Plaster of Paris containing zero valent iron particles Designing a permeable reactive barrier, used for remediation of 4-nitroaniline pollut

2.1. Chemicals

Plaster, in chemical form of CaSO4.0.5H2O, was a commercial powder (plaster of Paris quality) from Parexlanko, France; iron powder (ZVI, 325 mesh, 99%), H3PO4 (85%), and CaCO3 (99%) were from Acros (France), 4-nitroaniline (p-nitroaniline, PNA, 98%), p-phenylendiamine (PDA, 98%), o-phenanthroline (98%) and KSCN (97%) from Aldrich (France).

2.2. Plaster Samples Preparation

Plaster slabs were made by adding plaster powder (50 g) to appropriate amounts of distilled water. After 5 minutes of mixing, preparations were poured into PET (polyethylene terephtalate) or PE (polyethylene) cylindrical molds to make samples for flow experiments (in 2.8-cm- diameter, 2-cm-high) and for mechanical tests and density measurements (in 2.6-cm-diameter, 5-cm-heigh). All solid samples were dried at 60°C for 24 hours before their use. All samples were weighted and “apparent” density calculated from geometrical parameters.

2.3. Material Characterization

2.3.1. XRD and SEM Analysis Solid samples were characterized by powder X-ray dif-

fraction (XRD) with a Bruker D2 X’PertPRO diffractometer using Cu Kα radiation (40 kV and 40 mA). Crystallographic identification of CaSO4.2H2O was accomplished by comparing the experimental XRD pattern to COD 2300259 of gypsum standard (point group 2/m). Plaster samples were observed by scanning electron microscopy (SEM) Zeiss Supra 55VP. 2.3.2. Mechanical Tests Indirect tensile stress diametral compression tests were carried out with a Zwick Roll Z020 testing machine (Zwick Roell, France) at a cross-head speed of 10 mm/min, until failure. Using initial height (L [m]) and diameter (D [m]) of each sample, and measuring the reciprocated force (F [kg m/s2]) from compression, indirect (or Brazilian) tensile strength, σT ([kg/m s2]), was calculated from its definition in equation 1: σT=2F/(πLD)(1) 2.3.3. Porosity Determination Porosity was determined from the ratio of experimental “apparent” density of each sample and gypsum density, CaSO4.2H2O, 2.308 g/cm3 [16] with equation 2: porosity (%)=(densityexp/densitygypsum)*100(2) 2.3.4. Permeability Coefficient Measurements Permeability coefficient measurements were done on plaster samples (2.8-cm diameter, 2.0-cm heigth) with a water pressure column of 5 cm. Discharge of water (Q [m3/s]), percolating under a constant head difference (H [m]) through a sample of porous material of cross area S ([m2]) and length L ([m]), was determined for several samples. Permeability coefficients K (hydraulic conductivity, [m/s]) were calculated from Darcy’s law, following equation 3: K = (QL)/(SH)(3)

2.4. Bechamp Reduction

2.4.1. Spectroscopic Analysis Vis-UV spectra were used to quantify 4-nitroaniline (PNA) and p-phenylendiamine (PDA) concentrations. They were recorded between 200 and 700 nm in aqueous solution on a HP 8453 spectrophotometer with quartz cells of 0.2 and 1 cm of path-length. Spectra of pure 4-nitroaniline (PNA) and p-phenylendiamine (PDA) were recorded in aqueous solution giving the following spectroscopic features: PNA: 380 (ε=73.3 l mol-1cm-1), 227 (35.5), 204 (63.0) nm. PDA: 305 (ε=13.7 l mol-1cm-1), 240 (49.3), 210 (58.7) nm. PNA concentrations in solution were determined by measuring absorbance at 380 nm. PDA formation (in Bechamp reduction) was checked from band at 240 nm (PDA maximum absorption).

