 8.9⋅10-5 M is reached by corona in argon. For the observed time span, the production rate increases almost linearly with time, see Figure 4.34. [Formic acid] (mol/l) 2.5E-04 CON air CON argon 2.0E-04 1.5E-04 1.0E-04 5.0E-05 0.0E+00 0 30 60 90 120 Time (min) Figure 4.34 The production of formic acid during the oxidation of 100 ml 1.0 mM phenol solutions by corona in air and argon. The corona parameters are: V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Results 79 The estimation of the oxalic acid concentration involves an error margin, because oxalic acid appears to co-elute with a component leaving the column unretarded, as can be observed from Figures 4.29 and 4.30. This component is likely to be the nitrate ion, according to the discussion in section 4.1.4. Oxalic acid elutes rapidly, because it is the strongest organic acid observed in the phenol oxidation product mixture. Oxalic acid has only been detected during phenol oxidation by corona in air: after two hours of oxidation a concentration of about 3.9⋅10-4 M has been reached. Figure 4.35 shows the production of oxalic acid versus the oxidation time for corona discharges in air. [Oxalic acid] (mol/l) 5.0E-04 CON air 4.0E-04 3.0E-04 2.0E-04 1.0E-04 0.0E+00 0 30 60 90 120 Time (min) The production of oxalic acid during the oxidation of a 100 ml 1.0 mM phenol solution by corona in air. Oxalic acid has not been detected during oxidation by corona in argon. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Figure 4.35 The dialdehyde glyoxal has not been found in traceable amounts. Glyoxylic acid, an oxidation product of glyoxal, appears in small amounts in both the phenol solution oxidized by corona in air viz. 4⋅10-5 M to 5⋅10-5 M and the phenol solution oxidized by corona in argon viz. 3⋅10-5 M, see Figure 4.36. [Glyoxylic acid] (mol/l) 6.0E-05 4.0E-05 2.0E-05 CON air CON argon 0.0E+00 0 30 60 90 120 Time (min) Figure 4.36 The production of glyoxylic acid during the oxidation of 100 ml 1.0 mM phenol solutions by corona in air and argon. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. 80 Chapter 4. A very strong absorption signal has been found in the UV absorbance chromatograms of the phenol solution oxidized by corona in air, at the retention time tR=13.73 ±0.01 min and only pronounced at the wavelengths λ=255 nm and λ=270 nm. There is a corresponding conductivity signal at the retention time tR=13.97 ±0.03 min. The signal appears to be almost absent in the UV absorbance and conduction chromatograms of the phenol solution oxidized by corona in argon. Figure 4.37 shows the observed peak areas of both detectors versus the oxidation time. 6.0E+04 255 nm 270 nm CON 800 4.0E+04 400 2.0E+04 0 0.0E+00 0 Figure 4.37 30 60 90 Time (min) Conductivity (arb. u) Absorbance (mAU) 1200 120 The production of an unknown compound during the oxidation of a 100 ml 1.0 mM phenol solution by corona in air. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The identity of this component might be revealed in the following way. The component appears at the end of the carboxylic acid retention time range (3<tR<15 min) of the UV absorption chromatogram. It exists in partially dissociated form in aqueous solution, because it is detectable by the conductivity detector. Its strong UV absorption at wavelengths λ=255 nm and λ=270 nm indicates the presence of unsaturated carboncarbon bonds. The compound is formed under ring-cleavage conditions. It is known from literature that the oxidation of phenol, benzene or catechol yields the ring-cleavage product cis,cis-muconic acid (cis,cis-1,3-butadiene-1,4-dicarboxylic acid) [11,107,108]. According to the observations, the unknown compound possibly is cis,cis-muconic acid. Cis,cis-muconic acid has not been introduced in the standard series, because of its presumed but ambiguous limited stability [7]. Typical pulse voltage, current and power waveforms for corona in air and argon are shown by Figure 4.38. The corona current in argon is higher than the current in air. This is explained by the electronegative character of oxygen, that tends to inhibit the corona discharges in air compared to the discharges in argon. Results 81 1.0E+04 20 10 I(t) Icor(t) 0.0E+00 -100 0 100 Voltage (V) 2.0E+04 Current (A) Voltage (V) V(t) 300 2.0E+04 20 1.0E+04 10 I(t) Icor(t) -100 400 0 corona in air 0 200 300 400 corona in argon 6.0E+05 Power (W) Power (W) 100 Time (ns) Time (ns) 6.0E+05 30 V(t) 0.0E+00 0 200 corona in argon 3.0E+04 30 Current (A) corona in air 3.0E+04 4.0E+05 2.0E+05 4.0E+05 2.0E+05 0.0E+00 0.0E+00 -100 0 100 200 300 400 -100 Time (ns) 0 100 200 300 400 Time (ns) Typical pulse voltage, current and power waveforms for corona in air (left) and argon (right), recorded after 1 hour of oxidation. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Figure 4.38 From the pulse energy and conversion, the efficiency of phenol oxidation by corona in air and argon has been calculated, see Table 4.9 and Figure 4.39. Table 4.9 Efficiency (G), pulse energy (Ep) and phenol conversion at λ=270 nm (X270) during oxidation of 100 ml 1.0 mM phenol solutions by corona in air and by corona in argon. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Time (min) 15 Ep (mJ) 6.3 30 45 60 75 90 105 120 6.2 5.8 5.7 5.8 5.4 5.7 5.6 corona in air X270 G G Ep G ⋅108 (%) (mol/J) p100eV (g/kWh) (mJ) 14.4 2.60 0.25 8.8 10.2 26.1 2.35 0.23 8.0 10.3 35.0 2.23 0.21 7.5 10.8 42.1 2.01 0.19 6.8 10.4 48.0 1.79 0.17 6.1 10.7 52.6 1.74 0.17 5.9 10.8 56.2 1.48 0.14 5.0 10.9 59.2 1.38 0.13 4.7 11.0 corona in argon X270 G G G ⋅108 (%) (mol/J) p100eV (g/kWh) 18.8 2.09 0.20 7.1 43.1 2.35 0.23 8.0 50.2 1.72 0.17 5.8 65.5 1.72 0.17 5.8 75.5 1.53 0.15 5.2 82.2 1.36 0.13 4.6 85.8 1.20 0.12 4.1 88.5 1.05 0.10 3.6 82 Chapter 4. Only minor differences in efficiency appear to exist for phenol conversion by corona in air and argon. With increasing conversion, less phenol molecules are available for oxidation and a competition with intermediate oxidation products exists, thus the efficiency decreases. G (mol/J) 3.0E-08 2.0E-08 1.0E-08 air argon 0.0E+00 0 Figure 4.39 20 40 60 80 Conversion X270 (%) 100 Efficiency of phenol conversion during oxidation of 100 ml 1.0 mM phenol solutions by corona in air and argon. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The conversion has been measured by a UV absorbance detector set at λ= 270 nm. Nevertheless, the difference between the phenol oxidation pathways for corona in air and argon is distinct. This is explained by simplified degradation pathways of phenol by hydroxyl radicals, oxygen, ozone and argon ion bombardment. A detailed phenol oxidation mechanism is presented in section 5.3. The attack of the hydroxyl radical on phenol will initially produce para- and orthodihydroxycyclohexadienyl (DHCHD) radicals, see Figure 4.40. The meta-DHCHD radical is not considered here, because it is likely to be less stable than the others. OH OH HO • HO phenol Figure 4.40 OH • H OH H para-DHCHD radical ortho-DHCHD radical The formation of dihydroxycyclohexadienyl (DHCHD) radicals from the attack of the hydroxyl radical on phenol. In an air atmosphere, oxygen will attack these radicals to produce dihydroxycyclohexadienylperoxyl (DHCHDP) radicals, see Figure 4.41. The ortho-DHCHDP radical may produce catechol by elimination of the hydroperoxyl radical (HO2). The paraDHCHDP radical probably yields hydroquinone by dimerization to a tetraoxide followed by decomposition; hydroperoxyl elimination is not likely here, because of the larger -H to •OO- distance of the para-DHCHDP radical compared to the ortho-DHCHDP radical. By action of oxygen, the DHCHDP radicals are eventually converted to endoperoxides. These very instable intermediates decompose by ring-cleavage. Results 83 Direct ozone attack on phenol will also invoke ring-cleavage by addition, see Figure 4.42. In this way, a complex mixture of aliphatic unsaturated and saturated C1-C6 hydrocarbons will be produced, having polyfunctional groups like carboxyl, aldehyde, ketone or hydroxyl groups. Glyoxal, glyoxylic acid and oxalic acid may result from multifold attack of ozone on phenol. Formic acid may be produced by carbon monoxide loss from glyoxylic acid or by decarboxylation of oxalic acid. OH hydroquinone OH T OH HO OO • HO H O2 • O2 HO H OH HO H O2 OH • O O DHCHDP DHCHD OO • H OH OH O2 COOH, CHO, CO, OH OH OO• C1-C6 ring cleavage products endo peroxides - HO2 OH OH catechol Figure 4.41 A simplified mechanism of the oxidation of DHCHD and DHCHDP radicals to endoperoxides, followed by aromatic ring-cleavage. OH OH O O OH O O3 molozonides phenol O O3 OH Figure 4.42 ozonides OH O O O O O zwitterions O O HO H + C O O H2O HO C H O O H OOH HO OH -H2O H O hydroperoxy alkanols A simplified mechanism of phenol oxidation by ozone. O OH H O unsaturated polyfunctional aliphatic hydrocarbons 84 Chapter 4. On the contrary, when pulsed corona discharges take place in an argon atmosphere over an oxygen-free phenol solution, the argon ions/metastables created by the corona discharges will dissociate water molecules to produce hydroxyl radicals (and hydrogen atoms) but no ozone can be formed. Hydroxylation of phenol will be the main degradation pathway and hydroxybenzenes will be found in much higher amounts than for the case of corona oxidation in air, see Figure 4.43. OH HO • HO H HO HO OH -H2O H hydroquinone H HO DHCHD OH isomerization OH • OH O H OH Figure 4.43 HO HO O OH H OH -H2O OH H OH OH catechol A simplified mechanism of phenol oxidation by hydroxyl radicals under oxygen-free conditions. The fact, that still some ring-cleavage products are found during corona oxidation in argon, can be explained by ring fragmentation by the argon ions/metastables bombardment. Ring-cleavage also takes place by small amounts of oxygen, that cannot be removed by argon purging. The differences in amounts of polyhydroxybenzenes bring about the color difference between the oxidized solutions. As can be derived from both the chromatograms and a carbon mass balance, a certain number of unknown products remains, whose identity is difficult to resolve. They are likely to be carboxylic acids, because the conductivity detector is able to observe these partially ionic compounds. Candidates are C1-C6 mono- and dicarboxylic acids with alkanol-, aldehyde- or ketone-functional groups. Finally, by comparison of the conductivity chromatograms of oxidized phenol solutions and deionized water exposed to corona in air (section 4.1.4), the following observation has been made: the conductivity signal at tR=8.73 min from deionized water exposed to corona in air is present in the chromatogram of the phenol solution oxidized by corona in argon, but is absent in the chromatogram of the air-oxidized phenol solution. This might be explained by anode metal sputtering differences between corona in argon and air. Also it should be considered, that the regarding peak may be due to an oxidation product with equal retention time. Results 85 Gas chromatography By oxidation of phenol in aqueous solution, the following oxidation products may be introduced in the gas phase over the oxidized solution: carbon oxides, unsaturated hydrocarbons and volatile aldehydes. A preliminary direct gas chromatography analysis of a helium flow reactor purge did not reveal any information. Therefore a specific procedure has been chosen according to NIOSH for screening the presence of aldehydes. The identification of other candidate oxidation products has been performed by infrared spectroscopy and is discussed in section 4.2.3. The aldehyde screening test involves a chemical derivatization reaction of aldehydes. Gas sampling tubes containing the derivatization agent HMP are exposed to an argon 5.0 purge, originating from the reactor. A 500 ml 1.0 mM phenol solution has been oxidized for 3 hours, see reactor configuration 7. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Before oxidation, the reactor and phenol solution are purged with argon 5.0 and the purge gas is directed through a gas sampling tube; this is the background sample. Due to this purge the corona discharges take place in argon 5.0. After the oxidation the reactor is purged again and the collected purge gas is directed through a different fresh tube; this is the corona sample. The background and corona sample tube contents are extracted with toluene and the extract is analyzed by gas chromatography-mass spectrometry, see GC configuration 1. The obtained chromatograms are shown by Figure 4.44. FID signal (arb.u. ) 4.0E+06 corona sample 3.0E+06 2.0E+06 1.0E+06 background sample 0.0E+00 130 150 170 190 210 230 Retention time (s) Figure 4.44 GC chromatograms of toluene-extracted sampling tube contents. The tubes have been exposed to an argon 5.0 purge originating from the reactor, before oxidation (background sample) and after oxidation (corona sample). The reactor contains 500 ml 1.0 mM phenol solution. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. 86 Chapter 4. The chromatograms reveal non-derivatized HMP present in both samples, eluting between about 170 and 210 seconds. HMP-aldehyde derivatization products should occur just before and after the elution of non-derivatized HMP. Although many small peaks are visible, there appear to be no differences between the corona sample and the background sample. All peaks below 145 seconds definitely originate from the extraction solvent, viz. toluene together with xylene impurities. It is concluded that no volatile aldehydes have been detected during oxidation of a phenol solution by corona in argon, within the lower detection limit of this procedure. This limit is not specified, because this test is an overall screening technique. A specific acetaldehyde test (method 2538) using the same derivatization technique but different parameters has a lower detection limit of 0.74 ppm ≡ 1.3 mg/m3. Summary The conversion of phenol in aqueous solution by pulsed corona discharges increases by increasing the corona load voltage, corona pulse repetition rate and solution alkalinity. The location of the discharges is best at some distance from the liquid-gas interface. Ion-exclusion chromatography has proven to be considerably more powerful for separation of the complex phenol oxidation product mixture than reversed-phase HPLC. Identified phenol oxidation products are di- and trihydroxybenzenes, mono- and dicarboxylic acids. The phenol oxidation pathways strongly depend on the composition of the gas phase, where the corona discharges are produced. Corona in argon or helium also invokes oxidation by the formation of hydroxyl radicals due to ions/metastables bombardment of water. The maximum obtained phenol conversion efficiency is about G=2.7⋅10-8 mol/J ≡ 0.26 (100eV)-1 ≡ 9.1 g/kWh at X=9.5% conversion by corona in air. According to an aldehyde screening test, no volatile aldehydes have been detected during oxidation of a phenol solution by corona in argon. Results 87 4.2.2. Mass spectrometry The first attempts to identify phenol oxidation products have been performed by liquid chromatography coupled mass spectrometry (LC-MS) analyses on samples obtained by Solid Phase Extraction (SPE), see MS configuration 1. Phenol degradation by pulsed corona discharges involves oxidation by highly reactive, thus non-specific, hydroxyl radicals. This means, that a wide variety of oxidation products is formed. However, the oxidation of low content phenol solutions, according to the scope of this thesis, implies that the produced amounts of oxidation products are very low. The oxidation of high content phenol solutions may produce a different oxidation product mixture as a result of polymerization. Also these solutions require a long treatment time to achieve reasonable conversion. Pre-concentration of the oxidation product mixture of low content phenol solutions by SPE is attractive. Unfortunately, LC-MS analyses of SPEprocessed oxidation product samples have not revealed the identity of any of the components. Also it has been observed, that certain oxidation products will be lost to the SPE matrix. This has been concluded from chromatograms taken from an oxidized phenol solution before and after SPE processing. Therefore, SPE has not been applied to subsequent analyses of phenol oxidation product mixtures despite its advantages. Different LC-MS approaches have been made, but all of them have been performed offline i.e. without ICE column. The acidic eluent, required for separation according to the ion-exclusion mechanism, causes a high background noise that complicates the analysis of the weak signals of the oxidation product components. Also the spraying of the aqueous eluent has appeared to be problematic, due to the high surface tension of water. Only sparingly results have been achieved by application of negative IonSpray of a phenol oxidation product sample in a 100% acetonitrile flow. Prior to the MS analysis 0.1 v/v % concentrated ammonia has been added to the sample, to promote the ion formation by converting carboxylic acids and phenols into their anionic form. The sample has been prepared by oxidation of a 500 ml 1.0 mM (94 mg/l) phenol solution for 3 hours, see reactor configuration 6c. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. With regard to the interpretation of the mass-spectrum shown by Figure 4.45, it should be remarked that the indicated masses are equal to Mass-1, because [M-H]- ions are produced. 69 Intensity (arb.u.) 6.0E+05 62 113 4.0E+05 93 115 97 175 89 99 141 2.0E+05 0.0E+00 0 Figure 4.45 50 100 150 200 m/z (amu) 250 300 IonSpray mass spectrum of a 500 ml 1.0 mM phenol solution, oxidized during 3 hours. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. 88 Chapter 4. An extensive list of probable phenol oxidation products has been mainly derived from literature references (section 2.4.1), see Table 4.10. This list is used to identify the observed masses (FW). m/z=62 might be a (CO2)(H2O) cluster. m/z=69 has not been identified. If the peak at m/z=89 is no noise peak, it may be due to singly deprotonated oxalic acid or deprotonated lactic acid (90). m/z=93 is due to phenol (FW=94 g/mol, 76.6%C, 6.4%H, 17.0%O). m/z=97 and 99 have not been identified. m/z=113 is likely to be trifluoroacetic acid (114), that has been left behind in the IonSpray compartment from previous attempts. m/z=115 may originate from singly deprotonated maleic/fumaric acid or deprotonated dioxobutyric acid (116). m/z=141 is probably a singly deprotonated muconic acid enantiomer (142). The peak at m/z=175 may contain tetrahydroxybenzoquinone (172). Higher masses are likely to be polymerized benzoquinones/hydroquinones [34], although polymerization is not likely to be important during the oxidation of 1 mM phenol solutions. Other attempts have included the application of APCI mass spectrometry in combination with the TFA eluent and addition of acetonitrile to favour the sprayability. Also the sample preparatory options i.e. fractional collection and freeze drying have been applied. No spectra could be obtained from these experiments. There are several reasons for the fact, that identification by LC-MS has been problematic. Although the range of oxidation products is broad, the individual component concentration is low. Negative IonSpray, especially suitable for ionizing phenols and carboxylic acids is less sensitive than positive ion spray. The available interfaces IonSpray and APCI appear to be unsuitable for analyzing masses below 200 amu. Table 4.10a A list of possible phenol oxidation products, sorted by increasing molecular weight FW (g/mol). FW<FWphenol. Also shown are abbreviated compound structure and w/w-% carbon, hydrogen and oxygen. Compound name Acetylene Carbon monoxide Ethylene Formaldehyde Carbon dioxide Acetaldehyde Formic acid Butadiene Glyoxal Acetic acid Acrylic acid Malonaldehyde Glyoxylic acid Propionic acid Maleic aldehyde Pyruvic acid Butyric acid Oxalic acid Lactic acid Compound structure CH≡CH CO CH2=CH2 HCHO CO2 CH3-CHO HCOOH CH2=CH-CH=CH2 HC(O)-CHO CH3-COOH CH2=C(H)COOH HC(O)-CH2-CHO HC(O)-COOH C2H5-COOH HC(O)-CH=CH-CHO CH3-C(O)-COOH C3H7-COOH HOOC-COOH CH3-CH(OH)-COOH %C 92.3 42.9 85.7 40.0 27.3 54.5 26.1 88.9 41.4 40.0 50.0 50.0 32.4 48.6 57.1 40.9 54.5 26.7 40.0 %H 7.7 0.0 14.3 6.7 0.0 9.1 4.3 11.1 3.4 6.7 5.6 5.6 2.7 8.1 4.8 4.5 9.1 2.2 6.7 %O 0 57.1 0 53.3 72.7 36.4 69.6 0 55.2 53.3 44.4 44.4 64.9 43.2 38.1 54.5 36.4 71.1 53.3 g/mol 26 28 28 30 44 44 46 54 58 60 72 72 74 74 84 88 88 90 90 Results 89 Table 4.10b A list of possible phenol oxidation products, sorted by increasing molecular weight FW (g/mol). FW> FWphenol. Also shown are abbreviated compound structure and w/w-% carbon, hydrogen and oxygen. Compound name Oxobutyric acid Valeric acid Malonic acid o,p-Benzoquinone Muconaldehyde Dihydroxybenzenes Cyclohexadienediol Cyclohexanedione Dioxosuccinic aldehyde Maleic / fumaric acid Dioxobutyric acid Caproic acid Ketomalonic acid Succinic acid Hydroxybenzoquinone Trihydroxybenzenes Oxalacetic acid Glutaric acid Malic acid Dihydroxybenzoquinones Muconic acid Dioxosuccinic acid Adipic acid Tartaric acid Trihydroxybenzoquinone Dibenzofuran Tetrahydroxybenzoquinone Dibenzo-p-dioxin Dihydroxybiphenyl Dimer BQ-BQ Dimer HQ-BQ Purpurogallin Compound structure HC(O)-(CH2)2-COOH CH3-(CH2)3-COOH HOOC-CH2-COOH O=C6H4=O HC(O)-CH=CH-CH=CH-CHO C6H4(OH)2 HO(H)-C6H4-(H)OH C6H8(O)2 HC(O)-C(O)-C(O)-CHO HOOC-CH=CH-COOH HC(O)-C(O)-CH2-COOH CH3-(CH2)4-COOH HOOC-C(O)-COOH HOOC-(CH2)2-COOH O=C6H3(OH)=O C6H3(OH)3 HOOC-CH2-C(O)-COOH HOOC-(CH2)3-COOH HOOC-CH2-CH(OH)-COOH C6H2(OH)2(O)2 HOOC-CH=CH-CH=CH-COOH HOOC-C(O)-C(O)-COOH HOOC-(CH2)4-COOH HOOC-CH(OH)-CH(OH)-COOH C6H(O)2(OH)3 C6H4(O)C6H4 C6(O)2(OH)4 C6H4 (O,O)C6H4 HOC6H4-C6H4OH C6H3(O)2-C6H3(O)2 C6H3(O)2-C6H3(OH)2 C6H(OH)3C5H3(OH)(O) %C 47.1 58.8 34.6 66.7 65.5 65.5 64.3 64.3 42.1 41.4 41.4 62.1 30.5 40.7 58.1 57.1 36.4 45.5 35.8 51.4 50.7 32.9 49.3 32.0 46.2 85.7 41.9 78.3 77.4 67.3 66.7 60.0 %H 5.9 9.8 3.8 3.7 5.5 5.5 7.1 7.1 1.8 3.4 3.4 10.3 1.7 5.1 3.2 4.8 3.0 6.1 4.5 2.9 4.2 1.4 6.8 4.0 2.6 4.8 2.3 4.3 5.4 2.8 3.7 3.6 %O 47.1 31.4 61.5 29.6 29.1 29.1 28.6 28.6 56.1 55.2 55.2 27.6 67.8 54.2 38.7 38.1 60.6 48.5 59.7 45.7 45.1 65.8 43.8 64.0 51.3 9.5 55.8 17.4 17.2 29.9 29.6 36.4 g/mol 102 102 104 108 110 110 112 112 114 116 116 116 118 118 124 126 132 132 134 140 142 146 146 150 156 168 172 184 186 214 216 220 With regard to hydroxylated phenols and C1-C6 ring-cleavage products, the phenol oxidation product mixture exhibits a mass range of about 30 amu (formaldehyde) till 172 amu (tetrahydroxybenzoquinone). Next to off-line LC-MS, mass analyses have been performed by electron-impact mass spectrometry (EI-MS), see MS configuration 2. In order to have sufficient detector signal from the oxidation products, a 100 ml 5.0⋅10-1 M (concentrated) phenol solution has been oxidized for 3 hours, see reactor configuration 3b. The corona parameters are V=30 kV, C=1 nF, f=300 Hz, d=1.0 cm. By oxidation the initial colorless phenol solution turns deep brown. Mass spectra have been recorded for the mass range 1-500 amu, see Figure 4.46. 