Article pubs.acs.org/est
Novel Experimental−Modeling Approach for Characterizing Perfluorinated Surfactants in Soils Denis Courtier-Murias,*,† Eric Michel,‡ Stéphane Rodts,† and François Lafolie‡ †
Université Paris-Est, Laboratoire Navier (ENPC−IFSTTAR−CNRS), 77420 Champs-sur-Marne, France EMMAH, INRA, Université d’Avignon et des Pays de Vaucluse, 84000 Avignon, France
‡
S Supporting Information *
ABSTRACT: Soil contamination is still poorly understood and modeled in part because of the difficulties of looking inside the “black box” constituted by soils. Here, we investigated the application of a recently developed 1H NMR technique to 19F NMR relaxometry experiments and utilized the results as inputs for an existing model. This novel approach yields 19F T2 NMR relaxation values of any fluorinated contaminant, which are among the most dangerous contaminants, allowing us to noninvasively and directly monitor their fate in soils. Using this protocol, we quantified the amount of a fluorinated xenobiotic (heptafluorobutyric acid, HFBA) in three different environments in soil aggregate packings and monitored contaminant exchange dynamics between these compartments. A model computing HFBA partition dynamics between different soil compartments showed that these three environments corresponded to HFBA in solution (i) between and (ii) inside the soil aggregates and (iii) to HFBA adsorbed to (or strongly interacting with) the soil constituents. In addition to providing a straightforward way of determining the sorption kinetics of any fluorinated contaminant, this work also highlights the strengths of a combined experimental−modeling approach to unambiguously understand experimental data and more generally to study contaminant fate in soils.
■
INTRODUCTION Anthropogenic changes in the environment are creating regional combinations of environmental conditions causing huge impacts on biodiversity, climate, and human health.1−3 Among the different anthropogenic environmental transformations, aquifer and soil contaminations are of foremost importance. In soils, contaminant transport and redistribution are determined by a range of physical and chemical processes (e.g., convective−dispersive transport in the solute form, diffusion, contaminant binding to the soil constituents).4 As a consequence, understanding how and why contaminants move and bind in situ is paramount to environmental and human health and to the preservation and persistence of life on this planet. This has led to the development of many powerful environmental models used to predict contaminant fate and transport in soils.5−8 Usually, some model parameters are difficult to determine from independent experiments and are fitted from experimental data. Even though such models are useful for fitting a specific set of data, representing meaningful physical and chemical information, they alone cannot be used to make comprehensive predictions of contaminant fate in soils in general. Experimentally, the fate of contaminants is most often studied by technologies that do not provide information on the local dynamics of the processes (inside the samples). In general, © XXXX American Chemical Society
soils and contaminants are commonly studied prior to and at the final state of contamination (by conventional chemical extractions), but the information before equilibration of the contaminant is missed,9 with no information about contaminant behavior (e.g., transfer and adsorption mechanisms) at the pore scale. Sorption kinetics are determined from several “identical” soil or sediment samples as proxies, and each of them is analyzed destructively after different experimental times.10 The aqueous concentrations of the contaminant are determined for all samples, but only a subset of them are dried, extracted, and analyzed for adsorbed contaminant content. Because several different samples are needed to build each kinetics sorption curve, this implies a great effort of sample preparation and analysis and also increases the uncertainty of the study. Actually, the complexity of soils and the lack of techniques sufficiently capable of obtaining detailed information on the phenomena occurring inside the soil hinder the complete understanding of processes controlling contaminant fate in soils. In recent years, many of these issues have been ameliorated by the versatility and the nondestructive nature Received: Revised: Accepted: Published: A
November 10, 2016 January 30, 2017 February 6, 2017 February 6, 2017 DOI: 10.1021/acs.est.6b05671 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
studies have used 19F NMR to determine contaminant binding into soil,25 to identify free and soil-bound contaminants,12 or to understand soil−contaminant interactions at the molecular level and to follow kinetics in soils.31 To date, 19F NMR relaxometry and MRI studies have been limited to simple molecules in terms of their NMR properties22,25 [i.e., molecules containing equivalent 19F atoms or 19F atoms that do not interact with another 19F atom by indirect dipole−dipole couplings (J-couplings), e.g., trifluralin, a widely used herbicide]. These J-couplings could lead to a modulation of spin echoes called J-modulation (see Materials and Methods for further details). This phenomenon is of special importance for MRI and NMR relaxometry experiments because it produces image artifacts and impedes the measurement of NMR relaxation values, as has been shown for 19F and 1H nuclei.46−48 As a consequence, there are no examples of 19F MRI or NMR relaxometry studies of complex (in terms of their NMR properties) fluorinated molecules in soil samples. This work addresses this methodological gap. It presents the implementation of a general NMR protocol, here applied to 19F NMR, that was recently developed for 1H NMR and is distinguished by the name PROJECT (periodic refocusing of Jevolution by coherence transfer).49 It has been developed as an alternative to the Carr−Purcell−Meiboom−Gill (CPMG)50,51 NMR protocol commonly used to measure T2 relaxation (or spin−spin relaxation) values for systems not affected by Jmodulation. Here, we used this novel NMR method to distinguish heptafluorobutyric acid (HFBA) in three different environments in a soil aggregate packing, showing that we can quantify and follow the partition kinetics of complex fluorinated molecules using T2 NMR data. In addition, we used a model computing the dynamics of HFBA partition between different soil compartments to ascertain the nature of the three environments identified on the NMR data, emphasizing the added value of combined experimental−modeling approaches for studying soil contamination.
