Health Articles

Doi:10.1016/j.jcis.2004.08.150

Journal of Colloid and Interface Science 283 (2005) 160–170 Sorption of the antibiotic ofloxacin to mesoporous and nonporous alumina and silica Keith W. Goyne Jon Chorover James D. Kubicki Andrew R. Zimmerman Susan L. Brantley a Department of Soil, Water and Environmental Science, University of Arizona, 429 Shantz Building, Tucson, AZ 85721-0038, USA b Department of Geosciences, The Pennsylvania State University, University Park, PA 16802, USA c Department of Geological Science, University of Florida, Gainesville, FL 32611, USA Received 23 June 2004; accepted 17 August 2004 Available online 30 November 2004 Mesoporous and nonporous SiO2 and Al2O3 adsorbents were reacted with the fluoroquinolone carboxylic acid ofloxacin over a range of pH values (2–10) and initial concentrations (0.03–8 mM) to investigate the effects of adsorbent type and intraparticle mesopores onadsorption/desorption. Maximum ofloxacin adsorption to SiO2 surfaces occurs slightly below the pKa2 (pH 8.28) of the antibiotic andsorption diminishes rapidly at pH > pKa2. For Al2O3, maximum sorption is observed at pH values slightly higher than the adsorbent's pointof zero net charge (p.z.n.c.) and less than midway between the pKa values of ofloxacin. The effects of pH on adsorption and ATR–FTIRspectra suggest that the zwitterionic compound adsorbs to SiO2 solids through the protonated N4 in the piperazinyl group and, possibly,a cation bridge; whereas the antibiotic sorbs to Al2O3 solids through the ketone and carboxylate functional groups via a ligand exchangemechanism. Sorption edge and isotherm experiments show that ofloxacin exhibits a higher affinity for mesoporous SiO2 and nonporousAl2O3, relative to their counterparts. It is hypothesized that decreased ofloxacin sorption to mesoporous Al2O3 occurs due to electrostaticrepulsion within pore confines. In contrast, it appears that the environment within SiO2 mesopores promotes sorption by inducing formationof ofloxacin–Ca complexes, thus increasing electrostatic attraction to SiO2 surfaces.
 2004 Elsevier Inc. All rights reserved.
Keywords: Ofloxacin; Fluoroquinolone carboxylic acid; Mesoporosity; Sorption edge; Adsorption/desorption isotherms; ATR–FTIR spectroscopy; Molecular modeling; Mineral–organic interactions of FQCAs in wastewaters and streams with concentrationstypically reported in the range of ng L−1 to µg L−1 Fluoroquinolone carboxylic acids (FQCAs) are a class Due to the land application or discharge of wastes to streams of chemotherapeutic agents with antibacterial activity used and our limited knowledge of the fate and interactions of in human and veterinary medicines. Although absorptivity FQCAs in aquatic and terrestrial environments, these com- of orally administered FQCAs is high a portion of the pounds are of significant environmental concern dose passes through the body into human and animal ex- Within the large class of FQCAs, ofloxacin is used to treat crement. Thus, FQCAs have been detected in wastewaters urinary and respiratory tract infections in humans and ani- insufficiently treated by sewage treatment plants liq- mals Although a significant number of studies have in- uid animal manures and streams Recent studies in vestigated aqueous ofloxacin–metal complexation reactions the United States and Europe have documented the presence much less work has been done on the sorption ofofloxacin to minerals and soil. Djurdjevic et al. de- * Corresponding author. Fax: +1(520)-621-1647.
termined that sorption of ofloxacin to Al2O3 solids exhib- E-mail address: (J. Chorover).
ited S-shaped isotherms when experiments were conducted 0021-9797/$ – see front matter  2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcis.2004.08.150 K.W. Goyne et al. / Journal of Colloid and Interface Science 283 (2005) 160–170 from 19 to 140 µmol L−1 in neutral and very acidic (pH 1) In this work, studies were conducted to investigate the aqueous background solutions, and isotherms were L-shaped sorption of the FQCA ofloxacin to nonporous and meso- (Langmuir) in basic solutions (pH 11). The greatest extent porous Al2O3 and SiO2 mineral sorbents. The objectives of sorption occurred at neutral pH (0.7 mmol g−1) followed of this study were (a) to investigate differences in ofloxacin by sorption at pH 1 (0.5 mmol g−1) and sorption at pH 11 sorption to Al2O3 and SiO2 surfaces as a function of pH and (0.38 mmol g−1) However, ofloxacin sorption onto initial concentration, (b) to determine if mesoporosity influ- Al(OH)3 gel exhibited a C-shaped (linear) isotherm and 21% ences the amount of ofloxacin sorbed to Al2O3 and SiO2 of adsorbed ofloxacin was released from the mineral surface solids, and (c) to determine the mechanism(s) through which during desorption reactions Al dissolution and changes ofloxacin binds to Al2O3 and SiO2 surfaces.
in solution pH as a function of ofloxacin adsorption werenot measured in either study. Thus, it is unclear to what ex-tent aqueous Al may compete with Al surfaces for ofloxacin 2. Materials and methods
complexation, and the multifunctionality of this compoundmay equate to several possible bonding mechanisms.
