Environ. Sci. Technol. 2001, 35, 2710-2716
A Fluorescence Spectroscopic Study of Phenanthrene Sorption on Porous Silica† Z. WANG,* D. M. FRIEDRICH,‡ M. R. BEVERSLUIS,§ S. L. HEMMER, A. G. JOLY, M. H. HUESEMANN, M. J. TRUEX, R. G. RILEY, AND C. J. THOMPSON Pacific Northwest National Laboratory, Richland, Washington 99352 B. M. PEYTON Washington State University, Pullman, Washington 98376
Fluorescence spectroscopic characteristics of sorbed phenanthrene in porous silica provide information about its chemical state such as monomer vs dimer or higher aggregates, as well as a basis for high sensitivity detection. In this study, the chemical state and distribution of phenanthrene sorbed in two types of porous silica particles, mesoporous silica (365 µm particle diameter, 150 Å average pore diameter) and microporous silica (custom synthethized, 1 µm particle diameter, 20 Å pore diameter), is determined by fluorescence spectroscopy, fluorescence lifetime measurements, and scanning two-photon excitation fluorescence profiling. From the characteristic fluorescence emission spectra, it is found that at loading levels of e4.7 mg/g (phenanthrene/silica) phenanthrene exists as monomers in both meso- and microporous silica particles for phenanthrene loaded from super critical CO2 (SCF). Twophoton excitation fluorescence intensity distribution profiles indicate that for the mesoporous silica particles phenanthrene is adsorbed throughout the entire silica particle. Introduction of water into phenanthrene-loaded mesoporous silica particles causes instantaneous conversion of phenanthrene from monomer to crystalline form at phenantherene loading levels g4.7 µg/g due to hydration of the silica surface. In this process, sorption of water molecules expels phenanthrene from the surface sorption sites and causes localized phenanthrene concentration beyond its solubility limit, resulting in crystallization. In comparison this fast conversion is not observed for phenanthrene-loaded microporous silica particles that show extremely slow conversion even for phenanthrene loading levels as high as 4.7 mg/g. This difference is interpreted as reflecting hindered diffusion of phenanthrene in the nearly monodispersed micropores with pore sizes close to the molecular diameter of phenanthrene.
Introduction The slow desorption of hydrophobic organic contaminants, such as alkylbenzenes, chlorobenzenes, and polycyclic aromatic hydrocarbons (PAHs), from soils and sediments directly affects the subsurface transport and biodegradation of these contaminants. Much of the recent research in this area has been directed toward the elucidation of factors 2710
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 13, 2001
responsible for such slow desorption processes (1-8). It appears that the constrained diffusion of contaminants through soil organic matter and in the intraparticle mesopores and micropores plays major roles in these processes. Several models based on such considerations have been proposed to account for the experimental observations (1, 4, 9-11). However, reliable predictions for specific soil systems are currently unavailable due to the heterogeneity of soils and the complex interactions of organics in the soil matrix. At a molecular level, sorption and desorption processes involve direct chemical and physical interactions between sorbate and sorbent molecules at the solid surface. For ionic systems, such interactions may involve breaking and forming chemical bonds resulting in new chemical species. For nonionic compound sorption on ionic solids such as soils and silicates, interaction forces may involve van der Waals, dipole-dipole, dipole-induced dipole, and hydrogen bonding (8, 12). Often the sorption sites on mineral surfaces are heterogeneous with a wide distribution of sorption energy and multiple sorption mechanisms. In porous solids, such heterogeneity is further complicated by the size distribution of the pores and the degree of hydration (13-16). Conventionally, sorption studies are conducted by batch and column experiments. These techniques, while providing accurate sorption isotherm and rate data on a macroscopic scale, typically lack structural and molecular information. Recent advancement in molecular spectroscopy makes it possible to observe directly the structural changes involved in sorption/desorption processes. For samples with high sorbate concentration (g1%), several techniques, such as infrared (IR), Raman, nuclear magnetic resonance (NMR), and X-ray based techniques, can provide vital information about the structure, bonding, sorbate-solvent exchange rate, and sorption kinetics (17-19). At low sorbate concentration, molecular fluorescence-based techniques are favored for their high detection sensitivity (20-24). For instance, using a laser fluorescence technique, Ainsworth et al. (21) were able to study the bonding and structure of salicylate surface complexes on alumina in aqueous suspension at salicylate concentrations as low as 10-8 M. Another aromatic compound of environmental importance is phenanthrene. Phenanthrene is a significant component of crude oil and has been used as a model PAH in sorption and desorption studies involving mineral oxides (7, 10-12, 25). It shows rich spectroscopic and photophysical behavior that can serve as a probe of its chemical state and the nature of the interaction between phenanthrene and its chemical and physical environment (13, 26-29). In the present work, fluorescence spectroscopy and fluorescence lifetime measurements are applied to study the sorption of phenanthrene in two types of porous silica particles with distinctly different pore size distributions. The objectives of † This manuscript has been authored by Battelle Memorial Institute, Pacific Northwest Division, under Contract No. DE-AC0676RL0 1830 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. * Corresponding author telephone: (509) 376-6119; fax: (509) 3763650; e-mail:
[email protected]. ‡ Current address: Optical Coating Laboratory, Inc., 2789 Northpoint Parkway, Santa Rosa, CA 95407-7397. E-mail: dfriedrich@ ocli.com. § Current address: Department of Optics, University of Rochester, Rochester, NY 14627.
