Hydroxyl-Radical-Induced Degradative Oxidation of β-Lactam

J. Phys. Chem. A 2010, 114, 8391–8395

8391

Hydroxyl-Radical-Induced Degradative Oxidation of β-Lactam Antibiotics in Water: Absolute Rate Constant Measurements Michelle K. Dail and Stephen P. Mezyk* Department of Chemistry and Biochemistry, California State UniVersity at Long Beach, 1250 Bellflower BouleVard, Long Beach, California 90840 ReceiVed: May 18, 2010; ReVised Manuscript ReceiVed: July 13, 2010

The β-lactam antibiotics are some of the most prevalent pharmaceutical contaminants currently being detected in aquatic environments. Because the presence of any trace level of antibiotic in water may adversely affect aquatic ecosystems and contribute to the production of antibiotic-resistant bacteria, active removal by additional water treatments, such as using advanced oxidation and reduction processes (AO/RPs), may be required. However, to ensure that any AOP treatment process occurs efficiently and quantitatively, a full understanding of the kinetics and mechanisms of all of the chemical reactions involved under the conditions of use is necessary. In this study, we report on our kinetic measurements for the hydroxyl-radical-induced oxidation of 11 β-lactam antibiotics obtained using electron pulse radiolysis techniques. For the 5-member ring species, an average reaction rate constant of (7.9 ( 0.8) × 109 M-1 s-1 was obtained, slightly faster than for the analogous 6-member ring containing antibiotics, (6.6 ( 1.2) × 109 M-1 s-1. The consistency of these rate constants for each group infers a common reaction mechanism, consisting of the partitioning of the hydroxyl radical between addition to peripheral aromatic rings and reaction with the central double-ring core of these antibiotics. Introduction The removal of trace amounts of contaminant pharmaceuticals from aquatic environments has received considerable attention in recent years. It is now well recognized that these chemicals often persist at significant levels even after standard water treatment.1 Of the many contaminants identified to date, one of the most prevalent classes is the β-lactam antibiotics. These particular antibiotics have been extensively used for many years,2,3 not only for human consumption but also in agriculture to prevent infection in plants and livestock.4-6 Such widespread usage inevitably leads to the excretion of these antibiotics into water systems. Although no formal water restriction exists for these drugs, their presence, even at trace levels5,7-10 has contributed to the production of strains of bacteria that are resistant to important classes of antibiotics.6 In addition, some humans are hypersensitive to even trace levels of antibiotics in water. Therefore, active removal of antibiotic contaminants may be necessary under some planned water use, or reuse, applications. Many new technologies have been investigated for the purpose of quantitatively removing residual antibiotics from waters. Examples include photodegradation,11,12 membrane filtration,13,14 activated carbon filtration,15,16 chlorine dioxide treatment,17 and ozonation.18-24 It has been reported that membrane filtration and activated carbon adsorption will remove antibiotics from water, provided only low levels of dissolved organic carbon (DOC) are present. However, this approach also results in waste spent activated carbon and membrane retentate which subsequently require disposition.24 The effective use of ozone is also greatly reduced at higher levels of DOC, thereby increasing treatment cost.19 It has recently been shown that ClO2 treatment can remove some β-lactam antibiotics from water17 but the rates of these reactions vary considerably. * To whom correspondence should be addressed. Phone: 562-985-4649. Fax: 562-985-8557. E-mail: [email protected].

