Chemical fractionation of a terrestrial humic acid upon sorption on alumina by high resolution mass spectrometry

Abstract

Understanding the fractionation of humic acids (HA) during their sorption at mineral-solution interfaces is one of the major issues of soil and environmental sciences. Molecular-scale investigations have been conducted on the fractionation of a terrestrial HA (more precisely, of the water-soluble fraction of Aldrich HA denoted WSAHA) – rich in highly condensed aromatic compounds – during its sorption at acidic pH on alumina particles, which were taken as surrogates of Al oxide hydrates existing in soils. High-resolution mass spectrometry combined with electrospray ionization and atmospheric pressure chemical ionization was used for analysing WSAHA solutions before and after their contact with alumina. The sorption process was found to lead to enrichment of the highly reactive, acidic, oxygen-functionalized aromatic and aliphatic molecules, and of highly condensed aromatic compounds depleted in hydrogen carrying only a few oxygenated groups on the alumina surface. In contrast, the poorly oxygenated aliphatic constituents and aromatic compounds of O/C values in the range 0.2 to 0.5 remained preferentially in the solution. By comparing results obtained for homologous compounds whose elemental composition differed only by the number of CO₂ groups, evidence is found that both molecular acidity and degree of molecular hydrophobicity influence the degree of sorption (via ligand exchange on the surface and hydrophobic interactions, respectively) of WSAHA compounds of highly condensed aromatic type. Evidence at the molecular scale is provided that molecular acidity and hydrophobicity are the determining factors that control the size-fractionation of WSAHA during sorption on alumina.

---

Introduction

Humic substances (HS) such as humic acids (HA) and fulvic acids (FA) are components which are ubiquitous in soils and in surface waters and which originate from the chemical transformation of dead plants and animals and from microbiological activities. A key characteristic of HA and FA is their ability to bind on the surface of (nano) particles of clays and of metallic (oxihydr)oxides. It is widely recognized that during its sorption at the mineral-solution interface the HA/FA undergoes chemical fractionation, because it is a complex mixture made up by thousands of organic compounds showing a variety of compositions, of structures and of reactivities. The sorptive fractionation of HA and FA is expected to play a crucial role in the partitioning of heavy metals and organic micro-pollutants between the different compartments of the surface geochemical systems, as HS are known to form complexes with metals and to interact with hydrophobic organic molecules.^1–4 Therefore, acquiring a comprehensive description at the molecular level of the fractionation of HA and FA during the sorption processes is of interest for main issues of the environment such as the storage of CO₂ in soils and the coupling between the bio–geochemical cycles of carbon, of trace metallic elements and of organic pollutants in the natural systems.

The sorption of HS on mineral surfaces has been the subject of numerous studies, with many of them using spectroscopic techniques to gain information on the contrasted behaviour during the sorption processes of the different classes of organic compounds making up a HS.^2,5–8 The preferential sorption onto metallic oxides of the FA molecules containing a higher-than-average concentration of aromatic units activated by oxygenated functions has been reported.^8–11 It was found that among the dissolved organic molecules of a river, the molecules showing a high content of aromatic moieties and of carboxyl groups show for the surfaces of iron oxide an affinity that is stronger than that of the aliphatic fractions.¹² In a recent study by our group, the chemical fractionation of an aquatic fulvic acid (Suwannee River Fulvic acid, SRFA) in alumina-solution systems was described at the molecular scale by using ESI-FTMS for analysing the SRFA solutions, before and after the sorption process, and for deriving the chemical formulas of the molecules of SRFA adsorbed (or not adsorbed) on the alumina surface.¹³ Strong correlations were found to exist between the degree of sorption of a molecule of SRFA within a CO₂-series and its number of CO₂ groups, providing evidence that molecule acidity is a key parameter governing the sorptive fractionation. This finding added to the consistency of the aforementioned studies reporting a preferential sorption of the aromatic molecules of FA which carry carboxyl functions onto metallic oxides^8–12 and of other studies demonstrating the role of the strong acidity of simple organic compounds on their behaviour of sorption.^14,15 However, several authors have reported a behaviour of sorption of HS that contrasts with that mentioned above for FA. For example, Wang and Xing¹⁶ have shown with ¹³C NMR spectroscopy that the aliphatic fraction of the HA extracted from the Amherst peat was sorbed preferentially on the surfaces of kaolinite and montmorillonite, while the aromatic fraction remained in solution. Such variable behaviour points to the need of clarifying the influence of certain key parameters, such as surfaces of clay or a marked hydrophobic character of HS, on the sorptive fractionation of a humic substance in natural surface systems.

