8796
J. Phys. Chem. C 2007, 111, 8796-8804
Environmental Biogeochemistry Studied by Second-Harmonic Generation: A Look at the Agricultural Antibiotic Oxytetracycline† Patrick L. Hayes, Julianne M. Gibbs-Davis, Michael J. Musorrafiti, Amanda L. Mifflin,‡ Karl A. Scheidt, and Franz M. Geiger* Department of Chemistry, Northwestern UniVersity, 2145 Sheridan Road, EVanston, Illinois 60208 ReceiVed: NoVember 1, 2006; In Final Form: January 5, 2007
The binding behavior of the veterinary antibiotic oxytetracycline (OTC) to tailor-made environmental interfaces was investigated using second-harmonic generation (SHG). Specifically, four environmentally relevant chemical moieties were covalently tethered to fused quartz surfaces using silane and amide coupling chemistry to survey how amides, polyfunctional binding sites, and phenyl rings that are common in humic acids control OTC binding at pH 8. The model geochemical surfaces prepared for this study were the following: methylamide-terminated fused quartz as well as amide-linked carboxylic acid-, benzoic acid-, and benzylterminated fused quartz. We find that amide-linked benzoic acid-terminated silanes display the highest equilibrium binding constant (1.1 × 106 M-1) among the systems surveyed. A straightforward predictive method of using contact angle measurements and the corresponding interfacial energy densities to predict OTC mobility across humic acid-containing mineral/water interfaces is discussed in the context of the emerging bacterial antibiotic resistance development threat.
I. Introduction A. Agricultural Pharmaceuticals in Soils and Antibiotic Resistance. The emerging threat of bacterial antibiotic resistance due to the agricultural use of pharmaceuticals is now receiving much attention.1-6 This is underlined by last year’s FDA ban of the veterinary antibiotic Baytril because of its chemical similarity to ciprofloxacin, which is used in humans to combat anthrax and other bacterial infections.7 Additionally, the European Union banned the use of antibiotics for growth promotion in livestock in 2006, although antibiotics can still be used to treat ill animals.8 Interestingly, the practice of feeding healthy animals routine doses of antibiotics is not prohibited in the United States,5 even though many antibiotics administered to these farm animals are excreted 50-80% unmetabolized.9 The excreted portion of antibiotics poses an environmental and biological concern; recent monitoring studies, such as the one carried out by the United States Geological Survey, have reported low levels of a wide range of pharmaceuticals, including hormones, steroids, and antibiotics in soils, surface waters, and groundwaters.10,11 Although the measured pharmaceutical concentrations are usually less than 1 µg/L in surface waters, it appears that these levels are constant throughout the year and over a wide range of hydrological, climatic, and landuse settings.10 In addition, it has been found that veterinary antibiotics in manure can be transported via overland water flow into the surrounding soil during precipitation events. Specifically, oxytetracycline (OTC), a human and veterinary antibiotic, has been found at a concentration of 72 µg/L in surface waters near plots treated with OTC-contaminated manure.12 Here, we apply second-order spectroscopy to address the mobility of OTC across various environmental interfaces. †
Part of the special issue “Kenneth B. Eisenthal Festschrift”. * To whom correspondence should be addressed. E-mail:
[email protected]. ‡ Current address: Department of Chemistry, Bowdoin College, Brunswick, ME 04011.
The antibiotic oxytetracycline (Figure 1) was chosen for this study for multiple reasons. OTC is a member of the tetracycline family of antibiotics, a class of pharmaceuticals important for treating bacterial illnesses such as acne, pneumonia, and Lyme disease in humans.13 However, tetracycline antibiotics are widely used as nontherapeutic feed additives in the United States, with 3 million pounds employed annually for growth promotion of livestock.14 Not surprisingly, antibacterial resistance has already been measured in soil bacteria from sites that were treated with OTC-contaminated pig manure, and furthermore, the effectiveness of tetracyclines in treating illnesses has diminished significantly in recent years.13 Numerous studies have been conducted that examine OTC spectroscopy and sorption behavior.9,12,14-19 They indicate that OTC sorption is influenced by soil texture, cation exchange capacity, soil pH, soil clay content, and iron oxide content. Furthermore, a study by Kulshrestha et al. suggests that OTC sorption in alkaline soils is dominated by hydrophobic interactions rather than cationic exchange.17 Recently, our group reported the use of resonantly enhanced second-harmonic generation (SHG) to study the redox-inactive binding of oxytetracycline to fused quartz/water and chemically functionalized fused quartz/water interfaces.20 This work showed that OTC binding to fused quartz/water interfaces and those functionalized with methyl ester- and carboxylic acid-terminated silanes is fully reversible and highly dependent on solution pH, with the highest levels of adsorption occurring at pH 8. This pH dependence is consistent with the complex acid/base chemistry of the OTC molecule18 and suggests that OTC binds most effectively when in the monoanion form at pH 8. B. Humic Substances and Their Role In Pollutant Transport. Given the connection between agricultural antibiotic use and the development of antibiotic resistance in soil bacteria, it is critical to understand the mobility of antibiotics in soils. The mobility of an antibiotic in soil will determine the extent of the
10.1021/jp0672149 CCC: $37.00 © 2007 American Chemical Society Published on Web 03/22/2007
Environmental Biogeochemistry
Figure 1. Molecular structure of OTC‚HCl.
