Personal Use Only Not for Distribution

0 downloads 0 Views 7MB Size Report
... these species re- act much more quickly with green rust and do not transform into ammonium. ... *Address correspondence to this author at the Laboratoire de Chimie Physique et ... moving organic matter from water but the rates of denitrifica- tion and .... centrifugation to eliminate residual chloride or nitrate ions from the ...

Send Orders for Reprints to [email protected] Current Inorganic Chemistry, 2016, 6, 100-118 ISSN: 1877-9441 eISSN: 1877-945X

RESEARCH ARTICLE

Use of Ferrihydrite-Coated Pozzolana and Biogenic Green Rust to Purify Waste Water Containing Phosphate and Nitrate Christian Ruby1,*, Sébastien Naille1, Georges Ona-Nguema2, Guillaume Morin2, Martine Mallet1, Delphine Guerbois2, Kévin Barthélémy1,3, Marjorie Etique1, Asfaw Zegeye1, Yuhai Zhang2; Hella Boumaïza1, Muayad Al-Jaberi1, Aurélien Renard1, Vincent Noël2, Paul Binda1, Khalil Hanna6, Christelle Despas1, Mustapha Abdelmoula1, Ravi Kukkadapu7, Joseph Sarrias3, Magali Albignac3, Pascal Rocklin5, Fabrice Nauleau4, Nathalie Hyvrard4 and Jean-Marie Génin8 1

U

io n

se

O

nl y

Laboratoire de Chimie Physique et Microbiologie pour l’Environnement (LCPME), UMR 7564 CNRS – Université de Lorraine, 405 rue de Vandœuvre, 54600 Villers-lès-Nancy, France; 2Institut de Minéralogie, de Physique des Matériaux et de Cosmochimie (IMPMC), Sorbonne Universités – UPMC Univ. Paris 06, UMR 7590 CNRS, Muséum National d’Histoire Naturelle, IRD UMR 206, 4 place Jussieu, F-75005 Paris, France; 3Marion Technologies, Parc Technologique Delta Sud, 09340 Verniolle, France; 4SAUR, 1 rue Antoine Lavoisier, 78064 Guyancourt, France; 5Sous-Direction de la Valorisation de la Recherche de l’Université de Lorraine, 91 avenue de la Libération, 54001 Nancy Cedex, France; 6Ecole Nationale Supérieure de Chimie de Rennes (ENSCR), UMR 6226 CNRS – Ecole Nationale Supérieure de Chimie de Rennes), 11 allée de Beaulieu, 35708 Rennes cedex 7, France; 7Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, Richland, WA 99354 – USA and 8Ecole Supérieure des Sciences et Technologies de l’Ingénieur de Nancy (ESSTIN), 2 Rue Jean Lamour, 54519 Vandœuvre-lès-Nancy Cedex

DOI: 10.2174/18779441069991606031254 59

ut

rib

is t

rD

Received: June, 18, 2015 Revised: April 13, 2016 Accepted: April 29, 2016

Methods: The goal of this work is to identify the most suitable iron based materials for such treatments and to determine their optimal use conditions, in particular in hydrodynamic mode. The reactivity of the iron based minerals was measured either by using free particles in suspension or by depositing these particles on a solid substrate.

fo

ARTICLE HISTORY

rs on

al

Abstract: Background: The activated sludge treatments combined to the addition of ferric chloride is commonly used to eliminate nitrate and phosphate from waste water in urban area. These processes that need costly infrastructures are not suitable for rural areas and passive treatments (lagoons, reed bed filters…) are more frequently performed. Reed bed filters are efficient for removing organic matter but are not suitable for treating phosphate and nitrate as well. Passive water treatments using various materials (hydroxyapatite, slag…) were already performed, but those allowing the elimination of both nitrate and phosphate are not actually available.

Pe

Results: Pozzolana, a volcanic rock, that is characterized by a porous sponge-like structure suits for settling a high amount of iron oxides. In addition, the experimental conditions enabling to avoid any ammonium formation when green rust encounters nitrate were also determined within the framework of a full factorial design. The dephosphatation and denitrification process is divided into two steps that will be performed inside two separated reactors. Indeed, the presence of phosphate inhibits the reduction of nitrate by green rust and the dephosphatation process must precede the denitrification process.

N ot

Current Inorganic Chemistry

100

Conclusion: In order to remove phosphate, ferrihydrite coated pozzolana is the best materials. The kinetics of reaction of green rust with nitrate is relatively slow and often leads to the formation of ammonium. In this process, it is interesting to favor the accumulation of nitrite in a first step, these species react much more quickly with green rust and do not transform into ammonium.

Keywords: Water treatment, reed beds; iron oxide, ammonium, calcium. 1. INTRODUCTION Eutrophication is a leading cause of impairment of many fresh water and coastal marine ecosystems in the world with *Address correspondence to this author at the Laboratoire de Chimie Physique et Microbiologie pour l’Environnement (LCPME), UMR 7564 CNRS – Université de Lorraine, 405 rue de Vandœuvre, 54600 Villers-lès-Nancy, France; Tel: 00(0) 333 83685220; Fax: 00(0) 333 83275444; E-mail: [email protected] 5X

dramatic consequences for drinking water sources, fisheries, and recreational water bodies [1]. This phenomenon is most often a direct result of seepage of phosphate- and nitrate-rich groundwater into the aquifers. Common sources of such pollutants in these waters are due to over fertilization of agricultural soils and release of inadequately treated waste water into the aquifers. In urban and rural areas, treatments of waste waters containing nitrate and phosphate are processed by optimized ‹%HQWKDP6FLHQFH3XEOLVKHUV

Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Use of Ferrihydrite-Coated Pozzolana and Biogenic Green Rust

An average x value could lead to formula, FeII6(1-x) FeIII6x O12 H2(7-3x) CO3 • 3H2O; however, it is not a solid solution but rather a mixture of topotaxical domains in layers with different FeIII cation ordering [18,19]. Since dephosphatation and denitrification proceed by different pathways and the reactivity of a given material towards phosphate and nitrate varies, a sequential removal of the pollutants by using distinct filters in series may be a most attractive method. A detailed study of the reactivity of ferrihydrite and GR-related minerals is reported here. Understanding the adsorption efficiency of pure ferrihydrite and mössbauerite, GR*, towards phosphate around circumneutral pH under batch and flowthrough columns experiments in the presence of coexisting anions, is essential to build robust filtration plants. GR-related compounds, with various x values and those prepared abiotically and biotically, with varying amounts of phosphate in the media are studied in detail under batch conditions to optimize their reactivity towards nitrate. The selectivity of the reaction for transforming nitrate into ammonium is also systematically studied since NH4+ is an undesirable product [20] whereas the formation of nitrogen gas is harmless. Moreover, previous works demonstrated that a GRrelated compound can be synthesized by bioreduction of ferric oxyhydroxides such as lepidocrocite [21] or from mössbauerite [22]. The reactivity of this biologically generated green rust (GRbio) towards nitrate and nitrite species was studied very recently [23, 24]. Thus, if GRbio is envisaged to be used for a tertiary denitrification process, the following questions needs to be addressed: (i) how to decrease or avoid ammonium formation when GRbio reacts with nitrate?, (ii) does reactivity of GRbio towards nitrate and nitrite vary? (iii) what kind of role does nature of GR-related compounds (abiotic and biotic) play? In this paper an attempt was made to address these questions so that an optimum design for treatment of these pollutants will be finally proposed.

io n

ut

is t

rD

fo

N ot

Pe

rs on

al

Green rusts (GR) belong to the layered double hydroxide (HDL) family and are the special case where the divalent and trivalent cations are from the same element, iron. Therefore, they are mixed Fe(II)-Fe(III) hydroxysalts with general chemical formula [FeII(1-x)FeIIIx(OH)12]x+ [(x/n) An-, m H2O]xwhere x is the ferric molar fraction, {[FeIII] / [Fetotal]}, situated between 0.25 and 0.33 whereas An- are intercalated anions between the brucite-like layers made of six OH- ions at the apices of octahedral sites occupied by Fe cations. These compounds were found easy reductants of nitrate, NO3-, into ammonium, NH4+, and their reactivity towards nitrate species was dictated by both x and the nature of the intercalated anions. For example, green rust with low x values containing monovalent anions such as Cl- or F- displayed higher reactivity than GR containing divalent anions, e.g. SO42- or CO32[15-17]. In this paper, the carbonated green rust, GR(CO32-), with x = 1/3 and formula FeII4 FeIII2 (OH)12 CO3 • 3H2O, is only considered for two main reasons: (i) it is the most stable one, (ii) closely related compounds are found as minerals in natural environments explaining the bluish color of gleys of hydromorphic soils and are obtained in the laboratory by bacterial reduction of ferric species in anoxic conditions. It was previously demonstrated that GR(CO32-) can get oxidized in situ by deprotonation of OH- ions; this leads to closely related minerals which have been found in nature and have been denominated by the International Mineralogical Association (IMA2012-049 [18]):

rib

U

se

O

FeIII containing solid compounds are well known for their strong adsorption capacity for phosphate species. Compounds with high specific surface area such as ferrihydrite easily adsorb phosphate species but such pulverulent material can generally not be used directly for flow through treatments since the release of red water containing FeIII nano-crystallites is not desired. Nano-crystallites, however, when physically stabilized by different ways, e.g., (i) embedded inside a matrix [4-6], (ii) granulated to form millimetric sized aggregates [7] or (iii) deposited on various substrates [8-10], may be suitable hosts, for the process. Low cost materials such as steel slag [11], hydroxyapatite [1213], layered double hydroxides (LDH), limestone and sand shell sand [14], appear also suitable to treat phosphate under waste water treatment conditions. An attempt to compare the efficiency of these materials will be performed in section III.1.1.d.

(iii) mössbauerite with x = 1, FeIII6 (OH)8 O4 CO3 • 3H2O, a fully ferric compound, previously called “ferric green rust GR*”.

nl y

methods. For example, in urban areas nitrate is often treated in large scale plants by using the activated sludge process and phosphate is eliminated by precipitation processes using ferricor aluminum-based materials. Such treatments, however, are not always cost-effective for rural settings where smaller infrastructure with minimized maintenance is generally required. Waste water plants using passive treatment, e.g. low cost reed bed filters and lagooning, are often preferred for sparsely populated regions. Such installations are relatively efficient for removing organic matter from water but the rates of denitrification and dephosphatation are limited [2,3]. Therefore, tertiary passive treatments that eliminate both nitrate and phosphate from waste water are required.

101

(i) fougèrite with x = 1/3, GR(CO32-), FeII4 FeIII2 (OH)12 CO3 • 3H2O, (ii) trébeurdenite with x = 2/3, FeII2 FeIII4 (OH)10 O2 CO3 • 3H2O,

2. SYNTHESES OF THE INORGANIC IRON BASED MINERALS 2.1. Abiotic synthesis 2.1.1. The 2-Line Ferrihydrite and Lepidocrocite The 2-line ferrihydrite (Fh) is prepared by precipitation of ferric species ([Fe3+]0 = 0.8 M) with a 3 M NaOH solution until the pH reaches 7.5. Both FeIIICl3•8H2O and FeIIINO3•9H2O salts are used for the preparations and the resulting Fh samples are labelled as Fh(Cl) and Fh(NO3), respectively. The dry powders are then washed 3 times by centrifugation to eliminate residual chloride or nitrate ions from the surface [25]. Both the samples display two broad peaks in their X-ray diffraction patterns (more or less similar patterns), as expected. Only XRD pattern of Fh(Cl) is shown in Fig. 1, as an example. The samples are “gel like”, based on transmission electron microcopy (TEM) observations [26]. The specific surface areas (SSA) of Fh(Cl) and Fh(NO3) are 300 m2 g-1 and 210 m2 g-1, respectively.