In the present paper, we address the description at the molecular scale of the sorptive fractionation of a HS characterized by a pole of very hydrophobic molecules. This work is a part of an extensive study by ESI-FTMS undertaken by our group¹³ to clarify to what extent some key characteristics of the molecules of HS, and here, more particularly, the degree of molecular hydrophobicity, govern the sorptive fractionation of HS, a main issue for soil. We investigated here the sorption of a terrestrial humic acid rich in highly condensed aromatics (namely, the water soluble fraction of the Aldrich humic acid extracted from peatland, denoted WSAHA) on the surface of colloids of aluminum oxide, taken as surrogates of Al-oxyhydroxides of soil. Batch experiments of WSAHA sorption were performed at acidic pH (3.7), at alumina-to-solution ratios from 4.9 to 15 g L⁻¹. Basic information on the behaviour of sorption of WSAHA was gained from the analysis of the supernatants by using TOC measurements and UV-Vis spectroscopy. The work focused particularly on acquiring knowledge at the molecular scale of the chemical nature of WSAHA, and of its sorptive fractionation in the alumina-solution systems investigated, by use of an advanced analytical technique, which is high-resolution Orbitrap Fourier transform mass spectrometry combined with electrospray ionization (ESI) and atmospheric pressure chemical ionization (APCI). First, specific knowledge of the exact elemental composition and of the chemical characteristics (molecular weight, degree of aromaticity…) of thousands of molecules constitutive of WSAHA was derived from the FTMS analysis of the initial WSAHA solution. FTMS data were also acquired for supernatants of the batch experiments of sorption, and examination was made of their differences and similarities with the data obtained for the initial solution of WSAHA. The comparison aimed to (a) the identification of molecules of WSAHA that were removed totally from the aqueous phase during the experiment of sorption, and (b) the calculation of the value of the parameter I for each molecule identified in the initial solution of WSAHA, with I being a parameter describing quantitatively the relative degree of sorption of a molecule on alumina. The FTMS results obtained in the present study provided valuable insights into the chemical fractionation taking place among the distinct types of organic compounds of a terrestrial humic acid during its sorption on the surface of aluminum oxide, and into the relationships between the chemical characteristics of the molecules of HA and the order of relative affinity of the molecules for the surface of the oxide mineral. Second, much attention was paid to elucidate whether the chemical fractionation of WSAHA during sorption might have led to an apparent size fractionation of the sample of WSAHA, as this type of phenomenon has already been reported in the literature for several HS-mineral-solution systems.^17,18 To this end, the native solution of WSAHA was fractionated into several sub-samples by applying a procedure of ultracentrifugation and FTMS analysis was performed on the different molecular size fractions collected. The results reported here provide a basis for a molecular scale description of the process of chemical and size fractionation of humic acids during their sorption on the surfaces of metallic oxides. Such information is necessary for building a realistic model for HS-mineral-solution interactions in soils.

Materials and methods

Materials

Alumina colloids were Alfa Aesar's α-Al₂O₃ crystallites (chemical purity: 99.95%, surface area: 7.6 m².g⁻¹) with a particle size of 280 ± 20 nm. The humic acid was purchased from Aldrich. The stock solution was obtained as follows: Aldrich humic acid was put in contact with water (2 g L⁻¹). After mixing for 2 h, the water soluble fraction (WSAHA) was separated from the insoluble residue by centrifugation during 1.5 h at 8000 rpm. WSAHA was purified from Na⁺ by passing the solution through an AG50Wx8 resin.

Sorption experiments

Batch experiments of WSAHA (90 mg L⁻¹) sorption onto alumina, at solid-to-solution ratio ranging from 4.9 to 25 g L⁻¹ were conducted in HDPE tubes using α-Al₂O₃ suspensions at pH 3.7 and 298 K. No attempt was made to maintain the pH during an experiment or to fix the solution's ionic strength, as salt addition was observed to considerably alter the ESI-MS response. After shaking the tubes in the dark for 24 h, the suspension was centrifuged at 8000 rpm during 1.5 hours for solution – colloid separation. Concentration of WSAHA in the supernatant was determined by total organic carbon measurements using colorimetric detection at 610 nm after persulfate digestion (Hach method, 3 replicates). The UV-visible spectra of solutions were recorded using a Cary 100 spectrometer (Varian). The UV-Vis spectral ratios E₂/E₃ and E₃/E₅, i.e., the ratio of absorbance at 250 and 365, and 350 and 550 nm, respectively, were determined for each solution.

Ultrafiltration

Original WSAHA solution was fractionated by filtration with Omega ultrafiltration membranes (modified polyethersulfone), with nominal molecular weight cut-off of 3 kDa and 10 kDa. Centrifugation at 7500g provided the driving force for filtration.

FT-MS measurements

Native solution of WSAHA and supernatants collected at the end of sorption experiments were analysed by ESI-FTMS and APCI-FTMS in the negative ionisation mode by using a linear ion trap-Orbitrap mass spectrometer, LTQ Orbitrap XL (Thermo Scientific). The solutions were introduced directly into the ESI/APCI probe with a syringe pump at a flow rate of 10 μL min⁻¹. Nitrogen was used as the drying and spraying gas. The temperature of the transfer tube was set at 275 °C. Voltage applied to the spray, the capillary, the multipoles and the tube lens was tuned to favour transmission of high-molecular-weight ions. MS spectra were recorded using the Orbitrap analyser, by averaging 100 scans in the ranges 120–798 m/z (no peak at m/z values higher than 798 was detected). Acquisition and treatment of data were made using the Xcalibur™ software (Thermo Scientific, version 2.1.0). Only peaks whose intensity exceeded twice the signal-to-noise ratio were considered as true peaks of the constitutive WSAHA compounds.

Kendrick mass defect analysis of MS datasets was undertaken to find series of molecules that are separated by integer multiples of a given functional group (CH₂, CO₂). For chemical formulas assignment, the Xcalibur MS calculation software was used. All possible formula attributable to a given odd m/z value were calculated by considering ¹²C, ¹H, ¹⁶O, ³²S atoms (with an upper limit of 200, 600, 50 and 1 respectively, for number of these atoms). Compounds peaking at even m/z values were considered to contain at most one ¹⁴N atom, in addition to C, H, O (and S) atoms. All formulas whose theoretical mass differs from 3 ppm or more from measured mass were rejected. For even m/z (corresponding to odd masses), peaks whose positions coincided with those of species including one ¹³C isotope have been excluded.