Figure 2. Synthetic strategy utilized for the synthesis of methylamide-, carboxylic acid-, benzoic acid-, and benzyl-functionalized surfaces. Environmentally relevant moieties (ERMs) are covalently attached to the quartz surface through amide coupling chemistry.
spatial and temporal exposure of bacteria to a contaminant plume. Mineral oxide/water interfaces can play a crucial role in controlling contaminant mobility in soils.21-24 In the environment, natural organic matter, including humic substances, can aggregate on mineral oxide surfaces. This aggregation can create a heterogeneous solid/water interface where aqueous-phase pollutant species can adsorb to chemical functional groups within the immobilized organic adlayer.25 Through their interaction with geochemically important dissolved species, humic substances and natural organic matter immobilized on mineral oxide surfaces can have a profound impact on pollutant transport. The structure of humic substances has been the source of much debate in the scientific community, but a new concept of humic substances is now emerging in which low molecular weight and chemically diverse organic molecules form supramolecular clusters via hydrogen bonds.26 In our previous work, we addressed the chemical complexity characteristic of humic substances by quantifying the interactions between a toxic metal ion, namely, the EPA priority pollutant chromium VI,27-30 as well as OTC20 with fused quartz/water interfaces containing tailor-made organic moieties commonly found in humic substances from a thermodynamic and kinetic perspective. The interaction of OTC with carboxylic acid-, ester-, and alkylfunctionalized fused quartz/water interfaces was found to depend strongly on the chemical functionality present at the surface, with methyl esters exhibiting equilibrium binding constants approximately 7 times larger than those found for carboxylic acid-functionalized fused quartz/water interfaces. In this work, we explore the relationship between interfacial chemical composition and OTC binding at a solid/water interface further. Specifically, we utilize a versatile synthetic strategy where an environmentally relevant moiety is tethered to the fused quartz surface via an alkyl amide linkage (Figure 2).31,32 Amide-coupling synthetic chemistry was specifically chosen because it allows for facile access to new chemical moieties. This allows us to explore the possibilities of different OTC
J. Phys. Chem. C, Vol. 111, No. 25, 2007 8797 binding properties when compared to our earlier work,20 where environmentally relevant moieties were tethered directly to the fused quartz surface via alkyl silanes. In addition, amidecoupling is important from a geochemical perspective because amides are the most prevalent form of nitrogen in humic substances.26 Each tailor-made surface was designed so that specific types of OTC binding interactions with humic substances could be studied. In this work, we studied fused quartz/water interfaces functionalized with methylamide-terminated as well as amidelinked carboxylic acid-, benzoic acid-, and benzyl-terminated silanes. The aromatic systems were chosen because humic acid has been reported to be composed of approximately 30% aromatic carbons.26 In addition, the amide-linked carboxylic acid-terminated surface, which contains a multidentate chelation group, provides a model for polyfunctional binding sites in humic substances. (The amide-linked benzoic acid-terminated surface is also multidentate and serves as a model for polyfunctional binding sites with bulky, rigid phenyl rings that hinder chelation.) The third surface studied, the methylamide-functionalized fused quartz, accesses the amide group by itself, which is again important given the fact that amides are the most prevalent form of nitrogen in humic substances.26 Adsorption isotherms recorded using SHG yield surfacespecific thermodynamic binding parameters at environmentally relevant OTC concentrations, which allow us to elucidate the role that specific interactions and moieties play in the interaction of antibiotics with geosorbents. Prior to this work, neither aromatic acids nor chelating groups, such as the amide-linked carboxylic acid, had been studied with respect to their control of OTC mobility. Such information is expected to be vital for developing accurate transport models for oxytetracycline in the environment. More generally, this work is important for assessing the risk of antibiotic resistance development in soil biota due to the presence of possibly mobile veterinary antibiotics in various soil types and heterogeneous geochemical environments. II. Experimental Section In our earlier work,20 we showed that the well-known electronic transitions of OTC33,34 allow for resonantly enhanced SHG studies of OTC interacting with aqueous/solid interfaces. The second-harmonic signal intensity, ISHG, is related to the 35-38 As can second-order susceptibility of the interface, χ(2) int . (2) been seen below in eq 1, χint consists of both a nonresonant (2) and resonant contribution, χ(2) NR and χR , respectively. Further(2) more, χint also contains phase information linking the resonant and nonresonant contributions to the SHG signal.
xISHG ) ESHG ∝
x| | x| T(2) 2 ) χ int
T(2) T(2) χ NR + χ R ei∆φ
|
2
(1)
For resonantly enhanced SHG, the phase difference, ∆φ, is typically taken to be 90°,28,39,40 which eliminates the cross product resulting from the square modulus in eq 1. Equation 2 shows that the resonant contribution to second-order susceptibility, χ(2) R , can be modeled as the product of the adsorbate number density at the interface, Nads, and the second-order molecular hyperpolarizability, R(2), averaged over all molecular orientations.
〈 〉 (2)
T χ(2) R ) Nads R
(2)
8798 J. Phys. Chem. C, Vol. 111, No. 25, 2007
Hayes et al.
Figure 3. Experimental setup and flow cell used in the SHG measurements.