10.1021/es001658f CCC: $20.00
2001 American Chemical Society Published on Web 05/25/2001
TABLE 1. Physical Properties of the Silica Particles mesoporous silica average particle size average pore size total N2 BET surface area total organic carbon microporous silica average particle size average pore size total N2 BET surface area total organic carbon
360 µm 15 nm 314 m2/g 68 µg/g 1 µm 2 nm 868 m2/g 31 µg/g
this work are (i) to demonstrate the effectiveness of fluorescence spectroscopy in the determination of the characteristics of phenanthrene sorbed on porous silica surfaces and (ii) to use fluorescence spectroscopic techniques to determine what role pore size plays for phenanthrene sorption and desorption in porous silica.
Experimental Section Phenanthrene (99.5+%, zone-refined) and mesoporous silica particles (365 µm average particle diameter with 150 Å average pore size) were purchased from Aldrich. Microporous silica particles (1 µm particle diameter with pore size of 20 Å) were synthesized according to the procedures of Bruinsma et al. (30). The silica particles were equilibrated with water over a period of 3 to 4 days, filtered to remove bulk water and dried over vacuum at room temperature in a desiccator containing Drierite for 5 to 6 days. The removal of physically adsorbed water was achieved by evacuation at a more reduced pressure (e4 × 10-6 Torr) for 5 to 6 days (31). The pore size of the particles was analyzed using a Micrometrics surface area analyzer (model 2010 Micrometrics Instrument Corp., Norcross, GA). The carbon content of the particles was determined by gas chromatographic analysis of methane generated from the catalytic conversion of CO2 that was released when the particles were heated to 550 °C in the presence of oxygen. Properties of the silica particles are listed in Table 1. Deionized water (18 MΩ‚cm) from a Millipore reverse osmosis and ion exchange system with ultraviolet sterilization of the finished water was used for wetting the silica particles. The details of the technique of loading phenanthrene into porous silica particles in SCF has been described elsewhere (32). Briefly, a Dionex model SFE-703 supercritical extraction instrument was integrated with an Eldex model B-100-S high performance liquid chromatography pump system and a Shimadsu UV-2401 spectrophotometer equipped with a custom mounted high-pressure flow cell (Shimadzu SPDM6A) to facilitate supercritical CO2 circulation in a closed loop. Two high-pressure stainless steel vessels (Keystone Scientific) were included in the closed-loop. One vessel was used to dissolve the phenanthrene in SCF and the other to contain the silica particles. Seven Rheodyne 7000 valves and two on/off SSI needle valves were used to direct the flow of SCF and isolate the pressure vessels from the SFE system pump. First, the SCF was circulated through the phenanthrene vessel allowing the dissolution of phenanthrene and then the phenanthrene solution in SCF was circulated through the silica vessel to initiate the sorption of phenanthrene in the silica particles. The changes of the UV-visible absorption spectra (250-280 nm) recorded via the flow cell provide quantitative information about the status of phenanthrene dissolution and sorption. A constant reading of phenanthrene absorbance was considered as an indication that dissolution/sorption equilibrium was reached. Typically, the time required for phenanthrene dissolution was less than 10 min, and sorption equilibrium was reached within an hour. The amount of phenanthrene sorbed to silica was
determined from the difference of the phenanthrene absorbances in the SCF phase before and after the sorption. The calculated Kd values for phenanthrene sorbed to silica in SCF varied between 0.24 and 0.55, depending on the specific experimental conditions. Fluorescence excitation and emission spectra were recorded on a SPEX Flourolog II fluorimeter. The fluorimeter was equipped with a 450 W xenon arc lamp, double monochromators (SPEX 1680) for excitation and emission, and a cooled photomultiplier tube in photon counting mode. To correct for minor variations of the lamp intensity, light from the excitation monochromator was reflected by a thin quartz plate into a cell containing rhodamine dye. Red emission from this “quantum counter” was detected by a second photomultiplier tube and recorded by the fluorimeter in an analog reference channel. Fluorescence intensity from the sample was normalized at each point to the corresponding excitation intensity in the reference channel. Both the mesoand microporous silica particles displayed weak fluorescence background spectra with a maximum intensity around 450 nm. Therefore, fluorescence spectra of phenanthrene sorbed in silica particles were obtained by subtracting the background fluorescence spectra of the “blank” silica particles (no phenanthrene) recorded from the same batch of silica particles as those used for the phenanthrene sorption experiment. Fluorescence lifetime measurements were carried out on a conventional time-correlated single-photon-counting apparatus (33). Briefly, the sample was excited with frequencydoubled picosecond laser pulses at 292 nm derived from a mode-locked Coherent Antares Nd:YAG laser-pumped dye laser system operated between 580 and 610 nm. The fluorescence emission was collected with a pair of parabolic mirrors, focused into an American Holographic double monochromator and detected by a microchannel plate photomultiplier tube. The instrument response was typically between 30 and 40 ps. For each data set, collection was terminated when the peak intensity of the fluorescence decay curve was at least 104 counts. All experiments were performed at 25 °C. Phenanthrene distribution profiles were obtained by placing a sample particle on a microscope cover slip fixed on a translation stage and scanning the sample particle stage across the focal point of the laser focused by a 20× objective with a numerical aperture of 0.70. Fluorescence excitation was achieved via two-photon absorption from the 658 nm output of a Coherent Satori 774 dye (Kiton Red) laser (110 fs, 200 mW) pumped by the frequency-doubled output of a mode-locked Coherent Antares Nd:YAG laser operating at 76 MHz. The 658 nm laser beam was expanded by a set of matched fused silica lenses, fed into an inverted optical microscope (Nikon TE300) from the back and reflected into the objective by a short pass dichroic filter placed inside the filter cassette at 45°. The resulting fluorescence (350-450 nm) was collected by the same objective, transmitted through the dichroic filter and directed into the microscope side port and detected by a Hamamatsu R4220p photomultiplier tube. The fluorescence intensity data was collected in photon counting mode with a Stanford Research SR400 dual channel photon counter. Typically for each profile at least 200 data points were recorded across the particle using a step size of 0.88 µm.
