Role of pH in the Aqueous Phase Reactivity of Zerovalent Iron Nanoparticles with Acid Orange 7, a Model Molecule of Azo Dyes

Citation Freyria, F. S.; Esposito, S.; Armandi, M.; Deorsola, F.; Garrone, E. and Bonelli, B. "Role of pH in the Aqueous Phase Reactivity of Zerovalent Iron Nanoparticles with Acid Orange 7, a Model Molecule of Azo Dyes." Journal of Nanomaterials Volume 2017 2749575 (June 2017): 1-13 © 2017 F. S. Freyria et al As Published http://dx.doi.org/10.1155/2017/2749575 Publisher Hindawi Publishing Corporation


Introduction
Synthetic dyes are used in many fields, such as textile, plastic, cosmetic, and tannery industries: approximately 800.000 tons of dyes are produced every year all over the world, their global demand being predicted yet to grow in the next years [1][2][3][4].
For on field environmental applications, industrial quantities of NZVI slurries are, however, necessary: in the past, Toda Kogyo Corp. was the only company producing industrial quantities of NZVI slurries under the name of RNIP (Reactive Nanoscale Iron Particles), whereas more recently other companies started NZVI production [53].
Whatever its source, NZVI remains, so far, a poorly characterized system, normally consisting of an aqueous slurry featuring a solid phase and, possibly, a stabilizer [30].
As it concerns the solid phase, physicochemical characterization by means of several microscopic and spectroscopic techniques determined that NZVI mainly consists of coreshell particles, with a core of -Fe 0 particles and an outer shell of Fe  O  [54][55][56][57].The shell consists of mixed Fe 2+ /Fe 3+ oxides, usually richer in Fe 3+ ions (i.e., more oxidized) at the water/shell interface [58,59].On the one hand, totally oxidized surfaces (i.e., where only Fe 3+ ions occur) result in inactive towards pollutants degradation; on the other hand the core-shell structure is fundamental for reactivity, as both the defectiveness and disorder of Fe  O  layer make NZVI more active than bulk iron materials covered by "nondefective"-oxide layers [59].The reaction mechanism acknowledged by the literature for the process of contaminant removal encompasses both the metallic iron core, acting as the electron source, and the oxide shell [38], facilitating sorption of contaminants while ensuring electrons transport from the metal core to the target contaminant [60][61][62].
In a previous paper [30], we investigated the degradation of the azo dye Acid Orange 7 (AO7, Scheme 1) by two RNIP suspensions commercialized at pH values around 12. Those experiments were run in slightly basic conditions, as obtained after adding RNIP to a water solution of 0.67 mM AO7 (the latter having a natural pH equal to 6.8, i.e., close to neutrality).The performance of RNIP suspensions towards AO7 degradation resulted to depend on both aging (i.e., nanoparticles oxidation by exposure to air) and the presence of a protective organic polymer in one of them.The mechanism of interaction with AO7 was instead affected by the degree of surface oxidation [30].In summary, on oxidized (aged) surfaces, mere adsorption of AO7 moieties took place and surface oxidation was partially prevented in polymer protected RNIP, which was able to degrade AO7 through the complex redox mechanism reported in Scheme 3.
Surface oxidation occurs normally as a consequence of aging, not only decreasing NZVI efficiency in remediation, but also modifying its toxicity towards earthworms and other organisms [63,64].For this reason, in this work we studied a slurry treated with HCl with the purpose of (partially/totally) removing the oxide shell, eventually promoting the activity of inner metallic iron.
Concerning pH of the aqueous phase, another important parameter affecting remediation efficiency, other slurries are commercialized at nearly neutral pH values.The role of pH is crucial, as it may influence both the reactivity with the pollutants and the actual composition of NZVI.In this work, the overall role of pH in the reactions of AO7 with NZVI is investigated, with the purpose of drawing a comprehensive picture of the reactivity of the NZVI/AO7 system, in comparison with previous work where the degradation of AO7 was studied in basic conditions [30].