102 Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Ruby et al.

that the crystal morphology of mössbauerite was very similar to the morphology of the GR(CO32-) precursor [30].

Intensity (u.a.)

Fh(Cl)

*

GRa

*

20

30

40

50

(511)

(020)

(220)

lepidocrocite

60

70

80

2 Theta (°) Fig. (1). X-ray diffraction patterns of the ferric compounds synthesized in this study: Fh(Cl) is ferrihydrite synthesized in the presence of Cl-; GR*a and GR*c is synthetic mössbauerite synthesized by two different pathways (see section II-1-2); lepidocrocite is FeOOH.

ut

c

rD

Pe

2.1.2. Hydroxycarbonate Green Rust (Fougèrite) and “Ferric Green Rust” (Mössbauerite)

N ot

fo

FeII-III hydroxycarbonate green rust, GR(CO32-), FeII4 Fe 2 (OH)12 CO3 • 3H2O that is homologous to fougèrite mineral, is synthesized by the coprecipitation of a mixture of soluble ferrous and ferric species by an alkaline solution [27]; i.e. FeIISO4 •7H2O and FeIII2(SO4)3 • 5H2O salts ([FetoIII tal] = 0.4 M and x = {[Fe ] / [Fetotal]} = 0.33) are dissolved in demineralized water and precipitated by a carbonated alkaline solution where {[OH-] / [Fetotal]} = 2 and {[CO32-] / [SO42-]} = 1. Mössbauerite synthetic mineral is prepared in two ways: either, the GR(CO32-) suspension is immediately oxidized by dropwise addition of H2O2 solution (“GR*a” hereafter), or, it is aged during 24 hours prior to its oxidation by H2O2 (“GR*c” hereafter) [30]. The XRD pattern of GR*c presents some kind of analogy with the XRD pattern of ferrihydrite except for the presence of an additional intense low angle diffraction peak corresponding to an interlayer distance of about 7.4 Å (Fig. 1). In contrast, this peak is not observed for GR*a. Such a line corresponding to 7.4 Å is very close to the interlayer distance measured for GR(CO32-), i.e. 7.6 Å [33]; it strongly suggests that the carbonate anions remain present in the GR*c structure and hexagonal shape crystals are observed (Fig. 2a) similar to those of hydroxycarbonate GR(CO32-) [30]. In contrast, undefined crystal shapes are observed for GR*a (Fig. 2b). The SSA of GR*a and GR*c are ~120 m2 g-1 and ~ 65 m2 g-1, respectively. It was shown III

rib

is t

rs on

al

U

io n

se

O

On the contrary to ferrihydrite, lepidocrocite cannot be obtained by direct precipitation of ferric salt but is rather prepared by oxidation of ferrous species [27]. Lepidocrocite is prepared by vigorous air oxidation of hydrolyzed Fe2+ species in the presence of chloride anions [28] and exhibits XRD and TEM characteristic of pure and crystalline lepidocrocite (Fig. 1 and 2c). The XRD pattern contains a series of sharp diffraction peaks which can be indexed in the orthorhombic system, space group Bbmm. Crystallographic unitcell parameter values (a = 12.512 (5) Å, b = 3.879 (53) Å and c = 3.071 (84) Å) are obtained from the XRD data

nl y

10

(600)

(101)

(111)

(210)

(301)

(200)

GRc

200 nm Fig. (2). TEM images of the ferric compounds synthesized in this study. (a) Synthetic mössbauerite GR*c, (b) Synthetic mössbauerite GR*a, (c) Lepidocrocite -FeOOH.

2.1.3. Ferrihydrite- and Mössbauerite-Coated Pozzolana Pozzolana (Pz) with a grain size of 1-3 mm is furnished by "Pouzzolane des Dômes” (Saint Ours les Roches, France). Pozzolana is a composite of labradorite (Na,Ca)(Al,Si)4O8 (ICDD file # 831371), diopside CaMgSi2O6 (ICDD file # 78-1390), forsterite Mg2SiO4 (ICDD file # 75-1446) and hematite -Fe2O3 (ICDD file # 89-0599), based on XRD. It is highly porous (~70% porosity), which is evident from appearance of multiple alveoli in the size range  50 - 500 m in its SEM images (Fig. 3). The main chemical elements detected by Energy Dispersive X-Rays (EDX) for uncoated Pz are O, Si, Ca and Al (Table 1).

Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Use of Ferrihydrite-Coated Pozzolana and Biogenic Green Rust

103

N ot

fo

rD

is t

rib

al rs on Pe

Mass concentration of the chemical elements detected by SEM-EDX for uncoated pozzolana, ferrihydrite-coated pozzolana and mössbauerite-coated pozzolana.

ut

U

Table 1.

io n

se

O

nl y

The coated materials were obtained by mixing both compounds in a dry state [10]. This method allows maximum deposition of the iron oxides ( 8 g%), independent of nature of oxide, i.e., Fh or GR*. This quantity is higher than the amount of iron oxides that can be deposited on Fontainebleau sand ( 1 g% [34]) having much lower grain size of  200 m. Because the apparent density of Fontainebleau sand ( 1.5 g cm-3) is relatively close to the apparent density of Pz ( 1 g cm-3), the amount of iron oxide that can be introduced inside an identical volume is about 5 times higher when Pz is chosen as a supporting material instead of sand. Except for oxygen, the main chemical elements of Pz, i.e. Si, Ca and Al, are almost undetected with EDX when ferrihydrite-coated Pz and mössbauerite-coated Pz are analyzed and the expected increase of iron concentration is observed (Table 1). The good coverage of the Pz surface by the iron oxide coating is evidenced by SEM (Fig. 3). Aggregates of iron oxides are either embedded inside the Pz alveoli or form a coating deposited on smoother part of the Pz surface (Fig. 3b). It is even possible to estimate a thickness of  8 m for this coating by imaging the frontier between an uncoated and coated areas (Fig. 3c & d). It was checked that the Fh and GR* coatings are stable in lixiviation column experiments by using a descendant vertical flow rate as high as 1 L s-1: very minor quantity of iron oxides is released in the outflow in these experimental conditions.

Chemical Elements

Uncoated pozzolana

FerrihydriteMössbaueritecoated pozzolana coated pozzolana

O

44.2

22.7

32.9

Na

2.5

-

3.6

Mg

1.7

-

-

Al

10

0.2

0.5

Si

23.4

0.5

0.3

K

0.8

-

-

Ca

13

-

-

S

0.4

P

0.7

Fe

4.4

76.6

61.6

2.2. Synthesis of Hydroxycarbonate Green Rust by Bioreduction of Ferric Oxyhydroxides

Fig. (3). SEM images of mössbauerite coated pozzolana (Pz): (a) uncoated Pz, (b), (c) and (d) coated Pz at different scales.

Biogenic “green rust-related compounds” (referred to as “GRbio” hereafter) can be prepared by bioreduction under anoxic conditions at neutral pH of either mössbauerite or lepidocrocite, using an inoculum of Shewanella putrefaciens, as reported previously [21, 22, 35]. A recent study

104 Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Ruby et al.

1 µm

bio-GR

nl y

bio-GR

bio-GR

1 µm

1 µm

D1: Fe(II)GR 98.5 98 97.5

D3: Fe(III)GR

96.5 96

-4

-3

-2

-1

0

1

2

ut

rib

3.1. Phosphate Adsorption

is t

D2: Fe(II)GR

97

3. RESULTS AND DISCUSSION

rD

fo

99

Fig. (5). Scanning electron microscopy images of biogenic hydroxycarbonate green rust (bio-GR) obtained by bioreduction of lepidocrocite.

3.1.1. Experiments Performed in Homogeneous Suspensions (“Batch Experiments”)

N ot

Transmittance (%)

99.5

io n

se

U

al

rs on

Pe

100

1 µm

bio-GR

O

[35] has provided evidence that the rate of microbial Fe(II) production was faster when mössbauerite was used as an electron acceptor instead of lepidocrocite. Initial rates of Fe(II) production calculated according to a second-order kinetic model on the first five days were 5.3 and 3.2 mM-1 h-1 in experiments with mössbauerite and with lepidocrocite, respectively [35]. Fig. 4 displays an example of Mössbauer spectrum for a biogenic GRbio obtained upon reduction of lepidocrocite by Shewanella putrefaciens after one month of incubation. This 10 K Mössbauer spectrum exhibits three paramagnetic quadrupole doublets D1, D2 and D3. Doublets D1 and D2 with large center shift (δ) and quadrupole splitting (Δ) (δ1 and 2 ≈ 1.25 mm/s, Δ1 = 2.99 mm/s and Δ2 = 2.72 mm/s) are characteristic of Fe2+ (55%) in hydroxycarbonate GR, while doublet D3 with small hyperfine parameters (δ3 = 0.42 mm/s, Δ3 = 0.41 mm/s) is typical of Fe3+ (45%) in hydroxycarbonate GR. This 10 K Mössbauer analysis revealed that lepidocrocite was fully converted to green rust, and that the x = Fe(III)/Fe molar ratio of this GRbio was equal to 0.45. However, fitting of the Mössbauer spectrum with only Lorentzian shape lines is absolutely not satisfactory. This fitting procedure does not take into account the structural models of GR at x  0.33 proposed previously that involves constraints between the relative area of the ferrous and ferric doublets [36]. As it will be shown later for GR samples prepared abiotically at x values higher than 0.33 (section III.2.1.a), fitting the Mössbauer data by taking into account the structural models and by using a pseudo-Voigt profile analysis leads to a better agreement between the experimental and calculated spectra. SEM images of GRbio show hexagonal-shaped crystals with various diameters ranging from 2 to 5 µm (Fig. 5).

3

4

-1

Velocity (mm s )

Fig. (4). Transmission Mössbauer spectroscopy spectrum recorded at 10 K for biogenic hydroxycarbonate green rust sample obtained by bioreduction of lepidocrocite. Experimental and fit curves are displayed as dotted black and dashed red lines respectively. The fit components displayed as plain lines correspond to three doublets D1, D2 and D3.

a) Kinetics of Adsorption

Fh(Cl) and Fh(NO3) adsorb phosphate with a quite similar kinetics (Fig. 6). The saturation of the surface sites occurs at reaction time higher than 20 hours with a maximal phosphate adsorption capacity at equilibrium qe of  80 mg of PO4 per gram of Fh at a pH of 7. This value corresponds to an adsorption capacity, normalized to the surface area, of 0.27 mg m-2 and 0.38 mg m-2 for Fh(Cl) and Fh(NO3), respectively. Mössbauerite adsorbs phosphate more rapidly and the saturation of the surface sites is obtained after a few hours of reaction time (Fig. 6a). As expected from the specific surface area (SSA) values of GR*, the maximal phosphate adsorption capacities at equilibrium qe, i.e.  45 mg g-1 and  35 mg g-1 for GR*a and GR*c, respectively, are lower than the values measured for Fh. The adsorption capacity, normalized to the surface area, is 0.37 mg m-2 and 0.53 mg m-2 for GR*a and GR*c respectively. The normalized value of the adsorption capacity measured for GR*a is relatively close to the one measured for ferrihydrite. GR*c seems to have an intrinsic phosphate normalized adsorption capacity higher than ferrihydrite. This difference could be related to a higher affinity of phosphate for the lateral surface site of the hexagonal crystals of GR*c as already observed for hydroxycarbonate GR [29].

Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Use of Ferrihydrite-Coated Pozzolana and Biogenic Green Rust

directly correlated to an increase in the ionic strength of the aqueous solution and confirms the formation of an inner sphere complex between phosphate and the Fh surface [3739]. The important result shown here is that the phosphate adsorption on ferric oxyhydroxides is quasi-independent of the presence of other soluble anions; it is a major advantage for water treatment application.

50

-1

qt (mg g )

40 30 *

GRa

20

*

-

GRc

Fh(Cl ) * GRa

100

10

*

-1

qe (mg g )

80

60

40

60

nl y

20 -

4

al

9

100

65 %

80 % 68 %

71 %

63 %

is t

rs on

50

N ot

fo

rD

A decrease of pH induces a continuous increase of the phosphate adsorption capacity qe for all the three compounds Fh(Cl), GR*a, GR*c (Fig. 7). It confirms also that the adsorption of Fh is higher than that of GR* independently of the pH. Acidic conditions would be preferable for optimizing the adsorption process. In real treatment condition, the pH depends both on the type of water to be treated and on the chemical nature of the filtration material. Waste water has generally a slightly alkaline pH of  8 and demineralized water in equilibrium with ferric oxyhydroxides has a slightly acidic pH close to  6. Therefore circumneutral pH conditions are expected during the adsorption process of phosphate present in waste water by iron oxides. Note that even at pH = 9, where the point zero of charge of ferrihydrite and mössbauerite is overpassed (pzc  8), the adsorption of phosphate is still observed. In order to explain the pH dependence of qe, a multisite complexation (MUSIC) model was used and it was shown that essentially bidentate complexes form at the Fh surface [37].

Pe

8

io n

1200

Fig. (6). Kinetics of phosphate adsorption on different ferric compounds.

b) Influence of the pH

7

ut

800

Time (min)

rib

400

U

0

6

Fig. (7). Phosphate adsorption capacity of different ferric compounds (ferrihydrite and mössbauerite) as a function of pH.

b

0

5

pH

se

20

0

O

Fh(Cl ) Fh(NO3)

40

Percentage of adsorption (%)

-1

GRc

80

a

0

qt (mg g )

105

c) Influence of Co-Existing Anions other than Phosphate The adsorption of phosphate ([PO43-] = 70 mg L-1) on a 1 g L-1 Fh(Cl) suspension is studied in the presence of other soluble anions such as Cl-, NO3- and SO42-. In isoconcentration, these anions do not affect the phosphate adsorption (Fig. 8). Only the co-existence of chloride in very high concentration ([Cl-] = 3000 mg L-1) induces a slight increase of the phosphate adsorption (Fig. 8e). This effect is

(a) PO4

(b) PO4 + Cl

(c) PO4 + Cl + NO3 + SO4

(d) PO4 + Cl

(e) PO4 + Cl

Fig. (8). Percentage of phosphate adsorption on ferrihydrite as a function of the anions present in the aqueous medium. Phosphate concentration is constant in all the experiments, i.e. [PO43-] = 70 mgL-1. For the experiments (b), (c) & (d), the concentration of coexisting anion is identical to the phosphate concentration, i.e. [Cl-] = [SO42-] = [NO3-] = 70 mg L-1. For experiment (e), the chloride concentration is [Cl-] = 3000 mg L-1.

d) Comparison Between Various Sorption Materials The comparison of the phosphate adsorption capacities between different materials is more straightforward when values of qe (in mg PO4 g-1) are determined from kinetic or adsorption isotherm experiments performed in homogeneous suspensions (“batch” experiments). According to the previous sections, such a comparison should be performed in similar experimental conditions; in particular the pH of the experiments should be identical. Three categories of adsorbent could be distinguished: (i) chemically synthesized materials essentially iron oxides [25, 30, 31 41,42] and layered double hydroxides [43-45], (ii) minerals such

106 Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Ruby et al.

Comparison of the phosphate adsorption capacities of various materials.

O

Table 2.

lution-reprecipitation mechanism was clearly identified for Ca-Al LDH that transformed in a mixture of gibbsite and hydroxyapatite {Al(OH)3, Ca5(PO4)3(OH)} in the presence of phosphate [45]. This is probably the reason why Cabased LDHs present generally higher phosphate adsorption capacities than Mg-based LDHs [49]. Despite the fact that Ca-based LDHs present phosphate adsorption capacities about five times higher than the best iron oxides sorbents, e.g. ferrihydrite, their main drawback is their alkaline characters with equilibrium pH in water generally higher than 11 [45]. Such a trend is also observed for other calcium containing materials such as hydroxyapatite or steel slags that present, moreover, lower phosphate adsorption capacities with qe values situated in a range between 0.4 and 12.5 mg PO4 g-1. In summary, Fh that possesses a relatively high adsorption capacity (qe  80 mg PO4 g-1 at pH 7) and an equilibrium pH in water close to neutrality is a good compromise for water dephosphatation application.

nl y

as apatite [12,13] and dolomite [46] (iii) industrial byproducts or waste, e.g. steel slags [11], as illustrated in Table 2. The pH conditions and, if available, the values of the SSA (in m2 g-1) of the materials are mentioned. When the experiments are not performed in static pH conditions, initial and final pH are given. Among the ferric oxides family, the phosphate adsorption capacities are strongly correlated to the SSA and both mössbauerite and ferrihydrite represent the best sorbents. MII-MIII layered double hydroxides (LDH) are also very interesting materials as phosphate adsorbent since several mechanisms could potentially be involved: (i) phosphate adsorption on the external LDH crystals faces, (ii) anionic exchange between the anions present in the LDH interlayer and phosphate anions, (iii) dissolution of the LDH and reprecipitation of new compounds containing phosphate. The intercalation of phosphate anions replacing NO3- ions inside Mg-Al and Ca-Fe LDH structure was observed by following the position of the low angle (003) XRD peaks [47-49]. The disso-

Metallic oxides AND HYDROXIDES

Chemical formula

io n

(m g )

Ferric oxides

rib

Lepidocrocite

(mg PO4 g-1)

14

[41]

32

6.7

18

[40]

94

7

21

[41]

“Fe O4(OH)8CO33H2O”

65-120

7

35-48

[30]

Fe8.2O8.5(OH)7.4,3H2Ob

210-300

7

80

[25]

is t

rs on

a

fo

III 6

rD

-FeOOH

Pe

Ref.

7

-FeOOH

Akaganeite

qe

85

-FeOOH

Goethite

pH

-1

ut

al

Material name

Ferrihydrite

2

U

or

Mössbauerite

SSA

se

Mineral name

N ot

Layered Double Hydroxides

Pyroaurite

[Mg0.80Fe0.20(OH)2](CO3)0.14(H2O)0.62

-

7.5

15

[43]

Hydrotalcite

Mg0.683Al0.317(OH)1.995 (CO3)0.028 Cl0.226, 0.54H2O

-

7.8

143

[44]

Hydroxycalumite

Ca4Al2(OH)12Cl2,2H2O

-

7-11

400

[45]

MINERALS

Apatite

Ca5(PO4)3(OH,Cl,F).

-

7

4.8

[12]

Apatite

Ca5(PO4)3(OH,Cl,F).

-

8

0.3-1.1

[13]

Dolomite

CaMg(CO3)2

0.14

7

47

[46]

STEEL SLAGS

a

Electric arc furnace (EAF) steel slags

-

Particle size 510 mm

7-11.5

0.4-0.9

[11]

Basic oxygen furnace (BOF) steel slags

-

Particle size 510 mm

7-12.5

3.4-7.5

[11]

Chemical formula proposed by assuming that the oxidation of hydroxycarbonate green rust into Mössbauerite occurs with no release of carbonate anions. The excess of positive charge due to the FeIII cations is fully compensate by a deprotonation of the brucite-like sheets. b Chemical formula proposed by Michel et al. (Science, 2007,316,1726-1729)

Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Use of Ferrihydrite-Coated Pozzolana and Biogenic Green Rust

a) Influence of the Nature of the Coatings

qB (mg PO4 g-1)  -21 F (mL min-1) + 34 for Fh qB (mg PO4 g-1)  -9.5 F (mL min-1) + 17 for GR*a Table 3.

0.0 0

12

26

50

100

coated pozzolana

(demineralized (municipal waste water) water) 10.5

Section (cm2)

5.3

5.3

5.3

Volume (cm3)

56.7

72.3

55.6

Porous volume Vp (cm3)

40

57

40

Material mass (g)

55

70

55

io n

ut

rib

is t

N ot

-1

6 mg L

(demineralized water)

coated pozzolana

Ferrihydrite

13.6

1.0

0.8

C/Ci

-

Fh(Cl ) * GRa

0.2

coated pozzolana

Mössbauerite GR*a

10.7

se

0.4

Ferrihydrite

Length (cm)

rD

C/Ci

0.6

fo

Pe

0.8

(2)

Physical characteristics of the columns containing ferrihydrite and mössbauerite

Column characteristics

U

al rs on

1.0

(1)

and

O

The two best phosphate sorbents identified in homogeneous suspensions, e.g. Fh(Cl) and GR*a, are deposited onto Pz according to the procedure described in section II-1-3. The experiments are performed with a constant mass of  5 g of iron oxides deposited on Pz. These filter materials are introduced inside the columns characterized by the physical parameters given in Table 3. The breakthrough curves obtained at 0.5 mL min-1 flow rate shows clearly that Fh adsorbs more efficiently phosphate than GR*a, confirming the results obtained in batch experiments (Fig. 9). The breakthrough volume VB is chosen for a C/Ci value of 6 %, where C and Ci represent the concentration of phosphate present in the outflow and the inflow solutions, respectively (Ci = 100 mg L-1). The limit of 6 % corresponds therefore to a concentration of phosphate of 6 mg L-1 that should not be exceeded. The ratio between the volume at the breakthrough to the porous volume VB/VP is shifted by factor of 2.5 if Fh is compared to GR*a. The phosphate adsorption capacities at the breakthrough qB are 22 mg PO4 g-1 and 12.5 mg PO4 g-1 for Fh and GR*a, respectively. For comparison purpose to the experiments performed in batch, the adsorption capacities are normalized here relatively to the quantity of iron oxides and not to the total mass of filtration material. The results confirm those obtained in batch and show that Fh is also the best sorbent in hydrodynamic conditions.

reaction time is not completely sufficient to saturate the surface sites of iron oxides and it should be even truer under hydrodynamic condition. Experiments with higher residence times are not performed for practical reason: a column experiment with a residence time of 24 hours could last more than 4 months. A relative linear relationship between the values of capacities at the breakthrough qB and the flow rate is measured if data of Table 4 are considered and the following equations are obtained:

nl y

3.1.2. Experiments Performed Under Hydrodynamic Conditions (“Column Experiments”)

107

0.6 -1

0.1 mL min -1 0.5 mL min -1 1.0 mL min

0.4

0.2

150

V/Vp Fig. (9). Comparison of the phosphate breakthrough curves obtained in hydrodynamic condition between ferrihydrite (Fh) and mössbauerite (GR*a).

b) Influence of the Flow Rate The flow rate F has a strong influence on the breakthrough curves (Fig. 10). The breakthrough volume is shifted to higher values for decreasing flow rate and increasing residence time. This phenomenon is observed independently from the nature of the coating, i.e. ferrihydrite or mössbauerite (Table 4). The flow rate of 0.1 mL min-1 corresponds to residence time situated between 6.5 and 9 hours depending on the geometry of the columns (Table 3). According to the kinetics of adsorption performed under batch condition, such

0.0 0

50

100

150

V/Vp

Fig. (10). Influence of the flow rate on the phosphate breakthrough curve obtained for mössbauerite in hydrodynamic conditions.

