Environ. Sci. Technol. 2009, 43, 8264–8269
New Option for Characterizing the Mobility of Organic Compounds in Humic Acids K H A N N E H W A D I N G A F O M B A , * ,†,§ PETRIK GALVOSAS,‡ ULF ROLAND,† ¨ RG KA ¨ RGER,§ AND JO FRANK-DIETER KOPINKE† Department of Environmental Technology, UFZ - Helmholtz Centre for Environmental Research, Permoserstrasse 15, 04318 Leipzig, Germany, School of Chemical and Physical Sciences, Victoria University of Wellington, P.O. Box 600, Wellington 6140, New Zealand, and Department of Interface Physics, University of Leipzig, Linne´strasse 5, 04109 Leipzig, Germany
Received May 6, 2009. Revised manuscript received August 11, 2009. Accepted September 1, 2009.
A new NMR option for monitoring the mobility of organic contaminants in SOM in the solid state has been successfully applied for the first time. This recently available noninvasive technique, magic angle spinning pulsed-field gradient (MAS PFG) NMR, combines both NMR spectroscopy and diffusometry to selectively monitor the diffusion of compounds sorbed in porous media or polymer matrices. Using this technique, the diffusion of toluene in humic acid particles has been studied. Measurements were performed under varying temperatures from 25 to 80 °C. The obtained diffusion coefficients were found to be in good agreement with those obtained from computer simulations reported elsewhere. Our results show a strong influence of the interaction of toluene with humic acid on its diffusion in the matrix even at elevated temperatures of up to 80 °C. The Arrhenius plot of the diffusivities shows a decrease in the activation energy of diffusion above 50 °C by a factor of 3. This change of activation energy is attributed to a structural change in the humic acid matrix that influences the mobility of toluene.
Introduction Soil organic matter (SOM) is known to be relevant for a variety of environmental processes including the adsorption, immobilization and transport of anthropogenic organic chemicals and heavy metals in the soil, often representing hazardous pollutants. Its ability to bind hydrophobic organic contaminants (HOCs) due to its high carbon content, large surface area, and porous nature makes the understanding of processes within this matrix essential in evaluating and predicting the fate of contaminants in the environment, especially in the soil. One of the limitations to effective soil remediation is usually attributed to the slow release of organic contaminants from the soil. Processes related to this slow release include their diffusion in the micropores of soil particles (1, 2), their intraparticle diffusion in SOM (3, 4), and * Corresponding author e-mail:
[email protected], fomba@ tropos.de. † UFZ - Helmholtz Centre for Environmental Research. § University of Leipzig. ‡ Victoria University of Wellington. 8264
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 21, 2009
their slow migration inside the SOM matrix (5). Consequently, the understanding of the mechanism of the diffusion processes inside SOM under various temperature and humidity conditions is of great importance for optimizing remediation techniques including thermally supported methods (6). Until now, most investigations on the diffusion of contaminants in SOM have been carried out using macroscopic techniques, e.g., gravimetric analysis. With respect to these techniques, the application of sorption kinetic methods has widely been used (7-10). Although these methods are very useful to characterize the rate-limiting processes during uptake or release of contaminants from the soil, they are not able to provide information about the dynamic transport processes in steady state, i.e., after sorption equilibrium is achieved. Information about transport processes under (macroscopic) equilibrium conditions, related to the self-diffusion of these contaminants is also essential for understanding the rate-limiting steps in environmental processes like the bioaccessibility and bioavailability of these contaminants for microbial degradation. Hence, developing methods suitable for monitoring self-diffusion processes of contaminants in SOM is of great significance. So far, few experimental methods have been applied to monitor this process in a noninvasive manner. This explains the limited information on selfdiffusion processes of contaminants in SOM in the solid state. The implementation of the NMR technique on natural systems as presented in this paper is one option to reduce this scarcity. Due to the fact that the nuclear spins of the adsorbed contaminant molecules in SOM are sensitive to their chemical environment, expressed by the chemical shift, NMR spectroscopy and other NMR methods can be employed to study the interactions (11) as well as the diffusivities (12) of these contaminants in a noninvasive way. Nevertheless, because of the strong sorptive interaction between contaminant molecules and SOM leading to restricted mobility, the NMR signals of the