2.1. Properties of the adsorbate Nowara et al. investigated the sorption of several FQCAs, including levofloxacin (an active optical isomer of ofloxacin), to soil, soil clay fractions, and soil minerals. This study reported that adsorption of FQCAs to soil, soil clay ne-6-carboxylic acid; 98% minimum purity) was purchased fractions, and layer silicates is very high (95–99% removal from Sigma–Aldrich Co. (St. Louis, MO) and used as re- from initial aqueous concentrations ranging from 0.28 to ceived. The compound was stored at 4 ◦C in the dark to min- 28 µmol L−1) and desorption in 0.01 M CaCl2 is very low imize photolytically induced degradation Ofloxacin is (<2.6% of adsorbed amount was released into solution). In- a zwitterionic compound with acid dissociation constants of frared spectra and microcalorimetry data were interpreted 6.08 (pKa1) and 8.25 (pKa2) As shown in by Nowara et al. to suggest that FQCAs are bound to , the antibiotic is primarily cationic below pKa1 (N4 clays via a cation bridge between charged basal surfaces and in the piperazinyl group), anionic above pKa2 (3-carboxyl the carboxylate functional group of FQCAs However,cation bridging is a relatively weak sorption mechanism thatis not associated typically with irreversible adsorption asobserved by Norwara et al. In addition, experimentalpH values were generally equal to or less than the pKa1of the FQCAs, thus cationic forms of the compounds mayhave been adsorbed to mineral surfaces. High Koc values(40,000–71,000) suggest that sorption was also influencedby the amount of organic carbon present in the soil andothers have reported sorption of FQCAs to dissolved humicacids An additional factor that should be considered when studying the fate of organic compounds in soils and sedi-ments is substrate surface morphology. Recent studies havedemonstrated that mineral mesoporosity (2–50 nm in porediameter), as occurs in naturally weathered geosorbentscan impact organic compound sorption. Zimmermanet al. observed that nitrogenous organic compoundssmaller than one-half the average mesopore diameter exhib-ited significantly greater surface area-normalized adsorptionto mesoporous alumina and silica, relative to nonporous ana-logues. Sorption of larger compounds was inhibited due tocompound exclusion from the internal mesopore surfaces.
Goyne et al. documented increased adsorption of thepesticide 2,4-D to alumina sorbents with increased meso-porosity. However, it should be noted that porosity, in andof itself, is not always the most important governing fac- tor. For instance, 2,4-D did not adsorb to mesoporous or Fig. 1. (a) Ionization of aqueous ofloxacin and (b) distribution of cationic nonporous silica, presumably because of electrostatic repul- (ofx+), zwitterionic (ofx0), and anionic (ofx−) ofloxacin in aqueous solu- tion as a function of pH (pKa1 = 6.08 and pKa2 = 8.25) K.W. Goyne et al. / Journal of Colloid and Interface Science 283 (2005) 160–170 group), and zwitterionic (net neutral) between pKa1 and pH range from pH 2 to 10 for sorption edge experiments and pKa2. Due to solubility issues (Csat = pH 7.2 for isotherm experiments. Samples were spiked with ofloxacin stock solutions were not prepared at concentra- ofloxacin dissolved in 0.02 M CaCl2 to give initial ofloxacin tions greater than 9 mM The dimensions of ofloxacin as concentrations ranging from 0.03 to 8 mM for isotherm ex- determined by measuring interatomic distances and account- periments or 1 mM for sorption edge experiments. Although ing for van der Waals radii of the atoms and constrained by these concentrations are significantly greater than those de- energy minimization using the COMPASS force field are tected in natural waters our goal was to compare 1.2 × 0.95 × 0.6 nm.
ofloxacin sorption to mesoporous and nonporous silica andalumina, not to mimic solute concentrations found in im- 2.2. Adsorbent synthesis and characteristics pacted waters. For the isotherm experiments, the initial pH ofthe ofloxacin stock solution and CaCl2 solution was adjusted Four mineral adsorbents were used in the present work: to 7.20. Samples were reacted on an end-over-end shaker (1) mesoporous Al2O3 (Al-P242), (2) nonporous Al2O3(Al- (7 rpm) in the dark at 298 K for 30 min. Adsorbent-free NP37), (3) mesoporous SiO2 (Si-P700), and (4) nonporous controls (no mineral) were reacted concurrently to measure SiO2 (Si-NP8), where the subscripts refer to specific surface compound loss resulting from sorption to centrifuge tube area in m2 g−1. Al-NP37 and Si-NP8 were purchased from walls or volatilization. Neither of these was found to be sig- Alfa Aesar (Ward Hill, MA), Stock Nos. 40007 and 89709, respectively. Al-NP37 was washed and dried as described in After reaction, mineral suspensions were centrifuged at Goyne et al. to remove an N-containing soluble con- 15,290g and 298 K for 45 min. An aliquot of supernatant stituent associated with synthesis. Adsorbents Al-P242 and solution was removed by pipet, stored in 4 ml amber vials, Si-P700 were prepared using a neutral template route and refrigerated for measurement of ofloxacin concentra- The synthesis procedure and removal of the templates tion. The remaining solution was aspirated, filtered through from the fabricated adsorbents are detailed elsewhere a 0.02-µm nominal pore size filter, acidified to pH < 2 with All minerals, except Si-NP8, were ground gently prior trace metal grade HCl, and refrigerated at 4 ◦C. Concentra- to characterization and stored in polyethylene bottles prior tions of Al and Si were determined using a Perkin–Elmer to use. The physical and chemical characteristics were pub- Elan DRC II inductively coupled plasma-mass spectrome- lished previously and a summary is provided in ter (ICP-MS). The pH of unfiltered and unacidified super-natant solution was measured using a calibrated Orion Ross 2.3. Batch sorption edge and isotherm experiments semi-micro combination pH electrode attached to a Beck-man
390 pH meter.
