Surface Complexation of the Zwitterionic Fluoroquinolone Antibiotic

Article pubs.acs.org/est

Surface Complexation of the Zwitterionic Fluoroquinolone Antibiotic Ofloxacin to Nano-Anatase TiO2 Photocatalyst Surfaces Tias Paul,*,† Michael L. Machesky,‡ and Timothy J. Strathmann† †

Department of Civil and Environmental Engineering, University of Illinois at Urbana−Champaign, Urbana, Illinois 61801, United States ‡ Illinois State Water Survey, 2204 Griffith Drive, Champaign, Illinois 61820, United States S Supporting Information *

ABSTRACT: The surface complexation behavior of ofloxacin (OFX), a zwitterionic fluoroquinolone antibiotic, to nano-anatase titanium dioxide (TiO2) was characterized. OFX adsorption in aqueous TiO2 suspensions was measured as a function of pH, OFX concentration, and electrolyte type and concentration, and structural information was derived from in situ spectroscopic observations. An ultraviolet−visible spectral red shift upon OFX adsorption indicated formation of inner-sphere coordination complexes. Fourier transform infrared spectra of TiO2-adsorbed OFX were invariable over a wide concentration and pH range and were similar to measured spectra of dissolved species wherein the carboxylate group is deprotonated. A charge distribution surface complexation model constrained by spectroscopic observations was developed to describe macroscopic adsorption trends. A tridentate mode of adsorption involving bridging bidentate inner-sphere coordination of the deprotonated carboxylate group and hydrogen bonding through the adjacent carbonyl group on the quinoline ring resulted in successful predictions of observed adsorption trends. In NaClO4 electrolyte, spectroscopic data and model fitting suggested that OFX ion pairing with ClO4− enhanced adsorption under acidic conditions. Moreover, comparison of OFX adsorption data with the pH trend in the kinetics of OFX degradation by visible light (λ > 400 nm) photocatalysis suggested that adsorbed OFX−ClO4− ion pairs inhibit photodegradation.

■

(UVA; 320 nm < λ < 400 nm) and visible light (λ > 400 nm) photocatalysis using nanophase titanium dioxide (TiO2) materials.10,11 Previous work indicates that FQ adsorption to the TiO2 surface is a critical step in the proposed charge transfer mechanism.12 However, the molecular structures of FQ species adsorbed to TiO2 surfaces and environmental factors controlling FQ adsorption are not well understood. A number of recent studies have examined adsorption of FQs to mineral surfaces, including aluminum (hydr)oxides,13,14 iron (hydr)oxides,13,15,16 manganese oxides,17 silica,14 and aluminosilicate clays.18−20 Spectroscopic measurements have probed the structures of adsorbed FQ species, and various structures have been proposed, including those where carboxylate,13−15 carbonyl,13,14 and amino14 groups interact with mineral surface sites. However, spectroscopically derived structural information has not been incorporated into adsorption models, and currently no models can accurately predict the extent of FQ adsorption to TiO2 and related metal (hydr)oxide surfaces at widely variable solution conditions (e.g., pH, electrolyte composition). To date, attempts to model FQ adsorption

INTRODUCTION Fluoroquinolones (FQs) like ofloxacin (OFX) are a class of broad spectrum antibacterial agents that inhibit bacterial growth by blocking DNA replication enzymes.1 FQs have been extensively applied in human medicine and commercial animal rearing operations.2 Recently, FQs and other antibacterial agents have emerged as aqueous micropollutants of concern.3−5 Incompletely metabolized FQs have been detected in both wastewater sludge-treated agricultural soils and in surface waters influenced by effluent from wastewater treatment plants.6,7 FQ concentrations as low as 1 μg/L have been shown to adversely affect microbial ecology,8 and mixtures of FQs with other pharmaceutically active compounds at environmentally relevant concentrations have been shown to elicit negative responses in human embryonic cell development.9 In addition, there are serious concerns about environmental releases of antibacterials promoting emergence of antibiotic resistance in pathogenic microorganisms. Concerns about the environmental and health risks associated with contamination of aquatic systems by FQs and other antibacterials have prompted research on effective and efficient treatment technologies. Photocatalytic processes have been shown to be effective for many classes of organic pollutants, and recent work indicates that FQs can be effectively transformed to nonpotent products by both ultraviolet-A © 2012 American Chemical Society

Received: Revised: Accepted: Published: 11896

May 25, 2012 August 29, 2012 September 5, 2012 October 18, 2012 dx.doi.org/10.1021/es302097k | Environ. Sci. Technol. 2012, 46, 11896−11904

Environmental Science & Technology

Article

Figure 1. Acid−base speciation of OFX. Although the order of functional group deprotonation in zwitterionic FQs can vary, the dominant species result from sequential deprotonation of the carboxylic acid at pKa1 (5.927) to form the zwitterionic structure (middle left structure) followed by the amine at pKa2 (8.327).28−30

processes have been limited to solution-specific isotherms14,21 and calculation of aqueous/surface partition coefficients (Kd values)19 for individual FQ aqueous species. Modeling adsorption of FQs like OFX to variably charged surfaces is complicated by the zwitterionic structures that predominate at neutral pH conditions (Figure 1). Simultaneous presence of anionic and cationic functional groups leads to both attractive and repulsive electrostatic interactions with charged surfaces, the relative strengths of which vary depending on adsorbate orientation and charge state of the surface. Such complex interactions cannot be realistically accounted for by classical point-charge surface complexation modeling approaches that treat zwitterions as neutral molecules.22 This contribution reports the results of work designed to characterize OFX−TiO2 bonding interactions and to use structural information to formulate a surface complexation model (SCM) capable of describing OFX adsorption to TiO2 over wide-ranging solution conditions (FQ concentration, pH, and background electrolyte identity and concentration). The charge distribution (CD) approach of Hiemstra and Van Riemsdijk23 was used to incorporate structural information derived from in situ spectroscopic measurements and handle functional group-specific electrostatic interactions in the SCM. To our knowledge, the CD modeling approach has not been previously used to describe the adsorption of zwitterionic organic compounds. Finally, pH-dependent trends in the kinetics of visible light-TiO2 photocatalyzed oxidation of OFX

are interpreted in light of SCM predictions of OFX surface speciation.