Materials and Physicochemical Characterization.
The slurry (commercial name: NANOFER 25, provided by Nano Iron Future Technology, Czech Republic) was studied both as received and after 12 months.Both slurries were stored in a fridge to preserve their properties by preventing/slowing down oxidation phenomena, but in the latter case, the bottle was opened in order to study the effect of aging by natural oxidation in atmospheric conditions.
On the fresh slurry, an acidic treatment was carried out by sonication of ca.0.204 g slurry for 5 min, followed by filtration and washing with 5.0 mL of 1.0 M HCl solution and final drying.
For NPs physicochemical characterization, solid fractions were collected by filtrating the slurry (Anopore inorganic membrane with 0.02 m pore size, Whatman), washing the solid with acetone, and drying in air at ambient conditions [55].
The corresponding powder samples will be referred to as Fe_fresh and Fe_aged (for the solids recovered from the asreceived slurry and after 12 months) and as Fe_fresh-HCl (for the solid obtained after HCl treatment).
The physicochemical properties of the solids were characterized as follows: (i) Powders XRD patterns were measured on an X'Pert Philips diffractometer (PANalytical B.V.), using Cu K  radiation in the 10 : the peaks at 310 nm and 230 nm (and the shoulder at 260 nm) are due to aromatic ring absorptions.The peak at 484 nm is due to - * transition involving the lone pair of N atoms at the conjugated system extending over the two aromatic moieties and encompassing the N-N group of the hydrazone form [5,52,66].The shoulder at 403 nm has a similar nature, involving the N-N group of the azo form [8]. Degradation experiments were run either at natural pH of the solution (6.8) or by adjusting it to 10.0, 8.0, 4.0, and 2.0 by adding proper amounts of either 0.01 M NaOH or 0.01 M HCl.The latter solutions were prepared by dissolving either NaOH (anhydrous pellets from BioXtra, Italy, with purity ≥ 98%) or hydrochloric acid (37% w/w, ACS reagent, Sigma Aldrich, Italy) in doubly distilled water (Carlo Erba, Italy).
The interaction of AO7 with NZVI samples was studied in air at room temperature by adding 0.5-1 g L −1 NZVI dispersion to 400 ml of 0.67 mM AO7 solution, then by stirring with an orbital shaker (MPM instruments).Before reaction, the NZVI dispersions were sonicated for ca. 5 minutes in order to favour NPs dispersion.
Aliquots of the suspension were collected at different time intervals and the supernatant solution was separated by centrifugation (ALC centrifuge PK110, at 4000 rpm for 2 minutes).After each sampling, the UV-Vis spectrum of the supernatant was recorded in the 190-800 nm range on a Cary 5000 spectrophotometer (Varian instrument), equipped with a quartz cell with 1 mm path length.
The (residual) concentration of AO7 in the supernatant was evaluated by the intensity of the band at 484 nm, related to the dye hydrazone form in equilibrium with the azo form in water, as depicted in Scheme 1.
To have first-hand spectroscopic data concerning 1amino-2-naphthol (1A2N), one of the possible products of AO7 degradation [5], the corresponding chlorhydrate (Merck) was dissolved in water, and its UV-Vis spectrum was taken over a wide pH range.This procedure allowed us to discriminate between the spectrum of the neutral form, the protonated one, and the anionic one (Figure SI1 in Supplementary Material available online at https://doi.org/10.1155/2017/2749575).Similarly, the possibility also investigated that 1A2N may form complexes with either Fe 3+ or Fe 2+ ions in solution, by taking UV-Vis spectra in the presence of such cations (Figure SI2).

