Environ. Sci. Technol. 2004, 38, 4369-4376
Solid-State NMR Characterization of Pyrene-Cuticular Matter Interactions JOSEPH R. SACHLEBEN* Chemistry Department, Otterbein College, Westerville, Ohio 43081
Introduction
BENNY CHEFETZ Department of Soil and Water, Faculty of Agricultural, Food, and Environmental Quality Sciences, The Hebrew University of Jerusalem, P.O. Box 12, Rehovot, 76100, Israel ASHISH DESHMUKH AND PATRICK G. HATCHER Department of Chemistry, The Ohio State University, Columbus, Ohio 43210
One- and two-dimensional nuclear magnetic resonance (NMR) experiments were performed on Agave americana cutan and tomato cutin to examine the interactions between a hydrophobic pollutant, pyrene, and cuticular material. Variable-temperature NMR experiments show that cutan, an acid- and base-resistant cuticular biopolymer, undergoes the characteristic melting behavior of “polyethylene-like” crystallites, while the tomato cutin does not. The melting point of A. americana cutan was found to be approximately 360 K, which is consistent with the thickness of the polyethylene crystallites of 30-40 methylene units. Sorption models predict that the sorption behavior of hydrophobic pollutants should depend on the phase of the cuticular material. 13C NMR experiments on labeled pyrene were performed. The 13C T1 of pyrene decreases significantly from that of crystalline pyrene upon sorption to both tomato fruit cutin and A. americana cutan, indicating that the pyrene is mobile upon sorption. Magic angle spinning experiments at low spinning frequencies (2-4 kHz) provided the chemical shift anisotropy (CSA) parameters δ, the anisotropy, and η, the asymmetry parameter, for crystalline and sorbed pyrene. For crystalline pyrene, two types of crystallographically distinctive pyrenes were observed. The first had δ ) -97.4 ( 0.5 ppm and η ) 0.934 ( 0.006, while the second had δ ) -98.1 ( 0.5 ppm and η ) 0.823 ( 0.008. After sorption to cutan, these CSA parameters were found to be δ ) -78.9 ( 5.3 ppm and η < 0.70 independent of the length of time since completion of the sorption procedure. In tomato cutin, the CSA parameters were found to be dependent upon the time since completion of the sorption procedure. One and onehalf months after sorption, δ was found to have a value of -30.4 ppm < δ < 0.0 ppm and η was undeterminable, while after 22 months these values become δ ) -80.0 ( 3.3 ppm and η < 0.42. These changes in the CSA parameters demonstrate that upon sorption of pyrene to cutan, the pyrene undergoes anisotropic motion, while in cutin pyrene * Corresponding author phone: (614)823-1666; fax: (614)823-1968; e-mail:
[email protected]. 10.1021/es035362w CCC: $27.50 Published on Web 07/17/2004
initially can tumble isotropically, but after 22 months this motion also becomes anisotropic. 2D heteronuclear correlation experiments indicate that pyrene is in close proximity to aliphatic cuticular materials after sorption. This work is directly relevant toward understanding the physical and chemical mechanisms of pollutant sorption to soil organic matter and, thus, help develop improved sorption models and pollution remediation techniques.
2004 American Chemical Society
In both forest and agricultural soils, leaf derived plant litter constitutes a significant portion of soil organic matter (SOM) (1-5). This litter mainly contributes aliphatic compounds to SOM, many of which are derived from the plant cuticle (6, 7). Plant cuticle is a thin layer of predominantly lipid material, which is synthesized by the epidermal cells and deposited on the outer walls of leaves. Modeled as a bilayer, the outer region of the plant cuticle is composed mainly of aliphatic lipids, while the inner layer contains large amounts of various cell-wall polysaccharides. The principal lipid component of the cuticle is the polyester cutin (8, 9). Cutin is an insoluble polyester of cross-linked hydroxy-fatty acids, hydroxyepoxyfatty acids, and waxes with 16-18 carbons (9). Some plant cuticles have been shown to additionally contain an acid and base hydrolysis-resistant biopolymer known as cutan (7). Cutan is made up of long-chain fatty acids (n > 30) attached to an aromatic core via ester linkages (10). Both of these biopolymers are thought to be difficult to degrade microbiologically, and several studies showed that the aliphatic biopolymers could be selectively preserved in soils with little or no alteration (6, 11, 12). Simplified models of cutan and cutin are shown in Figure 1a and c, respectively. 13C NMR studies on bulk soil samples from numerous localities indicate a significant aliphatic fraction of SOM (1315). The increase of alkyl-carbon contents (0-50 ppm region in the 13C NMR spectra) during humification or plant litter decomposition can be explained by relative enrichment of extractable and bound lipids (16), plant-derived biopolyesters, and nonsaponifiable plant aliphatic polymers (1, 7). Recent studies (17) suggest that soils contain a significant fraction of a substance that acts like polyethylene crystallites with 13C NMR peaks at 32.9 and 31.0 ppm. According to this study, these two peaks correspond to the “rigid all-trans and the mobile gauche-containing (CH2)n units” (17) of a longchain polyethylene-like polymer. While some of this “crystalline” character could derive from normal, lipid-like, and extractable aliphatic materials in SOM, some could be associated with resistant soil biopolymers. The presence of the crystalline and amorphous regions in cuticular materials has a significant impact on their ability to sorb pollutants. Multiple models have