■

EXPERIMENTAL SECTION Materials. Hombikat UV100, a nanophase anatase TiO2 of average particles size pKa1. These peaks decrease in intensity at lower pH conditions and are replaced by a peak assigned to stretching of the CO bond in the protonated carboxylic acid group (νCO,carboxyl = 1710 cm−1).14 Together, these spectral trends indicate that the predominant molecular changes occurring near pKa1 involve the carboxylate functional group and that the zwitterionic species predominates over the neutral species when pKa1 < pH < pKa2. Other peaks have also been assigned previously for spectra of aqueous OFX and structurally related FQs.14,15,20 Of particular note, a prominent peak at 1622 cm−1, comprised of two closely spaced peaks of a Fermi doublet,31 has been assigned to stretching of the quinoline ring carbonyl group (νCO,carbonyl).14 A small upshift in this peak occurs at pH < pKa1, which has previously been attributed to a change in Fermi resonance as a result of intramolecular hydrogen bonding (H-bonding) between the quinoline ring carbonyl group oxygen and the carboxylic acid group’s proton.15 Bonding and Structures of Adsorbed OFX. UV−vis spectra of OFX collected in TiO2 suspension (Figure 2A) show a significant increase in absorption at λ > 300 nm in comparison to similar spectra of dissolved OFX, which can 11898

dx.doi.org/10.1021/es302097k | Environ. Sci. Technol. 2012, 46, 11896−11904

Environmental Science & Technology

Article

at pH conditions where the carboxylate group is protonated (pH 3.1) and deprotonated (pH 7.1). Unlike the pHdependent spectra of aqueous OFX, the spectral features of adsorbed OFX are relatively invariable with respect to changing pH, largely corresponding with the spectra of aqueous OFX collected pH > pKa1, where the carboxylate group is deprotonated. Asymmetric and symmetric stretching modes of the deprotonated carboxylate group, νCOO,as and νCOO,s, match closely with those measured for aqueous OFX species. A distinct difference, however, is that an absorption band at 1272 cm−1 attributed to the scissoring of the deprotonated carboxylate group15 is significantly broadened. Additionally, the position of a band assigned to stretching of the quinoline ring carbonyl group (νCO,carbonyl) is more consistent with that observed for aqueous OFX species when the neighboring carboxylate group is protonated (i.e., pH < pKa1) and believed to H-bond to the carbonyl oxygen. While spectral peak positions observed for OFX adsorbed to TiO2 are relatively invariable with pH, the spectral intensities at different pH values do vary due to changing extent of OFX adsorption to the TiO2 film. The observed spectral intensities are similar between pH 3−7, but drop off at pH ≥ 8.0, consistent with measured trends in the extent of OFX adsorption. The spectra shown in Figure 2B were collected using NaClO4 as the background electrolyte. Similar spectra were observed when NaCl was used as the background electrolyte, with two notable exceptions. First, overall spectral intensity also decreases somewhat at lower pH conditions in NaCl, which agrees with trends in the measured extent of OFX adsorption discussed below. Second, a broad ClO4− asymmetric stretching band observed at 1100 cm−132 at lower pH conditions is absent in NaCl spectra (Figure S2 of the SI). The most significant observations in the ATR-FTIR spectra of OFX adsorbed to TiO2 are the close similarity to spectra of dissolved OFX species where the carboxylate group is deprotonated, even at pH ≪ pKa1. Carboxylate deprotonation in adsorbed FQ species has been observed with other metal (hydr)oxides,13,15 and this indicates that the dominant adsorbed species involves interaction of a deprotonated carboxylate group with the TiO2 surface. Additionally, the absence of spectral shifts in the νCOO,as and νCOO,s peaks indicates that the two carboxylate oxygen atoms retain equivalent coordination environments like in bulk solution. The lack of a significant shift in the νCOO,as and νCOO,s bands suggests inner-sphere bridging bidentate coordination of the two carboxylate oxygen atoms with neighboring Ti(IV) surface sites.33 Previous studies of carboxylic acids on TiO2 have reported that spectra of bridging bidentate complexes are very similar to the spectra of the aqueous carboxylate species.34−36 A similar mode of carboxylate complexation has also been proposed for ciprofloxacin adsorption to iron oxide surfaces.15 Although FTIR data alone do not provide sufficient evidence to ascertain inner-sphere coordination in the absence of carboxylate spectral shifts, the UV−vis spectra showing formation of a new LMCT absorbance band strongly indicates inner-sphere complexation. Since UV−vis spectra were collected at lower OFX:TiO2 ratios than ATR-FTIR spectra shown in Figure 2B, it is important to consider the possible effects of this difference on the predominant adsorbed species. Separate ATR-FTIR experiments were conducted using different OFX:TiO2 ratios. Although ATR-FTIR is not sensitive enough to observe adsorbed OFX at the ratios used for UV−vis measurements, these data (Figure S4 of the SI) showed no