Therefore, from eqs 1 and 2, SHG can be used to probe adsorbate surface coverage and to perform adsorption isotherm experiments. Furthermore, polarization-resolved SHG experiments can provide information on adsorbate orientation.41-43 For SHG measurements of surface adsorbates, R(2) can be calculated from a summation over all electronic excited states:
〈 〉
2 3
T(2) ) -4π e Rijk h2
∑ b,c (ω
〈a|µ bi|b〉〈b|µ bj|c〉〈c|µ bk|a〉 ba - ω - iΓba)(ωca - 2ω - iΓca)
(3)
Here, e is the charge of the electron, h is Planck’s constant, Γ represents the damping coefficient for each electronic transition, b µ is the electric dipole moment operator and a, b, and c represent the ground, intermediate, and final states, respectively.36,37,44 Resonance enhancement of the second-harmonic signal results from the increase in second-order hyperpolarizability when 2ω approaches a resonant frequency, ωca, in the adsorbate. This increase in R(2) then corresponds to a higher SHG efficiency (eq 2). Resonance enhancement allows for interface-specific spectroscopic studies in the UV-vis spectral region. As described in detail previously,45,46 SHG studies were performed using a regeneratively amplified Ti:Sapphire laser (Hurricane, Spectra Physics) pumping an optical parametric amplifier (OPA-CF, Spectra Physics). After focusing the fundamental probe light (614 nm) onto the aqueous/solid interface under investigation, the fundamental light field was filtered out using Schott filters and the SHG signal at 307 nm was detected using a monochromator and gated photon-counting system. This SHG wavelength is in resonance with one of the electronic transitions of OTC20 and allows us to track the presence of OTC at the various interfaces studied. The quadratic power dependence of the SHG signal was verified regularly. The experimental setup is pictured in Figure 3. For the experiments at the aqueous/solid interface, a fused quartz hemispherical lens (ISP Optics) was clamped upon the open top of a custom-built Teflon flow cell. A Viton O-ring was used to ensure the flow cell remained leak-tight. Variable flow peristaltic pumps were used to pump Millipore water and aqueous oxytetracycline hydrochloride (VWR) solutions through the flow cell. All water and oxytetracycline solutions were kept at pH 8 using NaOH (Spectrum Chemicals) and HCl (VWR) solutions. At this pH and over the experimentally relevant OTC concentrations (5 × 10-7-3 × 10-5 M), the extinction coefficient measured at 360 nm is found to be 1.40(1) × 104 cm-1M-1. Quartz lenses functionalized with N-hydroxysuccinimide (NHS)-ester-terminated alkyl silanes were prepared via our
previously published synthesis.31 NHS groups are good leaving groups when reacted with primary amines47 and can be easily replaced with the chemical moiety of interest. Thus, the NHSester-functionalized lens was placed in a custom-built Teflon reaction vessel that confines solutions of the primary amine of interest to the flat surface of the hemisphere. Generally, the amine containing the chemical moiety of interest (0.372 mmol) was dissolved in pH 8 borate buffer (4 mL) and placed in the reaction vessel to react with the NHS-functionalized quartz surface for 15 h (Figure 2). In the present study, the compounds used for amide-coupling (all from Aldrich, used as received) were methylamine (1), γ-aminobutyric acid (2), 4-(aminomethyl)benzoic acid (3), and benzyl amine (4). For the functionalization with methylamine, a 40% by weight aqueous solution of methylamine was placed in the reaction vessel for 8 h to react with the functionalized quartz surface. The amines studied in this work were selected because they contain, after being covalently tethered to the surface, environmentally relevant moieties (carboxylic acids, amides and aromatic groups) that will be displayed toward the OTC-containing bulk solution in the SHG experiments. Following reaction, the lenses were rinsed with copious amounts of Millipore water and then stored under Millipore water. The functionalized surfaces were characterized by time-offlight secondary-ion mass spectroscopy (ToF-SIMS, Physical Electronics, PHI TRIFT III), ellipsometry (Sopra Inc., MOSS ES4G/OMA), and contact angle measurements (First Ten Ångstroms, FTÅ 125). Interfacial pKa measurements were taken using the χ(3) method.48,49 The results of these analyses were consistent with the synthesis presented in Figure 2 as well as a high surface coverage and linking efficiency: ToF-SIMS was performed on surfaces functionalized with 4-fluorobenzylamine and shows a positive ion peak at 109 m/z, which corresponds to the expected molecular fragment for cleavage between the carbonyl carbon and nitrogen of the amide group. Given the fact that the unreacted NHS-ester surface does not display the same peak at 109 m/z, amide coupling clearly occurs at the surface. Ellipsometry measurements for silicon wafers functionalized by the same method as the fused quartz substrates show an organic layer thickness for the amide-linked carboxylic acid-terminated surface and the unreacted NHS-ester-terminated surface of 1.9(2) nm and 1.9(1) nm, respectively. This thickness is consistent with the length of the NHS-ester, which contains 10 CH2 groups in addition to the N-oxysuccinimidyl ester. The ellipsometry results for the amide-linked carboxylic acid, which contains three CH2 groups between the amide moiety and the acid terminus, indicate that the butyric acid portion does not stand up straight but lies relatively flat on the surface. It is important to note, however, that the ellipsometry measurements were carried out in air and not under water. Contact angles measured for all of the surfaces discussed in this work are reported in Table 1 and indicate substantial increases in hydrophobicity with decreasing polarity of the nitrogen-bound portion of the silane molecule. III. Results and Discussion A. Adsorption/Desorption Traces. Adsorption/desorption experiments were performed at pH 8 by exposing the functionalized quartz/water interfaces to aqueous OTC solutions. After recording an SHG baseline for a few minutes while flowing water held at pH 8 across the interface, the OTC solution maintained at pH 8 replaces the water flow. Following about 10 min of OTC flow, water at pH 8 is sent again across the interface.