been suggested to explain linear and nonlinear sorption behavior of hydrophobic pollutants to SOM (18-22). Simple partition models represent the SOM as an organic phase into which the hydrophobic pollutant dissolves. These models predict linear sorption isotherms and suggest only a single type of average environment in the organic phase that the pollutant recognizes and that pollutant-pollutant interactions are small. Weber and co-workers have suggested the dual reactive domain model (DRDM) to describe nonlinear isotherms. In this model, the hydrophobic pollutant sorbs differently to the “glassy and rubbery domains” of SOM. DRDM is a superposition of Langmuir sorption to the glassy domains and partitioning into the rubbery domains (19). The Freundlich isotherm has also been used to describe nonlinear VOL. 38, NO. 16, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4369
FIGURE 1. Temperature dependence of the 13C CPMAS spectra of A. americana cutan and tomato cutin. (a) Model of cutan showing the rigid all-trans region and mobile gauche-containing region of the aliphatic chain. (b) 13C CPMAS spectra of the aliphatic region of A. americana cutan. The peaks at 31.0 and 32.9 ppm are due to the mobile and crystalline regions of this polymer. These data clearly show the melting of the crystalline region to form the mobile region. (c) Model of the cutin showing its branched polyester structure. (d) Temperature dependence of the 13C CPMAS spectra of tomato cutin. These data show broadening with decreasing temperature but no conversion of one peak into another. sorption behavior. This model can be rationalized in terms of interactions between the sorbed pollutants. Recently, Salloum et al. (23) and Chefetz et al. (24) have demonstrated the ability of plant cuticular matter to sorb a significant amount of hydrophobic organic pollutants. In addition, Mao and co-workers (25) examined the sorption of phenanthrene to various humic substances and found a strong correlation between the sorptive capacity of hydrophobic pollutants and the presence of amorphous aliphatic regions in SOM. Therefore, in this study, we examine the relationships between the physical state of SOM and its sorption properties by investigating two cuticular materials, 4370
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 16, 2004
Agave americana cutan and tomato cutin, and their sorption of pyrene. This connection is examined by variable-temperature, low and high spinning frequency magic angle spinning, and 2D-heteronuclear (1H, 13C), solid-state nuclear magnetic resonance (NMR) spectroscopy. These solid-state NMR techniques allow the determination of the physical state of SOM, the dynamics of sorbed pyrene, and the component of SOM with which the pyrene interacts.
Methods Cutin was extracted from tomato fruit and provided to us by Prof. Karl Espelie. Cutan was obtained from E. Tegelaar who
described its origin and isolation from Agave americana (7). Pyrene sorption was performed as described by Chefetz et al. (26). For completeness, the method will be summarized here again. First, 100-200 mg of cuticular material was mixed with 500 mL of a 0.1 mM solution of HgCl2 (antimicrobial), which was made from HPLC-grade water (J. T. Baker, Phillipsburg, NJ) that had been acidified to a pH of 4.5 using 6 M HCl. Samples were shaken in the dark for 48 h at 150 rpm. After this initial period, 12.5 µL of 1 mg/mL pyrene in methanol was added and allowed to equilibrate for 48 h. This was continued until the pyrene concentration in tomato cutin was 0.65% by mass while A. americana cutan was loaded to 1% by mass. The supernatant was decanted, and the cuticular material was loaded into 50 mL Teflon centrifuge tubes. Next, 50 mL of HPLC grade water was added, the tubes were agitated for 10 min and centrifuged, and the supernatant was discarded. This washing procedure was repeated three times. After being washed, the samples were freeze-dried and stored in the dark until the NMR experiments were performed. This dry storage period varied from within 1 week of freeze-drying to 22 months after freeze-drying. All solid-state cross polarization, magic angle spinning (27-29) (CPMAS) NMR spectra were obtained on either a Bruker DMX300 or a Bruker DMX400 NMR spectrometer operating at a 1H frequency of 300.13 and 400.13 MHz, respectively. This study was performed with a 4 mm highpower CPMAS probe with an extended temperature range and spinning speeds of up to 15 kHz. Spectra were acquired with ramped cross polarization (30, 31) and high power twopulse phase modulated (TPPM) decoupling (32). CP contact times were typically 1 ms, decoupling field strengths were between 80 and 100 kHz, the recycle delay was 1 s for all samples except for crystalline pyrene where it was increased to 20 s, and spinning rates were either approximately 3 kHz for low spinning frequency experiments or greater than 13 kHz for those at high spinning frequencies. Variabletemperature NMR was performed using the Bruker temperature controller and heat exchanger attached to an extended temperature range 4 mm MAS probe. The probe was retuned at every temperature at which a spectrum was measured, and the proton decoupling conditions were checked at the lowest temperature measured and were found not to have changed. The temperature was measured by a thermocouple near the sample and was calibrated to the temperature-dependent chemical shift of solid Pb(NO3)2 (33). We found that temperature gradients of less than 2 K could be obtained if we restricted the sample volume to the central one-third of the 4 mm MAS