Mineral adsorbents were suspended in a 0.02 M CaCl2 Ofloxacin concentrations in solution were determined by (0.06 M ionic strength) background electrolyte solution to measuring the concentration of nonpurgable organic carbon give a sorbent surface area to solution ratio of 2.86 × (NPOC) and total nitrogen (TN) present in solutions acidi- 103 m2 L−1 in PTFE centrifuge tubes. Sorption experiments fied to pH < 2 with trace metal grade HCl (Shimadzu Model were conducted in the absence of pH buffers to prevent TOC-VCSH, total organic carbon analyzer, equipped with a competitive sorption between buffer constituents (e.g., phos- TNM-1, total nitrogen measuring unit, and an ASI-V au- phate) and ofloxacin for available sorption sites and to al- tosampler). Standards were prepared by dissolving ofloxacin low for measurement of pH shifts often indicative of lig- in 0.02 M CaCl2. There were no significant differences and exchange reactions All stock solutions or sam- between ofloxacin concentrations calculated using NPOC ples containing ofloxacin were wrapped in aluminum foil or TN; thus all data shown are based on NPOC measure- and/or stored in amber glassware to prevent or minimize ments for simplicity. High-performance liquid chromatog- raphy (HPLC) was not used, due to decreased column re- Solutions of 0.06 M HCl or 0.02 M Ca(OH)2 were added tention of ofloxacin when aluminum was present in solu- to mineral suspensions (prior to reaction) to achieve a final tion (i.e., peaks were broadened and decreased in height, Table 1Physical characteristics and surface charge properties of the adsorbents SBET (m2 g−1) Dpore (nm) 3.4 ± 0.4 7.5 ± 0.1 8.2 ± 0.6 Note. See Goyne et al. for detailed methods and data analysis; SBET is the specific surface area ± std. dev. as measured by N2 BET; Dpore is the meanpore diameter ± std. dev. determined by the BJH method on the adsorption isotherm leg; Sip is the intraparticle surface area (within pores 2–20 nm in diameter)as percentage of total determined by BJH method; p.z.n.c. is the point of zero net charge ± 95% CI; values of p.z.n.c. not encountered in the pH range of theexperiment are expressed as the lowest pH values of the experiment; pKa1 and pKa2 are surface acidity intrinsic constants in accordance with the constantcapacitance model, using proton charging data from K.W. Goyne et al. / Journal of Colloid and Interface Science 283 (2005) 160–170 evidently because of ofloxacin complexation with Al in where the solver function of Microsoft Excel 2002 the aqueous phase). However, excellent agreement between is used to vary iteratively the three fitting parameters Nt, standards analyzed using HPLC and NPOC/TN was ob- A, and β to maximize the coefficient of determination served in the absence of aluminum. Mineral blanks (no (R2 = 1).
ofloxacin) were reacted concurrently for correction.
Surface excess of ofloxacin was calculated as 2.4. Infrared spectroscopy Γads = Cads,B − Cads,S , Attenuated total reflectance (ATR)–Fourier transform in- frared (FTIR) spectroscopy was employed to document where Γads is the surface excess of ofloxacin (µmol m−2), changes in ofloxacin spectra as a function of pH and ionic Cads,S and Cads,B are the equilibrium ofloxacin concentra- composition and to investigate the mechanism of adsorp- tions (µmol kg−1) in supernatant solutions of mineral sus- tion. For aqueous phase spectra, stock solutions contain- pensions (S) and for the corresponding blank (B) after the ing 9.0 mM ofloxacin were prepared in 0.06 M NaCl and reaction period, and SAS is the suspension concentration of 0.02 M CaCl2 background electrolyte solutions with pH val- adsorbent (m2 kg−1).
ues ranging from pH 5 to 10. A 3-ml aliquot of solution was Desorption reactions (isotherm experiments only) were then transferred into an ATR cell equipped with a 45◦ ZnSe initiated immediately after the adsorption step by adding a flat plate crystal (ARK cell, Thermo Spectra-Tech, Inc.), and mass of 0.02 M CaCl2 (pH 7.20) equivalent to that mass of spectra were obtained by averaging 400 scans at 2 cm−1 res- supernatant solution removed. Desorption reaction time was olution on a Nicolet Magna 560 spectrometer.
equal to that for adsorption (30 min). After the desorption Infrared spectra of adsorbed ofloxacin were obtained for period, supernatant solutions were again removed by pipet samples containing 11.80 g L−1 of mesoporous material re- and analyzed. Adsorbate retained was calculated from acted at pH 5.5 and 7.2 for 30 min with initial ofloxacin con- centrations of 0 and 9.0 mM, as described previously. After Γdes = Γads − adsorption, samples were centrifuged and most of the super-natant solution was removed, except ca. 3.0 ml which was where Γdes is the surface excess of ofloxacin (µmol m−2) af- left in the suspension to create a slurry. ATR–FTIR slurry ter the desorption step, Cdes,S is the ofloxacin concentration samples were immediately transferred into the ATR cell for in supernatant solution of mineral suspension (S) after the data collection. Spectra of adsorbed ofloxacin were obtained desorption reaction (µmol kg−1), Mtot,soln is the total mass by subtracting those of the ofloxacin-free slurry.
of solution (kg) in the reaction vessel during desorption,Ment is the mass of entrained solution (kg) remaining in the 2.5. Molecular modeling of infrared frequencies centrifuged adsorbent pellet after aspiration of adsorptionstep supernatant, and SA is the total surface area of adsor- Gas-phase, infrared frequencies of cationic, anionic, bent (m2) in the reaction vessel.
zwitterionic ofloxacin, and an Al–ofloxacin complex were Sorption data were fit to the Langmuir–Freundlich equa- calculated at the B3LYP/6-31G(d) level using the tion which has been shown to successfully model a Gaussian 98 program Frequency values were corrected number of other organic compounds on heterogeneous sur- by multiplying calculated values by 0.96 Model struc- tures of the cationic, zwitterionic, and anionic species of ofloxacin were modeled with and without explicit hydra- tion of the polar functional groups. In addition, the species (OH2)4Al–ofloxacin (in the zwitterionic state) was modeled.
The Al3+ was bonded in a bidentate fashion to one O atom i is adsorbate surface excess (µmol m−2), Nt is the total number of binding sites, A is a parameter related to of the carboxylate group and to the O atom of the adjacent the binding affinity (K ketone group. The output files were then used to view an- 0; K0 = A1/β ), ci is the equilibrium aqueous concentration of ofloxacin (µmol L−1), and β is a imated vibrational motions in Molden Version 3.9 for fitting parameter When β = 1, the Langmuir– band assignment.
Freundlich equation reduces to the Langmuir equation qi = NtAci (β = 1), 3. Results and discussion
whereas, if ci or A approach zero, it equation reduces to the 3.1. Ofloxacin adsorption as a function of pH Freundlich equation The effects of pH on ofloxacin adsorption to SiO2 sur- qi = Ac (ci or A → 0). faces are shown in Above pH 5.0, Si-P700 adsorbs The Langmuir–Freundlich equation was fit to the experi- significantly more ofloxacin than does Si-NP8. Maximum mental data using the method outlined in Umpleby et al.
ofloxacin sorption to these minerals (80.3 and 67.2% for K.W. Goyne et al. / Journal of Colloid and Interface Science 283 (2005) 160–170 Fig. 3. Ofloxacin sorbed on (a) Si-NP Fig. 2. Ofloxacin adsorbed (Γ 8 and Si-P700 or (b) Al-NP37 and ads) on (a) Si-NP8 and Si-P700 or (b) Al-NP37 242 after 30 min of adsorption (Γads ) or desorption (Γdes) reaction 242 as a function of pH after 30 min of reaction time (duplicate time (duplicate means are shown and error bars, where visible, represent means are shown and error bars, where visible, represent 95% CI). Sur- 95% CI). Surface excess is expressed as micromoles per square meter and face excess is expressed as micromoles per square meter and molecules per molecules per square nanometer.
square nanometer.