According to the first equation, increasing the residence time from 6.5 to 24 hours would only have a minor impact on the adsorption capacity of Fh coated Pz that would slightly increase to qB values close to  33 mg PO4 g-1. Finally, the integration of the upper part of the breakthrough curves allows the determination of the total phosphate adsorption capacity at saturation that is independent from the flow rate; qsat values of 45 and 53 mg PO4 g-1 are measured for GR*a and Fh, respectively (Table 4). These values are comparable to those obtained under batch condition in a similar pH range

108 Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Table 4.

Ruby et al.

Breakthrough and phosphate adsorption capacities determined in column experiments for different flow rates.

Filtration materials

Flow rate

Breakthrough

F

VB/VP

Adsorption capacity at the breakthrough qB (mg PO4 g-1)

Adsorption capacity at saturation qSat (mg PO4 g-1)

-1

(mL min )

Mössbauerite coated pozzolana

1

17

14

53

0.5

28

22

53

0.1

36

33

53

1

8

8

45

0.5

11

12.5

45

0.1

17

16

45

situated between 6 and 9, i.e.  59 mg PO4 g-1 ≤ qe(Fh) ≤  87 mg PO4 g-1 and  30 mg PO4 g-1 ≤ qe(GR*a) ≤  51 mg PO4 g-1.

O

c) Sorption of Phosphate Present in Municipal Waste Water

hydroxychloride and hydroxysulphate GR under circumneutral pH condition, nitrate was fully reduced into ammonium and GR was transformed into magnetite Fe3O4. If the same reaction is assumed here, the following transformation should occur:

nl y

Ferrihydrite coated pozzolana

4 FeII4FeIII2(OH)12CO3 + NO3- ↔ 8 FeIIFeIII2O4 + NH4+ + 6 (3) H+ + 4 CO32- + 19 H2O

io n

ut

Demineralized water Recycled water

rib

1.0

C/Ci

is t

0.8

rD

fo

N ot

Pe

rs on

al

U

se

A supplementary column experiment is performed by using pretreated municipal waste water withdrawn from the outflow of a SAUR® sewage water plant (Dombasle sur Meurthe, Lorraine, France). The characteristics of water are as following: pH = 6.9, BOD5 = 5.5 mg L-1, COD = 14 mg L1 , TSS = 5.5 mg L-1, [SO42-] = 300 mg L-1, [Cl-] = 100 mg L1 , [NO3-] = 5 mgL-1, [PO43-] = 1 mg L-1, [Ca2+] = 95 mg L-1 and [Mg2+] = 29 mg L-1. In order to mimic conditions that can be encountered inside the water outflow of reed bed filters, phosphate and nitrate ions are added to this solution in order to reach an equimolar concentration [NO3-] = [PO43-] = 100 mg L-1. Water is then introduced in a column containing Fh coated Pz (mFh = 5 g) with the physical characteristics given in Table 3. Despite of the presence of organic matter and other competitive anions, the breakthrough curve corresponding to municipal water is surprisingly shifted to higher V/Vp values in comparison with the curve obtained with demineralized water containing an identical concentration of phosphate (Fig. 11). In a recent study [50], it was clearly demonstrated that the increase of phosphate adsorption capacity observed in the presence of waste water was due to presence of soluble calcium. The co-adsorption of phosphate and calcium at the ferrihydrite surface or the surface precipitation of apatite may explain this observation, but further investigations are required to completely understand the mechanism of this synergetic effect. 3.2. Nitrate and Nitrite Reduction 3.2.1. Reactivity of Abiotically Synthesized Green Rust a) Kinetics of the Reaction Experiments performed at the pH of synthesis of GR (pH  10.5). Hydroxycarbonate GR synthesized by the coprecipitation method of Bocher et al. [29], where the FeII and FeIII species are precipitated by a {Na2CO3, NaOH} mixture leads to an equilibrium pH of  10.5. According to the previous works of Hansen et al. [15,16] who studied the reactivity of

0.6

0.4

0.2

Shift of breakthrough point

0.0 0

50

100

150

V/Vp Fig. (11). Comparison of phosphate breakthrough curves obtained either in demineralized water or in water coming out from the outflow of a sewage water plant.

The formation of magnetite was indeed confirmed by Raman and X-ray diffraction [20]. Note that this reaction corresponds to the stoichiometric ratio [NO3-]: [FeII]GR of 1: 16 = 6.25%. Two experiments, the first one in excess of NO3- species ([NO3-]: [FeII]GR = 12.7%) and the second one in excess of FeII species ([NO3-]: [FeII]GR = 2%), are performed to validate the occurence of reaction (3). The nitrate concentration is measured in the aqueous suspension by ion chomatography after filtration of the GR solid particles. The total concentration of ferrous species in the suspension present in both GR and the aqueous solution is measured by the ferrozine method (Viollier et al. 2000). In fact, at a pH of 10.5, FeII species are essentially incoporated into solid GR and the measured concentration is therefore written as [FeII]GR. The time evolution of the nitrate and fer-

Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Use of Ferrihydrite-Coated Pozzolana and Biogenic Green Rust

rous concentrations present in the suspension is monitored (Fig. 12). In the presence of an excess of FeII species ([NO3]: [FeII]GR = 2%), nitrate species are almost fully reduced in 7 days of reaction. As discussed in previous studies [15, 16, 18], the variation of the nitrate concentration can relatively well be adjusted with a pseudo-first order law obeying to equation (4): [NO3-]t = [NO3-]0 × exp(-kobs × t)

(4)

(a)

2

-

-

II

-

II

nl y

1

[NO3 ] : [Fe ] = 12.7 % [NO3 ] : [Fe ] = 2 % 0

2

4

6

8

se

Reaction time (days)

al

120

ut

U

160

-

II

(b)

II

[NO3 ] : [Fe ] = 12,7 % 0 0

2

4

6

rib

is t

2+

8

8

N ot

Reaction time (days)

rD

-

[NO3 ] : [Fe ] = 2 %

16

fo

40

Pe

80

[Fe ] (mM)

rs on

24

2+

[Fe ] (mM)

Influence of the phosphate concentration and pH. The kinetics of nitrate reduction by GR is strongly influenced by both the pH and the phosphate concentration (Fig. 13). A significant decrease of the reaction kinetics is observed for increasing amount of phosphate and the reaction is fully inhibited for a relative concentration ratio [PO4]: [FeII]GR of 2.4%. This behavior may be attributed to a stronger affinity of GR surface for phosphate anions rather than for nitrate anions. Indeed, it was shown that phosphate anions were preferentially adsorbed on the lateral faces of the hexagonal GR crystals [29]. Moreover, high resolution TEM experiments performed on hydroxysulphate GR reacting with chromate proved that the reaction began at the lateral edges of the GR crystals [51]. From these previous results, one may hypothesize that phosphate anions are quickly adsorbed on the lateral faces of the GR crystals and act as a physical barrier that restrains NO3- ions to interact with FeII species present in the solid.

O

0

(Fig. 12a). The initial concentration of FeII species decreases more significantly by a factor very close to 2. As described in reaction (3), only the half of the initial FeII species are oxidized into FeIII species if the final oxidation product is magnetite FeIIFeIII2O4. A value of P = 8.8 ± 0.7 is measured in this case that is even closer from the stoichiometric value of 8. As discussed earlier (Supporting information of [20]) the formation of nitrogen species other than ammonium would have led to significantly lower P values of 5, 4 and 3 for N2, N2O and NO, respectively. Moreover, complementary measurements of the concentration of ammonium produced confirm that this concentration produced after 7 days of reaction time ([NH4+]7days) is very close to the concentration of nitrate removed ([NO3-]0 - [NO3-]7days). All these results show that reaction (3) describes quite well the occurring reaction when hydroxycarbonate GR is in contact with nitrate at the pH of synthesis, i.e. pH = 10.5. As it will be seen later in section III-2-1-b, reaction (3) is not the unique occurring reaction if either extra GR stabilizing anions such as phosphate are introduced in the suspension or if the reaction is performed at a lower pH.

io n

[NO3 ] (mM)

3

109

Fig. (12). Evolution of the nitrate and total iron(II) concentrations during the reduction of nitrate by hydroxycarbonate green rust.

Where [NO3-]t and [NO3-]0 correspond to the nitrate concentration for a reaction time t and the initial nitrate concentration (here [NO3-]0 = 3.2 mM or 200 mg L-1), respectively. Equation 4 can be easily linearized and the fit of the data leads to a kobs value of 5.4 ± 0.2 x 10-6 s-1 in good agreement with previous works [15, 16, 20]. As expected, the initial concentration of nitrate is sufficient to reduce only partially the ferrous species (Fig. 12b). The ratio P between the concentration of FeII species removed relatively to the concentration of nitrate removed, i.e. P = {[FeII]7days - [FeII]0} /{[NO3-]7days - [NO3-]0}, is close to 9 ± 3, a value in relative good agreement with the stoichiometric value of 8 expected from reaction (3). In the presence of an excess of NO3- species ([NO3-]: [FeII]GR = 12.7%), nitrate species are only partially reduced

The dependence between the reactivity of GR towards nitrate and the pH is not linear (Fig. 13). The kinetics of the reaction can be followed by measuring the fraction of nitrate removed X(NO3-)t that corresponds to the following equation: X(NO3-)t = {[NO3-]0 - [NO3-]t} / [NO3-]0

(5)

The reaction kinetics is faster at a pH value of 10.5 than that observed at pH of 9 with nitrate consumption X(NO3)7days that can reach  97% and  14% of the initial concentration after 7 days of reaction, respectively. An intermediate situation is observed at a pH of 7.5 with a maximal X(NO3)7days value that can reach  40%. Choi and Batchelor [17] who studied the reactivity of hydroxyfluoride GR modified with copper towards nitrate ions found a higher reactivity under alkaline conditions (pH 9 and 11) in comparison with that measured under circumneutral pH conditions (pH = 7.8). All these results are a bit in contradiction with the point of zero charge measured for GR that was found recently to be 8.3 independently from the nature of the intercalated anions, i.e. carbonate or sulfate anions [52]. On the contrary to what is experimentally observed, at a pH higher than 8.3 the sur-

110 Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Transmittance %

face of the GR crystals would be negatively charged [52, 53] and this should limit the reactivity of GR towards negatively charge nitrate species. However, reaction (3) that is observed to be quite fast at a pH of 10.5 is clearly pH dependent and should also be favored under alkaline condition according to Le Chatelier principle.

Probability density (au)

Ruby et al.