spinner. When a filled sample spinner was used for variable-temperature measurements, the melting temperature was overestimated by approximately 10 K. 13C longitudinal relaxation times were measured by first cross polarizing the carbon magnetization from proton, inverting this magnetization for a variable time, and then recovering with a 90° pulse. High-frequency MAS is performed when acquiring 1H decoupled 13C NMR spectra of powered samples to remove the chemical shift anisotropy (CSA) and to give NMR spectra that are “liquidlike”, that is, have a single carbon resonance for each magnetically inequivalent 13C in the sample. However, the chemical shift actually depends on the molecular orientation with respect to the applied magnetic field, resulting in broad lines in static solids. This broadening is characteristic of the symmetry of the nuclear environment and thus the motion of the molecule and is characterized by two parameters, the anisotropy, δ, and the asymmetry parameter, η, which can have values between 0 and 1. δ measures the strength of the CSA interaction; larger values of δ result in broader resonance lines in the spectrum of the static solid. η probes the symmetry of the nuclear environment and affects the shape of the static spectral line. If η )
0, the nucleus sits in an axially symmetric environment, while the environment is highly asymmetric if η ) 1. Molecular motion can lead to decreases in the observed values of δ and η by averaging effects. Similarly, high-frequency MAS removes this broadening; however, at lower frequencies, spinning sidebands occur. Spinning sidebands are a series of additional peaks in the NMR spectrum centered on the liquidlike peak observed in the high-frequency MAS spectrum and spaced at integer multiples of the spinning frequency. The intensities of the sidebands allow the determination of δ and η (34, 35). Low spinning frequency magic angle spinning spectra were analyzed using a Bayesian sideband fitting code written in C and running on a UNIX-based SGI O2, a Linex-based PC with a Pentium III processor, or a G3-based MacIntosh iBook. The sideband intensities, order number, and MAS spinning frequency are input, and a contour plot of probability versus δ and η is produced. A peak in this plot indicates that the data fit the simulation with a high probability (36). 2D 1 H-13C solid-state heteronuclear correlation spectra (37) were obtained at 300 K. 2D spectra were obtained with pure phase using the method of States et al. (38). Data were processed on a Power Macintosh G3 or iMac using the program RMN (39).
Results and Discussion Figure 1b shows the variable-temperature 1D NMR spectra of Agave americana cutan in the temperature range of 200400 K. The aliphatic region of the spectrum displays two peaks: a broader peak at 31.0 ppm, attributable to mobile (CH2)n, and a narrower peak at 32.9 ppm, attributable to crystalline (CH2)n. These two peaks make up over 90% of the total observed carbon species, consistent with what has been previously observed (10). Similar to what was seen by Hu et al. (17), as the temperature is increased, the peak at 31.0 ppm is found to narrow, while the peak and 32.9 ppm decreases in intensity. At approximately 360 K, the 32.9 ppm peak completely disappears, while further increases in temperature cause the remaining peak at 31.0 ppm to narrow. This result is consistent with the melting behavior of polyethylene as measured by 13C solid-state NMR. It has been shown that the melting temperature of polyethylene increases with the size of the polymer crystallite (40). The melting temperature in A. americana cutan is 360 K. Differential scanning calorimetry measurements on A. americana cutan by Villena and co-workers show a phase transition at 363 K (41). The melting temperature of cutan is approximately 10 °C higher than that seen in other SOM (17). The previous measurement of Amherst humin (17), albeit a material different from our cuticular samples, gave a melting temperature of 350 K that was consistent with CH2 crystalline domains that have a thickness of 25-30 CH2 units. A. americana cutan is proposed to have an average chain length of greater than 30 units (10). The increased melting temperature of this sample suggests that the thickness of the crystals is approximately 30-40 units. These results demonstrate that there are amorphous and crystalline domains in cutan that have dimensions on the 10-100 nm size scale. 1H inversion recovery experiments by Hu et al. have shown that the humic substances they analyzed showed aliphatic domains in this same size range (17). Mao et al. performed spin-diffusion measurements that demonstrated that the amorphous and crystalline regions are in direct proximity and possibly that the CH2 chains pass through both regions (25). If these domains are truly on the 100 nm size scale, the surface area of the crystalline domains could be quite large (1-10 m2/g). The tomato cuticle 1D 13C CPMAS NMR spectrum (Figure 1d) shows the presence of carbonyl groups from esters at 173.0 ppm; polysaccharides at shifts of 63.1, 64.6, 72.5, 74.5, 84.2, and 105.5 ppm; and aliphatic peaks at 25.7, 29.8, 34.8, VOL. 38, NO. 16, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4371
TABLE 1. 13C T1 of Plant Polymers and Sorbed Pyrene at 25 °Ca 13C
sample pyrene
cutan cutin a
crystalline cutan sorbed cutin sorbed (1.5 months) cutin sorbed (22 months) amorphous peak crystalline peak (biexponential) 1.5 month sample 22 month sample
T1 (s)
80.1 0.68 0.56 0.16 0.23 0.45 and 6.1 0.27 0.19
The errors in these values are estimated to be (10%.