Results and Discussion The Chemical State of Phenanthrene Adsorbed at the Silica Surface. The absorption spectra of phenanthrene sorbed in porous silica particles at loading levels between 0.47 µg/g to 4.7 mg/g were recorded. For comparison, the absorption spectra of phenanthrene in cyclohexane was also recorded. Representative absorption spectra are shown in Figure 1 for VOL. 35, NO. 13, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2711
TABLE 2. Fluorescence Lifetimes of Phenanthrene Sorbed in Porous Silica Particles sample type mesoporous mesoporous mesoporous mesoporous microporous microporous microporous microporous
FIGURE 1. Absorption spectra of 1 × 10-4 M phenanthrene solution in cyclohexane (a), phenanthrene sorbed to mesoporous silica at 4.7 mg/g loading (b), and phenanthrene sorbed to microporous silica at 4.7 mg/g loading level (c).
FIGURE 2. Fluorescence emission spectra of phenanthrene: (a) 2 × 10-6 M cyclohexane solution; (b) SCF loaded microporous silica particles at 4.7 mg/g; (c) SCF loaded mesoporous silica particles at 4.7 mg/g; (d) SCF loaded onto microporous silica particles at 0.47 mg/g; (e) SCF loaded mesoporous silica particles at 0.47 mg/g; (f) SCF loaded microporous silica particles at 0.047 mg/g; (g) SCF loaded mesoporous silica particles at 0.047 mg/g; (h) SCF loaded microporous silica particles at 4.7 µg/g; (i) SCF loaded mesoporous silica particles at 4.7 µg/g. The spectral intensities were normalized to the peak intensity and offset for clarity. λex ) 292 nm. phenanthrene in cyclohexane solution (trace a) as well as sorbates at 4.7 mg/g loading level (traces b and c) in mesoand microporous silica particles, respectively. Absorption spectra of phenanthrene sorbed to the silica at lower loading levels are similar to those shown in Figure 1, but with lower signal-to-noise ratios. As seen in Figure 1, the absorption spectra of sorbed phenanthrene are comprised of two sets of bands, a stronger set located in the wavelength range of 260 to 300 nm with a spectral maximum of 292 nm and a weaker set in the wavelength range of 300 to 350 nm. These characteristics are the same as phenanthrene in dilute cyclohexane solution (34). Excitation at the absorption maximum (292 nm) results in fluorescence emission spectra in the range of 350 to 450 nm with an emission maximum located at 368.5 nm at all phenanthrene loading levels (Figure 2). The Aldrich zonerefined phenanthrene contains e0.5% anthracene as an impurity. Our previous work (26) has shown that molecular association among the sorbate molecules facilitates energy transfer from phenanthrene to anthracene upon phenanthrene excitation. This results in characteristic fluorescence emission spectra that have contributions from anthracene, which appears at longer wavelengths relative to that of phenanthrene. Therefore, the absence of any anthracene contribution in the emission spectra shown in Figure 2 indicates that phenanthrene sorbed in both mesoporous and microporous silica particles exists mainly as monomers and the formation of microcrystalline phenanthrene is negligible. 2712
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 13, 2001
loading level (µg/g)
lifetimes (ns)
4.70 47.0 470 4700 4.70 47.0 470 4700
4.08, 0.98 (m)a 6.79, 1.27 (m) 8.46, 1.32 (m) 6.42, 1.33 (m) 15.6, 3.34 (m) 32.8, 2.04 (m) 33.9, 2.58 (m) 31.1, 3.25 (m)
phenanthrene in water, 7.2 × 10-6 M phenanthrene in cyclohexane, 7.0 × 10-6 M
35.0b 45.8
a m indicates the major component. b The solution was saturated and the concentration is assumed to reach the solubility of phenanthrene in water.
