Time-lapse 3D imaging by positron emission tomography of Cu mobilized in a soil column by the herbicide MCPA

Phenoxyalkanoic acids like the 4-chloro-2-methylphenoxyacetic acid (MCPA) are the second highest used xenobiotic herbicides worldwide after glyphosate because of their apparently favorable environmental properties. Experimental batch equilibration data suggested a reduced Cu adsorption efficiency with the soil mineral goethite below pH 6 in presence of MCPA. This has been verified by advanced surface complexation adsorption modelling involving dissolved Cu-MCPA complexation constants. Positron emission tomography is a non-invasive molecular imaging method for time-resolved three-dimensional information commonly applied on non-retarded tracers in soil core scale experiments. Mineral surface reactive tracers like Cu-64 are too immobile for the relatively short observation times available with this advanced imaging technique. However, Cu-64 radiolabeled Cu-MCPA complex migration could be observed in as long as 10-cm artificial soil test columns where break-through occurred within a few days. For the first time, time-lapse movies of Cu migration in the opaque soil columns were recorded using this novel reactive transport process tomography approach.

. Complexation constants of Cu-MCPA in comparison with other dissolved organic Cu complexes like with fulvic acid (HFA). Data for Cu-MCPA from Masbouh et al. (2006), all other data from Visual MINTEQ 3.1 database (Gustafsson, 2016).
Batch Equilibrium Adsorption Experiments. The synthetic goethite used was Bayferrox 910 Z (LANXESS, Köln, Germany), which is a base material used in fixed-bed adsorption columns in waterworks worldwide (e.g., the Sorb 33™ system used by Severn Trent in the USA). MCPA (99.0% pure; Sigma-Aldrich, Gillingham, UK) was purified by recrystallizing it from a 1:1 mixture of water and ethanol. In the first suspension batch series, 30 mL of the Cu(MCPA)2 stock solution was poured into each of a series of 50 mL polypropylene centrifuge tubes. The mixtures in 10 tubes were adjusted to between pH 3 and 8 by adding 0.1 mol L −1 CO2-free HNO3 or NaOH dropwise. Other series of tubes were prepared each at 10 different pH values, but dry goethite was added to give a solid/solution ratio of 5 g L −1 when 30 mL of either the Cu or Cu(MCPA)2 stock solutions were added. All the tubes were allowed to equilibrate with shaking for 48 h at 25 °C in a dark room to avoid photodegradation of MPCA. Preliminary kinetic studies showed that adsorption equilibrium was reached within 48 h (data not shown). Each tube was centrifuged at 3000 g for 20 min, and the supernatant was passed through a 0.2 µm Minisart hydrophilic membrane filter (Sartorius, Göttingen, Germany). The pH of each supernatant was measured using a pH meter equipped with a WTW Inolab Level 2P combination electrode (Xylem Analytics, Weilheim, Germany), which had been calibrated using CertiPur buffers at three different pH values (Merck). The total dissolved Cu concentration was measured using a Series 7700 inductively coupled plasma mass spectrometer (Agilent Technologies, Santa Clara, CA, USA).
Setup of the CD-MUSIC surface complexation model. Using a multisite surface complexation model requires information about the proton affinities of the various types of surface groups on all relevant crystal planes of the adsorbent. The needle-like goethite crystals have a rather high aspect ratio, so are dominated by the basal prism (110) surfaces. The site density for the singly coordinated -OH groups and triply coordinated 3-OH groups relevant to proton binding has been found to be (3.0 ± 0.3) nm −2 for such surfaces (Hiemstra and Van Riemsdijk, 2006). In the 1-pK approach used, both types of negatively charged surface hydroxyl groups (FeOH −0.5 and Fe3O −0.5 in Table S2) become positively charged to +0.5 valence units (v.u.) by binding a proton. The proton affinity constants for both sites (pKH1 and pKH2) were set to equal the goethite pHpzc. An adsorbate, particularly an organic molecule, is large compared with a proton, so the charge was distributed over the electrical double layer to give a charge distribution multisite surface complexation (CD-MUSIC) model. In the extended Stern (three-plane) approach, the charges of the surface reactive functional groups of the adsorbate molecule are distributed over the surface plane (0-plane) and the first Stern plane (1-plane, Table S2). The precise location of the coordinating ligand charge at the solid-water interface is not explicitly known, and must be derived from structural information at the molecular level (EXAFS data). The adsorption model was parameterized using the speciation code VMINTEQ v3.1 coupled with the parameter fitting code PEST (Gustafsson, 2016). Activity corrections were made to the stability constants for the aqueous species using the Davies equation. The aqueous speciation database was extended using the dissolved MCPA-bearing species constants (Table S2). The background electrolyte ions can form ion pairs with both the singly and triply coordinated surface sites. The presence of the NaNO3 background electrolyte was therefore considered using corresponding surface reactions with the intrinsic equilibrium reaction constants fixed at logKNa = logKNO 3 − logKH1,2 = −0.6 for symmetrical behavior (Lützenkirchen et al., 2008).  Previously published models were used for both binary adsorption systems Cu-goethite and MCPA-goethite. The CD-MUSIC model was first used to calculate Cu adsorption considering bidentate inner-sphere Cu complexation with the singly coordinated surface sites of the goethite and allowing for hydrolysis according to reactions (3) and (4): The corresponding charge distribution as suggested by Weng et al. (2008) is shown in Table S2, and the surface complexation constants were fitted to the experimental binary Cu-goethite system data from this study, also shown in Table S2, with quite good match (R = 0.99). The model results indicated that the hydrolyzed bidentate surface species dominated the adsorption edge and the non-hydrolyzed Cu surface species contributed only to tailing of the adsorption edge toward acidic pH values. The MCPA ligand can form both outer-sphere and inner-sphere monodentate complexes. In inner-sphere complexation, a carbonyl oxygen atom exchanges with a protonated surface hydroxyl group to release a water molecule according to reaction (5): The carboxylic group act as a monodentate ligand towards Fe(III), thereby leaving the non-coordinated carboxylic oxygen free to bridge through hydrogen bonding to either water molecules in the bulk solution, or to water molecules on the surface as can be seen in molecular dynamics movies (Kersten et al., 2014).