Physicochemical Characterization of the Solid Fraction.
Figure 1 reports XRD patterns of samples both before (sections (a), (b), and (c)) and after reaction (sections (d) and (e)): the peaks due to -Fe 0 and Fe 3 O 4 phases are observed with all the samples.Additional peaks ascribable to another phase (vide infra) are observed to occur in the Fe_fresh-HCl powder after reaction.
Table 1 reports samples phase composition (wt.%), as obtained by Rietveld refinements: before reaction, -Fe 0 is the dominant phase, although it slightly decreases in the aged sample.Treatment with HCl brings about an apparent increase in the content of metallic iron, due to the dissolution of the basic oxide Fe 3 O 4(s) : With Fe_fresh-HCl after reaction, also peaks due to lepidocrocite (-FeO(OH)) are observed (Figure 1   of Fe 3 O 4 is still present after HCl treatment: the latter is likely a thin layer of magnetite that, acting as a semiconductor, could facilitate both charge transport within the layer and total oxidation of Fe 2+ to Fe 3+ species during reaction with AO7 and with dissolved oxygen, since the reaction mixtures were not deaerated [59,67].Morphological analysis of the samples (before reaction) was carried by means of both FE-SEM and TEM techniques.Figure 2(a) reports a FE-SEM micrograph of Fe_fresh where two families of particles were observed.Particles labelled as (i) are nearly spherical, with an average diameter of 110 nm and an iron and oxygen content of about 52 at.% and 38 at.%, respectively, as measured by EDX analysis, indicating that they are partially oxidized (vide infra).Particles (ii) are instead much larger and characterized by a platelet shape, with iron and oxygen contents of about 38 at.% and 52 at.%, respectively, that is, a chemical composition very close to that of magnetite (Fe 3 O 4 ).Fe_fresh, where (i) particles characterized by a metallic (darker) core surrounded by a brighter shell (arrow) can be appreciated.Morphological analysis in Figure 2 suggests that both larger magnetite particles and core-shell particles occur in the Fe_fresh sample, as previously observed with RNIP [15].Larger magnetite particles are likely present due to the kind of synthesis: the formation of fully oxidized particles for aging would be expected on the smaller particles, which undergo full oxidation first.
N 2 sorption isotherms at 77 K (not reported) were of type II for all the studied samples, as typical of nonporous materials.The corresponding BET specific surface area values, reported in Table 1, show that the surface area of Fe_fresh and Fe_aged samples does not differ much, indicating that it was almost unaffected by aging, whereas it markedly increased after treatment with HCl.This phenomenon was likely due to the formation of some porosity and/or surface roughness during reaction (1).
As a whole, aging has a limited effect on both surface area and phase composition, as determined by XRD: the aged sample (stored in the fridge, vide supra) shows a composition as similar as the as-received sample.HCl treatment, instead, markedly changes the surface properties of the studied materials, which contain both core-shell NPs and larger particles of magnetite.