and 38.3 ppm. The peaks at 64 and 72 ppm also have contributions from primary and secondary carbons neighboring ester oxygens of cutin (42). We tested whether these aliphatic peaks undergo the same behavior as the 31.0 and 32.9 ppm peaks of cutan. Variable-temperature NMR spectra from 180 to 350 K (Figure 1d) show that the aliphatic peaks simply narrow with increasing temperature and do not undergo the melting behavior seen in cutan. The broadening of the lines with decreasing temperature is due to motional interference with decoupling (43). The continuous change of the aliphatic peak line widths indicates that there is a slow, continuous freezing out of motions in the cutin. The polysaccharide line widths remain roughly constant, suggesting that a similar transition is not occurring in this component of the cuticular material.
13 C longitudinal relaxation time (T1) measurements were acquired on the cuticular samples to further probe motional regimes of the aliphatic moieties. Previous measurements of 13C and 1H T ’s and T ’s have indicated that lime cutin is 1 1F very mobile (42, 44). Table 1 shows the results of our 13C T1 measurements. The longitudinal relaxation time of tomato cutin is short, 0.27 and 0.19 s for the two samples examined, consistent with the previously published lime cutin data. The peak from the amorphous region of A. americana cutan has a T1 of 0.23 s, indicating the highly mobile nature of this region. The peak at 32.9 ppm shows biexponential recovery with time constants of 0.45 and 6.1 s. At this shift, a small portion of the amorphous peak overlaps the much sharper crystalline peak, suggesting that the 0.45 s time constant corresponds to the amorphous region while the 6.1 s T1 is that for the crystalline region. Thus, the crystalline region of cutan is much less mobile than the amorphous region.
These observations are consistent with Weber’s DRDM for sorption isotherms and the observance of phase changes in differential scanning calorimetry (DSC) (45). We directly see slowly relaxing and quickly relaxing domains in these two cuticular materials, which we interpret as corresponding to the “rubbery and glassy domains” postulated by DRDM. In the case of cutan, however, the portion of the sample that would be associated with the term glassy domain in DRDM is actually an all-trans crystalline domain of the polyethylene chains, while the rubbery domain is the mobile gauchecontaining region. The crystalline domain should show Langmuir-type surface adsorption onto the surface of the
FIGURE 2. 13C CPMAS spectra of crystalline pyrene. (a) CPMAS spectrum of crystalline 1,6 13C-labeled pyrene with a spinning frequency of 13 kHz. This sample shows two peaks at 125 and 127 ppm. (b) Low spinning frequency (2.5 kHz) of the same sample with the Baysian analysis and simulation of the 125 ppm sideband pattern. (c) Same as b but with Baysian analysis and simulation of the 127 ppm peak. The errors reported for the two sites are 95% confidence intervals. 4372
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 16, 2004
nanocrystals, while the mobile region should show linear partitioning of the pyrene. The cutin sample clearly shows a transition from a mobile rubbery phase to a true glassy phase as temperature is decreased, as indicated by the change in line width of the carbon spectra. Unfortunately, the current data do not allow an estimation of the glass transition temperature, but the spectra indicate it is probably below 275 K. There is not a mixture of these two phases at room temperature, however. DRDM predicts that the sorption behavior of these two samples should be considerably different. 1,6 13C-pyrene was sorbed to the cutan and cutin samples. Figure 2a shows the 13 kHz MAS spectrum of crystalline pyrene, which shows two peaks at 125 and 127 ppm, while Figure 2b and c shows its low spinning frequency (2.5 kHz) MAS spectra with the simulated spectra and posterior probability functions. From these data, we find that the CSA parameters for the peak at 125 ppm are δ ) -97.4 ( 0.5 ppm, η ) 0.934 ( 0.006 and for the peak at 127 ppm are δ ) -98.1 ( 0.5 ppm, η ) 0.823 ( 0.008, with 95% confidence. The 13C T1 of this sample, Table 1, was measured to be 80.1 s, indicating that the pyrene is unable to move in its crystalline solid. The values of the CSA parameters are their full, unaveraged values to which the sorbed parameters will be compared. Figure 3 shows the MAS spectra for cutan with sorbed pyrene analyzed 2 months after completion of the sorption procedure. The peak at 127 ppm, marked with a star in Figure 3a, is due to the 13C-labeled sites of the pyrene. Figure 3b shows the same sample as that in Figure 3a, but the spinning frequency has been reduced to 2.85 kHz. A clear pattern of spinning sidebands is visible. Bayesian analysis of this sideband pattern leads to the simulation in Figure 3c and the posterior probability function in Figure 3d. In pyrene sorbed to A. americana cutan, the CSA