The diameter of phenanthrene is ca. 1.2 nm. If the entire silica surface is accessible to phenanthrene and phenanthrene lies flat on the surface, it is estimated that 1 g of silica particles can accommodate 160 and 358 mg of phenanthrene with monolayer coverage for meso- and microporous silica particles, respectively. For the highest loading level of phenanthrene in the present work (4.7 mg/g), the surface coverage is ∼3% for mesoporous silica and 1.1% for microporous silica. At such low surface coverage, it is not surprising to see the absence of aggregate formation. Barbas et al. (20) studied phenanthrene sorption from cyclohexane solution of porous silica particles with an average pore size of 6 nm. They found that at a loading level of 18 mg/g (corresponding to 23.8% surface coverage), the formation of microcrystalline phenanthrene was minimal. However, at a loading level of 180 mg/g (238% surface coverage) the emission spectrum of the sorbed phenanthrene was dominated by the characteristics of the crystalline phenanthrene used to make the phenanthrene solutions. Our observations are completely consistent with these results, despite differences in the pore size distribution of the silica particles and the loading method between the present work and that of Barbas et al. (20). Fluorescence decay curves for phenanthrene sorbed on both meso- and microporous silica particles at different loading levels and for phenanthrene in dilute cyclohexane solution were recorded via excitation at 292 nm. The fluorescence lifetimes were calculated by fitting the decay data with exponential functions with either discrete or Lorentzian distributions of fluorescence lifetimes. The results are listed in Table 2, along with phenanthrene fluorescence lifetimes in aqueous solution. For dilute solutions of phenanthrene in either cyclohexane or deionized water, a single, discrete lifetime characterizes the fluorescence decay. However, for phenanthrene sorbed to porous silica the fluorescence decay is more complicated. Typically, two fluorescence lifetimes are needed to fit the data and the best fit is obtained when a Lorentzian type distribution of lifetimes is used in the data fitting procedures. Phenanthrene sorbed to mesoporous silica shows one lifetime of ca. 1 ns and a second lifetime between 4 and 8.5 ns, while phenanthrene sorbed to microporous silica particles shows a lifetime between 2 and 3.4 ns and a second component from 15 to 34 ns. The appearance of nondiscrete fluorescence lifetimes for surface sorbed PAHs has been reported by other groups (13-15, 35). For example, Ruetten and Thomas (35) found that in simple homogeneous systems, such as in solution and micelles, pyrene showed single-exponential fluorescence decay. Once pyrene was sorbed to a silica surface, a Gaussian type distribution of fluorescence lifetimes was needed to fit the fluorescence decay data. The wide distribution of fluorescence lifetimes reflects the heterogeneous nature of the silica
surface. In other words, there are a series or group of sorption sites on porous silica surfaces accessible by PAH molecules. The sorption energy of the individual sites varies and within each series or group of sorption sites the sorption energy can be described mathematically by a certain type of distribution. It is commonly accepted that surface characteristics of amorphous silica depend on the temperature and pressure at which the surface is treated (31, 36, 37). Under ambient conditions, an amorphous silica surface is completely hydroxylated with one or more layers of physi-sorbed water hydrogen-bonded to the hydroxyl groups. The physi-sorbed water can be removed by evacuation at 10-6 mmHg pressure at room temperature. Under ambient pressure, amorphous silica retains all surface hydroxyl groups at temperatures e500 °C. At higher temperatures, condensation of surface hydroxyl groups starts to occur, resulting in the formation of strained siloxane bridges that convert to stable siloxane groups at higher temperatures (e.g., 900 °C) (37). Rehydroxylation of a dehydroxylated silica surface is slow. At room temperature, complete rehydroxylation of silica calcined at 900 °C may take up to five years (36). The silica particles used in this work were calcined at 550 °C and were evacuated at 5 × 10-6 Torr for at least 5 days after rehydration such that it is reasonable to assume that the surface was mostly hydroxylated (31). Sorption of additional water was possible since the particles were intermittently exposed to ambient air and purged before the lifetime measurement with N2 that contained trace amounts of water vapor. The sorption of PAH on silica surfaces is caused by interaction between the π-electronic orbitals of the PAHs and either the surface hydroxyl groups or the tightly sorbed water molecules (14, 35). The interaction between the π-electronic orbitals of the PAHs and the well-aligned surface hydroxyl groups is expected to be stronger and, thus, significantly shortens the fluorescence lifetime. Therefore, the major component of the sorbed phenanthrene with shorter fluorescence lifetimes may be attributed to those sorbed to the fully hydroxylated silica surface with no additional physi-sorbed water. The minor component, with longer fluorescence lifetimes, is tentatively assigned to phenanthrene sorbed to hydroxyl groups that have additional physi-sorbed water(s), based on the fact that for phenanthrene sorbed in microporous silica particles, the fluorescence lifetime of the minor component is close to that of phenanthrene in water (Table 2). Assuming a hydroxyl density of 4.6 OH/nm2 (37) and a molecular diameter of 1.2 nm for phenanthrene, on average each phenanthrene