100 99 98 97 96 95 94

D2

x = 0.33

D3 D1

(a)

78 K

-4 -4

-3 -3

-2-2

-1-1 0 0

11

Velocity (mm s-1)

22

33

44

x = 0.33 D1 50 %

D3 33 %

D2 17 %

(b)

78 K -1

0

1

2

3

Quadrupole splitting (mm s-1)

Transmittance %

100

90 -

[NO3 ] / [NO3 ]0 Nitrate reduction rate X(NO3 )

80

(a) pH = 7.5 70

-

Probability density (au)

x ~ 0.50

100

96 x ~ 0.50

92 88

(c)

78 K 84

-4

-3 -2-2 -1 -1 00 -3

11

22

3

Velocity (mm s-1)

44

D1 38 % D4 16.5 %

(d)

78 K -1

D2 12.5 %

0

1

2

3

Quadrupole splitting (mm s-1)

-

60

D3 33 %

50

PO4 / Fe = 0 %

40

PO4 / Fe = 2.4 %

Fig. (14). (a) Mössbauer spectra of hydroxycarbonate green rust GR(x=0.33) and of (c) oxyhydroxycarbonate GR(x= 0.5) adjusted with Lorentzian shape lines (b) and with a pseudo-Voigt profile (d), respectively. (d) The position of the different ferrous (D1, D2) and ferric (D3 and D4) components of the pseudo-Voigt profile analyses is indicated.

PO4 / Fe = 0.6 %

nl y

30 20 0

2

4

6

8

0

2

4

100 90

-

6

rib

is t 8

N ot

Time (Days)

fo

40

80 70 60

(c) pH = 10.5

50 40

PO4 / Fe = 0 % 30

PO4 / Fe = 1.2 %

20

PO4 / Fe = 2.4 %

10 0

2

4

6

The lamellar structure of GR(CO32-) was preserved with slight contraction of the c parameter and a continuous deprotonation of the brucite-like layer was observed by X-ray diffraction later confirmed by photoelectron spectroscopy [55]. Mössbauer spectra of GR(CO32-) and GR-related material can be compared (Fig. 15). It is clear that the spectrum of GRbio is strikingly similar to that obtained by H2O2 addition when x lies around 0.5. Let us recall that the spectrum of GR(CO32-) with x = 1/3 is fitted with two ferrous doublets (D1 and D2) and one ferric doublet (D3) having Lorentzian shape lines (Fig. 14a & b). In agreement with the structural model proposed previously [36], the relative area of these doublets is within the experimental error 1/2, 1/6, 1/3. In contrast, the spectrum for x = 0.5 must be fitted with Voigt profile analysis (Fig. 14c & d) using four Gaussian distributions of quadrupole doublets; ferrous doublets D1 and D2 with large splitting and ferric doublets D3 and D4 with small splitting; the relative area of these doublets is within the experimental error 3/8, 1/8, 1/3, 1/6. It is clear that part of the FeII ions within GR(CO32-) got oxidized to form doublet D4. Therefore, D3 corresponds to the initial FeIII ions balancing the carbonate anions whereas D4 corresponds to FeIII ions due to the deprotonation of OH- ions. One notes also that D1 and D2 distinguish FeII cations whether they are out of the proximity of a CO32- anion or within it. The reason for using Voigt profile analysis proving that the quadrupole splitting has no longer a Dirac distribution but a Gaussian distribution is due to the deprotonation of some OH- ions at the apices of the octahedral sites that causes the electric field gradient (EFG) axis direction to fluctuate slightly around the c axis of layer stacking.The spectrum of the GRbio sample is presented as an example in Fig. 15b; it displays the features of the ref-

rD

PO4 / Fe = 2.4 %

20

[NO3 ] / [NO3 ]0

ut

U rs on

PO4 / Fe = 0.5 % 50

30

Nitrate reduction rate X(NO3 )

al

PO4 / Fe = 0 %

60

Pe

-

-

[NO ] / [NO ]

(b) pH = 9

70

io n

se

90

-

3 3 0 Nitrate reduction rate X(NO3 )

100

80

Influence of the ratio x = {[FeII]I / [Fetotal]}. Hydroxycarbonate GR(CO32-) with formula FeII4 FeIII2 (OH)12 CO3 • 3H2O at x = 1/3 and definitively called fougèrite is an interesting starting material to study the effect of ratio x on nitrate reduction. As soon as 2006, it was shown that the controlled oxidation by H2O2 of GR(CO32-) led to the formation of GR-related materials having a strong structural analogy with the initial GR(CO32-) [36, 54].

O

Time (Days)

8

Time (Days)

Fig. (13). Nitrate reduction rate in contact hydroxycarbonate green rust as a function of phosphate concentration at initial pH of 7.5, 9 and 10.5.

Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Use of Ferrihydrite-Coated Pozzolana and Biogenic Green Rust

111

GRabio (b) (a)

rib is t

rD

Pe

rs on

al

ut

U

io n

se

O

nl y

(c)

fo

Fig. (15). Fitting of the Mössbauer spectra of GRabio (a) and GRbio (b) with a pseudo-Voigt profile analysis. (c) Structural model of a mixture of fougèrite and trébeurdenite. The ferrous species (doublets D1 and D2 in Mössbauer spectroscopy) and ferric species (doublets D3 and D4) in octahedral coordination are represented.

droxysulfate and hydroxychloride GR synthesized at x values of 0.33 and 0.25 [16].

Fraction of nitrate removed X(NO3-) of 10 %, 9 %, 1.5 % and 0 % are measured for GR characterized by molar fraction of 0.33, 0.41, 0.5 and 0.65, respectively after a reaction time with nitrate of 7 days at a pH of 7.5. As expected, the fraction of nitrate removed decreases with increasing ferric molar fraction, i.e. a decrease of the FeII content of GR. At values of x higher than x = 0.5 the reduction of nitrate is completely inhibited. This result is in good agreement with previously published data concerning the reactivity of hy-

R(NH4+)t = {[NH4+]t - [NH4+]0} / [NO3-]0

N ot

erence (Fig. 14c & d); however, there exists a slight discrepancy in the doublets positions because the temperature of measurement is much lower (10 K vs. 78 K) and the value of x is slightly lower (about 0.45); anyhow all spectra of GRbio we obtained by bioreduction had a value of x in a range from 0.45 to 0.55. According to the structural model, which is validated by the International Mineralogical Association (IMA), GRbio samples are in fact mixtures of fougèrite and trébeurdenite according to the lever rule with x values close to 0.5 corresponding to half-half. The topotaxical reaction is illustrated in Fig. 15c. It does not need any migration of Fe cations or of CO32- anions, whatsoever.

b) Selectivity of the Reaction Towards Ammonium Definition of the ammonium selectivity S(NH4+)t. The measurement of the selectivity of the reaction toward ammonium is crucial if one considers the potential application, since its formation should be avoided. Ammonium is biologically available and a less mobile form of inorganic-N compared to nitrate. The ammonium production ratio R(NH4+) is defined in a similar way to the fraction of nitrate removed (equation 3) by the following equation: (6)

The ammonium selectivity is defined as the ratio S(NH4+) = R(NH4+) / X(NO3-) and corresponds to the ratio between the concentration of ammonium produced to the concentration of nitrate removed: S(NH4+)t = {[NH4+]t - [NH4+]0} / {[NO3-]t - [NO3-]0} -

(7)

The goal is to maximize the X(NO3 ) values close to 1 and to minimize the values of R(NH4+) and of S(NH4+) close

112 Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Ruby et al.

to zero. Ammonium is measured by the colorimetric salicylate/nitroprusside procedure ([56] or SI of [20]) after dilution of the sample in pure water and filtering (0.22 µm, Acrodisc Supor®, Gelman, Pall Corporation), or by ion chromatography DIONEX ISC 3000 after 0.22 μm filtration with detection limits of 0.1 mg/L. 60

40

PO4 / Fe = 0 % PO4 / Fe = 0.6 %

30

PO4 / Fe = 2.4 %

20

10

0 2

4

6

8

O

0

60

rib

al

ut

(b) pH = 9 40

U

+

50

rs on

PO4 / Fe = 0 %

30

is t

PO4 / Fe = 0.5 %

0 0

2

4

fo

Pe

10

rD

PO4 / Fe = 2.4 %

20

io n

se

Time (Days)

Ammonium production rate R(NH4 )

Ammonium selectivity of 30 independents experiments. Statistical experimental designs including 30 experiments performed in duplicate were performed in order to find experimental conditions favoring the reduction of nitrate by GR*c (x = 1/3) and limiting the formation of ammonium [20]. The influence of 4 parameters, i.e. the pH and the nitrate, ferrous and phosphate concentrations was studied. The lower and upper limits of concentrations were 50 - 200 mg L1 (0.8 - 3.2 mM), 0 - 95 mg L-1 (0 - 1 mM) and 0 - 3000 mg L-1 (42 - 167 mM) for nitrate, phosphate and ferrous species, respectively. The pH values were set initially at constant values of 7.5, 9 and 10.5. The experimental condition corresponding to each run is given in Table 5. Independently from the statistical analyses of the results that will be not discussed here (see section 3.4 of [20]), the experiments may be classified in 4 groups if the ammonium selectivity S(NH4+)7days is plotted as a function of the nitrate reduction rate (Fig. 17):

nl y

+

Ammonium production rate R(NH4 )

(a) pH = 7.5 50

Influence of the phosphate concentration and of pH. As already observed for the fraction of nitrate removed (Fig. 13), the ammonium production ratio R(NH4+) is strongly dependent on the pH and the phosphate concentration (Fig. 16). Independently from the values of the pH, the values of R(NH4+) decrease significantly when the phosphate concentration increases. Either, this diminution is directly linked to the diminution of nitrate reactivity already observed for increasing phosphate concentration, or phosphate anions play a role on the ammonium selectivity. The first case is observed if one compares Fig. 13a to Fig. 16a for [PO4]: [FeII] ratio of 0 and 0.6% where the diminution of the R(NH4+)7days value is proportional to the diminution of the X(NO3-)7days values. The second case is true if one compares Fig. 13c to Fig. 16c for [PO4]: [FeII] ratio of 0 and 1.2%, here the fraction of nitrate removed X(NO3-)7days value is diminished by a factor of  2 while the R(NH4+)7days value decreases by a factor of  8.

6



Group A corresponds to experiments where the reduction of nitrate by GR is insignificant, i.e. X(NO3-)7days = R(NH4+)7days = 0. According to equation (5) values of S(NH4+) are undetermined,



Group B corresponds to experiments where the reduction of nitrate by GR is relatively limited, i.e. X(NO3-)7days < 20%. The selectivity S(NH4+)7days values are relatively high and range between  25 and  100%. The experiments of group B may be divided in two subgroups (B1 and B2) for which the ammonium selectivity is situated above and below 50%, respectively,



Group C corresponds to experiments having a moderated nitrate reduction rate, i.e. 20% < X(NO3-)7days < 40%, and relatively low ammonium selectivity, i.e. 7% < S(NH4+)7days < 26%,



Group D corresponds to experiments having a high nitrate reduction rate, i.e. X(NO3-)7days > 70%, and high ammonium selectivity (S(NH4+)7days > 50%).

8

N ot

Time (Days) 60 +

Ammonium production rate R(NH4 )

(c) pH = 10.5 PO4 / Fe = 0 %

50

PO4 / Fe = 1.2 % PO4 / Fe = 2.4 %

40

30

20

10

0 0

2

4

6

8

Time (Days)

Fig. (16). Ammonium production rate during the reduction of nitrate by hydroxycarbonate green rust as a function of phosphate concentration.

As shown in Fig. 17, except for the experiments of groups D and B1, several experiments lead to ammonium selectivity much lower than 100% meaning that nitrate

Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Use of Ferrihydrite-Coated Pozzolana and Biogenic Green Rust

Table 5.

Parameters of the statistical experimental designs (adapted from Etique et al. [20]) -

[NO3 ]0

[FeII]0

-3

-3

3-

[PO4 ]0 -3

pHi

X(NO3-)

R(NH4+)

S(NH4+)

(%)

(%)

(%)

Group

(10 M)

(10 M)

1

0.8

42

0

7.5

0

0

u.v.a

A

2

3.2

42

0

7.5

0

0

u.v.