of nuclear magnetic resonance (NMR), which permit one to directly study opaque systems in situ and have boosted the development11−19 and application19−33 of the environmental NMR research field. In particular, NMR spectroscopy has provided a powerful framework to better understand contaminant fate, bioavailability, toxicity, sequestration, and remediation.19 In addition, some recent papers have used 1H magnetic resonance imaging (MRI) to follow relaxation contrast agents (or heavy metal ions) in porous media as a result of the reduction of the NMR relaxation time of water.34−37 Moreover, when 1H NMR chemical shift information is available, MRI experiments permit one to follow the differential transport of various molecules (e.g., oil and water) inside porous media by simply measuring the spin densities of each molecule separately.38 However, despite the potential of this technique, to date there is no example of contaminant transport studied by 1 H NMR or MRI in soils, probably because of their heterogeneity and 1H signal overlapping, which complicates experimental protocols and data interpretation. In the case of 19F NMR, for which sensitivity is almost equal to that of 1H NMR (in contrast to other less sensitive but commonly used NMR nuclei such as 13C), the absence of a natural background signal in soils simplifies signal interpretation in comparison to 1H NMR. Additionally, 19F exhibits a wider range of chemical shifts, which makes the chemical information on NMR spectra more robust to heterogeneities. This offers excellent selectivity for trace molecule analysis in these complex systems. In addition to the advantages of 19F NMR over other nuclei, the concern about fluorinated contaminants has rapidly increased in the recent years.39 Available data indicate organofluorines to be more persistent and more toxic than their nonfluorinated counterparts.40 Among these are poly- and perfluoroalkyl substances (PFASs), which are an emerging class of persistent organic pollutants and include perfluoroalkyl carboxylic acids (PFCAs) and perfluoroalkane sulfonic acids (PFSAs).41 Specifically, perfluorooctanoic acid (PFOA) and perfluorooctane sulfonic acid (PFOS) are well-know for their toxicity. Because of their persistence, toxic, and bioaccumulative characteristics, manufacturers have now shifted to shorter-chain perfluorinated chemicals [4 and 6 C atoms, e.g.. heptafluorobutyric acid (HFBA), studied in this work] that are less toxic and less bioaccumulative. However, they are still persistent, are nearly impossible to remove during water treatments, and have been detected in surface and drinking water, principally because they are more soluble and thus more mobile than longer-chain PFASs.42 For this reason, it has been recently claimed that the replacement of compounds with 8-carbon perfuoroalkyl groups with shorter-chain formulations requires external expert review.39 Data on interactions of PFASs with soil constituents are important for investigating the transport of PFASs in soils.43 They are inherently necessary both to model the fate of existing fluorochemicals released to aqueous environments and to evaluate the potential environmental impacts of new fluorochemicals as they are developed.10 However, despite their detection in water, sediments, soils, and oceans, data on the partitioning of fluorochemicals in soils are scarce. Because of the importance of fluorochemicals and the advantages of 19F NMR, several 19F NMR and MRI applications have been developed to monitor contamination in soils12,13,22,24−27,29,31 and model porous media.44,45 Recent
■
MATERIALS AND METHODS Soil Aggregates and Contaminant Information. Bulk soil was collected from the top ten centimeters of an experimental field in Avignon (43°54′57.9″ N, 4°53′00.0″ E). The soil is a silty clay loam calcisol (silt, 51%; clay, 33.3%; sand, 15.7%). The pH of the soil solution is 8.3; its cation exchange capacity 11.3 cmol kg−1, and its CaCO3 concentration 345 g kg−1. It was air-dried for 2 days and sieved to aggregates with size ranging between 2 and 3 mm. They were wetted to a water content of 14.7% prior to the beginning of the experiments. HFBA [also called perfluoro butanoic acid (PFBA)] was used to test the performance of the 19F PROJECT NMR experiment to measure T2 NMR relaxation in the presence of J-modulation. It was purchased from Sigma-Aldrich (CAS number 375-22-4) and used as is to prepare a solution of concentration C0 = 6.4 × 105 μg/L. This concentration is about 3 orders of magnitude lower than the HFBA critical micelle concentration52 and much higher than the detection limit of 19F NMR observed for a similar pollutant (10 μg/L for PFOS).53 A saturated NaF solution (Sigma-Aldrich, CAS number 7681-49-4) was used as a J-coupling-free standard to compare the performance of 19F CPMG and PROJECT NMR experiments to measure T2 relaxation values. Sample Preparation. Adsorption experiments kinetics were performed placing M = 41.3 g of soil aggregates in a plastic vial (50 mm diameter). They were not compacted. At B
DOI: 10.1021/acs.est.6b05671 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
determination by 1H NMR. The so-called PROJECT NMR experiment,49 which has shown clean exponential decays for all homonuclear spin systems in a broadband manner,48 has been developed to measure T2 relaxation values in 1H spin systems and has been used several times since. This NMR experiment integrates a well-known 90° refocusing pulse into the commonly used CPMG experiment, refocusing J-modulation and representing a simple and general method to measure T2 relaxation (the theoretical basis for this effect can be found in ref 49). This is the approach chosen to be implemented for 19F NMR relaxometry experiments in this work. Further details can be found in the Supporting Information. Modeling. We modeled HFBA adsorption kinetics onto the soil material hypothesizing that it may be located in three compartments: (i) as a solute, in the pores located above and in between the aggregates (volume Vb, concentration Cb, mol m−3); (ii) as a solute, in the pores located inside the aggregates (volume Vi, concentration Ci, mol m−3); and (iii) adsorbed to the soil constituents (concentration S, mol kg−1 of dry soil). Because initially part of the aggregate porosity is free of water (termed “dry region”) and part of it is filled with water, we further divide the aggregate porosity into initially dry and initially wet compartment (volumes Vi,wet and Vi,dry, respectively, with Vi = Vi,wet + Vi,dry), having concentrations of HFBA in solution and adsorbed Ci,wet, Ci,dry, Swet, and Sdry respectively. For the sake of simplicity, we considered that the aggregates were uniform and that HFBA, in solution or adsorbed, was uniformly distributed in the aggregates. At all times, the following mass balance equation must be verified:
time t = 0, a volume V0 (24.7 mL) of a HFBA solution was poured onto the soil aggregates. The HFBA solution covered completely all the aggregates, and a layer of about 7 mm of aggregate-free solution was left above the aggregates. The absence of entrapped bubbles in the aggregate packing was checked visually. Then, the plastic vial was closed and placed inside the NMR magnet. NMR Experiments. All the NMR experiments were carried out at 0.5 T in a vertical imaging spectrometer (DBX 24/80 Bruker) equipped with a birdcage radio frequency coil of 60 mm inner diameter. Such low magnetic field is suited for studies of porous media as the system maintains the susceptibility-induced field inhomogeneity inside the sample at a moderate level. The 19F PROJECT and CPMG NMR experimental pulse sequences (Figure S1) used in this work are P90,x−TE−(P180,y− TE−P90,y−TE−P180,y−TE′−acq−TE′)n and P90,x−TE−(P180,y− TE′−acq−TE′)n respectively.49 P90 and P180 are 90° and 180° pulses, respectively; TE is half of the echo time, acq the acquisition time (with TE′−acq−TE′ equal to the echo time), and n the number of echoes measured; x and y indexes refer to pulse phase in the rotating frame. Ideal relaxation data obtained by these experiments is a sum of exponential decays, owing to either various chemical groups, molecular interactions and dynamics close to neighboring structures, or diffusion−relaxation modes in the porous network:54 N