molecule may interact with at least four hydroxyl groups. The much shorter fluorescence lifetime of sorbed phenanthrene on the silica surface as compared to aqueous solution indicates a strong interaction between phenanthrene and the silica surface hydroxyl groups. It is unclear why the fluorescence lifetime is relatively longer on the microporous silica particle than on mesoporous silica particles. One possibility is that the small curvature of the micropores leads to less effective interaction between the hydroxyl groups and phenanthrene, since the steric constraints do not allow the phenanthrene to lie “flat” on the hydroxylated silica surface. Liu et al. (13-15) studied sorption of phenanthrene to both “dry” and “wet” Kieselgel-60 porous silica. To obtain “dry” silica particles, the silica was heated at 800 °C for 8 h under vacuum (e10-3 Torr), while for the “wet” particles the evacuation was performed at 25 °C. The sorption of phenanthrene on silica was carried out by phenanthrene evaporation in a sealed apparatus that also contained silica particles in a separate but connected chamber. For the dry silica particles, Liu et al. obtained a fluorescence lifetime of 6.8 ns, whereas for the wet silica particles the lifetime was centered at 29 ns. They attributed the lifetime of 6.8 ns to phenanthrene that interacts with isolated surface hydroxyl groups on the dried
silica particle surface where the hydroxyl density is much lower. The 29 ns lifetime on wet silica was attributed to phenanthrene sorbed to hydrogen-bonded hydroxyls on the “wet” silica particle surfaces. Because of the differences in sample preparation and phenanthrene loading mechanisms, quantitative comparison between these results and results of the current work may not be valid. However, the general trend is the same: sorption of phenanthrene on silica surfaces causes significant reduction of the phenanthrene fluorescence lifetime as compared to phenanthrene in aqueous solution unless there is enough local physi-sorbed water to mimic the solvation environment of phenanthrene in water. Such a difference in fluorescence lifetimes suggests that phenanthrene interacts with aligned hydroxyls on flat silica surfaces more effectively than with the hydroxyl groups of bulk water. Phenanthrene Distribution in Porous Silica. Phenanthrene distribution profiles in mesoporous silica particles were obtained by recording the fluorescence intensity across an individual particle by scanning the sample stage containing the particle through the focal point of an inverted optical microscope objective that serves to focus the laser beam. The chosen excitation wavelength (658 nm) corresponds to a two-photon absorption maximum of phenanthrene (38). Two-photon excitation is a process whereby an atom or molecule simultaneously absorbs two photons and promotes an electron from its ground state to an excited state with an energy equal to the sum of the absorbed photons (38-40). Fluorescence intensity recorded at a series of excitation laser powers indicates that the fluorescence intensity is proportional to the square of the laser power, confirming a twophoton excitation mechanism. The transition probability for two-photon absorption depends on the product of the instantaneous light intensities of the two photons, i.e., the square of the intensity if the two photons are from the same light source. Because of this square-law intensity dependence, predominantly molecules in the focal volume of the incident beam are excited, thus offering spatial resolution equivalent to that of confocal microscopy. The use of lower photon energy excitation light also alleviates the common bleaching problem encountered in one-photon excitation processes (41). Fluorescence intensities recorded for phenanthrene solutions in cyclohexane at various concentrations indicate that at a constant laser power, the two-photon excitation fluorescence intensity of phenanthrene was linearly proportional to its concentration over a wide range: 1 × 10-5 to 1 × 10-2 M. In the microporous silica particles, the phenanthrene fluorescence intensity was stable within the time frame of the measurement. However, for phenanthrene sorbed to mesoporous silica particles, the fluorescence intensity recorded at the same location fluctuated after several seconds. Therefore, a minimal integration time was used for the measurement of phenanthrene sorbed to these particles. Typical phenanthrene distribution profiles are shown in Figure 3 where the focal point is at or near the center of the particle. Figure 3 indicates that phenanthrene sorption occurs throughout the mesoporous silica particle with a highly nonuniform distribution. While some fluorescence is observed from all regions of each particle, there are regions of very bright fluorescence, indicating locally high concentrations of phenanthrene. Similar measurements were performed for blank mesoporous silica particles, which were treated with exactly the same procedure except without the addition of phenanthrene during the SCF loading process. It was found that the intensities of the fluorescence profiles of the phenanthrene-free mesoporous silica particles were negligible as compared with phenanthrene-loaded particles. In previous modeling studies involving organic compound sorption/desorption in soil and aquifer materials, the disVOL. 35, NO. 13, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2713
FIGURE 4. Fluorescence spectra of SCF-loaded phenanthrene in mesoporous silica at 47 µg/g; (a) dry particle; (b) water wetted particle; (c) crystalline particle. λex ) 292 nm. For clarity, the spectra were normalized to the peak intensity and offset in the vertical scale.