Results and discussion

1. Characterization of WSAHA

The description at the molecular level of the fractionation of WSAHA during its sorption on the surface of alumina requires at first a specific knowledge of the exact elemental composition and of the chemical characteristics of the thousands of molecules constitutive of the sample. Such information was derived here from the FTMS analysis of the native WSAHA solution. Fig. 1 shows the ESI and APCI mass spectra of the native WSAHA solution that were recorded in the negative ionization mode. Combining the analysis by ESI-FTMS and by APCI-FTMS gives a view of the heterogeneity of the sample of HS and allows to provide an accurate description of the humic substance. ESI is wellknown to ionize polar compounds with oxygen functions preferentially and APCI to facilitate the detection of oxygen-depleted molecules.

Fig. 1 FTMS spectra of WSAHA native solution recorded in the negative ionization mode, using an ESI (a) and an APCI (b) probe. Extended sections of the spectra in the range 500–798 m/z (c: ESI mode, d: APCI mode).

Both APCI(−) and ESI(−) FTMS spectra display a typical pattern, with a first series of intense peaks at odd m/z values and a second series (of peaks of lower intensity) at even m/z values. The full MS spectrum of WSAHA recorded using ESI reveals a multimodal distribution of peaks with m/z of the constituents not exceeding 750. The masses observable on the APCI(−)MS spectrum are in the range of m/z from 120 to 570. With the high resolving power of the Orbitrap analyser, 4980 and 2750 peaks were resolved in the ESI(−)FTMS and APCI(−)FTMS spectra, respectively. Of the 4980 peaks detected of the ESI(−)FTMS spectrum, 2049 peaks were also detected in the APCI-FTMS spectrum. It is to be noted that the magnitude of a peak is not necessarily correlated with the concentration of a compound due to biases related to the efficiencies of electrospray ionization and atmospheric pressure chemical ionization.

As is generally observed for natural organic matter,¹⁹ all species have been detected as singly charged entities with both modes of ionization. Indeed, the spacing between the peak of an ion containing uniquely ¹²C atoms and the peak for one containing a single ¹³C atom is equal to 1.0034 m/z in the case of a singly charged ion and equal to (1.0034/n) m/z (with n = 2, 3…) for multiply charged species. Only a spacing of 1.0034 m/z was observed on both the APCI-MS and ESI-MS spectra. Thus, no substances with a molecular weight over 750 Da were found in WSAHA. This finding is coherent with the concept proposed by Piccolo et al.²⁰ that humic substances are supramolecules, i.e. assemblies of low-molecular-weight units that are held together through van der Waals interactions and hydrogen bonding. Most of these weak interactions are disrupted during the ionization step in the ESI and APCI probes.

Thanks to the high mass accuracy of the Orbitrap analyzer, an elemental composition could be assigned to 71% of the compounds detected in both the ESI-FTMS and APCI-FTMS spectra. Most of the identified compounds contain only C, H and O atoms, which is consistent with the weak percentages of sulfur and of nitrogen (<2%) reported for the Aldrich humic acid.²¹ Identified compounds occupy broad fields of composition on van Krevelen's (VK) diagram, with O/C ratio and H/C ratio ranging from 0.04 to 0.95 and from 0.2 to 2, respectively (Fig. 2a and b). Combining analysis by ESI-FTMS and APCI-FTMS resulted in 12% more identification of elemental compositions in comparison with the use of electrospray ionization only, showing the interest of an approach using different and complementary methods of ionization to analyze HS. The large majority of compounds identified only on ESI(−) spectra contains multiple oxygenated functions, as they are characterized by an O/C ratio higher than 0.25 (Fig. 2c). The compounds detected by using both API sources, and whose elemental compositions include C, H, O and N atoms are gathered in the left part of the van Krevelen's diagram, with an O/C ratio exceeding hardly 0.5 (Fig. 2d and e). To compare the relative response of the ions when the mode ESI or the mode APCI is used, the intensity of the ions were normalized to the intensity of the molecular ion detected at m/z 241.0513 (this ion was chosen arbitrarily among those showing a signal-to-noise ratio higher than 10 in both modes of ionization). The most oxygenated compounds exhibit a higher intensity on the ESI-FTMS spectra than on the APCI-FTMS spectra (Fig. 2d). Conversely, compounds poor in oxygen and depleted in hydrogen give rise to a higher signal using APCI than by ESI (Fig. 2e). This trend is in agreement with the idea that the gas-phase ionization in APCI is more effective than the ESI to analyze the less polar species. Compounds detected only when using the APCI probe (Fig. 1f) occupy two areas of the van Krevelen's diagram. Most of them are located in the bottom-left part of the diagram and are highly unsaturated (H/C < 0.3). A few species lie in the VK region that is characteristic of fatty acids, as they are very poor in oxygen and exhibit only a few unsaturated bonds (with H/C ratio ranging from 1.8 to 2).

Fig. 2 van Krevelen diagrams showing the elemental composition of the identified WSAHA constituents: (a) compounds derived from ESI-MS datasets, (b) derived from APCI-MS datasets, (c) detected on ESI(−) spectra only, compounds detected on both ESI(−) and APCI(−) spectra and exhibiting (d) a higher or (e) a lower relative intensity when using ESI mode than the APCI mode, and (f) compounds detected only on APCI(−) spectra (cf. text). The symbols ● and o represent organosulfur molecules and S-free molecules, respectively.