A

3

0.8

167

0

7.5

40.5

8.1

20

C

4

3.2

167

0

7.5

18.5

15.6

84

B

5

0.8

42

1.0

7.5

0

0

u.v.

A

6

3.2

42

1.0

7.5

0

0

u.v.

A

7

0.8

167

1.0

7.5

10.4

13

100

B

8

3.23

167

1.0

7.5

9.6

7.9

82

B

9

0.8

42

0

10.5

72.8

55.6

76

D

10

3.2

42

0

10.5

74.6

62.7

84

D

11

0.8

167

0

10.5

92.1

86.8

94

D

12

3.2

167

0

10.5

97.1

86.9

90

D

13

0.8

42

1.0

10.5

0

0

u.v.

A

14

3.2

42

10.5

0

0

u.v.

A

15

0.8

167

1.0

10.5

92.1

76.3

83

D

16

3.2

167

1.0

10.5

79.8

89.6

100

D

17

2

104

0.5

9.0

13.9

4.9

35

B

18

2

42

0

9.0

6.4

4.4

69

B

19

2

9.0

7.6

6.5

85

B

20

2

21

2

167

22

2

23

io n

ut

rib

is t

rD

fo

0 1.0

9.0

0

0

u.v.

A

1.0

9.0

0

0

u.v.

A

42

0.5

7.5

0

0

u.v.

A

2

167

0.5

7.5

12.6

4.8

38

B

24

2

42

0.5

10.5

32.6

8.5

26

C

25

2

167

0.5

10.5

91.7

84

92

D

26

2

104

0

7.5

8

3.5

44

B

27

2

104

1.0

7.5

22.5

1.7

7

C

28

2

104

0

10.5

91.2

50.2

55

D

29

2

104

1.0

10.5

0

0

u.v.

A

30

2

104

0.5

9.0

6.9

1.9

27

B

N ot

42

se

U al

1.0

rs on Pe 167

nl y

(10 M)

O

Runs

113

species are not fully reduced into ammonium species. Therefore, other chemical reactions than reaction (3) occur and nitrate may be reduced in other N containing species such as NO2-, N2O, NO or N2. NO2- is not detected by ion

chromatography and a small bubbling is observed in most of the suspensions. Therefore, the formation of N-gaseous species is strongly suspected in our experimental conditions.

Ruby et al.

0.33), the rate of nitrite reduction reaches a value of 39 % after 20 minutes of reaction, and stabilizes at 43% after a reaction time of one hour (Fig. 18b). A maximum rate of 46% is observed after two days. The maximum ammonium production rate observed in this experiment is ~ 11% (Fig. 18b). In contrast, when GRbio (x = 0.45) is used as a reducer, the rate of nitrite reduction increases progressively with increasing the reaction time and reaches a value of 44% after two days of interaction. No ammonium production is observed in that case (Fig. 18c).

100

+

B1

B2

C

100

100

(a) abio-GR(Cl) 4

+

0

Ammonium production rate R(NH )

50

D

2

-

2 rate 0 Nitrite2reduction X(NO )

A 60

80

100

-

40

0.5

1

1.5

2

io n

se

Time (days) 100

(b) abio-GR(CO3) 80

80

rib

ut

2

[NO2 ]reduction / [NO[NO ] 2-] / X(NO [NO2-]-0) Nitrite 2 0 rate

60

40

40

20

20

-

60

-

is t

rD

We have therefore decided to compare the reactivity toward nitrite ions of such a GRbio to that of abiotically synthesized GR (GRabio) such as hydroxycarbonate GR and hydrochloride GR. Fig. 18a indicates that nitrite ions are rapidly removed with a reduction rate X(NO2-) of 100% after one hour of reaction with hydroxychloride GR (x = 0.25) at neutral pH. Kinetics of ammonium production R(NH4+) is also very fast with a maximum value of 29% reached after two hours of reaction. When nitrite ions interact with GRabio (x =

0

0

0

0

fo

0

0.5

1

1.5

2

Time (days)

100

100

(c) bio-GR(CO3) 4

+ 2

[NO2reduction ] / [NO2 ]rate Nitrite X(NO ) 0

80

60

60

40

40

20

20

-

80

-

N ot

Because of its low Fe content (i.e. x = 0.45), the reactivity of the biogenic hydroxycarbonate green rust (referred to as GRbio) towards nitrate under circumneutral pH conditions is very slow. However, after 4 months of reaction between GRbio and nitrate under anoxic conditions, the color of the suspension changed finally from green to brown (Fig. SI-1), suggesting a full oxidation of green rust to ferric oxyhydroxides. Results reveal that 5.84 mM of nitrate ions are fully converted to nitrogen gaseous species, without ammonium production after a reaction time of 4 months (Table SI-1).

20

100

U

al

rs on

Pe

40

0

3.2.2. Reactivity Towards Nitrite Ions of Biotically Synthesized Hydroxycarbonate Green Rust Compared to that of Abiotically Synthesized Hydroxycarbonate and Hydroxychloride Green Rusts II

60

20

O

The most interesting experiments are those of group C that give some indication about the condition that favors low ammonium selectivity. S(NH4+) may be lowered in the presence of a small quantity of phosphate, typically [PO4]: [FeII]GR  1 % or even in the absence of phosphate at a pH of 7.5 (run 3, Table 5). From the data presented in Table 5, it is difficult to find a direct correlation between the variation of an independent parameter (pH, [NO3-], [PO43-] or [FeII]GR) and the variation of S(NH4+). Nevertheless, Fig. 17 shows very clearly that low ammonium selectivity are only observed when the kinetics of nitrate reduction rate is relatively slow (X(NO3-]7days < 40 %). This result will have important implication in terms of practical application and will be discussed later in section IV. Finally, note that an ammonium selectivity S(NH4+) of 0% is never observed. Therefore, it seems rather difficult to completely avoid the formation of residual ammonium by this way.

60

nl y

Fig. (17). Ammonium selectivity as a function of the nitrate reduction rate for different experiments of the full factorial design (see text section III-2-1-b and Table 5).

-

-

[NO ] / [NO ]

Nitrate reduction rate X(NO3 ) %

+

40

4

20

80

Ammonium production rate R(NH )

0

80

Ammonium production rate R(NH )

Ammonium selectivity S(NH4 ) %

114 Current Inorganic Chemistry, 2016, Vol. 6, No. 2

0

0

0

0.5

1

1.5

2

Time (days)

Fig. (18). Nitrite reduction rate X(NO2-) and ammonium production ratio R(NH4+) obtained during interaction of abiotic hydroxychloride green rust (referred to as ‘GR(Cl)’), abiotic hydroxycarbonate green rust (referred to as ‘abio-GR(CO3)’) and biogenic hydroxycarbonate green rust (referred to as ‘bio-GR(CO3)’) with NO2ions under circumneutral conditions.

Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Use of Ferrihydrite-Coated Pozzolana and Biogenic Green Rust

As shown in Fig. 19, all experiments conducted with hydroxychloride GRabio and hydroxycarbonate GRabio lead to a similar ammonium selectivity of 25 ± 6% and 26 ± 2%, respectively. The most interesting case is obtained with hydroxycarbonate GRbio (x = 0.45) for which an ammonium selectivity S(NH4+) of 0% is observed. These findings suggest that nitrate ions should be reduced to nitrite ions prior to their reaction with GRbio.

80

60

40

nl y

abio-GR(CO 3)

Abio-GR(Cl) GR(Cl) 20

bio-GR(CO 3) 0 80

100

-

U

Pe

4. ENVIRONMENTAL IMPLICATIONS

Table 6.

Phosphate, nitrate and ammonium concentrations measured in the outflow of the Rhyzostep® reed bed filter of Burdignes (Loire, France) between April and August 2009.

is t

Date of sampling

rD

rs on

al

Fig. (19). Ammonium selectivity S(NH4+) measured when nitrite ions react with various GRs under circumneutral conditions: hydroxychloride GR (x = 0.25), abiogenic hydroxycarbonate GR (x = 0.33) and biogenic hydroxycarbonate GR (x = 0.45). Note that GR(Cl) = abiotic hydroxychloride green rust, abio-GR(CO3) = abiotic hydroxycarbonate green rust, and bio-GR(CO3) = biogenic hydroxycarbonate green rust.

io n

60

ut

40

rib

20

Nitrite[NO reduction X(NO ) % ] / [NOrate ] (%) 2 2 2 0

se

0

sions of the potential filtration reactor. Let us consider a given quantity of waste water containing a phosphate pollution of 1 population equivalent (PE), each PE producing approximately 3 g of phosphate per day. Moreover, if one considers that the passive treatment should last at least 4 years before the material would be replaced, the total quantity of phosphate that should be adsorbed represents a total mass of 4.38 kg PE-1. The column experiment performed with municipal waste water (Fig. 11) is certainly the most appropriate to estimate the quantity of filtration material that has to be used to catch this amount of phosphate. The breakthrough occurs at a value of  50 V/Vp corresponding to a phosphate adsorption capacity at the breakthrough of qB  40 mg g-1. This experiment is performed at a flow rate F = 0.5 mL min-1 and if equation (1) is considered, a higher adsorption capacity qB  57 mg g-1 should be obtained at a lower flow rate F of  0.03 mL min-1, i.e. a residence time of  24 hours. The value of qB can also be normalized to the total quantity of filtration material (FM) and it gives q’B  5.2 mg PO4 per g of FM (or  1.7 mg P-PO4 per g of FM). Therefore a mass of  840 kg PE-1 of FM would be necessary and the corresponding volume of the reactor would be  0.84 m3 PE-1 since the apparent density of FH coated pozzolana is very close to 1 (Table 3). Considering a horizontal filter of  0.5 m depth, this volume corresponds to a surface of  2 m2 PE-1. This surface corresponds clearly to an extensive water treatment that could only be performed for treating waste water in rural area where sufficient ground superficies should be available.

O

4

+

Ammonium selectivity S(NH ) %

100

115

fo

4.2. Passive Water Treatment Processes for Reed Beds Filters

N ot

Waste water coming out from vertical reed bed filters contains both phosphate and nitrate ions that should be eliminated from water by an adequate tertiary treatment. As an example, the relative concentrations of phosphate and nitrate measured between January and August 2009 in the outflow of the Rhyzostep® reed bed filters of Burdignes (Loire, France) were situated in a range of 5 - 8.5 mg P-PO4 L-1 and 34 - 56 mg NNO3 L-1, respectively (Fig. 20 and Table 6). The results presented in section III-2-1 show clearly that for these relative concentrations ranges, the reactivity of GR towards nitrate could be significantly lower due to the presence of coexisting phosphate anions. Therefore, it seems more cautious to consider tertiary treatments where phosphate and nitrate anions would be separately treated; of course the dephosphatation step has to precede the denitrification step. Each step will be therefore considered in the following sections. Dephosphatation process. In comparison to the literature data (Table 2), our study shows clearly that ferrihydrite is the best sorbent both in batch and hydrodynamic conditions. The preconized filtration material is therefore a ferrihydrite coated pozzolana [10] and it is possible to extrapolate the dimen-

P-PO4

N-NO3

N-NH4

-1

-1

mg L-1

mg L

mg L

04-15-2009

6,6

56

10

04-28-2009

5.2

34

2.5

05-26-2009

5.4

55

1

06-23-2009

6.0

44

2.9

07-20-2009

8.5

53

1

08-26-2009

6.1

50

2.1

Fig. (20). Plant of vertical reed bed filters (Rhyzostep©, Burdignes, Loire, France).