Radical-based treatment processes through contaminant oxidation (advanced oxidation processes, AOPs) continue to gain interest as the preferred technology for the removal of trace amounts of contaminant chemicals in different quality waters.5 These treatment technologies utilize the in situ formation of radicals that react with and destroy chemical contaminants in the water. Common AOPs include ozone, UV/ozone, and UV/ H2O2, which involve oxidation via the generated hydroxyl radical (•OH). Other processes also generate other species; advanced oxidation and reduction processes (AO/RPs) such as heterogeneous catalysis by TiO2, sonolysis, and γ radiation and electron beams produce a mixture of oxidizing •OH radicals as ), and hydrogen well as the reducing hydrated electrons (eaq • atoms ( H). The kinetics and mechanisms of all the chemical reactions involved under the conditions of use must be fully understood to ensure that any AO/RP treatment process occurs efficiently and quantitatively.25 In support of such AO/RP utilization, a recent study of the kinetics and mechanisms of hydroxyl radical and hydrated electron reactions with three common β-lactam antibiotics (Penicillin-G, Penicillin-V, Amoxicillin) was reported.26 However, in aerated antibiotic-contaminated waters, it would be expected that any AO/RP-formed hydrated electron would react primarily with dissolved oxygen ([O2] ) 2.5 × 10-4 M, k ) 1.9 × 1010 M-1 s-1 27) to give the relatively unreactive superoxide radical. Therefore, the oxidizing hydroxyl radical reactions would be expected to dominate antibiotic degradation. These measured hydroxyl radical reaction rate constants were fast, k ∼ 7-9 × 109 M-1 s-1, and the similarity of the transient spectra obtained upon this radical-induced oxidation for all three antibiotics suggested that it occurred predominately at the peripheral aromatic ring common to these species. Although this oxidation would result in hydroxylation of the aromatic ring, the product might still retain some antimicrobial properties. Furthermore, only three 5-member ring β-lactam antibiotics were investigated in this past work.26 Therefore, in

10.1021/jp104509t  2010 American Chemical Society Published on Web 07/27/2010

8392

J. Phys. Chem. A, Vol. 114, No. 32, 2010

Dail and Mezyk The radiolysis of neutral pH water in the absence of dissolved oxygen produces radicals and molecular species according to27 radiolysis

H2O 98 (0.28)•OH + (0.06)•H + (0.27)e-(aq) + (0.05)H2 + (0.07)H2O2 + (0.27)H3O+

(1)

where the number in parentheses preceding each species is its absolute yield (G value) in micromoles per gray of deposited energy. The isolation of hydroxyl radicals occurs through the reactions of the hydrated electron and hydrogen atom with the added N2O:27 • eaq + N2O(+H2O) f OH + N2 + OH

k2 ) 9.1 × 109 M-1 s-1 (2) H• + N2O f •OH + N2

k3 ) 2.1 × 106 M-1 s-1

(3)

Figure 1. Antibiotics of this study. Top: 5-member β-lactams have single R1 substituent as listed. Bottom: 6-member β-lactams having two substituents listed in R1, R2 order.

this study, we report on our kinetic measurements involving the oxidation of a larger library of 11 representative β-lactam antibiotics containing both 5- and 6-member ring species in water. Through a comparison of these kinetic data, along with concomitant transient spectral measurements, we have established the site of hydroxyl radical reactivity in this family of antibiotics.

Absolute dosimetry29 was performed daily using N2Osaturated, 1.00 × 10-2 M KSCN solutions at λ ) 475 nm, (Gε ) 5.2 × 10-4 m2 J-1) with average doses of 3-5 Gy per 2-3 ns pulse. Throughout this paper, G is defined in micromoles per gray, and ε is in units of per molar per centimeter. The total radical concentrations typically used in our kinetics radiolysis experiments were ∼2-4 µM per pulse. For nonambient temperature measurements, room-temperature solutions were flowed through a short condenser tube located immediately below the irradiation cell and fed by a temperaturecontrolled water bath. The actual solution temperature was measured directly above the cell using a thermocouple in the flow path, and the temperature variation of this method at any particular bath setting was found to be better than 0.3 °C. The rate constant error limits reported here are the combination of experimental precision and compound purities.

Experimental Section The chemicals used in this study (see Figure 1) were obtained from either Sigma-Aldrich Chemical Co. or ICC (Indofine Chemical Company) at the highest purity available (>98%), and used as received. Antibiotic solutions were made in high-quality Millipore Milli-Q charcoal-filtered (TOC < 13 ppb), deionized (>18.2 MΩ) water, which was constantly illuminated by a Xe arc lamp (172 nm) to keep organic contaminant concentrations below 13 µg L-1. All solutions were buffered with 5.00 mM phosphate, with the pH adjusted to 7.40 ( 0.05 using perchloric acid or sodium hydroxide, and were extensively sparged with high-purity N2O to remove dissolved oxygen before irradiation. Electron Pulse Radiolysis. The linear accelerator (LINAC) electron pulse radiolysis system at the Radiation Laboratory, University of Notre Dame, was used for the •OH rate constant determinations of this study. This irradiation and transient absorption detection system has been described in detail previously.28 During rate constant measurements, the solution vessels were sparged with the minimum amount of N2O necessary to prevent air ingress. Solution flow rates in these experiments were adjusted so that each irradiation was performed on a fresh sample. Each kinetic trace was the average of 10-15 pulses at the respective λmax determined from each compounds’ transient absorption spectra.