FIGURE 3. Phenanthrene distributions in a representative mesoporous silica particle at 0.47 mg/g loading level. The distribution is obtained by scanning the sample stage containing the particle along the horizontal lines indicated in the inset. The focal point of the microscope objective is at or near the center of the particle. Areas of increased intensity represent regions with high local phenanthrene concentrations. tribution of organic compounds has been assumed to be uniform throughout the entire particle (4, 11, 42). These studies demonstrated that such an assumption is valid for the simulation of experimental data at a macroscopic level. However, the present results (Figure 3) suggest that, on the basis of individual particles, the distribution of phenanthrene in mesoporous silica sorbed from SCF is nonuniform. From the emission spectra of the sorbed phenanthrene (Figure 2), we can rule out any significant phenanthrene aggregation. Therefore, an explanation for the nonuniform distribution of phenanthrene in mesoporous silica particle is that the distribution of pores is nonuniform within the particle or there is preferential sorption at certain pore surfaces. The latter may be caused by the geometrical difference among the pores, SCF flow characteristics in the loading system or the presence of organic matter. However, as will be shown in the next section, it is unlikely that the presence of a small amount of organic matter is the main source of phenanthrene sorption in the porous silicas used in this work. Attempts were made to record the phenanthrene distribution profiles inside the microporous silica particles. As expected, the size of the particles (∼1 µm) was below the spatial resolution limit of the method. Hence, distribution of phenanthrene inside these particles could not be resolved. However, the high intensity of the fluorescence emission from phenanthrene sorbed to the microporous silica particles supports the conclusion that phenanthrene sorption did occur. Since the sorbed phenanthrene is less than monolayer coverage, as indicated by the fluorescence emission spectra, phenanthrene must have entered the micropores. If phenanthrene sorption had only occurred on the external surface, the surface coverage would have been greater than a monolayer and crystal-like fluorescence emission would have been observed. Hydration-Forced Desorption of Phenanthrene from Silica Surfaces. The presence of water greatly affects the sorption and desorption of phenanthrene on porous silica surfaces due to hydroxylation and hydration. Both hydroxylation and hydration can either change the number of 2714
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 13, 2001
FIGURE 5. Fluorescence spectra of SCF-loaded phenanthrene in microporous silica at 47 µg/g; (a) dry particle (dotted line); (b) water wetted particle (solid line). λex ) 292 nm. Inset shows the relative intensity change as a function of time for water wetted microporous particles at 47 µg/g loading. λex ) 292 nm. For clarity, the spectra were normalized to the peak intensity and offset in the vertical scale. sorption sites or block the access of phenanthrene to some of the pores. Equally possible is that the interfacial water may cause a dramatic reduction of the rate of diffusion of phenanthrene into the pores. While conducting sorption experiments at extended periods of time may be difficult, hydration of porous silica particles which already have phenanthrene sorbed through other methods, such as SCF, and at higher loading levels, offers a unique way to study the effect of water on the interaction of phenanthrene with silica surfaces. In this work, the mesoporous silica particles have an average pore diameter of 15 nm with a pore diameter range from a few nanometers to a few thousand nanometers. Therefore, hydroxylation and hydration will modify the surface characteristics, but will hardly affect the pore size. However, the microporous silica particles have a uniform pore diameter of two namometers, which is very close to the molecular diameter of phenanthrene. Hydroxylation and hydration will not only change the silica surface properties but also may reduce the pore size to an extent that hinders the diffusion of phenanthrene inside the micropores. Fluorescence spectra were recorded immediately after a small amount of water was added to the silica particles with phenanthrene loaded in SCF. Representative emission spectra are shown in Figures 4 and 5 at a loading level of 47.0 µg/g for meso- and microporous silica particles, respectively. The spectral changes were similar for microporous silica particles at all loading levels and for mesoporous silica particles at loading levels g47.0 µg/g. As seen in Figure 4, for mesoporous silica particles, a majority of the sorbed phenanthrene converted from a monomer to a crystalline form immediately after contact with water. Clearly, the presence of bulk water drastically modifies the silica surface through hydroxylation and hydration of the silica surface and causes fast desorption of phenanthrene. Such desorption causes a sudden increase
