Development of new cement based matrices for safe disposal of hazardous metals: Cadmium and caesium

Synthesis of fly ash BC and zeolite

Spanish ASTM class F fly ash, here denominated FA-0, was used as a secondary raw material to obtain a gismondine type NaP1 zeolite (Z) and fly ash BC. Zeolite NaP1 (Z) was synthesised as described by Guerrero et al. 2000; Guerrero et al. 2004,13, 14 i.e. by stirring FA-0 in a 1M NaOH alkaline solution (ratio, 1∶3, w/w) for 2 h at 150°C. The solid phase was subsequently washed three times with demineralised water and dried to a constant weight at 50°C. The X-ray diffraction (XRD) and thermogravimetric findings for this material are shown in the section on ‘Characterisation of starting materials’ (Figs. 1 and 2).

OPEN IN VIEWER

Figure 1

X-ray diffraction patterns for starting materials (M: mullite; Q: quartz; H: hematite; G: gehlenite; Z: zeolite; α′: α′L-C2S; C: calcite): a class F fly ash; b gismondine type zeolite; c fly ash BC

OPEN IN VIEWER

Figure 2

Thermogravimetric–differential thermal analysis curves for gismondine type NaP1 zeolite

Fly ash BC was synthesised from FA-0 as described in a prior paper.13 The chemical and mineralogical compositions of FA-0 and BC are given in Table 1 and in Fig. 1 respectively.

OPEN IN VIEWER

Table 1

Chemical composition of starting fly ash (FA-0) and fly ash BC/wt-%

	FA-0	BC
LOI*	5·6	1·6
IR*	0·3	2·3
CaO	4·6	50·7
SiO2	48·9	27·2
Fe2O3	7·5	3·5
Al2O3	26·8	13·6
MgO	1·9	1·4
SO3	…	0·6
Na2O	0·7	0·3
K2O	3·7	1·1

*LOI: loss on ignition; IR: insoluble residue.

Matrix mix design

Designing these new cement based matrices called for prior optimisation of the components: BC, zeolite (Z) and nanosilica (Ns). The Ns used was a commercial colloidal nanosilica with a specific surface area of 100 m² g⁻¹, a particle size of 30 nm and an SiO₂ content of 45 wt-%. The matrices were prepared with or without 0·1M solutions of reagent grade CsCl or CdCl₂.

This prior optimisation entailed first mixing 10 g of 0∶100 to 70∶30 blends of zeolite (Z) and BC with CsCl or CdCl₂ at solution/solid ratios ranging from 0·7 to 1·2. The procedure was subsequently repeated, replacing 8% of the cement with nanosilica (Ns). The samples were cured for 1–7 days at 60°C, the latter proving to be the optimum curing time. Optimisation was established on the grounds of the waste solution/solid ratio, workability and hardening. In other words, the optimal design sought immobilisation of the maximum possible volume of CsCl or CdCl₂ solution and the highest Z/BC cement ratio that would ensure good workability and hardening.

The mix was scantly workable at solution/solid ratio values of <1·1 of cadmium solution, while for values of >1·1, it failed to harden. A zeolite free matrix, denominated ZBC-0 and ZBC-0-Ns, was consequently prepared for this ion. Similar results were obtained with the caesium ion, except that the matrix with a 50% replacement ratio exhibited good workability and hardening at a waste solution/solid ratio of 1·1. The leaching trials for this ion were conducted with matrices ZBC-0, ZBC-0-Ns, ZBC-5 and ZBC-5-Ns. Table 2 lists the matrices chosen for each ion and their respective denominations.

OPEN IN VIEWER

Table 2

Multicomponent matrices chosen for leaching test

Zeolite–BC–Ns/g	Cd solution Solution/solid = 1·1	Cs solution Solution/solid = 1·1
0–100–0	ZBC-0-Cd	ZBC-0-Cs
50–50–0		ZBC-5-Cs
0–92–8	ZBC-0-Ns-Cd	ZBC-0-Ns-Cs
50–42–8		ZBC-5-Ns-Cs

The samples were poured into 5×10 cm cylindrical moulds and removed after curing for 7 days at 60°C and >95% relative humidity. Reference matrices were prepared with water instead of the waste solution.

Leaching test

Leaching was conducted as specified in ANSI/ANS-16·1-1986.5 The leaching solution used was demineralised water with a conductivity of <5 μΩ cm⁻¹. The specimens were placed in individual plastic containers containing a sufficient volume to attain a solution/geometric surface area (of the specimen: 196·35 cm²) ratio of 10 cm and stored at 40°C for 0·083, 0·3, 1, 2, 3, 4, 5, 19, 47 and 90 days. Duplicate specimens of leachate were analysed.