Effect of pH on the Reactivity with AO7.
Due to the poor effect of aging on both surface area and composition (Table 1), sample Fe_aged was not considered for AO7 degradation experiments, and only the interaction between Fe_fresh and AO7 was studied.
The effect of pH on the process in water was considered by studying the phenomenon either at the natural pH of AO7 solution (=6.8) or after bringing pH to 2.0, 4.0, 8.0, and 10.0, the 484 nm band of the dye being used to measure the concentration of residual AO7 in the supernatant solution.
Figure 3(a) reports efficiency towards AO7 removal measured as the percentage of decrease of the 484 nm band at different pH values after 15 and 60 min of contact, pH values after 60 min being reported in parenthesis.It can be noticed that the effect of starting pH is substantial, as the AO7 removal efficiency ranges from ca. 20% (in basic conditions) to 100% (in acidic conditions).Nonetheless, in acidic conditions the supernatant solution pH increases with time, whereas it remains almost constant (or slightly decreases) in basic conditions, so indicating the occurrence of different processes.This may be appreciated in Figure 3(b) reporting the UV-Vis spectra of the supernatant solutions obtained at pH = 2.0 and pH = 8.0 after 15 and 60 min contact with Fe_fresh.In slightly basic conditions, the overall spectrum of AO7 decreases, without formation of any other band, indicating that mere adsorption takes place, and indeed UV-Vis spectra recorded after 15 and 60 min nearly coincide.The behaviour at pH = 10.0 is entirely similar (spectra not reported).Concerning AO7 adsorption, it must be considered that the point of zero charge (PZC) of magnetite is 6.8, so at higher pH, as in the present cases, the particles surface is negatively charged, and adsorption may only occur through some uncoordinated iron ions at the surface of Fe_fresh.Such ions may interact with the sulphonic group of the dye and/or coordinate N atoms of the azo bond, in agreement with the literature [68,69].Since AO7 molecules may not adsorb on fully oxidized (i.e., Fe 2 O 3 ) surfaces [30], interaction most probably involves Fe 2+ surface ions of the magnetite phase.
The experiment run at the initial pH of 2.0 has different features, in that a redox process is implied: the band at 484 nm disappears, while a band at ca. 249 nm, with a shoulder at 280 nm, firstly grows (after 15 min, black continuous line in Figure 3(b)) and then decreases somewhat (after 60 min, black dashed line in Figure 3(b)).As shown in Scheme 3, the expected products of the reaction between AO7 and metallic iron in acidic conditions are sulphanilic acid (SA) and 1A2N, arising from the reductive cleavage of the azo bond.Unfortunately, both substances (as well as many other organics) absorb around 250 nm.In detail, SA in solution shows both a band at ca. 245 nm and a shoulder at 285 nm.1A2N is reported to show a band at 228 nm [5], not observed in this case.Instead, in the separate experiment described in SI, the band for protonated 1A2N was measured at 245-250 nm, shifting to 257 nm in basic conditions.We are therefore inclined to ascribe the band at ca. 250 nm mainly to 1A2N.Being p a of SA equal to 3.23, the moiety in solution should be protonated at the beginning of the reaction and deprotonated at pH = 4.58, so that after 60 min reaction it should adsorb at the surface of Fe_fresh through the sulphonic group, from which the observed decrease in intensity of the band at ca. 249 nm.It is noteworthy that 1A2N is not adsorbed on Fe_fresh NPs in the present conditions.In the following, it will be shown that adsorption of 1A2N may be observed under different circumstances.
The reaction in acidic conditions was accompanied by a marked increase in pH, due to consumption of hydronium ions.Two mechanisms may be envisaged for this process.The reduction of AO7 is described for simplicity in Scheme 2 as implying H atoms. Indeed, as already suggested in the literature [46,50], hydronium ions are removed from solution, and electrons are provided by metallic iron, which can form in the first instance Fe 2+ species: Taking into account the fact that 4 hydronium ions are consumed per AO7 molecule, the change in pH expected on the basis of reaction (2) results to be much less than what observed.Another more extensive mechanism for the consumption of hydronium ions most probably consists in the attack of the (basic) solid Fe 3 O 4 by hydronium ions, the Figure 4: Section (a): UV-Vis spectra of 0.67 mM AO7 solution and of the supernatant solutions in contact with Fe_fresh at pH = 6.80 after 90 min (grey curve) and after 168 h (black curve).Section (b): trend in the intensity of the AO7 band at 484 nm as a function of time.
same reaction occurring during reactivation of aged samples by HCl solutions: yielding both ferrous and ferric ions in solution.
Finally, the experiments run at initial pH values of 4.0 and 6.8 yielded UV-Vis spectra (not reported) similar to those observed at basic pH values, so showing that the process is adsorption of AO7.The moderate increase in pH along the reaction, however, indicates that reaction (2) is also taking place (vide infra), providing fresh surfaces for adsorption, and thus yielding a removal efficiency above 20%.
In summary, as depicted in Scheme 2, with basic solutions simple adsorption takes place.With pH values close to neutrality, adsorption is also the predominant process, but acidic attack to the magnetite surface takes place, finally providing new anchoring points for AO7 molecules.With acidic solutions, the erosion of the oxide layer is extensive enough to activate the chemistry of metallic iron, so that the phenomenon switches from the mere adsorption of AO7 to cleavage of the azo bond, with probably firstly adsorption of SA and release of 1A2N into the solution.

Long-Term Behaviour of Fe_fresh.
At the natural pH of the solution only, further evolution of the system was observed at longer times with Fe_fresh: Figure 4(a) shows a minimum in the intensity of the 484 nm band after 90 minutes reaction, corresponding to a decrease around 24%.At pH = 6.80 adsorption is the main phenomenon taking place, although some reaction occurs, in agreement with both Figure 5: Section (a): UV-Vis spectra of 0.67 mM AO7 solution and of the supernatant solutions in contact with Fe_fresh-HCl sample at pH = 6.80 after 105 min (grey curve), 150 min (black curve), and 5 h (light grey curve).Section (b): selected UV-Vis spectra of the supernatant solutions after 5 h (light grey curve), 72 h (grey curve), and 216 h (black curve).Section (c): trend in the intensity of the 1A2N band at 249 nm (black symbols) and of pH (white symbols) as a function of time.
the observed increase in pH and the increased absorption at 249 nm (asterisk) due to the azo bond cleavage described above.Figure 4(b) shows that, at longer contact times, the intensity of the 484 nm band increases again and after 168 h the overall decrease is ca.15%.Such an apparent "reversible phenomenon" may be ascribed to further oxidation of the surface from magnetite to Fe 3+ oxo-hydroxides, causing desorption of AO7, which does not adsorb on fully oxidized surfaces [30].