Environmental Biogeochemistry
J. Phys. Chem. C, Vol. 111, No. 25, 2007 8799
TABLE 1: Summary of Thermodynamic Binding Parameters, Contact Angle Measurements, Partition Coefficients, and Retardation Factors for OTC Binding to Surfaces Investigated at pH 8
a Error in parentheses is one standard deviation and is calculated from the Langmuir isotherm fit. b Free-energy values are referenced to the molarity of water under standard conditions (55.5 M). c Thermodynamic parameters, partition coefficients, and retardation factors taken from Mifflin et al.20
The results of these adsorption/desorption experiments using bulk OTC concentrations resulting in monolayer surface coverage (see below) are shown in Figure 4. From eq 2, the SHG E-field is proportional to the interfacial OTC number density. Thus, the SHG E-field increase observed for the methylamideand the amide-linked benzoic acid- and the carboxylic acidfunctionalized fused quartz/water interfaces indicates that OTC adsorbs to the interface. The SHG E-field reaches steady state within 5 min of switching the water flow to the aqueous OTC solution flow, and a similar equilibration time is needed when switching back to water. Clearly, Figure 4 shows that the adsorption and desorption processes are fully reversible and suggest a high mobility of OTC when humic substances containing carboxylic acid and amide groups are present in silica-rich soil environments. B. Orientation Studies. As shown in eq 2, adsorption isotherms measured by SHG depend not only on the adsorbate number density but also on the adsorbate orientation. Using polarization-resolved SHG experiments41 we tested whether the SHG E-field polarization changed with OTC surface coverage or with the chemical nature of the interface. This experiment was carried out by performing null angle measurements using fundamental probe light polarized at a 45° angle from the plane of incidence while recording the SHG signal intensity as a function of the polarizer-analyzer angle.41 These null angle measurements were performed first on OTC monolayers on bare fused quartz/water interfaces and those functionalized with the amide-linked carboxylic acid and the direct-linked methyl ester. The results are shown in Figure 5, which indicates that the null angles are located within 15° from one another for the three interfaces at monolayer OTC coverage. This suggests that the average orientation of the chromophore giving rise to the SHG resonance enhancement within OTC does not depend significantly on the chemical identity of the interface. We then carried out null angle measurements on the amide-linked carboxylic acid-functionalized fused quartz/water interface containing
submonolayer (50%) and monolayer amounts of OTC, as determined from the adsorption isotherms (see below). The results are shown in the inset of Figure 5, which indicates that for both surface coverages the SHG response is lowest at a polarizer-analyzer angle between 120° and 130°. This indicates that for this surface the average orientation of the chromophore that gives rise to the SHG resonance enhancement within OTC does not depend significantly on OTC surface coverage. Although the structural complexicity of OTC prevents us from representing this molecule as a rod-like chromophore to obtain molecular orientation parameters, such as tilt angles,40 the coverage-independent null angles are consistent with little or no changes in the average OTC orientation with increasing OTC surface coverage. C. Relative Surface Coverages. Given our finding that the SHG null angles are quite similar for three interfaces under investigation in this work and independent of surface coverage, the SHG E-field increases observed for OTC monolayers at the various interfaces studied here can be used to compare the extent of adsorption. This information is important for assessing the extent to which the various chemical functional groups control the uptake of antibiotics at environmental interfaces. Figure 6 shows that the average background-subtracted SHG E-field obtained for OTC monolayers on the methylamide-, carboxylic acid-, and benzoic acid-functionalized surfaces is approximately identical, indicating similar OTC coverages on those surfaces. When OTC is in contact with a fused quartz/water interface functionalized with an amide-linked phenyl ring (no carboxylic acid groups present), the SHG E-field is about 40% of that obtained for the polar functional groups. This difference underscores that terminal polar chemical moieties, such as carboxylic acids or methylamides, are needed for effective antibiotic binding. The relative background-subtracted SHG E-fields observed for the amide-linked functionalized surfaces studied in this work are approximately 2 times lower than those observed in our
8800 J. Phys. Chem. C, Vol. 111, No. 25, 2007
Hayes et al.
Figure 5. SHG intensity for 45-in polarized probe light as a function of polarizer-analyzer angle for OTC monolayers at fused quartz/water interfaces functionalized with direct-linked methyl esters (crosses and dashed sinusoidal fit line) and amide-linked carboxylic acids (empty circles and dotted line). Solid squares and the solid line correspond to the bare fused quartz/water interface. Inset: Normalized SHG intensity for 50% (filled circles) and monolayer (empty circles) OTC coverage on the amide-linked carboxylic acid-terminated functionalized fused quartz/water interface. The solid and dashed lines are fits of the data to sine functions.
Figure 4. OTC adsorption/desorption traces for (from bottom) amidelinked benzyl-, methylamide-, carboxylic acid-, and benzoic acidfunctionalized fused quartz/water interfaces recorded at pH 8 using monolayer regime bulk OTC concentrations (6.0 × 10-6 M, 7.1 × 10-6 M, 3.1 × 10-5 M, and 3.6 × 10-6 M for benzyl-, methylamide-, carboxylic acid-, and benzoic acid-terminated surfaces, respectively). Traces are offset for clarity. The brackets immediately to the right of the traces approximately represent the change in average SHG signal when OTC adsorbs to an interface.
previous work on OTC interaction with carboxylic acid- and methyl-ester functionalized fused quartz/water interfaces.20 It is important to note that in our previous work the surfaces studied did not contain amide bonds, which could form extended hydrogen bond networks, similar to β-sheets recently reported by Song et al. on similar amide-functionalized surfaces.50 This lateral hydrogen-bond network could prevent OTC from intercalating into the hydrophobic portion of the organic layers, and thus decrease the surface coverage. Alternatively, the lower surface coverages may also be due to the high pH-dependence of OTC-surface interactions.9,15,17,18 Although pH 8 was found to be ideal for binding to the direct-linked surfaces,20 the presence of the amide bond may perturb the interfacial pKa values and thus the optimum bulk solution pH for OTC binding. D. Adsorption Isotherms. Plotting the SHG E-field as a function of increasing OTC concentration in the aqueous solution yields OTC adsorption isotherms for the methylamide-, carboxylic acid-, and benzoic acid-functionalized fused quartz/ water interfaces. Adsorption isotherms recorded for the three interfaces at pH 8 are shown in Figure 7 with the corresponding Langmuir adsorption model fits. Each data point in Figure 7 corresponds to a single adsorption/desorption experiment (as
Figure 6. Normalized and background-subtracted SHG E-fields obtained for monolayer OTC coverages on amide-linked methylamide-, carboxylic acid-, benzoic acid-, and benzyl-functionalized fused quartz/ water interfaces at pH 8. The error bars represent one standard deviation from the average monolayer SHG signal.
described in part A of this section) run at the concentration specified along the x axis. After subtracting out the nonresonant contribution (see eq 1), the equilibrium resonant SHG signal measured upon adsorption can be plotted versus OTC concentration to obtain a plot that has the appearance of a Langmuir isotherm. Normalizing this plot to its monolayer region then converts the SHG signal to surface coverage (θ) providing the isotherms in Figure 7. A detailed analysis of the Langmuir adsorption model51-53 applied to SHG adsorption isotherms while paying specific attention to the phase relationship between the nonresonant and resonant nonlinear susceptibilities has been
Environmental Biogeochemistry
Figure 7. Adsorption isotherms measured at pH 8 and 300 K for OTC on methylamide-, carboxylic acid-, and benzoic acid-functionalized fused quartz/water interfaces. Solid lines are fits of the Langmuir model to the data.