Figure 2. (A) UV−vis spectra of OFX in aqueous solution and adsorbed to nanophase TiO2 (7 μM OFX, 0.5 g/L TiO2, pH 4.6, 0.01 M NaCl). (B) ATR-FTIR spectra of OFX adsorbed to TiO2 at different pH conditions in comparison to spectra of aqueous OFX at pH conditions where the carboxylate group is protonated (pH 3.1) versus deprotonated (pH 7.1). Conditions for ATR-FTIR: adsorbed OFX in equilibrium with an overlying solution containing 100 μM dissolved OFX, IS = 0.01 M (NaClO4 + HClO4). Aqueous OFX spectra reproduced from full data set shown in the SI.

be attributed to the formation of a ligand-to-metal charge transfer (LMCT) band upon OFX coordination with one or more surface Ti(IV) ions. A similar spectral red shift was also observed in ex situ diffuse reflectance UV−vis measurements previously reported for ciprofloxacin adsorption to TiO2.11 The observed red shift is believed to play a crucial role in the proposed surface complex-mediated charge transfer mechanism for visible light-TiO2 photocatalytic oxidation of FQs since neither FQs nor TiO2 absorb light in the visible region.11 Spectra of dissolved OFX collected at other pH conditions do not exhibit similar spectral shifts, confirming that the spectral change does not result from changes in acid−base speciation of OFX (e.g., deprotonation of the carboxylate group upon adsorption). Moreover, the low concentration of OFX added to TiO2 suspensions in these experiments makes it unlikely that observed red shift is an artifact resulting from differences in suspension dispersion properties when OFX is introduced. Figure 2B shows the ATR-FTIR spectra of OFX adsorbed to TiO2 surfaces at different pH conditions. For reference comparison, spectra are also shown for aqueous OFX collected 11899

dx.doi.org/10.1021/es302097k | Environ. Sci. Technol. 2012, 46, 11896−11904

Environmental Science & Technology

Article

Figure 3. Adsorption pH-edge measurements and surface complexation model fits for OFX adsorption to TiO2 when NaCl (A,B) and NaClO4 (D− F) are used as background electrolytes. (A, D) Effects of pH and OFX concentration on adsorption to 0.5 g/L TiO2 at a fixed ionic strength (IS = 0.01 M). (B,E) Effects of pH and ionic strength on the adsorption of 100 μM OFX to 0.5 g/L TiO2. (C) Structures with interfacial charge distributions of OFX−TiO2 surface complexes without (top) and with (bottom) ion pairing to perchlorate anion at the Stern plane. (F) Detailed model fits showing the distribution of adsorbed OFX species in NaClO4 (for the 100 μM OFX, 0.5 g/L TiO2, IS = 0.01 M).

interaction, the protonated tertiary piperazine amine is a plausible site for ClO4− ion pairing as similar ionic interaction between these two groups is observed in the OFX−ClO4 salt.40 Additional support for formation of strong perchlorate ion pairs with cationic amines can be found in the common use of perchlorate as an ion pairing reagent for improving chromatographic analysis of amines.41 Adsorption pH Edges. Figure 3 summarizes the results of adsorption pH-edge experiments for OFX using NaCl (parts A,B) and NaClO4 (parts D−F) as background electrolytes. In general, OFX adsorption to TiO2 is characterized by negatively sloped adsorption edges with minimal adsorption at higher pH conditions, consistent with previous reports on FQ adsorption to specific metal (hydr)oxides.14,16 At fixed ionic strength (0.01 M), the absolute loading of adsorbed OFX (μmole OFX adsorbed per m2 of TiO2 surface) increases with total OFX concentration (20−500 μM) (Figure 3A,D), but the percentage of added OFX that adsorbs decreases. Somewhat more unexpectedly, at a fixed total OFX concentration (100 μM), increasing solution ionic strength from 0.001 to 0.1 M using either NaCl (Figure 3B) or NaClO4 (Figure 3E) leads to significant increases in OFX adsorption under neutral and acidic pH conditions. This contrasts with reports of adsorption for most classes of organic ligands, where increasing ionic strength tends to either decrease or have little effect on adsorption.42 Decreased adsorption at higher ionic strength has usually been attributed to increased shielding of attractive electrostatic interactions between negatively charged ligands and positively charged metal (hydr)oxide surfaces that predominate at acidic conditions. Consequently, the increased OFX adsorption with increasing ionic strength suggests

major shifts in the spectra when conditions were varied from 20 to 500 μM, suggesting little effect of OFX:TiO2 ratio on the structures of the predominant adsorbed OFX species. ATR-FTIR spectral trends also support involvement of the quinoline ring carbonyl group in OFX−TiO2 surface complex formation via H-bonding interactions. The observed upshift in νCO,carbonyl upon OFX adsorption is similar to the band shift observed for aqueous OFX when the neighboring carboxylate group protonates. This suggests that the coordination environment of the quinoline ring’s carbonyl group is affected in a similar manner. The shift in νCO,carbonyl upon protonation of the neighboring carboxylate group has been attributed to the effects of intramolecular H-bonding between the quinoline ring carbonyl group (H-bond acceptor) and the protonated carboxylate group (H-bond donor). Thus, the ATR-FTIR spectra are consistent with the carbonyl oxygen H-bonding with surface hydroxyl groups. The added contribution of the carbonyl group to OFX surface complex formation is also supported by previous studies showing that monocarboxylate ligands (e.g., benzoic acid) exhibit weak adsorption to TiO2 surfaces in comparison to similar compounds with additional adjacent Lewis base donor groups (e.g., amino, hydroxyl).37−39 The presence of a strong perchlorate absorption band at lower pH conditions indicates that perchlorate coadsorbs with OFX under these conditions, suggesting the formation of an OFX−ClO4− ion-paired adsorbed species. Strong ionic interaction between OFX and ClO4− is supported by the reduced solubility of OFX in NaClO4 (in comparison to the same concentration of NaCl) and apparent formation of an OFX−ClO4 salt (Figure S3 of the SI) under acidic conditions. While FTIR data cannot confirm the nature of the ion pairing 11900