90 Figure 4.46 Chapter 4. EI-mass spectra of a phenol oxidation product mixture, obtained by oxidation of a 100 ml 0.5 M phenol solution by corona in air. The corona parameters are: V=30 kV, C=1 nF, f=300 Hz, d=1.0 cm. The most evident masses found are m/z=94 (phenol) and m/z=95 (protonated phenol). Protonation is caused by the ionization of water molecules by the 100 eV electron beam. m/z=110 may originate from the hydroxybenzene isomers, but muconic aldehyde is also possible. m/z=105 might be protonated malonic acid. m/z=106 cannot be identified. The observed masses below 94 amu may contain phenol oxidation products, but are mainly due to EI-fragmentation of phenol. Results 91 The mass analyses from 110-500 amu sometimes reveal traces of higher molecular weight species, but these results are not reproducible when the detector integration time is increased. This is remarkable, because the deep brown color of the oxidized solution clearly implies the presence of polymerized benzoquinones/hydroquinones i.e. synthetic humic acids [34]. It is possible that the observed suspended polymeric particles cannot enter the mass spectrometer by means of the applied capillary tubing. Summary Several IonSpray/APCI-MS and EI-MS measurements have not confirmed the presence of both theoretically possible and literature-reported oxidation products, in spite of the fact that significant phenol conversion has been demonstrated. This has limited the quantitative measurements, because the identification of compounds from complex ICE chromatograms by comparison of retention times is problematic. As main reason for the negative results can be mentioned that, although the product range is broad, the concentration of individual oxidation products is low. 92 Chapter 4. 4.2.3. Spectroscopy Laser-Induced Fluorescence spectroscopy Next to chromatography, the conversion of phenol has also been measured by LaserInduced Fluorescence (LIF) spectroscopy [109,110]. Regarding the fact, that LIF literature only regards gas phase studies, the application of LIF for liquid phase analysis is considered to be a new approach. LIF spectroscopy enables in-situ and time-resolved measurements, which is very favourable to batch-wise sampling and analysis time inherent to liquid chromatography. In addition to phenol conversion measurements, the in-situ monitoring of total hydroxybenzenes fluorescence is an approach to oxidation or detoxification progress. A typical LIF spectrum of phenol in aqueous solution at 1.0⋅10-5 M concentration is shown by Figure 4.47. The excitation is performed by the fourth harmonic wavelength of a Nd:YAG laser i.e. λ=266 nm. No fine structure of rotational lines can be observed, because a low resolution grating (150 mm-1) has been applied to scan the broad wavelength range. Figure 4.47 A typical LIF spectrum of an 1.0⋅10-5 M aqueous phenol solution, excited at λ=266 nm. The fluorescence of phenol excited states in aqueous solution will be quenched by water, phenol and oxidation product molecules. The measured dependence of the LIF peak intensity at 298 nm on the phenol concentration is shown by Figure 4.48. The intensity appears to be linear for the range 1.0⋅10-6 M to 1.0⋅10-5 M. The extent of quenching by oxidation products has been investigated as follows. The fluorescence peak intensity of phenol-resorcinol and phenol-pyrogallol mixtures has been measured at constant total concentration of 1.0⋅10-5 M while the partial concentrations have been mutually varied, see Figure 4.48. Resorcinol and pyrogallol are the polyhydroxybenzenes that are both most stable and most fluorescent. By increasing the concentration of resorcinol or pyrogallol thus decreasing the phenol concentration, the LIF peak intensity at 298 nm decreases linearly. Collisional quenching of phenol molecules by the oxidation products resorcinol and pyrogallol thus appears not to influence linearity at these concentration levels. Also, the tested concentration range of oxidation products is far higher than actually observed during phenol oxidation. Results 93 1.0 LIF signal (arb.u.) 0.8 0.6 0.4 0.2 0.0 0.0 Figure 4.48 phenol phenol-resorcinol phenol-pyrogallol 0.2 0.4 0.6 0.8 -5 Phenol concentration (10 mol/l) 1.0 The LIF peak intensity of phenol versus the concentration. Also shown is the LIF peak intensity of several phenol-resorcinol and phenol-pyrogallol aqueous mixtures at a total concentration of 1.0⋅10-5 M. A 100 ml 1.0⋅10-5 M phenol solution has been oxidized for 20 minutes, see reactor configuration 8b. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The LIF spectrum of the phenol solution after 4 minutes of oxidation is shown by Figure 4.49. Also indicated is the LIF spectrum due to the remaining amount of phenol after 4 minutes of oxidation. This decomposed spectrum has been derived by fitting the LIF spectrum of the unoxidized phenol solution at t=0 minutes to the LIF spectrum of the oxidized solution, over the wavelength range 250-280 nm. Figure 4.49 The LIF spectrum of a 100 ml 1.0⋅10-5 M phenol solution, oxidized for 4 minutes. Also shown is the decomposed spectrum of non-oxidized phenol. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. With regard to Figure 4.48, it is justified to assume that the decomposed phenol fluorescence intensity is equal to the phenol concentration. Figure 4.50 shows the phenol concentration versus the oxidation time. 94 Chapter 4. The phenol concentration seems to decrease according to two different decay time constants. For the first 10 minutes the decay time constant is about 25 minutes, from 10 to 17 minutes the decay time constant is about 9.8 minutes. -5 Concentration (10 mol/l) 1 0.1 0 5 10 15 20 Time (min) Figure 4.50 The phenol concentration versus time during oxidation of a 100 ml 1.0⋅10-5 M phenol solution. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Next, the oxidation of 300 ml 1.0⋅10-4 M phenol solutions has been studied under similar conditions, see reactor configuration 8a; the oxidation time is 3 hours. The effect of the laser on phenol degradation has been determined by measuring the LIF peak intensity at λ=298 nm of a 300 ml 1.0⋅10-4 M phenol solution versus the laser exposure time at a laser beam pulse rate of 10 Hz, see Figure 4.51. LIF signal (arb.u.) 1 0.1 0 50 100 150 200 Time (min) Figure 4.51 Decrease of the LIF peak intensity of a 300 ml 1.0⋅10-4 M phenol solution due to degradation by exposure to a Nd:YAG laser beam at wavelength λ=266 nm and 10 Hz pulse rate. The degradation effect is substantial for long exposure times. The 266 nm photons have an energy of 4.7 eV, which is very close to the bond strength of a hydrogen atom attached to a benzene ring (465 ±3 kJ/mol [24]). Therefore, for all experiments the exposure time has been kept very short viz. 5 seconds and the laser pulse energy has been minimized to 0.8-1.1 mJ. Results 95 The influence of quenching effects with regard to the 1.0⋅10-4 M phenol solution has been tested for the range 10-6-10-4 M, see Figure 4.52. The LIF intensity is linear to the phenol concentration for the range 1⋅10-6-4⋅10-5 M. At higher concentrations deviation from linearity occurs due to absorption of the excitation photons; the phenol data points fit well to Equation 2.22 where kQ=0, as is shown by the dotted line. Both resorcinol and pyrogallol quench the fluorescence of phenol excited states significantly; their degree of quenching is mutually comparable. 1.0 LIF signal (arb.u.) 0.8 0.6 0.4 phenol phenol-resorcinol phenol-pyrogallol 0.2 0.0 0.0 0.2 0.4 0.6 0.8 1.0 -4 Phenol concentration (10 mol/l) Figure 4.52 The LIF peak intensity of phenol versus the concentration. The dotted line shows the absorption effect of the laser intensity. Also shown is the LIF intensity of several phenol-resorcinol and phenol-pyrogallol aqueous mixtures at a total concentration of 1.0⋅10-4 M. The LIF peak intensity of the 300 ml 1.0⋅10-4 M phenol solution is plotted versus the oxidation time by Figure 4.53. The solution fluorescence has been decomposed into the fluorescence of phenol and its oxidation products for the time range 0-150 minutes. With regard to this plot, no corrections have been performed for either laser absorption by the solution or quenching due to oxidation products. 1 LIF signal (arb.u.) Phenol Oxidation products 0.1 0.01 0 50 100 150 Time (min) Figure 4.53 The decomposed LIF peak intensity due to phenol and oxidation products, during oxidation of a 300 ml 1.0⋅10-4 M phenol solution. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. 96 Chapter 4. It is observed, that the oxidation products reach a maximum concentration after about 20 minutes of oxidation. After this time these intermediate products are further oxidized. The intermediate products comply -next to phenol- with the requirements for fluorescence and are thus likely to be polyhydroxybenzenes. Application of the absorbance correction to the decomposed phenol LIF intensity versus time course according to Equation 2.22, yields the absolute phenol concentration, see Figure 4.54. The error bars mark the deviation in the concentration due to quenching effects. Characteristic decay time constants τ apply for three intervals, viz. τ0-50min=19 minutes, τ50-120min=25 minutes and τ120-150min=100 minutes. These different decay rates originate from the production and consumption of intermediate polyhydroxybenzenes; after disappearance of phenol non-fluorescent saturated carboxylic acids remain that eventually mineralize. -4 Phenol concentration (10 mol/l) 1.0 0.8 0.6 0.4 0.2 0.0 0 20 40 60 80 100 120 140 160 Time (min) Figure 4.54 The absolute phenol concentration versus time during oxidation of a 300 ml 1.0⋅10-4 M phenol solution. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Figure 4.55 shows the LIF spectra of 1.0⋅10-4 M solutions of the dihydroxybenzenes and trihydroxybenzenes. Also included is the quinone 1,4-benzoquinone. The LIF intensity decreases by the following order: phenol > hydroquinone > resorcinol > catechol > hydroxyhydroquinone > pyrogallol > 1,4-benzoquinone, phloroglucinol. The trihydroxybenzenes are much weaker fluorescent than the dihydroxybenzenes and their maximum fluorescence intensity is at about 350 nm. Especially hydroquinone and hydroxyhydroquinone seem to account for the observed increased LIF intensity at higher wavelengths. Results 97 1200 30000 Hydroquinone Hydroxyhydroquinone Resorcinol Pyrogallol LIF signal (arb.u.) LIF signal (arb.u.) 40000 Catechol 20000 1,4-Benzoquinone 10000 0 250 300 350 400 450 800 400 0 250 500 Wavelength (nm) Figure 4.55 Phloroglucinol 300 350 400 450 500 Wavelength (nm) LIF spectra of 1.0⋅10 M dihydroxybenzene (left) and trihydroxybenzene (right) solutions. -4 air t=40 min 1 LIF signal (arb.u.) LIF signal (arb.u.) Figure 4.56 shows LIF spectra due to the phenol oxidation products, recorded after 40 minutes and 100 minutes of exposure time for corona in air and argon. These spectra have been derived by subtraction of the decomposed phenol spectrum from the observed solution LIF spectrum. Ar t=40 min 1 Ar t=100 min air t=100 min 0 250 300 350 400 Wavelength (nm) Figure 4.56 450 500 0 250 300 350 400 Wavelength (nm) 450 500 LIF spectra due to phenol degradation products, recorded after 40 min and 100 min, during degradation of 1.0⋅10-4 M phenol solutions by pulsed corona discharges in an air (left) and argon (right) atmosphere. The corona parameters are V=25 kV, V=1 nF, f=100 Hz, d=1.0 cm. The LIF spectra of oxidized phenol solutions reveal fluorescence at higher wavelengths compared to the LIF spectrum of phenol. This is especially the case for phenol degraded by corona in argon. From the ICE measurements (section 4.2.1) it has been concluded, that the degradation of phenol by corona in argon produces considerably higher amounts of dihydroxybenzenes. According to Figure 4.55 the fluorescence at higher wavelengths is likely due to the dihydroxybenzene hydroquinone. The contribution of the trihydroxybenzenes to fluorescence is negligible. 98 Chapter 4. 300 ml 1.0⋅10-4 M polyhydroxybenzene solutions and a 1,4-benzoquinone solution have been oxidized by pulsed corona discharges in air. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Figure 4.57 shows the LIF peak intensity of the solution versus the oxidation time. Peak wavelengths are according to Figure 4.55. Although no spectrum decomposition has been performed here, the LIF intensity decay can be related to the disappearance of the hydroxybenzenes by oxidation. The oxidation stability of the DHB’s decreases by the order resorcinol > 1,4-benzoquinone, hydroquinone > catechol while the stability of the THB’s decreases according to the order phloroglucinol > pyrogallol > hydroxyhydroquinone. The mutual stability within the DHB and THB classes may be explained by the resonance structures and by ring strain due to sterical hindrance of hydroxyl groups. It is remarkable that the THB stability is rather similar to the DHB stability, because THB’s are much stronger reducing agents than DHB’s [61]. This contradiction may arise from the fact that the oxidation of the hydroxybenzenes in aqueous solution by ozone is mass transfer limited. 0 10 resorcinol catechol hydroquinone 1,4-benzoquinone -1 10 -2 10 0 30 60 Time (min) Figure 4.57 90 0 pyrogallol hydroxyhydroquinone phloroglucinol LIF signal (arb.u.) LIF signal (arb.u.) 10 120 10 -1 10 -2 0 30 60 90 120 Time (min) The solution LIF peak intensity versus time during the oxidation of 300 ml 1.0⋅10-4 M dihydroxybenzene (left) and trihydroxybenzene (right) solutions. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Summary Although LIF spectroscopy is generally applied for gas phase studies, it has been demonstrated that it is also applicable to in-situ conversion measurements of phenol in aqueous solution at concentrations up to 1.0⋅10-4 M. For that case, the phenol LIF peak intensity has been obtained from the solution LIF peak intensity by spectrum decomposition, as a correction with regard to quenching effects. The LIF spectrum due to the phenol oxidation products indicates the presence of polyhydroxybenzenes, which is consistent with the results from ICE measurements. The polyhydroxybenzenes are less resistant to oxidation than phenol, but the decrease of the LIF intensity of DHB’s and THB’s has appeared to be comparable, although these compounds differ in reducing strength. This may be due to mass transfer limitation of the oxidation reaction by ozone. Results 99 Infrared spectroscopy corona in air corona in argon 0.06 0.04 0.02 Absorbance (arb.u.) In addition to the aldehyde screening test, the gas phase over oxidized phenol solutions has also been analyzed by Fourier-transform infrared spectroscopy (FTIR). 250 ml 1.0 mM phenol solutions have been oxidized by pulsed corona discharges in both an air and argon gas phase, see reactor configuration 9. Figure 4.58 shows the FTIR spectra recorded after three hours of oxidation. 0 3000 2500 2000 -1 Wavenumber (cm ) 1500 Figure 4.58 FTIR spectra of the gas phase over 250 ml 1.0 mM phenol solutions, oxidized by pulsed corona discharges in air or argon during 3 hours. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The strong absorption at 2349 cm-1 is due to carbon dioxide (anti-symmetric stretch vibration). Carbon monoxide (2143 cm-1) has not been detected. The very weak absorption at 2224 cm-1, observed in the spectrum from corona in air, may originate from traces of nitrous oxide (N2O). Within the scanned wavenumber range, no other compounds have been identified. Table 4.11 shows carbon dioxide concentrations. Table 4.11 Time (min) 0 30 60 120 180 The production of carbon dioxide during oxidation of 250 ml 1.0 mM phenol solutions by pulsed corona discharges in air or argon. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. [CO2] (ppm v/v) air argon 330 0 160 590 230 330 1100 410 The mineralization of phenol by corona in air is faster than the mineralization of phenol by corona in argon. 100 Chapter 4. For the case of air the oxidizers ozone, hydroxyl radicals and oxygen can produce carbon oxides from phenol. In argon, oxidizers can only be produced from the dissociation of water by argon ions and from trace amounts of oxygen dissolved in the water. Although carbon monoxide has not been detected, the production of this compound is possible from the decomposition of endoperoxides [19] produced by phenol oxidation, see Figure 4.59. However carbon monoxide is oxidized to carbon dioxide by ozone or by hydroxyl radicals and oxygen, see Equations 4.2a-d. OH HO O2 HO O H O O H O OH phenol Figure 4.59 OH H H O + CO OH endoperoxide The production of carbon monoxide from the decomposition of an endoperoxide produced during the oxidation of phenol by hydroxyl radicals and oxygen. CO + O3 → CO2 + O2 CO + HO• → HCO2• HCO2• + O2 → CO2 + HO2• HCO2• + HO• → CO2 + H2O (4.2a) (4.2b) (4.2c) (4.2d) The production of carbon dioxide has also been measured during the oxidation of 250 ml 0.1 mM phenol solutions, under the same conditions. The observed carbon dioxide concentrations are shown by Table 4.12 and Figure 4.60. Table 4.12 Time (min) 0 30 60 90 120 180 The production of carbon dioxide during oxidation of 250 ml 0.1 mM phenol solutions by pulsed corona discharges in air or argon. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm [CO2] (ppm v/v) air argon 330 18 467 132 591 173 687 240 787 287 922 350 It is clear, that the carbon dioxide production depends on the initial phenol concentration. The concentration increase versus the oxidation time is non-linear. Results 101 [CO2] (ppm v/v) 1000 800 corona in air 600 400 200 corona in argon 0 0 30 60 90 120 150 180 Time (min) Figure 4.60 The production of carbon dioxide during oxidation of 250 ml 0.1 mM phenol solutions by pulsed corona discharges in air or argon. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The theoretical amount of carbon dioxide that can be produced from phenol is calculated according to Equation 4.3. C6H5OH + 7O2 → 6CO2 + 3H2O (4.3) 1 mol phenol can produce a maximum amount of 6 moles carbon dioxide. The molar gas volume of carbon dioxide is about Vm=24.3 liter at p=p0 and T=293.15 K [111]. The 250 ml 1.0 mM phenol solution contains 2.5⋅10-4 mol phenol that corresponds to 36.5 ml carbon dioxide, while the 0.1 mM phenol solution can produce only one tenth of this amount. The gas phase volume is also equal to 250 ml. The percentage of carbon converted by oxidation of both 0.1 mM and 1 mM phenol solutions during 180 minutes in air and argon is shown by Table 4.13. Table 4.13 Percentage of carbon converted into carbon dioxide during oxidation of 250 ml 0.1 mM and 1.0 mM phenol solutions by corona in air or argon. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. t=180 min air argon 0.1 mM 922 / 14600 ≡ 6.3 % 350 / 14600 ≡ 2.4 % 1.0 mM 1100 / 146000 ≡ 0.8 % 410 / 146000 ≡ 0.3 % However, if the solubility of carbon dioxide in water [44] is taken into account viz. 7.07⋅10-4 mol CO2/mol H2O ≡ 3.92⋅10-2 mol/l ≡ 1.72 g/l ≡ 952.0 ml/l at T=293.15 K and p=p0, the liquid phase may also contain large amounts of dissolved carbon dioxide. Then the percentage of carbon dioxide converted is much higher. The 1.0 mM phenol solution is clearly less mineralized than the 0.1 mM solution. The converted carbon ratio air-to-argon is about 2.7 for both phenol concentrations. Summary Only CO2 has been detected in the gas phase over oxidized phenol solutions. The amounts produced by corona in argon are distinctly lower than the amounts produced by corona in air. The observed CO2 levels possibly imply a low level of mineralization. 102 Chapter 4. 4.2.4. Electrical conductometry Electrical conductometry has been applied in order to monitor oxidation progress of phenol and some of its intermediate oxidation products, by the increase of solution conductivity due to the production of carboxylic acids. The electrical conductivity of the hydroxybenzene solutions has been determined as a function of the oxidation time. Simultaneously, the hydroxybenzene conversion has been determined by ion-exclusion chromatography, see LC configuration 8. Experiments have been performed using 250 ml 1 mM aqueous hydroxybenzene solutions, see reactor configuration 5b. Tested compounds are phenol (PHE), catechol (CAT), hydroquinone (HQ), pyrogallol (PG) and hydroxyhydroquinone (HHQ). The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm and the maximum oxidation time is 40 minutes. A single tungsten anode tip has been used. The following relative temperature increase (∆T/T0) values have been measured after 40 minutes of corona exposure: PHE: 0.10, CAT: 0.08, HYD: 0.09, PYR: 0.10, HHQ: 0.11. As reported in section 4.1.4, the solution conductivity also increases by the formation of nitrate ions, due to the application of corona discharges in air. Therefore the conductivity measurements on oxidized hydroxybenzene solutions have been reported with regard to the following background: the conductivity of deionized water as a function of the oxidation time, under the same conditions the hydroxybenzenes have been oxidized. The conductivity measurements as a function of the oxidation time are shown by Figure 4.61. Conductivity (µS/cm) 500 PHE CAT HYD PYR HHQ 400 300 200 100 0 0 10 20 30 40 Time (min) Figure 4.61 The conductivity of 250 ml 1.0 mM hydroxybenzene solutions versus the oxidation time. Tested hydroxybenzenes are phenol (PHE), catechol (CAT), hydroquinone (HYD), pyrogallol (PYR) and hydroxyhydroquinone (HHQ). The conductivity is reported with regard to the conductivity of deionized water oxidized under the same conditions. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Results 103 It has been observed that the conductivity increases with the oxidation time for all tested compounds. Also, the conductivity at a fixed oxidation time increases according to the order phenol, catechol/hydroquinone, pyrogallol, hydroxyhydroquinone. This order resembles the increase in reducing properties of the regarded hydroxybenzenes. Hydroxyhydroquinone is the strongest reducing agent of the tested hydroxybenzenes and the oxidation of this compound i.e. production of carboxylic acids will thus be fastest. The untreated hydroxyhydroquinone solution already shows conductivity of about 35 µS/cm, which is caused by oxidation of this compound by oxygen dissolved in the water. The conversion of the hydroxybenzenes is shown by Table 4.14. Table 4.14 Time (min) 0 10 20 30 40 Conversion X of 250 ml 1.0 mM hydroxybenzene solutions. A tungsten single-pin anode has been used. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. XPHE (%) 0.0 1.7 3.3 5.0 49.3 XCAT (%) 0.0 3.0 6.1 9.4 12.2 XHYD (%) 0.0 3.1 6.2 9.2 12.4 Unfortunately the conversion determination of hydroxyhydroquinone and pyrogallol failed. The phenol conversion at 40 minutes is unlikely. The pulse energy has been rather constant during all oxidation experiments, so the oxidation conditions are comparable. Average pulse energies with 95% confidence interval are as follows: for phenol oxidation 5.1 ±0.3 mJ, for catechol oxidation 5.5 ±0.0 mJ and for hydroquinone oxidation 6.1 ±0.1 mJ. Figure 4.62 shows the hydroxybenzene conversion versus the solution conductivity observed during oxidation. 15 Conversion (%) PHE CAT 10 HYD 5 0 0 Figure 4.62 100 200 Conductivity (µS/cm) 300 The hydroxybenzene conversion versus the solution conductivity. PHE=phenol, CAT=catechol and HYD=hydroquinone. The conductivity is reported with regard to the conductivity of deionized water oxidized under the same conditions. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. 