J Sustain Const Mater Technol, Vol. 6, Issue. 4, pp. 124–134, December 2021

2.4.2. Determination of PNA Degradation Degradation of PNA, meaning the disappearing of its characteristic absorption band at 380 nm, was calculated from the UV-Vis spectra (Fig. S1) with equation 4: degradation (%)=(A380-A600)/(A°380-A°600)*100(4) where A380 and A600 are the absorption values of each sample at the noted wavelengths, and A°380 and A°700 the absorption values of starting solutions. Absorption values were corrected at 600 nm in order to avoid diffusion effects from cloudy solutions. At 380 nm the absorption contribution for presence of PDA in aqueous solutions was insignificant. 2.4.3. ZVI in Batch Batch experiments were conducted inside 250 ml Erlemeyer flasks under continuous magnetic stirring (500 rpm) and the overall time of experiments was 7 hours. Batch experimental initial conditions were: temperature; 25°C, PNA concentration; 30 mg/l, water volume; 100 ml. Specific variables in batch experiments included pH (initial pH acidic or neutral) and the amounts of ZVI (1, 2 and 3 g). Sampling was conducted within accumulative time of 15 minutes during first three hours, then of 1 hour for last four in batch experiments. For analysis, 2 ml samples were withdrawn by a 5 ml syringe, their Vis-UV spectra registered using a 1 cm quartz cell, then returned to batch. Initial pH was adjusted, or not, using 95% phosphoric acid. Degradation percentage was calculated from absorbance values at 380 nm. 2.4.4. Column Preparation 100 g of plaster, mixed with appropriate amounts of ZVI and 0.2% CaCO3, were added to 70 ml of distilled water, containing 0.4 g of phosphoric acid, then all reagents were mixed together and the viscous mixture poured into PET (polyethylene terephtalate) cylindrical molds (in 5.0-cm-diameter, 15-cm-heigh) used as percolator system. PET shells were thermally fixed to plaster columns in order to avoid leaking from their extremities. 2.4.5. ZVI in Column PNA-containing aqueous solutions (100 ml aliquots) were passed by percolation through columns. Fractions of 100 ml were recovered post-column: 2 ml of each fraction were checked by UV-Vis spectrum between 200 and 700 nm and absorbance values at 380 nm for used to calculate PNA concentration in each fraction. 2.4.6. Identification of Ionic Species Samples of plaster-based composites (2.8-cm diameter, 2.0-cm heigth) were soaked in distilled water (25 ml) and pH was measured as a function of time. KSCN (1%) and o-phenathroline (0.5%) in aqueous solution were used to check by colorimetry presence of Fe(II) or Fe(III) species, respectively.

3.1. Properties of Plaster-Based Composites

3.1.1. Effect of Water Water/plaster ratio was modified, and physical and mechanical properties were measured for each sample (Fig. 1a, Table A.1). It was reported that water amount determined porosity in plasters blocks [17, 18]; as reported in Figure 1a, indirect tensile strength of plaster was reduced as a function of water content whereas porosity increased. During reaction between plaster (CaSO4.0.5H2O) and water to form gypsum (CaSO4.2H2O), part of water was consumed by hydration reaction during material setting and remainder slowly evaporated leading to porosity [19]; hardening of plaster paste lead to porous structure, made of entangled needle-shape gypsum crystals [20]. 3.1.2. Effect of Porogen (CaCO3+H3PO4) Porous samples were prepared by adding fine calcium carbonate with plaster powder, then pouring the powder in phosphoric acid–containing aqueous solution. Calcium carbonate was used as porogen in presence of acid in order to generate open porosity and increase permeability through the column (Fig. 1b); phosphoric acid was chosen because formation of calcium phosphate as secondary product gave composites with enhanced mechanical properties and lower dissolution rates [21]. In presence of increasing amount of H3PO4, porosity increased while density and mechanical properties decreased. These results are in agreement with cavity formation following CO2 evolution during material setting (Table A.2). Starting from 0.5% acid, porosity was not significantly modified, as well as mechanical properties. However, permeability increased three times in magnitude compared to “compact” plaster. Calcium phosphate, formed by reaction of phosphoric acid with calcium carbonate [22], was dispersed in gypsum matrix with no effects on physical and mechanical parameters. It was reported that weak acids (such as citric acid) play a role as plastifier, retarding setting [23]. In fact, citric acid partially inhibits binder hydration from hemihydrate to dihydrate. Total porosity seemed little affected by increasing the amount of CaCO3, pore dimensions were bigger (Fig. 2) and SEM images of porogen-containing samples showed 1 mm pores.