dx.doi.org/10.1021/es302097k | Environ. Sci. Technol. 2012, 46, 11896−11904

Environmental Science & Technology

Article

Table 1. Model for OFX−TiO2 Surface Complexation reactions/parameters Aqueous

log K

Source

H2−OFX+ ⇌ H-OFX± + H+

−5.9a

ref 27

H-OFX± ⇌ OFX− + H+

−8.3a

ref 27

Surface chemistry reactionsb/parameters

logK

Value

Source

specific surface area (m2/g)

290

ref 12

adsorption site density (sites/nm2)

5.2c

crystal structure

0.65 ± 0.03d

titration fits

Stern layer capacitance (F/m2) ≡TiOH−0.5 + H+ ⇌ ≡TiOH2+0.5

5.6

intersection of titration curves

−1.31 ± 0.8d

titration fits

−1

ref 25

≡TiOH−0.5 + H+ + ClO4− ⇌ ≡TiOH2−ClO4−0.5

−1.71 ± 0.6d

titration fits

OFX surface complexation reactions/parameter

log K

CD

Δz0

Δzs

OFX− + 4H+ + 3 ≡TiOH−0.5 ⇌ [ ≡Ti3OH2(H-OFX)]+1.5 + 2 H2O

16.9 ± 0.1d

−1.2

1.8

1.2

OFX− + 4H+ + 3 ≡TiOH−0.5 + ClO4− ⇌ [≡Ti3OH2(H-OFX)+1.5...ClO4−] + 2H2O

17.4 ± 1.7d

−1.2

1.8

0.2

≡TiOH−0.5 + Na+ ⇌ ≡TiOH-Na+0.5 ≡TiOH−0.5 + H+ + Cl− ⇌ ≡TiOH2−Cl−0.5

a

The intrinsic logKa values, corrected to zero ionic strength, were determined from conditional pKa values reported in the literature.27 bSurface reactions written as combinations of component species. Protons included in the reactions account for change in charge at relevant surface planes resulting from surface reactions. cTotal site density of the (101) surface of anatase is twice this value but the bridged oxygen groups (≡Ti2O) are assumed to not participate in ligand exchange with OFX. However, the total site density (10.4 sites/nm2) was used in fitting the electrolyte binding constants. dUncertainty represents one standard deviation.

used here included a basic Stern layer description of the electric double layer as depicted in Figure 3C. The surface charging behavior of TiO2 was first characterized by conducting acid−base titrations of TiO2 suspensions prepared in 0.001, 0.01, and 0.1 M NaClO4 solutions (see SI). The number of surface sites that can participate in proton exchange reactions was derived from the anatase crystal structure. X-ray diffraction patterns previously reported for the nanoanatase material used (Hombikat UV100) indicate the predominant exposed crystal surface to be the (101) plane.43 In vacuum, the anatase (101) surface plane consists of 5-fold and 6-fold Ti atoms bonded with 2- and 3-fold coordinated oxygen atoms.44 In water, the under-coordinated 5-fold Ti hydrate, giving rise to singly coordinated exchangeable water molecules and hydroxyl groups. From crystallographic considerations, the total density of these singly coordinated sites (≡TiOH) sites is 5.2 sites/nm2. The anatase (101) surface also contains protolyzable bridged oxygen sites (≡Ti2O) at a density of 5.2 sites/nm2. According to Pauling’s bond valence concept,45 these groups possess different fractional charges (−1/3 and −2/3 for the singly coordinated and bridged groups, respectively) and protonation constants. However, following Bourikas et al.,25 a simplifying assumption was made that both sites have equivalent acid−base chemistry thereby reducing the surface charging behavior to the 1-pKa model,46

shielding of repulsive electrostatic interactions between OFX and TiO2. For OFX and other zwitterionic FQs, this can be rationalized by considering electrostatic interactions between the charged TiO2 surface and the amine group on the piperazine ring. Under acidic pH conditions, TiO2 surfaces exhibit a net positive charge and the amine group is protonated at pH < pKa2, leading to an unfavorable electrostatic interaction which will be shielded at higher ionic strength by electrolyte anions (i.e, ClO4− or Cl−). Adsorption edge shape and surface coverage are also influenced to some extent by the background electrolyte. When NaCl is used (Figure 3A,B), OFX surface coverage generally peaks at ∼pH 6 and then steadily decreases with decreasing pH. In contrast, when NaClO4 is used (Figure 3D,E), the extent of OFX adsorption tends to plateau or even increase at lower pH conditions. Surface Complexation Model. OFX adsorption data were fit using a SCM that incorporates the CD approach first developed by Hiemstra and Van Riemsdijk.23 This approach is well suited for modeling the adsorption of multifunctional groups like OFX and other zwitterionic FQs that contain both surface-bonding and nonbonding ionizable functional groups. The CD formulation allows for charge in the adsorbing species structure to be distributed across the solid/aqueous interface, enabling a more molecularly realistic depiction of electrostatic interactions in complex molecules like FQs. The CD model