parameters are δ ) -79.8 ( 5.2 ppm, η < 0.70, showing that the magnitude of the anisotropy, |δ|, and the asymmetry parameter, η, have decreased significantly from that seen in crystalline pyrene. The centerband intensity is overestimated when the spectrum is simulated using the most probable values, Figure 3c, indicating either exchange or a distribution of CSA parameters for the sorbed pyrene. T1 of the sorbed pyrene has decreased to 0.68 s, indicating that the motion of the pyrene increases significantly upon sorption. Assuming a simple two-spin dipolar model for T1 (46), we estimate that the correlation time for the motion of pyrene in the crystalline sample is at least 10-4 s, while after sorption, this correlation time reduces to either 10-6 or 10-10 s depending upon whether it is on the slow motion or extreme narrowing side of the T1 minimum. This indicates that the pyrene is much more free to move once sorbed to the cutan; however, the values of the CSA parameters indicate that this motion is highly anisotropic, leading to incomplete averaging of the CSA parameters. The peak at 125 ppm in Figure 4a shows pyrene sorbed to tomato cutin. Interestingly, the slow spinning spectra of this sample depend on the length of time that the pyrene is sorbed to the cutin, Figure 4b and d. When the slow spinning spectra are taken within a month and a half of sorption, few, if any, sidebands are observed, Figure 4b. Analysis of this spectrum, Figure 4f, indicates that the anisotropy has a value between -30.4 and 0.0 ppm. The asymmetry parameter is undeterminable from this spectrum. The 13C longitudinal relaxation time of this sample is 0.56 s, similar to that seen for pyrene sorbed to the cutan sample. Twenty-two months after completion of sorption, the slow spinning spectrum changes markedly. Sidebands are now clearly visible in Figure 4d, and Bayesian analysis, Figure 4g, indicates that δ ) -80.0 ( 3.3 ppm and η < 0.42. The 13C T1 experiment of the 22month stored sample shows a single-exponential magnetization recovery with a T1 of 0.16 s, significantly shorter than
FIGURE 3. 13C CPMAS spectra of pyrene sorbed to cutan. (a) 13 kHz CPMAS spectrum of pyrene sorbed to cutan. The peak at 127 ppm (marked with *) is due to the 13C-labeled pyrene. (b) Low spinning frequency (2.85 kHz) CPMAS spectrum of pyrene sorbed to cutan. Peaks marked with asterisks are the spinning sideband pattern of the pyrene. (c) Simulation of the sideband pattern using the most probable values of the CSA parameters. (d) Posterior probability function for the CSA parameters to describe the sideband patterns shown in (b). The magnitudes of δ and η significantly decrease upon sorption to the cutan. that seen in the other sorbed samples. This behavior is inconsistent with pyrene crystallization during the storage period, as crystallization would lead to an increase of T1 to 80 s and the mixed sample would show a distinct biexponential recovery. These experiments indicate that after aging the pyrene motion becomes restricted, leading to a lower probability of wobbling of the plane of the ring than in the freshly sorbed samples. η is significantly reduced from that seen in crystalline pyrene, indicating rotation of the molecule about the axis perpendicular to the plane of pyrene. Heteronuclear correlation experiments (37) with and without proton spin diffusion were used to demonstrate that the pyrene was sorbed to the aliphatic plant polymers. Figure 5a shows the heteronuclear correlation experiment without proton spin diffusion as applied to pyrene sorbed to cutan. Shifts are reported in Table 2 as (1H shift in ppm, 13C shift in ppm). The amorphous and crystalline regions of the cutan show peaks at (30.3 ppm, 1.3 ppm) and (32.8 ppm, 1.3 ppm), respectively. These peaks indicate that the protons of the aliphatic chains are responsible for cross polarizing their attached carbons. The peak at (127.3 ppm, 7.8 ppm) is that due to the proton of pyrene cross polarizing its attached carbon. In Figure 5b, the heteronuclear correlation experiment with 100 ms of 1H spin diffusion was run on the same VOL. 38, NO. 16, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4373
FIGURE 4. 13C CPMAS of pyrene sorbed to tomato cutin. (a) 14.0 kHz spinning frequency CPMAS spectrum of pyrene sorbed to cutin. The peak marked with an asterisk is the peak from the pyrene. (b) 3.5 kHz spinning frequency CPMAS spectrum of pyrene sorbed to cutin after 1.5 months. The centerband of the pyrene peak is marked with an asterisk; however, spinning sidebands are not evident in this spectrum. (c) Simulation of the data in (b) using the most probable values of δ and η. (d) 2.6 kHz spinning frequency CPMAS spectrum 22 months after sorption of pyrene to cutin. Spinning sidebands are evident, indicating the presence of CSA. (e) Simulation of the data in (d) using the most probable values of δ and η. (f) Posterior probability distribution for 1.5 months after sorption data. (g) Probability distribution for 22 months after sorption data. 4374