Instruments

The matrices were characterised by a number of techniques. X-ray diffraction analyses were recorded on a Philips PW 1730 diffractometer fitted with a graphite monochromator using Cu K_α1 radiation. Thermal analyses were conducted on samples weighing ∼50 mg using a simultaneous analysis system (model 409 STA; Netzch, Inc., Exton, PA, USA). Specific surface area measurements,16 pore size distribution17 and nanoporosity were found from nitrogen adsorption isotherms generated by a Micromeritics ASAP 2010 analyser. The cadmium and caesium concentrations in the leachate were determined by inductively coupled plasma optical emission spectrometry on a Perkin-Elmer Optima 3300 D/V spectrometer.

Characterisation of starting materials

The chemical compositions of starting fly ash, FA-0 and fly ash BC determined as specified in Spanish standard UNE-EN 196-2 are given in Table 1. The FA-0 was found to conform to ASTM class F and European and Spanish standard EN-UNE 450 specifications, with a SiO₂+Al₂O₃+Fe₂O₃ content of >70% and a low CaO content. The majority crystalline phases in the starting FA-0 (Fig. 1a) were quartz (α-SiO₂), mullite (Al₆Si₂O₁₃) and hematite (α-Fe₂O₃); the amorphous halo between 15 and 35° in the 2θ angular zone was generated by the vitreous component in the fly ash.

The XRD pattern for the zeolite synthesised from FA-0 is shown in Fig. 1b. The majority crystalline phases detected were gismondine type NaP1 zeolite (Na₆Al₆Si₁₀O₃₂.12H₂O) (Joint Committee on Powder Diffraction Standards card no. 39-219) (the main product of hydrothermal synthesis) and mullite, present in the starting fly ash, FA-0. A thermogravimetric analysis (TG/ATD) was conducted from 25 to 1100°C to confirm zeolite formation. The first continuous massive weight loss observed on the thermogravimetric curve for the sample, between 25 and 400°C, concurred with a minimum differential thermal analysis value of ∼120°C, attributed to the release of water molecules from the NaP1 zeolite (Z). A second weight loss observed between 500 and 800°C was due to the release of CO₂ molecules, possibly the result of sample carbonation or weathering. This loss concurred with a differential scanning calorimetry curve nadir at 690°C.

The anhydrous BC obtained from FA-0 (Fig. 1c) showed broad reflections in the 32–33° (2θ) angular zone, attributed to α_L-Ca₂SiO₄, a poorly crystallised variety of belite; traces of gehlenite (Ca₂Al₂SiO₇) appeared, while the absence of free lime suggested 100% reactivity.

Leaching analyses

The quantities of cadmium ion in the leachate were below the inductively coupled plasma detection limit (6·65 ppb) throughout the trial. The leaching study could consequently not be conducted, an indication of the good performance of these new cement based matrices (ZBC-0-Cd and ZBC-0-Ns-Cd). In the cement based matrices in which ordinary Portland cement was the main component, the 5 day effective diffusion coefficient D_e for this ion was ∼1·7×10⁻¹⁰. This finding will be discussed in a subsequent paper.

The cumulative fraction leached (CFL) from matrices ZBC-0-Cs, ZBC-5-Cs, ZBC-0-Ns–Cs and ZBC-5-Ns-Cs at 40°C is plotted versus the square root of leaching time in Fig. 3.

OPEN IN VIEWER

Figure 3

Caesium: CFL versus square root of leaching time

The caesium concentration in the reference specimens was the same as in the aggressive solution: 6·91, 6·85, 6·83 and 6·96 g kg⁻¹ for matrices ZBC-0-Cs, ZBC-5-Cs, ZBC-0-Ns-Cs and ZBC-5-Ns-Cs respectively. The matrices containing Z or Ns (ZBC-5-Cs and ZBC-5-Ns) consistently exhibited CFL values one order of magnitude lower than observed in the plain ZBC-0-Cs matrix across all test times. In ZBC-0-Cs, the CFL values varied with the square root of time along a straight line (up to 5 days), which would suggest that up to that time, diffusion was the mechanism that governed leaching. In this case, however, an inflection was observed at 5 days in the CFL line. That may have been due to the formation of a film on the surface of the sample, which would hinder leaching, a development observed by other authors (Goñi, 1996; Hernández, 1997) in immobilisation systems for similar ions, such as calcium or sulphate.