Organosulfur compounds occupy two clusters on the VK diagram. One cluster gathers condensed aromatics with O/C < 0.65 and the other gathers aliphatic compounds with an O/C ratio higher than 0.33 and an H/C ratio ranging from 0.7 to 2.0 (Fig. 2a and b). The organosulfur compounds detected only with the APCI probe (Fig. 2f) are poorly or moderately oxygenated (Fig. 2f) whereas their congeners detected with the ESI probe show values of the O/C ratio extending up to 0.9. Most of the organosulfur compounds associated with the humic fraction of soils are thought to contain C-bonded sulfur.²²

Koch and Dittmar²³ proposed a mathematical formula for calculating the aromaticity index (A.I.) of a molecule (assuming that half of the oxygen is σ-bonded), and two threshold values of A.I. as unequivocal and minimum criteria for the existence of condensed aromatics at A.I. ≥ 0.67 and aromatics at 0.5 < A.I. < 0.67. Compounds whose A.I. is lower or equal to 0.5 have a more or less pronounced aliphatic character. More than 70% of the identified WSAHA constituents show an aromaticity index higher than 0.67 and belong thus to the family of condensed aromatics. Compounds having a more or less pronounced aliphatic character represent about 12% of the total of the identified species. The most oxygenated WSAHA constituents lie in this class of compounds. It must be underlined however that aliphatic compounds may be under-represented in the ESI and APCI mass spectra, because hydrophilic species are little transferred into the gas phase in the presence of hydrophobic species.

2. Sorption of WSAHA on α-alumina

2.1. Macroscopic description of the sorption

Before exploring at the molecular scale the chemical fractionation of WSAHA taking place during its sorption on alumina, information was acquired on the macroscopic behaviour of sorption of the humic acid. In the investigated acidic condition (pH = 3.7), the percentage of sorption of WSAHA on alumina was found to increase dramatically when the value of the solid-to-solution ratio (r) used for the experiment was increased from 4.9 to 25 g L⁻¹ (Fig. 3). For the latter value of r, it was observed a quasi-total removal of the organic matter from solution, that corresponds to a surface concentration of WSAHA of 750 μg m⁻²or ∼3 μ eq. m⁻² (given the approximate value of 4 meq. g⁻¹ reported in the literature for the total proton exchange capacity of the Aldrich humic acid ²⁴). Such a value of surface coverage by WSAHA is expected to exceed the density of the proton-active sites on the alumina surface (a reported estimation of the density value is ∼1.7 μmole per site per m^{2 25}). This suggests that the sorption of WSAHA cannot be explained only by interactions occurring between WSAHA compounds and the ionisable, singly-coordinated hydroxyl groups which are present on the surface of the alumina.

Fig. 3 Influence of the Al₂O₃-to-solution ratio (r) on the percentage of WSAHA sorption (◆) and on the overall ion current recorded during ESI(−)FTMS analysis of supernatants (o), for batch experiments of the sorption of WSAHA on alumina at pH 3.7 (r = 0 represents the native WSAHA solution). Experimental uncertainties on TOC measurements and OIC values were calculated on the base of 3 replicates.

UV-visible analyses of the supernatants collected at the end of the sorption experiments show that the overall absorbance in the UV-visible region decreased as the percentage of sorption increased (Fig. S1 in ESI†). A previous study has provided evidence for the existence of an inverse relationship between the E₃/E₅ ratio (with E₃/E₅ defined in the Experimental section) and the molecular size of fractionated HS.²⁶ It was also found that the E₂/E₃ ratio and the aromaticity of HS show a moderate linear correlation,²⁷ with increasing values of E₂/E₃ for a decrease of the aromaticity. Thus, the E₃/E₅ and E₂/E₃ ratios can be taken as global indicators of the size fractionation and of the chemical fractionation of HS, respectively. In the present experiments of the sorption of WSAHA on the alumina surface, the E₃/E₅ ratio of the initial solution was found to be equal to 6.5, whereas the supernatant solutions (i.e. the unbound WSAHA fraction) showed values ranging from 5.1 to 5.5, depending on the percentage of sorption. We note that the absorbance at 550 nm was too low to allow an accurate determination of the value of the E₃/E₅ ratio for the experiment at r = 25 g L⁻¹ which shows a 98% sorption. It was observed no significant difference in the value of the E₂/E₃ ratio (2.7 ± 0.1) between the initial solution of WSAHA and the supernatants, for r in the range 4.9–15 g L⁻¹ (which corresponds to an HS-to-solid ratio from 2.4 to 0.8 mg_HS m⁻²). Janot et al.²⁸ have previously reported an invariable E₂/E₃ index for the sorption of the purified Aldrich humic acid on α−Al₂O₃ at pH 6.8 and 0.1 M NaClO₄, when the experimental HS-to-solid ratio was greater than 1.3 mg_HS m⁻². For lower values of the HS-to-solid ratio, the authors found that the E₂/E₃ ratio was higher for the supernatants than for the initial solution. For the present experiments, UV-visible analyses did not show any enrichment of either aromatic or aliphatic moieties on the surface of alumina. However, they gave evidence that compounds characterized by a relatively low molecular weight are preferentially removed from solution during sorption. Such a preferential sorption of the low-molecular-weight fraction was already reported for the sorption of the Aldrich HA on hematite.¹⁷ Investigations at the molecular level are required to explore the relationships existing between the molecular weight, the chemical characteristics and the affinity for the alumina surface of the compounds of WSAHA, as well as to identify the mechanisms responsible for the trend to the fractionation in size observed here.