Results and Discussion Hydroxyl Radical Reactions. The reaction of hydroxyl radicals with each of these antibiotics gave transient spectra with peaks in the region 300-380 nm (see, for example, Penicillin-G in Figure 2, inset, and Figure S1 in the Supporting Information). Absolute absorption coefficients were calculated following the method of LaVerne and Pimblott30 to determine intraspur scavenging yields. Under N2O-saturated conditions, the calculated initial •OH yield was 0.60 ( 0.01 µmol Gy-1 for the 250 µM antibiotic solutions used for these measurements. For Penicillin-G, the measured λmax was at 320 nm with εmax ) 3050 ( 150 M-1 cm-1, slightly lower than the previously reported value of 3300 M-1 cm-1.26 All the transient spectra maximum wavelengths and corresponding absorption coefficients are also given in Table 1. Antibiotics containing a phenyl ring in their substituents show fairly consistent λmax and εmax values, similar to Penicillin-G (Figure S1 of the Supporting Information). However, greater variation is seen for antibiotics with different aromatic substituents, such as Cefazolin, which has an εmax of 7000 M-1 cm-1 at 310 nm. From the rate of change of these pseudo-first-order growth kinetics of this transient absorption with antibiotic concentration, specific reaction rate constants for the oxidative hydroxyl radical reaction with each compound was determined at various

Oxidation of Β-Lactam Antibiotics in Water

J. Phys. Chem. A, Vol. 114, No. 32, 2010 8393

Figure 2. Growth kinetics at 320 nm for hydroxyl radical induced oxidation of Penicillin-G in 5.00 mM phosphate buffered solution at pH 7.40 and 23.0 °C. Solid lines are fitted first-order growths, corresponding to rate constants of (2.37 ( 0.04) × 106, (1.54 ( 0.02) × 106, (9.80 ( 0.18) × 105, and (5.13 ( 0.07) × 105 s-1 for 250.0 (0), 150.1 (O), 100.4 (∆), and 51.1 (3), respectively. Inset: Maximum transient absorption spectrum of oxidized 250.0 µM Penicillin-G at pH 7.40, and 23.0 °C.

temperatures (see Figure 3). All the rate constant data of this work are summarized in Table 1. Our hydroxyl radical rate constant value for reaction with Penicillin-G of (8.70 ( 0.32) × 109 M-1 s-1 at 23.0 °C is in reasonable agreement with the originally reported value of (7.97 ( 0.11) × 109 M-1 s-1 at room temperature.26 Even better agreement is seen between the two measurements for Penicillin-V (see Table 1). A comparison of all our measured rate constants in Table 1 for the 5-member β-lactams shows only a small variation in reactivity. However, it is important to note that these rate constants, although fast, are not diffusion-controlled. The upper limit, kdiff, rate constant value can be calculated from the Smoluchowski equation31

kdiff ) 4(1000)π(D•OH + DPen-G)(R•OH + RPen-G)N

(4) where N is Avogadro’s number, D is the diffusion coefficient of the individual species in water, and R is its effective reaction

Figure 3. (a) Temperature-dependent second-order kinetics obtained from the pseudo-first-order growth fits of Figure 2 for Penicillin-G oxidation at 12.7 (O), 23.0 (0), 30.6 (∆), and 36.4 (3) °C. Solid lines are weighted linear fits, with corresponding values of (6.52 ( 0.23) × 109, (8.71 ( 0.32) × 109, (1.04 ( 0.02) × 1010, and (1.10 ( 0.04) × 1010 M-1 s-1, respectively. (b) Corresponding Arrhenius plot for data of panel a. Solid line is weighted linear fit, giving a slope of -2064.8 ( 162.2 and intercept of 29.84 ( 0.54 (R2 ) 0.994) and corresponding to an activation energy of 17.2 ( 1.3 kJ mol-1.

radius. Literature values for diffusion coefficients have been obtained for some antibiotics (e.g., Penicillin-G32 is 9.7 × 10-10 m2 s-1) and for the hydroxyl radical33-35 (2.1 × 10-9 m2 s-1). From the diffusion coefficients, we can also obtain the corresponding R values from the Stokes-Einstein equation:31