I (t ) =
∑ M0,me−t /T
2, m
m=1
VbC b(t ) + Vi,dryC i,dry(t ) + Vi,wetC i,wet(t ) + Mdry Sdry(t )
In this equation, I(t) is the experimental data at detection time t (constrained as a multiple of 4TE or 2TE for PROJECT and CPMG, respectively); M0,m and T2,m are the amplitude value and the time constant associated with each exponential decay m, respectively. The influence of diffusion can be minimized by a sufficiently short TE, and it is assumed that the information regarding the amount of a specific NMR signal is carried by the M0 values and the information about the environment of each specific NMR signal by the T2 values (see section S.1 of the Supporting Information). As previously mentioned, undesirable J-modulation jeopardizes NMR relaxometry experiments. Because J-modulation arises from couplings between spins with chemical shift differences Δν ≪ 1/2TE, it can be suppressed, as in diffusion experiments, using short echo times in the CPMG experiment but at the cost of high radio frequency (RF) power deposition. Different NMR strategies have been proposed to reduce the effect of RF power deposition due to short echo times, but they have their own limitations and do not represent a general method (see sections S.2 and S.3 in the Supporting Information for further details). In the case of 19F NMR, the issue related to the use of short echo times is further exacerbated, as the 19F NMR chemical shift range is much higher than that of 1H NMR. To circumvent the use short echo times, 19F Jmodulation has been avoided by using selective RF pulses, but this is possible for only one multiplet at a time.46,47 In addition to missing some of the chemical shift information, this approach is possible only in certain favorable cases (e.g., when chemical shift resolution is good enough for selecting certain NMR signals). Several broadband and general NMR strategies have been therefore recently developed and evaluated48 for T2 relaxation
+ M wetSwet(t ) = V0C0
(1)
where C0 and V0 are the concentration and volumes of the HFBA solution introduced in the aggregate packing, respectively, and Mdry and Mwet are the dry mass of soil (g) in the dry and wet compartments computed as follows: Mdry = M(Vi,dry/Vi) and Mwet = M(Vi,wet/Vi). Most of the soil surface area available for interactions with HFBA is located in the aggregate porosity.55 We thus considered that only HFBA in solution inside the aggregates could adsorb to soil material. In a first modeling attempt, we assumed that (i) HFBA adsorption was kinetic and reversible and that (ii) only one type of adsorption site was involved in HFBA sorption. This simple one-site model failed to reproduce the observed adsorption dynamics (Figure S4). We thus assumed that two types of sorption sites characterized by a fast and a slow adsorption kinetics were involved in HFBA sorption. Moreover this assumption is supported by previously reported data of PFASs sorption kinetics that were well-described by a biexponential model: this model can be interpreted as an adsorption process occurring onto two different sorption sites.10,56,57 We use Swet,fast, Swet,slow, Sdry,fast, and Sdry,slow to denote the fast and slow kinetically sorbed HFBA concentration in the initially dry and initially wet compartments, respectively (mol kg−1 of dry/wet soil) with Swet = Swet,fast + Swet,slow and Sdry = Sdry,fast + Sdry,slow. Kinetic adsorption and desorption were characterized by fast and slow adsorption and desorption rates (s−1), identical in each compartment (initially wet or dry): ka,fast, ka,slow and kd,fast, kd,slow respectively. The rates of variation of HFBA adsorbed in the initially dry and wet compartments are described by C
DOI: 10.1021/acs.est.6b05671 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology dSdry,fast dt
= ka,fast
dSdry,slow
dt
Mdry
= ka,slow
dt dSwet,fast
Vi,dry
= ka,fast
dSwet,slow dt
C i(t ) − kd,fastSdry,fast(t )
Vi,dry Mdry
Vi,wet M wet
= ka,slow
C i(t ) − kd,slowSdry,slow(t )
C i(t ) − kd,fastSwet,fast(t )
Vi,wet M wet
C i(t ) − kd,slowSwet,slow(t )
slow and intermediate relaxing HFBA. See section S.4 of the Supporting Information for details on the model parametrization fitting procedure.
(2a)
■
RESULTS AND DISCUSSION Comparison of 19 F PROJECT and CPMG NMR Protocols in the Absence of J-Modulation. Both 19F PROJECT and CPMG NMR experiments were carried out using a NaF sample, and data were fitted monoexponentially (Figure S2). Resulting amplitudes are 100.0 ± 0.2 and 99.0 ± 0.6 au, and 19F T2 relaxation values obtained are 0.96 ± 0.01 and 0.97 ± 0.02 s for the CPMG and PROJECT experiments, respectively. As expected, the error associated with the PROJECT NMR measurement is slightly higher than that of the CPMG measurement. Because of extra pulses needed to carry out the PROJECT experiment (compared to the CPMG experiment), the effects of pulse imperfections are likely to be more important in this experiment, which could lead to higher uncertainty in its measurements. The uncertainty for both CPMG and PROJECT measurements may be minimized by careful pulse calibration, but the objective of this work was to compare their performance rather than to establish accuracy and precision limits. In the following, to assess the suitability of the 19F PROJECT NMR protocol for relaxometry analysis on systems affected by J-modulation, we compare its performance to that of the 19F CPMG NMR protocol. Comparison of 19 F PROJECT and CPMG NMR Protocols in the Presence of J-Modulation. Figure 1
(2b)
(2c)
(2d)
The fast and slow partition coefficient (m3 kg−1) of HFBA between soil and water can be expressed as Kh = (Ka,hVi)/ (Kd,hM) where h = fast or slow. Finally, we hypothesized that (i) diffusion of HFBA in between the aggregates was fast compared to adsorption time, but that (ii) the diffusion inside the tortuous aggregate pore space was a rate-limiting step. We accounted for diffusion inside the aggregates using a rate constant 1/τD (s−1) that depends on both the aggregate radius R (m) and the HFBA molecular diffusion coefficient D (m2 s−1): τD = R2/(15D/τ) with τ (unitless) the tortuosity of the aggregate pores. 8 The rate of change of HFBA concentration in solution inside the aggregate porosity is written as dC i,dry dt
=
dSdry,slow ⎞ Mdry ⎛ dSdry,fast 1 (C b − C i,dry ) − + ⎟ ⎜ dt ⎠ Vi,dry ⎝ dt τD (3a)
dC i,wet dt
=
dSwet,slow ⎞ M ⎛ dSwet,fast 1 (C b − C i,wet) − wet ⎜ + ⎟ τD Vi,wet ⎝ dt dt ⎠ (3b)
We hypothesized that at t = 0, the HFBA solution instantaneously fills (i) the volume Vb between the aggregates and by capillarity (ii) the water free pore space inside the aggregates. The kinetic adsorption sites are initially free of HFBA. The initial conditions are thus C b(0) = C i,dry(0) = C0
(4a)
C i,wet(0) = Swet,fast(0) = Sdry,fast(0) = Swet,slow(0) = Sdry,slow(0) = 0
(4b)
Equations 1−4 were numerically solved with the Mathematica computational software. To compare the model outputs with the NMR data, the solute and adsorbed concentrations computed by the model were transformed into percent of the number moles of HFBA initially introduced in the aggregate packing as follows: HFBAbetween_aggregates = 100[Cb(t)Vb]/C0V0; HFBAinside_aggregates = 100[Ci,dry(t)Vi,dry + Ci,wet(t)Vi,wet]/C0V0; HFBA adsorbed = 100{M dry [S dry,fast (t) + S dry,slow (t)] + Mwet[Swet,fast(t) + Swet,slow(t)]}/C0V0. The variations with time of these three quantities were compared with those of the slow relaxing, intermediate relaxing, and fast relaxing HFBA, respectively. The model requires 14 parameters. Eight of them where estimated from the experimental conditions or literature values. The tortuosity of the aggregates pores, the aggregate density, and the four rate constants were determined by a simultaneous least-square fit of HFBAbetween_aggregate and HFBAinside_aggregates to
Figure 1. (a) 19F NMR spectrum and a sketch of HFBA molecule with the respective transverse signal decay for the peak at −120 ppm using CPMG (white squares) and PROJECT NMR (white dots) experiments. The black line shows the best monoexponential fit for the 19F PROJECT NMR data. Measured T2 values are about 1.5 s for the three groups of 19F atoms (see Figure S3 for the other peaks’ signal decays).