J Sustain Const Mater Technol, Vol. 6, Issue. 4, pp. 124–134, December 2021

Figure 1. (a) Indirect tensile strength (closed signet) and porosity (empty signet) as a function of plaster composition as a function of: (b) water, (b) H3PO4, (c) ZVI amounts; experimental conditions: plaster (50.0 g). 3.1.3. Effect of ZVI Different amounts of ZVI were dispersed in solid components (plaster and calcium carbonate) before adding water and phosphoric acid, in order to uniformly scatter the heavier iron particles in the slab. Physical and mechanical properties of different samples obtained were reported in Fig. 1c. Depending on ZVI amounts (Table A.3), density and mechanical properties were increased, and porosity reduced. 3.1.4. Effect of Acid In classical Bechamp reduction hydrochloric acid is used; for this reason, effect of mineral acid on plaster setting was tested comparing H3PO4 and HCl. Acidic solutions increased porosity compared to distilled water, but samples with hydrochloric acid were more dense and compact than those made with phosphoric acid (Table A.4). Presence of mineral acids influenced the gypsum crystallization step, promoting formation of large bulky crystals [24]. Particle size was due to increased growth rate [25]. For hydrochloric acid-containing sample the mechanical strength was increased whereas porosity was similar to samples in absence of acid (Table A.3), However, the goal of our research was to increase plaster permeability and not composite mechanical properties. For

these reasons hydrochloric acid was not used in further experiments. In presence of H3PO4 and in absence of carbonate, porosity was higher with respect to samples in Table A.2, indicating that CaCO3 could also act as buffer stopping acid reactivity and controlling porosity. 3.1.5. Permeability Coefficient Measurements Permeability coefficients were measured for all sample and results reported in Table 1. In order to allow percolation rapidly, porosity obtained only by water evaporation was insufficient to have an acceptable permeability. For samples containing “compact” plaster (series 1 and 3), discharge was very slow and permeability coefficients were estimated from water amount recovered after 24 hours. Using CaCO3 as porogen, plaster permeability increased by three orders of magnitude allowing to run experiments in short times. In a 1H NMR relaxation study of water, confined in porous medium of hardening gypsum (without porogen), coexistence of two water populations were described in permeable and disordered porous plaster structure. The first population (P1) extended uniformly in space while the second one (P2) was more confined and isolated in some clusters of gypsum needles [20]. Distribution of the two populations was related to water-to-plaster ratio (w/p)

J Sustain Const Mater Technol, Vol. 6, Issue. 4, pp. 124–134, December 2021

Figure 3. Porosity optimization as a function of different components of BRP (water, phosphoric acid and ZVI) plaster (50.0 g, 40.0%); water (56.0%); CaCO3 (0.2%); H3PO4 (0.2%).

Figure 2. SEM photographs showing a view of porogen-containing hardened gypsum samples, prepared with a water-to-plaster ratio of 0.7, as a function of carbonate amounts (with 1.1 % phosphoric acid, w/w). CaCO3: A=0.2%, B=0.5g, C=0.7g, D= 0.9% (w/w); length scale bar is 1 mm and magnification x25). values: at low w/p value, P2 increased and P1 decreased, in agreement with total porosity observed as a function of w/p. Moreover, from a micro-structural point of view, an intricate percolation networks of needles with different packing density exists in hardened plaster structure where percolation threshold allowing water exchange was about w/p=0.6. Below w/p=0.6, there was an extremely slow exchange rate (almost zero) and the two water populations were almost independent. Above w/p=0.7, there was a small but finite exchange rate between the two water populations, but composite permeability was not sufficient to run PRB kinetic studies in short times. Porosity of different plaster slabs, obtained from sample densities, with and without porogen, was not so different (62% vs 55%, Table S2), however permeability was extremely dependent on porogen presence, differing by about three orders of magnitude (Table 1).