In contrast, indicates that ofloxacin sorption to Si-P700 and Si-NP8, respectively) occurs slightly below the mesoporous Al2O3 (Al-P242) was consistently lower than pKa2 (pH 8.28) of the antibiotic and diminishes rapidly at sorption to nonporous Al2O3 (Al-NP37) over the full pH pH > pKa2. Thus, we presume that cationic and zwitterionic range investigated. Sorption of the antibiotic to Al2O3 solids ofloxacin are adsorbed to the negatively charged silica sur- increases significantly above pH 5.0, concurrent with in- faces (i.e., ≡SiO− functional groups) via the protonated N4 creased aqueous concentrations of zwitterionic ofloxacin. In in the piperazinyl group. At the pH of maximum measured this case, Γ max (88.2 and 73.3% for Al-NP37 and Al-P242, sorption, only 45 and 9% of the dissociated silanol groups respectively) occurs at pH slightly less than midway be- on the surface of Si-P700 and Si-NP8 respectively, are tween the pKa values of ofloxacin (pH 7.16) and slightly occupied by the compound, suggesting that sorption was not higher than the adsorbent point of zero net charge (p.z.n.c.; limited by the availability of ≡SiO− sites.
At pH greater than that of Γmax, adsorption dimin- The fact that Si-NP8 adsorbs more ofloxacin below pH 5 ishes as alumina surfaces become increasingly negatively than does Si-P700 can be attributed to 50% higher density charged, thus repelling zwitterionic and anionic ofloxacin of dissociated surface silanol groups on Si-NP8 at pH 3–5 from the surface. These data lead us to hypothesize that according to surface charge data reported previously ofloxacin sorption to ≡AlOH+ surface sites via the dissoci- However, at this pH, Si-P700 has a greater fraction of disso- ated 3-carboxyl group (COO−) of the zwitterion is initiated ciated silanol groups occupied by ofloxacin. For instance, between pH 4.5 and 5.5. Initiation likely occurs closer to pH ofloxacin sorbed onto Si-P700 at pH 3.32 occupies 62% 5.5 based on . However, the same mechanism of sorp- of the ≡SiO− groups, whereas at pH 3.40 ofloxacin occu- tion may not be applicable below pH 5.0.
pies only 34% of the ≡SiO− groups on Si-NP8. Overall, it Ofloxacin adsorption to Al2O3 decreases from pH 2 to appears that the presence of intraparticle mesoporosity in- 4.5, and the same reproducible trend appears in the data creases ofloxacin sorption to SiO2 surfaces.
sets for both alumina solids We are unable to ex- K.W. Goyne et al. / Journal of Colloid and Interface Science 283 (2005) 160–170 Table 2Langmuir–Freundlich parameters for SiO2 and Al2O3 isotherms Nt (µmol m−2) A (µM−1) K0 (µM−1) R2 (n) 4.6 × 10−5 1.3 × 10−3 5.7 × 10−5 1.8 × 10−3 1.2 × 10−3 8.2 × 10−4 4.8 × 10−4 1.1 × 10−3 1.3 × 10−3 2.3 × 10−3 7.0 × 10−3 4.5 × 10−4 9.5 × 10−4 1.1 × 10−3 5.6 × 10−4 4.1 × 10−3 Note. Nt is adsorption capacity; A is a parameter related to mean binding affinity (K0); β is an exponent related to the heterogeneity of binding site energydistribution; K0 is the mean binding affinity; R2 is the coefficient of determination, and n is the number of sample points utilized to calculate the parameters.
plain the cause of this occurrence, other than to suggest that measurement of proton or hydroxide production, but varied the mechanism of ofloxacin adsorption is likely different among the four adsorbents.
than above pH 5.0. Below pH 5, the fraction of zwitterionic Equilibrium pH values for SiO2 samples were relatively ofloxacin is very low and surface complexation via constant (pH 7.20–7.40) except for those reacted at the high- COO− is likely insignificant. It is possible that dissolution est initial concentration of ofloxacin (pH 6.77 and 6.65 for of Al at acidic pH promotes the formation of Al-bridged Si-NP8 and Si-P700, respectively). This proton production dimers whose adsorption could be enhanced relative to likely decreased the quantity of ofloxacin sorbed to these monomeric ofloxacin. Irrespective of the mechanism(s) of samples, relative to that adsorbed at pH 7.20 (see adsorption, it is clear that intraparticle porosity does not in- and may slightly skew the isotherm shapes shown in crease sorption of ofloxacin to alumina surfaces. This is de- However, this shift to more acidic pH is indicative of proton spite the fact that ofloxacin is smaller (1.2 × 0.95 × 0.6 nm) displacement from ≡SiOH surface functional groups. This than the nominal pore size of Al-P242 (8.2 nm) and that these is verified by comparing the moles of ofloxacin adsorbed alumina minerals have nearly identical surface charge prop- to the predicted density of ≡SiO− present at equilibrium pH for samples with the highest initial ofloxacin concentra-tion. At pH 6.78, 0.79 µmol m−2 of ofloxacin is adsorbed 3.2. Ofloxacin adsorption/desorption isotherms to Si-NP8 and the predicted density of ≡SiO− at this pHis 0.74 µmol m−2. However, 1.15 µmol m−2 of ofloxacin Adsorption/desorption experiments were conducted to in- is adsorbed to Si-P700 with a predicted dissociated silanol vestigate ofloxacin sorption and retention as a function of site density of 0.34 µmol m−2 at pH 6.65. Thus cationic initial sorptive concentration at a target equilibrium pH 7.20 ofloxacin is displacing adsorbed protons upon adsorption: The isotherm data agree with findings from sorp- ≡SiOH + ofx+
≡SiO–ofx + H+.