116 Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Ruby et al.

Denitrification process and global tertiary treatment. Let us consider again the particular case of a potential tertiary denitrification process performed at the up-front of a vertical reed bed filter (Fig. 21). Such water treatments plants are often equipped with several air inlet stacks allowing a better decomposition of organic matter. Because of these oxic conditions the concentrations of nitrogen total Kjeldhal (N-NTK) and ammonium (N-NH4+) species present in waste water decrease generally significantly in the filter (Table 7). However, it leads also to a global increase of the nitrate concentrations (N-NO3-) that largely exceed the ammonium concentrations (N-NH4+) measured in the outflow of the reed bed filter (Table 6). This excess of nitrate species and the remaining ammonium ions should be eliminated by the tertiary treatment. Therefore, a potential tertiary treatment where 3 anoxic horizontal filters are in series could be imagined (Fig. 21). The essential role of the first horizontal filter consisting of ferrihydrite coated pozzolana (Fh-Pz) is to eliminate the phosphate contamination. Because of the long residence time of the water ( 24 hours) in this first reactor, anoxic condition may take place and

favors the biological reduction of nitrate species into either Ngaseous species or nitrite ions NO2-. Since the water coming out from the vertical reed bed filter contains relatively low amount of organic matter (Table 7), the biological denitrification cannot be fully completed in the first reactor. The second horizontal filter may consist of green rust coated pozzolana (GR-Pz), green rust could be synthesized either by a biological pathway (GRbio) or by a chemical route (abio-GR). As demonstrated previously, the advantage to use GRbio would be that it fully transforms nitrite species into N-gaseous species. Moreover, the oxidation products of GRbio in contact with nitrite species are ferric oxyhydroxides that can easily be reduce by iron-reducing bacteria in order to regenerate GRbio. GRabio presents the advantage to exhibit a much higher reactivity for reducing nitrate species than GRbio, however the NO3- reduction leads to the partial formation of NH4+ ions. Therefore, the use of a third horizontal filters consisting of pozzolana coated with an NH4+ sorbent, e.g. clays such as smectite, could be necessary to eliminate residual ammonium species before water is finally released in the natural environment. HORIZONTAL REED BED FILTERS

nl y

VERTICAL REED BED FILTERS

Anoxic condition

O

Oxic condition

PO4-NO3-NH4 treatments

ut rib is t

PO4

rD

Pe

rs on

al

U

Screening

io n

se

Organic matter treatment

Fh-Pz

NO3

NH4

GR-Pz

Clay-Pz

natural aquifer

N ot

Table 7.

fo

Fig. (21). Passive water treatments proposed for rural area. Organic matter is eliminated inside vertical reed bed filters. Phosphate, nitrate and residual ammonium are removed by a series of horizontal filters. Physico-chemical characteristics of the waste water of a Rhyzostep® reed bed filter (Burdignes, Loire, France) measured during a 8 months period from august 2006 to march 2007. BOD5: five-day biochemical oxygen demand; COD: chemical oxygen demand; TSS: total suspended solids; NTK: nitrogen total Kjeldhal. Inflow water quality Date of sampling

BOD5 mg L

-1

COD mg L

-1

Outflow water quality NTK

TSS mg L-1

mg L

-1

TSS

N-NH4 mg L-1

BOD5 mg L-1

COD mg L-1

mg L

-1

NTK mg L

-1

N-NH4 mg L-1

08-23-2006

177

580

204

131.6

120.4

4

59

10

9.8

4.2

10-19-2006

406

937

198

158.2

135.8

10

95

16

9.1

8.4

11-30-2006

401

1065

450

152.6

119

16

68

18

11.9

4.9

12-14-2006

245

705

148

124.6

93.8

14

85

56

25.2

14.7

01-10-2007

277

622

154

197.4

121.8

20

54

18

7.7

4.9

01-25-2007

214

598

137

113.4

98

9

63

16

7

7

02-08-2007

90

421

99

68.6

52.3

6

57

11

6.3

5.6

03-08-2007

244

706

230

172.4

130.6

3

32

3

21

3.5

Current Inorganic Chemistry, 2016, Vol. 6, No. 2

Use of Ferrihydrite-Coated Pozzolana and Biogenic Green Rust

117

4.2. FUTURE RESEARCHES

SUPPLEMENTARY MATERIAL

The eutrophication of aquifers is a serious environmental problem and therefore the research of new materials suitable to perform passive tertiary treatments of water dephosphatation and denitrification is a real challenge. This study demonstrated that iron oxides coated pozzolanas are very promising filter materials for such a purpose. Ferrihydrite coated Pz adsorbs efficiently orthophosphate in hydrodynamic condition and a surface of  2 m2 per PE would be required for performing the treatment inside a horizontal filter. Increasing the phosphate adsorption capacity by modifying the surface properties of the filter material could be the subject of future works. For instance, one may study with more details the synergetic effect induced by calcium on the phosphate adsorption, either by increasing artificially the amount of soluble calcium present in waste water or by doping the Fh coating with Ca. Ca doped Fh or Ca-Fe layered double hydroxide could be other promising materials to sequester a higher amount of phosphate. Moreover, finding chemical treatments that lead to the regeneration of the reactive surface sites of Fh by inducing phosphate desorption could also be studied in order to increase the life time of the filter material. One of the difficulties is to avoid the destabilization of the coating during such recycling treatment since Fh can easily be dissolved in acidic condition or it may also recrystallize into other ferric oxyhydroxides, e.g. goethite FeOOH, in alkaline condition.

Supplementary material is available on the publisher’s web site along with the published article. REFERENCES [1] [2]

[3]

[4]

[9]

ut

[8]

is t

[10]

[11]

rD

fo

N ot

Pe

rs on

al

In this study, the reactivity of green rust toward nitrate ions was limited to homogeneous “batch” reactor. Further works concerning the synthesis of GR coated pozzolana by either chemical or biological pathways are necessary. Preliminary experiments demonstrated that it is possible to induce the mineralogical transformation of Fh-coated Pz into GR coated Pz by the addition of controlled amount of hydroxylated FeII species [57]. This study showed that the kinetics of reaction of GR towards nitrite ions was much faster than towards nitrate. Interestingly, the nitrite species were exclusively transformed into N-gaseous species when GR was synthesized by a biological pathway. Therefore, it would be very interesting to find reacting conditions that favors the preliminary reduction of nitrate species into nitrite species in the denitrification process involving GR.

[7]

rib

U

se

O

[6]

io n

nl y

[5]

Carpenter, S.R.; Caraco, N.F.; Correll, D.L.; Howarth, R.W.; Sharpley, A.N.; Smith, V.H. Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecol. Appl., 1998, 8(3), 559-568. Tanner, C.C.; Sukias, J.P.S.; Headley, T.R.; Yates, C.R.; Stott, R. Constructed wetlands and denitrifying bioreactors for on-site and decentralised wastewater treatment: Comparison of five alternative configurations. Ecol. Eng., 2012, 42(1), 112-123. Torrens, A.; Molle, P.; Boutin, C.; Salgot, M. Impact of design and operation variables on the performance of vertical-flow constructed wetlands and intermittent sand filters treating pond effluent. Water Res., 2009, 43(7), 1851-1858. Savina, I.N.; English, C.J.; Whitby, R.L.D.; Zheng, Y.; Leistner, A.; Mikhalovsky, S.V.; Cundy, A.B. High efficiency removal of dissolved As(III) using iron nanoparticle-embedded macroporous polymer composites. J. Hazard. Mater., 2011,192(3), 1002-1008. Mandel, K.; Drenkova-Tuhtan, A.; Hutter, F.; Gellermann, C.; Steinmetz, H.; Sextl, G. Layered double hydroxide ion exchangers on superparamagnetic microparticles for recovery of phosphate from waste water. J. Mater. Chem. A, 2013, 1(5), 1840-1848. Xie, J.; Wang, Z.; Wu, D.; Kong, H. Synthesis and properties of zeolite/hydrated iron oxide composite from coal fly ash as efficient adsorbent to simultaneously retain cationic and anionic pollutants from water. Fuel, 2014, 116(1), 71-76. Driehaus, W. Method for producing a sorption material that contains iron. Patent WO2002047811 A1, June 20, 2002. Arias, M.; Da Silva-Carballal, J.; García-Río, L.; Mejuto, J.; Núñez, A.; Retention of phosphorus by iron and aluminum-oxidescoated quartz particles. J. Colloid Interf. Sci., 2006, 295(1), 65-70. Shi, Z.-l.; Liu, F.-m.; Yao, S.-h. Adsorptive removal of phosphate from aqueous solutions using activated carbon loaded with Fe(III) oxide. New Carbon Mater., 2011, 26(4), 299-306. Naille, S.; Barthélémy, K.; Mallet, M.; Ruby, C.; Procédé de fabrication d'un matériau de filtration. Patent WO2014032934 A1, August 8, 2013. Barca, C.; Gérentea, C.; Meyera, D.; Chazarenc, F.; Andrès;Y. Phosphate removal from synthetic and real wastewater using steel slags produced in Europe. Water Res., 2012, 46(7), 2376-2384. Molle, P.; Liénard, A.; Grasmick, A.; Iwema, A.; Kabbabi; A. Apatite as an interesting seed to remove phosphorus from wastewater in constructed wetlands. Water Sci. Technol., 2005, 51(9), 193-203. Bellier, N.; Chazarenc, F.; Comeau, Y. Phosphorus removal from wastewater by mineral apatite. Water Res., 2006, 40(15), 29652971. Lyngsie, G.; Borggaard, O.K.; Hansen H.C.B. A three-step test of phosphate sorption efficiency of potential agricultural drainage filter materials. Water Res., 2014, 51, 256-265. Hansen, H.C.B.; Koch, C.B.; NanckeKrogh, H.; Borggaard, O.K.; Sorensen, J. Abiotic nitrate reduction to ammonium: Key role of green rust. Environ. Sci. Technol., 1996, 30(6), 2053-2056. Hansen, H.C.B.; Guldberg, S.; Erbs, M.; Bender Koch, C. Kinetics of nitrate reduction by green rusts - effects of interlayer anion and Fe(II):Fe(III) ratio. Appl. Clay Sci., 2001, 18(1-2), 81-91. Choi, J.; Batchelor, B. Nitrate reduction by fluoride green rust modified with copper. Chemosphere, 2008, 70(6), 1108-1116. Mills, S.J.; Christy, A.G.; Génin, J.-M.R.; Kameda, T.; Colombo, F. Nomenclature of the hydrotalcite supergroup: natural layered double hydroxides. Mineral. Mag., 2012, 76(5), 1289-1336. Génin, J.-M. R;.Mills,S. J.; Christy, A. G., Guérin, O.;Herbillon, A. J.; Kuzmann, E.; Ona-Nguema, G.; Ruby, C., Upadhyay, C. Mossbauerite, Fe6O4(OH)8[CO3]·3H2O, the fully oxidized ‘green rust’ mineral from Mont Saint-Michel Bay, France. Mineral. Mag., 2014, 78(2), 447-465. Etique, M.; Zegeye, A.; Grégoire, B.; Carteret, C., Ruby C. Nitrate reduction by mixed iron(II-III) hydroxycarbonate green rust in the presence of phosphate anions: The key parameters influencing the ammonium selectivity. Water Res., 2014, 62, 29-39.