3.2. Degradation of 4-Nitroaniline

3.2.1. Column Formulation Only composites with a permeability allowing the study of PRB chemical reactivity in a one-day period, were used in pollution removal tests. Figure 3 reported optimization of all composite components (water, calcium carbonate, phosphoric acid, ZVI) for a fixed amount of plaster to improve different physical properties. Composite formulation, used to mold columns, contained plaster and water with 0.7 ratio, 0.2% CaCO3 and 0.4% H3PO4 (per 50 g of plaster). 3.2.2. ZVI Reactivity in Batch Bechamp reduction was checked batchwise. Experiments were run as a function of ZVI amount, in absence and in presence of phosphoric acid to activate reduction. “Classical” reduction [6] was run with acetic acid (or hydrochloric acid), which does not form insoluble species after oxidation of ZVI to Fe(II) or Fe(III). Yellow PNA aqueous solutions were expected to fade as a function of time, following reduction to PNA (Fig. A1). As a function of time, yellow solutions turned pale and were cloudy: also in absence of H3PO4, reduction occurred and after 7 hours almost no PNA remained in solution (Fig. 4). As a function of time, protons were consumed in reduction reaction and solution pH increased (Fig. 5).

Table 1. Permeability coefficients of plaster samples as a function of composition. Experimental conditions: plaster (50.0 g, 41.0%); experiments were an average of five samples

J Sustain Const Mater Technol, Vol. 6, Issue. 4, pp. 124–134, December 2021

Figure 4. Degradation of PNA in Bechamp reduction as a function of time and of ZVI percentage: 1.2 (О); 2.3 (☐); 3.4 (∆); 1.0% + 1.0% H3PO4 (●); [PNA]=30 mg/l.

Figure 6. Plaster column composition: plaster (50.0 g, 41.0%); water (57.4%); CaCO3 (0.2%); H3PO4 (1.4%); ZVI (2.3%). Figure 5. Final pH values of solutions in batch Bechamp reduction after 7 hours, as a function of ZVI: without acid (dark grey); with 1 % H3PO4 (light grey). In presence of H3PO4 (0.4%) or higher amount of ZVI (>3%), reaction was faster and total reduction occurred in 1 hour. Solutions were cloudy because of formation of insoluble iron hydroxydes or phosphates. Samples recovered after reduction, with or without acid, became violet by standing: redox reaction was not halted to PDA and polymerization of aromatic molecules also occurred, as usual in presence of metal ions [26, 27]. 3.2.3. ZVI Reactivity in Column PNA solutions were added to columns (Fig. 6) and, after percolation, fractions (100 ml) were recovered and analyzed by Vis-UV spectroscopy. In all experiments, yellow color of PNA solutions disappeared indicating that Bechamp reduction occurred in plaster columns. In order to verify chemical effects in colour fading, UV-Vis spectra were registered between 200 and 700 nm: PNA was reduced to PDA indicating that Bechamp reaction occurred in plaster column. As observed in batch experiments, PDA aqueous solutions were not stable as a function of time: pale yellow solutions

became hazy and violet colored, indicating polymerization of aromatic amines. Degradation profiles, as a function of time at different concentrations of PNA and ZVI, were reported in Figure

1.6. L of aqueous PNA solution at 100 mg/L were treated by

3 g of ZVI, meaning a complete inactivation at around 55 mg of PNA per gram of ZVI, or about 2.2x10-2 mol PNA per 1 mol Fe(0).

3.3. Analysis After Remediation

Plaster columns and solutions recovered were analyzed at the end of remediation process in order to elucidate the role and reactivity of ZVI trapped in solid. 3.3.1. Solution Analysis Because of the presence of colored species (PNA and

J Sustain Const Mater Technol, Vol. 6, Issue. 4, pp. 124–134, December 2021

Figure 7. Degradation profiles of p-nitroaniline as percentage (PNA=30 mg/L (a), 100 mg/L (b)) as a function of elution volume and ZVI percentage. ZVI: 1.2 (☐), 2.3 (●), 3.4 % (X); column composition (without ZVI): plaster (200.0 g, 41.0%); water (57.4%); CaCO3 (0.2%); H3PO4 (1.4%). PDA), a colorimetric analysis of the solutions in order to follow the fate of ZVI, was not possible. However, this analysis was possible in ZVI composite before any reaction. Samples were soaked in distilled water and pH was measured as a function of time. KSCN and o-phenathroline in aqueous solution were used to check presence of Fe(II) or Fe(III) species, respectively. Because of the low amount of iron ionic species in solution, reagents were used in strong excess for ion identification. Presence of Fe(II) was detected in plaster columns (Table A.5) following the reaction between ZVI and phosphoric acid, whereas Fe(III) was not found by colorimetric method. In ZVI-containing column, during addition of PNA solutions, the first volume eluted was slightly acid (pH=5.22), probably because of traces of phosphoric acid, whereas, later on, pH was determined by the major organic compound present in solution (PDA, pH=8.07; PNA, pH=6.10 for 30 mg/l aqueous solutions). 3.3.2. Solid Analysis At the end of reduction process, reddish colour of column indicated presence of Fe3O4 species, formed because of instability of Fe(II) species in aqueous solution in presence of dioxygen,