104 Chapter 4. There appears to be a relationship between conversion and conductivity. The monohydroxybenzene phenol shows a different course than the dihydroxybenzene isomers catechol or hydroquinone with mutually comparable course. Theorizing, because of the missing conversion data, the slope of the trihydroxybenzene conversion versus conductivity relationship will be much smaller than the slope of the dihydroxybenzenes conversion course. The relationship is applicable for a component-specific oxidation progress range where the conversion increases with conductivity, knowing the initial conductivity. Due to mineralization, the carboxylic acids will eventually disappear thus the conductivity will decrease from that time. Also it should be noted, that after 100 % conversion of the target compound, still carboxylic acids may be present that contribute to conductivity. Finally the conversion efficiency values are reported for phenol, catechol and hydroquinone by the G yield, see Table 4.15. Table 4.15 The conversion efficiency of phenol, catechol and hydroquinone; average values and 95% confidence intervals are presented. Compound Phenol Catechol Hydroquinone G mol/J ⋅108 1.4 ±0.07 2.3 ±0.07 2.2 ±0.06 G (100 eV)-1 0.13 ±0.01 0.22 ±0.01 0.21 ±0.01 G (g/kWh) 4.6 ±0.3 9.1 ±0.3 8.6 ±0.2 From these data it is confirmed, that the dihydroxybenzenes catechol and hydroquinone are stronger reducing agents than phenol. The observed efficiency values are generally less favourable than the efficiency values observed in section 4.2.1; this may be explained by application of a single pin anode. Summary Electrical conductometry has been tested as an alternative conversion measurement. There appears to be a relationship between solution conductivity and target compound conversion for a certain oxidation progress range. This relationship is based on the formation of carboxylic acids providing conductivity by deprotonation in aqueous solution. Results 105 4.2.5. Microtox ecotoxicity The degree of detoxification during oxidation of phenol by pulsed corona discharges has been investigated by Microtox ecotoxicity tests. The ecotoxicity is expressed as an effect concentration EC20, defining the concentration at which 20% inhibitory effect takes place. The EC20 value is determined from a concentration series by interpolation of the dose-effect relationship, see Equation 2.8. In order to quantify the ecotoxicity of phenol during oxidation, an EC20 value has been determined before and after oxidation of a series of different phenol concentrations. 250 ml solutions at 0.02 mM, 0.05 mM, 0.1 mM, 0.2 mM and 0.4 mM initial phenol concentration [PHE]0 have been oxidized for 30 minutes, see reactor configuration 5c. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Table 4.16 shows the phenol conversion X and absolute converted phenol amount mX, see LC configuration 9. Table 4.16 [PHE]0 (mM) 0.02 0.05 0.10 0.20 0.40 Phenol conversion (X) and absolute converted amount (mX) after 30 minutes of oxidation of 250 ml solutions. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. X (%) 64.6 72.8 53.0 20.2 25.3 mX (mg) 1.4 3.5 5.0 3.8 9.6 The untreated phenol solutions show a dose-effect relationship according to Equation 2.8. The EC20 values, dose-effect parameters {b,a} and correlation coefficient r2 are shown by Table 4.17. As the EC20 values do not change with the Vibrio fischeri exposure time tVF, the ecotoxicity effect is acute. Table 4.17 tVF (min) 5 15 30 EC20 values of phenol in deionized water versus the Vibrio fischeri exposure time. Also shown are the dose effect parameters {b,a} and correlation coefficient r2. EC20 (mM) 0.06 0.07 0.07 b (-) 0.70 0.90 0.91 a (-) -0.79 -0.62 -0.62 r2 (-) 0.9999 0.9825 0.9784 On the contrary, no EC20 value can be determined for the oxidized phenol solutions, because there appears to be no dose-effect relationship. All samples except for the 0.1 mM sample show very low bioluminescence intensity, implying high ecotoxicity of these oxidized phenol solutions. Table 4.18 shows the observed effects, indicated by the gamma value, for the different phenol solutions and Vibrio fischeri exposure times. It should be noted that the effect increases with the exposure time tVF. This means that the ecotoxicity effect of the oxidized samples is not acute, which may be explained by mass transfer-limited diffusion of the oxidation products within the Vibrio fischeri bacterium. 106 Table 4.18 [PHE]0 (mM) 0.02 0.05 0.10 0.20 0.40 Chapter 4. Ecotoxicity effects of oxidized phenol solutions, indicated by the gamma value, for different Vibrio fischeri exposure times; average values (AVG) and standard deviations (σn-1) are presented. 250 ml solutions have been oxidized for 30 minutes. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Γ 5 min (-) AVG σn-1 24.9 13.1 85.6 59.6 1.1 0.0 6.2 0.1 30.9 8.0 Γ 15 min (-) AVG σn-1 31.3 16.7 105.2 57.0 1.4 0.0 15.9 0.1 52.5 20.3 Γ 30 min (-) AVG σn-1 49.3 29.9 151.4 91.2 2.1 0.1 51.0 1.8 94.5 50.3 An explanation for the absence of the dose-effect relationship after oxidation is the fact, that all oxidized samples contain a complex range of oxidation products. Especially hydroquinone and 1,4-benzoquinone are highly toxic towards Vibrio fischeri, see Table 2.1. By oxidation progress, these compounds disappear but the oxidation product mixture increases in acidity which is also unfavourable to these bacteria. As part of the Microtox test, the solution acidity and oxygen content have been measured for both untreated and oxidized samples, see Table 4.19. If the sample acidity exceeds the range 6<pH<8.5 and/or the sample has a low oxygen content (<0.5 mg/l), these effects are inhibitory and bias the effect of the target compound. It appears, that oxidation increases the acidity of both the Millipore water and the phenol solutions. For the case of Millipore water, the production of nitrogen oxides by corona discharges in air results in the formation of nitric acid. The acidity increase by oxidation of phenol is caused by the formation of carboxylic acids. Table 4.19 [PHE]0 (mM) 0 0.02 0.05 0.10 0.20 0.40 Solution acidity and oxygen content (expressed as percentage of the saturation concentration) for untreated and oxidized samples. 250 ml solutions have been oxidized for 30 minutes. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Untreated pH (-) [O2] (%) 5.7 65 5.9 63 5.9 62 6.2 61 6.2 51 6.2 60 Oxidized pH (-) [O2] (%) 4.2 71 4.0 60 4.0 61 3.9 48 3.8 65 3.8 51 The small differences in oxygen content between the oxidized and untreated phenol solutions, if significant, might be caused by different amounts of polyhydroxybenzenes produced during the oxidation. These compounds are strong reducing agents and thus have a high affinity for oxygen. The oxygen content of oxidized Millipore water is somewhat higher than the oxygen content of untreated Millipore water. This might be due to the presence of hydrogen peroxide. Results 107 The influence of the solution acidity on the Microtox test has been verified by determination of the gamma values of non-pH adjusted oxidized phenol solutions at an exposure time tVF =5 minutes. The acidity of these solutions is about pH=4, see Table 4.19. The gamma values of oxidized phenol solutions that have not been pH-adjusted, are shown by Table 4.20. Table 4.20 Gamma values of non-pH adjusted oxidized phenol solutions, after 5 minutes exposure time; average values (AVG) and standard deviations (σn-1) are presented. 250 ml solutions have been oxidized for 30 minutes. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Γ 5 min [PHE]0 (mM) 0.02 0.05 0.10 0.20 0.40 AVG 13.7 1.4 5.9 395.6 193.1 σn-1 6.1 0.1 1.2 127.5 25.5 The gamma values are highest for the 0.2 mM and 0.4 mM samples, which are more acidic than the other samples. Also, these gamma values are very different from the gamma values of pH-adjusted oxidized solutions, illustrated by Table 4.18. The pH effect is thus very relevant. Finally, it has been observed, that the oxidation of Millipore water also induces ecotoxicity. The bioluminescence intensity at tVF=5 minutes of fresh Millipore water is It5=100.8 / σn-1=0.9. The bioluminescence intensity of pH-adjusted oxidized Millipore water is It5=100.1 / σn-1=2.5. However, if oxidized Millipore water is not pH-adjusted, the bioluminescence intensity is It5 = 67.8 / σn-1=5.2. This effect is also caused by acidity. The corona discharges produce nitrogen oxides in air, which dissolve in the water and form nitric acid. By adjusting the acidity of the solution the ecotoxicity vanishes, so the effect is not caused by the nitrate ion. Although hydrogen peroxide is produced by the corona discharges and hydrogen peroxide exhibits ecotoxicity (EC5015min=16 mg/l) [78], this effect has not been observed: the bioluminescence intensities of untreated Millipore water and pH-adjusted oxidized Millipore water are equal. Hydrogen peroxide at low concentrations decomposes rapidly into oxygen and water. Summary After oxidation during a fixed time (30 min) of a series of different concentration phenol solutions, no dose-effect relationship appears to exist anymore. Although mutually different conversions have been measured, the Microtox ecotoxicity effect is high for both low and high initial phenol concentration. This effect is especially caused by hydroquinone and 1,4-benzoquinone, that are highly toxic to Vibrio fischeri bacteria. Although these compounds disappear by oxidation progress, carboxylic acids are produced that cause pH-toxicity. 108 Chapter 4. 4.2.6. Total organic carbon During the oxidation of a 498 ml 1.0 mM phenol solution, see reactor configuration 10, the total organic carbon level (TOC) has been measured every 15 minutes during 2 hours. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. In addition, the conversion has been measured by ICE chromatography, according to LC configuration 7. Table 4.21 shows the TOC and conversion (X) values versus the oxidation time. X2 is a repeated conversion measurement, 21 hours after the first measurement X1. Table 4.21 t(min) 0 15 30 45 60 75 90 105 120 Total organic carbon (TOC) level and conversion (X) during the oxidation of a 498 ml 1.0 mM phenol solution. The time span between the two conversion measurements is 21 hours. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. TOC (mg/l) 52.0 58.4 62.8 57.2 59.2 53.6 67.6 62.4 56.8 X1 (%) 0.0 3.3 7.0 10.8 13.9 17.0 19.9 21.3 25.2 X2 (%) 0.0 3.2 7.1 11.0 14.4 17.2 20.3 22.3 25.7 It should be remarked, that the theoretical TOC level of an 1.0 mM phenol solution equals 72.1 mg/l. The measured value at t=0 however is somewhat lower. This may be due to an incidental sample preparation error, because the samples had to be diluted before analysis. The TOC level appears to be rather constant during oxidation, while the conversion increases. This means, that the oxidation products are likely to remain in the liquid phase for the observed conversion range. This is in accordance with the results from the infrared analyses of the gas phase over oxidized phenol solutions (section 4.2.3) and the aldehyde screening test (section 4.2.1). From the conversion measurements (X1 and X2) it is clear, that after stopping of the oxidation the conversion of phenol does not significantly proceed within 21 hours. Possible conversion progress after stopping the corona discharges might be due to remaining amounts of ozone and hydrogen peroxide. In between these two conversion measurements the TOC measurements have been performed. Summary At a conversion up to 25%, the TOC level of an oxidized 1 mM phenol solution appears to be rather constant. This implies that the oxidation products are likely to remain in the liquid phase for the observed conversion range. Results 109 4.3. Oxidation of other model compounds In this section a global survey is given of the oxidation of atrazine herbicide, malachite green dye and dimethyl sulfide odor component. The conversion of atrazine in aqueous solution has been measured by reversed-phase HPLC, while the oxidation progress of malachite green in aqueous solution has been determined by in-situ absorption spectrometry. Dimethyl sulfide has been oxidized in the gas phase and the conversion has been measured by gas chromatography. 4.3.1. Atrazine A 500 ml 25 mg/l (0.12 mM) atrazine solution (tap water-based) has been oxidized for 5 hours, see reactor configuration 6b. The corona parameters are V=30 kV, C=100 pF, f=50 Hz, d=1.0 cm. Every 15 minutes (0-3 hours) or 30 minutes (3-5 hours) the conversion has been determined in threefold using a rp-HPLC column according to LC configuration 10. Figure 4.63 shows the conversion versus the oxidation time. The conversion efficiency versus time is shown by Table 4.22. Conversion (%) 100 80 60 40 20 0 0 60 120 180 240 300 Time (min) Figure 4.63 The conversion of a 500 ml 25 mg/l (0.12 mM) atrazine solution (tap water-based). The corona parameters are V=30 kV, C=100 pF, f=50 Hz, d=1.0 cm. Table 4.22 Conversion (X), pulse energy (Ep) and efficiency (G) during oxidation of a 500 ml 25 mg/l (0.12 mM) atrazine solution (tap water-based); average values and 95% confidence levels are presented. The corona parameters are V=30 kV, C=100 pF, f=50 Hz, d=1.0 cm. t (min) 60 120 180 240 300 X (%) 12.7 ±5.5 27.2 ±15.3 38.5 ±21.0 47.8 ±28.8 56.5 ±32.5 Ep (mJ) 10.8 ±2.1 10.7 ±0.5 10.4 ±1.2 11.2 ±1.9 10.2 ±0.6 G (mol/J)⋅1010 7.67 ±5.14 8.22 ±5.17 7.89 ±3.29 6.83 ±1.93 7.21 ±4.86 G (100eV)-1⋅103 7.40 ±4.96 7.93 ±4.99 7.62 ±3.18 6.59 ±1.86 6.96 ±4.69 G (g/kWh) 0.60 ±0.40 0.64 ±0.40 0.61 ±0.26 0.53 ±0.15 0.56 ±0.38 110 Chapter 4. As a function of the oxidation time, the error bars grow to considerable size. The deviation increase cannot be due to sample inhomogenity because the reactor content is continuously mixed by a magnetic stirring bar. The time dependent effect may be a related to a UV absorbance detector stability problem. The conversion time relationship seems to obey first order kinetics with a rate constant k1≈2.7⋅10-3 min-1. After 5 hours of oxidation the atrazine conversion is about 57%. The efficiency values clearly show that atrazine is a very stable compound. At least seven oxidation products have been found in the chromatogram, but the identification of these products has not been possible. Pelizzetti [36] has concluded from detailed GC-MS experiments, that photocatalytic oxidation of aqueous atrazine will not destroy the triazine ring and that the final oxidation product is cyanuric acid. Cyanuric acid (2,4,6trihydroxy-1,3,5-triazine) is a questionable carcinogenic compound [112]. Summary Atrazine is a rather stable herbicide. The conversion efficiency is about one order of magnitude lower than the conversion efficiency of phenol. It has not been possible, to identify the variety of oxidation products. 4.3.2. Malachite green The influence of different electrode configurations on the decolorization of malachite green dye has been determined by absorption spectrometry. From a LED light source, a photodiode detector, an amplifier and a recorder a simple setup has been constructed that may be a reasonable alternative to a precious and laboratorybound liquid chromatograph or absorption spectrometer. 125 ml 1.94 mg/l malachite green solutions have been exposed to pulsed corona discharges during 1.5 hours, see reactor configuration 11. The load voltage has been set to V=20-30 kV, depending on the configuration. The time span t24 has been determined, after which 24% decrease in absorption at wavelength λ=590 nm has been measured. This percentage is based on the configuration which showed slowest decolorization. The configuration variables are as follows: the anode has been located at a distance d=-4 mm, +1 mm, +5 mm, +10 mm relative to the solution surface; the used anode types are single pin, 4 pins, 8 pins and a wire anode; in one case an air flow has been supplied at a single pin anode tip (configuration no. VII); the cathode location is in the water on the bottom of the reactor or directly outside/underneath the reactor. Table 4.23 shows the power input (P), decolorization time span (t24) and efficiency (G). It can be observed that capacitive configurations, where the cathode plate is situated outside/underneath the glass reactor vessel, are generally more favourable to the efficiency than configurations where the cathode is inside the aqueous phase. This is caused by the fact that for the case of a capacitive configuration, conductive current is not possible. Results 111 Table 4.23 Conf. nr. I II III IV V VI VII VIII IX Power input (P), decolorization time span (t24) and efficiency (G) for the oxidation of malachite green in aqueous solution. The corona parameters are V (configuration dependent), C=5 pF, f=150 Hz, d (indicated). Cathode Anode position type d (mm) outside 4 pins +1 outside 8 pins +1 outside wire +1 outside 1 pin +1 outside 1 pin -4 inside 1 pin +1 inside 1 pin+air -4 inside 1 pin +10 inside 1 pin +5 P (mW) 395 366 493 383 432 1789 6944 3963 6301 t24 (min) 12.1 15.9 32.2 57.9 94.0 26.4 35.0 61.9 66.7 G (mg/kWh) 412 337 123 89 48 67 16 13 7 Immersion of the anode in the aqueous solution is less favourable than situating the anode in the gas phase over the solution. The production of corona discharges at an anode tip immersed in an aqueous phase, demands extra energy for evaporation of the liquid, which is less efficient than the formation of the corona discharge in the gas phase. A multipin anode is more efficient than a single pin anode. The multipin anode seems to produce a more efficiently-dimensioned reactive volume (plasma) for the production of oxidizers than a single pin, while the energy input differences are comparable. The anode wire efficiency is in between the performances of multipin and single pin; although it spans a favourable space, it lacks the pin shape necessary for the creation of high electric field strengths. Although the addition of air to the discharge is likely to favour the production of ozone by provision of oxygen, configuration VII has appeared to show low performance. In this way, large amounts of oxygen-based ions may have been produced, that have caused high conduction currents within this noncapacitive configuration. It should be noted, that the decolorization of malachite green does not imply that detoxification takes place. Decolorization is caused by the destruction of chromophoric groups; at that time, mineralization of the large dye molecule is certainly not obvious. The observed decolorization efficiency is about one to two orders of magnitude lower than the conversion efficiency of phenol. The degradation of malachite green is complex, because the number of probable attack sites for oxidizers like the hydroxyl radical, ozone and oxygen is very large. The methyne central carbon is prone to attack by the hydroxyl radical or ozone yielding bond cleavage, thus removal of the main chromophoric part i.e. C=C6H4=N+(CH3)2 from the dye molecule. Also, the aromatic rings can be hydroxylated or cleaved. In addition to the oxidation products mentioned in section 2.4.3, the following products may be formed: nitro/hydroxybenzophenones, nitrophenols, benzoic acid derivatives, aliphatic aldehydes/carboxylic acids and nitrate ions. Summary A multipin anode over the aqueous phase, combined with a cathode directly outside and underneath the reactor is the most favourable configuration for the decolorization of malachite green. 112 Chapter 4. 4.3.3. Dimethyl sulfide This experiment differs from the others, because the target compound dimethyl sulfide (DMS) has been oxidized in the gas phase using a concentric electrode tubular reactor according to reactor configuration 12. A gas chromatograph equipped with sulfur chemiluminescence detector has been applied for analysis, see GC configuration 2. 10 ppm and 100 ppm standard solutions of DMS in methanol have been prepared. The corona parameters are V=30 kV, C=100 pF, f=50 Hz. Before every corona experiment, the reactor contains about 2.5 ppm DMS in ambient air. Three oxidation times have been investigated. Every experiment has been performed in threefold. Table 4.24 shows the results. Table 4.24 t (s) 15 30 60 Conversion (X) of dimethyl sulfide in air versus the oxidation time. Also shown are initial and final average concentrations [DMS] and 95% confidence intervals. The corona parameters are V=30 kV, C=100 pF, f=50 Hz. [DMS]0 (ppm) 2.33 ±0.60 2.59 ±1.05 2.54 ±0.22 [DMS]t (ppm) 1.28 ±0.25 0.77 ±0.25 0.14 ±0.03 X (%) 45 70 94 Next to the oxidation products mentioned in section 2.4.4, oxidative demethylation likely yields formaldehyde and formic acid. The aim of DMS removal from industrial waste gas flows is odor destruction. Although destruction of dimethyl sulfide has been demonstrated, thorough oxidation progress is necessary, because the oxidation products are still harmful. Dimethyl sulfoxide is an odourless, strongly hygroscopic liquid that is readily absorbed by the skin; it causes especially irritation of the mucous membranes; dimethyl sulfoxide is reported to exhibit mutagenic and teratogenic properties [113]. Formaldehyde is strongly irritating and has mutagenic, teratogenic and probably carcinogenic properties [1]. Formaldehyde is oxidized to formic acid and formic acid will eventually disappear by mineralization, but sulfuric acid has to be removed by a gas scrubber. Summary 2.5 ppm Dimethyl sulfide in 330 ml ambient air has been converted for about 94 % after 60 seconds using gas phase corona discharges at V=30 kV and f=50 Hz. The destruction of this organosulfur compound initially yields harmful intermediates; although qualitative product analysis has not been performed, likely final oxidation products are sulfuric acid, carbon dioxide and water. 5. Discussion A survey of the results is presented in combination with points of special interest and new approaches. The discussed topics are general pulsed corona discharge application, corona-induced phenol oxidation, the oxidation mechanism of phenol, a presentation of the applied analysis techniques and a comparison of some advanced oxidation processes. 5.1. Pulsed corona discharges Oxidation by corona in air For the case of corona in air, possibly nitrate ions have been identified in the oxidation product mixture, see section 4.1.4. The amounts of nitrogen oxides produced by corona in air are small viz. a few ppm’s [114]. Nitrate seems to exhibit no ecotoxicity, according to the Microtox tests described in section 4.2.5 on corona exposed deionized water; the observed toxicity appearing from corona-exposure of deionized water is due to H3O+ from nitric acid. The question arises, whether nitrogen might be introduced into the oxidation products by means of nitro groups, resulting in a very significant increase of toxicity. This should be accomplished by the attack of radical species like nitrogen oxide radicals, produced by oxidation of nitrite or nitrate ions. However, the electron affinities of NO2 and NO3 are very high viz. 2.27 eV and 3.94 eV respectively [115], therefore the production of these radicals is highly unlikely. In addition, nitrite ions are oxidized to nitrate ions by oxidizers like the hydroxyl radical and ozone. NO has a low electron affinity (0.03 eV), but will also be oxidized to nitrate and exhibits poor solubility in water. Compared to corona in oxygen, the application of corona in air, thus has no extra negative impact on the toxicity of the oxidation products. The production of nitrate can be avoided by application of corona in an oxygen atmosphere, of course for the case of nitrogen-free target compounds. Liquid/gas phase corona oxidation products By application of corona in aqueous solution or in the gas phase (air, oxygen) over the solution it is reasonable to argue, whether the way of application may give rise to oxidation product mixtures with different chemical composition. For the case of aqueous phase corona, the discharge channels in the solution will locally create transient extreme temperatures and pressures; in addition to oxidation by hydroxyl radicals, the target compound may undergo pyrolytical decomposition, comparable to the supercritical conditions favoured by ultrasonic irradiation or wet oxidation. For the case of corona in air or oxygen, in addition to hydroxyl radicals and ions, the corona discharges produce ozone that diffuses into the liquid phase and oxidizes the target compound. Due to their high reactivity, hydroxyl radicals produced in the gas phase cannot reach the liquid, but may produce hydrogen peroxide over the solution. 