published previously.27 The thermodynamic parameters obtained from the isotherms are summarized in Table 1. When referenced to a 55.5 M aqueous solution,53 the adsorption free energies, obtained from the Langmuir model fits to the data, ranged from -42 to -45 kJ/mol. These values are similar to the free energies observed previously for the direct-linked functionalized fused quartz and bare fused quartz at pH 8 20 and are consistent with a combination of hydrophobic and hydrogen-bonding interactions (e.g., physisorption). Table 1 shows that the amide-linked benzoic acid-functionalized surface displays the highest binding constant, 1.1(1) × 106 M-1, among the different interfaces studied in this and our previous20 work. Not surprisingly, this interface also reaches monolayer coverage at much lower OTC bulk concentrations (5 × 10-6 M) when compared to the methylamide and the amide-linked carboxylic acid-functionalized fused quartz/water interfaces. This indicates that the binding propensity of carboxylic acid groups toward OTC is very different depending on whether the acid is aromatic or aliphatic. The more efficient binding of OTC may be due to π-stacking interactions between the surface-bound benzoic acid moiety and the conjugated structures in oxytetracycline. Such π-stacking interactions have already been discussed in the literature as a possible mechanism for the aggregation of aromatic organic compounds with dissolved organic carbon.25 Alternatively, the geometry and rigid
J. Phys. Chem. C, Vol. 111, No. 25, 2007 8801 structure of the aromatic ring may improve binding to oxytetracycline. However, it is important to note that amide-linked phenyl rings (no carboxylic acid groups present) exhibit low relative surface coverages, which prevent us from collecting adsorption isotherm data below ∼50% of a monolayer. Surfacebound aromatic structures other than those studied in this work, and their effect on the adsorption of OTC to the solid/water interface, will be examined in future experiments. An enhancement in the interaction of OTC with the various interfaces under investigation in this work is observed when OTC samples the carboxylic acid and the amide moiety within the same organic adlayer. Specifically, the binding constant of the amide-linked carboxylic acid-functionalized surface, 6(1) × 105 M-1, is about 6 times larger than that obtained for the direct-linked carboxylic acid-functionalized surface, 1.1(3) × 105 M-1.20 Thus, even if the amide moiety is not the terminus of the organic adlayer, it still enhances OTC binding to a carboxylic acid-terminated surface. Future work will investigate the carbon-chain length dependence of this binding enhancement by increasing the distance between the amide and the acid groups. The methylamide- and the amide-linked carboxylic acidfunctionalized fused quartz/water interfaces exhibit binding constants of 4.9(8) × 105 M-1 and 6(1) × 105 M-1, respectively. These two binding constants are within error, even though the chemical nature of these two amide-linked systems differs substantially. It should be noted that these two binding constants are comparable to the one obtained for OTC interaction with the direct-linked methyl ester-functionalized fused quartz/water interface,20 7(1) × 105 M-1. Given the structural similarities between the methyl ester and the methylamide, this similarity in the binding constants can be understood. However, our finding that the addition of a butyric acid group to an amide linker does not result in significantly stronger OTC binding indicates that considerations other than simple hydrogen-bonding interaction between OTC and the organic adlayers should be considered, which is discussed in the following section. E. Interfacial Energy Density. The key thermodynamic driving force for physisorption is the lowering of the interfacial free-energy density upon adsorption.52,54 Provided hydrophilic and hydrophobic interactions are possible for a given adsorbatesurface system, high interfacial free-energy densities should result in strong adsorbate-surface interactions and therefore high binding constants. In contrast, low interfacial free-energy densities should result in weak interactions and therefore low binding constants, even if hydrophilic and hydrophobic interactions are possible. To test whether the binding constants depend on the interfacial free-energy densities of the systems under investigation in this work, we plotted the binding constants determined from the Langmuir model fits to the experimentally determined adsorption isotherms against the measured sessile contact angles for the various interfaces studied in this work (no OTC present) and related those contact angles to the interfacial energy density. Figure 8A indicates that the binding constants increase with increasing contact angles for all of the surfaces with hydrogenbonding capability. The binding constants from the Langmuir fits were then plotted versus the liquid-solid (e.g., water-solid) interfacial free-energy densities for all of the surfaces (top x axis in Figure 8A). The liquid-solid interfacial free-energy density was calculated from the contact angles using Young’s equation54 and the FTÅ-modified extended Girifalco, Good, Fowkes, and Young model.55-58 The water-air interfacial freeenergy density at 298 K was taken to be 72 mJ/m2.59
8802 J. Phys. Chem. C, Vol. 111, No. 25, 2007
Hayes et al. IV. Environmental Implications
Figure 8. (A) Equilibrium binding constants plotted vs the sessile contact angles (bottom x axis) for each of the surfaces studied. The corresponding water-solid interfacial free-energy density is marked on the top x axis. Results are shown for (a) bare fused quartz/water and (b) direct-linked carboxylic acid-, (c) methylamide-, (d) amide-linked carboxylic acid-, (e) direct-linked methyl ester-, (f) amide-linked benzoic acid-, and (g) octadecyl silane-functionalized fused quartz/water interfaces. Inset: CCD camera images of the sessile water drops for contact angles of 49°, 67°, 75°, and 96°. (B) Retardation factors plotted vs the water-solid interfacial free-energy density, same letter notation as in Figure 8A. The solid line is the result of a linear leastsquares analysis of the data. The dashed lines show the interfacial freeenergy densities for Aldrich and IHSS humic acids and the expected retardation factors.
