2.1 Chemicals

Carbofuran, certified for laboratory use and 99.5% pure, was supplied by Dr Ehrenstorfer, GmbH (Augsbourg, Germany). Its solubility limits are available in the literature: for example, it is 320 mg kg⁻¹ (certificate of analysis) at 25 °C in pure water.

Nearly saturated solutions of carbofuran were, first, prepared at the temperature chosen for the study and then diluted. A certain quantity of known weight and estimated to be less than the soluble one was, first, mixed with the chosen weight of purified water, then gently agitated in a thermoregulated oven for 15 days, the duration required for a complete dissolution of crystals [7,8]. All the prepared solutions were stored in the dark to prevent carbamate decomposition. Spontaneous pH was checked to fall within 4.7 and 5.5 and be far from the basic pH that induces decomposition.

Water was purified on a milliQ device from Millipore. Lead nitrate of analytical grade (more than 99.9% pure) was from Merck.

Silica was Degussa Aerosil 200 prepared by pyrogenation. Its specific area measured by BET adsorption isotherms for nitrogen was 196 m² g⁻¹. According to the supplier, silica particles had a mean diameter of about 15 nm and carried 3 OH groups per nm². To always operate in the same surface conditions, Aerosil 200 was dried at 195 °C during 1 week.

There were 10¹⁷ particles in 1 g of Aerosil. Most of the experiments were carried out on silica samples of 0.4 g mixed with about 20 cm³ of solution, or 0.2 g for 10 cm³. A rough calculation showed that the shortest distance between two particles in solution varied between 64.0 and 40.4 nm, distances sufficient to avoid capillary condensation.

The silica porosity, measured by BET adsorption with different gases was very low, the mean radius of pores was 66.88 Å (mesoporosity), and the microporosity was almost absent. According to the supplier, the distribution of particle diameters has a maximum at d = 100 Å and the particles are almost spherical. All the tests we have been able to do (with other types of silica, several very different adsorption values) showed that there was no interference, during centrifugation, between adsorption (molecules fixed on the surface) and the trapping of molecules between the packed silica particles.

2.2 Pb speciation in the studied samples

The Pb²⁺ ion is known to very easily form complexes [10]. This process is highly dependent on the solution pH. In the samples of concern here and prepared with de-ionised water, the most probable ligands were likely carbofuran, NO₃^– ion, CO₃^2– and HCO₃^– ions issued from dissolution of atmospheric CO₂. In water equilibrated with CO₂ (pH 5.5), addition of carbofuran and lead nitrate slightly elevated or lowered the pH by 1 or 2 tenth, depending on the salt concentration, whereas the solutions of pure carbofuran were at pH 7. According to the literature, such a pH is incompatible with the formation of a nitrate complex or that of a hydroxide [10] The polarographic analysis described hereafter showed no appreciable formation of Pb²⁺–carbofuran complex; as Pb²⁺ and NO₃^– were the only ionic species to be considered, the use of the equations given hereafter is thus justified.

2.3 Adsorption isotherm measurement

We used the classical depletion method to measure adsorption isotherms. Samples of silica powder (0.4 g) were weighed into centrifuge tubes before addition of 18 g of an aqueous solution of either carbofuran alone, or lead nitrate alone, or a mixture of both and smooth agitation during the time required for equilibrium. Previous determinations of this duration by adsorption kinetics experiments drove us to let adsorption develop for 40 h, a time sufficient to reach equilibrium for the three systems. During agitation, the samples were kept in a thermoregulated oven, at 25 °C; this temperature was known at more or less 0.2 K. Prior to adsorption determination, silica powder was removed from the suspensions by centrifugation at 2000 × g in a FirLabo centrifuge till getting a clear supernatant, which was then titrated.

Carbofuran was titrated at 276 nm with a KONTRON Uvikon 930 spectrophotometer; this wavelength corresponds to a peak on the spectrum. Let n⁰ be the total number of moles in the sample, x₂⁰ be the initial mole fraction of the solute, and x₂ be the equilibrium mole fraction of the supernatant after adsorption; then, the relative adsorption of solute 2 with respect to the solvent 1 is:

Lead nitrate was analysed by Differential Pulse Polarography (DPP) on a Hanging Mercury Drop Electrode (HMDE) [11]. To avoid interactions of metallic ions with carbonate and hydroxyl ions and to play the supporting electrolyte role, 15 μl of 10⁻² nitric acid were added to the sample before analysis. Other working conditions were as follows: a 50-mV pulse height and a 50-ms pulse duration. Before measurements, the solutions were always purged for 15 min with a nitrogen stream saturated with water vapour to remove oxygen.

