Evolution of iron speciation during hydration of C4AF

Waste Management 26 (2006) 720–724 www.elsevier.com/locate/wasman

Evolution of iron speciation during hydration of C4AF J. Rose a

e

a,b,*

, A. Be´nard

b,c

, S. El Mrabet a,b, A. Masion L. Olivi f, J.-Y. Bottero a,b

a,b

, I. Moulin d, V. Briois e,

CEREGE Equipe physico-chimie des interfaces, UMR 6635 CNRS/Universite´ Paul Ce´zanne Aix-MarseilleIII-IFRE PSME 112, Europole Me´diterrane´en de l’Arbois, BP 80, 13545 Aix en Provence Cedex 4, France b ARDEVIE, Europole Me´diterrane´en de l’Arbois, BP 80, 13545 Aix en Provence Cedex 4, France c INERIS, Domaine du Petit Arbois, BP 33, 13545 Aix en Provence, France d LERM, 10, rue Mercoeur, 75011 Paris, France LURE Laboratoire pour l’Utilisation du Rayonnement Electromagne´tique, Universite´ Paris-Sud, Orsay, France European Synchrotron Radiation Facility, BP 220, F-38043 Grenoble Cedex, France f ELETTRA Sincrotrone Trieste S.c.p.A. S. S. 14, Km. 163.5 in AREA Science Park, I-34012 Basovizza, TS, Italy Accepted 31 January 2006

Abstract It is now well accepted and demonstrated that calcium silicate, calcium aluminate and calcium sulfo aluminate (ettringite, AFm) phases exhibit a good capability to fix metals and metalloids. Unfortunately the role of minor phases and especially calcium-ferric aluminate phase, shorthand C4AF is not well defined. In other systems like in soils or sediments iron phases play a key role in the fixation of pollutant. In cement sorption isotherms, indicated that various metals can be retained by the C4AF hydrated products. Therefore the capabilities of those phase to retain heavy metal should not be neglected. Previous investigations have shown that the minerals formed during the hydration of C4AF are similar to those formed from C3A (pure tri-calcium aluminate) under comparable conditions. Nevertheless no investigation was conducted at the molecular level and there is still a controversy whether Fe substitutes for Al in the hydrated minerals in whole or in part, or if it forms FeOOH clusters scattered throughout the matrix. In this context we have conducted XAS experiments using synchrotron radiation. It was found that the hydration of C4AF forms C3AH6 (hydrogarnet) in which Fe randomly substitutes for Al as well as an amorphous FeOOH phase. Intermediate products like AFm (i.e., an ill organized lamellar phase) are also formed but rapidly evolve to C3AH6; iron does not seem to be incorporated in the AFm structure. Ó 2006 Elsevier Ltd. All rights reserved.

1. Introduction Although the calcium silicate and calcium aluminate phases are good candidates for metal retention in real cement, the role of the ferric-calcium aluminate phase (C4AF) should not be neglected. Indeed sorption isotherms indicated that metals showed a strong affinity for the hydrated C4AF phase (Moulin, 1999). The ferric-calcium aluminate content in Portland cement is 2–7% and the Fe2O3 content is in the 1–3.5% range. It is therefore reason*

Corresponding author. Tel.: +33 442971529; fax: +33 442971559. E-mail address: [email protected] (J. Rose).

0956-053X/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.wasman.2006.01.021

able to assume that a non-negligible proportion of the trace metals in an actual cement will be associated with Fe phases. However, the metal retention capabilities of those phases (either isolated or within a cement) have received only little attention. Moreover, the accurate speciation of iron in cement is not fully determined. Many investigations have shown that the minerals formed during the hydration of C4AF are essentially similar to those formed from tricalcium aluminate (C3A) under comparable conditions (Chatterji and Jeffrey, 1962; Collepardi et al., 1979; Emanuelson and Hansen, 1997; Taylor, 1997; Meller et al. (2004)). The first minerals formed in the absence or presence of CaSO4 are the ill organized AFm and AFt phases,