Figure 8A shows that for the functionalized surfaces capable of hydrogen bonding and hydrophobic interactions the OTC binding constants scale with the liquid-solid interfacial freeenergy density. Clearly, the interaction efficiency for OTC binding to the various surfaces studied in this work is high when hydrogen-bonding and hydrophobic moieties are present at interfaces with high-energy density. Other properties besides energy density, such as the acidity of the surface-bound organic groups, may determine the interaction efficiency as well. For instance, the interaction of hydrophobic portions of the OTC molecule with completely hydrophobic organic adlayers, such as those present at the octadecylsilane-functionalized fused quartz/water interface, is clearly not sufficient to promote substantial binding even though those surfaces may display high interfacial energy densities.
In the complex structures of humic substances, many binding sites that consist of many different functional groups exist. Clearly, attempts to obtain thermodynamic information on each of these sites may be experimentally and synthetically unfeasible. Grouping interactions into classes represents one strategy for understanding and predicting OTC binding to humic substances. Indeed, in our earlier work, we proposed using a weighted average of the experimentally determined binding constants for three interaction classes to predict the binding constants for humic acids.20 In light of the isotherm reported for the amide-linked benzoic acid-terminated surface, we add a fourth interaction type to the original three classes, all of which are based on the nature of the OTC binding interaction, to obtain the following interaction classes: 1. Hydrogen-bond donor and acceptor such as fused quartz and carboxylic acid-containing surfaces; 2. Hydrogen-bond acceptor only such as methyl ester- and methylamide-containing surfaces; 3. Hydrophobic-only interactions such as alkyl- and benzylcontaining surfaces; and 4. Interactions with polar aromatic surface structures. The binding constants determined from the adsorption isotherms can be used to predict the mobility of OTC in soil environments. The Kd model23,60 is commonly used by regulatory organizations, including the Environmental Protection Agency, to calculate the extent to which pollution transport is slowed due to heterogeneous binding events at solid/water interfaces. In this model, one calculates a Kd binding parameter from an isotherm plot by converting the OTC surface coverage and bulk OTC concentration to units of gOTC/gquartz and g/mL, respectively.45 A linear least-squares analysis of the submonolayer regime in this plot then yields a slope, which is the Kd value in units of mL/g. The Kd value can then be incorporated into an expression for the retardation factor, Rf:23
Rf ) 1 +
F K n d
(4)
Here, F is the bulk density of the soil, and n is its porosity. Given the typical porosity values corresponding to unconsolidated granular deposits, F/n values for silica-rich soil environments range from 4 to 10 g/cm3.23 Limitations of this model are substantial and discussed, for instance, in our earlier work.27,28,48 However, the simplicity of the Kd model allows us to calculate the Kd and Rf values for OTC binding to the interfaces studied in this and our earlier work. The results are listed in Table 1 and follow the same trend as the thermodynamic parameters because the Kd and Rf values are calculated from the adsorption isotherms. The benzoic acid-functionalized fused quartz/water interface displays the highest Kd value of 0.11(1) mL/g (or Rf ) 1.44-2.10) and the methylamidefunctionalized fused quartz/water interface displays the lowest Kd value of 0.053(4) mL/g (or Rf ) 1.21-1.53). These values indicate that OTC will be least mobile in benzoic acid-rich soil environments at pH 8. Specifically, the retardation factor for the benzoic acid system suggests that OTC will move 50-70% as far as the free-flowing groundwater. For the surfaces containing hydrogen-bonding moieties, the retardation factors calculated from the adsorption isotherms were plotted versus the solid-liquid interfacial energy (Figure 8B). It appears that the retardation factors increase in an approximately linear fashion with respect to the interfacial energy. A linear fit of the data, when forced through one, yields a slope
Environmental Biogeochemistry of 0.028(3) m2mJ-1. This fit provides a straightforward method for predicting the interaction of OTC with humic acids, with the slope of the linear fit to the data presented in Figure 8B being a proportionality constant that directly relates interfacial energies to retardation factors. To this end, we obtained the contact angles and the interfacial free-energy densities for water on two humic acid samples (Aldrich: humic acid sodium salt technical grade, and International Humic Substances Society (IHSS): Elliot Soil Humic Acid Standard). The humic acid samples were prepared by pressing ∼100 mg of the humic acid between two stainless-steel metal plates of an infrared pellet press, and the resulting flattened surfaces were then used for contact angle measurements. The contact angles for the Aldrich and IHSS humic acids were found to be 57° and 59°, respectively. The corresponding solid-liquid interfacial energies were 13 and 14 mJ/m2, respectively. Figure 8B shows that these interfacial energies correspond to retardation factors of 1.3 for the Aldrich humic acid and 1.4 for the IHSS humic acid. The contact angle method presented here allows us to predict binding constants for those geosorbents that are difficult to study with SHG because of their light-absorbing optical properties, such as humic acids. Thus, this major challenge in geochemistry, and complex systems in general, can be quantitatively addressed. The retardation values predicted in this laboratory study have important implications on the environmental transport of OTC. We have shown that in the presence of amide-linked aliphatic and aromatic carboxylic acids and for (redox inactive) physisorption interactions geochemical interfaces that display high interfacial free energies will slow the transport of pollutants through soils to a greater extent than those interfaces with low interfacial free energies. This result is consistent with equilibrium batch experiments that have demonstrated the dependence of antibiotic mobility (i.e., Kd and Rf values) on soil composition.61 The relatively low antibiotic mobility in heterogeneous geomedia with high interfacial energies implies that soil microbes, which are typically associated with mineral surfaces,62 will have extended contact with antibiotics provided that the soil microbe and the antibiotic-enriched organic adlayers are close in proximity. In this scenario, the resulting risk for antibiotic resistance development in the microbe is high but localized. In contrast, OTC is expected to be mobile at geochemical interfaces with low interfacial free energy and the exposure of bacteria to the antibiotic is expected to be widespread. In this scenario, the risk of antibiotic resistance development is no longer localized. If the exposure of soils to agricultural antibiotics continues at its current rate, then the risk to human health due to antibacterial resistance development will be elevated in regions characterized by high OTC mobility in soils.