XRD diffractograms for plaster samples (Fig. 8) were compatible with those of gypsum (COD 2300259). Typical diffraction angle at 44.8° (2θ) for Fe(0) [28] was not detected by XRD, because in these experimental conditions, iron metal, dispersed in a gypsum matrix, could be observed only at high percentage (>50%). However, in plaster composites containing Fe(0), a diffraction angle at 39.5° was present but disappeared after PNA reduction (Fig. 8). 3.3.3. Iron Reactive Species Bechamp reduction is a stoichiometric reaction, and not a catalytic one, resulting in a series of parallel reactions, with generation of one or more intermediates and byproducts [29]. ZVI is “consumed” in redox system and passivated during flow of aqueous solutions. Noubactep reported to use the lowest possible Fe(0) loading (e.g. <5 g/l) and work at low temperature (e.g. 15°C) [30]. In an environmental application, it would be possible to alternate in PRB porous plaster blocks (in order to absorb more polluted solution) and ZVI-containing compact plaster blocks (in order to act as reducing agent) to increase the residence time. However, Bechamp reaction in PRB seems to be more complex than in solution. In fact, reduction by Fe(0) occured and contributed significantly to contaminant removal, but the

J Sustain Const Mater Technol, Vol. 6, Issue. 4, pp. 124–134, December 2021

phosphoric acid acted also on ZVI surface, partially generating Fe(II) species. It means that Fe(II) species could be involved in order to accelerate reduction rate, but remains the question of Fe(II) species stability in presence of dioxygen. In ZVI reactions, concomitant precipitation of iron hydroxydes absorbing organic compounds has been also suggested [32]. From our experience in pollutant co-precipitation methods [33], we cannot totally reject this mechanism [34], but recovering PDA in solution strongly suggested that redox mechanism was predominant. Finally, we observed that, as a function of time, downflow through the column increased by 25% indicating that plaster columns were embrittled: columns were sensitive to percolation because of calcium sulfate solubility in water and because iron corrosion products formed in situ were partially disrupting their structure.

Figure 8. DRX of plaster composites, plaster, plaster + Fe(0) before reduction, and plaster + Fe(0) after reduction. Composites were formulated as in Tables S1 and S4. mechanism remained unclear. It is matter of debate in ZVI depolluting systems, if, between the Fe(II)/Fe(0) or Fe(III)/ Fe(II) couples, which one is involved in the redox reactions [31]. In batch experiments, PNA reduction occurred also without acid suggesting that Fe(0) would be directly involved in redox system; however kinetics, in batch and in column, were not the same (in column the reaction was instantaneous) suggesting that preliminary treatment with

3.4. Environmental Impact and Cost

An ideal PRB should present the following advantages [35]: • stability during PRB lifetime in order to minimize lost of efficiency: Plaster of Paris containing zero valent iron particles: is a stable composite • chemical reactivity, which is related to reactant contact time, depending on flow rate and amount of reactant: ZVI is still actif when included in plaster. • hydraulic and mechanical performances: PRB permeability has to be higher than aquifer flow in order to not disturb underground flow and not induce polluted flow out of barrier. The supporting material should also be shaped in blocks in order to easily build PRB. Permeability, mechanical properties and shapes of plaster-based PRB can be modulated in formulation step. • reduced environmental impact in order to avoid pollution effects. PBR has to be mainly constituted by inert and non-toxic materials, producing non-toxic and harmless by-products: if the main problem in using calcium sulfate-based PRB for geological applications is their relative solubility in water, however they will break up with time discharging in environment no toxic species. • cost of PRB building (in terms of materials and construction process) as low as possible in order to be economically feasible: the cost of plaster-based PRB is estimated to be very low and convenient. Finally, efficiency of ZVI in plaster based-columns suggests use of these composite as precast plaster wall blocks in environmental applications.