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 16, 2004
sample. In this experiment, after proton frequency encoding but before 13C cross polarization, the 1H magnetization is stored along the external magnetic field. This allows energy conserving flip-flops of the protons (spin diffusion), allowing the encoded magnetization to move to another type of proton. Short contact time cross polarization then transfers the magnetization to the attached carbon, allowing identification of the site to which the magnetization has spin diffused. In this spectrum, the aliphatic peaks remain unchanged; however, the peak at the pyrene carbon frequency is now correlated to the 1H of the aliphatics (127.3 ppm, 1.3 ppm). This demonstrates that the pyrene is intimately associated with the aliphatic regions of the cutan. Figure 5 also shows similar spectra for pyrene sorbed to tomato cutin. Figure 5c and d shows the spectra without and with 1H spin diffusion, respectively. The heteronuclear peaks, for these two experiments, are given in Table 3. Significantly, the proton frequencies of the aliphatic, polysaccharide, and pyrene components of the sample are well resolved. After allowing 100 ms of proton spin diffusion, only two significant changes occur. The peak at (127.3 ppm, 7.8 ppm) is that due to pyrene, which, in the spin diffusion experiment, gives a peak at (127.3 ppm, 1.3 ppm), indicating that the pyrene protons exchange magnetization with the aliphatic protons of the cutin and not the protons of the polysaccharides. This clearly indicates that the pyrene is primarily interacting with the cutin and not the polysaccharides. Previous measurements show that the relative amounts of polysaccharide and aliphatic carbon are about equal, each making up 45% of the total carbon (26). Our results clearly demonstrate pyrene’s preference for the more hydrophobic aliphatic region. This might not seem surprising considering the makeup of our cutin sample (e.g., aliphatic and carbohydrate); however, previous studies (26) demonstrated that extremely high KOC values are observed in cutin. The spin diffusion experiment also clearly shows that the carbonyl peak is associated with two types of distant protons, (172.3 ppm, 1.3 ppm) and (172.3 ppm, 4.6 ppm). The first of these cross-peaks is due to the interaction of the methylene groups of the cutin with its ester carbonyl carbon. The second indicates the free or esterified carbonyls of polysaccharides in the form of pectins. These experiments probe the proximity of pyrene to cuticular materials. The heteronuclear correlation spectra demonstrate that the pyrene is preferentially in close proximity to aliphatic biopolymers rather than polysaccharides. This is important for interpreting the slow spinning MAS spectra because it indicates that the measured anisotropic motions are caused by the interactions between the pyrene and the aliphatic cuticular material. In freshly sorbed pyrene to cutin, the interactions between the pyrene and the biopolymer are small, allowing nearly isotropic motion of the sorbed pyrene molecule. After a prolong period after sorption, the pyrene molecules undergo anisotropic motion and either a decrease in correlation time or an increased dipolar coupling to the cutin due to the smaller void space. The T1 of the methylene groups of cutin decreases slightly but not significantly over the 20.5-month period, implying that the change in motion of the pollutant is unrelated to the motions of the biopolymer. We suggest that during prolonged sorption, the pyrene migrates into smaller, more restricted voids in the cutin, limiting its motion and directly resulting in its anisotropic motion. The published isotherm for this sorption (26) is linear but was only equilibrated for 1 week between pyrene additions. Our current measurements predict that hysteresis will be observed in desorption experiments. Once pyrene migrates into one of these smaller void regions, desorption is slower and with a different equilibrium constant than from the larger voids. These data also raise significant questions about the time scale necessary to reach true equilibrium in a sorption experiment. Diffusion within the
FIGURE 5. 2D 1H-13C heteronuclear correlation spectra of sorbed pyrene. (a) 2D spectrum of pyrene sorbed to cutan taken without 1H spin diffusion. Conditions were chosen such that peaks only occur between directly bonded hydrogens and carbon. (b) Same as (a), but with 100 ms of 1H spin diffusion. 13C peak of pyrene is now correlated to the aliphatic 1H values of cutan, indicating that pyrene and cutan are intimately mixed. (c) 2D spectrum of pyrene sorbed to cutin taken without 1H spin diffusion. (d) Same as (c), but with 100 ms of 1H spin diffusion. 13C peak of pyrene is correlated to the aliphatic 1H values of cutin and not cell wall polysaccharides, indicating that pyrene is in close proximity to cutin and not the cell wall polysaccharides.
