The volcanic material (VM) was collected from natural accumulations in Villa La Angostura, Argentina (40˚45'48"S, 71˚38'46"W), five months after the eruption. In this area, the depth of accumulated pyroclasts was close to 30 cm. Bulk samples were dried (60˚C) and sieved (American Society for Testing and Materials mesh) to obtain fractions of particle size between 0.105 - 1.680 mm.

Adsorption experiments were done with water prepared from a standard As(V) solution containing 1 g of As L⁻¹ which was obtained by dissolving sodium hydrogen arsenate (Na₂HAsO₄∙7H₂O, Mallinckrodt) in distilled water. Water samples with minor arsenic concentration were obtained from the standard solution by dilution with tap water in order to ensure the mineralization of real water. In this sense, different results, attributed to the ionic strength of natural mineralization, are reported for the use of natural or deionized water [17] . Chemical composition of the tap water is given in Table 1.

The chemical activation of VM was carried out at room temperature by treatment with Fe(III) salts, following the technique described by Schwertmann and Cornell [23] to get ferrihydrite. The method consists in the combination of ash with solutions of FeCl₃∙6H₂O and KOH (conc. 1M). For each 100 g of VM, it was used 20 g of FeCl₃∙6H₂O dissolved in 250 mL water and 155 mL KOH to maintain Fe⁺³/OH⁻ = 1/3 ratio. KOH was added to adjust the pH to neutrality. The modified sample called Fe-VM was washed until complete removal of chloride, centrifuged and then oven dried at 60˚C. With comparative purposes, pure ferrihydrite was synthesized using the same technique.

The VM and Fe-VM chemical analyses were performed by ICP-AES technique for major elements. The ICP- MS analysis was employed for arsenic element (ALS Chemex Lab, Canada). X-ray Diffraction patterns were collected with a PHILIPS PW 1710 diffractometer using Cu KαNi-filtered radiation. SEM-EDS measurements were performed in an ESEM (FEI Quanta 200), with tungsten filament and ETD (high vacuum secondary electron) detector. Microanalysis was carried out with an EDAX Detector Apollo 40. Chemical results were expressed as %w/w oxides. The BET surface area was measured by N₂ adsorption using a Micromeritics ASAP 2020 automated Braunauer-Emmet-Teller Sorptometer. Samples were degassed for 10 h (P < 10 µm of Hg, temperature 100˚C). The specific surface (BET) was obtained from the N₂ adsorption isotherm (P/P₀ between 0.01 - 0.3) whereas the pore size distribution done by the DFT (Density Functional Theory) method was obtained in the pressure P/P₀ 0.01 - 0.99 range.

The As concentration in aqueous solutions was analyzed by means of an atomic absorption spectrometer (GF-AAS Perkin Elmer Analyst 200) equipped with a Perkin Elmer HGA 900 graphite-furnace. Turbidity was determined by using a turbidimeter Hanna HI 93703 whereas the pHmeter Denver Instrument Ultrabasic Benchtop was used to measure the water pH.

Batch experiments were carried out at room temperature (20˚C ± 2˚C) in jar test equipment (Velp Scientifica JLT6). The experiments were performed in duplicate and the mean values were considered (results show a variation of ±3%). Assays were performed to evaluate the effect of contact time with shaking at 150 rpm as well as the sedimentation time, the dose of adsorbent and the pH.

In order to find out the optimal contact time, experiments were carried out using an initial concentration As(V) of 0.500 mg∙L⁻¹ and a solid/liquid ratio 1/100 (7.5 g of adsorbent and 750 ml of solution). The pH was maintained at pH 7.75. The time was varied between 30 and 240 minutes using an agitation of 150 rpm. The turbidity and the solution pH were controlled. An aliquot of the solution was filtered through a 0.45 µm nylon filter and the filtrate was used for analysis of remaining arsenic concentration.

To determine the optimal adsorbent dose, experiments were carried out using adsorbent doses between 5-100 g adsorbent L⁻¹. The arsenic concentration was 0.500 mg∙L⁻¹, the solution pH 7.75 and the contact time 60 minutes at 150 rpm. After stirring, turbidity, pH and As concentration were measured in a procedure similar to that mentioned above.