Behaviour of Reactivated Samples.
With samples reactivated by HCl, two types of experiments were run, one at nearly neutral pH and the other at pH = 10.
Figure 5 describes the experiment at nearly neutral pH: spectra recorded within the first 105 minutes show a plain decrease in intensity (with an overall decrease around 80%, section (a)).No new bands were observed, so suggesting again the occurrence of a mere adsorption process.
After 150 minutes, a band at ca. 249 nm was observed to grow as in Figure 3(b) (black curve in Figure 5(a)), reaching its maximum after 72 hours, then decreasing in intensity, and finally reaching a plateau (Figure 5(b)).Interestingly, the solution pH followed the same trend, reaching after 72 hours the highest value at 6.8 and then decreasing to 4.8 within 216 h (Figure 5(c)).
The second stage of the process seems to correspond to the cleavage of the azo bond, and indeed the pH increases because protons are consumed.Simultaneously, however, oxidation of the surface takes place, beyond the formation of magnetite, likely yielding lepidocrocite [5], probably through the agency of some dissolved oxygen: The surface so becomes unable to host SA, which is adsorbed on magnetite, but not on more oxidized surfaces (in the present case lepidocrocite).
In contrast, 1A2N appears to have affinity for the fully oxidized surface and to adsorb by releasing the phenolic proton.This accounts for the decrease in pH in the latter stages of reaction.
Two facts are worth of note.On the one hand, the acidity of phenolic groups may markedly increase by effect of substituents.On the other hand, the 1A2N anion given rise may behave as a bidentate ligand because of the simultaneous presence of -NH 2 and the phenolic O − species, whereas SA only features the sulphonic group as a docking moiety, apparently able to interact with Fe 2+ ions, but not with Fe 3+ ones.The presence in the liquid phase of SA accounts for the presence of a plateau in Figure 5(b).Experiments run with Fe_fresh-HCl at pH = 10 showed similar features (Figures not reported).
It is now appropriate to make a comparison with the reactivity in slightly basic solutions, that is, with the RNIP slurries studied in a previous paper [15], the overall chemistry observed being reported in Scheme 3. RNIP promotes the cleavage of azo group in basic conditions [15]: with particularly active (polymer protected) RNIP samples, the 1A2N given rise partially further oxidizes to -naphthoquinone, which condenses back with SA to yield a molecule close in structure to AO7, so mimicking a partial reversal of the degradation reaction.
It is worth noting that such reaction is not observed with the samples studied in this work.The question arises whether a basic nature of the solution is required.Separate experiments, reported in Figure SI2, showed that the 1A2N molecule, when allowed to stand in a basic solution, does not show reactions as in Scheme 3, that is, the substitution of the amino group with a hydroxyl species.Much more vital seems to be the nature of the surface, which provides sites for the ligand displacement: it is not straightforward, however, to propose a more detailed explanation.
IR spectra of powders outgassed at room temperature were recorded before and after interaction with AO7 solution in order to further investigate the presence of adsorbed species: Figure 6 reports difference spectra of both Fe_fresh and Fe_fresh-HCl samples after reaction obtained by subtraction of the corresponding spectra before interaction with AO7.It should be noticed that the slope of the curves is simply due to some scattering phenomena and that, being difference spectra reported, positive bands are due to species formed after interaction with AO7, whereas negative bands are due to bands disappeared after such interaction.
In the lower wavenumbers region, bands between 630 and 530 cm −1 are assigned to Fe-O symmetric stretching modes [70,71].The characteristic bands of lepidocrocite (-FeO(OH)) [5] at 1157, 1020, and 749 cm −1 are observed in the IR spectrum of Fe_fresh-HCl (black curve), in fair agreement with the previously discussed results; both samples exhibited signals at ca. 1090, 900, and 750 cm −1 , likely due to presence of other iron oxy-hydroxide surface species [5].
No bands ascribable to adsorbed AO7 molecules are seen with Fe_fresh (grey curve), in agreement with the observed desorption of AO7 from oxidized surfaces.The IR spectrum concerning Fe_fresh-HCl (black curve) shows instead bands in the 3600-2800 cm −1 and in the 1640-1540 cm −1 range: the former are assigned to NH-stretching and bending vibration of adsorbed species [72]; no bands ascribable to sulphonic groups modes, usually absorbing at 1250-1140 cm −1 (asymmetric stretching vibration) and 1070-1030 cm −1 (symmetric stretching vibration), are observed [8], in agreement with the previous hypothesis of desorption of SA and adsorption of 1A2N on more oxidized surfaces, as suggested in Scheme 2(b).