J. Rose et al. / Waste Management 26 (2006) 720–724

respectively. Both types of phases contain Al3+ and Fe3+ and tend to undergo further changes to form hydrogarnet phases. Rogers and Aldridge (1977), Fukuhara et al. (1981), and Brown (1987) indicated the possible formation of an amorphous FeOOH phase. Nevertheless no investigation was conducted at the molecular level and a controversy still exists: does all the Fe substitute for Al or do FeOOH clusters also formed? The aim of our work was to monitor the evolution of iron speciation during the hydration of C4AF in lime water. The structure of iron at the atomic scale has been investigated using X-ray absorption spectroscopy. 2. Materials and methods 2.1. C4AF hydration Starting materials consisted of a anhydrous ferrite (brownmillerite) C4AF obtained from Lafarge Ciments (Vivier, France) hydrated with lime water (LW) with a water/solid (w/s) ratio of 10 and 0.5 on a rotary shaker and filtered (0.2 lm). The pore solution was exchanged by acetone and the solid then dried at room temperature. The C4AF was hydrated for 24, 48 h and 7 days. Three samples were prepared in bi-distilled water at three initial pH (9, 11 and 10) with a w/s = 10. 2.2. X-ray diffraction Solids were ground to fine powder, and analyzed by Xray diffraction (XRD). A Philips PW3710 X-ray diffractometer with a Co Ka radiation at 40 kV and 40 mA was used. The diffractograms were scanned in the 2h 3.5–78° range with a counting time of 12 s per 0.02° step. 2.3. Electron microscopy SEM-EDX images and elemental analysis were obtained using a Philips XL30 SFEG-SEM coupled to an Oxford Instruments energy dispersive spectrometer (EDS). The SEM was operated at 20 keV. 2.4. X-ray absorption spectroscopy X-ray absorption spectroscopy (XAS) experiments were performed at the Laboratoire pour l’Utilisation du Rayonnement Electromagnetique (LURE, Paris-France) on the D44 beamline and at the ELLETRA synchrotron facility on the BL 11.1 beamline. A Si(1 1 1) monochromator crystals were used to scan energy from 100 eV below to 800 eV above the Fe K-edge. The samples were measured in the transmission mode. EXAFS data reduction was accomplished according to a procedure described previously (Manceau and Calas, 1986) by means of a series of programs developed by Michalowicz (1991). The extracted EXAFS were k3 weighted to enhance the high-k region and Fourier transformed over