114 Chapter 5. Hydroxylation in aqueous solution is explained by the reaction of ozone, water and UV photons or hydrogen peroxide and UV photons. At the gas-liquid interface, the discharge streamers produce limited amounts of reactive species by ions/metastables bombardment; pyrolytical decomposition of the target compound will not be relevant, due to the small contact area between the discharge channels and the target compound solution. Additives The conversion efficiency of pulsed corona discharges can likely be improved by the application of additives. Fe(II,III) salts invoke the decomposition of produced hydrogen peroxide into hydroxyl radicals according to Fenton chemistry, see Equations 1.3b-d. Extra hydrogen peroxide can be added, which is decomposed into hydroxyl radicals by UV photons produced by the corona. Application of corona in an oxygen atmosphere increases the production of ozone, atomic oxygen and oxygen ions. In alkaline solution, ozone reacts by hydroxyl radical chemistry and weakly acidic compounds (e.g. phenols) then exist in anionic form that is preferentially attacked by the electrophilic oxidizer species. Nitrous oxide addition, as applied in radiolysis, may also yield extra hydroxyl radicals for the case of corona treatment, according to Equation 1.6e. By turbulent mixing using a compressed gas, the gas-liquid interface area is increased, which is likely to be favourable to conversion. Corona-induced synthesis It may be suggested that pulsed corona discharges can also be utilized for the synthesis of organic compounds. Application of pulsed corona discharges in an argon atmosphere produces polyhydroxybenzenes in considerably higher amounts than corona in air, as has been described in section 4.2.1. Although unwanted ring-cleavage takes place, this may be suppressed by removal of remaining oxygen from the reactor by degassing or additional purging with inert gas and minimizing the corona energy input. Ring-cleavage also inevitably occurs during hydrogen peroxide-based commercial production of polyhydroxybenzenes from benzene or phenol. The advantages of pulsed corona discharges over ex-situ hydrogen peroxide addition are a fine-tunable hydroxyl radical dosage and the in-situ production of hydrogen peroxide which is very safe and convenient. By application of pulsed corona discharges to non-aqueous media under oxygen-free conditions, the medium is dissociated into radical species, while the absence of oxidizers prevents mineralization. For instance the exposure of a monomer to pulsed corona discharges under oxygen-free conditions is likely to invoke polymerization; also coupling reactions may be performed between different reactants. The advantages of corona-induced synthesis to conventional synthesis may be simplified synthesis pathways and reduced waste. Of course, radical formation can also be achieved by pulse radiolysis or electron-beam treatment. However these technologies are expensive, complex and demand shielding of radiation, while corona application is straightforward and comparatively inexpensive. Discussion 115 The synthesis of several organic compounds has been applied already for several years by electrochemical reactions viz. by electrolysis. Pathways include cathodic coupling (adiponitrile from acrylonitrile), cathodic hydrogenation (aniline from nitrobenzene) and anodic functionalization by hydroxylation (hydroquinone from benzene), halogenation (prefluorinated dialkyl ethers from dialkylethers) or oxygenation (dimethyl sulfoxide from dimethyl sulfide) [116]. Corona discharges versus electrolysis A major application of electrical energy for chemical purposes is electrolysis. The differences between electrolysis and corona are very distinct: Electrolysis deals with low voltage (a few volts) and high current densities: typical values for inorganic electrolysis are j=1-10 kA/m2 and for organic electrosynthesis j=0.1-1 kA/m2 [116]. Pulsed corona features high voltage and low time-averaged currents: typical values are a voltage of 10-100 kV and a time-averaged current of a few milliamperes in gaseous dielectrics. For the case of electrolysis both oxidation and reduction take place at the anode and cathode respectively; for the case of corona either oxidizing or reducing species can be produced, depending on the constitution of the dielectric medium where the discharges take place. Electrolysis is applied in molten salts, aqueous solution with electrolytes or for some cases in pure organic liquid phases; corona is applied either in the gas phase or in aqueous solution. The location where reactions occur is the electrode surface for the case of electrolysis; the corona discharge channel tips and -to a less extent- the discharge channels form a variable volume where reactive species are produced. For electrolysis, ion migration in solution and electron transfer at the electrodes take place; the application of pulsed corona discharges in the gas phase minimizes ion migration, electron transfer occurs from the streamer head by the discharge channel towards the electrode. Electrolysis is operated at DC voltage, while corona can be applied as either fast positive polarity pulses (gas and water remediation) or at AC voltage (ozonizers). An unconventional form of electrolysis is contact glow discharge electrolysis (CGDE) [117]. This phenomenon represents DC glow discharges across a gaseous envelope between anode or cathode and the surrounding aqueous, non-aqueous or molten phase. CDGE is initiated from normal electrolysis by increasing the voltage to a level, where a distinct current drop and growth of a gaseous sheath over an electrode are observed. The applied voltage is much higher than the voltage of normal electrolysis viz. some hundreds of volts, while the current is several milliamperes. By application of CGDE to aqueous solutions of e.g. ammonia [118] or ethanol [119], products are formed, which are explained by the dissociation and ionization of liquid phase molecules by electron and ion bombardment. As far as known, CGDE has not been qualified as AOP, probably because of the requirement of a high solution conductivity and an unfavourable conversion efficiency due to solvent evaporation by resistive heating. 116 Chapter 5. 5.2. Corona-induced phenol oxidation Ozone to phenol ratio An estimation has been made of the ratio R of ozone molecules consumed by phenol to the amount of converted phenol molecules. Data have been obtained from ozone measurements over 100 ml deionized water and over a 100 ml 1.0 mM phenol solution and a phenol oxidation experiment using a 100 ml 1.0 mM phenol solution. All experiments have been performed under equal conditions: V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. The amount of ozone molecules consumed by phenol is calculated from the difference of the ozone production efficiency for the case of 100 ml deionized water (GO2water) and the ozone production efficiency for the case of a 100 ml 1.0 mM phenol solution (GO21mM phenol) . The phenol conversion efficiency (Gphenol1mM) is known from Table 4.9. The ratio calculation is shown by Equation 5.1. Regarding this estimation, the assumption is made that phenol is exclusively converted by ozone. R= GO3 water mM phenol − GO3 1 Gphenol 1mM (5.1) The ozone production rate is determined from the initial slope (S) of the ozone density versus time plots given by Figure 4.11. The determined slopes are about Swater=1.54⋅1020 m-3s-1 and S1mM phenol=1.22⋅1020 m-3s-1. The ozone production efficiency is calculated from the initial slope S (m-3/min), the pulse energy (Epwater=3.5 mJ, Ep1mM phenol=5.6 mJ), the pulse repetition rate (f=100 Hz) and the gas phase volume (Volg=253 cm3). The ozone production efficiency for the case of 100 ml deionized water is about GO3water=1.12⋅1017 J-1≡1.85⋅10-7 mol/J≡1.79 (100eV)-1, while the efficiency regarding the 100 ml 1.0 mM phenol solution is about GO31mM phenol=5.50⋅1016 J-1≡9.13⋅10-8 mol/J≡ 0.88 (100 eV)-1. The amount of ozone molecules consumed by phenol equals ∆GO3= 5.67⋅1016 J-1≡9.42⋅10-8 mol/J≡0.91 (100 eV)-1. The initial phenol conversion efficiency is about Gphenol1mM=1.44⋅1016 J-1≡2.39⋅10-8 mol/J≡0.23 (100 eV)-1. From these values it has been calculated that about 4 ozone molecules are used to convert 1 phenol molecule: R=4. A theoretical stoichiometric ratio is impossible to determine. The reason for this is the complex phenol oxidation mechanism, as will be discussed in section 5.3. Hydroxyl radical production efficiency An estimation of the order of magnitude of the hydroxyl radical production efficiency can be made as follows. The initial slope of the phenol concentration versus time course indicates the rate at which phenol is converted at the start of corona oxidation. At that time, produced oxidizers are exclusively consumed by phenol and by recombination but not yet by oxidation products. Discussion 117 For the case of oxidation in argon, it may be assumed that phenol is mainly converted by hydroxylation. The exact stoichiometric ratio is unknown but ranges theoretically from OH:phenol=1-5, although ratios higher than 1 are unlikely. Then, including radical loss processes and polyfold OH attack on phenol, a lower limit estimation of the hydroxyl radical production rate equals 1/5th of the phenol decrease rate in argon at t=0. This production rate and the measured energy input, which is assumed to be completely used for radical production, yield an approximation of the hydroxyl radical production efficiency. The first order exponential fit of the phenol conversion versus time has been determined from Figure 4.31, while the initial phenol concentration C(0)=1.0 mM, see Equation 5.2. The pulse energy at time t=0 is about 10 mJ, see Table 4.9. Therefore at a pulse repetition rate f=100 Hz, 60 Joules are consumed every minute. The reactor volume is 100 ml. C(t) = C(0) exp(−k1 ⋅ t) k1 ≈ 1.8 ⋅ 10−2 min−1 dC(t) −5 −3 −1 dt = − k1 ⋅ C(0) ≈ − 1.8 ⋅ 10 mol dm min t =0 (5.2) Then, the lower limit of the hydroxyl radical production efficiency GOH is estimated to be 1/5x1.8⋅10-5 mol dm-3 min-1x0.1 dm3x1 min/60 J ≈ 6⋅10-9 mol OH/J ≡ 0.06 OH /100 eV. For the case of OH/phenol=1 stoichiometry, the efficiency is about 3⋅10-8 mol/J ≡ 0.29 OH/100 eV and is equal to about 345 eV/OH. By comparison of this value to the phenol conversion efficiency values listed by Tables 4.4, 4.5 and 4.9 and assuming that phenol exclusively reacts with hydroxyl radicals, the OH-to-phenol ratio is globally 1-2. Phenol conversion efficiency A comparison of the observed efficiency values for the conversion of phenol is given below. The relevant data are given by Tables 4.4, 4.5 and 4.9. It has appeared, that the efficiency of oxidation of a 500 ml 0.05 mM phenol solution at V=30 kV and f=50 Hz is considerably less favourable than the efficiency of oxidation of a 500 ml 1 mM phenol solution at V=25 kV and f=100 Hz. The removal of phenol from diluted aqueous solutions is probably less efficient, because here the highly reactive oxidizers have less probability to encounter a phenol molecule before they disappear by recombination or decomposition. Ozone may decompose into oxygen by wall recombination, by homogeneous catalysis by light and by nitrogen oxides produced from corona in air. In humid air, hydroxyl radicals react with nitrogen and oxygen to produce nitrogen oxides. Hydroxyl radicals oxidize water to hydrogen peroxide. Within this context it is also important to state, that the removal of target compounds is likely to be more efficient, when pulsed corona discharges are applied to thin films of the aqueous target solution. This has been demonstrated by the conversion measurements on two oxidized phenol solutions having equal initial amounts of phenol but a different solution volume and phenol concentration, see section 4.2.1. The influence of the gas-liquid interface area on the mass transfer of oxidizers may contribute to the differences between the efficiency values of Table 4.5 and 4.9. 118 Chapter 5. Phenol conversion efficiency in corona literature The phenol efficiency values, determined from this thesis, have been compared to some literature references on phenol oxidation by liquid phase corona and ozonation. The efficiency has been determined at 50% phenol conversion (G50). Table 5.1 shows the literature references, an overview of the applied conditions and the phenol conversion efficiency [104]. Table 5.1 A comparison of phenol conversion efficiency G50 values, obtained from liquid phase corona literature and from this thesis. Symbol description: C0=initial phenol concentration Vol=phenol solution volume, Ep=pulse energy, f=pulse repetition rate, t50=oxidation time to reach 50% phenol conversion. corona in solution corona in air over solution ozone References [100]1) [101] [102] [120] Thesis [45] C0 (mM) 0.03 0.53 0.53 0.02 0.05 1.0 1.0 0.5 Vol (cm3) 550 1000 2502) 500 500 100 500 4000 Ep (mJ) 1750 800 8802) 30 13 6 10 f (Hz) 60 50 48 50 50 100 100 t50 (min) 180 260 7 180 26 83 237 8 G50 (mol/J) 7.3·10-12 4.2·10-10 3.7·10-9 3.1·10-10 1.2⋅10-8 1.7⋅10-8 1.8·10-8 5.2·10-8 G50 (100eV)-1 7.0·10-5 4.1·10-3 3.6·10-2 3.0·10-3 1.2⋅10-1 1.6⋅10-1 1.7·10-1 5.0·10-1 1) 2) The energy consumption inside the power supply is included in the pulse energy. Additional information [121]. In this reference the yield is given as G = 1.0·10-2 (100 eV)-1. The phenol conversion efficiency by ozonation has been calculated using an ozone consumption of 61.5 mg/min x 8 minutes and assuming an ozone production efficiency of 100 g/kWh. It has been observed that the efficiency of phenol conversion by application of corona in air over the solution is one to two orders of magnitude higher than the efficiency of liquid phase corona. In references [100-102] it is mentioned, that the liquid phase is externally cooled during application of the corona discharges. The temperature increase of the aqueous solution due to exposure to gas phase corona, has been measured to be unsignificant, see section 4.1.4, 4.2.1 and 4.2.4. It is expected, that further optimization of the corona reactor will lead to a comparable or even higher efficiency than that of ozonation. First trial runs by corona in oxygen have already yielded higher phenol conversion efficiency values than ozonation [104]. This is explained by the broad range of reactive species that are produced by corona discharges in air viz. hydroxyl radicals, ozone, atomic oxygen, UV photons, ions and metastables. Discussion 119 LIF spectra of oxidized phenol solutions Oxidized phenol solutions, especially those exposed to corona in an argon atmosphere, show an increased fluorescence intensity at wavelengths in the range 400-500 nm, see Figure 4.56. This fluorescence is not due to phenol, see Figure 4.47. This fluorescence is partially explained by the presence of polyhydroxybenzenes. The strongest fluorescent polyhydroxybenzene is hydroquinone, which shows a fluorescence maximum at about 335 nm but does not show fluorescence at wavelengths higher than 450 nm. The trihydroxybenzene hydroxyhydroquinone shows a similar fluorescence course but at a much lower intensity. Other molecules that can explain for the fluorescence have to be in accordance with the requirements for fluorescence, see section 2.5.3. Muconic acid may likely be no candidate fluorescent product, although it has a conjugated system of π electrons (it is a diene) and the carbon chain is cyclic and flat, because the carbon chain is not linkedup. Regarding the fluorescence range a remaining possibility is, that synthetic humic acids account for this fluorescence behaviour. These polymeric molecules viz. contain hydroquinone monomeric units, see section 2.4.1. On the contrary the occurrence of these compounds in oxidized 1 mM phenol solutions is likely to be limited. Microtox toxicity units Although it has not been possible to determine the toxicity of the phenol oxidation product mixture by an effect concentration EC20 value, yet a quantitative estimation of ecotoxicity can be made by means of a toxicity units (TU) calculation [122]. The TUx value describes the amount of dilution that is needed for a sample to reach a defined effect percentage x. High TU values imply high toxicity. The TUx value is calculated by summation of the quotients (tux,i) of component concentration and component ECx value for each product mixture component i, see Equation 5.3. The general assumption is made that the toxicity of the individual components is additional. TUx = ∑ components C(t) = ∑ tux, i ECx i components (5.3) A TU50 calculation has been performed with regard to the quantitative ICE analysis of 1.0 mM phenol solutions exposed to pulsed corona discharges in air and argon, described in section 4.2.1. However, this calculation is a rather rough estimation, because some oxidation products have not been identified. Also the actual catechol and resorcinol concentrations are not known, because these components could not be separated; the same is true for hydroquinone and 1,4-benzoquinone. For a worst case approach it has been assumed that only hydroquinone and catechol represent the dihydroxybenzenes; resorcinol viz. has the lowest ecotoxicitity among the dihydroxybenzenes. 1,4-benzoquinone and hydroquinone have comparable ecotoxicity. Muconic acid is likely to be present but neither Microtox ecotoxicity nor concentrations are known. Finally, the addition of individual toxicity units is only justified, if the components show comparable toxicity effects. Although this assumption is questionable, the essence of ecotoxicity change during oxidation is demonstrated. 120 Chapter 5. A global ecotoxicity estimation has been performed on the phenol oxidation product mixture, obtained after 1 hour and 2 hours of exposure to corona in air and argon, using the EC50 values from Table 2.1. The following compounds are concerned: phenol, hydroquinone, catechol, formic acid, oxalic acid and glyoxylic acid. From the different hydroquinone effect values reported in literature [78], an average value EC505min ≈ 0.061 mg/l has been used for the calculations. Table 5.2 shows the individual component toxicity contributions (tu50) and the total mixture ecotoxicity (TU50). Table 5.2 An estimation of the toxicity unit value TU50 of 100 ml 1.0 mM phenol solutions, after exposure to pulsed corona discharges in air or argon for t=60 min and t=120 min. The corona parameters are V=25 kV, C=1 nF, f=100 Hz, d=1.0 cm. Corona in air Phenol Hydroquinone Catechol Formic acid Oxalic acid Glyoxylic acid Corona in Ar Phenol Hydroquinone Catechol Formic acid Oxalic acid Glyoxylic acid t=60 min C (mg/l) tu50 (-) 58.0 2.0 0.4 6.6 13.6 0.4 5.9 0.7 31.2 2.8 3.1 0.3 TU50= 12.8 t=120 min C (mg/l) tu50 (-) 40.8 1.4 0 0 8.6 0.3 10.8 1.4 34.9 3.1 4.0 0.4 TU50= 6.6 t=60 min C (mg/l) tu50 (-) 34.0 1.2 2.1 34.4 35.2 1.1 1.7 0.2 0 0 0 0 TU50= 36.9 t=120 min C (mg/l) tu50 (-) 11.4 0.4 1.5 24.6 29.1 0.9 4.0 0.5 0 0 2.2 0.2 TU50= 26.6 The total ecotoxicity of the phenol oxidation product mixture seems to be mainly caused by hydroquinone or benzoquinone, due to their very low EC value. The oxidation of phenol by corona in argon produces much higher amounts of dihydroxybenzenes than the oxidation of phenol by corona in air, which results in high ecotoxicity as is reflected by the high TU values. Regarding the fact, that at t=0 minutes TU50=3.5, the total ecotoxicity initially increases considerably but decreases again during oxidation progress. Finally, the ecotoxicity is discussed for some hypothetical situations, where a 1.0 mM phenol solution is completely converted into one of the described oxidation products, according to stoichiometry, see Table 5.3. By this approach, a comparison of phenol oxidation products ecotoxicity is illustrated. Discussion Table 5.3 121 Ecotoxicity of hypothetical situations, where a 1 mM phenol solution is completely converted into one of the described oxidation products, according to stoichiometry. Oxidation product from 1 mM phenol Hydroquinone Catechol Resorcinol Formic acid Oxalic acid Glyoxylic acid Stoichiometry (mM) 1 1 1 6 3 3 TU50 (-) 1805.1 3.4 0.3 34.9 23.9 19.8 The stoichiometric conversion of a 1 mM phenol solution (TU50=3.3) into a 1 mM hydroquinone solution thus creates the most dramatic ecotoxicity, while the conversion of a 1 mM phenol solution into a 1 mM resorcinol solution should result in a decrease of ecotoxicity according to literature EC50 values [78]. The ecotoxicity increase by conversion of a 1 mM phenol solution into a solution of the mentioned carboxylic acids at stoichiometric concentration, is maximum 2 % of the ecotoxicity increase due to the formation of a 1 mM hydroquinone solution. Final oxidation products Phenol oxidation has only been performed during a limited time range in order to ensure stable oxidation conditions, which is important for accurate efficiency measurements. By continued oxidation of the investigated product mixtures by corona in air, the observed polyhydroxybenzenes disappear by ring-cleavage, which is favourable to ecotoxicity. Also the possibly present quinones will be converted. The produced ringcleavage products, viz. polyfunctional aliphatic hydrocarbons, gradually degrade and finally all organic carbon has been converted into carbon dioxide. If all carbon dioxide has left the water, ecotoxicity has disappeared. Theoretically, remaining minor ecotoxicity may appear from carbonic acid i.e. pH toxicity. However, carbonic acid is a weak acid: pKa,I=6.46 and pKa,II=10.22; only about 1 % of the dissolved carbon dioxide appears to react with water [123]. Continued degradation by corona in argon will also eventually result in mineralization, due to hydroxylation and argon ions & metastables bombardment. By absence of oxygen, oxygenation now mainly proceeds by hydroxyl radicals, originating from the dissociation of water. Without detailed information about the composition of the phenol oxidation product mixtures, it is not justified to state that oxidation of phenol by corona in argon results in oxidation products with a different oxygen content than the products obtained from oxidation of phenol by corona in air. On the contrary it is clear, that the mineralization rate due to corona in argon is different from the mineralization rate due to corona in air. Previous to complete mineralization, a large variety of organic oxidation products exists, which will be discussed in section 5.3. 122 Chapter 5. 