3.1 Adsorption of lead nitrate alone

Fig. 1 shows the adsorption isotherm of lead nitrate on silica from aqueous solutions at 25 °C. Conversely to the case of solution/air interfaces and despite the still rather small adsorption, the salt is positively adsorbed instead of being depleted [2,9]. At this stage and on condition to agree with the classical explanation by Wagner–Onsager–Samaras, this positive adsorption is induced by the difference between the dielectric constant of silica and that of the air. One should remind that the profile of an ion i, at the interface between an aqueous solution (w) and the air (a), along an axis z perpendicular to the interface is [4]:

Then, the Gibbs excess is given by:

On condition to identify the surface salt content as Γ_Pb++ and whatever the bulk concentration, its concentration profile given by Eq. (4) is of the form displayed on Fig. 2. Indeed, the dielectric constant of silica, the medium where the image charge appears, is lower than that of water, like in the case of contact with air. Then, one gets:

Then, the water (w) excess at the silica/solution interface in presence of salt (s) is:

So, the surface layer contains a lot of water.

3.2 Adsorption of carbofuran alone

Adsorption of a single solution of carbofuran is rather well-known [6–8]. The adsorption isotherm obtained in this study and displayed on Fig. 3 is consistent with our previous experimental data [7,8]. The finding of a very low adsorption due to the negative Gibbs relative adsorption does not mean that the surface layer is devoid of carbofuran, but indicates that this layer is richer in water than bulk. This explains why carbofuran is quoted in the literature as “non-adsorbed” (implied “on soils”). The physico-chemical reason for that is that the organic molecule, which has affinity with the interface with air, considered as hydrophobic [7,8], is far less attracted by the silica surface, known to be covered with water molecules, particularly in aqueous suspensions.

3.3 Coadsorption of carbofuran and lead

Figs. 4 and 5, respectively, display relative adsorptions of lead nitrate and carbofuran, with respect to the concentrations of both components. The concentrations given on the axes are the equilibrium data, i.e. the supernatant concentrations once adsorption equilibrium had been reached. Fig. 4 shows an increase of salt adsorption with both salt and carbofuran concentrations; on the other hand, Fig. 4 highlights the weakness of carbofuran adsorption together with the lack of strong variations.

A comparison of Fig. 4 with Fig. 1 shows that lead nitrate adsorption is favoured by the presence of carbofuran: indeed, for the same bulk content, the relative adsorption values are approximately twice the initial ones. An examination of Fig. 5 obtained with data collected in the same run of experiments (coadsorption) indicates that the relative adsorption of carbofuran is still negative, but its absolute value is larger, which means a lower relative adsorption in presence of salt. This experimental result is very intriguing. Indeed, the solubility of the organic compound is reduced by the presence of salt in the solution, and theoretical and experimental demonstrations have shown that, at the air–solution interface, it usually results in an enhancement of the organic molecule adsorption [1,2,8].

3.4 Discussion

As clearly shown with solutions of lead nitrate alone or carbofuran alone, the surface layer in contact with silica is water-rich; this water-richness explains why ion adsorption is far much greater than at the solution/air interface: the transition between the solution and silica bulk (respectively, considered as media 1 and 2) seems to be more progressive than at the solution–air contact. We previously [1,2,9] argued that the value of the dielectric constant of a medium was affected by the existence of an organic layer between the air and a solution and that the reduction of the repulsive effect between a given ionic charge and its image could be explained by successive image-charges formed through the different interfaces (solution/organic layer, organic layer/air). Here, the situation seems to be far more complicated: indeed, the presence of organic molecules induces a consequent enhancement of ionic adsorption together with a reduction of organic adsorption. However, relation (5) indicates that the enhancement of ionic adsorption could be obtained by negative values of Γ_w.

Then the enhancement of relative ionic adsorption could be explained on assuming the presence of a sufficient amount of organic molecules in the ionic layer responsible for a hydrophobic effect at the origin of the repulsion of some among the water molecules.

If one considers the effect of a negative value of Γ_w, on the Γ_ow relative adsorption of the organic molecule (o) with respect to water (w), according to Eq. (6):

Fig. 6 summarises these views about the molecular organisation, and Fig. 7 highlights schematically the result of these hypothetical arrangements on the values of the image charges finally included in the equivalent of formula (2) when the presence of organic molecules is taken into account. Fig. 7a depicts the air–solution interface, where ions are imbedded in an organic layer (ε ≈ 30 as if carbofuran was pure) in the close vicinity of the surface, and Fig. 7b shows a water layer against silica followed by an ionic layer embedded in an organic layer. Image charges were calculated according to relation (7) applied to the different interfaces:

Finally, in the way of more or les statistical modelling, statistical simulations as done in Ref. [12] as well as quantum description of interactions [9] would be both extremely helpful. Of course, fundamental energetic data as immersion enthalpies obtained by microcalorimetry or following from adsorption studies with temperature could bring experimental confirmation to the theoretical arguments. Because of the intriguing particle arrangements in surface, kinetics measurements could also help to clarify our view of the problem.