2.2. Molecular scale description of the sorption

2.2.1. Mechanisms involved in the fractionation of WSAHA. MS analysis showed a diminution of the overall ion current (OIC) from the initial solution of WSAHA to the supernatant solutions, which were collected at the end of the sorption experiments. The diminution was becoming larger consistently with the alumina-to-solution ratio (r) of an experiment and with the percentage of sorption of WSAHA (Fig. 3). Fig. 4a and b show the ESI-FTMS spectra of the supernatants of the experiments, for r = 4.9 and 10 g L⁻¹, respectively. Fig. 4c gives the APCI-FTMS spectrum of the supernatant of the experiment performed at r = 4.9 g L⁻¹. The comparison of these spectra with those of the native WSAHA solution revealed that the magnitude of the decrease of signal intensity from the native solution to the supernatant varies considerably from one compound to another, indicating that the constituents of the humic substance are partitioned between the solution and alumina surface to varying degrees. The species showing a decrease in the intensity of their signal in solution (D_IS) which is more twice the decrease of the overall ion current (D_OIC) are considered as the species with a strong affinity for alumina. Fig. 5–7 give the VK diagrams derived from examination of the ESI(−)MS supernatants datasets of experiments at r = 4.9 g L⁻¹ and 10 g L⁻¹ (in Fig. 4a and b, respectively) and the APCI(−)MS supernatant dataset of experiment at r = 4.9 g L⁻¹ (Fig. 4c). The diagrams show plots of the elemental compositions of the identified constituents of WSAHA, sorted by a decrease of their relative affinity for the surface of alumina, i.e., in decreasing order of reduced signal from native solution to supernatant. One observes that the number of species totally sorbed at the alumina surface increases with percentage of WSAHA sorption. For example, all the species identified as having a strong affinity for the surface of alumina (Fig. 5b) in the experiment at r = 4.9 g L⁻¹ (for which the percentage of sorption of WSAHA was 31%) were sorbed totally (Fig. 6a) in the experiment at r = 10 g L⁻¹ (for which the percentage of sorption was 52%). This trend suggests that a limited number of reactive sites are available on the surface of alumina and that high-affinity compounds compete for these sites at a low solid-to-solution ratio.

Fig. 4 ESI(−)FTMS spectra of the supernatants collected at the end of WSAHA sorption experiments conducted at pH 3.7 for an Al₂O₃-to-solution ratio of 4.9 g L⁻¹ (a) and 10 g L⁻¹ (b) and APCI(−)FTMS spectrum for pH = 3.7 and r = 4.9 g L⁻¹.

Fig. 5 ESI(−)FTMS results on the sorptive fractionation of WSAHA in alumina-solution systems at pH 3.7 for r = 4.9 g L⁻¹: van Krevelen plots for compounds (a) totally sorbed, (b) with a strong affinity for the surface (with D_IS > 2D_OIC), (c) with D_OIC < D_IS < 2D_OIC, (d) with and (e) with . D_IS and D_OIC: decrease, from WSAHA solution to supernatant, in the intensity of the signal of the species and in overall ion current, respectively.

Fig. 6 ESI(−)FTMS data on the sorptive fractionation of WSAHA on alumina at pH 3.7 and at r equal to 10 g L⁻¹: van Krevelen plots for compounds (a) quantitatively sorbed, (b) with a strong affinity for the surface (with D_IS > 2D_OIC), (c) with D_OIC < D_IS < 2D_OIC, (d) with and (e) with . D_IS and D_OIC: decrease, from WSAHA solution to supernatant, in the intensity of the signal of the species and in overall ion current, respectively.

Fig. 7 APCI(−)FTMS data on the sorptive fractionation of WSAHA on alumina at pH 3.7 for r = 4.9 g L⁻¹: van Krevelen plots for compounds (a) totally sorbed or (b) not totally sorbed and detected only using the APCI probe.

In previous ESI-FTMS work,¹³ we studied the sorption on the aluminum oxide under acidic conditions of an aquatic fulvic acid (SRFA) which contains a small quantity of polycyclic aromatic compounds. Evidence was given for a preferential sorption of the acidic, oxygen-functionalized molecules (O/C > 0.25) and of condensed aromatics depleted in oxygen, while the aliphatic compounds carrying few oxygenated groups remained mainly in the dissolved phase. There are similarities in the sorptive fractionation trends of SRFA and WSAHA. The data in Fig. 5a and b, 6a and b and 7a show for both solid-to-solution ratios of 4.9 and 10 g L⁻¹ that a major part of the aromatics (condensed and not condensed) containing only a few oxygenated functions (O/C < 0.15) was sorbed on the alumina, or they showed a strong affinity for surface. Many constituents of WSAHA (both aliphatic and aromatic) characterized by high values of the O/C ratios (O/C > 0.6) carrying multiple oxygenated functional groups are prone to sorption. In contrast, it is observable in Fig. 5c, 6c and 7b that most of the less oxygenated (O/C < 0.4) aliphatic species were comparatively poorly sorbed. Compounds bearing N or S heteroatoms show a behavior of sorption that is similar to that of S- or N-free molecules. However, an important difference exists between the trends in the sorptive fractionation of the aquatic fulvic acid SRFA and the terrestrial humic acid WSAHA. Contrary to SRFA, most of the condensed aromatics with an intermediate O/C ratio (0.15 ≤ O/C ≤ 0.6) display in the WSAHA sample a low affinity for alumina surface. Their affinity is even lower than that observed for the poorly oxygenated (O/C < 0.4) aliphatic species (Fig. 5d and 6d).