shows that 19F NMR chemical shift information is available for a HFBA solution sample even at the relative low magnetic field provided by our system. This permitted us to assign each of the three 19 F NMR peaks observed in Figure 1a to its corresponding group of equivalent 19F atoms found in HFBA (see Figure S3). Then, we carried out both 19F CPMG and PROJECT NMR experiments and measured the signal amplitude for each of these three peaks. We observed a similar D
DOI: 10.1021/acs.est.6b05671 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology behavior of the T2 signal decays for the three peaks (Figure S3): the signal obtained by the CPMG NMR experiment (white dots) decreases fast and then oscillates, but the signal obtained by the PROJECT NMR experiment (white squares) shows monoexponential decay (black line). This is explained by the fact that the CPMG NMR experiment suffers from severe Jmodulation, but the PROJECT NMR experiment removes this J-modulation. From this data it is clear that it is not possible to obtain the T2 relaxation values and to perform relaxometry analysis with these types of molecules by CPMG NMR experiments. To our knowledge this is first time that the 19F PROJECT NMR protocol has been carried out, showing its suitability as a general method for 19F NMR relaxometry studies of any fluorinated molecule. We show below that by using this method we are able to follow the partition kinetics of contaminants in different environments in soil aggregates. Dynamics of HFBA Fate in Soil Aggregates Studied by 19 F NMR Relaxometry. Three well-separated T2 relaxation distributions were observed when the HFBA solution was mixed with the soil aggregates (Figure 2). The maxima values
Figure 3. NMR determined (big symbols) and modeled (small symbols) percentage of HFBA in each compartment (error in NMR data is about ±1%). NMR data: slow relaxing (big triangles), intermediate relaxing (big circles), and fast relaxing HFBA (big squares). Model: HFBA in solution above and between the aggregates (small triangles); HFBA in solution inside the aggregate porosity (small circles); and HFBA adsorbed onto soil material, all (small squares), fast (+), and slow (*). Fitted parameters: ka,fast = 11.0 ± 0.1 s−1, ka,slow = 4.0 × 10−5 ± 0.1 × 10−5 s−1, kd,fast = 3.52 ± 0.02 s−1, kd,slow = 5.8 × 10−5 ± 0.2 × 10−5 s−1, τ = 0.72 ± 0.01, ρa = 1602 ± 4 kg m−3.
relatively fast during 150 min and then more moderately, reaching a plateau (at about 34%) about 950 min after HFBA introduction. The contribution of intermediate relaxing HFBA to the signal amplitude was about 15%, and it stayed almost constant during the entire experiment. At time t = 0, in the aggregate-free HFBA solution there was not intermediate relaxing HFBA. This means that the amplitude increase from 0 to 15% occurred very fast, before the first data point was measured. Considering it carefully, the signal amplitude of the intermediate relaxing HFBA reveals a slight increase and then decrease during the first 70 min to reach a constant value (14%) for the rest of the experiment. Finally, fast relaxing HFBA, at the time the first experimental data point was recorded, represented about 43% of the total HFBA, revealing a rapid increase of HFBA in this compartment during the first 7 min of the experiment. Then, the fast relaxing HFBA amount kept increasing relatively quickly for about 200 min and then more moderately up to about 950 min, when it reached a plateau. At the end of the experiment, it constituted the main type of HFBA found in the soil aggregates sample, accounting for 52% of total HFBA (Figure 3). The important differences of the three T2 values (Figure 2) suggest that they correspond to HFBA in different environments rather than to different 19F groups in the same HFBA molecule where small differences in T2 should be expected (as seen for data in Figure S3). The different evolution of the three amplitudes over time supports this hypothesis. Slowly relaxing HFBA has a T2 relaxation value similar to that found for HFBA in solution. This corresponds likely to HFBA located in the supernatant above the aggregates. Also, because the soil aggregates have not been compacted, the pores located in between the aggregates were large (about one millimeter). This suggests that the slow relaxing HFBA was also located in these interaggregate pores where it was almost unaffected by its environment and relaxed similarly to HFBA free in solution. We have considered that all the HFBA having a T2 lower than 10 ms is adsorbed or strongly interacting with soil
Figure 2. 1D T2 distributions of HFBA in the soil aggregates sample. Two different data post-treatments were carried out: using a T2 relaxation distribution with 200 values from 0.1 to 10 000 ms (solid line) and from 1 to 10 000 ms (dashed line). The differences of signal NMR for adsorbed HFBA for the two post-treatments show that NMR data do not allow adsorbed HFBA to be quantified and that a better way consists of subtracting the sum of the NMR signal corresponding to HFBA inside aggregates and free in solution from the total NMR signal measured for the HFBA added to the sample.
of the relaxation time distribution shown in Figure 2 were about 1.3 s (slowly relaxing HFBA), 150 ms (intermediate relaxing HFBA), and lower than 10 ms (fast relaxing HFBA). The NMR relaxation times of water (or of any molecule) in pores vary because of a variety of factors, including the geometric confinement (surface interactions) and molecular interactions. Surface interactions describe the interactions between molecules observed by NMR and the pore walls (here the soil constituents). Interactions with soil organic matter are also known to play a role in the NMR relaxation values of contaminants.32 Here, we interpret the different NMR relaxation times as HFBA having three different geometric confinements and/or surface interactions with soil and use the associated amplitudes to quantify the fraction of HFBA experiencing each environment. The amplitude evolution of these three relaxation components (the associated error is about 1%), as a function of elapsed time after pouring the HFBA solution into soil aggregates, is shown in Figure 3. The first data point was measured 7 min after HFBA introduction inside the aggregates. At that time, slowly relaxing HFBA accounted for only 43% of the total HFBA added in the sample. This value decreased E
DOI: 10.1021/acs.est.6b05671 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