in local phenanthrene concentrations in the mesopores to levels beyond its solubility and thus results in the conversion of phenanthrene from monomeric form to the microcrystalline form. Such a conversion is in agreement with the expected decreasing hydrophobicity of the surface from a less hydrated state to a more hydrated state, which repels phenanthrene from the surface. In contrast to the mesoporous silica particles, spectral changes were not observed for phenanthrene sorbed to microporous silica particles at any loading levels, as shown in Figure 5 for the 47.0 µg/g phenanthrene-sorbed microporous silica particles. Even after the particles had been in contact with water for more than a week, formation of crystalline phenanthrene was still negligible (Figure 5 insert). From the low sorption capacity of phenanthrene in microporous particles achieved in water (42), it is expected that desorption would occur at loading levels g47 µg/g. Therefore, the absence of a spectral change from monomer to microcrystalline indicates that movement (e.g., diffusion) of phenanthrene inside the micropores is greatly hindered, prohibiting crystallization. In comparison, in mesoporous silica particles the pore sizes are much larger than the molecular diameter of phenanthrene so that phenanthrene can diffuse out of the pores without much resistance. The effect of micropores on the slow sorption/desorption of organic contaminants in porous minerals and sediments has been the focus of other works (2-6, 8, 12) with no consistent conclusions. The results are further complicated by the presence of organic matter. Werth and Reinhard (3) measured the rate of exchange between deuterated trichloroethylene in fast desorbing sites and nondeuterated trichloroethylene in slow desorption sites in a silica gel, a Santa Clara sediment, and a Livermore clay/silt fraction. The results obtained with different exchange times supported the hypothesis that slow sorption kinetics can be controlled by diffusion in micropores. Cornelissen et al. (2) determined the desorption rate of PCBs and chlorobenzenes from model sorbents in which either micropore diffusion or organic matrix diffusion/entrapment can occur. Their results suggest not only that organic matter entrapment determines slow desorption, but that diffusion through hydrophobic pores cannot be ruled out. In a study of phenanthrene sorption to eight particulate mineral solids with varied porosity and pore geometries in aqueous solution, Huang et al. (12) found that sorption coefficients were an order of magnitude lower for particles with an internal pore surface as compared to those with external surfaces only. Such results suggest that nearsurface particle or pore geometry, as well as the preferential sorption of water on a silica surface, can affect the accessibility of phenanthrene to the internal pores. PAHs interaction with organic matter is an important factor in the sorption and mass transport of PAHs in natural sediments and soils. It has been suggested that the intraorganic particle diffusion limits the sorptive uptake of phenanthrene in small rock fragments containing large organic particles (43). Several authors have proposed to use the concept of “soft” vs “hard” carbon to describe the expanded and condensed soil organic matter with different diagenetic origins and sorption properties for organic contaminants (8, 10). However, in the present work the organic matter content in both types of silica was low (Table 1). More importantly, the relative amount of total organic carbon (TOC) in the mesoporous silica was more than twice that in the microporous silica. If sorption of phenanthrene were at organic matter sites, slower desorption would be expected in the mesoporous silica upon addition of water because of its higher TOC content. The experimental results were just the opposite (Figures 4 and 5). Therefore, it is safe to conclude that sorption of phenanthrene at organic carbon is not the dominant mechanism for the porous silica used
in the present work. Instead, it is the size of the pore that plays a dominant role. The microporous silica particles used in this work have a pore size comparable to the size of phenanthrene. The slow diffusion of phenanthrene observed in these microporous silica particles further suggests that only when the pore size is close to the molecular size of the PAHs, does hindered diffusion become critical. For most PAHs such effects may only occur in micropores (pore diameter e 2 nm). For most silica gels, a large majority of pores are mesopores or macropores that are much larger than the molecular size of PAHs, and thus hindered diffusion in micropores becomes less important. The results of this work also indicate that the role of solvent in the sorption and desorption of PAHs in porous silica depends on its interaction with porous silica surfaces. The presence of water will hinder the diffusion of PAHs in micropores due to surface hydration. Consequently, dehydration of sorbent should accelerate the diffusion rate of PAHs. This is consistent with the observations of Grathwohl and Reinhard for the removal of sorbed trichloroethylene (TCE) in columns packed with either wet or oven-dry Santa Clara aquifer material by soil-gas venting (44). For the dry material, the removal rate of TCE was proportional to that of the gas flow. However for the wet material, the TCE removal rate was independent of the gas flow rate and limited by the slow intraparticle mass transfer. In other solvents, such as in supercritical fluid or organic solvent, the blocking effect in the micropores is reduced due to the weaker interaction between the solvent and the mineral surface. The transport of PAHs in micropores can also readily occur. Therefore, for wet sorbents, such as natural soil and sediments, the effectiveness of decontamination of PAHs closely depends on the contact time (aging) and the moisture content. The higher the degree of sorbent hydration, the slower the sorption/desorption rate of PAHs will be due to slower diffusion of PAHs in the micropores. Aside from other factors such as sorption and diffusion in organic matter, aging allows diffusion of PAHs into micropores (sequestration), and thus their removal will be more difficult.
Acknowledgments This research was supported by the Environmental Technology Partnerships (ETP) Program, Office of Biological and Environmental Research (BER), Office of Energy Research, U.S. Department of Energy. Pacific Northwest National Laboratory is operated for the U.S. Department of Energy by Battelle under Contract DE-AC06-76RLO 1830.