721

˚ 1, to R space using a kaiser apothe k range 2.4 to 13–14 A dization window with s = 2.5. The resulting pseudo-radial distribution functions (p-RDF) are uncorrected for phase ˚ . The pshift leading to a shift of the peaks by 0.3–0.4 A RDFs consist of n peaks representing n coordination spheres at Rn distances from the central atom. Distances are uncorrected for phase shift and have to be displaced ˚ from the crystallotoward long distances by 0.3–0.4 A graphic position. The structural and chemical parameters Rj, Nj (N: number of atoms) and nature of atomic neighbors in the jth shell around Fe were determined by leastsquares fitting of partial EXAFS spectra. We have used the phase (/backscatterer(k), dcentral atom(k)) and amplitude (|fbackscatterer(h, k, R)|) functions calculated ab initio with FEFF8 (Zabinsky et al., 1995). For each atomic shell of the unknown samples, the interatomic distance R, the coordination number N, and the Debye–Waller factor r were adjusted. The limit on the number of free parameters (Nind = (2DkDR)/p 1) was in no case exceeded during ˚ and the fits. The uncertainties of R and N are 0.02 A 20%, respectively. 3. Results and discussion 3.1. Hydration kinetics of C4AF The key point to study the reactivity of the C4AF mineral phase is to determine the dissolution kinetics of the anhydrous mineral. In order to characterize the effect of pH, hydration experiments were performed at various liquid/solid ratio and hydration pHs (lime water, pH 11, 10 and 9 with NaOH). Fig. 1A presents the XRD pattern of C4AF hydrated for 48 h with a liquid/solid ratio of 10. The different peaks shown in Fig. 1A correspond to anhydrous C4AF and hydrogarnet (C3AH6). After 48 h of hydration with a liquid/solid ratio = 10, almost all the anhydrous C4AF has been dissolved. The C3AH6XRD peak intensities are slightly higher at pH 10 than at either pH 9 or lime water, indicating that formation of C3AH6 is faster, or the crystallinity of the hydrogarnet is better, at pH 10 than in lime water. However, it should be noted that in the present case the relevant parameter is not the pH but the Ca concentration. Scrivener and Pratt (1984) indicated that lime water delays the C4AF hydration, presumably because of the sorption of calcium at the surface of C4AF thus creating a barrier to further hydration. After 7 days of hydration with a l/s = 0.5 the various samples exhibit the same diffraction patterns. In the case of the l/s = 10 samples, it is worth noting that after 24 h AFm phases are formed (Fig. 1B). The diffraction peaks at 7.5, ˚ can be attributed to carbonated phases. 7.7 and 8.2 A Indeed the samples were hydrated in presence of carbon˚ XRD peaks correspond to the ates and the 7.5 and 8.2 A C3A Æ CaCO3 Æ 11 H2O and C4A Æ (CO3)0.5 Æ 12H2O, respectively (Taylor, 1997). The peak at 7.7 can be attributed to ˚ ; Taylor, 1997). Such phases are tranthe C4AH13 (7.94 A sients since they are not detected after 48 h hydration. Dur-

722

J. Rose et al. / Waste Management 26 (2006) 720–724

C4AF L/S=60 7days lime water C4AF L/S=60 48 H lime water

A

Lepidocrocite Goethite

X-ray counts

C4AF

Fe garnet (andradite)

Hydrated C4AF 48 hours Lime Water Hydrated C4AF 48 hours NaOH pH=1

0

C3AH6

10

20

30 40 2 theta (˚) (Co k α)

50

60

C4AF hydrated 24 H

B

C4AF hydrated 48 H

Fourier transform (magnitude)

Hydrated C4AF 48 hours NaOH pH=9

C4AF anhydrous 7.7 Å 7.5 Å

Anhydrous C4AF C3AH6

8.2 Å

0

0.5

1

1.5

2

2.5

3

3.5

4

R- ΔR (Å)

Fig. 2. Pseudo-radial distribution functions of C4AF hydrated during 48 h and 1 week and three reference compounds: Fe-garnet and two iron oxyhydroxides goethite and lepidocrocite. 8

10

12

14 16 18 2 theta (˚) (Co k α)

20

22

Fig. 1. (A) XRD pattern of anhydrous C4AF, C3AH6 and various hydrated C4AF after 48 h with l/s = 10; (B) XRD pattern of C4AF hydrated after 24 and 48 h.

ing the hydration processes and in the absence of calcium sulfate AFm minerals converts to hydrogarnet. Therefore, the hydration of C4AF leads to the formation of hydrogarnet regardless of the pH of hydration. After 7 days, some AFm still exists but no crystallized iron oxide or hydroxide phases have been detected. 3.2. Iron speciation For a more in-depth investigation, EXAFS spectroscopy has been used to determine the atomic environment of iron. This technique provides complementary information to XRD since amorphous phases are also identified by EXAFS. From XRD we know that AFm and C3AH6 are formed. The question is whether or not Fe can replace Al in the hydrogarnet and AFm structures and if other Fe phases are present. Fig. 2 shows the pseudo-radial distribution function (pRDF) of C4AF hydrated for 1 week; it is compared to the p-RDFs of Fe-garnet, and two iron oxy-hydroxides (goethite and lepidocrocite). The various peaks of the p-RDF correspond to the probability to find atoms around the central atom (Fe) at the distance (R) of the peaks. In the case of iron the first peak corresponds to the presence of oxygen ˚ while the atoms with a Fe–O interatomic distance 2 A