CONFLICT OF INTEREST

[12]

[13] [14] [15]

The authors confirm that this article content has no conflicts of interest.

[16]

ACKNOWLEDGMENTS

[17]

We would like to thank the following organizations for their support: The Agence Nationale de la Recherche (ANR program, ECOTECH2009 - N°0994C0103), the Region Lorraine, the Institut Carnot Energie et Environnement de Lorraine (ICEEL) and the Environmental and Molecular Sciences Laboratory, a US-DOE Office of Science User Facility sponsored by the Office of Biological and Environmental Research and located at Pacific Northwest National Laboratory. K. Barthélémy expresses his sincere gratitude to SAUR for the Ph.D. position grant.

[18] [19]

[20 ]

118 Current Inorganic Chemistry, 2016, Vol. 6, No. 2

[31] [33]

[34]

[35]

[36]

[37]

[38] [39]

[44] [45]

[48]

[49]

[50]

ut

O

[47]

io n

nl y

[46]

rib

[30]

[43]

the goethite–water interface. J. Colloid Interf. Sci., 2005, 285(2), 476-486. Rahnemaie, R.; Hiemstra, T.; van Riemsdijk, W.H. Geometry, charge distribution and surface speciation of phosphate on goethite. Langmuir 2007, 23(7), 3680-3689. Kim, J.; Li, W.; Philips B. L.; Grey, C. P. Phosphate adsorption on the iron oxyhydroxides goethite (α-FeOOH), akaganeite (βFeOOH), and lepidocrocite (γ-FeOOH): a 31P NMR Study. Energy Environ. Sci., 2011, 4, 4298-4305. Peleka, E.N.; Deliyanni, E.A. Adsorptive removal of phosphates from aqueous solutions. Desalination, 2009, 245(1-3), 357-371. Chitrakar, R.; Tezuka, S.; Sonoda, A.; Sakane, K.; Ooi, K.; Hirotsu, T. Adsorption of phosphate from seawater on calcined MgMnlayered double hydroxides. J. Colloid Interf. Sci., 2005, 290(1), 4551. Kuzawa, K.; Jung, Y.-J.; Kiso, Y.; Yamada, T.; Nagai, M.; Lee, T.G. Phosphate removal and recovery with a synthetic hydrotalcite as an adsorbent. Chemosphere, 2006, 62(1), 45-52. Xu, Y.; Dai, Y.; Zhou, J.; Xu, Z.P.; Qian, G.; Lu, G.Q.M. Removal efficiency of arsenate and phosphate from aqueous solution using layered double hydroxide materials: intercalation vs. precipitation. J. Mater. Chem., 2010, 20, 4684-4691. Karaca, S.; Gürses, A.; Ejder, M.; Açıkyıldız, M. Adsorptive removal of phosphate from aqueous solutions using raw and calcinated dolomite. J. Hazardous Mater., 2006, 128(2-3), 273-279. Woo, M.A.; Kim, T.W.; Paek, M.-J.; Ha, H.-W.; Choy, J.-H.; Hwangn S.J. Phosphate-intercalated Ca-Fe-layered double hydroxides: Crystal structure, bonding character, and release kinetics of phosphate. J. Solid State Chem., 2011, 184(1), 171-176. Gillman, G.P.; Noble, M.A.; Raven, M.D. Anion substitution of nitrate-saturated layered double hydroxide of Mg and Al. Appl. Clay Sci., 2008, 38(3-4) 179-186. Ashekuzzaman, S.M.; Jiang, J.-Q. Study on the sorptiondesorption-regeneration performance of Ca-, Mg and CaMg-based layered double hydroxides for removing phosphate from water, Chem. Eng. J., 2014, 246, 97-105. Ruby, C.; Barthélémy, K.; Hanna, K.; Mallet, M.; Naille, S. Ferrihydrite-coated pozzolana as an efficient material for water dephosphatation, Mater. Design, 2015, 86, 168-177. Skovbjerg, L.L.; Stipp, S.L.S.; Utsunomiya, S.; Ewing, R.C. The mechanisms of reduction of hexavalent chromium by green rust sodium sulphate: Formation of Cr-goethite, Geochim. Cosmochim. Ac., 2006, 70(14), 3582-3592. Guilbaud, R.; White, M.L.; Poulton, S.W. Surface charge and growth of sulphate and carbonate green rust in aqueous media. Geochim. Cosmochim. Ac., 2013, 108, 141-153. Remy, P.-Ph.; Etique, M.; Hazotte, A.A.; Sergent, A.-S.; Estrade, N.; Cloquet, C.; Hanna, K.; Jorand, F.P.A. Pseudo-first-order reaction of chemically and biologically formed green rusts with HgII and C15H15N3O2: Effects of pH and stabilizing agents (phosphate, silicate, polyacrylic acid, and bacterial cells) Water Res. 2015, 70, 266-278. Ruby, C.; Abdelmoula, M.; Naille, S.; Renard, A.; Khare, V.; OnaNguema, G.; Morin, G.; Génin J.-M. R. Oxidation modes and thermodynamics of FeII-III oxyhydroxycarbonate green rust: Dissolution–precipitation versus in situ deprotonation. Geochim. Cosmochim. Ac., 2010, 74(3) 953-966. Mullet, M.; Guillemin, Y.; Ruby, C. Oxidation and deprotonation of synthetic FeII-FeIII (oxy)hydroxycarbonate Green Rust: An X-ray photoelectron. J. Solid State Chem., 2008, 181(1), 81-89. Baethgen, W.E.; Alley, M.M. A manual colorimetric procedure for measuring ammonium nitrogen in soil and plant Kjeldahl digests. Commun. Soil Sci. Plan.,1989, 20(9-10), 961-969. Boumaiza, H.; Naille, S.; Renard, A.; Grégoire, B., Mallet, M., Ruby, C. Chemical transformation of ferrihydrite coating into green rust followed by Raman and X-ray photoelectron spectroscopies, Desalin. Water Treat., 2015, 53(4), 1031-1036.

[51]

is t

[29]

[42]

se

[28]

[41]

rD

[27]

[40]

U

[26]

fo

[25]

al

[24]

N ot

[23]

rs on

[22]

Ona-Nguema, G.; Abdelmoula, M.; Jorand F.; Benali, O.; Géhin, A.; Block, J.-C.; Génin J.-M. R.; Iron(II,III) Hydroxycarbonate green rust formation and stabilization from lepidocrocite bioreduction. Environ. Sci. Technol., 2002, 36(1), 16-20. Jorand, F.; Zegeye, A.; Landry, F.; Ruby; C. Reduction of ferric green rust by Shewanella putrefaciens. Lett. Appl. Microbiol., 2007, 45(5), 515-521. Etique, M.; Jorand, F.P.A.; Zegeye, A.; Grégoire, B; Despas, C; Ruby, C. Abiotic process for Fe(II) oxidation and green rust mineralization driven by a heterotrophic nitrate reducing bacteria (Klebsiella mobilis). Environ. Sci. Technol., 2014, 48, 3742-3751. Guerbois, D.; Ona-Nguema, G.; Morin, G., Abdelmoula, M.; Laverman, A.M. Mouchel, J.-M.; Barthelemy, K.; Maillot, F.; Brest, J. Nitrite Reduction by Biogenic Hydroxycarbonate Green Rusts: Evidence for Hydroxy-nitrite Green Rust Formation as an Intermediate Reaction Product. Environ. Sci. Technol., 2014, 48(8), 4505-4514. Mallet, M., Barthélémy, K., Ruby, C., Renard, A., Naille, S. Investigation of phosphate adsorption onto ferrihydrite by X-ray Photoelectron Spectroscopy. J. Colloid Interf. Sci., 2013, 407, 95-101. Usman, M.; Hanna, K.; Abdelmoula, M.; Zegeye, A.; Faure, P.; Ruby, C. Formation of green rust via mineralogical transformation of ferric oxides (ferrihydrite, goethite and hematite). Appl. Clay Sci., 2012, 64, 38-43. Jolivet, J.-P.; Tronc, E.; Chanéac, C. Iron oxides: From molecular clusters to solid. A nice example of chemical versatility. C. R. Geosci., 2006, 338(6-7), 488-497. Ona-Nguema, G.; Abdelmoula, M.; Jorand, F.; Benali, O.; Géhin, A.; Block, J.-C.; Génin, J.-M. R. Microbial reduction of lepidocrocite γ-FeOOH by shewanella putrefaciens; the formation of green rust. Hyperfine Interact., 2002, 139(1), 231-237. Bocher, F.; Gehin, A.; Ruby, C.; Ghanbaja, J.; Abdelmoula, M.; Genin, J.M.R. Coprecipitation of Fe(II-III) hydroxycarbonate green rust stabilised by phosphate adsorption. Solid State Sci., 2004, 6(1), 117-124. Barthélémy, K.; Naille, S.; Despas, C.; Ruby, C.; Mallet, M. Carbonated ferric green rust as a new material for efficient phosphate removal. J. Colloid Interf. Sci., 2012, 384(1), 121-127. Refait, Ph.; Benali, O.; Abdelmoula, M.; Génin; J.-M.R. Formation of ‘ferric green rust’ and/or ferrihydrite by fast oxidation of iron(II-III) hydroxychloride green rust. Corros. Sci., 2003, 45(11) 2435-2449. Aissa, R.; Francois, M.; Ruby, C. Fauth, F., Medjahdi, G., Abdelmoula, M., Génin J.-M., Formation and crystallographical structure of hydroxysulphate and hydroxycarbonate green rusts synthetised by coprecipitation. J. Phys. Chem. Solids, 2006, 67(5-6), 10161019. Khare, V.; Mullet, M.; Hanna, K.; Blumers, M.; Abdelmoula, M.; Klingelhöfer, G.; Ruby, C. Comparative studies of ferric green rust and ferrihydrite coated sand: Role of synthesis routes. Solid State Sci., 2008, 10(10), 1342-1351. Guerbois, D., Ona-Nguema, G., Morin G., Abdelmoula, M., Laverman, A.L., Mouchel, J.-M., Barthelemy, K., Maillot, F., Brest, J. (2014) Nitrite Reduction by biogenic hydroxycarbonate green rusts: Evidence for hydroxy-nitrite green rust formation as intermediate reaction product. Environ. Sci. Technol., 2014, 48(8), 4505-4514. Génin J.-M. R., Christian Ruby, Chandan Upadhyay(2006), Structure and thermodynamics of ferrous, stoichiometric and ferric oxyhydroxycarbonate green rusts; redox flexibility and fougerite mineral, Solid State Sci., 2006 8(11), 1330-1343 Antelo, J.; Fiol, S.; Pérez, C.; Mariño, S.; Arce, F.; Gondar, D.; López, R. Analysis of phosphate adsorption onto ferrihydrite using the CD-MUSIC model. J. Colloid Interf. Sci., 2010, 347(1), 112119. Navrotsky, A.; Mazeina, L.; Majzlan, J. Size-driven structural and thermodynamic complexity in iron oxides. Science, 2008, 319, 1635-1368. Antelo, J.; Avena, M.; Fiol, S.; Lopez, R.; Arce, F. Effects of pH and ionic strength on the adsorption of phosphate and arsenate at

Pe

[21]

Ruby et al.

[52]

[53]

[54]

[55] [56] [57]

DISCLAIMER: The above article has been published in Epub (ahead of print) on the basis of the materials provided by the author. The Editorial Department reserves the right to make minor modifications for further improvement of the manuscript.