tion edge experiments in that mesoporous SiO2 consis-tently sorbs more ofloxacin than nonporous SiO Calculations show that for Si-NP 8 the stoichiometry of this whereas, the opposite is true for Al cation exchange reaction is 1:1 as shown in Eq. For Si- The Langmuir–Freundlich isotherm results in in- P700, the measured release of H+ is somewhat lower than the dicate that N moles of ofloxacin adsorbed in excess of negatively charged t (sorption maximum) and A (measure of bind- ing affinity) for the two minerals adsorbing greater amounts sites, suggesting the possibility of additional sorption mech- of ofloxacin are very similar and higher than the lower affin- anisms (e.g., cation bridging).
ity sorbents. In addition, values of N In contrast, pH values for Al t indicate that whereas 2O3 reacted samples show the sorption maximum was reached by Si-NP an increase in pH with increased ofloxacin adsorption, re- to the other minerals is below the predicted maximum. No gardless of whether acid or base was added to reach the other discernible trends are apparent within the Langmuir– target equilibrium pH of 7.20. The range of Al2O3 sam- Freundlich isotherm parameters.
ple pH values (pH 6.45–7.58) are located very near maxi- As mentioned under Section isotherm experiments mum adsorption (see and the shape of the Al2O3 were conducted in the absence of pH buffers. Thus, ofloxacin isotherms should be very similar to that of isotherms where stock solution and 0.02 M CaCl pH = 7.20 for all samples. These results are suggestive of 2 background electrolyte so- lution were adjusted with base (0.02 M Ca(OH) ligand exchange reactions between ≡AlOH+ and the COO− 7.20, and additional acid (0.06 M HCl) or base was added at the beginning of each experiment to offset any pH changes ≡AlOH + ofx0
≡Al–ofx+ + OH−.
resulting from buffering of the minerals themselves The amount of acid/base added to reaction vessels was This could explain the slight hysteresis between the Al2O3 constant for the full isotherm of each adsorbent to permit adsorption/desorption isotherms and adsorption K.W. Goyne et al. / Journal of Colloid and Interface Science 283 (2005) 160–170 via the COO− functional group agrees with the sorp- background electrolyte solutions at a particular pH from tion mechanism inferred from sorption edge experiments their counterparts containing dissolved ofloxacin.) Ofloxacin ). Dissolution data for Al adsorbents are very simi- dissolved in the different background electrolyte solutions lar and show elevated Al concentrations in solution (582 and were compared because it is known that ofloxacin forms 593 µM for Al-P242 and Al-NP37, respectively) for samples strong bonds with divalent cations possi- reacted with the highest initial concentration of ofloxacin, bly through interaction between the carboxylic and ketone relative to mineral controls (<6.4 µM) and solubility data published previously for these materials (<1.4 µM) In we observe that the C=O stretch of COOH This is also consistent with adsorption via a ligand exchange (1710 cm−1) is lost as pH increases. Subse- reaction, which would tend to promote Al dissolution quently, intensity of the asymmetric (1585 cm−1) andsymmetric (1340 cm−1) stretch of COO− increases 3.3. Infrared spectra of dissolved ofloxacin and with increasing pH. There is also an increase in the inten- sity of the wavenumber that we assign as vibrations as-sociated with protonation of N4 in the piperazinyl group Although several studies have used infrared spectroscopy (1400 cm−1) Assignments of the remaining vibrations to investigate the interaction of ofloxacin with polyva- observed in are as follows: C=O stretch of ketone lent cations these studies have not documented group (1620 cm−1) C=C aromatic stretching changes in aqueous phase spectra as a function of pH (i.e., (1530 cm−1) and C–O–C stretching of the ether group functional group protonation) and cation composition (i.e., aqueous phase complexation). Thus, as noted by Macias Molecular orbital models of ofloxacin calculations on et al. assigning functional groups appropriately to vi- the explicitly solvated models generally produce similar fre- brations observed in powder IR spectra of ofloxacin–metal quencies (±35 cm−1; for the main peaks of inter- complexes can be challenging. In order to help interpret est. The C=O stretch of the carboxylic acid for explicitly ATR–FTIR spectra of ofloxacin adsorbed to the mineral sur- solvated cationic ofloxacin is an exception because it is cal- faces, we first collected ATR–FTIR difference spectra of culated to occur at 80 cm−1 lower than the experimental ofloxacin (9 mM) dissolved in 0.06 M NaCl and 0.02 M observation. The main reason for this discrepancy is the ap- CaCl2 from pH 5 to 10 as shown in (Dif- proximate representation of the electron correlation in the ference spectra were obtained by subtracting spectra of the Fig. 4. Attenuated total reflectance–Fourier transform infrared (ATR–FTIR) Fig. 5. Attenuated total reflectance–Fourier transform infrared (ATR–FTIR) difference spectra of ofloxacin in 0.06 M NaCl from pH 5 to 10. Difference difference spectra of ofloxacin in 0.02 M CaCl2 from pH 5 to 10. Differ- spectra were obtained by subtracting spectrum of 0.06 M NaCl at a partic- ence spectra were obtained by subtracting spectrum of 0.02 M CaCl2 at a ular pH from spectrum of ofloxacin dissolved in 0.06 M NaCl at the same particular pH from spectrum of ofloxacin dissolved in 0.02 M CaCl2 at the K.W. Goyne et al. / Journal of Colloid and Interface Science 283 (2005) 160–170 Fig. 7. Attenuated total reflectance–Fourier transform infrared (ATR–FTIR) Fig. 6. Attenuated total reflectance–Fourier transform infrared (ATR–FTIR) difference spectra of (a) ofloxacin in 0.02 M CaCl2 at pH 7.2 and difference spectra of (a) ofloxacin in 0.02 M CaCl2 at pH 5.5 and ofloxacin-adsorbent slurry at pH 7.2 for (b) Al-P242 and (c) Si-P700. Dif- ofloxacin-adsorbent slurry at pH 5.5 for (b) Al-P242 and (c) Si-P700. Dif- ference spectra were obtained by subtracting spectrum of 0.02 M CaCl2 ference spectra were obtained by subtracting spectrum of 0.02 M CaCl2 from ofloxacin dissolved in 0.02 M CaCl2 (a) and by subtracting 0.02 M from ofloxacin dissolved in 0.02 M CaCl2 (a) and by subtracting 0.02 M CaCl2-adsorbent slurry from ofloxacin–0.02 M CaCl2-adsorbent slurry CaCl2-adsorbent slurry from ofloxacin–0.02 M CaCl2-adsorbent slurry molecule. Although the LYP gradient-corrected correlation Difference spectra of ofloxacin and ofloxacin–mineral functional is adequate for most of the bonds in