Conclusions
The type of interaction between Acid Orange 7 and Nanoscale Zerovalent Iron mostly depends on surface composition of the solid nanoparticles and pH of the aqueous phase.
Concerning surface oxidation of iron nanoparticles, two main results are obtained: (i) exposure to air does not lead to remarkable iron oxidation, likely because the slurry was kept in a fridge, the low temperature definitely preventing iron oxidation, and (ii) treatment with HCl satisfactorily restores nanoparticles reactivity towards Acid Orange 7, by the likely dissolution of the outer magnetite layer.Surface oxidation is an important issue, in that when only Fe 3 O 4 is available adsorption of the dye is the dominant type of interaction in this system.The same result was obtained by some of us in a previous work where another Nanoscale Zerovalent Iron commercialized at basic pH was studied (Reactive Nanoscale Iron Particles).
In the presence of (reactive) core-shell nanoparticles, cleavage of N=N bond occurs: such phenomenon leads to different reaction products depending on pH of the aqueous phase.
In the present work, where a slurry commercialized at nearly neutral pH was employed, lowering the pH led to a remarkable enhancement of dye removal.The reaction basically consists in the hydrogenation cleavage of the N=N bond, but in acidic condition, the source of hydrogen atoms seems to be the reaction H + + Fe 0 → H + Fe + and indeed the pH is observed to increase during the process.The resulting products (sulphanilic acid and 1-amino-2-naphthol) are rather stable and suffer a different fate, as sulphanilic acid is firstly adsorbed, and 1-amino-2-naphthol is in solution and only adsorbs when the surface is thoroughly oxidized.
With Reactive Nanoscale Iron Particles, instead, the cleavage of azo group was possible also in strongly basic conditions (Scheme 3).Moreover, with the polymer protected Reactive Nanoscale Iron Particles, the produced 1-amino-2naphthol partially further oxidizes to -naphthoquinone and condenses back with sulphanilic acid to yield a molecule close in structure to Acid Orange 7, then mimicking a partial reversal of the degradation reaction.Such reaction seems to be typical of the polymer protected Reactive Nanoscale Iron Particles, as it is not observed with Nanoscale Zerovalent Iron particles studied in the present paper.

Figure 2 (
b) reports a FE-SEM micrograph of Fe_fresh-HCl sample: particles (i) have undergone slight morphology changes, appearing after HCl treatment with a more irregular shape; particles (ii) look almost unaffected by the acidic treatment.Figure 2(c) reports a TEM magnification of sample

Figure 3 :
Figure 3: Section (a): removal efficiency of AO7 (%) obtained with sample Fe_fresh at different initial pH values after 15 min (grey columns) and 60 min (light grey columns); numbers in parenthesis refer to pH values measured after 60 min.Section (b): UV-Vis spectra of 0.67 mM AO7 solution and of the supernatant solutions after 15 min and 60 min at both pH = 8.00 (grey curves) and pH = 2.00 (black curves).

Scheme 3 :
Scheme 3: Mechanism of AO7 degradation by NZVI particles in aqueous solution.Dotted framed reaction takes place only in presence of polymer modified NZVI particles described in[15].

Figure 6 :
Figure 6: IR spectra of Fe_fresh (grey curve) and Fe_fresh-HCl after reaction, obtained by subtraction of the IR spectra of the corresponding samples before reaction.

Table 1 :
BET specific surface area (SSA) values as obtained by N 2 sorption isotherms at −196 ∘ C and calculated phase composition (% by weight) as obtained by Rietveld refinement.