TABLE 2. 1H-13C Heteronuclear Correlation Results on Cutan Sorbed Pyrene
TABLE 3. 1H-13C Heteronuclear Correlation Results on Cutin Sorbed Pyrene
peak position (13C ppm, 1H ppm)
peak position (13C ppm, 1H ppm)
Cutan and Pyrene with No Spin Diffusion (from Figure 5a) (30.3 ppm, 1.3 ppm) (32.8 ppm, 1.3 ppm) (127.3 ppm, 7.8 ppm)
Cutin and Pyrene with No Spin Diffusion (from Figure 5c) (25.6 ppm, 1.3 ppm) (29.6 ppm, 1.3 ppm) (33.9 ppm, 1.3 ppm) (172.6 ppm, 2.6 ppm) (62.3 ppm, 3.9 ppm) (63.7 ppm, 3.9 ppm) (72.3 ppm, 4.6 ppm) (104.0 ppm, 3.9 ppm) (127.3 ppm, 7.8 ppm)
Cutan and Pyrene with 100 ms of Spin Diffusion (from Figure 5b) (30.3 ppm, 1.3 ppm) (32.8 ppm, 1.3 ppm) (127.3 ppm, 1.3 ppm)
cutin to the small void sites is on at least the month to year time scale. The behavior of cutan is surprisingly different from that of cutin. A bulk sample of cutan is composed of two phases of methylene groups, rigid crystalline and mobile amorphous; however, there does not seem to be two distinct types of sorbed pyrene molecules as would be expected. This is explained by pyrene having a much larger equilibrium constant for sorption to one of the two phases. Unfortunately, the heteronuclear correlation experiments do not have high enough 1H resolution to distinguish which phase is preferred. Loading cutan to higher levels might lead to populating the other phase. It is also interesting that pyrene sorbed to cutan is undergoing anisotropic motion independent of time from 2 to 22 months. Pyrene quickly finds a preferred environment in cutan and remains in it over the time scale studied. The physical interactions between hydrophobic pollutants and SOM are important for determining sorption behavior
Cutin and Pyrene with 100 ms of Spin Diffusion (from Figure 5d) (25.6 ppm, 1.3 ppm) (29.6 ppm, 1.3 ppm) (34.2 ppm, 1.3 ppm) (172.6 ppm, 1.3 ppm) (172.6 ppm, 4.6 ppm) (62.3 ppm, 4.6 ppm) (63.7 ppm, 4.6 ppm) (72.3 ppm, 4.6 ppm) (104.3 ppm, 4.6 ppm) (127.3 ppm, 1.3 ppm)
and understanding the basis of sorption isotherm behavior. These results establish constraints on sorption and desorption isotherm models and potentially allow the development of improved models. Such improved models are important for understanding the environmental effects of SOM uptake and release of hydrophobic pollutants and will allow the development of improved remediation techniques. VOL. 38, NO. 16, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4375
Acknowledgments We thank Karl Espelie (University of Georgia, Athens) and Jan de Leeuw (NIOZ, Texel, The Netherlands) for samples of tomato cutin and A. americana cutan, respectively. Dr. Jack Richman (University of Minnesota, St. Paul, MN) kindly provided the 13C-labeled pyrene. We would also like to thank R. E. Stark for stimulating discussions about the NMR of cuticular materials. This work was supported by the NSF through the Ohio State Environmental Molecular Science Institute Grant CHE0089147.
Literature Cited (1) Kogel-Knabner, I.; Hatcher, P. G.; Tegelaar, E. W.; de Leeuw, J. W. Sci. Total Environ. 1992, 113, 89. (2) Stevenson, F. J. Humus Chemistry; John Wiley & Sons Inc.: New York, 1994. (3) Zech, W.; Senesi, N.; Guggenberger, G.; Kaiser, K.; Lehmann, J.; Miano, T. M.; Miltner, A.; Schroth, G. Geoderma 1997, 79, 117. (4) Chefetz, B.; Salloum, M. J.; Desmukh, A. P.; Hatcher, P. G. Soil Sci. Soc. Am. J. 2002, 66, 1159. (5) Chefetz, B.; Tarchitzky, J.; Deshmukh, A. P.; Hatcher, P. G.; Chen, Y. Soil Sci. Soc. Am. J. 2002, 66, 129. (6) Nierop, K. G. J. Org. Geochem. 1998, 29, 1009. (7) Tegelaar, E. W.; de Leeuw, J. W.; Largeau, C.; Derenne, S.; Schulten, H.-R.; Mu ¨ ller, R.; Boon, J. J.; Nip, M.; Sprenkels, J. C. M. J. Anal. Appl. Pyrolysis 1989, 15, 29. (8) Holloway, P. J. Pestic. Sci. 1993, 37, 203. (9) Jeffree, C. E. Plant Cuticle: An Integrated Functional Approach; Environmental Plant Biology Series; Bios Scientific Publishers Ltd.: Oxford, 1996; p 33. (10) McKinney, D. E.; Bortiantynski, J. M.; Carson, D. M.; Clifford, D. J.; de Leeuw, J. W.; Hatcher, P. G. Org. Geochem. 1996, 24, 641. (11) Alemendros, G.; Guadalix, M. E.; Gonzlez-Vila, F. J.; Martin, F. Org. Geochem. 1996, 24, 651. (12) Lichtfouse, E.; Leblond, C.; da Silva, M.; Behar, F. Naturwissenschaften 1998, 85, 497. (13) Hatcher, P. G.; Breger, I. A.; Dennis, L. W.; Maciel, G. E. Aquatic and Terrestrial Humic Material; Ann Arbor Science: Ann Arbor, MI, 1983; p 37. (14) Kogel-Knabner, I. Geoderma 1997, 80, 243. (15) Wilson, M. A. NMR Techniques and Applications in Geochemistry and Soil Chemistry; Pergamon Press: Oxford, 1987. (16) Ziegler, F.; Zech, W. Z. Pflanzenernaehr. Bodenkd. 1989, 152, 287. (17) Hu, W.-G.; Mao, J.; Xing, B.; Schmidt-Rohr, K. Environ. Sci. Technol. 2000, 34, 530. (18) Huang, W.; Weber, W. J. Environ. Sci. Technol. 1997, 31, 2562.