adsorbed species are usually very broad (13, 14), and exhibit very short transverse relaxation times (T2). This renders the application of NMR-based diffusometry methods, such as conventional pulsed-field gradient (PFG) NMR, successfully applied to other systems (15), difficult. Although much has been done in solution state NMR to enhance the relaxation time and resolution of the signals for suitable diffusion measurements in similar systems as, for example, shown by Steinbeck et al. (16), such applications on solid state systems are far from common. However, by aligning pulsed magnetic field gradients along the magic angle of spinning, it is possible to combine both high-resolution NMR spectroscopy and diffusometry (17, 18). The method presented in this work as discussed in the works of Pampel et al. (17) shows great prospects in resolving such difficulties in soil systems. This so-called MAS PFG NMR (magic angle spinning pulsed-field gradient) is able to enhance the spectral resolution of the NMR signal, decrease T2 relaxation, and thereby increase the time interval during which magnetic pulsed-field gradients can be applied (19). This facilitates the measurement of diffusivities in systems that are hardly accessible by standard NMR techniques without sample spinning. This technique is capable of monitoring the mean squared displacements of compounds sorbed in a (porous) media selectively via their chemical shifts, within a given observation time, from which their diffusivities can be determined. Applying this process, it has been possible to follow molecular displacements from 10.1021/es901358s CCC: $40.75
2009 American Chemical Society
Published on Web 09/29/2009
some hundreds of nanometres up to hundreds of micrometers (ref 20 and references therein). In this study, the effect of temperature on the diffusion of sorbed toluene (as a contaminant probe compound) in humic acid is presented, and the difficulties in the application of this method and its limits with respect to the range of diffusivities that can be determined will be discussed. Furthermore, the application of this method to a system with relatively low probe compound concentration in humic acid is shown.
Experimental Section Chemicals and Sample Preparation. The humic acid and toluene were obtained from Sigma-Aldrich GmbH, Germany. The Aldrich humic acid (ALD-HA) was obtained as a sodium salt and was purified by the standard procedure stipulated by the International Humic Substance Society (IHSS). About 50 g of ALD-HA sodium salt was suspended in 250 mL of 0.1 M HCl/0.1 M HF. The suspension was stirred overnight, centrifuged, and the supernatant was decanted. The obtained solid matter was repeatedly rinsed with distilled water until the AgNO3 test did not detect the presence of residual chloride ions in the wash solution. The obtained humic acid was airdried and stored in closed vials. For the preparation of NMR samples, the humic acid was oven-dried at approximately 95 °C for about 12 h. After heating, the humic acid was quickly filled into a vial in order to avoid readsorption of water. About 15 mg of the ovendried humic acid was filled into a 3.8 mm NMR Pyrex glass tube. Thereafter, about 0.68 mg of toluene (corresponding to approximately 4.5 wt.-%) was injected into the humic acid. The sample tube was cooled in dry ice and sealed. The sealed sample was weighed and held at 100 °C overnight to enable a homogeneous distribution of toluene in the sample. For the first set of investigated samples, the sorbate concentration used was high in comparison to natural contaminant concentrations but for the sake of validation of the method, these concentrations were chosen. In the second set of samples, lower concentrations were used. About 140 µg of toluene was sorbed in a 36 mg pressed ALD-HA disk via the gas phase using a thermogravimetry analyzer (TGA 50A from Shimadzu, Japan) corresponding to approximately 0.4 wt.-% toluene in ALD-HA. After sorption equilibrium was achieved, about 18 mg of the toluene loaded ALD-HA was filled into a 3.8 mm Pyrex glass tube and sealed. NMR Spectroscopy and MAS PFG Experiments. All 1H NMR spectra were acquired using a Bruker AVANCE spectrometer operating at 750 MHz. A Bruker 4-mm MAS probe, with magnetic pulsed-field gradient capabilities along the direction of the magic angle that can support maximum field gradient amplitudes (gmax) of 0.54 T/m was used. The upper temperature limit of this probe was 80 °C. The typical 90° radio frequency (rf) pulse length was 2.5 µs. MAS PFG NMR data were collected with a stimulated-echo pulse sequence with bipolar sine-shaped gradient pulses and an eddy current delay before signal acquisition according to Wu et al. (21), as graphically illustrated in the Supporting Information (SI). Using this sequence, the obtained NMR signal attenuation (Ψ) for isotropic diffusion is related to the signal intensity (S) and the diffusion coefficient according to ψ ) S/S0 ) exp(-bD) where b )