a slightly lower value of R, typically R = 1.0 × 10−3 m, does not change the diffusion rate constant and yields a value of τ of about 1.12. Altogether, the good agreement of modeled and experimental data supports the interpretation made above of the three relaxation processes observed in the NMR data. Interestingly, the unsuccessful preliminary attempt to fit the data with a one-site adsorption model indicates that the abundant internal NMR data were helpful to discriminate between the two candidate models. The affinity of PFASs for soil materials is generally estimated by determining their partition coefficient, K, between soil and water from measurements of the soil and water concentration at equilibrium. Values ranging between 0.004 and 0.4 cm3 g−1 have been reported in riverbank quartz sand41 and of about 11.7 cm3 g−1 in river sediment in china.61 The values of Kfast = 0.77 ± 0.05 cm3 g−1 and Kslow = 0.17 ± 0.01 cm3 g−1 determined in this study fall in the range of those reported previously. The experimental data and the success of the two-site model to reproduce them indicate that at least two mechanisms lead to HFBA adsorption. The question is now to determine the nature of these mechanisms. As highlighted by recent experimental and theoretical results on PFASs fate in soils, PFASs are likely to be involved in two types of interactions with soil materials.62 First, it was shown that the partition coefficient of PFASs is strongly influenced by (i) the amount of organic matter contained in the sorbent and (ii) the length of the PFAS perfluorinated moiety, suggesting that hydrophobic interaction of this moiety with the soil organic matter accounts for a large part of the sorption.10,31,33,62,63 This hypothesis is supported by the good performances of a mechanistic model predicting the partition coefficient of PFASs with organic matter based on a balance between attractive hydrophobic and repulsive Coulombic forces between negatively charged organic matter and PFASs.64 Second, because at environmental pH the carboxylic group of PFASs is deprotonated (the pKa of HFBA is close to 0.08),65 electrostatic interactions were shown to lead to adsorption onto positively charged soil constituents such as metal oxides and clay minerals.66,67 Because fast adsorption constitutes more than three-quarters of the total adsorbed HFBA, and in view of the above literature results, we surmise that in the model, the fast adsorption sites account for hydrophobic interactions between organic matter and the perfluorinated moiety of PFASs, while the slow adsorption sites account for attractive Coulombic forces between the negatively charged PFAS polar moiety and scattered positively charged soil minerals. This assumption is compatible with the observed fast sorption kinetics of PFOS onto negatively charged chitosan beads at neutral to basic pH, when adsorption was driven by hydrophobic interactions only, and the much slower kinetics observed at acidic pH, when attractive Coulombic interactions between the negatively charged PFOS and positively charged chitosan dominated the sorption.56 Future work will aim at confirming whether this provisional attribution of the fast and slow adsorption processes to hydrophobic and Coulombic interactions is valid. Benefits, Limitations, and Potential of the 19F PROJECT NMR Experiment. To our knowledge, this is the first time that the 19F PROJECT NMR experiment has been implemented. However, this experiment is not proposed as a general method for T2 measurements. Because of the slightly better performance of the CPMG NMR sequence, we would still recommend it for studying systems where J-modulation is
aggregates. However, to differentiate between adsorbed HFBA and HFBA strongly interacting with soil aggregates, an independent experimental study that quantifies only adsorption could be useful.42 Finally, intermediate relaxing HFBA was tentatively assigned to HFBA in solution inside the soil aggregates. The pores inside the aggregates are smaller that those in between the aggregates. Thus, we expect the relaxation in these small intra-aggregate pores to be faster than that in the large interaggregate pores but slower than the relaxation of HFBA strongly interacting with or adsorbed to soil materials. The decreasing amount of slow relaxing HFBA with time and the corresponding increasing amount of fast relaxing HFBA supports the hypothesis that fast relaxing HFBA corresponds to HFBA adsorbed to, or strongly interacting with, soil material. These strong interactions are indeed expected to increase the relaxation rates.12,31 In addition, although HFBA does not adsorb as easily to soil as longer-chain PFCAs do, it was shown recently during transport experiments in 5 cm diameter, 50 cm long columns packed with loamy sand that its adsorption onto soil material was far from negligible: 40−55% of applied HFBA was retained in the columns.41 This fraction of retained HFBA is similar to that recorded at the end of the present batch experiment. The 19F NMR data shows that there is HFBA in at least three different environments, but we cannot rule out the fact that some HFBA could be in other environments that are not distinguished by T2 NMR relaxometry. For instance, if exchange between HFBA in different environments is fast in comparison to the NMR time scale or if their NMR relaxation values are too similar, we would not be able to differentiate them and would observe an average of their values. Thus, the 19 F NMR data alone may not unequivocally characterize HFBA in all different environments in soil aggregates, but it contains key information about contaminant fate during penetration into soil aggregates that was not previously accessible experimentally. In the following we show how the partition model may help us to (i) test the above interpretation of the three NMR relaxation values, (ii) improve our understanding of this experimental data, and (iii) gain insights into processes that may not be well-observed experimentally because of signal overlapping or fast exchange on the NMR time scale. Modeling. Figure 3 compares the experimental data (large symbols) with the outputs of the model (small symbols). The modeled HFBA repartition in the porous medium (small triangles = HFBA b e t w e e n _ a g g r e g a t e , small circles = HFBAinside_aggregates, and small squares = HFBAadsorbed) fits the variations with the slow, intermediate, and fast relaxing HFBA amplitudes, respectively, reasonably well. The optimized value of the density, 1602 kg m−3, falls close to the lower bound of the expected range. This low density value finds a simple explanation when considering the method used to prepare the sample. Upon pouring, the structure of the aggregates undergoes a number of transformations including swelling, microcracking, and breakdown.58−60 The overall result of these transformations is a packing of aggregates having a much opened, and consequently less dense, structure than before pouring HFBA. These transformations probably also result in aggregates smaller than the arithmetic mean of the aggregate size bounds (R = 1.25 × 10−3 m) used in the calculation together with the optimized pore tortuosity to compute the diffusion rate constant. This may explain the unphysical (lower than 1) optimized value of the tortuosity τ = 0.7. Indeed, using F
DOI: 10.1021/acs.est.6b05671 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
HFBA affinity with soil constituents. The 19F PROJECT NMR experiment allowed the simultaneous recording of unique timelapse information on the partition of a PFAS in different compartments of the soil. This information and its interpretation resorting to an adequate model open a unique method of determining the partition of PFASs between these environments, paving the way for a separation of the contribution of each type of interaction. The combined experimental−modeling approach proved valuable because (i) the good agreement of the model outputs with the NMR data supported the initial interpretation of the experimental data and (ii) the data helped in discriminating a two-site candidate model. In addition, the combined experimental−modeling approach contributed to a better understanding of the partition dynamics by providing (i) estimates of the dynamics of fast and slow adsorbed HFBA separately, which could not be distinguished experimentally probably because of fast exchange in the NMR time scale, and (ii) an estimate of the early time dynamics that could not be observed by the experiment alone. We believe that this combination of modeling and internal experimental NMR data is promising and may facilitate future studies of contaminant fate in soil.