5.3. Phenol oxidation pathways A complete description of the oxidation mechanism of phenol, containing all possible intermediate products is probably unattainable, because of the huge amount of likely reactions involved in radical-induced oxidation. Simplification of the oxidation mechanism is only possible, when the most relevant intermediates have been identified in the oxidation product mixture. Although a considerable number of oxidation products has been found by ICE chromatography, the revelation of the identity by mass spectrometry has appeared to be very complicated. Therefore, from a theoretical point of view the oxidation of phenol by hydroxyl radicals/oxygen and ozone is described in this section. The deduction has been performed by analogy with the studies of Pan and Von Sonntag on the oxidation of benzene & alkenes [19,50] and fundamental ozone chemistry according to Bailey [49]. The attack of the hydroxyl radical and oxygen on phenol The action of the hydroxyl radical is always reported in combination with oxygen. The hydroxyl radical concentration is very low due to its high reactivity, while oxygen is present in high excess amounts: at T=293K the solubility of oxygen in water is about 1.4 mM [44]. Attack of the hydroxyl radical on the benzene ring of phenol, produces 1,2- and 1,4dihydroxycyclohexadienyl (DHCHD) radicals, see Figure 5.1. The formation of the 1,3DHCHD radical is not discussed in literature, probably because this radical is less stable. Oxygen adds to these radicals to form dihydroxycyclohexadienylperoxyl (DHCHDP) radicals. For easy reference, the intermediates and products are coded with a number according to the attack position of the hydroxyl radical and oxygen. OH O2 • HO OH 1 5 4 HO H 1,4-DHCHD radical 2 3 4 OH HO • H OH 1,2-DHCHD radical Figure 5.1 HO OO • HO H DHCHDP-14 radical O2 HO OO• H OH DHCHDP-12 radical The production of 1,2- and 1,4-dihydroxycyclohexadienyl (DHCHD) radicals from the attack of the hydroxyl radical on phenol. Also shown is the production of dihydroxycyclohexadienylperoxyl (DHCHDP) radicals from the attack of oxygen on the DHCHD radicals. Discussion 123 The DHCHDP radical may react in different ways, see Figure 5.2. Catechol can be produced from the DHCHDP-12 radical by elimination of a hydroperoxyl radical. For the case of the DHCHDP-14 radical, hydroperoxyl elimination is less probable due to the larger -H to •OO- distance compared to the DHCHDP-12 radical; hydroquinone is likely to be formed from the decomposition of a tetraoxide (T) that is produced from the dimerization of two DHCHDP-14 radicals. Also possible is the formation of α,α’-endoperoxyalkyl (EPA) radicals. This reaction is reversible, because of the very weak endoperoxyl bond. The EPA radicals may scavenge oxygen again and then endoperoxyalkylperoxyl (EPAP) radicals are produced according to an irreversible reaction. OH T 2x HO OO • HO H OH hydroquinone OH H O O DHCHDP-14 HO • H HO O2 O O H H OO • H OH EPA-14 radical EPAP-14 radical OH OH HO + HO2• catechol OO• H OH OH DHCHD-12 H • O O OH H OH EPA-12 radical Figure 5.2 O2 OO • H H O O H OH EPAP-12 radical The production of dihydroxybenzenes or α,α’-endoperoxyalkyl (EPA) radicals from dihydroxycyclohexadienyl (DHCHD) radicals. Also illustrated is the formation of endoperoxyalkylperoxyl (EPAP) radicals from the attack of oxygen on EPA radicals. The produced dihydroxybenzenes hydroquinone and catechol will also undergo oxidation and yield trihydroxybenzenes and ring-cleavage products. These pathways are not discussed, because they are comparable to the described mechanisms. 124 Chapter 5. The dimerization of EPAP radicals produces a tetroxide (T), that decomposes into two endoperoxides (EP) and oxygen, see Figure 5.3. The unstable endoperoxides undergo ring-cleavage; regarding EP-14b/12b, ring-cleavage occurs together with the elimination of carbon monoxide. In this way, aliphatic monounsaturated C5 & C6 hydrocarbons (P14/12) with carboxyl-, aldehyde-, ketone- or alkanol-functional groups are produced. H HO H HO O O HO H H O O 2 H OO • OH H HO O P-14d O H O H O O H H OH OH P-14e O2 H H HO EP-14a T -H2O P-14c OH HO O H O H OH OH EPAP-14 OH H HO O O -CO O OH P-14f P-14g OH EP-14b HO HO H OO 2 H H O O H OH T HO O H O O O EP-12b OH H O H OH -H2O O H H O H OH P-12d P-12c OH H H OH P-12e OH OH O Figure 5.3 H OH O2 EPAP-12 OH O H O H EP-12a OH • O O H H O OH O OH O OH H OH -CO H O H OH O P-12f The dimerization of endoperoxyalkylperoxyl (EPAP) radicals produces a tetroxide (T) that decomposes into endoperoxides and oxygen. The endoperoxides undergo ring-cleavage and aliphatic monounsaturated C5 & C6 hydrocarbons (P-14/12) with polyfunctional groups are produced. Discussion 125 The monounsaturated products P-14/12 can also be attacked by a hydroxyl radical and oxygen. Also ozone attack is very likely but is not discussed here. Every product P can produce two hydroxyperoxyl C5 & C6 radicals HP-14/12, because the hydroxyl radical and oxygen can attack on both sides of the alkene bond, see Figures 5.4 and 5.5. H HO O O H O H HO H •OO H HO HO +O2 H O H O H O 2x T 2 P-14d H O O H O H H HO O O H H OH OH HO +O2 H HO H HO H •OO O O H H OH OH HP-14e2 H OO H HO • H O O H H OH OH HP-14e1 P-14e O H OH H HO O OH +O2 OH H + 2x T H + 2 T 2 H HO H O T O H + H T 2 H HO H OH saturated hydroxyperoxyl C5,C6 radicals 2x OH + T O O O2 OH + H O2 H OH H OH 2 O • H H H 2 + O αHA-14e2 OH αHA-14g1 P-14g1-I OH H H • O O H O2 αHA-14e1 + O H + HO P-14e2-I 2 O2 O • OH 2x + H OH 2 H O O2 O αHA-14d2 H 2 + HO H H HO OH P-14e1-I 2x O2 • O P-14d2-I 2x + H O αHA-14d1 O H O • HO H O 2 H 2 O O OH HP-14g2 monounsaturated polyfunctional C5,C6 hydrocarbons Figure 5.4 H O OH HP-14g1 H HO H •OO P-14g O O P-14d1-I HP-14d2 HO H •OO H HO H HO O HP-14d1 HO H HO H •OO H + OH P-14-g2-I saturated polyfunctional C2,C3,C4 hydrocarbons O 2 H • HO H αHA-14g2 α-hydroxyalkyl C2,C3,C4 radicals The attack of the hydroxyl radical and oxygen on the monounsaturated ring-cleavage products (P-14) produces hydroxyperoxyl C5 & C6 radicals (HP-14). Dimerization of HP-14 produces a tetraoxide intermediate (T) that decomposes into polyfunctional saturated hydrocarbons (P-14-I), αhydroxyalkyl radicals (αHA-14) and oxygen. 126 Chapter 5. The HP-14/12 radicals dimerize to a tetraoxide, that decomposes into saturated aliphatic polyfunctional C2-C4 hydrocarbons (P-14/12-I), saturated aliphatic αhydroxyalkyl-functional C2-C4 radicals (αHA-14/12) and oxygen. O O H O H O H O OH H O H H OH •OO H H OH HO +O HP-12d1 2 O H O H P-12d O 2x T H O OH H H OH HO +O2 2 H OH O H O H OH HO 2x T H T monounsaturated polyfunctional C5,C6 hydrocarbons saturated hydroxyperoxyl C5,C6 radicals H OH H O O 2 T 2 + H O2 H OH + O2 + O2 + O2 • H αHA-12e2 OH O O H OH O H P-12f2-I O OH O + 2 H 2 + H • 2 H OH H O HO • αHA-12f1 H OH T O2 αHA-12e1 P-12f1-I 2x H OH O H 2x + • HO P-12e2-I HP-12f1 H O OH H H O OO • HO H HP-12f2 + OH 2 O2 H αHA-12d2 H P-12e1-I 2x + OH O O OH O O OH H 2 H 2 P-12d2-I OH P-12f Figure 5.5 +O2 H • H OH H αHA-12d1 H OH+ O OH H O OH H H O OH •OO H H O HO O HP-12e1 HO O H H OH H H O OO • HO H HP-12e2 O 2 P-12d1-I OH P-12e H H HO H HP-12d2 HO O + O H OH 2x H T OO • OH O HO H H OH H H O H OH •OO O 2 + saturated polyfunctional C2,C3,C4 hydrocarbons 2 O OH • H αHA-12f2 α-hydroxyalkyl C2,C3,C4 radicals The attack of the hydroxyl radical and oxygen on the monounsaturated ring-cleavage products (P-12) produces hydroxyperoxyl C5 & C6 radicals (HP-12). Dimerization of HP-12 produces a tetraoxide intermediate (T) that decomposes into polyfunctional saturated hydrocarbons (P-12-I), αhydroxyalkyl radicals (αHA-12) and oxygen. Discussion 127 Finally, by scavenging oxygen, the α-hydroxyalkyl radicals (αHA-14/12) are converted into α-hydroxyalkylperoxyl radicals (αHAP-14/12), see Figure 5.6. H O O2 • H HO O αHA-14d1 O2 O • HO H O2 O HO αHA-14e1/14g1 H HO H • O2 O H αHA-14e2 OH H • O2 HO H αHA-14g2/12e2 O H O HO • H OH H O2 OH O H OH O HO • H αHA-12e1/12f1 α-hydroxyalkyl radicals Figure 5.6 O H •OO HO HO2• + HO2• + HO2• + HO2• + HO2• + HO2• + HO2• H O HO P-14e1/14g1-II H HO H O O H OH H P-14e2-II H H O H OH H P-14g2/12e2-II O O H OH O HO H OO• αHAP-12d1 H O H OH H O P-12d1-II H H OH OO• H αHAP-12d2/12f2 O O H P-12d2/12f2-II OH OH O + O OH H H O2 HO2• P-14d2-II H OH OH O H αHA-12d2/12f2 OH O O O2 • O O αHAP-14g2/12e2 αHA-12d1 H H + HO H H H O HO H O P-14d1-II H αHAP-14e2 O H H •OO O O OH OO H • HO αHAP-14e1/14g1 H OH OH H αHAP-14d2 OH H H HO H H O •OO HO H αHA-14d2 • O O αHAP-14d1 HO H H H OO HO • H OH HO H OO • αHAP-12e1/12f1 α-hydroxyalkylperoxyl radicals H OH O O H P-12e1/12f1-II saturated polyfunctional C2,C3,C4 hydrocarbons The attack of oxygen on α-hydroxyalkyl radicals produces α-hydroxyalkylperoxyl radicals that decompose into saturated polyfunctional C2-C4 hydrocarbons by elimination of a hydroperoxyl radical. 128 Chapter 5. The α-hydroxyalkylperoxyl radicals (αHAP-14/12) decompose into saturated aliphatic C2-C4 hydrocarbons with carboxyl-, aldehyde-, ketone- or alkanol-functional groups by elimination of hydroperoxyl radicals. A summary of all presented saturated polyfunctional oxidation products of hydroxyl radical & oxygen - induced phenol oxidation is given by Figure 5.7. The occurrence frequency of the compounds is indicated by the compound code used in the oxidation pathways. O H H P-12d1-I, P-12f1-I, P-12d2/12f2-II glyoxal (58 g/mol) P-14e2-I, P-14g2-I, P-14e1/14g1-II glyoxylic acid (74 g/mol) H P-14d2-I, P-14d1-II ketomalonaldehyde (86 g/mol) O P-14d1-I, P-14d2-II hydroxymalonaldehyde (88 g/mol) OH P-14g1-I, P-12e1-I, P-14g2/12e2-II 3-hydroxy-1,2-propanedione (88 g/mol) P-12e2-I, P-12f2-I, P-12e1/12f1-II 2-hydroxy-3-oxopropionic acid (104 g/mol) P-12d2-I, P-12d1-II 2-hydroxy-1,3,4-butanetrione (116 g/mol) P-14e1-I, P-14e2-II 2,4-dihydroxy-1,3-butanedione (118 g/mol) O H O O OH O O H O H H HO O H O H H O H OH H OH H O O O H O H H OH H O HO H O H O H OH Figure 5.7 Saturated polyfunctional C2, C3 & C4 hydrocarbons produced from the oxidation of phenol by hydroxyl radicals and oxygen. Discussion 129 The attack of ozone on phenol Next, the electrophilic addition of ozone on phenol is described. Although the phenolate anion (C6H5O-) reacts considerably faster with ozone than phenol [7], these reactions have not been described here, because this anion only exists in an alkaline environment. The two most likely ozone attack positions of the phenol molecule are the 2,3- and 3,4position. The 1,2- position is not considered here, because the hydroxyl group is likely to cause sterical hindrance. For easy reference, now all intermediates and products are coded with a number according to the attack position of ozone. The attack of ozone on phenol produces the molozonides M-23 and M-34, see Figure 5.8. These molozonides rearrange immediately to the ozonides O-23 and O-34. The unstable ozonides decompose by ring-cleavage to zwitterion-functional products. The zwitterion group can be formed on both sides of the former attack position, resulting in four compounds viz. Z-23a/b and Z-34a/b. In water, the zwitterion groups hydrolyze to hydroxyalkylhydroperoxyde-functional ring-cleavage products (H-23a/b and H34a/b). OH OH O + O C H OH O O O O3 M-23 H O Z-23a OH O O O OH O-23 OH 1 5 4 H H-23a OH H O + C H O Z-23b O 2 OOH OH O H H H H-23b O OH OOH 3 4 OH OH + O3 OH O O H OH O M-34 O Z-34a O O C O H O O H OH O-34 H + C HO H O OOH OH O H-34a OH O Z-34b Figure 5.8 H HOO H O OH H-34b The attack of ozone on the 2,3- or 3,4-position of phenol initially produces molozonides (M-23/34), which rearrange immediately to ozonides (O-23/34). The unstable ozonides decompose to zwitterionfunctional ring-cleavage products (Z-23/34). Hydrolysis of the zwitterionfunctional products yields hydroxyalkylhydroperoxide-functional ringcleavage products (H-23/34). 130 Chapter 5. The first generation hydroxyalkylhydroperoxides (H-23ab and H-34a/b) decomposes into monounsaturated C6 ring-cleavage products (P-23c/d/e and P-34c/d/e) by release of water or hydrogen peroxide, see Figure 5.9. Every set of hydroperoxides produces three different products P, because P-23d and P-34d can be produced from both of the hydroperoxides. The markers c/d/e identify the six possible products by the position of a carboxyl or aldehyde endgroup. O O OH H O OH H O H-23a OH H H-23b OOH OH H O H OH OOH + H2O + H2O2 + H2O + H2O + H2O2 + H2O P-23c O H O H O P-23d O O H OH O P-23e O O H OH OH H H O H-34a OOH OH O P-34c O H O H OH O H HOO OH H-34b H Figure 5.9 O P-34d O O H O OH P-34e The decomposition of the first generation hydroxyalkylhydroperoxides (H23 and H34) into monounsaturated C6 ring-cleavage products (P23 and P34) with carboxylic acid- or aldehyde-functional endgroups by release of water or hydrogen peroxide respectively. Discussion 131 The monounsaturated C6 ring-cleavage products P23c/d/e and P34c/d/e can also be attacked by ozone. Attack by a hydroxyl radical and oxygen is also possible, but is not discussed here. Then, a second generation molozonides M-45c/d/e, M-56c/d/e and ozonides O-45c/d/e, O-45c/d/e is produced, see Figures 5.10 and 5.11. Decomposition of the second generation of ozonides splits the former phenol molecule into two parts. A second generation zwitterions (Z-45f/g/h/i and Z-56f/g/h/i) is produced, of which the Z-45g/56g and Z-45h/56h zwitterions can be formed in two ways. Together with the zwitterions a first generation of saturated polyfunctional C2 & C4 hydrocarbons is produced (P-45c1/c2/d1/d2/e1/e2-I and P-56c1/c2/d1/d2/e1/e2-I). O O O O OH H O O OH H O O3 O O P-23c H C O O O O OH H O O O H + O P-45c1-I Z-45f H O HO O H C O + O Z-45g O O + O H H O O H O3 O O O P-23d O H O H O O M-45d O O O H O H C + Z-45h H O + O O O-45d O H P-45d1-I O O + O H O O O H C H P-45c2-I O O O H O O + O-45c M-45c O OH + H O H P-45d2-I Z-45g O O H + O O H OH O O O3 O O P-23e O M-45e O H OH O O O H C O O Z-45h O O O O-45e H OH O H OH O + C O O Z-45i Figure 5.10 OH + O O H P-45e1-I O O + O H H P-45e2-I Ozone attack on the 4,5-position of the monounsaturated C6 ring-cleavage products (P-23) yielding a second generation molozonides (M-45) and ozonides (O-45). The ring-cleavage of the 4,5-ozonides produces a second generation zwitterions (Z-45) and a first generation saturated C2 & C4 hydrocarbons (P-45-I) 132 Chapter 5. H O O O O H OH O3 O O O H OH O O O M-56c P-34c O H C O O O O O O + O OH + H + C O O-56c H O H P-56c1-I Z-56f O H OH O O O O O + H OH O Z-56g P-56c2-I H O O O3 H O H O O O O O P-34d H O H O O O M-56d O C O H H O O + O H H + C O O-56d H O O H Z-56h O H O + P-56d1-I O O O O + O Z-56g H H P-56d2-I H O O3 O H OH O P-34e O O O O H O OH M-56e O O O O H O OH O-56e OH + C O H O O Z-56h O O O O H + C O OH Z-56i Figure 5.11 O + + O O O H P-56e1-I O O H H P-56e2-I Ozone attack on the 5,6-position of the monounsaturated C6 ring-cleavage products (P-34), yielding a second generation molozonides (M-56) and ozonides (O-56). The ring-cleavage of the 5,6-ozonides produces a second generation zwitterions (Z-56) and a first generation saturated C2 & C4 hydrocarbons (P-56-I) The second generation zwitterions (Z-45 and Z-56) hydrolyzes to a second generation hydroxyalkylhydroperoxides (H-45f/h, H-56f/h, H-4556g/i), which decompose into a second generation saturated polyfunctional C2 & C4 hydrocarbons (P-45f1/f2/h1/h2-II, P-56f1/f2/h1/h2-II, P-4556g1/g2/i1/i2-II) by elimination of water or hydrogen peroxide, see Figure 5.12. Discussion 133 O OH O H C O O + OH O O OH O OH HO O O H P-45f1-II + H2O P-45f2-II + H2O2 P-56f1-II + H2O P-56f2-II + H2O2 P-45g/56g1-II + H2O P-45g/56g2-II + H2O2 P-45h1-II + H2O P-45h2-II + H2O2 P-56h1-II + H2O P-56h2-II + H2O2 P-45i/56i1-II + H2O P-45i/56i2-II + H2O2 OH OOH H O O H-45f Z-45f O O OH H O O HOO H + C O O OH O O O OH OH Z-56f H-56f H H HO H H C+ O O O Z-45g=Z-56g OH O O O H OH H O O O O OH H OOH H-45g=H56g O O H O O O H H C O O H + O O O HO H OOH Z-45h H H-45h O O + C O H O O HOO H O H O O O OH H HO H O O O O H H-56h Z-56h H OH O O H O H OH H + C O OH O O Z-45i=Z-56i OH HO H O O O OH OOH H-45i=H-56i OH O O H Figure 5.12 Hydrolysis of the second generation zwitterions (Z-45 and Z-56). Also illustrated is the decomposition of the second generation hydroxyalkylhydroperoxides (H-45 and H-56) into a second generation saturated C2 & C4 ring-cleavage products (P-45/56-II) with carboxylic acidor aldehyde-functional endgroups, by release of water or hydrogen peroxide respectively. 134 Chapter 5. A summary of all presented saturated polyfunctional oxidation products of ozoneinduced phenol oxidation is given by Figure 5.13. The occurrence frequency of the compounds is indicated by the compound code used in the oxidation pathways. H O O H P-45c1-I, P-45d1-I, P-56c1-I, P-56d1-I,P-45g/56g2_II glyoxal (58 g/mol) P-45e1-I, P-56e1-I, P-45g/56g1-II, P-45i/56i2-II glyoxylic acid (74 g/mol) P-45i/56i1-II oxalic acid (90 g/mol) P-45d2-I, P45-e2-I, P-56d2-I, P-56e2-I, P-45h2-II, P-56h2-II 1,2,4-butanetrione (100 g/mol) P-45c2-I, P-45f2-II, P-56h1-II 2,4-dioxobutyric acid (116 g/mol) P-56c2-I, P-56f2-II, P-45h1-II 3,4-dioxobutyric acid (116 g/mol) P-45f1-II, P56f1-II oxo-succinic acid (132 g/mol) OH O O H OH O O OH O O H O H O HO O H O O O O H OH O OH O OH O Figure 5.13 Saturated polyfunctional C2 & C4 hydrocarbons produced from the oxidation of phenol by ozone. Discussion 135 The described reaction chemistry of the oxidation of phenol by hydroxyl radicals, oxygen and ozone accounts for the formation of a broad range of polyfunctional oxidation products, but is still incomplete. Many of the described products are not stable and will further degrade due to the rigorous oxidizing conditions. With regard to the possible oxidation products, a following summary can be made: The oxidation of phenol by hydroxyl radicals and oxygen initially yields dihydroxybenzenes and monounsaturated aliphatic C5 and C6 hydrocarbons. The dihydroxybenzenes are either oxidized to trihydroxybenzenes or undergo ring-cleavage like phenol. Repeated attacks of the hydroxyl radical and oxygen result in the formation of saturated aliphatic C2, C3 and C4 hydrocarbons with carboxyl-, aldehyde-, ketoneand/or alkanol-functional groups. The oxidation of phenol by ozone initially yields di- and monounsaturated aliphatic C6 hydrocarbons. Attack of ozone on these ring-cleavage products yields saturated aliphatic C2 and C4 hydrocarbons with carboxyl-, aldehyde- and/or ketone-functional groups. Glyoxal, glyoxylic acid and oxalic acid are well-known stable oxidation products of phenol, see section 2.4.1. In addition these products form an oxidation decay series to carbon dioxide. Ketomalonaldehyde and hydroxymalonaldehyde are precursors to ketomalonic acid, which is described in literature, see section 2.4.1. The products with aldehyde-functional endgroups will be oxidized to carboxylic acids. Carboxyaldehydes are mentioned in literature [124] as products from the cleavage of aromatic compounds by ozone. With regard to the produced oxocarboxylic acids (α: RC(O)COOH, β: RC(O)CH2COOH, γ: RC(O)-(CH2)2COOH) the α-form is relatively stable, the β-form decomposes by decarboxylation to corresponding ketones and the γ-form is again stable and does not decarboxylate [125]. β-hydroxy aldehydes (R-C(OH)-CH2-CHO) generally dimerize or polymerize [126]. Oxalic acid, glyoxylic acid and polyhydroxybenzenes have been identified in this study, see section 4.2.1. Although the other observed products viz. formic, acetic, propionic, malonic, succinic and maleic acid cannot be directly traced in the presented oxidation pathways, these compounds are eventually explained from continued degradation of the products described in the models or from combined OH/O3 oxidation reactions. The phenol degradation pathways have been described for either oxidation by the hydroxyl radical or ozone, in order to present a clear model. For the case of practical conditions for corona oxidation in humid air, all described reactions occur mixed up and also other reactive species will be present like e.g. singlet oxygen, metastables, ions and UV photons. Modeling of the reaction kinetics of phenol oxidation by pulsed corona discharges is only meaningful, if the relevant reaction pathways are known from experimental observations. This has been the main problem, because both separation and identification of the theoretically possible intermediates/products is analytically considered to be very difficult. Detailed clarification of the phenol oxidation mechanism by e.g. using labelled compounds, blocking of reactive molecular sites and chemical derivatization reaches beyond the scope of this thesis. 136 Chapter 5. Toxicological aspects of phenol degradation The discharge of oxidized phenol solutions into surface waters may lead to oxygen depletion, due to the presence of polyhydroxybenzenes, which are strong reducing agents. Hydroquinone and 1,4-benzoquinone show very high ecotoxicity according to the Microtox test. Although several carboxylic acids produced during mineralization of phenol occur in nature, a concentrated discharge flow causes ecotoxicity due to acidity. Although direct human intoxification by phenol oxidation products is not likely, indirect exposure via the food chain may eventually occur. Therefore a short discussion about human toxicity is presented [1,61,127-129]: All hydroxybenzenes cause irritation or etching of skin and mucous membranes and are skin poisons. After resorption respirational paralysis takes place. 1,4-benzoquinone causes severe eye injury. Global fatal doses are: phenol 5-10 g, resorcinol 12 g, hydroquinone 5-12 g, pyrogallol 2-3 g. Hydroxybenzenes exhibit mutagenic properties, this means that these compounds cause chromosomic and genetic abberations; mutagenic activity is reported according to the order hydroquinone > hydroxyhydroquinone > pyrogallol. For the dimerization products 2,2’- and 4,4’-dihydroxybiphenyl human mutation data are reported. The multiring condensation product dibenzo-p-dioxin is a questionable carcinogenic compound. Although endoperoxides cannot be isolated from the oxidation product mixture because of their instability thus reactivity, these compounds probably are the most harmful intermediates. Endoperoxides may also release singlet oxygen which is highly reactive towards physiological materials [43]. Unsaturated carboxylic acids like muconic, maleic, fumaric and acrylic acid are harmful due to their reactivity of the alkene bond. By metabolysis, the alkene bond may be converted into a carcinogenic epoxide, as has been explicitly described for benzene and polyaromatic hydrocarbons. The epoxide causes DNA damage, because it reacts with nucleic acids. Aldehydes exhibit mutagenic properties, cause strong irritation of the mucous membranes and skin and act on the central nervous system; examples are formaldehyde, acetaldehyde, acrolein and glyoxal. Unsaturated aldehydes (enals) are reported to be particularly reactive with some biological molecules [124]. Formic and acetic acid are entitled as etching protoplasma poisons due to their acidity, but their salts show low toxicity. Skin poisoning is possible for the case of formic acid, because of fat-dissolving properties. Acetic acid exhibits a lethal dose value of about 20-50 g. In nature, formic acid occurs in stinging-nettle and ants, while acetic acid is an important ingredient of vinegar. Oxalic acid shows a very strong etching effect (low pKa), combines calcium and causes kidney damage or cardial paralysis; the lethal dose value is about 5-15 g. Oxalic acid salts are toxic. In nature it occurs in small amounts in rhubarb and spinach. Glyoxylic acid is a metabolite in mammalian biochemical pathways. It also occurs in in plants e.g. in unripe food e.g gooseberries. It is oxidized in the human body to oxalic acid and thus causes comparable effects. The lethal dose LD50=2500 mg/kg (oral,rat). Glyoxylic acid has possibly mutagenic properties. Discussion 137 5.4. Analysis techniques Liquid chromatography The identification of corona-induced liquid phase oxidation products of organic compounds is best performed by ion-exclusion chromatography (ICE). The oxidation mechanism induced by pulsed corona discharges results in a diverse product range, consisting of several aliphatic carboxylic acids, which can be properly separated by ionexclusion chromatography. The polyhydroxybenzene intermediate oxidation products do not interfere with the carboxylic acids. The application of a conductivity detector in series with a UV absorbance detector offers specific sensitivity for the carboxylic acids. For the case of a target compound with fluorescent properties like e.g. polycyclic aromatic hydrocarbons, benzene, phenol and laser dyes, a fluorescence detector offers conversion measurements with much higher sensitivity, although the non-fluorescent degradation products require yet identification by a UV absorbance or conductivity detector. The required analysis time of phenol-containing oxidation products is long viz. about 30 minutes, due to strong interaction of phenol with the partially-crosslinked polystyrene-divinylbenzene stationary phase resin of the ICE column. Depending on the column type, it may be possible to introduce organic modifier into the acidic aqueous eluent by means of a gradient elution program. In this way the retention of phenol and/or polyhydroxybenzenes may be forced after elution of the complex carboxylic acid range, in order to shorten the analysis time. However, this option has not been investigated, because identical column performance cannot be guaranteed anymore after operation with organic modifier-based acidic eluents. Mass spectrometry The product range complexity rapidly increases with the complexity of the target compound molecular structure. The fact, that the number of possible oxidation products from a very simple molecular structure like a benzene ring is already extensive, inevitably demands the application of mass spectrometry for identifcation after separation. Mass spectrometry has been applied both directly by electron-impact MS and via IonSpray and APCI interfaces with the liquid chromatograph (off-line MS). Direct mass analysis of oxidized high initial concentration phenol solutions has not revealed significant information, although the appearance of the oxidized solution implies thorough chemical changes. Concerning off-line MS, the signals of separated components are overruled by the high background signal due to the acidic ICE eluent. The available interfaces are not suitable for handling the separated low molecular weight components. 