δ+τ ∆( 4δgγ π )( 2 δ - p ) (1) 6 2
π
where D denotes the self-diffusion coefficient of the component under study, S0 is the intensity at g ) 0, ∆ is the observation time, δ and g are the gradient pulse duration and amplitude, respectively. pπ is the duration of the 180° rf
FIGURE 1. (a) Static and (b) 10 kHz 1H MAS NMR spectra of toluene (4.5 wt.-%) in dry humic acid at 25 °C showing a dramatic increase in spectra resolution due to spinning. pulse, γ is the gyromagnetic ratio of protons (1H), and τ is the delay after each gradient pulse and the subsequent rf pulse which was set at 500 µs for these measurements. Diffusion experiments were carried out by varying the gradient amplitude between 5 and 95% of its maximum value (usually in 12 increments) while keeping all other parameters constant. From eq 1, the diffusivity is evaluated from the slope of the plot of ln S/S0 vs b, where S is obtained from the integral over the spectral line “peak area” under consideration. 1 H MAS NMR spectra were acquired using 256 scans and a delay of 4 s between pulses using 4096 data points. The chemical shift scale was calibrated using a poly(dimethylsiloxane) (PDMS) sample. The MAS PFG NMR spectra were collected using 256 or 1024 scans (depending on the toluene concentration in the sample). A diffusion time (∆) of 50 ms, sinusoidal gradient pulse durations (δ) of 2 and 2.3 ms were used to investigate the mobility of toluene in the 4.5 wt.-% and 0.4 wt.-% samples, respectively. Further details about the NMR system used for these experiments can be found in the literature (20).
Results and Discussion In addition to the already mentioned limitations of line broadening of contaminant signals and their short transverse relaxation times (T2), 1H NMR signals of humic acids in the solid state even under MAS conditions are very broad because spinning (even at 10 kHz) is usually not enough to completely average out all interactions leading to line broadening. For instance, see the 10 kHz 1H MAS NMR signal of humic acid in Figure 1. Unfortunately, most contaminant probe compounds possess chemical shifts within the same regions leading to an overlap of the probe compound and the matrix signals, which is one of the reasons most researchers prefer working with deuterated or fluorinated solvents (22). Sample spinning, however, significantly decreases the signal broadening, thus, increasing the T2 of probe molecules in the matrix. Although chemical shifts expand even into negative numbers due to remaining line broadening, the smallest central position of any peak lies at 0.4 ppm. Figure 1 shows both static and 10 kHz high-resolution 1H MAS NMR spectra of toluene in ALD-HA at 25 °C. From the spectra, a significant decrease in the spectral line width of both toluene and humic acid signals due to spinning and an enhancement in the spectral resolution of toluene signals can be observed. In particular, the aromatic signal of toluene at about 6.7 ppm can be clearly identified. Although ALDHA does contribute to the aromatic signal, its lines are usually very broad even under MAS conditions and its visibility VOL. 43, NO. 21, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8265
FIGURE 2. High-resolution 1H NMR PFG spectrum of toluene, 1H MAS PFG NMR spectra of ALD-HA and toluene (4.5 wt.-%) in dry ALD-HA, obtained with a T1 filter(∆) of 300 ms and a T2 filter (δ) of 2 ms. (compared to the strong toluene signal in the same chemical shift region) therefore remains very poor. Despite the aforementioned enhancement, the toluene signals still overlap with the humic acid signals. In order to minimize the overlapping of the signals and to further increase the resolution of the toluene signals, a relaxation time filter (either T2 or T1 filter) was used. Applying a relaxation time filter leads to the filtering out of signals having very short T2 values, i.e., broad spectral line width. Since the acquired PFG NMR signals are usually relaxation time weighted, such T2 and T1 filters are intrinsic parts of the NMR pulse sequence used for performing MAS PFG NMR experiments. They are incorporated through the implementation of the gradient pulse duration time δ (T2 filter) and the diffusion observation time ∆ (T1 filter). Thus, signals acquired during a MAS PFG NMR experiment have relaxation times longer than those set by the relaxation time filters. Figure 2 shows high-resolution 1H MAS PFG NMR spectra of bulk toluene, ALD-HA, and toluene in ALD-HA acquired with a T1 filter (∆) of 300 ms and T2 filter δ + τ ) 2.5 ms. Due to filtering, the spectrum of toluene in ALD-HA reveals an enhancement in the spectral resolution (i.e., a better identification) of the aromatic and methyl group signals of toluene and some aliphatic domains signals of humic acid. Although the ALD-HA spectrum still shows a very small signal around 1.7 ppm, which may overlap with the toluene methyl group signal, this overlap is not significant in the evaluation of the