Literature Cited (1) Luthy, R. G.; Aiken, G. R.; Brusseau, M. L.; Cunningham, S. D.; Gschwend, P. M.; Pignatello, J. J.; Reinhard, M.; Traina, S. J.; Weber, W. J., Jr.; Westall, J. C. Environ. Sci. Technol. 1997, 31, 3341. (2) Werth, C. J.; Reinhard, M. Environ. Sci. Technol. 1999, 33, 730. (3) Werth, C. J.; Reinhard, M. Environ. Sci. Technol. 1999, 31, 6697. (4) Wu, S. C.; Gschwend, P. M. Environ. Sci. Technol. 1986, 20, 717. (5) Cornelissen, G.; Van Noort, P. C. M.; Parsons, J. R.; Govers, H. A. J. Environ. Sci. Technol. 1997, 31, 454. (6) Cornelissen, G.; Van Noort, P. C. M, Govers, H. A. J. Environ. Sci. Technol. 1998, 31, 33124. (7) Huang, W.; Weber, W. J., Jr. Environ. Sci. Technol. 1997, 31, 3238. (8) Pignatello, J. J.; Xing, B. Environ. Sci. Technol. 1996, 30, 1. (9) Young, T. M.; Weber, W. J. Jr. Environ. Sci. Technol. 1997, 31, 1692. (10) Young, T. M.; Weber, W. J. Jr. Environ. Sci. Technol. 1995, 29, 92. (11) Ball, W. P.; Roberts, P. V. Environ. Sci. Technol. 1991, 25, 1237. (12) Israelachvili, J. N. Intermolecular and Surface Forces, 2nd Ed.; Academic Press: New York, 1997. (13) Huang, W.; Schlautman, M. A.; Weber, W. J., Jr. Environ. Sci. Technol. 1996, 30, 2993. VOL. 35, NO. 13, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2715
(14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) (30) (31)
Liu, Y. S.; Ware, W. R. J. Phys. Chem. 1993, 97, 5980-5986. Liu, Y. S.; Ware, W. R. J. Phys. Chem. 1993, 97, 5987. Liu, Y. S.; de Mayo, P.; Ware, W. R. J. Phys. Chem. 1993, 97, 5995. Foster, A. L.; Brown, G. E. Jr.; Parks, G. A. Environ. Sci. Technol. 1998, 32, 1444. Little, L. H. Infrared Spectra of Adsorbed Species; Academic Press: New York, 1966. Casey, W. H.; Phillips, B. L.; Nordin, J. P.; Sullivan, D. J. Geochim. Cosmochim. Acta 1997, 62, 1789. Barbas, J. T.; Sigman, M. E.; Dabestani, R. Environ. Sci. Technol. 1996, 30, 1776-1780. Ainsworth, C. C.; Friedrich, D. M.; Gassman, P. L.; Wang, Z.; Joly, A. G. Geochim. Cosmochim. Acta 1998, 62, 595-612. Wang, Z.; Ainsworth, C. C.; Friedrich, D. M.; Gassman, P. L.; Joly, A. G. Geochim. Cosmochim. Acta 2000, 64, 1159. Hering, J. G.; Stumm, W. Langmuir 1991, 7, 1567. McCarroll, M.; Joly, A. G.; Wang, Z.; Friedrich, D. M.; von Wandruszka, R. J. Colloid Interface Sci. 1999, 218, 260. Mader, B. T., Uwe-Goss, K., Eisenreich, S. J. Environ. Sci. Technol. 1997, 31, 1079. Wang, Z.; Hemmer, S. L.; Friedrich, D. M.; Joly, A. G. J. Phys. Chem. A, in press. Rhodes, T. A.; Fox, M. A. J. Phys. Chem. 1997, 100, 17931. Ramamurthy, V. J. Am. Chem. Soc. 1994, 116, 1345. Tro, N. J.; Nishimura, A. M.; George, S. M. J. Phys. Chem. 1989, 93, 3276. Bruinsma, P. J.; Kim, A. Y.; Liu, J.; Baskaran, S. Chem. Mater. 1998, 9, 2507. Young, G. J. J. Colloid Sci. 1958, 13, 67.
2716
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 13, 2001
(32) Riley, R. G.; Thompson, C. J.; Huesemann, M. H.; Truex, M. J.; Friedrich, D. M.; Wang, Z.; Peyton, B. M., unpublished data. (33) Holtom, G. R. Proc. SPIE-Int. Soc. Opt. Eng. 1990, 1204. (34) Berlman, Isadore B. Handbook of Fluorescence Spectra of Aromatic Molecules, 2nd ed.; Academic Press: New York, 1971. (35) Ruetten, S. A.; Thomas, J. K. J. Phys. Chem. 1998, 102, 598. (36) Zhuravlev, L. T. Colloids Surf. 1993, 74, 71. (37) Bergna, H. E. in The Colloid Chemistry of Silica; Bergna, H. E., Ed.; American Chemical Society: Washington, D. C., 1994; pp 1-50. (38) Dick, B.; Hohlneicher, G. Chem. Phys. Lett. 1983, 97, 324. (39) Go¨ppert-Mayer, M. Ann. Phys. 1931, 9, 273. (40) Friedrich, D. M. J. Chem. Educ. 1982, 59, 472. (41) Denk, W.; Piston, D. W.; Webb, Watt W. in Handbook of Biological Confocal Microscopy, 2nd Ed.; Pawley, J. B., Ed.; Plenum Press: New York, 1995; pp 445-458. (42) Huesemann, M. H.; Fortman, T.; Riley, R. G.; Thompson, C. J.; Wang, Z.; Truex, M. J.; Peyton, B. M., unpublished data. (43) Ru ¨ gner, H.; Kleineidam, S.; Grathwohl, P. Environ. Sci. Technol. 1999, 33, 1645-1651. (44) Grathwohl, P.; Reinhard, M. Environ. Sci. Technol. 1993, 27, 2360-2366.
Received for review September 11, 2000. Revised manuscript received April 9, 2001. Accepted April 17, 2001. ES001658F