138 Chapter 5. Gas chromatography Although GC-MS technology is more straightforward than LC-MS technology, the GC analysis of high water content samples is problematic. Therefore the oxidation products have to be transferred from the aqueous phase to an organic phase, by solid-phase extraction or freeze-drying. Several attempts have failed, because certain compounds within the complex mixture have been lost during this sample preparation step. For application of GC-MS, also chemical derivatization of the oxidation product compounds may be necessary, for the production of thermally stable compounds and for improved separation and detectability. Aldehyde screening test For gas phase analysis, both gas chromatography-mass spectrometry (GC-MS) and Fourier transform Infrared spectroscopy (FTIR) have been applied. Chemical derivatization has been applied to trap possible aldehydes present in the gas phase. Gas sampling can be performed in-field; the storage life of tubes after sampling depends on the analyte and the sampling/derivatization technique: sampled aldehyde screening tubes are stable for at least one week at 25°C. Nevertheless, here the performed tests have been directly processed. Direct gas sampling may be disadvantageous, due to the presence of water vapour that is produced by the corona discharges over the aqueous solution. Hydroxyl radical identification The fact that in-situ ESR has not revealed hydroxyl radicals can be explained by the low sensitivity of ESR compared to fluorescence spectrometry. Also it is reported by Sun [130], that the spin trapping efficiency η of DMPO for hydroxyl radicals is only η=33 ±3 % for OH generated by photolysis of hydrogen peroxide and η=28 ±3 % for OH generated by photocatalytic oxidation of water. The trapping of hydroxyl radicals by DMPO is thus likely to be far from effective, although DMPO is a well-known and widely applied spin trap. Fourier transform infrared spectroscopy Gas phase analysis by FTIR can be performed in-situ and is highly sensitive. Simple molecules like oxides of carbon, nitrogen and sulfur are completely identified. Unfortunately, full revelation of the identity of a complex hydrocarbon, especially in a mixture, is nearly impossible. On the contrary, the presence of functional groups like e.g. hydroxyl (carboxylic acids, alkanols), carbonyl (aldehydes, carboxylic acids, esters, ketones), nitrogen-based groups (nitriles, amines, amides, nitro compounds), sulfurbased groups (thiols) and the presence of aromaticity (mono, polycyclic) or aliphatic unsaturated hydrocarbons (alkenes, alkynes) can be monitored accurately. It should be noted, that in-situ FTIR measurements may encounter disturbance by water vapour. Discussion 139 Liquid chromatography versus LIF-spectroscopy Reversed-phase HPLC features a very powerful separation of the oxidation product mixture components combined with identification by a wide range of detectors. On the contrary, samples have to be analyzed off-line and the analysis times are always longer than that of in-situ spectroscopic measurements. Sample ageing is relevant for the case of e.g. the phenol oxidation product mixture, because after stopping the corona discharge experiment i.e. the oxidation time, residual generated oxidizers continue the degradation in the time lag before analysis. Especially the trihydroxybenzenes are susceptible to further degradation after sampling. In addition, all hydroxybenzenes are light-sensitive. LIF spectroscopy features a high sensitivity, time- and space-resolved analysis of fluorescent priority compounds like e.g. polyaromatic hydrocarbons, benzene and aniline. Although LIF spectroscopy is actually applied for the study of gas phase reactions, its application for the study of fluorescent molecules in aqueous solution is possible, but requires some points of attention. Only low concentration solutions can be studied, because laser absorption by a high concentration solution results in conversiondependent excitation. Despite fluorescence quenching, the conversion determination can be justified by decomposition of the solution fluorescence spectrum into the spectrum of the target compound and oxidation products. Analyte photolysis due to the excitation photons can be minimized, by application of short exposure times and low laser power. By electronic excitation, the acid dissociation constant pKa of phenol is reported to change considerably that is, phenol becomes a much stronger acid [131]. In the ground state S0 pKa=10, while for the first excited singlet state S1 pKa=3.6-4.0. For the lowest triplet state T1 pKa=8.5. By laser excitation, excited phenol molecules may be produced that will dissociate into phenolate anions according to the Equations 5.4ab. {C6H5OH}* + H2O H3O+ + {C6H5O-}* Ka = [H3O + ][{C6 H5O − }*] [{C6 H5OH }*] (5.4a) (5.4b) The phenolate anions are preferentially attacked by electrophilic oxidizer species. In liquid chromatography, the detection of separated components by UV absorbance or fluorescence indeed also creates excited phenol molecules, but this is an ex-situ measurement. If phenol conversion measurements by in-situ LIF spectroscopy are biassed by the mentioned effect, the excitation volume should be sufficiently high and also oxidizers should be present in that volume. It is likely, that oxidation mainly takes place at the gas-liquid interface; the laser beam enters the reactor vessel some centimeters below the water surface. The excitation volume is estimated from a 3 mm laser beam waist and 10 cm optical path length to be about 0.7 ml which is only 0.2 v/v% or 0.7 v/v % of the total reactor volume (300 or 100 ml). Therefore this effect can be ignored. 140 Chapter 5. Conductometry Conductometry measurements in general may be applied as very cheap and simple insitu tests to globally monitor oxidation progress of C/H/O-based target compounds infield, by the increase of the carboxylic acid concentration. For the case of halogen/ nonmetal-based organic target compounds (viz. chlorobenzene and aniline), conductivity increase during oxidation progress is also due to the formation of inorganic ions (chloride and nitrate respectively) and then specific pH conductometry is to be preferred for monitoring the disappearance of total carbon. For that case, a certain conductivity thus remains after complete mineralization of the target compound. Microtox ecotoxicity The toxicity change due to degradation of phenol by pulsed corona discharges has been monitored by a bacteria test viz. the Microtox ecotoxicity test. Actually, the determination of environmental toxicity should also take place for other trophic levels like for fish, crustaceans, algae and activated sludge, however this approach reaches beyond the scope of this thesis. The toxicity effects revealed by the Microtox test cannot be related to human toxicity, because these are based on bacteria viz. Vibrio fischeri. This can be illustrated by the EC505min effect values [78] and human lethal dose values [127] of some compounds with very high human toxicity: potassium cyanide: EC505min= 4.77 mg/l CN- while 0.15-0.25 g is an average lethal dose for adults; aflatoxin B1: EC505min=21.97 mg/l while 5 µg/kg is an upper exposure limit; arsenic (III) oxide: EC505 min=73.73 mg/l while 0.12-0.3 g is a lethal dose for adults. General remarks It can be summarized that all applied analysis techniques have generally yielded very consistent results. Hydroxyl radicals account for the formation of polyhydroxybenzenes from phenol and have been demonstrated by fluorescence spectrometry using the molecular probe CCA. Ozone is consumed by phenol in aqueous solution according to UV absorbance spectrometry and accounts for the formation of carboxylic acids together with hydroxyl radicals and oxygen. Polyhydroxybenzenes have been identified by rp-HPLC, ICE, LIF and the Microtox test. Carboxylic acids have been demonstrated by ICE and conductometry. No other gaseous phenol oxidation products than carbon dioxide have been identified according to FTIR, the aldehyde test and the TOC measurements. The less consistent results, viz. radical identification by ESR and the identification of oxidation products by MS, are explained by sensitivity problems and/or non-optimal conditions. Of particular interest for global in-field monitoring of oxidation progress are in-situ UV absorbance or fluorescence spectrometry and conductometry. UV absorbance is generally applicable to organic compounds, while fluorescence spectroscopy has a very high dynamic range but is only applicable to fluorescent molecules; on the contrary several aromatic compounds exhibit fluorescent properties [85-87,131]. Unfortunately, spectroscopic analysis of turbid waste water is not possible; TOC measurements then are more favourable but these measurements are not in-situ and are more complicated. Conductometry is cheap, simple and suitable for monitoring oxidation progress of any organic compound, but the results are rather emperical. Discussion 141 5.5. AOP comparison Fundamental merits and problems are described for the advanced oxidation processes [132,12] discussed in section 1.1 together with corona discharge technology. A financial comparison of these technologies has not been part of this study. General problems concerning oxidation are the conversion of nitrogen from air into nitrogen oxides and nitrous oxide. Nitrogen oxides contribute to environmental acidification; nitrogen dioxide is suspected to exhibit human reproductive toxicity [133]. Nitrous oxide is a greenhouse gas. Oxidation of halide-containing waste water may result in halogenation of organic target compounds [124]; also problematic is the oxidation of bromide to carcinogenic bromate (BrO3-). Ozone/UV Pro: Ozone is a powerful oxidizer that can be produced from a simple ozonizer setup and air. Also, hydrogen peroxide is produced from the oxidation of water by ozone. Contra: The reactions of ozone in aqueous solution are mass transfer limited. Molecular ozone reacts much more slowly than hydroxyl radicals. Unreacted/undecomposed ozone leaving the reactor has to be detoxified e.g. by chemical reduction. The UV-lamp power efficiency and lamp life are limited; the penetration depth of UV radiation in turbid aqueous solutions is low. Hydrogen peroxide/UV Pro: Hydrogen peroxide is a pure source of hydroxyl radicals. Activation can be applied by UV photons and/or iron (II,III) salts. The quantum yield for generation of hydroxyl radicals from photolysis of hydrogen peroxide is about 1.0. Contra: The transport, storage and handling of hydrogen peroxide require special safety precautions. Hydrogen peroxide shows only weak absorption in the range 200-300 nm and also absorption at wavelengths higher than 300 nm is not significant; by addition of iron salts the hydroxyl radical production efficiency is greatly enhanced, but a high iron salt concentration is required according to stoichiometry. The application of other salts than iron hydroxo/carboxyl chelates causes unnecessary release of anions like e.g. sulfate, chloride or nitrate. Lamp power efficiency and lamp life are limited. Turbid waste water is problematic. Photocatalytic oxidation Pro: A simple setup is required. Contra: The quantum yield for the generation of hydroxyl radicals on the surface of the most generally applied photocatalyst titanium dioxide is low. Indicated numbers are only 4-8% for TiO2 slurries or even lower for immobilized photocatalyst particles. Also the quantum yield is dependent on the light intensity. Mainly for these reasons, the scale-up procedure from laboratory setup to industrial application is problematic. Mass transfer limitation occurs for immobilized photocatalysts. After reaction, suspended photocatalyst particles have to be separated from the oxidation product mixture. Again, lamp power efficiency and lamp life are limited. Turbid waste water is problematic. 142 Chapter 5. Wet oxidation Pro: Supercritical water conditions favour high solubility of organic compounds plus oxygen and extreme chemical reactivity. Contra: The setup requires autoclave-proof housing including e.g. high-pressure pumps, (pre)heaters, compressors thus only has large scale applicability. Oxygen addition is necessary. After oxidation, vapour/liquid/solids separation and cooling of the product mixture are necessary. Due to depressurization, only batch flow operation is possible. The vigorous conditions may enable polymerization and multiring condensation of organic compounds. Electron beam treatment Pro: The generation efficiency of hydroxyl radicals by high energy electrons is about 2.7 OH per 100 eV. The E-beam technology is especially suitable for the degradation of halogenated hydrocarbons, because these compounds react rapidly with solvated electrons. Waste water containing solid matter up to 5 % is accepted without pretreatment. Continuous flow operation is possible. Contra: The setup involves complex equipment and the technology only has large scale applicability. The shielding of β - radiation is necessary. The appearance of electrons from the setup through the titanium foil high vacuum separator causes energy dissipation. Idle E-beam setups require power due to vacuum maintenance and cathode heating. The application of nitrous oxide for improved hydroxyl radical production from aqueous electrons requires additional facilities. Ultrasonic irradiation Pro: With a simple setup, vigorous conditions can be created in aqueous solution. Continuous flow operation is possible. Contra: A main issue is the fact that the scale-up procedure is complex. The shielding of ultrasound is important, because the intense first subharmonic of the applied driving frequency and white noise cause hearing impairment; also, a potential hazard is the formation of aerosols from the harmful solution by surface wave activity [134]. The transducer device undergoes erosion by intense cavitation. Pulsed corona Pro: The pulsed corona technology features a simple setup and produces a broad range of oxidizers. Oxidizers are produced by highly efficient processes. Target compounds can be oxidized in both the liquid phase and gas phase. Continuous flow operation is possible. Contra: Shielding of electromagnetic radiation is necessary to avoid interference with adjacent electronic devices. The life of pulsed-corona circuits is limited, because longterm generation of steep pulses will eventually strain the electrical components. Anode corrosion may occur. Discussion 143 Electric arc Pro: The thermal plasma offers the highest destruction power. Conversion of gas, liquid and solid phase target compounds is possible. Contra: The energy consumption is very high. After decomposition, cooling and stripping of the gaseous reaction products is necessary. The required facilities require large scale operation. The process is batch-wise. The electrodes life is limited. General remarks It is sensible to state, that one particular ideal AOP does not exist. The applicability depends on e.g. the nature of the target compound(s), the pollution magnitude and concentration, geographical location of the pollution and AOP performance stability. Extremely dangerous materials like 2,3,7,8-tetrachlorodibenzo-p-dioxin [1] have to be destructed regardless of the costs. With regard to high quantities, continuous-flow operation is preferable to batch-wise processing. Treatment of low concentration intermediate toxic waste flows should be performed with maximum efficiency. The AOP setup complexity and pollution magnitude determine whether it is affordable to build or position the setup at/near the location where the pollution is situated. The AOP performance stability is dependent on e.g. the process-technological complexity but also on the input flow composition stability; therefore every AOP should be continuously monitored by chemical-analytical measurement techniques on the processed waste flow to guard performance stability. With regard to the obtained results described in this thesis and mentioned corona literature references, pulsed corona technology can be considered as a true AOP. Its strongest feature is the efficient conversion of low concentration target compounds from both the liquid or gas phase, using simple technology. It is worth-wile to study scale-up possibilities. Then also a cost analysis can be derived. At this experimental stage of investigation, it is not possible to determine the process costs for a specific detoxification: unknown parameters are viz. the size of power supply, power charges, reactor design, target compound(s), destruction level, flow rate, pollution magnitude, maintenance costs, setup depreciation and analytical monitoring costs. 144 Chapter 5. 6. Conclusions 6.1. Pulsed corona discharges It has been demonstrated that the application of pulsed corona discharges in the gas phase over aqueous target compound solutions is able to degrade the target compound by oxidation. This application of corona in air has a more powerful effect than an ozone generator, because apart from ozone a wide variety of highly reactive species are produced like hydroxyl radicals, singlet oxygen, metastables, ions and UV photons. Hydroxyl radicals are produced in the gas phase by ionization and dissociation of water molecules at the corona discharge streamer tips. Also, hydroxyl radicals are produced indirectly, in the aqueous phase from ozone and UV photons, where ozone has been produced by the dissociation of oxygen molecules. The production of hydroxyl radicals in water has been investigated by ex-situ fluorescence spectrometry using a molecular probe and in-situ electron spin resonance. The molecular probe coumarin-3-carboxylic acid (CCA) has revealed a linear fluorescence increase with the corona exposure time, implying a constant hydroxyl radical production rate for the observed time span. Quantitative hydroxyl radical measurements using CCA-hydrogen peroxide standards have failed, due to instability of dilute hydrogen peroxide solutions and poor CCA solubility in water. In-situ electron spin resonance using the spin trap 5,5-dimethyl-1-pyrroline N-oxide (DMPO) has not revealed hydroxyl radicals, which has been attributed to an insufficient sensitivity compared to fluorescence spectrometry. The production rate of ozone has been determined by UV absorbance spectrometry. The ozone concentration increases with the load voltage, while negative corona always yields less ozone than positive corona at the same absolute value of the load voltage. The ozone concentration over aqueous phenol solutions has appeared to be lower than the ozone concentration over deionized water, which has been explained by the reaction of ozone with phenol. In the applied setups, ozone concentrations up to about 5.5⋅1022 m-3 in air and about 8.0⋅1022 m-3 in oxygen have been measured. The obtained ozone production efficiency is about 40 gO3/kWh which is quite favourable, because it has been achieved in humid air. The action of radicals, metastables and ions has been illustrated by the application of corona in argon and helium for the degradation of phenol in aqueous solution. Although under inert gas conditions no ozone can be produced, considerable conversion has been measured. This conversion is caused by the production of hydroxyl radicals from the dissociation of water by ions and metastables bombardment. From analyses of corona-exposed deionized water by electrical conductometry, ionexclusion chromatography and spectrochemical ICP analysis it has been derived, that nitric acid is formed from nitrogen oxides, produced by the corona discharges in air. From experiments on the effect of different electrode configurations on the conversion of phenol and decolorization of malachite green, it has been concluded, that a favourable electrode configuration consists of a capacitive configuration with a multipin anode. The dielectrical separation of cathode and anode avoids conductive currents in the water, which dissipate energy at the expense of contributing to radical production. The multipin anode produces a more efficiently-dimensioned reactive volume for the production of oxidizers than a single pin. 146 Chapter 6. With regard to the generally applied corona conditions, these are a voltage V=25 kV, a pulse repetition rate f=100 Hz, an anode-tip-to-water distance d=1.0 cm and a solution volume Vol≤500 ml, the measured pulse energy range is about 5 -15 mJ. The application of corona in the gas phase over aqueous target compound solutions has appeared to be more efficient than application of corona in the aqueous phase. This has been demonstrated by both a comparison of phenol efficiency values from literature with experimentally obtained values and pulse energy measurements in aqueous solution. Calorimetry-based pulse energy measurements in deionized water have yielded values of about 55-61 ±9 mJ at a voltage V=19 kV and pulse repetition rate f=100 Hz. The application of corona discharges in the liquid phase indeed produces radicals in the direct vicinity of the target compound. However, for creation of the discharges the liquid phase at the anode tip has to be evaporated, which is less efficient than the production of an oxidative environment in the gas phase over the aqueous target compound solution. The order of magnitude difference in pulse energy between liquid phase and gas phase corona is in accordance with the observed efficiency differences, see Table 5.1. 