diffusivity of toluene since the aromatic groups are distinctly observed. However, a good identification of both toluene signals is needed. By filtering out humic acid signals having very short relaxation times as compared to that of toluene, it is possible to perform highly resolved 1H MAS PFG NMR experiments on toluene in ALD-HA, with a satisfactory separation between the toluene and humic acid signals. It is worth mentioning that although filtering leads to loss of information of some structural components of humic acid and some strongly bound toluene fraction in humic acid, in this case, it is necessary for a better identification of the signals originating from toluene diffusing in the ALD-HA matrix. This is crucial for carrying out accurate 1 H MAS PFG NMR measurements. The humic acid spectrum shows that only the aliphatic chains (between 0.5 and 1.9 ppm) of humic acid have 8266
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 21, 2009
relaxation times longer than those set by the filters. This suggests that contrary to the postulate of Kahlaf et al. (13), but in line with those of Mao et al. (23, 24), the most flexible domains in ALD-HA are the aliphatic domains. In accordance with Mao et al. (23, 24), and in our opinion, this finding clarifies somewhat the debatable question of which structural domains for ALD-HA are responsible for the postulated condensed and flexible regions of humic substances. Diffusion of Bulk Toluene and of Toluene in Humic Acid. 1 H PFG NMR and high-resolution 1H MAS PFG NMR measurements have been performed for temperatures between 25 and 80 °C, on bulk toluene and on about 4.5 wt.-% toluene in ALD-HA, respectively. Figure 3a shows a stack plot of 1H MAS PFG NMR spectra of toluene in ALD-HA with increasing gradient amplitude and hence gradient strength b at 70 °C, revealing a signal attenuation of the peaks with increasing gradient strength. In Figure 3b, the corresponding signal intensity of the toluene aromatic group (i.e., area under the 6... 7 ppm peaks) with increasing b within this temperature range is shown. Usually, after taking into consideration the small overlap of the matrix signal with the toluene methyl group signal, similar signal attenuation is obtained for the methyl and the aromatic groups as expected. For the sake of simplicity, only the signal intensities of the toluene aromatic group are shown (Figure 3b). Due to the application of the relaxation time filters, only about 75% of the toluene signal could be monitored during the diffusion measurements. Figure 3b shows an increase in the signal attenuation with increasing temperatures from 25 to 80 °C. From 50 °C and above, a nonlinear attenuation of the signal intensity (in log scale) with increasing gradient strength is observed. Such deviations of the NMR signal attenuation from the simple exponential dependence of eq 1 indicate that, over the given observation time, the molecules may diffuse through different regions with notable differences in their structural features, giving rise to differences in the diffusivities of the “guest” molecules (20, 25). This observation concurs with the heterogeneous nature of humic acid since the molecular interactions with different moieties of humic acids could lead to different resistances along their diffusion path. However, only the average diffusion coefficient (average D as shown in Figure 3b) which corresponds to the slope of the initial attenuation of the curves has been evaluated. The Arrhenius plot of the obtained diffusion coefficients for toluene in humic acid and for bulk toluene is shown in Figure 4. For the case of toluene in humic acid, the error bars are shown to represent the uncertainty in the determination of the slopes of the attenuation curves. These errors originate from the uncertainties in the baseline correction of the spectrum which results in uncertainties in the evaluation of the peak areas under the curves. In particular, this presented a difficulty at 25 °C, where the signal attenuation was less than 8% and was of the same order of magnitude as the uncertainty in the measurement. Thus, the diffusion coefficient at 25 °C is considered as an estimate of the lower limit of the measurable diffusion coefficient at this temperature. The uncertainties in the determination of the diffusivity of bulk toluene were less than 0.5% and have not been indicated on the curve. As expected, the diffusivity of bulk toluene and those of toluene in ALD-HA differ by 1-2 orders of magnitude. In particular, at 80 °C it is observed that the toluene molecules diffuse about 20 times slower in humic acid than in the bulk liquid. Although this is to be expected for any diffusion in a rigid matrix (26-28) due to the resistance to diffusion introduced by the matrix, an influence of sorptive interactions on the mobility in the matrix should not be excluded. Such interactions between diffusant and soil matrices had been postulated by Piatt and Brusseau (29).