not present. However, in the case of systems affected by Jmodulation, 19F CPMG NMR experiments do not permit relaxometry analysis for a wide range of environmentally relevant molecules, and the 19F PROJECT NMR experiment emerges as an indispensable tool paving the way for a range of novel applications. The novel 19F experiment represents a complementary tool to recent NMR developments showing a great potential to understand soil contamination. For instance, PFOA (whose structure differs from HFBA only on the carbon chain length) has been monitored in a soil sample,31 following the evolution over time of the so-called solution, gel, and solid phases (depending on the NMR experiment used). These show a trend similar to the kinetics of slow, intermediate, and fast relaxing HFBA, respectively. Because the signal intensity multiexperimental approach depends on relaxation, nuclear tumbling, and/or dipolar interactions, quantitative signal comparison between different environments is not straightforward. The approach proposed here uses a single NMR experiment, which overcomes these limitations and permits one to quantify the signal evolution over time of contaminants in different local environments. Moreover, it can be implemented in most NMR spectrometers. However, when the same molecule in different environments is measured with one single NMR experiment, it is difficult to extract chemical shift information. This is mainly due to spectral overlapping, and it is exacerbated in heterogeneous systems as in our case. In the case of kinetics studies, this does not represent an issue as the HFBA amplitude is fitted as a whole (i.e., representing all the 19F atoms of HFBA) as it is done in multiexperimental approaches at high magnetic fields.31 The great advantage of the chemical shift information finds great use in different types of analyses (i.e., epitope mapping).31 Even though chemical shift information is not used in this work, Figure 1 shows that 19F PROJECT NMR could be an excellent tool to investigate the molecular dynamics of fluorinated systems as it is commonly done in 1H and 13C NMR studies by combining relaxation and chemical shift information. Thus, it could be a complementary tool for the study of contamination in environmental samples by solutionstate 19F NMR.26,27,30 The PROJECT block could also be useful when J-modulation artifacts jeopardize other NMR or MRI pulse sequences. For instance, the NMR protocol used in this work may be valuable for removing J-modulation artifacts on 19F MRI experiments.46 This would permit the monitoring of PFASs displacement during a transport experiment. The 19 F PROJECT experiment has been tested by monitoring the adsorption kinetics in soils of a short-chain PFAS. The novel 19F NMR experiment overcomes the limitations of regular kinetics studies in which different samples are studied destructively: it provides simultaneous time-lapse information on adsorbed and in solution PFASs, in a nondestructive manner, using a single sample. This represents a time- and cost-effective way to accurately determine the partition coefficient K of PFASs. These partition coefficients could subsequently be used to parametrize PFAS transport models and to predict the fate in soils of existing and new PFASs as they appear. It was recently suggested that the partition coefficient K is not the best descriptor of the sorption behavior of PFAS because it is a lumped parameter that takes into account both hydrophobic and electrostatic interactions.62 However, this is the only quantitative estimate that could be made, to date, of
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.est.6b05671. NMR experimental information and data processing, approaches to reduce the effect of RF power deposition in NMR relaxation experiments, NMR approaches to remove J-modulation in spin echo based experiments, model parametrization, and Figures S1−S4 (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*Université Paris-EstLaboratoire Navier (UMR 8205 IFSTTAR-ENPC-CNRS), 2 allée Kepler, 77420 Champs-surMarne, France. Phone: +33 (0) 1 81 66 84 76. Fax: +33 (0) 1 81 66 84 50. E-mail:
[email protected]. ORCID
Denis Courtier-Murias: 0000-0001-8023-738X Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS Authors thank Pascal Moucheront for his assistance with 19F NMR coil installation and Dr. Hashim Farooq for reviewing the manuscript.
■
REFERENCES
(1) Pimm, S. L.; Russell, G. J.; Gittleman, J. L.; Brooks, T. M. The Future of Biodiversity. Science 1995, 269 (5222), 347−350. (2) Huber, M.; Knutti, R. Anthropogenic and natural warming inferred from changes in Earth’s energy balance. Nat. Geosci. 2012, 5 (1), 31−36. (3) Myers, S. S.; Patz, J. A. Emerging Threats to Human Health from Global Environmental Change. Annu. Rev. Environ. Resour. 2009, 34 (1), 223−252. (4) Kundas, S.; Gishkeluk, I.; Grinchik, N. Application of computer modeling for the analysis and prediction of contaminant behavior in groundwater systems. In Strategies to Enhance Environmental Security in Transition Countries; Hull, R., Barbu, C.-H., Goncharova, N., Eds.; G
DOI: 10.1021/acs.est.6b05671 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology NATO Science for Peace and Security Series C: Environmental Security; Springer: Netherlands, 2007; pp 57−68. (5) Lambert, S. M.; Porter, P. E.; Schieferstein, R. H. Movement and Sorption of Chemicals Applied to the Soil. Weeds 1965, 13 (3), 185. (6) Lambert, S. M. Functional relation between sorption in soil and chemical structure. J. Agric. Food Chem. 1967, 15 (4), 572−576. (7) Chiou, C. T.; Peters, L. J.; Freed, V. H. A Physical Concept of Soil-Water Equilibria for Nonionic Organic Compounds. Science 1979, 206 (4420), 831−832. (8) Lafolie, F.; Hayot, C.; Schweich, D. Experiments on Solute Transport in Aggregated Porous Media: Are Diffusions Within Aggregates and Hydrodynamic Dispersion Independent? Transp. Porous Media 1997, 29 (3), 281−307. (9) Kelsey, J. W.; Alexander, M. Declining bioavailability and inappropriate estimation of risk of persistent compounds. Environ. Toxicol. Chem. 1997, 16 (3), 582−585. (10) Higgins, C. P.; Luthy, R. G. Sorption of Perfluorinated Surfactants on Sediments. Environ. Sci. Technol. 2006, 40 (23), 7251− 7256. (11) Mao, J.-D.; Schmidt-Rohr, K. Separation of aromatic-carbon 13C NMR signals from di-oxygenated alkyl bands by a chemical-shiftanisotropy filter. Solid State Nucl. Magn. Reson. 2004, 26 (1), 36−45. (12) Courtier-Murias, D.; Farooq, H.; Masoom, H.; Botana, A.; Soong, R.; Longstaffe, J. G.; Simpson, M. J.; Maas, W. E.; Fey, M.; Andrew, B.; Struppe, J.; Hutchins, H.; Krishnamurthy, S.; Kumar, R.; Monette, M.; Stronks, H. J.; Hume, A.; Simpson, A. J. Comprehensive multiphase NMR spectroscopy: Basic experimental approaches to differentiate phases in heterogeneous samples. J. Magn. Reson. 2012, 217, 61−76. (13) Longstaffe, J. G.; Courtier-Murias, D.; Soong, R.; Simpson, M. J.; Maas, W. E.; Fey, M.; Hutchins, H.; Krishnamurthy, S.; Struppe, J.; Alaee, M.; Kumar, R.; Monette, M.; Stronks, H. J.; Simpson, A. J. InSitu Molecular-Level Elucidation of Organofluorine Binding Sites in a Whole Peat Soil. Environ. Sci. Technol. 2012, 46 (19), 10508−10513. (14) Shirzadi, A.; Simpson, M. J.; Kumar, R.; Baer, A. J.; Xu, Y.; Simpson, A. J. Molecular Interactions of Pesticides at the Soil−Water Interface. Environ. Sci. Technol. 2008, 42 (15), 5514−5520. (15) Kinchesh, P.; Samoilenko, A. A.; Preston, A. R.; Randall, E. W. Stray field nuclear magnetic resonance of soil water: development of a new, large probe and preliminary results. J. Environ. Qual. 2002, 31 (2), 494−499. (16) Smernik, R. J.; Oades, J. M. Solid-state 13C-NMR dipolar dephasing experiments for quantifying protonated and non-protonated carbon in soil organic matter and model systems. Eur. J. Soil Sci. 2001, 52 (1), 103−120. (17) Courtier-Murias, D.; Farooq, H.; Longstaffe, J. G.; Kelleher, B. P.; Hart, K. M.; Simpson, M. J.; Simpson, A. J. Cross polarizationsingle pulse/magic angle spinning (CPSP/MAS): A robust technique for routine soil analysis by solid-state NMR. Geoderma 2014, 226− 227, 405−414. (18) Farooq, H.; Courtier-Murias, D.; Soong, R.; Masoom, H.; Maas, W.; Fey, M.; Kumar, R.; Monette, M.; Stronks, H.; Simpson, M. J.; Simpson, A. J. Rapid parameter optimization of low signal-to-noise samples in NMR spectroscopy using rapid CPMG pulsing during acquisition: application to recycle delays. Magn. Reson. Chem. 2013, 51 (3), 129−135. (19) Simpson, A. J.; Simpson, M. J.; Soong, R. Nuclear magnetic resonance spectroscopy and its key role in environmental research. Environ. Sci. Technol. 2012, 46 (21), 11488−11496. (20) Simpson, A. J.; Kingery, W. L.; Shaw, D. R.; Spraul, M.; Humpfer, E.; Dvortsak, P. The application of 1H HR-MAS NMR spectroscopy for the study of structures and associations of organic components at the solid-aqueous interface of a whole soil. Environ. Sci. Technol. 2001, 35 (16), 3321−3325. (21) Šmejkalová, D.; Piccolo, A. Host-Guest Interactions between 2,4-Dichlorophenol and Humic Substances As Evaluated by 1 H NMR Relaxation and Diffusion Ordered Spectroscopy. Environ. Sci. Technol. 2008, 42 (22), 8440−8445.