6.2. Oxidation of model compounds The oxidation of phenol in aqueous solution has been studied in detail. The phenol oxidation product mixture has a complex composition. Observed oxidation products are dihydroxybenzenes, trihydroxybenzenes, carboxylic acids and carbon dioxide. The following polyhydroxybenzenes have been identified: catechol, resorcinol, hydroquinone, pyrogallol and hydroxyhydroquinone. The possible existence of quinonoid compounds could not be confirmed. Several aliphatic carboxylic acids have been observed: formic, acetic, propionic, oxalic, malonic, maleic, succinic and glyoxylic acid; possibly also a muconic acid enantiomer. Except for the carboxyaldehyde glyoxylic acid, no aldehydes have been detected. Pathways have been constructed for the oxidation of phenol by the hydroxyl radical and oxygen and for phenol oxidation by ozone. The described mechanisms account for the production of polyhydroxybenzenes and a broad range of aliphatic carboxylic acids with polyfunctional groups. The measured conversion efficiency for a 500 ml 1.0 mM phenol solution by corona in air at a conversion X=24% is G=2.2⋅10-8 J/mol ≡ 0.21 (100eV)-1 ≡ 7.5 g/kWh. After 2 hours of oxidation of a 100 ml 1.0 mM phenol solution by corona in air and argon using the generally applied corona parameters, a quantitative product analysis has yielded the following approximate concentrations (air/argon values in mmol/l): phenol: 0.43/0.12 mM; hydroquinone: 0/0.01 mM; total dihydroxybenzenes: 0.08/0.28 mM; formic acid: 0.23/0.09 mM; oxalic acid: 0.39/0 mM; glyoxylic acid: 0.05/0.03 mM; glyoxal: 0/0 mM. The oxidation mechanism of phenol by corona in argon appears to be very different from the oxidation of phenol by corona in air. Degradation of phenol by corona in argon mainly yields polyhydroxybenzenes due to hydroxyl radicals produced from the dissociation of water by ion/metastables bombardment; in argon, ring-cleavage is of minor importance because of the absence of oxygen and ozone. Conclusions 147 Application of a Microtox ecotoxicity test on a series of 1 mM phenol solutions exposed to corona in air for different times, has yielded a substantial toxicity increase. This effect is mainly due to the presence of polyhydroxybenzene- and possibly also quinoneintermediate products; especially hydroquinone and 1.4-benzoquinone show very high ecotoxicity. The formation of carboxylic acids also accounts for the toxicity increase, but to a far less extent. The toxicity increase is only temporal, because the polyhydroxybenzenes, quinones and eventually also the carboxylic acids will be mineralized. The temporal toxicity increase is not a specific problem of corona technology, but is inherent to oxidation. Carbon dioxide has been detected in the gas phase over oxidized phenol solutions, but carbon monoxide and volatile aldehydes have not been identified. After 3 hours of oxidation using the standard corona conditions, about 0.8% carbon from a 1.0 mM phenol solution and about 6.3% carbon from a 0.1 mM phenol solution have been converted into CO2. The amount of CO2 produced by phenol oxidation by corona in air is about 2.7 times higher than the amount produced by corona in argon. The total organic carbon content appears to be rather constant during oxidation, which implies that for the observed oxidation time span the oxidation products mainly remain in the liquid phase, except for the small amounts of CO2 produced. The degradation of the atrazine herbicide, malachite green dye and dimethyl sulfide odor component by pulsed corona discharges has resulted as follows. The conversion efficiency of a 500 ml 0.12 mM atrazine solution oxidized by corona in air at a conversion level X=13 % is about 7.7⋅10-10 mol/J ≡ 7.4⋅10-3 (100eV)-1 ≡ 0.6 g/kWh. This low efficiency is inherent to persistent triazine herbicides. In order to achieve a 24% decolorization of malachite green at λ=590 nm, the degradation efficiency using several electrode configurations has been determined. An immersed electrode configuration with a single pin anode has yielded the lowest efficiency viz. about 7 mg/kWh; a capacitive electrode configuration with a 4-pin anode has yielded the highest efficiency: about 412 mg/kWh. Dimethyl sulfide has been degraded in the gas phase. The conversion of air containing 2.5 ppm dimethyl sulfide has been measured to be 45% after 15 seconds, 70% after 30 seconds and 94% after 1 minute. Corona treatment appears to be very effective for stench abatement. 6.3. Analytical techniques Ion-exclusion chromatography (ICE) has appeared to be much more powerful than reversed-phase HPLC for separation of the phenol oxidation product mixture. Although ICE has been designed for the separation of complex mixtures of carboxylic acids, also the polyhydroxybenzenes are well-separated mutually and do not interfere with the acids. On the contrary, the separation of individual polyhydroxybenzene isomers is best performed by reversed-phase HPLC. Identification of the complex phenol oxidation product mixture by mass spectrometry is necessary before quantitative analyses can be performed. However, both off-line LC-MS and electron-impact MS have yielded very limited information about the composition of the oxidation product mixture. The applied IonSpray and APCI LC-MS interfaces appear to be not applicable to the relatively low molecular weight phenol oxidation products. Also, the individual product concentrations are low. 148 Chapter 6. The applied solid phase extraction technique has appeared to be not successful. Therefore, based on literature, major oxidation products have been identified by comparison of retention times. For global in-field monitoring of oxidation progress of an organic compound, acidity increase by electrical conductometry is a very unpretentious option. In addition, simple devices based on detection by UV absorbance or fluorescence (when applicable) can be applied to monitor the conversion of organic materials. However for the case of turbid waste water, spectroscopic analysis techniques fail; then, TOC measurements are to be preferred. Due to sensitivity differences, in-situ ESR has not been able to resolve hydroxyl radicals, while fluorescence spectrometry using the CCA molecular probe has distinctly demonstrated hydroxyl radical production in aqueous solution. 6.4. Outlook The very favourable conversion efficiency justifies the continuation of pulsed coronainduced water treatment; topics of interest may be described as follows. Optimization of the reactor/electrode configuration can be achieved by increasing the contact area between the target compound solution and the corona discharges. Also an investigation of corona pulse duration and pulse shape may result in an even higher oxidizer production efficiency. Continuous flow application, scale-up possibilities and process stability should be studied. With regard to commercially applied organic electrosynthesis, it is interesting to investigate the possibilities of corona-induced synthesis. As electric discharge source for waste water treatment, the application of a flat dielectric barrier lamp may be promising; these xenon-excimer-based lamps produce 172 nm radiation at a very high efficiency of about 60%. High resolution LIF spectroscopy should be studied, in order to investigate the possibility to resolve the oxidation products by their rotational spectra. The identification of unresolved oxidation products might be improved by application of chemical derivatization prior to liquid- or gaschromatography / mass spectrometry. 7. References 1. Patnaik P., A comprehensive guide to the hazardous properties of chemical substances 1999, 2nd edn., New York: Wiley, ISBN 0-471-29175-7, 835-865 2. Bratsch, S.G, Standard electrode potentials and temperature coefficients in water at 298.15 K., J. Phys. Chem. Ref. Data 1989, 18 (1), 1-21 3. Lide D.R., Handbook of Chemistry and Physics 1999, 79th edn., Cleveland (OH): Chemical Rubber Co., ISBN 0-8493-9720-0, Electrochemical series, 8-21 - 8-31 4. Atkins P.W., Physical Chemistry 1982, 2nd ed., Oxford: Oxford University Press, ISBN 0-19-855151-7, 346-375 5. Rook J.J., Formation of haloforms during chlorination of natural waters, Water Treat. Exam. 1974, 23, 234-243 6. Atkinson R., Carter W.P.L., Kinetics and mechanisms of the gas-phase reactions of ozone with organic compounds under atmospheric conditions, Chem. Rev. 1984, 84, 437-470 7. Singer P.C., Gurol M.D., Dynamics of ozonation of phenol – 1 Experimental observations, Water Res. 1983, 17 (9), 1163-1171 8. Ullmann, Encyclopedia of Industrial Chemistry 1991, 5th edn. Vol. A18, Weinheim: Verlag Chemie, ISBN 3-527-20118-1, Ozone, 349-357 9. Kirk R.E., Othmer D.F., Encyclopedia of chemical technology 1996, 4th ed. Vol. 17, London: Wiley-Interscience, ISBN 0-471-52686-X, Ozone, 953-994 10. Masuda S., Akutsu K., Kuroda M., Awatsu Y., Shibuya Y., A ceramic-based ozonizer using high-frequency discharge, IEEE Trans. Ind. Appl. 1988, 24 (2), 223-231 11. Scheck C.K., Frimmel F.H., Degradation of phenol and salicylic acid by ultraviolet radiation/hydrogen peroxide/oxygen, Water Res. 1995, 29 (10), 2346-2352 12. Safarzadeh-Amiri A., Bolton J.R., Cater S.R., The use of iron in advanced oxidation processes, J. Adv. Oxid. Technol. 1996, 1 (1), 18-26 13. Ullmann, Encyclopedia of Industrial Chemistry 1989, 5th edn. Vol. A13, Weinheim: Verlag Chemie, ISBN 3-527-20113-0, Hydrogen peroxide, 443-466 14. Kirk R.E., Othmer D.F., Encyclopedia of chemical technology 1995, 4th ed. Vol. 13, London: Wiley-Interscience, ISBN 0-471-52682-7, Hydrogen peroxide, 961-995 150 Chapter 7. 15. Okamoto K., Yamamoto Y., Tanaka H., Tanaka M., Itaya A., Heterogeneous photocatalytic decomposition of phenol over TiO2 powder, Bull. Chem. Soc. Jpn. 1985, 58, 2015-2022 16. Bouquet-Somrani C., Finiels A., Graffin P. Olivé J.-L., Photocatalytic degradation of hydroxylated biphenyl compounds, Appl. Cat. B: Environm. 1996, 8, 101-106 17. Gopalan S., Savage P.E., Reaction mechanism for phenol oxidation in supercritical water, J. Phys. Chem. 1994, 98, 12646-12652 18. Devlin H.R., Harris I.J., Mechanism of the oxidation of aqueous phenol with dissolved oxygen, Ind. Eng. Chem. Fundam. 1984, 23, 387-392 19. Pan X.-M., Schuchmann M.N., Von Sonntag C., Oxidation of benzene by the OH radical. A product an pulse radiolysis study in oxygenated aqueous solution, J. Chem. Soc. Perkin. Trans. 2 1993, 289-297 20. Lin K., Cooper W.J., Nickelsen M.G., Kurucz C.N., Waite T.D., Decomposition of aqueous solutions of phenol using high energy electron beam irradiation – A large scale study, Appl. Radiat. Isot. 1995, 46 (12), 1307-1316 21. Hoffmann M.R., Hua I., Höchemer R., Application of ultrasonic irradiation for the degradation of chemical contaminants in water, Ultrason. Sonochem. 1996, 3, 163-172 22. Makino K., Mossoba M., Riesz P., Chemical effects of ultrasound on aqueous solutions. Formation of hydroxyl radicals and hydrogen atoms, J. Phys. Chem. 1983, 87, 1369-1377 23. Raizer Y.P., Gas discharge physics 1991, Berlin: Springer, ISBN 3-540-19462-2 24. Lide D.R., Handbook of Chemistry and Physics 1999, 79th edn., Cleveland (OH): Chemical Rubber Co., ISBN 0-8493-9720-0, Strengths of chemical bonds, 9-51 - 9-73 25. Lide D.R., Handbook of Chemistry and Physics 1999 79th edn., Cleveland (OH): Chemical Rubber Co., ISBN 0-8493-9720-0, Ionization energies of gas phase molecules, 10-181/182 26. Creyghton Y.L.M., Pulsed positive corona discharges, fundamental study and application to flue gas treatment, Ph.D. thesis, Eindhoven University of Technology, 1994, Eindhoven, The Netherlands, ISBN 90-9007232-2 27. Van Veldhuizen E.M., Electrical discharges for environmental purposes: fundamentals and applications 1999, New York: Nova Science, ISBN 1-56072-743-8 28. Mätzing H., Chemical kinetics of flue gas cleaning by irradiation with electrons, Adv. Chem. Phys. 1991, 80, 315-402 References 151 29. Klimkin V.F., Investigation of initial stage of electric discharge evolution in liquids and dense gases by the methods of superfast optical recording, Ph.D. Thesis 1975, Novosibirsk 30. Alkhimov A.P., Vorobyov V.V., Klimkin V.F., Ponomarenko A.G., Soloukhin R.I., Doklady AN SSSR 1970, 194, 1052-1054 31. Clements J.S., Sato M., Davis R.H., Preliminary investigation of prebreakdown phenomena and chemical reactions using a pulsed high-voltage discharge in water, IEEE Trans. Ind. Appl. 1987, IA-23, 224-235 32. Ullmann, Encyclopedia of Industrial Chemistry 1991, 5th edn. Vol. A19, Weinheim: Verlag Chemie, ISBN 3-527-20119-X, Phenol, 299-312 33. Kirk R.E., Othmer D.F., Encyclopedia of chemical technology 1996, 4th ed. Vol. 18, London: Wiley-Interscience, ISBN 0-471-52687-8, Phenol, 592-644 34. Patai S., The Chemistry of Functional Groups: The Chemistry of quinonoid compounds, Part 2, 1974, London: Wiley, ISBN 0-471-66932-6, 794-816 35. US Environmental Protection Agency, Drinking water health advisory: Pesticides 1989, Chelsea (MI): Lewis publishers, ISBN 0-87371-235-8, 43-67 36. Pelizzetti E., Maurino V. Minero C., Carlin V., Pramauro E. Zerbinati O., Tosato M.L., Photocatalytic degradation of atrazine and other s-triazines, Environ. Sci. Technol., 1990, 24, 1559-1565 37. Ullmann, Encyclopedia of Industrial Chemistry 1996, 5th edn. Vol. A27, Weinheim: Verlag Chemie, ISBN 3-527-20127-0, Cationic dyes, 189-226 38. Zollinger H., Color chemistry; syntheses, properties and applications of organic dyes and pigments, Weinheim: Verlag Chemie, ISBN 3-527-26200-8 39. Ullmann, Encyclopedia of Industrial Chemistry 1995, 5th edn. Vol. A26, Weinheim: Verlag Chemie, ISBN 3-527-20100-9, Sulfides, polysulfides and sulfanes, 773-785 40. Henschler D., Toxikologisch-arbeitsmedizinische Begründungen von MAK-Werten, gesundheitsschädliche Arbeitsstoffe 1975, Band IV, Dimethylsulfid, Weinheim: Verlag Chemie, ISSN 0930-1984 41. Rees J.A., Electrical breakdown in gases 1973, London: Macmillan, ISBN 0-333-13832-5 42. Buxton G.V., Greenstock C.L., Helman W.P., Ross A.B., Critical review of rate constants for reactions of hydrated electrons, hydrogen atoms and hydroxyl radicals (OH/O-) in aqueous solution, J. Phys. Chem. Ref. Data 1988, 17 (2), 513-531 43. Haugland R.P., Handbook of fluorescence probes 1996, 6th edn., Eugene (OR): Molecular Probes, ISBN 0-9652240-1-5, 484-493 152 Chapter 7. 44. Lide D.R., Handbook of Chemistry and Physics 1999 79th edn., Cleveland (OH): Chemical Rubber Co., ISBN 0-8493-9720-0, Solubility of selected gases in water, 8-86 - 8-89 45. Gurol M.D., Vatistas R., Oxidation of phenolic compounds by ozone and ozone and u.v. radiation: a comparative study, Water Res. 1987, 21 (8), 895-900 46. Perry R.H., Perry’s Chemical engineers handbook 1997, 7th edn., New York: McGraw-Hill, ISBN 0-07-049841-5, Physical and chemical data, solubilities, 2-128 47. Moeller T., Inorganic chemistry 1957, 6th edn., New York: Wiley, LCCC Nr. 52-7487, 484; 505 48. Patai S., The Chemistry of Functional Groups: The Chemistry of the hydroxyl group, Part 1, 1971, London: Interscience, ISBN 0-471-66939-3, 138-139 49. Baily P.S., Organic Chemistry - a series of monographs, Vol. 39-I, Ozonation in Organic Chemistry I 1978, London: Academic Press, ISBN 6-12-073101-0,111-113 50. Von Sonntag C., Schuchmann H.-P., Aufklärung von Peroxyl-Radikalreaktionen in wäßriger Lösung mit strahlenchemischen Techniken, Angew. Chem. 1991, 103, 1255-1279 51. Getoff N., Radiation and photoinduced degradation of pollutants in water, a comparative study, Radiat. Phys. Chem. 1991, 37, 673-680 52. Patai S., The Chemistry of Functional Groups: The Chemistry of the hydroxyl group, Part 1, 1971, London: Interscience, ISBN 0-471-66939-3, 284, 505-592 53. Gordon P.F., Gregory P., Organic chemistry in colour, Berlin: Springer, ISBN 3-540-11748-2, 281-298 54. McGraw-Hill, Encyclopedia of science and technology 1992, 7th edn., Vol. 14, ISBN 0-07-909206-3, Pyrolysis, 547-551 55. Guillard C., Pichat P., Huber G., Hoang-Van C., The GC-MS analysis of organic intermediates from the TiO2 photocatalytic treatment of water contamined by lindane, J. Adv. Oxid. Technol. 1996, 1 (1), 53-60 56. Hashimoto S., Miyata T., Washino M., Kawakami W., A liquid chromatographic study on the radiolysis of phenol in aqueous solution, Environ. Sci. Technol. 1979, 13 (1), 71-75 57. Omura K., Matsuura T., Photo-induced reactions-IX The hydroxylation of phenols by the photodecomposition of hydrogen peroxide in aqueous media, Tetrahedron. 1968, 24, 3475-3487 58. Swern D., Organic peroxides Vol. II 1970, London: Wiley-Interscience, ISBN 0-471-83961-2, 280 References 153 59. Joshek H.-I., Miller S.I., Photooxidation of phenol, cresols and dihydroxybenzenes, J. Am. Chem. Soc. 1966, July 20, 88:14, 3273-3281 60. Gopalan S., Savage P.E., A reaction network model for phenol oxidation in supercritical water, J.AIChE. 1995, 41 (8), 1864-1873 61. Kirk R.E., Othmer D.F., Encyclopedia of chemical technology 1996, 4th ed. Vol. 19, London: Wiley-Interscience, ISBN 0-471-52688-6, Polyhydroxybenzenes, 778-806 62. Roberts J.D., Caserio M.C., Basic principles of organic chemistry 1977, 2nd edn., Menlo Park (CA): W.A. Benjamin Inc., ISBN 0-8053-8329-8, 1306 63. Arnold S.M., Talaat R.E., Hickey W.J., Harris R.F., Identification of Fenton’s reagent-generated atrazine degradation products by HPLC and megaflow electrospray ionization tandem mass spectrometry, J. Mass. Spectr. 1995, 30, 452-460 64. Pulgarin C.O., Schwitzguebel J.-P., Peringer P.A., Pajonk G.M., Bandara J.S., Kiwi J.T., Abiotic oxidative degradation of atrazine on zero-valent iron activated by visible light, J. Adv. Oxid. Technol. 1996, 1 (1), 94-101 65. Adewuyi Y.G. and Carmichael G.R., Kinetics of oxidation of dimethyl sulfide by hydrogen peroxide in acidic and alkaline medium, Environ. Sci. Technol. 1986, 20, 1017-1022 66. Poole C.F. and Poole S.K., Chromatography today, 1995 4th edn., Amsterdam: Elsevier Science b.v., ISBN 0-444-89161-7 67. Heftmann E., Journal of Chromatography Library v 51A-B, A: Fundamentals and techniques 1992, 5th edn., Amsterdam: Elsevier, ISBN 0-444-88404-1 68. Snyder L.R., Kirkland J.J., Introduction to modern liquid chromatography 1974, New York: Wiley-Interscience, ISBN 0-471-81019-3 69. Lide D.R., Handbook of Chemistry and Physics 1999 79th edn., Cleveland (OH): Chemical Rubber Co., ISBN 0-8493-9720-0, Solvents for UV spectrophotometry, 8-120 70. Cole R.B., Electrospray ionization mass spectrometry: fundamentals, instrumentation and applications 1997, New York: Wiley-Interscience, ISBN 0-471-14564-5 71. Niessen W.M.A., Liquid Chromatography-Mass Spectrometry principles and applications 1992, New York: Dekker, ISBN 0-8247-8635-1 72. Eller, P.M., NIOSH manual of analytical methods, Vol. 1, NIOSH method 2539 15-5-1989, Cincinnati (OH): National Institute for Occupational Safety and Health 154 Chapter 7. 73. Bullock A.T, Gavin D.L., Ingram M.D., Electron spin resonance detection of spin-trapped radicals formed during the glow-discharge electrolysis of aqueous solutions, J.C.S. Faraday 1, 1980, 76, 846-653 74. Madden K.P., Taniguchi H., In situ radiolysis time-resolved ESR studies of spin trapping by DMPO: Reevaluation of hydroxyl radical and hydrated electron trapping rates and spin adduct yields, J. Phys. Chem. 1996, 100, 7511-7516 75. Pou S., Ramos, C.L., Gladwell T., Renks E., Centra M., Young D., Cohen M.S., Rosen G.M., A kinetic approach to the selection of a sensitive spin trapping system for the detection of hydroxyl radical, Anal. Biochem. 1994, 217, 76-83 76. Sibille J.-C., Doi K., Aisen P., Hydroxyl radical formation and iron-binding proteins, J. Biol. Chem. 1987, 262 (1), 59-62 77. ISO 11348-3 standard 1998: Water quality - Determination of the inhibitory effect of water samples on the light emission of Vibrio fischeri (Luminescent bacteria test). Part 3: Method using freeze-dried bacteria 78. Kaiser K.L.E., Devillers J., Handbooks of Ecotoxicological Data, Vol. 2: Ecotoxicity of Chemicals to Photobacterium Phosphoreum 1994, Gordon and Breach Science Publishers, ISBN 2-88124-974-4 79. DIN EN 1484 standard 1997: Water analysis – Guidelines for the determination of total organic carbon (TOC) and dissolved organic carbon (DOC) 80. Lide D.R., Handbook of Chemistry and Physics 1999 79th edn., Cleveland (OH): Chemical Rubber Co., ISBN 0-8493-9720-0, Dissociation constants of organic acids and bases, 8-46 - 8-56 81. Creyghton Y.L.M., Van Veldhuizen E.M., Rutgers W.R., Diagnostic techniqes for atmospheric streamer discharges, IEE Proc. Sci. Meas. Technol. 1994, 141 (2), 141-152 82. Lide D.R., Handbook of Chemistry and Physics 1999 79th edn., Cleveland (OH): Chemical Rubber Co., ISBN 0-8493-9720-0, Ionic conductivity and diffusion at infinite dilution, 5-93 - 5-95 83. Hesse M., Spektroskopische Methoden in der organischen Chemie 1995, 5. Aufl., Stuttgart: Thieme, ISBN 3-13-576105-3, 4 84. Seinfeld J.H., Atmospheric chemistry and physics of air pollution 1986, Chichester: Wiley-Interscience, ISBN 0-471-82857-2 85. Bowen E.J., Luminescence in chemistry 1968, London: D. van Nostrand, LCCC no. 68-16129, 77-117 86. Williams R.T., The fluorescence of some aromatic compounds in aqueous solution, J. Roy. Inst. Chem. 1950 (nov), 611-626 References 155 87. Stark J., Steubing W., Fluoreszenz und lichtelektrische Empfindlichkeit organischer Substanzen, Physik. Zeitschr. 1908, 15, 481-495 88. Bist H.D., Brand J.C.D., Williams D.R., The 2750-Å electronic band system of phenol, part I. The in-plane vibrational spectrum, J. Mol. Spec. 1966, 21, 76-98 89. Bist H.D., Brand J.C.D., Williams D.R., The 2750-Å electronic band system of phenol, part II. Extended vibrational assignments and band contour analysis, J. Mol. Spec. 1967, 24, 413-467 90. Christoffersen J., Hollas J.M., Kirby G.H., Rotational band contours in electronic spectra of large asymmetric top molecules: the 2750 Å system of phenol, Proc. Roy. Soc. A. 1968, 307, 97-110 91. Biaglow J.E., Kachur A.V., The generation of hydroxyl radicals in the reaction of molecular oxygen with polyphosphate complexes of ferrous ion, Rad. Res. 1997, 148, 181-187 92. Sherman W.R., Robins E., Fluorescence of substituted 7-hydroxycoumarins, Anal. Chem. 1968, 40 (4), 803-805 93. Atkins P.W., Physical chemistry, 1984 2nd edn., Oxford: Oxford University Press, ISBN 0-19-855151-7, 563-598 94. Lide D.R., Handbook of Chemistry and Physics 1999 79th edn., Cleveland (OH): Chemical Rubber Co., ISBN 0-8493-9720-0, Fundamental vibration frequencies of small molecules, 9-76 95. Morrison R.T., Boyd R.N. Organic chemistry, 1983 4th ed., Boston: Allyn and Bacon, ISBN 0-205-07802-8, 680-688 96. Lide D.R., Handbook of Chemistry and Physics 1999 79th edn., Cleveland (OH): Chemical Rubber Co., ISBN 0-8493-9720-0, Standard thermodynamic properties of chemical substances, 5-5 - 5-61 97. Bilski P., Reszka K., Bilska M. Chignell C.F., Oxidation of the spintrap 5,5-dimethyl-1-pyrroline N-oxide by singlet oxygen in aqueous solution, J. Am. Chem. Soc. 1996, 118, 1330-1338 98. Samoilovich V.G., Gibalov V.I., Kozlov K.V., Physical chemistry of the barrier discharge 1997, 2. Aufl., Düsseldorf: DVS Verlag, ISBN 3-87155-744-7 99. Lide D.R., Handbook of Chemistry and Physics 1999 79th edn., Cleveland (OH): Chemical Rubber Co., ISBN 0-8493-9720-0, Permittivity of liquids, 6-139 - 6-161 100. Joshi A.A., Locke B.R., Arce P., Finney W.C., Formation of hydroxyl radicals, hydrogen peroxide and aqueous electrons by pulsed streamer corona discharge in aqueous solution, J. Hazard. Mat. 1995, 41, 3-30 156 Chapter 7. 101. Šunka P., Babický V., Člupek M., Lukeš P., Šimek M., Schmidt J., Černák M., Generation of chemically active species by electrical discharges in water, Plasma Sources Sci. Technol. 