J. Phys. Chem. C, Vol. 111, No. 25, 2007 8803 display a free energy of adsorption on the order of -40 kJ/mol for the interfaces studied. Our results are consistent with our previously reported work on OTC binding to fused quartz and alkyl-, methyl ester-, and carboxylic acid-functionalized fused quartz/water interfaces.20 However, there are notable differences. It appears that polar functional groups, in this case an amide, can play a significant role in OTC binding even when that group is not the terminal surface moiety. Another notable result is the remarkably higher binding constant for an aromatic acid versus an aliphatic acid. It therefore appears that the mobility of OTC in silica-rich soil environments containing natural organic matter abundant in polar aromatic functional groups can be significantly reduced. By incorporating the data from our isotherm measurements into the Kd model, which is commonly used for predicting pollutant mobility in the environment, we developed important predictive capabilities for assessing how OTC is transported in various soil environments. The method of determining retardation factors via contact angle measurements proposed in this work represents a feasible technique for rapidly evaluating the relative mobility of OTC for soils rich in humic substances. This information is vital for understanding the environmental risk posed by the use of antibiotics in agriculture. Clearly, understanding the molecular origin of OTC mobility in soils is critical for evaluating how the use of agricultural pharmaceuticals impacts antibiotic resistance in bacteria and human health. Future work will focus on studying the structure-mobility relationships of aromatic and nonaromatic pollutants with surfaces functionalized with phenols, catechols, and other types of substituted benzene rings commonly found in humic substances. Acknowledgment. Ken Eisenthal inspires all of us and we dedicate this work to him. J.M.G.-D. gratefully acknowledges a fellowship from the Camille and Henry Dreyfus Postdoctoral Program in Environmental Chemistry. A.L.M. gratefully acknowledges an EPA Science To Achieve Results (STAR) fellowship. M.J.M. gratefully acknowledges a Wender fund fellowship. This work is supported by the National Science Foundation through the CAREER program in Experimental Physical Chemistry. We also gratefully acknowledge support from the Chemical Sciences, Geosciences and Biosciences Division, Office of Basic Energy Sciences, Office of Science, U.S. Department of Energy, the DOE-funded Northwestern University Institute for Environmental Catalysis, the American Chemical Society Petroleum Research Fund, and a Dow Chemical Company professorship to F.M.G. F.M.G. acknowledges an Alfred P. Sloan Foundation fellowship. References and Notes
V. Conclusions Second-harmonic generation was used to study the binding behavior of the antibiotic oxytetracycline to tailor-made solid/ liquid interfaces. To create chemical models for heterogeneous geochemical systems, fused quartz surfaces were functionalized using amide-coupling chemistry so that the bulk aqueous media was presented with environmentally relevant organic moieties. The flexibility of the synthetic method used in this study allows for the facile functionalization of fused quartz surfaces with different chemical functional groups. Using SHG, OTC binding to fused quartz/water interfaces functionalized with methylamide-, amide-linked carboxylic acid-, and benzoic acid-terminated silanes is found to be fully reversible, to follow a Langmuir adsorption model, and to
(1) King, D. A.; Peckham, C.; Waage, J. K.; Brownlie, J.; Woolhouse, M. E. Science 2006, 313, 1392. (2) Schwarzenbach, R. P.; Escher, B. I.; Fenner, K.; Hofstetter, T. B.; Johnson, C. A.; von Gunten, U.; Wehrli, B. Science 2006, 313, 1072. (3) D’Costa, V. M.; McGrann, K. M.; Hughes, D. W.; Wright, G. D. Science 2006, 311, 374. (4) Ferber, D. Science 2000, 288, 792. (5) Ferber, D. Science 2003, 301, 1027. (6) Witte, W. Science 1998, 279, 996. (7) Hileman, B. Chem. Eng. News 2005, 83, 16. (8) Kummerer, K.; Velo, G. Drug Saf. 2006, 29, 371. (9) Sassman, S. A.; Lee, L. S. EnViron. Sci. Technol. 2005, 39, 7452. (10) Boxall, A. B. A. EMBO Rep. 2004, 5, 1110. (11) Barnes, K. K.; Kolpin, D. W.; Meyer, M. T.; Thurman, E. M.; Furlong, E. T.; Zaugg, S. D.; Barber, L. B. “Water-Quality Data for Pharmaceuticals, Hormones, and Other Organic Wastewater Contaminants in Streams, 1999-2000,” U.S. Geological Survey Open-File Report 0294, 2002.