second peak corresponds to atoms in the next nearest coordination spheres. The differences in position and amplitude suggest that the atomic environment of Fe is different in the sample and in the reference compounds, i.e., pure Fe-garnet or oxy-hydroxides are not formed. The modelling of the first peak indicates that each iron is ˚ . Therefore each iron surrounded by 6 oxygen atoms at 2 A is octahedrally coordinated. The results of the modeling of the second peak are presented in Table 1. These results indicate each iron to be surrounded by 2.1 ± 0.4 Fe atom ˚ , 0.2 ± 0.04 Fe atoms at 3.45 ± 0.02 A ˚ at 3.03 ± 0.02 A ˚ and 2.1 ± 0.4 calcium atoms at 3.59 ± 0.02 A. The interpretation of EXAFS modeling is deduced by comparison with atomic structures of reference compounds. Table 2 presents the atomic structure of different reference compounds and Fig. 3 illustrates the corresponding structures of Fe at the molecular scale. The comparison between structural results obtained in the case of C4AF hydrated 1 week and the structural data of reference compounds suggests that the Fe–Fe interatomic distances determined for Table 1 Structural results of the analysis of the second coordination sphere of Fe in C4AF hydrated for 24 h and 1 week in lime water with l/s = 10 ˚) ˚) Sample Atomic pair R (A r (A N Residue C4AF, 24 h

Fe–Fe Fe–Fe

3.03 3.46

0.111 0.093

3.1 1.3

0.012

C4AF, 1 week

Fe–Fe Fe–Fe Fe–Ca

3.03 3.45 3.59

0.104 0.070 0.091

2.1 0.2 2.1

0.0389

J. Rose et al. / Waste Management 26 (2006) 720–724

723

Table 2 Iron atomic environment of different reference compounds Mineral

Atomic pair

˚) Distance (A

Number

Reference

Lepidocrocite (c-FeOOH)

Fe–O Fe–O Fe–O Fe–Fe

1.91 2.02 2.12 3.058

1 3 2 6

Christensen and Christensen (1978)

Goethite (a-FeOOH)

Fe–O Fe–O Fe–Fe Fe–Fe Fe–Fe

1.95 2.09 3.01 3.28 3.46

3 3 2 2 4

Szytula et al. (1968)

Ferrihydrite (amorphous-FeOOH)

Fe–Fe Fe–Fe

3.01 3.43

4.5 3.9

Manceau and Drits (1993)

Hemathite (a-Fe2O3)

Fe–O Fe–O Fe–Fe Fe–Fe Fe–Fe Fe–Fe

1.94 2.11 2.90 2.97 3.36 3.70

3 3 1 3 3 6

Blacke et al. (1966)

Andradite Fe-garnet

Fe–O Fe–Ca Fe–Si

2.02 3.36 3.36

6 6 6

Hazen and Finger (1989)

Hydrogarnet C3(A/F)H6

Al/Fe–O Al/Fe–Ca

1.91 3.51

6 6

Lager and Von Dreele (1996)

AFm

Al/Fe–O Al/Fe–Ca

1.90 3.35

6 5

˚ < Fe–Fe < 3.36 A ˚ ) and double corner Fig. 4. Face, edge (2.97 A ˚ ˚ (3.40 A < Fe–Fe < 3.66 A) sharing between Fe–O octahedra.