≡TiOH−0.5 + H+ ⇌ ≡ TiOH 2+0.5 log KH = 5.6 11901

(1)

dx.doi.org/10.1021/es302097k | Environ. Sci. Technol. 2012, 46, 11896−11904

Environmental Science & Technology

Article

Model fitting results are summarized in Table 1. Figure 3A,B shows the results of the SCM fits of OFX adsorption experiments conducted in NaCl electrolyte solutions. Measured data were generally well described considering formation of a single surface complex depicted in Figure 3C (upper structure) with the overall stoichiometry described in eq 2. Simultaneous fitting of all five measured pH adsorption edges yielded an optimized binding constant (log K = 16.9 ± 0.1). Considering that only a single adsorbed complex (one adjustable fitting parameter) was used to fit the entire data set, the measured adsorption trends over a wide range of pH (3 − 11), OFX concentration (20 − 500 μM), and ionic strength (0.001 to 0.1 M) are reproduced reasonably well. Most notably, the spectroscopically constrained CD value led to correct predictions of the unexpected ionic strength effect, that is, adsorption increasing with ionic strength at low pH. Small fit improvements could be achieved by fitting each pH edge experiment separately, but we chose to use a single model parametrization to best describe all measurements made. The fact that the single surface complex with a structurally constrained CD value is capable of describing the measured adsorption trends supports the pH-invariant nature of the OFX−TiO2 surface species observed in ATR-FTIR spectra. It is also noteworthy that similar fits and a similar binding constant value (log K = 16.85) and CD value (−1.25 v.u.) were obtained if the CD value was also allowed to float as an adjustable parameter when fitting the measured adsorption data. Attempts at modeling the OFX−TiO2 surface species as an outer-sphere complex involving H-bonding of a deprotonated carboxylate group with TiO2 surface sites were unsuccessful in predicting observed adsorption trends. As already noted, slightly different adsorption trends were observed in experiments conducted in NaClO4 electrolyte. A single tridentate surface complex was not sufficient for describing all of the experimental data. On the basis of FTIR evidence, successful fits in the NaClO4 system (Figure 3D,E) were obtained by adding a second tridentate complex with similar surface binding and charge distribution, but with the additional inclusion of ClO4− ion pair formation with the protonated amine group on the piperazine ring functional group (Figure 3C lower structure):

where the ±0.5 surface site charges represent the average of the two site types. According to this model, the log KH value of the surface group is equivalent to pH of zero point charge (pHzpc), which was determined from the common intersection point of the surface acid−base titration curves (Figure S5 of the SI). Surface titration data fits also yielded values of the Stern Layer capacitance and binding constants for background electrolytes (Na+ and ClO4−), which were assumed to reside at the Stern plane as point charges. The surface charging model parameters are summarized in Table 1, and experimental and modeling details are described in SI. These parameters were then fixed in the overall SCM when fitting the OFX adsorption data. OFX adsorption measurements shown in Figure 3 were fit by considering formation of one or more OFX−TiO2 surface complexes that were structurally constrained by spectroscopic observations. Figure 3C shows the proposed dominant modes of OFX−TiO2 surface complexation derived from the collective in situ spectral observations along with formulated charge distributions in the adsorbed complexes. In the proposed tridentate surface complexes, both oxygen atoms from the carboxylate group exchange with surface (hydr)oxide ligands from neighboring TiOH2+0.5 groups to form inner-sphere coordination complexes, and a third TiOH2+0.5 surface site donates an H-bond to the quinoline ring carbonyl oxygen. The only difference between the two structures is ion pairing of the protonated nonbonding amine group with ClO4− in the lower structure. Formation of the upper structure in Figure 3C (nonion-paired) can be formulated as follows: OFX− + 4H+ + 3 ≡ TiOH−0.5 ⇌ [ ≡ Ti3OH 2(H − OFX)]+1.5 + 2H 2O

(2)

The bridging oxygen atoms on the anatase surface are assumed to be nonexchangeable and therefore do not participate in ligand exchange reactions. Describing charge distribution in the surface complexes involves calculating the changes in charge that occur at the 0plane (Δz0) and the Stern plane (Δzs) during surface complex formation. Inner-sphere bidentate binding of the carboxylate group places −1 valence units (v.u.) of charge at the 0-plane. Formation of an H-bond where the adsorbing functional group acts as an H-bond acceptor is expected to place an additional −0.2 v.u. of charge at the surface.42 Thus, the proposed tridentate structure is expected to have a CD value at the surface near −1.2 v.u. The net change in Δz0 and Δzs then are determined by the collective consideration of the CD value from the complex, the overall charge of the adsorbing molecular species, the number of protons involved in the surface complexation reaction, and H2O displaced from the surface by OFX surface complexation. For example, for surface complexation of zwitterionic OFX species by the proposed binding mode (Figure 3C, top structure), three protons are brought to the surface (+3 v.u.) to convert three ≡TiOH−0.5 to ≡TiOH2+0.5. This is followed by the exchange of two of the resulting water molecules by the incoming carboxylate oxygen atoms, bringing −1 v.u. to the surface. Formation of a H-bond between the third ≡TiOH2+0.5 group and the quinoline ring carbonyl then brings an additional −0.2 v.u. to the surface, resulting in Δz0 = +3 − 1 − 0.2 = +1.8 v.u. Since the adsorbing zwitterion species of OFX has a net charge of zero, and because −1.2 v.u. is placed at the 0-plane, an equal and opposite amount of charge must be placed at the Stern plane (Δzs = +1.2 v.u.) to maintain net neutrality of the zwitterionic molecule.