the mole- complexes formed at pH 5.5 and 7.2 are shown in cule, C=O bonds depend more strongly on electron corre- respectively. (Difference spectra were obtained by lation so this calculated mode ends up with a larger subtracting a spectrum of mineral suspended in 0.02 M discrepancy with experiment. A more accurate method for electron correlation could possibly decrease this error, but 2 at a specific pH from corresponding samples re- acted with ofloxacin.) The spectra indicate that sorption of methods such as (MP2, Møller Plesset second-order pertur- ofloxacin occurs via similar mechanisms for a given min- bation) require far more computational time and are not eral type, irrespective of pH. For example, ofloxacin sorbed practical for a model of this size. However, without explicitsolvation, COO− stretching was predicted to occur in the to Al2O3 at pH 5.5 and 7.2 shows a dra- 1725–1710 cm−1 region, an error of 100 cm−1 or more. In matic decrease in the intensity of the ketone (1620 cm−1) addition, linear regression of calculated versus predominant and asymmetric COO− (1590 cm−1) vibrations. However, experimental frequencies (n  16) for the three explicitly the same spectra show a large increase in the intensity of solvated molecules yielded slopes of 1 (±0.03) and R2 val- peaks at 1530 and 1275 cm−1.
ues of 0.99. Based on this good correlation, we conclude that We attribute the change at 1530 cm−1 to a downward the model calculations produce realistic vibrational frequen- shift in frequency of the ketone and/or asymmetric COO− cies for ofloxacin and can be used to help interpret spectra stretch upon innersphere complexation with an Al center on of unknown structures such as surface complexes.
the mineral surface. Molecular modeling of an ofloxacin–Al Ofloxacin spectra collected in 0.02 M CaCl complex, whereby Al is bound via bidentate complexation show trends similar to those in as pH increases, but vi- to the ketone and carboxylate functional groups brations associated with the ketone and asymmetric stretch indicates that the downward shift is attributable to of COO− are more intense and broadened in the presence of the ketone vibration (predicted at 1525 cm−1; ). Al- Ca2+. We propose, as have others that this is indica- though others have suggested that the ketone group vibration tive of a weak Ca–ofloxacin complex that forms between the may decrease in frequency upon strong complexation with ketone and carboxylate functional groups. Apparently, simi- a metal ion our experimental data and calculations lar complexes are not formed in the presence of monovalent demonstrate this occurrence, as do calculations performed ions with a large hydrated radius (e.g., Na+; by Sagdinc and Bayarı Perhaps more importantly, the

K.W. Goyne et al. / Journal of Colloid and Interface Science 283 (2005) 160–170 Table 3Selected experimental and calculated vibrational frequencies for ofloxacin Experimental (cm−1) Calculated (cm−1) C=O stretch of carboxylic acid 1705 C=O stretch of ketone group COO− asymmetric stretching COO− symmetric stretching C–O–C stretching of ether group 1058 Fig. 8. Molecular structure of (OH2)4Al–ofloxacin (zwitterionic) complex.
C=O stretch of ketone group This complex has the same basic orientation as the ofloxacin molecules COO− asymmetric stretching COO− symmetric stretching C–O–C stretching of ether group 1054 ketone frequency (1620 cm−1). This suggests that ofloxacin may be bonded to SiO2 surfaces via a cation bridge atpH 5.5, although the same occurrence is not observed at C=O stretch of ketone group COO− asymmetric stretching pH 7.2 This supports our contention that ofloxacin sorbs to SiO2 through the protonated piperazinyl group, via COO− symmetric stretching weaker electrostatic interaction, and through a cation bridge.
C–O–C stretching of ether group 1053 Note. Experimental assignments are reported for ofloxacin in 0.06 M NaCl; 3.4. Mechanisms for enhanced or reduced ofloxacin N.C.V. is no calculated vibration. However, the calculated results where sorption to mesoporous adsorbents N.V.C. is reported are reasonable. This is due to the presence of COOH,only, in the modeled cationic molecule; whereas, the experimental data con- Enhanced ofloxacin sorption to mesoporous SiO tain both cationic (dominant) and zwitterionic (minor) species in solution.
pothesized to occur due to higher concentration of Ca2+([Ca2+]) within the confines of a pore. Our calculations IR data strongly support our hypothesis that ofloxacin sorbs indicate that electric double layer (EDL) thickness extend- to Al2O3 surfaces via a ligand exchange process ing from a planar solid–water interface in 0.02 M CaCl2 is In contrast, IR spectra of ofloxacin sorbed to SiO2 1.2 nm Given that the mean pore diameter of Si-P700 are similar to those of dissolved ofloxacin, is 3.4 nm, EDLs within pore confines would not quite over- despite the fact that Si-P700 adsorbed 2–3 times more lap near the center of this circular pore. However, the pres- ofloxacin per unit mass than did Al-P242 ence of EDLs extending from pore walls into the pore cen- However, the spectra of the ofloxacin–SiO2 complex col- ter should increase [Ca2+] and decrease [Cl−] in the pore, lected at pH 5.5 shows an increase and broadening of the relative to concentrations of these ions near surfaces exter- Table 4Correlation of calculated vibrations and assignments of an Al–ofloxacin complex to experimental vibrations Calculated (cm−1) Experimental (cm−1) C=O stretch of carboxylate–Al complex 1593 (m), 1572 (m) C=O stretch of ketone group 1488 (s), 1468 (m) 1415 (m), 1413 (s) 1378 (m), 1372 (m) 1325 (m), 1322 (m) 1106 (w), 1075 (m) C–O–C stretch of ether group Note. Vibration assignments are based on model results and experimental vibrations are reported for ofloxacin–Al-P242 slurry spectra collected at pH 7.2 K.W. Goyne et al. / Journal of Colloid and Interface Science 283 (2005) 160–170 nal to pores The higher [Ca2+] within pores would sion from specific, but as yet undetermined, aluminol sites promote ofloxacin–Ca2+ complexation via the carboxylate within pore cavities. Overall, the data indicate that intraparti- group and effectively increase the amount of pos- cle mesopores can enhance ofloxacin sorption to adsorbent itive charge on an ofloxacin ion. Therefore, a zwitterionic minerals, but the degree of enhancement may be diminished ofloxacin molecule could bind to the surface via the proto- or even reversed by other mitigating factors (e.g., surface nated N4 of the piperazinyl group and, at the same time, bind charge properties).