The mean effective diffusion coefficients D_e for the four matrices studied in each leaching interval [(Δt)_n (s)] are given in Table 3.

OPEN IN VIEWER

Table 3

Effective diffusion coefficients (cm² s⁻¹) for matrices ZBC-0-Cs, ZBC-0-Ns-Cs, ZBC-5-Cs and ZBC-5-Ns-Cs

Time/days	(Δt)n/s	ZBC-0-Cs	ZBC-0-Ns-Cs	ZBC-5-Cs	ZBC-5-Ns-Cs
		De/cm2 s−1	De/cm2 s−1	De/cm2 s−1	De/cm2 s−1
0·083	7·2×103	5·3×10−8	1·0×10−7	8·8×10−10	1·1×10−9
0·29	1·8×104	4·2×10−8	3·9×10−9	9·6×10−10	7·2×10−10
1	6·1×104	3·8×10−8	1·0×10−9	6·8×10−10	7·1×10−10
2	8·6×104	3·0×10−8	4·4×10−10	4·3×10−10	7·0×10−10
3	8·6×104	3·7×10−8	1·8×10−10	5·8×10−10	6·8×10−10
4	8·6×104	2·7×10−8	2·1×10−10	4·9×10−10	8·5×10−10
5	8·6×104	1·8×10−8	1·4×10−10	3·3×10−10	7·2×10−10
19	1·1×106	4·6×10−8	9·6×10−11	5·1×10−11	9·4×10−11
47	2·4×106	2·4×10−8	7·8×10−11	4·5×10−11	4·8×10−11
90	4·0×106	1·4×10−8	2·0×10−10	2·9×10−11	2·5×10−11

The mean effective diffusion coefficient D_e in the first 5 days of leaching (8·6×10 s) (abbreviated test) was 1·8×10⁻⁸ cm² s⁻¹ for matrix ZBC-0-Cs. After 90 days of leaching (4·0×10⁶ s), the D_e for this matrix was very similar, 1·4×10⁻⁸, an indication that leaching continued over time. By contrast, the 5 day D_e value for the Ns containing, zeolite free matrix was 1·4×10⁻¹⁰ cm² s⁻¹, i.e. two orders of magnitude lower than the 1·8×10⁻⁸ cm² s⁻¹ calculated for matrix ZBC-0-Cs. This behaviour, indicative of a lower diffusion rate for the ZBC-0-Ns-Cs matrix, is observed to be present throughout the 90 days of the test.

In matrices containing zeolite with and without Ns (ZBC-5-Cs and ZBC-5-Ns-Cs), D_e was higher in the first 5 days than in the zeolite free matrix, with values of ∼3·3×10⁻¹⁰ and 7·2×10⁻¹⁰. D_e decreased after 90 days to 2·9×10⁻¹¹ and 2·5×10⁻¹¹ for matrices ZBC-5-Cs and ZBC-5-Cs-Ns respectively; these values, which were greater than reported by other authors2 in matrices with zeolite but no Ns, denoted the existence of different kinetic mechanisms depending on the presence or otherwise of Z and Ns in the cement based system.

The mean leachability index L is used to assess the efficiency with which a material immobilises waste; the threshold value for acceptance is 6. In all the matrices studied, the L value was above this threshold, at 7·5, 9·3, 9·6 and 9·6 for ZBC-0-Cs, ZBC-0-Ns-Cs, ZBC-5-Cs and ZBC-5-Ns-Cs respectively. Consequently, the four BC based matrices studied efficiently immobilised the caesium in nuclear waste. The highest efficiency was exhibited by matrices containing zeolite or nanosilica, however.

The pH values measured in the leachates obtained in the leaching test for cadmium and caesium ions are shown in Fig. 4. For the cadmium matrices, ZBC-0-Cd and ZBC-0-Ns-Cd, the pH value measured after 5 days was <11, and after 90 days, it remained between 11 and 12. While the pH was lower in the Ns containing matrix at both times, it was consistently >7, the critical value for cadmium immobility.7 In the caesium matrices, by contrast, the pH was >11 in all cases, reaching alkaline values upward of 12·5 for matrices lacking Ns, in particular the Z and Ns free matrix, ZBC-0-Cs.

OPEN IN VIEWER

Figure 4

Change in leachate pH versus leaching time

In light of these findings, the authors are conducting further studies of the leachates, the results of which will be published in a future paper.

Matrix characterisation

X-ray diffraction

Given the large number of diffraction results obtained for each matrix, only the most significant findings are discussed here. Figures 5 and 6 show the most prominent differences in the XRD patterns for matrices ZBC-0-Cs and ZBC-5-Cs before and after the leaching attack. The diffractograms for the equivalent matrices prepared with demineralised water instead of the CsCl solution are shown for comparison.