Dx )

kT 6πηR

(5)

where η ) 8.88 × 10-4 Pa s is the viscosity of water at 25 °C.36 Substituting in the diffusion coefficients from above gives RPen-G ) 0.25 nm. The radius of the hydroxyl radical in water has been reported as R•OH ) 0.22 nm.37 Substituting these values into eq 4 gives kdiff ) 1.1 × 1010 M-1 s-1 at 298 K. Because this value is faster than our measured values, we believe that

TABLE 1: Summary of Hydroxyl Radical Reaction Rate Constants for 5- And 6-Member β-Lactam Antibioticsa 5-member β-lactams

6-member β-lactams

species λmax, εmax (nm, M-1cm-1)

10-9 k•OH M-1 s-1

temp °C

species λmax, εmax (nm, M-1 cm-1)

10-9 k•OH M-1 s-1

temp °C

Ampicillin (320, 2475) Carbenicillin (330, 2850) Cloxacillin (320, 3050) Penicillin-G (320, 3050)

8.21 ( 0.29 10.7 ( 0.20 7.31 ( 0.11 9.70 ( 0.21 6.27 ( 0.15 9.11 ( 0.58 6.52 ( 0.23 8.70 ( 0.32 10.42 ( 0.15 11.00 ( 0.40 7.97 ( 0.11b 8.54 ( 0.27 10.74 ( 0.27 8.76 ( 0.28b 7.84 ( 0.49 10.1 ( 0.30 8.18 ( 0.99 9.68 ( 0.21

22.7 34.0 18.0 37.4 23.0 37.5 12.7 23.0 30.6 36.4 RT 21.3 37.4 RT 22.5 36.9 21.8 36.5

Cefaclor (300, 3900) Cefazolin (310, 7000) Cephalothin (380, 3750) Cefatoxime (330, 2900)

6.00 ( 0.13 10.79 ( 0.20 6.48 ( 0.48 10.50 ( 0.16 5.51 ( 0.29 12.78 ( 0.40 8.22 ( 0.14 10.55 ( 0.33

22.7 37.4 22.4 38.1 19.2 37.4 21.2 37.4

Penicillin-V (320, 3200) Piperacillin (310, 2550) Tircarcillin (330, 2350) a

Values of this study in bold. b Value of Song et al.,26 at room temperature (RT).

8394

J. Phys. Chem. A, Vol. 114, No. 32, 2010

these reaction rate constants are still predominately activationcontrolled and that the similarity of the measured values implies a common reaction mechanism. By comparison of our measured rate constants for the 5-member ring antibiotics (average rate constant is k ) (7.9 ( 0.8) × 109 M-1 s-1) to literature data for model aromatic chemicals that correspond to the R1 substituents (see Figure 1) such as toluene (k ) 3.0 × 109 M-1 s-1 38) and thiophene (k ) 3.3 × 109 M-1 s-1 39), as well as the previously established value of (2.40 ( 0.05) × 109 M-1 s-1 for (+)-6-aminopenicillanic acid core moiety,26 we conclude that the initial hydroxyl radical reaction for all seven of the antibiotics of this study would be partitioned between the peripheral aromatic ring on the lactam substituent and the central double-ring system, with only ∼30% reacting with the latter. This is consistent with previous measurements on some of these antibiotics.26 The slightly slower rate constant for Cloxacillin could be due to the presence of the electron-withdrawing chloride substituent on its peripheral ring. However, these data imply that some of the hydroxyl radical oxidation does occur at, or close to, the β-lactam ring, which might better deactivate its overall microbial activity. These data contrast previous hydroxyl radical measurements performed for sulfonamide antibiotics,40 in which the radicalinduced oxidation occurred quantitatively at the sulfanilic acid moiety with a consistent rate constant value of k ) 8.5 × 109 M-1 s-1, effectively independent of any sulfanilic acid substitutuent. An investigation into the temperature-dependence of the hydroxyl radical oxidation for Penicillin-G was also conducted in this study. Specific rate constants over the range 12.7-36.4 °C were determined; the data are summarized in Table 1 and shown in Figure 3. From the increase in the second-order rate constant with temperature (Figure 3a), the Arrhenius plot in Figure 3b was constructed, giving a straight line plot corresponding to an activation energy of Ea ) 17.2 ( 1.3 kJ mol-1. Rate constants for all the other six 5-member ring antibiotics of this study were also determined at near physiological temperatures, and again were shown to be fairly consistent (see Table 1), indicating that no mechanistic change occurred for the hydroxyl radical oxidation over this short range. For the four 6-member ring antibiotics, the overall rate constant for hydroxyl radical oxidation measured was slightly slower, k ) (6.6 ( 1.2) × 109 M-1 s-1 at near room temperature. This again suggests that some partitioning of the hydroxyl radical reaction between the central moiety and a peripheral aromatic ring would occur for these antibiotics. Again, the rate constants measured at physiological temperature were faster, but generally consistent, implying that the same oxidation pathways still occurred. Because the majority of the hydroxyl radical oxidation occurs at the peripheral aromatic ring systems in these antibiotics, ultimately, it would be expected that phenol compounds would be formed.26 This implies that radical-generating systems could be useful for the quantitative removal of antibiotics in realworld waters, provided sufficient oxidation occurred. It would also be necessary to evaluate the toxicity of the degradation products before any practical implementation would occur. Conclusions The reaction rate constants for hydroxyl radical reaction with 11 different β-lactam antibiotics have been directly determined at room and physiological temperatures. For the 5-member ring species, the oxidation was effectively independent of substituent, and the measured rate constant, k ) (7.9 ( 0.8) × 109 M-1 s-1