For each ion detected in the native solution of WSAHA, we calculated the value of the parameter I, previously proposed by Galindo and Del Nero¹³ as a trend indicator for sorption among HS ions. The parameter I is defined as the ratio between the ion intensity in the supernatant divided by the sum of the intensities for all detected ions in the supernatant to the normalized intensity in the native solution. The lower the value of I, the greater the affinity of the ion for the surface of alumina, with I = 0 for the species totally sorbed on the mineral surface. Fig. 8 presents a plot of parameter I, calculated from the ESI-FTMS datasets of the experiment of sorption at r = 4.9 g L⁻¹, as a function of the O/C ratio of a molecule in a –CO₂ homologous series, i.e., in a series of species whose chemical formulas differ only by their number of –CO₂ groups. Several CO₂ series, containing up to 7 members that are observable in the ESI-FTMS spectrum of WSAHA as 43.9898 Da repeating units, are considered here (species detected only by using the APCI probe were not taken into consideration as the CO₂ series contains only a small number of members). Fig. 8 shows that for most of the series composed of either condensed polycyclic aromatics or of not condensed aromatics, the member of the series with the lowest molecular weight (i.e., the first member) is characterized by I < 0.58. Such low values correspond to the species that showed a diminution after the process of sorption of the intensity of their signal in solution greater than that of the overall ion current (D_IS > D_OIC). Values of I lower than 0.58 designate thereby species with a strong affinity for the surface of alumina. In contrast, for the series containing compounds with a more or less pronounced aliphatic character, the first member is characterized by I > 0.58. For both types of series, there is observed systematically an increase of the value of I with increasing O/C ratio of a compound in a series, as long as the O/C ratio is lower than 0.4–0.5. Above this value, the opposite trend is observed. That is, the value of I is inversely correlated with the O/C ratio. Hence, for series including compounds with an O/C ratio in the range 0–0.6, the last member of the series generally shows a value of I > 0.58, so its affinity for alumina is moderate or low. For the series including compounds with O/C ratios up to 1, the most oxygenated members of the series show a high degree of sorption (I < 0.58). Sometimes, they are even totally sorbed (I = 0). The finding of good correlations between the parameter I and the O/C ratio of a compound in a CO₂-series hints that the number of the oxygenated functional groups carried by a molecule is a key parameter governing the sorptive fractionation of WSAHA. The opposite trend in the dependence of I on the O/C ratio, which are observable for compounds having O/C below and above a value of 0.4–0.5, respectively, suggest the existence of at least two types of mechanisms of sorption having opposite effects on the variation of the degree of affinity for the surface as a function of number of carboxyl groups carried by a molecule.

Fig. 8 Variation of parameter I as a function of the O/C ratio of molecules in the CO₂-homologous series characterized by a modulus of −44 (for MS data collected during the sorption experiment performed at an Al₂O₃-to-solution ratio of 4.9 g L⁻¹). Full symbols represent compounds with an aromatic character and empty symbols represent compounds with an aliphatic character. Experimental uncertainties on the values of parameter I were estimated at 11% (3 replicates).

The trend of decreasing I with an increasing O/C ratio observed for molecules with an O/C ratio higher than 0.4–0.5 (Fig. 8) supports the idea that the most oxygenated organic molecules are sorbed on the surface of alumina by chemical bonding involving the carboxyl groups of WSAHA molecules, which possibly depends on exchange of surface ligands between the aluminol sites of alumina and the carboxyls. However, it is expected that only surface complexes with a structure involving one, two, or potentially three carboxyl groups would form on the mineral surface. It has been previously suggested that the enhanced sorption observed for simple organic acids with an increasing number of carboxyl groups does not result from multiple bonding at surface, but is rather due to changes in molecular acidity.¹⁵ In a previous ESI-FTMS study by our group,¹³ evidence was provided that molecular acidity is a main parameter governing the fractionation of an aquatic fulvic acid (SRFA) depleted in condensed aromatics during its sorption on alumina under acidic conditions. It is likewise concluded in the present work that the degree of sorption of the most oxygenated molecules in WSAHA is controlled by a molecular acidity.

In contrast, the poorly oxygenated aliphatic species and the less polar condensed aromatics, which show a trend of increasing I value with increasing O/C ratio (up to 0.4–0.5) in a CO₂-series (Fig. 8), are more likely driven to the surface of α-alumina by hydrophobic interactions (i.e., the repulsion of hydrophobic moieties from aqueous solution). The inverse correlation observable between the O/C ratio of such molecules and their degree of sorption (Fig. 8) is well accounted for by this interpretation, because increasing of the number of oxygenated functions on the carbon backbone has the effect of reducing a molecule's hydrophobic character. Hydrophobic interactions are likely favoured at low pH of solution where many carboxylic groups and phenolic/alcoholic groups are protonated and uncharged. The highest sorption degree of the less polar condensed polycyclic aromatics compared to aliphatic species having similar O/C values can be explained by the fact that condensed polycyclic aromatics with π-donor properties also engage in π–π type associations with π-acceptor congeners sorbed on the alumina. More generally, π-donor and π-acceptor compounds show a higher sorption compared to non π-donors/acceptors. Reported energies for π–π interactions are larger than expected for van der Waals forces typically associated with hydrophobic interactions.^29,30 As outlined by Keiluweit and Kleber²⁹ in their review on the mechanisms of retention of aromatic molecules on minerals, π–π interactions have the potential of rendering the attachment of aromatic molecules to sorbent more favourable (even if other mechanisms govern the sorption behaviour).

The bell shape of I = f(O/C) clearly indicates that the sorptive fractionation pattern of WSAHA is controlled by both ligand exchange reactions and hydrophobic interactions (enhanced by π–π interactions for molecules with aromatic moieties), with the predominant mechanism of the sorption of a WSAHA compound depending on its chemical nature. This explains why the molecules exhibiting an intermediate value of the O/C ratio (i.e., the molecules that are neither too hydrophobic, nor too acidic) are the less prone to sorption among WSAHA constituents.

2.2.2. Effect of the chemical fractionation on WSAHA size distribution between surface and solution. Further molecular scale investigations were made to clarify the relations existing between the molecular weight of the WSAHA compounds, their chemical characteristics and their affinity for the alumina surface. As mentioned in Section 2.1, UV-vis analysis of macroscopic sorption samples of WSAHA suggests that the WSAHA–alumina surface interactions taking place under acidic conditions results in a size fractionation of WSAHA, with the molecules of lower molecular weight (MW) being preferentially sorbed onto alumina. Hur and Schlautman³¹ studied the sorption of purified Aldrich humic acid on hematite under acidic and near neutral conditions by means of size-exclusion chromatography with UV detection. The authors observed a similar fractionation pattern and postulated that the adsorption of the lower MW fractions of humic acid is governed by ligand-exchange/electrostatic interactions, on the basis of the observation that the preferential sorption of small MW components was reduced in the presence of phosphate ions.