8804 J. Phys. Chem. C, Vol. 111, No. 25, 2007 (12) Kay, P.; Blackwell, P. A.; Boxall, A. B. A. Chemosphere 2005, 59, 951. (13) Chopra, I.; Roberts, M. Microbiol. Mol. Biol. ReV. 2001, 65, 232. (14) Simon, N. S. EnViron. Sci. Technol. 2005, 39, 3480. (15) Figueroa, R. A.; Leonard, A.; MacKay, A. A. EnViron. Sci. Technol. 2004, 38, 476. (16) Halling-Sorensen, B.; Nors Nielsen, S.; Lanzky, P. F.; Ingerslev, F.; Holten Lutzhoft, H. C.; Jorgensen, S. E. Chemosphere 1998, 36, 357. (17) Kulshrestha, P.; Giese, R. F., Jr.; Aga, D. S. EnViron. Sci. Technol. 2004, 38, 4097. (18) Tolls, J. EnViron. Sci. Technol. 2001, 35, 3397. (19) Jones, A. D.; Bruland, G. L.; Agrawal, S. G.; Vasudevan, D. EnViron. Toxicol. Chem. 2005, 24, 761. (20) Mifflin, A. L.; Konek, C. T.; Geiger, F. M. J. Phys. Chem. B 2006, 110, 22577. (21) Brown, G. E., Jr. Science 2001, 294, 67. (22) Evangelou, V. P. EnVironmental Soil and Water Chemistry; John Wiley & Sons: New York, 1998. (23) Langmuir, D. Aqueous EnVironmental Geochemistry; Prentice Hall, Inc: New Jersey, 1997. (24) Morel, F. M. M.; Hering, J. G. Principles and Applications of Aquatic Chemistry; John Wiley & Sons: New York, 1993. (25) Wershaw, R. L. Soil Sci. 1999, 164, 803. (26) Sutton, R.; Sposito, G. EnViron. Sci. Technol. 2005, 39, 9009. (27) Mifflin, A. L.; Musorrafiti, M. J.; Konek, C. T.; Geiger, F. M. J. Phys. Chem. B 2005, 109, 24386. (28) Al-Abadleh, H. A.; Mifflin, A. L.; Bertin, P. A.; Nguyen, S. T.; Geiger, F. M. J. Phys. Chem. B 2005, 109, 9691. (29) Al-Abadleh, H. A.; Mifflin, A. L.; Musorrafiti, M. J.; Geiger, F. M. J. Phys. Chem. B 2005, 109, 16852. (30) Al-Abadleh, H. A.; Voges, A. B.; Bertin, P. A.; Nguyen, S. T.; Geiger, F. M. J. Am. Chem. Soc. 2004, 126, 11126. (31) Boman, F. C.; Musorrafiti, M. J.; Gibbs, J. M.; Stepp, B. R.; Salazar, A. M.; Nguyen, S. T.; Geiger, F. M. J. Am. Chem. Soc. 2005, 127, 15368. (32) Voges, A. B.; Yin, G.; Gibbs-Davis, J. M.; Lettan, R. B., II; Bertin, P. A.; Pike, R. C.; Nguyen, S. T.; Scheidt, K. A.; Geiger, F. M. J. Phys. Chem. B, in press, 2006. (33) Tetracyclines in Biology, Chemistry and Medicine; Nelson, M, Hillen, W, Greenwald, R. A., Eds.; Birkhauser Verlag: Boston, 2001. (34) Leypold, C. F.; Reiher, M.; Brehm, G.; Schmitt, M. O.; Schneider, S.; Matousek, P.; Towrie, M. Phys. Chem. Chem. Phys. 2003, 5, 1149. (35) Eisenthal, K. B. Chem. ReV. 2006, 106, 1462. (36) Eisenthal, K. B. Chem. ReV. 1996, 96, 1343. (37) Shen, Y. R. The Principles of Nonlinear Optics; John Wiley & Sons: New York, 1984.
Hayes et al. (38) Shen, Y. R. Nature 1989, 337, 519. (39) Al-Abadleh, H. A.; Mifflin, A. L.; Musorrafiti, M. J.; Geiger, F. M. J. Phys. Chem. B 2005, 109, 16852. (40) Heinz, T. F. Nonlinear Surface Electromagnetic Phenomena, 1991 ed.; Elsevier: Amsterdam, 1991. (41) Hicks, J. M.; Kemnitz, K.; Eisenthal, K. B.; Heinz, T. F. J. Phys. Chem. 1986, 90, 560. (42) Kemnitz, K.; Bhattacharyya, K.; Hicks, J. M.; Pinto, G. R.; Eisenthal, K. B.; Heinz, T. F. Chem. Phys. Lett. 1986, 131, 285. (43) Simpson, G. J.; Westerbuhr, S. G.; Rowlen, K. L. Anal. Chem. 2000, 72, 887. (44) Boyd, R. W. Nonlinear Optics; Academic Press: New York, 2003. (45) Mifflin, A. L.; Gerth, K. A.; Geiger, F. M. J. Phys. Chem. A 2003, 107, 9620. (46) Mifflin, A. L.; Gerth, K. A.; Weiss, B. M.; Geiger, F. M. J. Phys. Chem. A 2003, 107, 6212. (47) Ege, S. Organic Chemistry, Structure and ReactiVity, 4th ed.; Houghton Mifflin: Boston, MA, 1999. (48) Yan, E. C. Y.; Liu, Y.; Eisenthal, K. B. J. Phys. Chem. B 1998, 102, 6331. (49) Zhao, X.; Ong, S.; Wang, H.; Eisenthal, K. B. Chem. Phys. Lett. 1993, 214, 203. (50) Song, S.; Ren, S.; Wang, J.; Yang, S.; Zhang, J. Langmuir 2006, 22, 6010. (51) Somorjai, G. A. Chemistry in Two Dimensions; Cornell University Press: Ithaca, NY, 1981. (52) Masel, R. I. Principles of Adsorption and Reaction on Solid Surfaces; Wiley: New York, 1996. (53) Adamson, A. W. Physical Chemistry of Surfaces, 5th ed.; John Wiley & Sons: New York, 1990. (54) Somorjai, G. A. Introduction to Surface Chemistry and Catalysis; John Wiley & Sons: New York, 1994. (55) Application Notes, Surface Energy Calculations. In First Ten Angstroms Technical Documentation; First Ten Angstroms: Portsmouth, VA, 2001; Vol. 2006. (56) Good, R. J. AdV. Chem. Ser. 1964, 43, 74. (57) Good, R. J. Nature 1966, 212, 276. (58) Good, R. J.; Girifalco, L. A. J. Phys. Chem. 1960, 64, 561. (59) Zisman, W. A. AdV. Chem. Ser. 1964, 43, 1. (60) Understanding Variation in Partition Coefficient, Kd, Values: Introduction; United States Environmental Protection Agency, 2003. (61) Rabolle, M.; Spliid, N. H. Chemosphere 2000, 40, 715. (62) Stumm, W.; Morgan, J. J. Aquatic Chemistry, Chemical Equilibria and Rates in Natural Waters, 3rd ed.; John Wiley & Sons: New York, 1996.