Dail and Mezyk showed that most (∼70%) of the reactivity was at the peripheral aromatic rings in these compounds. The temperature dependence was detailed for Penicillin-G, giving an activation energy of Ea ) 17.2 ( 1.3 kJ mol-1. Similar temperature behavior was observed for all of these antibiotics. For the 6-member ring species, the hydroxyl radical reactivity at room temperature was slightly slower, k ) (6.6 ( 1.2) × 109 M-1 s-1, but again showed consistent temperature behavior for hydroxyl radical oxidation. Acknowledgment. This work was performed at the Radiation Laboratory, University of Notre Dame, which is supported by the Office of Basic Energy Sciences, U.S. Department of Energy. Financial support is also acknowledged from the Water Reuse Foundation Grant WRF 04-017, CSULB Women and Philanthropy Undergraduate Research Grant, CSULB Provost Undergraduate Summer Stipend Research Grant, and Target Specialty Products, Inc. Scholarship. Supporting Information Available: Transient spectra comparisons for all 5- and 6-member ring β-lactam antibiotics are available in the Supporting Information. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Kolpin, D. W.; Furlong, E. T.; Meyer, M. T.; Thurman, E. M.; Zaugg, S. D.; Barber, L. B.; Buxton, H. T. EnViron. Sci. Technol. 2002, 36, 1202. (2) Hirsch, R.; Ternes, T. A.; Haberer, K.; Kratz, K. Sci. Total EnViron. 1999, 225, 109. (3) Bailo´n-Pe´rez, M. I.; Garcı´a-Campan˜a, A. M.; Cruces-Blanco, C.; del Olmo Iruela, M. J. Chromatogr. A 2008, 1185, 273. (4) Al-Ahmad, A.; Daschner, F. D.; Ku¨mmerer, K. Arch. EnViron. Contam. Toxicol. 1999, 37, 158. (5) Ku¨mmerer, K. Chemosphere 2001, 45, 957. (6) Ku¨mmerer, K. J. Antimicrob. Chemother. 2004, 54, 311. (7) Petrovic, M.; Hernando, M. D.; Diaz-Cruz, M. S.; Barcelo, D. J. Chromatogr. A 2005, 1067, 1. (8) Richardson, S. D.; Ternes, T. A. Anal. Chem. 2005, 77, 3807. (9) Zwiener, C.; Gremm, T. J.; Frimmel, F. H. Pharmaceuticals in the EnVironment: Sources Fate, Effects, and Risks; Springer-Verlag: Weinheim, Germany, 2004. (10) Hirscha, R.; Ternesa, T.; Haberera, K.; Kratzb, K. L. Sci. Total EnViron. 1999, 225, 109. (11) Prabhakaran, D.; Sukul, P.; Lamshoeft, M.; Maheswari, M. A.; Zuehlke, S.; Spiteller, M. Chemosphere 2009, 77, 739. (12) Avisar, D.; Lester, Y.; Mamane, H. J. Hazard. Mater. 2010, 175, 1068. (13) Koyuncu, I.; Arikan, O. A.; Wiesner, M. R.; Rice, C. J. Membr. Sci. 2008, 309, 94. (14) Hylton, K.; Sangwan, M.; Mitra, S. Anal. Chim. Acta 2009, 653, 116. (15) Choi, K.-J.; Kim, S.-G.; Kim, S.-H. EnViron. Sci. Technol. 2008, 29, 333. (16) Choi, K.-J.; Kim, S.-G.; Kim, S.-H. J. Hazard. Mater. 2008, 151, 38. (17) Navalon, S.; Alvaro, M.; Garcia, H. Water Res. 2008, 42, 1935. (18) Huber, M. M.; Canonica, S.; Park, G.-Y.; von Gunten, U. EnViron. Sci. Technol. 2003, 37, 1016. (19) Ternes, T. A.; Stu¨ber, J.; Herrmann, N.; McDowell, D.; Ried, A.; Kampmann, M.; Teiser, B. Water Res. 2003, 37, 1976. (20) Andreozzi, R.; Canterino, M.; Marotta, R.; Paxues, N. J. Hazard. Mat. 2005, 122, 243. (21) Ikehata, K.; Naghashkar, N. J.; El-Din, M. G. Ozone Sci. Eng. 2006, 28, 353. (22) Yargeau, V.; Leclair, C. Water Sci. Technol. 2007, 55, 321. (23) Beltran, F. J.; Aguinaco, A.; Garcia-Araya, J. F.; Oropesa, A. Water Res. 2008, 42, 3799. (24) Ikehata, K.; El-Din, M. G.; Snyder, S. A. Ozone Sci. Eng. 2008, 30, 21. (25) Crittenden, J. C.; Hu, S.; Hand, D. W.; Green, S. A. Water Res. 1999, 33, 2315. (26) Song, W.; Chen, W.; Cooper, W. J.; Greaves, J.; Miller, G. E. J. Phys. Chem. A 2008, 112, 7411. (27) Buxton, G. V.; Greenstock, C. L.; Helman, W. P.; Ross, A. B. J. Phys. Chem. Ref. Data 1988, 17, 513.