OFX− + 4H+ + 3 ≡ TiOH−0.5 + ClO4 − ⇌ [ ≡ Ti3OH 2(H − OFX)+1.5 ...ClO4 −] + 2H 2O

(3)

Simultaneous fit of all five measured pH adsorption edges in NaClO4 using the two tridentate surface complexes, where the only adjustable parameter was the binding constant of the ionpaired complex (the log K and CD values for nonion-paired complex were fixed at values determined from the NaCl system and the CD values for both complexes were assumed identical), yielded an optimum log K = 17.4 ± 1.7. Figure 3F shows an example of the predicted distribution of the ion-paired (IP) and nonion-paired (Non-IP) species formed as a function of pH. The relative contribution of the IP complex increases with decreasing pH and increasing ionic strength and OFX concentration. The increasing concentration of the IP complex at low pH is consistent with ATR-FTIR data showing increased coadsorption of ClO4− with OFX at low pH. Implications for FQ Photocatalysis. The proposed innersphere complexation mechanism for adsorption of OFX and related FQ antibacterial agents to TiO2 is consistent with the proposed visible light-sensitive charge transfer mechanism.11 11902

dx.doi.org/10.1021/es302097k | Environ. Sci. Technol. 2012, 46, 11896−11904

Environmental Science & Technology

■

Strong coupling between the ground state of the chromophore (FQ) with Ti surface states lying in the semiconductor band gap can give rise to a new red-shifted band in the spectrum of TiO2.47 The red shift allowing visible light photoexcitation of adsorbed FQ was confirmed by UV−vis spectra (Figure 2A). Studies of other organic TiO2 photosensitizers show that the chromophores are often bound to Ti atoms through innersphere complexation involving oxygen-containing functional groups, often carboxylic acids.47 Thus, the red shift in the UV− vis, the visible light photoexcitation of surface complexed OFX, and CD model predictions support inner-sphere ligand exchange reactions as the dominant mode of OFX adsorption to TiO2. The SCM predictions can also help rationalize observed pHdependent trends in the kinetics of visible light photocatalytic oxidation of FQs. FQ photocatalysis rates measured in NaClO4 electrolyte solutions (NaCl was not used since Cl− can act as a radical scavenger in aqueous systems)48 do not exactly follow the pH-dependent trend in total adsorbed FQ. However, application of CD SCM reveals that measured pseudo firstorder rate constants (kobs) for visible light-TiO2 photocatalysis of OFX track closely with the model-predicted concentration of the nonion-paired surface complex described by eq 2 (Figure 4). It remains unclear why ion pairing with ClO4− would inhibit

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This article was developed under a STAR Research Assistance Agreement No. 91683701 awarded by the U.S. Environmental Protection Agency to T.Paul. It has not been formally reviewed by the EPA. The views expressed in this document are solely those of the authors and the EPA does not endorse any products or commercial services mentioned in this publication. This study was also supported in part by a National Science Foundation CAREER award to T.J. Strathmann (CBET074645 CAREER). Tamar Kohn is acknowledged for use of UV−vis spectrophotometer. Moira Ridley (Texas Tech) and James Kubicki (Penn State) are acknowledged for their helpful discussions and critiques.

■

REFERENCES

(1) Walsh, C. Antibiotics: Actions, Origins, Resistance; ASM Press: Washington, D.C., 2003. (2) WHO. Use of quinolones in food animals and potential impact on human health. Report of WHO meeting, Geneva, Switzerland, 1998. (3) Nakata, H.; Kanaan, K.; Jones, P. D.; Giesy, J. P. Determination of fluoroquinolone antibiotics in wastewater effluents by liquid chromatography-mass spectroscopy and fluorescence detection. Chemosphere 2005, 58, 759−766. (4) Kolpin, D. W.; Furlong, E. T.; Meyer, M. T.; Thurman, E. M.; Zaugg, S. D.; Barber, L. B.; Buxton, H. T. Pharmaceuticals, hormones, and other organic wastewater contaminants in U.S. streams, 1999− 2000: A national reconnaissance. Environ. Sci. Technol. 2002, 36, 1202−1211. (5) Xiao, Y.; Chang, H.; Jia, A.; Hu, J. Trace analysis of quinolone and fluoroquinolone antibiotics from wastewaters by liquid chromatography-electrospray tandem mass spectrometry. J. Chromatogr., A 2008, 1214, 100−108. (6) Giger, W.; A.C., A.; Golet, E. M.; Kohler, H.-P. E.; McArdell, C. S.; Molnar, E.; Siegrist, H.; Suter, M.J.-F. Occurrence and fate of antibiotics as trace contaminants in wastewaters, sewage sludges, and surface waters. Chimia 2003, 57, 485−491. (7) Golet, E. M.; Xifra, I.; Siegrist, H.; Alder, A. C.; Giger, W. Environmental exposure assessment of fluoroquinolone antibacterial agents from sewage to soil. Environ. Sci. Technol. 2003, 37, 3243−3249. (8) Yang, L. H.; Ying, G. G.; Su, H. C.; Stauber, J. L.; Adams, M. S.; Binet, M. T. Growth-inhibiting effects of 12 antibacterial agents and their mixtures on the freshwater microalga Pseudokirchneriella subcapitata. Environ. Toxicol. Chem. 2008, 27, 1201−1208. (9) Pomati, F.; Castiglioni, S.; Zuccato, E.; Fanelli, R.; Vigetti, D.; Rossetti, C.; Calamari, D. Effects of a complex misture of therpeutic drugs at environmental levels on human embryonic cells. Environ. Sci. Technol. 2006, 40, 2442−2447. (10) Calza, P.; Medana, C.; Carbone, F.; Giancotti, V.; Baiochhi, C. Characterization of intermediate compounds formed upon photoinduced degradation of quinolones by high performance liquid chromatography/high-resolution multiple-stage mass spectrometry. Rapid Commun. Mass Spectrom. 2008, 22, 1533−1552. (11) Paul, T.; Miller, P. L.; Strathmann, T. J. Visible-light-mediated TiO2 photocatalysis of fluoroquinolone antibacterial agents. Environ. Sci. Technol. 2007, 41, 4720−4727. (12) Paul, T.; Dodd, M. C.; Strathmann, T. J. Photolytic and photocatalytic decomposition of aqueous ciprofloxacin: transformation products and residual antibacterial activity. Water Res. 2010, 44, 3121− 3132.