FIGURE 3. (a) Stack plots of 1H MAS PFG NMR spectra at 10 kHz of toluene (4.5 wt.-%) in ALD-HA with increasing gradient strength b at 70 °C. For the sake of clarity only nine spectra are shown. (b) Toluene aromatic signal intensity with increasing gradient strength at temperatures from 25 to 80 °C. Measurements were performed with a relaxation delay of 5 s, a δ value of 2.0 ms, and a ∆ value of 50 ms.
FIGURE 4. Plot of ln D vs 103/T for the self-diffusivity of toluene in bulk phase and, in ALD-HA, at temperatures between 25 and 80 °C (4.5 wt.-%) and at 50 °C (0.4 wt.-%). The dotted lines are best fits and are indicative of the activation energy. Thermodynamics. Toluene diffusivity increases with temperature in both the bulk phase and in humic acid. In humic acid, increasing temperature enhances the fluctuation frequency of the humic acid segments, thereby, facilitating the dynamics of the toluene molecules. According to the Arrhenius equation, the effect of temperature on diffusivity can be quantified as D ) D0exp(-Ea /RT)
(2)
where D0 is the diffusivity of the reference state and Ea is the activation energy of diffusion. The plot of ln D against 1/T (Figure 4) gives an activation energy of diffusion for bulk toluene of (15.7 ( 0.1) kJ/mol. The obtained diffusivities and activation energy of diffusion for bulk toluene are, within the experimental error, in good agreement with those reported in the literature (30). In particular, O’Reilly et al. (30) reported activation energy of diffusion of about (13.4 ( 0.5) kJ/mol in the temperature range from -95 to 105 °C with very similar diffusivities. In the case of toluene in humic acid, a nonlinear increase corresponding to a change in slope above 50 °C is observed within the same temperature range. This change corresponds to a decrease in the Ea of toluene in humic acid by a factor of about 3, i.e., from 79.7 to 21.6 kJ/mol when increasing the
temperature above 50 °C. Such a decrease in Ea is not expected in matrices that do not undergo structural changes within this temperature range. This has been shown for diffusion in polymer matrices without structural changes, e.g., benzene in polystyrene (31) and hexafluorobenzene in polypropylene (data available as SI). We therefore presume that the decrease in the activation energy of diffusion of toluene above 50 °C indicates a structural change occurring in the matrix above this temperature that strongly affects the mechanism of diffusion of toluene, and that such a structural change may be related to conformational changes in the matrix. Although structural changes related to glass transitions and melting transitions have been widely discussed for ALD-HA within this temperature range, we do not have sufficient evidence to attribute our findings in an unambiguous way to the abovementioned transitions. To better understand such effects a combination of NMR and DSC measurements will be helpful. Measurements at Lower Concentrations. 1H MAS PFG NMR diffusion measurements on about 0.4 wt.-% toluene in ALD-HA were performed at 50 °C. The apparent diffusivity of toluene at this concentration and temperature was about (3.4 ( 0.5) × 10-11 m2/s, which is about 3 times lower than the diffusion coefficient at approximately 4.5 wt.-% toluene in this matrix (Figure 4). This difference is significant but not huge. We attribute it to a “softening effect” of toluene in the higher concentrated sample. The successful realization of the method on approximately 0.4 wt.-% toluene in ALD-HA opens up a perspective for its application on even lower probe compound concentrations in humic acids. Thus, by utilizing higher magnetic field gradients, as demonstrated in ref 17 and at temperatures at or above 50 °C, the current concentration limit might be lowered between 0.1 and 0.01wt.-%. The diffusivities of toluene in humic acid obtained in this work are of the same order of magnitude as those reported elsewhere (32, 33). Within the margin of error, the average diffusivity of toluene obtained in the presented work at 25 °C with D ) (7.8 ( 2.5) × 10-12 m2/s is in consonant with that predicted by Shih et al. (32) at 27 °C by means of computer simulation (D ) 8.4 × 10-12 m2/s). However, it is faster than that reported by Chang et al. (33) at 25 °C from sorption kinetic measurements of toluene in pressed ALD-HA disk at toluene concentrations less than 1 wt.-% (D ) 6.6 × 10-13 m2/s). The activation energy of diffusion obtained in this work below 50 °C lies within an order of magnitude of those reported by Chang et al. (33) EA ) 42.3 kJ/mol for adsorption and 65.7 kJ/mol for desorption. VOL. 43, NO. 21, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8267
The differences in the diffusivities and the activation energies of diffusion in the presented work and those reported by Chang et al. (33) is likely related to the differences in concentrations used (approximately 4.5 wt.-% in this work, and