Oxidation of Β-Lactam Antibiotics in Water (28) Whitman, K.; Lyons, S.; Miller, R.; Nett, D.; Treas, P.; Zante, A.; Fessenden, R. W.; Thomas, M. D.; Wang, Y. Proceedings of the ’95 Particle Accelerator Conference and International Conference on High Energy Accelerators, Dallas, TX, 1996. (29) Buxton, G. V.; Stuart, C. R. J. Chem. Soc., Faraday Trans. 1995, 91, 279–282. (30) LaVerne, J. A.; Pimblott, S. M. Radiat. Res. 1993, 135, 16. (31) North, A. M. The Collision Theory of Chemical Reactions in Liquids; Meuthen: London, 1964. (32) Cussler, E. L.; Breuer, M. M. AIChE J. 1972, 18, 2517. (33) Dorfman, L. M.; Adams, G. E. Reactivity of the Hydroxyl Radical in Aqueous Solution; Nat. Stand. Ref. Data Ser., National Bureau of Standards, 1973; Vol. 46.

J. Phys. Chem. A, Vol. 114, No. 32, 2010 8395 (34) Hotop, H.; Lineberger, W. C. J. Phys. Chem. Ref. Data 1975, 4, 539. (35) Busi, F.; D’Angelantonio, M.; Betoli, G.; Concialini, V.; Tubertini, O.; Barrer, G. C. Inorg. Chem. 1984, 84, 105. (36) Lide, D. R., Ed.; Handbook of Chemistry and Physics, 71st ed.; CRC Press; Boca Raton, FL, 1991. (37) Braams, R.; Ebert, M. AdV. Chem. Ser. 1968, 81, 464. (38) Dorfman, L. M.; Taub, I. A.; Harter, D. A. J. Chem. Phys. 1964, 41, 2954. (39) Lilie, J. Z. Naturforsch. Teil B 1971, 26, 197. (40) Mezyk, S. P.; Neubauer, T. J.; Cooper, W. J.; Peller, J. R. J. Phys. Chem. A 2007, 111, 9019.

JP104509T