4376
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 16, 2004
(19) Huang, W.; Young, T. M.; Schlautman, M. A.; Yu, H.; Weber, W. J. Environ. Sci. Technol. 1997, 31, 1703. (20) LeBoeuf, E. J. Environ. Sci. Technol. 2000, 34, 3632. (21) Winget, P.; Cramer, C. J.; Truhlar, D. G. Environ. Sci. Technol. 2000, 34, 4733. (22) Weber, W. J.; LeBoeuf, E. J.; Young, T. M.; Huang, W. Water Res. 2001, 35, 853. (23) Salloum, M. J.; Chefetz, B.; Hatcher, P. G. Environ. Sci. Technol. 2002, 36, 1953. (24) Chefetz, B. Environ. Toxicol. Chem. 2003, 22, 2492. (25) Mao, J.-D.; Hundal, L. S.; Thompson, M. L.; Schmidt-Rohr, K. Environ. Sci. Technol. 2002, 36, 929. (26) Chefetz, B.; Deshmukh, A.; Hatcher, P. G.; Guthrie, E. A. Environ. Sci. Technol. 2000, 34, 2925. (27) Andrew, E. R.; Bradbury, A.; Eades, R. G. Nature 1958, 182, 1659. (28) Pines, A.; Gibby, M. G.; Waugh, J. S. J. Chem. Phys. 1973, 59, 569. (29) Stejskal, E. O.; Schaefer, J.; Waugh, J. S. J. Magn. Reson. 1977, 28, 105. (30) Hediger, S.; Meier, B. H.; Kurur, N. D.; Bodenhausen, G.; Ernst, R. R. Chem. Phys. Lett. 1994, 223, 283. (31) Metz, G.; Wu, X.; Smith, S. O. J. Magn. Reson., Ser. A 1994, 110, 219. (32) Bennett, A. E.; Rienstra, C. M.; Auger, M.; Lakshmi, K. V.; Griffin, R. G. J. Chem. Phys. 1995, 103, 6951. (33) Bielecki, A.; Burum, D. P. J. Magn. Reson., Ser. A 1995, 116, 215. (34) Maricq, M. M.; Waugh, J. S. J. Chem. Phys. 1979, 70, 3300. (35) Herzfeld, J.; Berger, A. E. J. Chem. Phys. 1980, 73, 6021. (36) Sachleben, J. R. Manuscript in preparation. (37) Ernst, R. R.; Bodenhausen, G.; Wokaun, A. Principles of Nuclear Magnetic Resonance in One and Two Dimensions; Oxford University Press: Oxford, 1990. (38) States, D. J.; Haberkorn, R. A.; Ruben, D. J. J. Magn. Reson. 1982, 48, 286. (39) Courtesy of Professor P. Grandinetti, Department of Chemistry, The Ohio State University. (40) Wunderlich, B. Macromolecular Physics; Academic Press: New York, 1980. (41) Villena, J. F.; Dominguez, E.; Stewart, D.; Heredia, A. Planta 1999, 208, 181. (42) Zlotnik-Mazori, T.; Stark, R. E. Macromolecules 1988, 21, 2412. (43) Mehring, M. High-Resolution NMR Spectroscopy in Solids, 2nd ed.; Springer: Berlin, 1983. (44) Garbow, J. R.; Stark, R. E. Macromolecules 1990, 23, 2814. (45) LeBoeuf, E. J.; Weber, W. J. Environ. Sci. Technol. 2000, 34, 3623. (46) Harris, R. K. Nuclear Magnetic Resonance Spectroscopy; Longman Scientific and Technical: Essex, 1986.
Received for review December 5, 2003. Revised manuscript received May 11, 2004. Accepted June 3, 2004. ES035362W