via a cation bridge as others have observed This ap-pears to be a reasonable explanation for increased ofloxacinsorption to mesoporous Si-P700 and it agrees with our other data that suggest ofloxacin binds to Si-P700 surfaces by morethan one mechanism.
The authors thank Mary Kay Amistadi for laboratory as- It is unclear why adsorption to Al-NP sistance, Sridhar Komarneni, Bharat Newalkar, and Stephen 37 exceeds that of Stout for mineral synthesis and preparation, and Chad Trout 242. Ofloxacin sorption to Al-P242 is not reduced due to exclusion of the antibiotic from the pores, differences in for assistance with molecular modeling. Financial support surface charge properties, or increased dissolution of Al- was provided by the Penn State Biogeochemical Research Initiative for Education (BRIE) sponsored by NSF (IGERT) 242 resulting in aqueous and solid phase Al competition for ofloxacin. In addition, ATR–FTIR and molecular orbital cal- Grant DGE-9972759 and by the Penn State Materials Re- culations indicate that ofloxacin forms a strong innersphere search Science and Engineering Center (MRSEC) spon- complex with surface aluminum molecules. Thus, we con- sored by NSF Grant DMR-0080019. Andrew Zimmerman clude that the coexistence of positive- and negative-charged acknowledges donors to the American Chemical Society Petroleum Research Fund for partial support of this re- 2O3 may inhibit sorption of ofloxacin within the pore confines, relative to external surfaces. If ofloxacin search, and James Kubicki acknowledges the support of sorbs to Al centers via the ketone and carboxylate groups, Stony Brook-BNL collaboration to establish a Center for then over a wide range of pH the protonated N Environmental Molecular Sciences (CEMS). Computation azinyl group will be positioned toward the pore center. This was supported, in part, by the Materials Simulation Center, may, depending on the local surface charge properties, re- a Penn State MRSEC and MRI facility.
sult in an energy barrier that prohibits or hinders ofloxacinsorption to particular aluminol sites. In other words, repul-sive forces between the protonated N 4 of the piperanzinyl groups and ≡AlOH+ surface functional groups may inhibit [1] A.K. Sharma, R. Khosla, A.K. Keland, V.L. Mehta, Indian J. Pharm.
ofloxacin sorption within pore confines.
26 (1994) 249.
[2] E.M. Golet, A.C. Alder, W. Giger, Environ. Sci. Technol. 36 (2002) [3] C.G. Daughton, T.A. Ternes, Environ. Health Perspect. 107 (1999) [4] J. Tolls, Environ. Sci. Technol. 35 (2001) 3397.
Intraparticle mesoporosity in SiO2 solids was found to re- [5] D.W. Kolpin, E.T. Furlong, M.T. Meyer, E.M. Thurman, S.D. Zaugg, sult in increased uptake of ofloxacin when adsorption was L.B. Barber, H.T. Buxton, Environ. Sci. Technol. 36 (2002) 1202.
normalized to sorbent surface area. Relative to the non- [6] A. Hartmann, E.M. Golet, S. Gartiser, A.C. Alder, T. Koller, R.M.
porous solid, the presence of intraparticle porosity resulted Widmer, Arch. Environ. Contam. Toxicol. 36 (1999) 115.
[7] E.R. Campagnolo, K.R. Johnson, A. Karpati, C.S. Rubin, D.W. Ko- in a statistically significant sorption enhancement through- plin, M.T. Meyer, J.E. Eseeban, R.W. Currier, K. Smith, K.M. Thu, out the isotherms and over most of the sorption edge for the M. McGeehin, Sci. Total Environ. 299 (2002) 89.
porous silica adsorbent (Si-P700). Observations of proton re- [8] S.E. Jørgensen, B. Halling-Sørensen, Chemosphere 40 (2000) 691.
lease in association with ofloxacin sorption and sorption in [9] P.T. Djurdjevic, M. Jelikic-Stankov, J. Pharm. Biomed. Anal. 19 excess of the surface site density of ≡SiO− groups (mea- [10] M. Sakai, A. Hara, S. Anjo, M. Nakamura, J. Pharm. Biomed. Anal. 18 sured in absence of ofloxacin) indicate that the compound (1999) 1057.
is capable of displacing protons complexed with silanol [11] D.L. Ross, C.M. Riley, Int. J. Pharm. 93 (1993) 121.
groups as it sorbs to SiO2 surfaces via the protonated N4 [12] H.-R. Park, K.-Y. Chung, H.-C. Lee, J.-K. Lee, K.-M. Bark, Bull. Ko- of the piperazinyl group. Conversely, ofloxacin adsorption rean Chem. Soc. 21 (2000) 849.
[13] P.T. Djurdjevic, M. Jelikic-Stankov, I. Lazarevic, Bull. Chem. Soc.
2O3 was significantly higher than that to Jpn. 74 (2001) 1261.
mesoporous Al2O3 in both sorption edge and isotherm ex- [14] B. Macías, M.V. Villa, I. Rubio, A. Castiñeiras, J. Borrás, J. Inorg.
periments. The observed hydroxyl release concurrent with Biochem. 84 (2001) 163.
ofloxacin adsorption and shifts in the frequency of ketone [15] M. Tanaka, T. Kurata, C. Fujisawa, Y. Ohshima, H. Aoki, O. Okazaki, and increased intensity of carboxylate stretching vibrations H. Hakusui, Antimicrob. Agents Chemother. 37 (1993) 2173.