The VK diagrams for the three size fractions (<3 kDa, <10 kDa and >10 kDa) collected after an ultracentrifugation procedure of the native WSAHA solution (cf. Experimental section) are presented in Fig. 9. The diagrams show that the elemental composition of the compounds detected in the filtrates differed from those of the retentate. An overwhelming majority of the aliphatic species carrying multiple carboxyl groups and alcoholic groups was retrieved in the 3 kDa-and 10 kDa filtrates and was only sparingly detected in the retentate. As shown previously, these compounds show a high affinity for the alumina surface and interact with it through reactions of ligand exchange involving the aluminol surface groups on Al₂O₃ (Section 2.2.1). Therefore, the highly-sorbed acidic compounds exist in the native WSAHA solution as individual entities or small-size supramolecules.

Fig. 9 van Krevelen diagram derived from the ESI-FTMS datasets obtained for the filtrate using a 3 kDa-membrane (a) or a 10 kDa-membrane (b) and for the retentate (cut-off of 10 kDa), (c) after ultrafiltration of WSAHA native solution. The symbols ● and o represent organosulfur molecules and S-free molecules, respectively.

In contrast, the condensed aromatics characterized by an intermediate O/C ratio were not detected in the <3 kDa-filtrate (Fig. S2 in ESI†), indicating that in the WSAHA solution such compounds participate in the formation of supramolecular assemblies too large to pass through the membrane porosity. Therefore, they account for a significant part of the high molecular weight fraction. These compounds were poorly sorbed (Fig. 5d and 6d), as they were neither hydrophobic enough to be retained on the surface of alumina via hydrophobic interactions, nor acidic enough to form strong surface complexes by surface ligand exchange (Fig. 8 and Section 2.2.1).

In summary, by identifying specifically the compounds of WSAHA having a high affinity for the surface of alumina, evidence at the molecular scale is provided here that the differences observed in the chemical and structural trends of the low-molecular-weight and high-molecular-weight fractions of WSAHA are the determining factors that direct the fractionation in size of WSAHA during its sorption.

Conclusion

The present study provides data at the molecular scale from which conclusions could be drawn with regard to the interactions occurring between the molecules constitutive of the WSAHA humic acid and the surface of alumina. Within a CO₂-series of WSAHA compounds, the degree of sorption of a molecule was found to increase with increased oxygen content of the most polar organic compounds, whereas the opposite trend was observed for the nonpolar species. Such behaviours are consistent with a sorptive fractionation pattern that is controlled by both the acidity and the hydrophobic character of the molecules constituting HS, the organic molecules being driven to the alumina surface either by ligand-exchange reactions or by hydrophobic interactions. By comparing these results with those reported for the sorption of an aquatic fulvic acid on alumina under similar experimental conditions,¹³ there was observed a heightened contribution of the hydrophobic interactions for WSAHA, attributable to the highest content in highly condensed aromatic compounds of this terrestrial humic substance. This implies that the variety of interactions taking place on an oxide surface is largely determined by the amount, nature and reactivity of the functional groups present in HS.

The present study highlights that acquiring thorough knowledge of the chemical characteristics of individual HS compounds is of paramount importance to understand and predict the sorptive fractionation patterns of complex mixtures such as humic substances in geochemical systems.

Acknowledgements

This work was supported financially by the IPHC and the REALISE network of the Alsace Region. We thank the two anonymous reviewers and Dr A. Pape for their helpful comments.