Fig. 3. Atomic environment around Fe in different Fe minerals. These differences in term of structure lead to the interatomic distances shown in Table 2.

the hydrated C4AF are similar to those of amorphous FeOOH. From the 3D structure it is possible to determine the type of linkage between Fe octahedra corresponding to ˚ . The short Fe–Fe interatomic distances of 3.03 and 3.45 A

Fe–Fe distance corresponds to edge sharing between Fe octahedra and the longer one corresponds to double corner linkages, as illustrated in Fig. 4. The fact that Fe octahedra are linked mainly through edge sharing indicates that Fe hydroxide or oxy-hydroxides exist in the sample. If Fe was only present in the hydrogarnet or AFm, no Fe–Fe contribution would be detected. The Fe–Ca atomic pair determined in the hydrated C4AF corresponds to the hydrogarnet structure and not to the AFm phase. Indeed the Fe–Ca interatomic distances in AFm and hydrogarnet are very different and the distance determined in the hydrated sample is similar to the Fe–Ca distance of the hydrogarnet. In the case of the C4AF hydrated for 24 h, XRD analysis indicate that AFm phases are formed. EXAFS results indicate that no Fe–Ca interaction exist, thus suggesting that Fe is not associated with AFm phases but that amorphous FeOOH phases are formed.

724

J. Rose et al. / Waste Management 26 (2006) 720–724

Therefore, it appears that the hydration leads to the formation of hydrogarnet in which Fe can replace Al, and the additional formation of Fe amorphous oxy-hydroxides. Acknowledgments We thank the European Community for supporting this work through the INERWASTE Craft European program, and specially the YPREMA company. References Blacke, R.Z., Hessevick, R.E., Zoltai, T., Finger, L.W., 1966. Refinement of the hematite structure. American Mineralogist 51, 123–129. Brown, P.W., 1987. Early hydration of tetracalcium aluminoferrite in gypsum and lime–gypsum solutions. Journal of the American Ceramic Society 70 (7), 493–496. Chatterji, S., Jeffrey, J.W., 1962. Studies of early stages of paste hydration of cement compounds. Journal of the American Ceramic Society 45, 536–543. Christensen, H., Christensen, A.N., 1978. Hydrogen bonds of c-FeOOH. Acta Chemica Scandinavica A32, 87–88. Collepardi, M., Monosi, S., Moriconi, G., Corradi, M., 1979. Tetracalcium aluminoferrite hydration in the presence of lime and gypsum. Cement and Concrete Research 9 (4), 431–437.

Emanuelson, A., Hansen, S., 1997. Distribution of iron among ferrite hydrates. Cement and Concrete Research 27 (8), 1167–1177. Fukuhara, M., Goto, S., Asaga, K., Daimon, M., Kondo, R., 1981. Mechanisms and kinetics of C4AF hydration with gypsum. Cement and Concrete Research 11, 407–414. Hazen, R.M., Finger, L.W., 1989. High-pressure crystal chemistry of andradite and pyrope: revised procedures for high-pressure diffraction experiments. American Mineralogist 74, 352–359. Lager, G.A., Von Dreele, R.B., 1996. Neutron powder diffraction study of hydrogarnet to 9.0 GPa. American Mineralogist 81, 1097– 1104. Manceau, A., Calas, G., 1986. Nickel-bearing clay-minerals. 2. Intracrystalline distribution of nickel – an X-ray absorption study. Clay Mineralogy 21, 341–360. Manceau, A., Drits, V.A., 1993. Local structure of ferrihydrite and feroxyhite by EXAFS spectroscopy. Clay Minerals, 165–184. Meller, N., Hall, C., Jupe, A.C., Colston, S.L., Jacques, S.D., Barnes, P., Philipps, J., 2004. The paste hydration of brownmillerite with and without gypsum: a time resolved synchrotron diffraction study at 30, 70, 100 and 150 °C. Journal of Material Chemistry 14, 428–435. Moulin, I., 1999, Cristallochimie des me´taux dans les phases hydrate´es des ciments. Ph’D University Paul Ce´zanne-Aix Marseille III Aix-en Provence, France. Rogers, D.E., Aldridge, L.P., 1977. Hydrates of calcium ferrites and calcium aluminoferrites. Cement and Concrete Research 7 (4), 399–409. Scrivener, K.L., Pratt, P.L., 1984. Backscattered electron images of polished cement sections in the SEM. In: Proceedings of the 6th Annual International Conference on Cement Microscopy, pp. 145–155.