(22) Simpson, M. J.; Simpson, A. J.; Gross, D.; Spraul, M.; Kingery, W. L. 1H and 19F nuclear magnetic resonance microimaging of water and chemical distribution in soil columns. Environ. Toxicol. Chem. 2007, 26 (7), 1340−1348. (23) Kohl, S. D.; Toscano, P. J.; Hou, W.; Rice, J. A. Solid-State 19F NMR Investigation of Hexafluorobenzene Sorption to Soil Organic Matter. Environ. Sci. Technol. 2000, 34 (1), 204−210. (24) Khalaf, M.; Kohl, S. D.; Klumpp, E.; Rice, J. A.; Tombácz, E. Comparison of sorption domains in molecular weight fractions of a soil humic acid using solid-state 19F NMR. Environ. Sci. Technol. 2003, 37 (13), 2855−2860. (25) Strynar, M.; Dec, J.; Benesi, A.; Jones, A. D.; Fry, R. A.; Bollag, J.-M. Using 19F NMR Spectroscopy to Determine Trifluralin Binding to Soil. Environ. Sci. Technol. 2004, 38 (24), 6645−6655. (26) Longstaffe, J. G.; Simpson, M. J.; Maas, W.; Simpson, A. J. Identifying Components in Dissolved Humic Acid That Bind Organofluorine Contaminants using 1H{19F} Reverse Heteronuclear Saturation Transfer Difference NMR Spectroscopy. Environ. Sci. Technol. 2010, 44 (14), 5476−5482. (27) Longstaffe, J. G.; Courtier-Murias, D.; Simpson, A. J. The pHdependence of organofluorine binding domain preference in dissolved humic acid. Chemosphere 2013, 90 (2), 270−275. (28) Todoruk, T. R.; Litvina, M.; Kantzas, A.; Langford, C. H. LowField NMR Relaxometry: A Study of Interactions of Water with Water-Repellant Soils. Environ. Sci. Technol. 2003, 37 (13), 2878− 2882. (29) Dixon, A. M.; Mai, M. A.; Larive, C. K. NMR Investigation of the Interactions between 4‘-Fluoro-1‘-acetonaphthone and the Suwannee River Fulvic Acid. Environ. Sci. Technol. 1999, 33 (6), 958−964. (30) Šmejkalová, D.; Spaccini, R.; Fontaine, B.; Piccolo, A. Binding of Phenol and Differently Halogenated Phenols to Dissolved Humic Matter As Measured by NMR Spectroscopy. Environ. Sci. Technol. 2009, 43 (14), 5377−5382. (31) Masoom, H.; Courtier-Murias, D.; Soong, R.; Maas, W. E.; Fey, M.; Kumar, R.; Monette, M.; Stronks, H. J.; Simpson, M. J.; Simpson, A. J. From Spill to Sequestration: The Molecular Journey of Contamination via Comprehensive Multiphase NMR. Environ. Sci. Technol. 2015, 49 (24), 13983−13991. (32) Simpson, M. J.; Simpson, A. J.; Hatcher, P. G. Noncovalent interactions between aromatic compounds and dissolved humic acid examined by nuclear magnetic resonance spectroscopy. Environ. Toxicol. Chem. 2004, 23 (2), 355−362. (33) Shirzadi, A.; Simpson, M. J.; Xu, Y.; Simpson, A. J. Application of Saturation Transfer Double Difference NMR to Elucidate the Mechanistic Interactions of Pesticides with Humic Acid. Environ. Sci. Technol. 2008, 42 (4), 1084−1090. (34) Lehoux, A. P.; Rodts, S.; Faure, P.; Michel, E.; Courtier-Murias, D.; Coussot, P. Magnetic resonance imaging measurements evidence weak dispersion in homogeneous porous media. Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top. 2016, 94 (5), 053107. (35) Ramanan, B.; Holmes, W. M.; Sloan, W. T.; Phoenix, V. R. Investigation of Nanoparticle Transport Inside Coarse-Grained Geological Media Using Magnetic Resonance Imaging. Environ. Sci. Technol. 2012, 46 (1), 360−366. (36) Baumann, T.; Werth, C. J. Visualization of colloid transport through heterogeneous porous media using magnetic resonance imaging. Colloids Surf. Colloids Surf., A 2005, 265 (1−3), 2−10. (37) Nestle, N.; Wunderlich, A.; Niessner, R.; Baumann, T. Spatial and Temporal Observations of Adsorption and Remobilization of Heavy Metal Ions in a Sandy Aquifer Matrix Using Magnetic Resonance Imaging. Environ. Sci. Technol. 2003, 37 (17), 3972−3977. (38) Chardaire-Rivière, C.; Roussel, J. C. Use of a high magnetic field to visualize and study fluids in porous media. Magn. Reson. Imaging 1991, 9 (5), 827−832. (39) Sedlak, D. Fool Me Once. Environ. Sci. Technol. 2016, 50 (18), 9803−9804. (40) Adams, D. E. C.; Halden, R. U. Fluorinated Chemicals and the Impacts of Anthropogenic Use. In Contaminants of Emerging Concern H
DOI: 10.1021/acs.est.6b05671 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology in the Environment: Ecological and Human Health Considerations; ACS Symposium Series; American Chemical Society: Washington, DC, 2010; Vol. 1048, pp 539−560. (41) Buck, R. C.; Franklin, J.; Berger, U.; Conder, J. M.; Cousins, I. T.; de Voogt, P.; Jensen, A. A.; Kannan, K.; Mabury, S. A.; van Leeuwen, S. P. Perfluoroalkyl and polyfluoroalkyl substances in the environment: Terminology, classification, and origins. Integr. Environ. Assess. Manage. 2011, 7 (4), 513−541. (42) Vierke, L.; Möller, A.; Klitzke, S. Transport of perfluoroalkyl acids in a water-saturated sediment column investigated under nearnatural conditions. Environ. Pollut. 2014, 186, 7−13. (43) Gellrich, V.; Stahl, T.; Knepper, T. P. Behavior of perfluorinated compounds in soils during leaching experiments. Chemosphere 2012, 87 (9), 1052−1056. (44) Zhang, C.; Werth, C. J.; Webb, A. G. A Magnetic Resonance Imaging Study of Dense Nonaqueous Phase Liquid Dissolution from Angular Porous Media. Environ. Sci. Technol. 