The adsorption reaction for the outer-sphere surface complex of the anionic MCPA molecule is shown in reaction (6): The charge distribution for this monodentate inner-sphere complex of the neutral MCPA molecule was explained in detail previously (Kersten et al. 2014, Table S2). The monodentate inner-sphere surface complex of the protonated MCPA was dominant in the acidic pH range. Since MCPA adsorption was largely unaffected by NaNO3 concentrations (10-100 mM), the contribution of this outer-sphere complex was insignificant and responsible only for some tailing of the adsorption edge toward the neutral pH range. Similarly, outer-sphere surface complex formation was assumed for the ternary complex, facilitated by the co-adsorption of one proton to render the surface charge positive, as shown in reaction (7): A good fit to the data was found if the charge on the ternary complex was not kept fully in one plane but was equally distributed over all three planes, leading to CD coefficients of z0 = 1 (surface protonation), and z1 = z2 = −0.5 (Table S2). No inner-sphere surface complexation was assumed for this ternary complex because of steric hindrance. Any attempts to trace inner-sphere complexation using EXAFS failed (data not shown), and it was also not necessary to invoke such complex formation to improve the fit quality.
Cu-64 radiotracer production using a cyclotron. The radionuclide 64 Cu (T1/2 = 12.7 h) was produced by means of the nuclear reaction 64 Ni(p,n) 64 Cu (Szelecsényi et al., 1993). The target was prepared by electrodeposition of metallic [ 64 Ni]Ni (isotope enrichment for 64 Ni 99.09%, STB Isotope, Germany) from a 0.22 M (NH4)2SO4 solution at pH 9, a voltage of 2.60 V and a current of 25 mA on a gold disc (Avila-Rodriguez et al., 1997). A target thickness of 38.8 mg/cm 2 was achieved. The Ni deposit was covered with a 100 µm thick Al foil to avoid Ni loss during the irradiation. Routine production runs of 64 Cu were performed at an initial energy of 12 MeV protons and at currents of ~20 µA lasting for 3-5 h at the Leipzig cyclotron Cyclone ® 18/9 (IBA Molecular, Belgium) equipped with a COSTIS ® solid target system (IBA Molecular, Belgium, see Fig. S5 below). The separation of 64 Cu from the target material 64 Ni was performed by anion exchange chromatography (Thieme et al., 2012). The irradiated Ni deposit was dissolved in boiling 12.5 M HCl, and the received solution was evaporated to dryness. The residue was re-dissolved in 6 M HCl (target solution). The anionic exchanger AG ® 1-X8 resin (BIO-RAD, USA) was used to separate the target material 64 Ni from the radionuclide 64 Cu. The target solution was loaded on the resin column (   The PET experiment. The ClearPET is a high-performance small animal PET scanner that has been developed by the Crystal Clear Collaboration (CCC) and was commercially available through the Raytest group (Sempere-Roldan et al., 2007). The ClearPET is characterized by the following performance data: • 10240 double-layered miniaturized (2mm×2mm×10 mm) detector crystals, L(Y)SO/LuYAP:Ce in phoswich configuration providing DOI-Information • Modular, 20 cassettes with each 4 PMTs (Hamamatsu R7600-M54) • Rotating gantry (1×360° rotation per min = minimum frame length 60 s) • Adjustable gantry diameter: (125 -220) mm • Field of View (FOV): adjustable diameter (70 -144) mm, length 110 mm • Activity range in FOV: ca. 100 kBq to 30 MBq (small) / 70 MBq (large) • Detection threshold in the order of 3 kBq mL -1 • Resolution CFOV 1.1 mm • Max. sensitivity 4.5 % • Open-Source software STIR, available via http://stir.sourceforge.net/ S 8 Figure S4. Principal workflow of the PET experiment Figure S5. Photograph of the PET gantry with the soil column mounted vertically on the slide. Figure S6. Global break-through curve for the 64 Cu tracer (decay-corrected) as measured at the outflow by a gammacounter equipped with fraction sampler. The experiment #3 was terminated just after 50% of the tracer came out. The curve is extrapolated under assumption of 44% of tracer trapped (red curve) or no trapping (blue curve).