1999, 8, 258-265 102. Sun B., Sato M. Clements J.S., Use of a pulsed high-voltage discharge for removal of organic compounds in aqueous solution, J. Phys. D: Appl. Phys. 1999, 32, 1908-1915 103. Van Veldhuizen E.M., Rutgers W.R., Hoeben W.F.L.M., Baede A.H.F.M., Babaeva N.Yu., Progress in Plasma Processing of Materials 1999, eds. Fauchais J., Amouroux J., New York: Begell House, ISBN 1-56700-126-2, 493 104. Hoeben W.F.L.M., Van Veldhuizen E.M., Rutgers W.R., Kroesen G.M.W., Gas phase corona discharges for oxidation of phenol in an aqueous solution, J. Phys. D: Appl. Phys. 1999, 32, L133-L137 105. Hoeben W.F.L.M., Van Veldhuizen E.M., Rutgers W.R., Cramers C.A.M.G., Kroesen G.M.W., The degradation of aqueous phenol solutions by pulsed positive corona discharges, accepted for publication in Plasma Sources Sci. Technol. (2000) 106. Ullmann, Encyclopedia of Industrial Chemistry 1989, 5th edn. Vol. A12, Weinheim: Verlag Chemie, ISBN 3-527-20112-5, Glyoxal, 491-494 107. Merck & co., The Merck Index, an encyclopedia of chemicals, drugs and biologicals 1997, 12th edn., New York: Chapman & Hall, ISBN 0-412-82910-X, M. nr. 6381 108. Beilsteins Handbuch der organischen Chemie 4. Aufl., EW 3, Band II, 2. Teil, Syst. nr. 163-194, 671/1995-1996, Berlin: Springer 109. Hayashi D., Hoeben W.F.L.M., Dooms G., Van Veldhuizen E.M., Rutgers W.R., Kroesen G.M.W., LIF diagnostic for pulsed corona induced degradation of phenol in aqueous solution, J. Phys. D: Appl. Phys. 2000, 33, 110. Hayashi D., Hoeben W.F.L.M., Dooms G., Van Veldhuizen E.M., Rutgers W.R., Kroesen G.M.W., Laser induced fluorescence spectroscopy for aqueous phenol and intermediate products in degraded solutions by pulsed corona discharges, submitted to Appl. Opt. (2000) 111. Lide D.R., Handbook of Chemistry and Physics 1999 79th edn., Cleveland (OH): Chemical Rubber Co., ISBN 0-8493-9720-0, Van der Waals constants for gases, 6-43 112. Lewis R.J., Sax’s dangerous properties of industrial materials 1992, 8th edn., New York: Van Nostrand Reinholt, ISBN 0-442-01132-6 113. Ullmann, Encyclopedia of Industrial Chemistry 1994, 5th edn. Vol. A25, Weinheim: Verlag Chemie, ISBN 3-527-20125-4, Dimethyl sulfoxide, 498 References 157 114. Van Veldhuizen E.M., Rutgers W.R., Bityurin V.A., Energy efficiency of NO removal by pulsed corona discharges, Plasma Chem. Plasma Process. 1996, 16(2), 227-247 115. Lide D.R., Handbook of Chemistry and Physics 1999 79th edn., Cleveland (OH): Chemical Rubber Co., ISBN 0-8493-9720-0, Electron affinities, 10-147 - 10-159 116. Ullmann, Encyclopedia of Industrial Chemistry 1987, 5th edn. Vol. A9, Weinheim: Verlag Chemie, ISBN 3-527-20109-2, Electrochemistry, 183-254 117. Sengupta S.K., Singh O.P., Contact glow discharge electrolysis: a study of its onset and location, J. Electroanal. Chem. 1991, 301, 189-197 118. Mazzocchin G.A., Magno F., Bontempelli G., Glow discharge electrolysis on ammonia in aqueous solution, Electroanal. Chem. and Interfacial Electrochem. 1973, 45, 471-482 119. Almubarak M.A., Wood A., Chemical action of glow discharge electrolysis on ethanol in aqueous solution, J. Electrochem. Soc. 1977, 124, 1356-1360 120. Van Veldhuizen E.M., Rutgers W.R., Pulsed positive corona in water, Proc. 12th Int. Symp. Plasma Chem. (ISPC-12), eds. Heberlein J.V., Ernie B.W., Roberts J.T., Minneapolis USA 1995, Vol. 2, 1083-1089 121. Sun B., Sato M., 1999, private communication 122. Harmsen G.H., Toxiciteitsaspecten bij chemische oxidatie in combinatie met biologische behandeling, Directoraat Generaal RWS-RIZA, rapport 98.004 123. Moeller T., Inorganic chemistry 1957, 6th edn., New York: Wiley, LCCC Nr. 52-7487, 689 124. Glaze W.H., Reaction products of ozone: a review, Environ. Health Perspect. 1986, 69, 151-157 125. Ullmann, Encyclopedia of Industrial Chemistry 1991, 5th edn. Vol. A18, Weinheim: Verlag Chemie, ISBN 3-527-20118-1, Oxocarboxylic acids, 313-319 126. Ullmann, Encyclopedia of Industrial Chemistry 1985, 5th edn. Vol. A1, Weinheim: Verlag Chemie, ISBN 3-527-20101-7, Aldehydes, aliphatic and araliphatic, 321-352 127. Wirth W., Gloxhuber C., Toxikologie 1981, 3. Aufl., Stuttgart: Thieme, ISBN 3-13-421103-3 128. Henschler D., Toxikologisch-arbeitsmedizinische Begründungen von MAK-Werten, gesundheitsschädliche Arbeitsstoffe 1994, Band IV – 1,4-Dihydroxybenzol, Weinheim: Verlag Chemie, ISSN 0930-1984 129. Sigma-Aldrich Material Safety Data Sheets; product no. 260150, 8-10/1998; regular MSDS updates of applied reagents listed by Table 3.1 158 Chapter 7. 130. Sun L., Hoy A.R., Bolton J.R., Generation efficiency of the hydroxyl radical adduct of the DMPO spin trap in homogeneous and heterogeneous media, J. Adv. Oxid. Technol. 1996, 1 (1), 44-52 131. Becker R.S., Theory and interpretation of fluorescence and phosphorescence 1968, New York: Wiley-Interscience, SBN 471-06126-3, 239 132. Bolton J.R., Valladares J.E., Zanin J.P., Cooper W.J., Nickelsen M.G., Kajdi D.C., Waite T.D., Kurucz C.N., Figures-of-merit for Advanced Oxidation Technologies: a comparison of homogeneous UV/H2O2, heterogeneous UV/TiO2 and electron beam processes, J. Adv. Oxid. Technol. 1998, 3 (2), 174-181 133. Barlow S.M., Sullivan F.M., Reproductive hazards of industrial chemicals 1982, London: Academic Press, ISBN 0-12-078960-4, 417-421 134. Pal S.B., Handbook of laboratory health and safety measures 1990, 2nd edn., Dordrecht: Kluwer Academic Publishers, ISBN 0-7462-0077-3, 427-455 Summary Since recently, some advanced oxidation processes (AOP’s) have been applied for the degradation of harmful materials, as alternative to microbiological waste water treatment or as add-on technology. The AOP focuses on the degradation of persistent toxic materials by radical-induced oxidation; for this purpose, halogen- and metal-free oxygen-based oxidizers are utilized, especially hydroxyl radicals. By advanced oxidation a persistent toxic compound is converted into microbiologically degradable products, or is even mineralized to carbon dioxide, water and -depending on the nature of the target compound- inorganic ions like e.g. nitrate, sulfate and phosphate. Hydrogen peroxide/UV, ozone/UV and wet oxidation AOP’s have already been applied on modest scale; the status of electrical discharge, electron-beam and photocatalytic AOP’s is still experimental. This thesis describes the degradation of organic materials in aqueous solution by pulsed corona discharges. These electrical discharges have been applied in the gas phase over the target compound solution. As a result of the extremely high electric field strength (about 200 kV/cm) at the head of the discharge channels, locally present molecules of the dielectric are dissociated, excited or ionized. In humid air the following reactive species are produced: hydroxyl radicals, oxygen atoms, ozone, nitrogen metastables, ions and UV photons. The pulsed corona technology has been studied in detail using the model compound phenol (hydroxybenzene), which is an important precursor in organic chemical synthesis. Also the degradation of some other model compounds has been studied viz. atrazine (herbicide), malachite green (dye) and dimethyl sulfide (odor component). A favourable electrode configuration has been derived from the experiments. In addition to the oxidation of model compounds, also the production of oxidizers has been investigated. The production of hydroxyl radicals in aqueous solution has been studied by fluorescence spectrometry in combination with the fluorescent molecular probe coumarin-3-carboxylic acid (CCA); also in-situ electron spin resonance has been applied using the spin trap 5,5-dimethyl-1-pyrroline N-oxide (DMPO). UV absorption spectrometry has been applied for the quantitative analysis of ozone. The conversion level, conversion efficiency and oxidation mechanisms of the model compound phenol have been investigated. Conversion and oxidation pathways have been determined by means of the liquid chromatographic techniques reversed-phase HPLC and ion-exclusion chromatography; in addition, IonSpray and electron-impact mass spectrometry have been applied. Also, phenol oxidation has been studied in-situ using laser-induced fluorescence spectroscopy. The gas phase over the oxidized aqueous phenol solution has been analyzed by infrared spectroscopy and an aldehyde screening test. From an environmental point of view, Microtox ecotoxicity tests and total organic carbon measurements have been performed on fresh and oxidized phenol solutions. Reaction mechanisms have been derived for phenol oxidation by hydroxyl radicals, oxygen and ozone. 160 __ The conversion of atrazine has been performed by reversed-phase HPLC. Dimethyl sulfide has been oxidized in the gas phase and the conversion has been determined by gas chromatography. The decolorization of malachite green due to oxidation by corona has been determined for several electrode configurations using absorption spectrometry. The most efficient electrode configuration consists of a multipin anode situated in the gas phase over the aqueous solution of the target compound, while the cathode is dielectrically separated from the anode. In this way an effectively dimensioned plasma is produced and energy dissipation due to vapour formation is avoided. During oxidation of aqueous solutions of the fluorescent molecular probe CCA, the fluorescence intensity appears to increase linearly with the oxidation time, which implies a constant hydroxyl radical production rate in water. Calibration of the fluorescence intensity as a function of the amount of hydroxyl radicals added using hydrogen peroxide and CCA has appeared to be impossible. Identification of hydroxyl radicals by in-situ ESR has been unsuccessful, although the spin trap DMPO has been utilized to produce a stable adduct. The ozone concentration in air over water increases with the applied load voltage, while positive corona produces more ozone than negative corona. The measured ozone production efficiency is about 40 g/kWh which is high, because the ozone has been produced in humid air. During oxidation the ozone concentration over aqueous phenol solutions is significantly lower than the ozone concentration over deionized water, which is explained by the reaction of ozone with phenol. The oxidation of phenol in aqueous solution by pulsed corona discharges in air yields a complex mixture of oxidation products. Polyhydroxybenzenes are produced, which yield aliphatic aldehydes and carboxylic acids by ring-cleavage. The derived oxidation pathways explain the formation of polyhydroxybenzenes and aliphatic polyfunctional hydrocarbons. The intermediate oxidation products are more harmful than phenol, therefore thorough oxidation progress is required; the temporal toxicity increase is no specific problem occurring from corona technology, but is inherent to oxidation. Except for small amounts of carbon dioxide, no other gaseous phenol oxidation products have been identified. This is in accordance with an observed nearly constant total organic carbon level of the aqueous solution during oxidation. There appear to be large differences between oxidation products obtained by corona in air and corona in argon. By corona in argon mainly hydroxylation of phenol takes place, while in the presence of air also ring-cleavage takes place. The measured efficiency at 24% phenol conversion by corona in air is about 22 nanomol/J, which corresponds to 7.5 g/kWh. Atrazine and malachite green are very stable compounds, which appears from the conversion efficiency: the efficiency of atrazine is 0.6 g/kWh at 27% conversion, while the efficiency of malachite green is 0.4 g/kWh at 24% absorption decrease at λabs=590 nm with regard to the most favourable electrode configuration. On the contrary, dimethyl sulfide is readily converted. Samenvatting Ten behoeve van de afbraak van schadelijke chemische verbindingen worden sinds enige tijd enkele geavanceerde oxidatieprocessen (AOP’s) toegepast als vervanger voor -of in combinatie met- conventionele microbiologische afvalwaterbehandeling. Een AOP beoogt de degradatie van persistente toxische verbindingen via radicalaire oxidatie; hierbij wordt gebruik gemaakt van halogeen- en metaalvrije, op zuurstof gebaseerde oxidatoren, met name hydroxylradicalen. Door geavanceerde oxidatie wordt een persistente toxische verbinding gedegradeerd tot microbiologisch afbreekbare verbindingen, of eventueel gemineraliseerd tot koolstofdioxide, water en -afhankelijk van de aard van de verbinding- anorganische ionen zoals bijvoorbeeld nitraat, sulfaat en fosfaat. Van de AOP’s worden waterstofperoxide/UV, ozon/UV technologie en nattelucht oxidatie reeds op beperkte schaal toegepast; elektrische ontladingen, elektronenbundel technologie en fotokatalytische oxidatie bevinden zich nog in de experimentele fase. In dit proefschrift wordt de afbraak beschreven van organische verbindingen in waterige oplossing met behulp van gepulste corona-ontladingen. Deze elektrische ontladingen zijn toegepast in de gasfase boven the oplossing van de doelcomponent. Ten gevolge van de extreem hoge elektrische veldsterkte (ongeveer 200 kV/cm) aan de kop van de corona-ontladingskanalen, worden aldaar aanwezige moleculen van het diëlectricum gedissocieerd, geëxciteerd of geïoniseerd. In vochtige lucht worden aldus de volgende reactieve deeltjes geproduceerd: hydroxylradicalen, zuurstofatomen, ozon, stikstofmetastabielen, ionen en UV fotonen. De gepulste corona technologie is gedetailleerd getoetst op de modelstof fenol (hydroxybenzeen), een belangrijke precursor in de organisch-chemische synthese. Tevens is de degradatie van enkele andere modelcomponenten bestudeerd, te weten atrazine (herbicide), malachiet groen (kleurstof) en dimethyl sulfide (geurcomponent). Uit de experimenten is een gunstige elektrodenconfiguratie naar voren gekomen. Naast de oxidatie van doelcomponenten is ook de produktie van oxidatoren onderzocht. De produktie van hydroxylradicalen in waterige oplossing is bestudeerd met fluorescentie spectrometrie in combinatie met de fluorescerende moleculaire probe coumarine-3-carboxylzuur (CCA); tevens is in-situ elektron spin resonantie toegepast met behulp van de spin trap 5,5-dimethyl-1-pyrroline N-oxide (DMPO). UV absorptiespectrometrie is toegepast voor de kwantitatieve analyse van ozon. Van de modelstof fenol zijn omzettingsgraad en efficiëntie alsmede de oxidatiemechanismen onderzocht. De conversie en oxidatieroutes zijn bepaald met behulp van de vloeistofchromatografische technieken reversed-phase HPLC en ionexclusie chromatografie; ook zijn IonSpray en electron-impact massaspectrometrie toegepast. Tevens is in-situ de oxidatie van fenol bestudeerd met laser-geïnduceerde fluorescentiespectroscopie. De gasfase boven de geoxideerde fenoloplossing is geanalyseerd met infrarood spectroscopie en een aldehyde test. Vanuit milieutechnisch oogpunt zijn Microtox ecotoxiciteitstests en totaal organisch koolstof analyses uitgevoerd op onbehandelde en geoxideerde fenoloplossingen. Reactiemechanismen zijn opgesteld voor de oxidatie van fenol door hydroxylradicalen, zuurstof en ozon. 162 __ De conversie van atrazine is bepaald met reversed-phase HPLC. Dimethylsulfide is geoxideerd in de gasfase en de conversie is gemeten met gaschromatografie. De ontkleuring van malachietgroen ten gevolge van oxidatie door corona ontladingen is gemeten voor diverse elektrodenconfiguraties door middel van absorptiespectrometrie. De meest efficiënte elektrodenconfiguratie bestaat uit een meerpunts anode in de gasfase boven de waterige oplossing van de doelcomponent, waarbij de kathode diëlektrisch gescheiden is van de anode. Aldus wordt een effectief gedimensioneerd plasma gevormd en wordt energieverlies door dampvorming voorkomen. Bij oxidatie van waterige oplossingen van de fluorescerende moleculaire probe CCA, blijkt de fluorescentie intensiteit lineair toe te nemen met de oxidatietijd, waaruit een constante produktiesnelheid van hydroxylradicalen in water is gebleken. Calibratie van de fluorescentie-intensiteit als functie van de hoeveelheid toegevoegde hydroxylradicalen met behulp van waterstofperoxide en CCA is niet mogelijk gebleken. Het is niet gelukt, om hydroxylradicalen aan te tonen met behulp van in-situ ESR, ondanks dat een spin trap is toegepast voor de vorming van een stabiel adduct. De ozonconcentratie in lucht boven water neemt toe met de aangelegde spanning, terwijl positieve corona meer ozon produceert dan negatieve corona. De gemeten produktie efficiëntie van ozon bedraagt ongeveer 40 g/kWh en is hoog, temeer daar de ozon in vochtige lucht geproduceerd is. Tijdens oxidatie blijkt de ozonconcentratie boven fenoloplossingen significant lager dan de ozonconcentratie boven gedeïoniseerd water, hetgeen verklaard wordt door de reactie van ozon met fenol. De oxidatie van fenol in waterige oplossing door middel van gepulste corona ontladingen in lucht levert een complex mengsel van oxidatieprodukten. Geproduceerd worden polyhydroxybenzenen, waaruit door ringopening alifatische aldehyden en carboxylzuren ontstaan. De opgestelde modellen verklaren de produktie van polyhydroxybenzenen en alifatische polyfunctionele koolwaterstoffen. De intermediaire oxidatieprodukten blijken schadelijker dan fenol, waardoor grondige oxidatie dus vereist is. De aanvankelijke toxiciteitstoename is geen specifiek probleem van corona technologie, doch is inherent aan oxidatie. Behalve kleine hoeveelheden koolstofdioxide, zijn geen andere gasvormige oxidatieprodukten van fenol aangetroffen. Dit stemt overeen met een waargenomen nagenoeg constante waarde van het totaal organisch koolstofgehalte van de vloeistoffase tijdens oxidatie. Er blijken grote verschillen te bestaan tussen de oxidatieprodukten van fenol, die ontstaan zijn door behandeling met corona in lucht en corona in argon. Voor het geval van corona in argon treedt hoofdzakelijk hydroxylering van fenol op, terwijl in aanwezigheid van zuurstof ook ringopening optreedt. De gemeten conversie efficiëntie bij 24% fenol conversie door corona in lucht bedraagt ongeveer 22 nanomol/J, wat overeenkomt met 7.5 g/kWh. Atrazine en malachietgroen zijn zeer stabiele verbindingen, hetgeen blijkt uit de conversie efficiëntie: deze bedraagt voor atrazine 0.6 g/kWh bij 27% conversie en voor malachietgroen 0.4 g/kWh bij 24% absorptievermindering bij λabs=590 nm voor de meest gunstige elektrodenconfiguratie. Dimethylsulfide wordt daarentegen snel omgezet. Dankwoord / Acknowledgements Dit multidisciplinaire proefschrift is tot stand gekomen mede dankzij de inzet van een groot aantal personen. Allereerst dank ik eerste promotor Wijnand Rutgers en co-promotor Eddie van Veldhuizen bijzonder voor toekenning van de promotieplaats, intensieve begeleiding, enthousiasme en een zeer prettige samenwerking. Tevens dank ik Gerrit Kroesen, Frits de Hoog en Bram Veefkind voor hun belangrijke bijdrage aan mijn promotieonderzoek. Tweede promotor Carel Cramers en begeleider Henk Claessens (beiden faculteit Scheikundige Technologie / capaciteitsgroep Instrumentele Analyse) dank ik voor hun ondersteuning op chemisch-analytisch gebied en het beschikbaar stellen van laboratoriumfaciliteiten. René Janssen (faculteit Scheikundige Technologie / capaciteitsgroep Macromoleculaire en Organische chemie) dank ik voor de diverse discussies en voor de leiding bij de ESR experimenten. Voor chemisch-analytische assistentie ben ik verder dank verschuldigd aan Marion van Straten, Eric Vonk, Jan Jiskra, Ber Vermeer, Marc van Lieshout, Joost van Dongen, Gius Rongen, Henri Snijders en Hans Damen. Ook dank aan Denise Tjallema. Het was geweldig vertoeven bij de capaciteitsgroep Elementaire Processen in Gasontladingen; bedankt voor de leuke tijd en samenwerking: Loek Baede (speciaal voor de technische realisatie van corona en voor fotografie), Lambert Bisschops, Leon Bakker, Rina Boom, Jean-Charles Cigal, Jurgen van Eck, Marjan van de Elshout, Hans Freriks, Marc van de Grift, Charlotte Groothuis, Gerjan Hagelaar, Daiyu Hayashi (special thanks for his important contribution to LIF spectroscopy), Marcel Hemerik, Carole Maurice, Gabriela Paeva, Koen Robben, Eva Stoffels, Winfred Stoffels en Geert Swinkels. En niet te vergeten mijn oud-collega’s van Elektrotechniek (EG): Vadim Banine, Hub Bonné, Frank Commissaris, Ad Holten, Ad van Iersel, Herman Koolmees, Roel Moors en Mariet van Rixtel. Dank voor jullie bijdrage: afstudeerders Roy Janssen, Hens Renierkens, Ralph van Eijk; stagiairs Geert Dooms, Coen van de Vin, Chris Peters, Marco van Steen, Martijn Siffels. Verder dank ik: Marius Bogers en medewerkers van de faculteitswerkplaats N voor de vervaardiging van onderdelen ten behoeve van experimentele opstellingen. Marco van de Sande (N/ETP) voor uitvoering van de spectrochemische ICP metingen. Gerard Harmsen (Directoraat-Generaal RWS-RIZA, Lelystad) voor advies over ecotoxicologische tests. KEMA KPG/CET Arnhem voor uitvoering van de TOC metingen. Thanks to Pavel Šunka (Institute of Plasma Physics of the Czech Academy of Sciences, Prague CR) and Bruce Locke (Florida State University, Tallahassee USA) for received hospitality and pleasant discussions. Tenslotte dank ik speciaal mijn ouders, broer en zus voor hun grote belangstelling en ondersteuning. Curriculum Vitae: Daar op de basisschool al behoefte bestond aan een eigen chemisch laboratorium, is de auteur (geboortedatum 02-07-1966) na afronding van het VWO Peelland College te Deurne in 1985 Scheikundige Technologie gaan studeren aan de toenmalige TH Eindhoven. Na zijn afstuderen in 1990 als polymeertechnoloog is hij werkzaam geweest bij NKF Kabel in Delft als chemisch technoloog en ook als plaatsvervangend chef van het materialen laboratorium energie aldaar. In november 1995 is hij gestart met dit promotieonderzoek bij de faculteit Elektrotechniek / vakgroep Elektrische Energiesystemen, in samenwerking met de faculteit Scheikundige Technologie / capaciteitsgroep Instrumentele Analyse, waarna het onderzoek is voortgezet en afgerond bij de faculteit Technische Natuurkunde / capaciteitsgroep Elementaire Processen in Gasontladingen (EPG). Het promotieonderzoek heeft geleid tot dit proefschrift. A youthful passion to own a private chemical laboratory convinced the author (date of birth 02-07-1966) to apply for the chemical engineering curriculum at Eindhoven University of Technology, after finishing his pre-university education at Peelland College Deurne in 1985. After his MSc degree in polymer engineering in 1990, he joined NKF Kabel in Delft as chemical engineer and also as deputy supervisor of the materials laboratory. In November 1995 he started this PhD project at the faculty of Electrical Engineering / Electrical Energy Systems group in cooperation with the faculty of Chemical Engineering / Instrumental Analysis group; he continued and finished the project at the faculty of Applied Physics / department of Elementary Processes in Gas Discharges (EPG). The PhD project has resulted in this dissertation. Stellingen behorende bij het proefschrift “Pulsed corona-induced degradation of organic materials in water” door W.F.L.M. Hoeben 1. De gepulste corona-ontladingen technologie omvat meer dan alleen de produktie van ozon uit zuurstof. [dit proefschrift] 2. Ten behoeve van waterreiniging is het toepassen van gepulste corona-ontladingen in de gasfase efficiënter dan toepassing in de vloeistoffase. [dit proefschrift] 3. Bij toepassen van gepulste corona-ontladingen in de gasfase boven een waterige oplossing van de doelverbinding, is de chemische samenstelling van de gasfase van grote invloed op het degradatiemechanisme van de doelverbinding. [dit proefschrift] 4. Een ion-exclusie kolom is vanwege het carboxylzuur-specifieke retentie mechanisme zeer geschikt voor de scheiding van oxidatieprodukten mengsels van organische verbindingen. [dit proefschrift] 5. Bij degradatie van een organische verbinding door oxidatie kunnen intermediaire oxidatieprodukten ontstaan, die schadelijker zijn dan de doelverbinding. [dit proefschrift] 6. Simulatie beperkt milieuvervuiling. 7. Kunstmatige mineralisatie is energetisch gezien inefficiënt. 8. Het potentieel van geavanceerde oxidatie technologie is hoofdzakelijk gelegen in het onschadelijk maken van organische probleemstoffen. 9. Milieubewustheid komt nog steeds niet altijd gelegen. 10. Het is niet mogelijk, om milieukundig onderzoek te verrichten zonder enig chemisch afval te produceren. 11. World Wide Web adressen zijn vanwege het vluchtige karakter geen bruikbare literatuurreferenties. 12. Het weggooien van een proefschrift is een vorm van energiedissipatie. 13. Hoge schoorstenen verlangen veel wind. 14. Allesreinigers reinigen niet alles. 15. Een goed tandarts is zeker zo belangrijk als een goed gebit.