[16] A. Nowara, J. Burhenne, M. Spiteller, J. Agric. Food Chem. 45 (1997) are indicative of ligand exchange between ofloxacin and ≡AlOH or ≡AlOH+ surface sites. Decreased adsorption on [17] H.-C.H. Lützhøft, W.H.J. Vaes, A.P. Freidig, B. Halling-Sørensen, porous Al2O3 is postulated to result from electrostatic repul- J.L.M. Hermens, Environ. Sci. Technol. 34 (2000) 4989.
K.W. Goyne et al. / Journal of Colloid and Interface Science 283 (2005) 160–170 [18] Ph. Schmitt-Kopplin, J. Burhenne, D. Freitag, M. Spiteller, A. Kettrup, J.V. Ortiz, B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Ko- J. Chromatogr. A 837 (1999) 253.
maromi, R. Gomperts, R.L. Martin, D.J. Fox, T. Keith, M.A. Al- [19] L.M. Mayer, Chem. Geol. 114 (1994) 347.
Laham, C.Y. Peng, A. Nanayakkara, C. Gonzalez, M. Challacombe, [20] A.R. Zimmerman, K.W. Goyne, J. Chorover, S. Komarneni, S.L. Bran- P.M.W. Gill, B. Johnson, W. Chen, M.W. Wong, J.L. Andres, C. Gon- tely, Org. Geochem. 35 (2004) 355.
zalez, M. Head-Gordon, E.S. Replogle, J.A. Pople, Gaussian 98 (Re- [21] K.W. Goyne, J. Chorover, A.R. Zimmerman, S. Komarneni, S.L.
vision A.10), Gaussian, Pittsburgh, PA, 2001.
Brantley, J. Colloid Interface Sci. 272 (2004) 10.
[36] M.W. Wong, Chem. Phys. Lett. 256 (1996) 391.
[22] E. Fasani, A. Profumo, A. Albini, Photochem. Photobiol. 68 (1998) [37] G. Schaftenaar, J.H. Noordik, J. Comput. Aided Mol. Design 14 [23] D.L. Ross, C.M. Riley, Int. J. Pharm. 63 (1990) 237.
[38] G. Sposito, The Chemistry of Soils, Oxford Univ. Press, New York, [24] K.W. Goyne, A.R. Zimmerman, B.L. Newalkar, S. Komarneni, S.L.
Brantley, J. Chorover, J. Porous Mater. 9 (2002) 243.
[39] B. Macías, M.V. Villa, M. Sastre, A. Castiñeiras, J. Borras, J. Pharm.
[25] S. Komarneni, R. Pidugu, V.C. Menon, J. Porous Mater. 3 (1996) 99.
Sci. 91 (2002) 2416.
[26] P.T. Tanev, T.J. Pinnavaia, Science 267 (1995) 865.
[40] B.M. Sánchez, M.M. Cabarga, A.S. Navarro, A.D.G. Hurlé, Int. J.
[27] W. Zhang, T.R. Pauly, T.J. Pinnavaia, Chem. Mater. 9 (1997) 2491.
Pharm. 106 (1994) 229.
[28] W. Stumm, Chemistry of the Solid–Water Interface, Wiley, New York, [41] S. Sagdinc, S. Bayari, J. Mol. Struct. 691 (2004) 107.
[42] R.M. Silverstein, G.C. Bassler, T.C. Morrill, Spectrometric Identifica- [29] R.J. Umpleby, S.C. Baxter, Y. Chen, R.N. Shah, K.D. Shimizu, Anal.
tion of Organic Compounds, fifth ed., Wiley, New York, 1991.
Chem. 73 (2001) 4584.
[43] A.L. Mattioda, D.M. Hudgins, C.W. Bauschlicher Jr., M. Rosi, L.J.
[30] J.-Y. Yoon, H.-Y. Park, J.-H. Kim, W.-S. Kim, J. Colloid Interface Allamandola, J. Phys. Chem. A 107 (2003) 1486.
Sci. 177 (1996) 613.
[44] J. Gauss, J. Chem. Phys. 99 (1993) 3629.
[31] J.-Y. Yoon, J.-H. Kim, W.-S. Kim, Colloids Surf. A 153 (1999) 413.
[45] C. Møller, M.S. Plesset, Phys. Rev. 46 (1934) 618.
[32] A.D. Becke, J. Chem. Phys. 98 (1993) 5648.
[46] B. Gu, J. Schmitt, Z. Chen, L. Liang, J.F. McCarthy, Geochim. Cos- [33] C.T. Lee, W.T. Yang, R.G. Parr, Phys. Rev. B 37 (1988) 785.
mochim. Acta 59 (1995) 219.
[34] W.J. Hehre, R. Ditchfield, J.A. Pople, J. Chem. Phys. 56 (1972) 2257.
[47] J. Chorover, M.K. Amistadi, Geochim. Cosmochim. Acta 65 (2001) [35] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, V.G. Zakrzewski, J.A. Montgomery Jr., R.E. Strat- [48] K. Vermöhlen, H. Lewandowski, H.-D. Narres, E. Koglin, Colloids mann, J.C. Burant, S. Dapprich, J.M. Millam, A.D. Daniels, K.N.
Surf. A 170 (2000) 181.
Kudin, M.C. Strain, O. Farkas, J. Tomasi, V. Barone, M. Cossi, [49] D.L. Sparks, Environmental Soil Chemistry, Academic Press, San R. Cammi, B. Mennucci, C. Pomelli, C. Adamo, S. Clifford, J. Ochter- Diego, 1995.
ski, G.A. Petersson, P.Y. Ayala, Q. Cui, K. Morokuma, D.K. Mal- [50] Y. Wang, C. Bryan, H. Xu, P. Pohl, Y. Yang, C.J. Brinker, J. Colloid ick, A.D. Rabuck, K. Raghavachari, J.B. Foresman, J. Cioslowski, Interface Sci. 254 (2002) 23.

Source: http://swes.cals.arizona.edu/chorover_lab/pdf_papers/Goyne_et_al_2005.pdf