Figure 4. Comparison of pH-dependent trends for the measured rate constants for visible light-TiO2 photocatalysis of OFX (symbols; right axis) and model predictions of OFX adsorption and the distribution of ion-paired (IP) and nonion-paired (non-IP) species (lines, left axis). Conditions: 100 μM OFX, 0.5 g/L TiO2, 0.01 M NaClO4, irradiation with visible light (λ > 400 nm, 450 W Xe lamp) using setup previously described.11

visible light photocatalysis, but it may be associated with the fact that the piperazine ring (where ion pairing is believed to be centered) is the primary site for oxidative transformations of FQs resulting from visible light photocatalysis.11,12 Further study is needed to examine this issue in greater detail.

■

Article

ASSOCIATED CONTENT

S Supporting Information *

Listing of chemical reagents; additional ATR-FTIR spectra of aqueous, adsorbed, and precipitated OFX; and TiO2 surface potentiometric titration data with model fits are provided. This material is available free of charge via the Internet at http:// pubs.acs.org. 11903

dx.doi.org/10.1021/es302097k | Environ. Sci. Technol. 2012, 46, 11896−11904

Environmental Science & Technology

Article

TiO2: Implications for the biocompatibility of titanium. J. Colloid Interface Sci. 2000, 227, 48−54. (35) Dobson, K. D.; McQuillan, A. J. In situ infrared spectroscopic analysis of the adsorption of aliphatic carboxylic acids to TiO2, ZrO2, Al2O3, and Ta2O5 from aqueous solutions. Spectrochim. Acta, Part A 1999, 55, 1395−1405. (36) Paszti, Z.; Guczi, L. Amino acid adsorption ot hydrophilic TiO2: A sum frequency generation vibrational spectroscopy study. Vib. Spectrosc. 2009, 50, 48−56. (37) Dobson, K. D.; McQuillan, A. J. In situ infrared spectroscopic analysis of the adsorption of aromatic carboxylic acids to TiO2, ZrO2, Al2O3, and Ta2O5 from aqueous solutions. Spectrochim. Acta, Part A 2000, 56, 557−565. (38) Tunesi, S.; Anderson, M. A. Surface effects in photochemistry: An in situ cylindrical internal reflection-fourier transform infrared investigation of the effect of ring substituents on the chemisorption onto TiO2 ceramic membranes. Langmuir 1992, 8, 487−495. (39) Weisz, A. D.; Regazzoni, A. E.; Blesa, M. A. ATR-FTIR study of the stability trends of carboxylate complexes formed on the surface of titaniumd dioxide particles immersed in water. Solid State Ionics 2001, 143, 125−130. (40) Yoshida, A.; Moroi, R. Crystal structure of ofloxacin perchlorate. Anal. Sci. 1991, 7, 351−352. (41) Roberts, J. M.; Diaz, A. R.; Fortin, D. T.; Friedle, J. M.; Piper, S. D. Influence of the Hofmeister series on the retention of amines in reversed-phase liquid chromatography. Anal. Chem. 2002, 74, 4927− 4932. (42) Filius, J. D.; Hiemstra, T.; Van Riemsdijk, W. H. Adsorption of small weak organic acids on goethite: Modeling and mechanisms. J. Colloid Interface Sci. 1997, 195, 368−380. (43) Colon, G.; Hidalgo, M. C.; Navio, J. A. Photocatalytic deactivation of commercial TiO2 samples during simultaneous photoreduction of Cr(VI) and photooxidation of salicylic acid. J. Photochem. Photobiol., A 2001, 138, 79−85. (44) Lazzeri, M.; Vittadini, A.; Selloni, A. Structure and energetics of stoichiometric TiO2 anatase surfaces. Phys. Rev. B: Condens. Matter 2001, 63, 155409. (45) Pauling, L. The principles determining the structure of complex ionic crystals. J. Am. Chem. Soc. 1929, 51, 1010−1026. (46) Bolt, G. H.; Van Riemsdijk, W. H. In Soil Chemistry, Part B: Physico-chemical models; Bolt, G. H., Ed.; Elsevier: Amsterdam, 1982; pp 459−504. (47) Duncan, W. R.; Prezhdo, O. V. Theoretical studies of photoinduced electron transfer in dye-sensitized TiO2. Annu. Rev. Phys. Chem. 2007, 58, 143−184. (48) Liao, C.-H.; Kang, S.-F.; Wu, F.-A. Hydroxyl radical scavenging role of chloride and bicarbonate ions in the H2O2/UV process. Chemosphere 2001, 44, 1193−1200.