2002, 36 (15), 3310− 3317. (45) Zhang, C.; Werth, C. J.; Webb, A. G. Characterization of NAPL Source Zone Architecture and Dissolution Kinetics in Heterogeneous Porous Media Using Magnetic Resonance Imaging. Environ. Sci. Technol. 2007, 41 (10), 3672−3678. (46) Giraudeau, C.; Flament, J.; Marty, B.; Boumezbeur, F.; Mériaux, S.; Robic, C.; Port, M.; Tsapis, N.; Fattal, E.; Giacomini, E.; et al. A new paradigm for high-sensitivity 19 F magnetic resonance imaging of perfluorooctylbromide. Magn. Reson. Med. 2010, 63 (4), 1119−1124. (47) Kimura, A.; Narazaki, M.; Kanazawa, Y.; Fujiwara, H. 19F Magnetic resonance imaging of perfluorooctanoic acid encapsulated in liposome for biodistribution measurement. Magn. Reson. Imaging 2004, 22 (6), 855−860. (48) Segawa, T. F.; Bodenhausen, G. Determination of transverse relaxation rates in systems with scalar-coupled spins: The role of antiphase coherences. J. Magn. Reson. 2013, 237, 139−146. (49) Aguilar, J. A.; Nilsson, M.; Bodenhausen, G.; Morris, G. A. Spin echo NMR spectra without J modulation. Chem. Commun. 2012, 48 (6), 811−813. (50) Carr, H.; Purcell, E. Effects of Diffusion on Free Precession in Nuclear Magnetic Resonance Experiments. Phys. Rev. 1954, 94 (3), 630−638. (51) Meiboom, S.; Gill, D. Modified Spin-Echo Method for Measuring Nuclear Relaxation Times. Rev. Sci. Instrum. 1958, 29 (8), 688. (52) Henriksson, U.; Ö dberg, L. A 19F nuclear magnetic relaxation study of micelle formation in aqueous solutions of heptafluorobutyric acid and sodium pentadecafluorooctanoate. J. Colloid Interface Sci. 1974, 46 (2), 212−219. (53) Moody, C. A.; Kwan, W. C.; Martin, J. W.; Muir, D. C. G.; Mabury, S. A. Determination of Perfluorinated Surfactants in Surface Water Samples by Two Independent Analytical Techniques: Liquid Chromatography/Tandem Mass Spectrometry and 19 F NMR. Anal. Chem. 2001, 73 (10), 2200−2206. (54) Grebenkov, D. S. Multiexponential attenuation of the CPMG spin echoes due to a geometrical confinement. J. Magn. Reson. 2006, 180 (1), 118−126. (55) Zachara, J.; Brantley, S.; Chorover, J.; Ewing, R.; Kerisit, S.; Liu, C.; Perfect, E.; Rother, G.; Stack, A. G. Internal Domains of Natural Porous Media Revealed: Critical Locations for Transport, Storage, and Chemical Reaction. Environ. Sci. Technol. 2016, 50 (6), 2811−2829. (56) Zhang, Q.; Deng, S.; Yu, G.; Huang, J. Removal of perfluorooctane sulfonate from aqueous solution by crosslinked chitosan beads: Sorption kinetics and uptake mechanism. Bioresour. Technol. 2011, 102 (3), 2265−2271. (57) Chiron, N.; Guilet, R.; Deydier, E. Adsorption of Cu(II) and Pb(II) onto a grafted silica: isotherms and kinetic models. Water Res. 2003, 37 (13), 3079−3086. (58) Legout, C.; Leguedois, S.; Le Bissonnais, Y. Aggregate breakdown dynamics under rainfall compared with aggregate stability measurements. Eur. J. Soil Sci. 2005, 56 (2), 225−238.
(59) Michel, E.; Majdalani, S.; Di-Pietro, L. A novel conceptual framework for long-term leaching of autochthonous soil particles during transient flow: Long-term soil particle leaching: a conceptual framework. Eur. J. Soil Sci. 2014, 65 (3), 336−347. (60) Le Bissonnais, Y. Aggregate stability and assessment of soil crustability and erodibility: I. Theory and methodology: Aggregate stability and assessment of soil crustability and erodibility. Eur. J. Soil Sci. 2016, 67 (1), 11−21. (61) Wang, P.; Lu, Y.; Wang, T.; Zhu, Z.; Li, Q.; Zhang, Y.; Fu, Y.; Xiao, Y.; Giesy, J. P. Transport of short-chain perfluoroalkyl acids from concentrated fluoropolymer facilities to the Daling River estuary, China. Environ. Sci. Pollut. Res. 2015, 22 (13), 9626−9636. (62) Gellrich, V.; Knepper, T. P. Sorption and Leaching Behavior of Perfluorinated Compounds in Soil. In Polyfluorinated Chemicals and Transformation Products; Knepper, T. P., Lange, F. T., Eds.; Springer: Berlin, 2012; Vol. 17, pp 63−72. (63) Longstaffe, J. G.; Simpson, A. J. Understanding solution-state noncovalent interactions between xenobiotics and natural organic matter using 19F/1H heteronuclear saturation transfer difference nuclear magnetic resonance spectroscopy. Environ. Toxicol. Chem. 2011, 30 (8), 1745−1753. (64) Higgins, C. P.; Luthy, R. G. Modeling Sorption of Anionic Surfactants onto Sediment Materials: An a priori Approach for Perfluoroalkyl Surfactants and Linear Alkylbenzene Sulfonates. Environ. Sci. Technol. 2007, 41 (9), 3254−3261. (65) Agency for Toxic Substances and Disease Registry (ATSDR) Toxicological Profile for Perfluoroalkyls. (Draft for Public Comment); U.S. Department of Health and Human Services, Public Health Service: Atlanta, GA, 2009; http://www.atsdr.cdc.gov/toxprofiles/tp.asp?id= 1117&tid=237. (66) Hellsing, M. S.; Josefsson, S.; Hughes, A. V.; Ahrens, L. Sorption of perfluoroalkyl substances to two types of minerals. Chemosphere 2016, 159, 385−391. (67) Tang, C. Y.; Shiang Fu, Q.; Gao, D.; Criddle, C. S.; Leckie, J. O. Effect of solution chemistry on the adsorption of perfluorooctane sulfonate onto mineral surfaces. Water Res. 2010, 44 (8), 2654−2662.
I
DOI: 10.1021/acs.est.6b05671 Environ. Sci. Technol. XXXX, XXX, XXX−XXX