Additionally, taking into account the turbidity reference value of ≤3 NTU and results obtained in previous works [12] , the stirring process for 60 min was supplemented by a 23 h settling period, so completing a total contact time of 24 hours (1 h at 150 rpm and 23 hours at rest). After this time, the same determinations were performed.

Studies were also conducted to find out the optimum pH range. The pH effect was evaluated in the range from 3.77 to 8.95. The pH of solutions was adjusted by adding 0.1M solutions of HCl or NaOH. Assays were performed at an initial As(V) concentration of 0.500 mg∙L⁻¹ for a stirring period of 60 minutes at 150 rpm and 23 sedimentation hours with an adsorbent dose of 40 g∙L⁻¹, according to previous results. Turbidity, pH and As concentration were determined.

In all experiments, the concentration of arsenic adsorbed on the solid was calculated by subtracting the final measured concentration (C_e) to the initial concentration of arsenic introduced in the solution (C₀). Results are given in percentage of arsenic removed (%Re):

Langmuir isotherms were conducted in order to investigate the maximum adsorption capacity Fe-VM taking into account the behavior of the unmodified material. Operating conditions optimized in the above items were used. Arsenic concentrations between 0.450 and 18.3 mg∙L⁻¹ were employed.

Table 2 presents the ICP-AES chemical data for VM and Fe-VM. The iron content increase was slightly lesser than 5% (expressed as %w/w of Fe₂O₃), while the SiO₂/Al₂O₃ ratio was comparable in both cases (4.4 and 4.5 for VM and Fe-VM respectively). These results can be correlated with some others obtained for natural species including aluminosilicates and volcanic ash [12] [17] . The low specific surface of these minerals (S_BET < 12 m²∙g⁻¹) limits the magnitude of the surface deposit of iron phase caused by the fast alkaline hydrolysis of Fe(III) salts.

Likewise, the As content in the original sample, measured by the ICP-MS technique, was 12.4 ppm. This value can be correlated with the original iron content, as it is observed in other aluminosilicates already studied (richer and poorer in iron) [12] .

On the other hand, the selected pyroclastic fraction was predominantly amorphous to X-ray corroborating that vitreous particles prevail respect to crystaloclastic and lithoclastic phases. XRD-patterns of iron activated samples did not present differences with those of natural samples. This fact could be attributed to the small amount of added iron phase and/or to its low crystallinity. Similar behavior is also observed in the activation of other aluminosilicates [12] [24] . Ferrihydrite is structurally poorly crystalline [25] . The two forms, named two and six line ferrihydrite, show a reduced number of XR signals, being the most intense at ~22˚ of 2θ [26] . Although the stability of two-line ferrihydrite is increased in certain conditions, preventing its transformation to crystalline Fe-oxide phases, hematite is the resulting alteration product, observed in nature or in laboratory. The transformation depends particularly on temperature and time [24] [26] .

SEM results for VM and Fe-VM are shown in Figure 1 a and b respectively. The typical vesicular morphology of the VM material was replaced by covered particles resulting from the deposition of iron phase on the surface and vesicles of Fe-VM material. EDS chemical data for original and activated materials is given in Table 3. The comparison between the bulk and surface results for VM and Fe-VM reveals the marked increase on the surface iron content after the iron activation, whereas SiO₂/Al₂O₃ ratios remain comparable. These values

LOI: weight loss on ignition at 1000˚C.

clearly show the effect of hydrated oxy-hydroxide iron phase deposition.

Data of specific surface by BET method follow the sequence VM < Fe-VM < ferrihydrite (0.76 < 4.94 < 322.44 m²∙g⁻¹). Figure 2 shows comparatively the pore size distribution for the three materials according to DFT method. Whereas the ferrihydrite used as reference presented prevalence of mesopores (average pore width 28.47 Å), VM material presented a wide distribution of meso and macropores (average pore width 107.29). Fe- VM activated sample showed an intermediate behavior, with an increase of mesopores and a pronounced decrease of macropores. This aspect was correlated to the higher specific surface by effect of iron-nanophase deposition, covering the original macro and mesopores. Likewise, the total pore volume was considerably increased (from 0.00178 to 0.00317 cm³∙g⁻¹).