(13) Gu, C.; Karthikeyan, K. G. Sorption of the antibacterial ciprofloxacin to aluminum and iron hydrous oxides. Environ. Sci. Technol. 2005, 39, 9166−9173. (14) Goyne, K. W.; Chorover, J.; Kubicki, J. D.; Zimmerman, A. R.; Brantley, S. L. Sorption of the antibiotic ofloxacin to mesoporous and nonporous alumina and silica. J. Colloid Interface Sci. 2005, 283, 160− 170. (15) Trivedi, P.; Vasudevan, D. Spectroscopic investigation of ciprofloxacin speciation at the goethite-water interface. Environ. Sci. Technol. 2007, 41, 3153−3158. (16) Zhang, H.; Huang, C.-H. Adsorption and oxidation of fluoroquinolone antibacterial agents and structurally related amines with goethite. Chemosphere 2007, 66, 1502−1512. (17) Zhang, H.; Huang, C.-H. Oxidative transformation of fluoroquinolone antibacterial agents and structurally related amines by manganese dioxide. Environ. Sci. Technol. 2005, 39, 4474−4483. (18) Nowara, A.; Burhenne, J.; Spiteller, M. Binding of fluoroquinolone carboxylic acid derivatives to clay minerals. J. Agric. Food Chem. 1997, 45, 1459−1463. (19) Carrasquillo, A. J.; Bruland, G. L.; Mackay, A. A.; Vasudevan, D. Sorption of ciprofloxacin and oxytetracycline zwitterions to soils and soil minerals: Influence of compound structure. Environ. Sci. Technol. 2008, 42, 7634−7642. (20) Pei, Z.; Shan, X.-Q.; Kong, J.; Wen, B.; Owens, G. Coadsorption of ciprofloxacin and Cu(II) on montmorillonite and kaolinite as affected by solution pH. Environ. Sci. Technol. 2010, 44, 915−920. (21) Pei, Z.-G.; Shan, X.-Q.; Zhang, S.-Z.; Kong, J.-J.; Wen, B.; Zhang, J.; Zheng, L.-R.; Xie, Y.-N.; Janssens, K. Insight to ternary complexes of co-adsorption of norfloxacin and Cu(II) onto montmorillonite at different pH using EXAFS. J. Hazard. Mater. 2011, 186, 842−848. (22) Sheals, J.; Persson, P.; Hedman, B. IR and EXAFS spectroscopic studies of glyphosate protonation and copper(II) complexes of glyphosate in aqueous solution. Inorg. Chem. 2001, 40, 4302−4309. (23) Hiemstra, T.; Van Riemsdijk, W. H. A surface structural approach to ion adsorption: the charge distribution (CD) model. J. Colloid Interface Sci. 1996, 179, 488−508. (24) Bekbolet, M.; Suphandag, A. S.; Uyguner, C. S. An investigation of the photocatalytic efficiencies of TiO 2 powders on the decolourisation of humic acids. J. Photochem. Photobiol., A 2002, 148, 121−128. (25) Bourikas, K.; Hiemstra, T.; Van Riemsdijk, W. H. Ion pair formation and primary charging behavior of titanium oxide (anatase and rutile). Langmuir 2001, 17, 749−756. (26) Lindegren, M.; Loring, J. S.; Persson, P. Molecular structures of citrate and tricarballylate adsorbed on α-FeOOH particles in aqueous suspensions. Langmuir 2009, 25, 10639−10647. (27) Ross, D. L.; Riley, C. M. Aqueous solubilities of some variously substituted quinolone antimicrobials. Int. J. Pharm. 1990, 63, 237−250. (28) Lizondo, M.; Pons, M.; Gallardo, M.; Estelrich, J. Physicochemical properties of enrofloxacin. J. Pharm. Biomed. Anal. 1997, 15, 1845−1849. (29) Langlois, M.-H.; Montagut, M.; Dubost, J.-P.; Grellet, J.; Saux, M.-C. Protonation equilibrium and lipophilicity of moxifloxacin. J. Pharm. Biomed. Anal. 2005, 37, 389−393. (30) Sun, J.; Sakai, S.; Tauchi, Y.; Deguchi, Y.; Chen, J.; Zhang, R.; Morimoto, K. Determination of lipophilicity of two quinolone antibacterials, ciprofloxacin and grepafloxacin, in the protonation equilibrium. Eur. J Pharm. Biopharm. 2002, 54, 51−58. (31) Dorofeev, V. L. Infrared spectra and the structure of drugs of the fluoroquinolone group. Pharm. Chem. J. 2004, 38, 693−697. (32) Soulen, J. R.; Schwartz, W. F. Infrared spectrum of nitryl perchlorate. J. Phys. Chem. 1962, 66, 2066−2067. (33) Deacon, G. B.; Phillips, R. J. Relationships between the carbonoxygen stretching frequencies of carboxylate complexes and the type of carboxylate coordination. Coord. Chem. Rev. 1980, 33, 227−250. (34) Roddick-Lanzilotta, A. D.; McQuillan, A. J. An in situ infrared spectroscopic study of glutamic acid and aspartic acid adsorbed on 11904

dx.doi.org/10.1021/es302097k | Environ. Sci. Technol. 2012, 46, 11896−11904