References

1. L. P. Weng, W. H. van Riemsdijk, L. K. Koopal and T. Hiemstra, Adsorption of humic substances on goethite: comparison between humic acids and fulvic acids, Environ. Sci. Technol., 2006, 40, 7494–7500.
2. X. Feng, A. J. Simpson and M. J. Simpson, Chemical and mineralogical controls on humic acid sorption to clay mineral surfaces, Org. Geochem., 2005, 36, 1553–1566.
3. K. D. Jones and C. L. Tiller, Effect of Solution Chemistry on the Extent of Binding of Phenanthrene by a Soil Humic Acid:  A Comparison of Dissolved and Clay Bound Humic, Environ. Sci. Technol., 1999, 33, 580–587.
4. S. A. Wood, The role of humic substances in the transport and fixation of metals of economic interest (Au, Pt, Pd, U, V) Ore, Ore Geol. Rev., 1996, 11, 1–31.
5. K. Kaiser, G. Guggenberger, L. Haumaier and W. Zech, Dissolved organic matter sorption on sub soils and minerals studied by 13C-NMR and DRIFT spectroscopy, Eur. J. Soil Sci., 1997, 48, 301–310.
6. T. H. Yoon, S. B. Johnson and E. Brown Jr, Adsorption of Suwannee River Fulvic Acid on Aluminum Oxyhydroxide Surfaces:  An In Situ ATR-FTIR Study, Langmuir, 2004, 20, 5655–5658.
7. F. Claret, T. Schäfer, J. Brevet and P. E. Reiller, Fractionation of Suwannee River Fulvic Acid and Aldrich Humic Acid on α-Al₂O₃: Spectroscopic Evidence, Environ. Sci. Technol., 2008, 42, 8809–8815.
8. X. H. Guan, C. Shang and G. H. Chen, ATR-FTIR investigation of the role of phenolic groups in the interaction of some NOM model compounds with aluminum hydroxide, Chemosphere, 2006, 65, 2074–2081.
9. H.-B. Fu, X. Quan, S. Chen and Y.-Z. Zhao, Interaction of humic substances and hematite : FTIR study, J. Environ. Sci., 2005, 17(1), 43–47.
10. S. Ghosh, Z. Y. Wang, S. Kang, P. C. Bhowmik and B. S. Xing, Sorption and fractionation of a peat derived humic acid by kaolinite, montmorillonite, and goethite, Pedosphere, 2009, 19(1), 21–30.
11. K. Kaiser, Sorption of natural organic matter fractions to goethite (α-FeOOH): effect of chemical composition as revealed by liquid-state ¹³C NMR and wet-chemical analysis, Org. Geochem., 2003, 34(11), 1569–1579.
12. D. M. McKnight, K. E. Bencala, G. W. Zellweger, G. R. Aiken, G. L. Feder and K. A. Thorn, Sorption of dissolved organic carbon by hydrous aluminum and iron oxides occurring at the confluence of deer creek with the Snake River, Summit County, Colorado, Environ. Sci. Technol., 1992, 26, 1388–1396.
13. C. Galindo and M. Del Nero, Molecular Level Description of the Sorptive Fractionation of a Fulvic Acid on Aluminum Oxide Using Electrospray Ionization Fourier Transform Mass Spectrometry, Environ. Sci. Technol., 2014, 48, 7401–7408.
14. M. Edwards, M. M. Benjamin and J. N. Ryan, Role of organic acidity in sorption of natural organic matter (NOM) to oxide surfaces, Colloids Surf., A, 1996, 107, 297–307.
15. C. R. Evanko and D. A. Dzombak, Influence of structural features on sorption of NOM-analogue organic acids to goethite, Environ. Sci. Technol., 1998, 32, 2846–2855.
16. K. Wang and B. Xing, Structural and Sorption Characteristics of Adsorbed Humic Acid on Clay Minerals, J. Environ. Qual., 2005, 34, 342–349.
17. J. Hur and M. A. Schlautman, Molecular weight fractionation of humic substances by adsorption onto minerals, J. Colloid Interface Sci., 2003, 264, 313–321.
18. M. Meier, K. Namjesnik-Dejanovic, P. A. Maurice, Y.-P. Chin and G. R. Aiken, Fractionation of aquatic natural organic matter upon sorption to goethite and kaolinite, Chem. Geol., 1999, 157, 275–284.
19. A. C. Stenson, A. G. Marshall and W. T. Cooper, Exact Masses and Chemical Formulas of Individual Suwannee River Fulvic Acids from Ultrahigh Resolution Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectra, Anal. Chem., 2003, 75, 1275–1284.
20. A. Piccolo, The supramolecular structure of humic substances : a novel understanding of humus chemistry and implcations in soil science, Adv. Agron., 2002, 75, 57–134.
21. A. Rodrigues, A. Brito, P. Janknecht, M. F. Proença and R. Nogueira, Quantification of humic acids in surface water: effects of divalent cations, pH, and filtration, J. Environ. Monit., 2009, 11, 377–382.
22. W. B. McGill and C. V. Cole, Comparative aspects of cycling of organic C, N, S and P through soil organic matter, Geoderma, 1981, 26, 267–286.
23. B. Koch and T. Dittmar, From mass to structure: an aromaticity index for high-resolution mass data of natural organic matter, Rapid Commun. Mass Spectrom., 2006, 20, 926–932.
24. S. Pompe, M. Bubner, K. H. Heise, G. Bernhard and H. Nitsche, Influence of humic acids on the migration behavior of radioactive and non-radioactive substances under conditions close to nature, Forschungszent. Rossendorf, Wiss.-Tech. Ber., 2000, http://nbn-resolving.de/urn:nbn:de:bsz:d120-qucosa-30193.
25. X. Yang, Z. Sun, D. Wanga and W. Forsling, Surface acid–base properties and hydration/dehydration mechanisms of aluminum (hydr)oxides, J. Colloid Interface Sci., 2007, 308, 395–404.
26. D. P. Dick and P. Burba, Extraction kinetics and molecular size fractionation of humic substances from two Brazilian soils, J. Braz. Chem. Soc., 1999, 10, 146–152.
27. J. Peuravuori and K. Pihlaja, Molecular size distribution and spectroscopic properties of aquatic humic substances, Anal. Chim. Acta, 1997, 337, 133–149.
28. N. Janot, P. E. Reiller, X. Zheng, J.-P. Croue and M. F. Benedetti, Characterization of humic acid reactivity modifications due to adsorption onto a-Al₂O₃, Water Res., 2012, 46, 731–740.
29. M. Keiluweit and M. Kleber, Molecular-level interactions in soils and sediments: the role of aromatic π-systems, Environ. Sci. Technol., 2009, 43, 3421–3429.
30. T. Reemtsma, A. These, A. Springer and M. Linscheid, Differences in the molecular composition of fulvic acid size fractions detected by size-exclusion chromatography–on line Fourier transform ion cyclotron resonance (FTICR–) mass spectrometry, Water Res., 2008, 42, 63–72.
31. J. Hur and M. A. Schlautman, Effects of pH and phosphate on the adsorptive fractionation of purified Aldrich humic acid on kaolinite and hematite, J. Colloid Interface Sci., 2004, 277, 264–270.

---

Footnote

† Electronic supplementary information (ESI) available: Fig. S1 compares the UV-visible spectra of WSAHA native solution and those of the supernatants obtained after the sorption experiments performed at various Al₂O₃-to-solution ratio. Fig. S2 presents the van Krevelen diagram of the WSAHA constituents that show a low affinity for alumina surface and their partitioning between the filtrate and the retentate upon ultrafiltration using a 3 kD membrane. See DOI: 10.1039/c5ra12091h

---