OPEN IN VIEWER

Figure 5

X-ray diffraction patterns for ZBC-0-Cs: a anhydrous BC, b after 7 days of curing at 60°C, without Cs, c after 7 days of curing at 60°C, with Cs and d after 90 days of leaching {St: strattlingite (Ca₂Al₂SiO₇.8H₂O); K: katoite [Ca₃Al₂(SiO₄)(OH)₈]; g: C–S–H gel (Ca_1·5SiO_3·5.xH₂O); m: calcium monosulfoaluminate [Ca₄Al₂(SO₄)O₆.10H₂O]; 3: α-Ca₄Al₂O₆Cl₂.10H₂O}

OPEN IN VIEWER

Figure 6

X-ray diffraction patterns for ZBC-5-Cs: a anhydrous BC, b starting zeolite (Z), c after 7 days of curing at 60°C, without Cs, d after 7 days of curing at 60°C, with Cs and e after 90 days of leaching {St: strattlingite (Ca₂Al₂SiO₇.8H₂O); K: katoite [Ca₃Al₂(SiO₄)(OH)₈]; g: C–S–H gel (Ca_1·5SiO_3·5.xH₂O); m: calcium monosulfoaluminate [Ca₄Al₂(SO₄)O₆.10H₂O]; 3: α-Ca₄Al₂O₆Cl₂.10H₂O}

After 7 days of curing at 60°C with demineralised water (Fig. 5b), the main hydrated compounds detected in matrix ZBC-0-Cs were C–S–H gel (Ca_1·5SiO_3·5.xH₂O), strätlingite (Ca₂Al₂SiO₇.8H₂O) and katoite [Ca₃Al₂(SiO₄)(OH)₈]. The main change induced by the presence of the CsCl solution (Fig. 5c) was the presence of calcium monosulfoaluminate [Ca₄Al₂(SO₄)O₆.10H₂O] and α-Ca₄Al₂O₆Cl₂.10H₂O. No changes were observed after leaching (Fig. 5d).

For matrix ZBC-5-Cs (Fig. 6), the main difference observed in the hydrated compounds was the absence of katoite (see Fig. 6b). In the presence of the CsCl solution, the intensity of the aforementioned hydrated compounds declined, while traces of α-Ca₄Al₂(Cl₂)O₆.10H₂O appeared as the result of the uptake of chlorides in the Ca₄Al₂(SO₄)O₆.10H₂O molecule. Note that crystalline caesium and cadmium compounds were not formed in any of the matrices studied, suggesting that these ions were not chemically bonded. The presence of Ns had no material effect that could be detected with XRD. The use of other techniques would be needed to track the effect of Ns in cement based matrices.

Nanoporosity

The BET-N₂ surface area, nanoporosity (micropore volume) and differential pore size distribution of matrices determined with nitrogen adsorption, along with the changes induced by leaching, are depicted in Figs. 7 and 8. For the cadmium ion, no notable change was observed in the BET-N₂ surface area over time (Fig. 7): only a slight increase was recorded in this parameter after 90 days of leaching, from 26 to 30 m² g⁻¹ for ZBC-0-Cd and from 61 to 65 m² g⁻¹ for ZBC-0-Ns-Cd.

OPEN IN VIEWER

Figure 7

Leaching induced changes in surface area

OPEN IN VIEWER

Figure 8

Effect of leaching on micropore volume and differential pore size distribution (dV/dD) in caesium matrices

Before leaching, caesium ion containing matrices ZBC-0-Cs and ZBC-0-Ns-Cs exhibited high surface areas of ∼50 m² g⁻¹, compared to 32 and 33 m² g⁻¹ in the matrices where part of the BC was replaced with zeolite (Fig. 7). The 90 day surface area values were higher in the matrices containing Z and Ns, however. The use of these new cement based matrices may therefore be said to constitute a safe and innovative immobilisation system, with surface area values of ∼99 m² g⁻¹ compared to a preleaching area of 33 m² g⁻¹.

A study of the variations in micropore volume (nanoporosity) revealed a number of developments (Fig. 8a). After leaching, this parameter rose in matrix ZBC-0-Cd from a negative (–0·002621 cm³ g⁻¹) to a positive (0·001356 cm³ g⁻¹) value. Moreover, the value attained with the preleaching microstructural refinement induced by the inclusion of Ns (matrix ZBC-0-Ns-Cd matrix) was practically unaffected after 90 days of leaching (varying from 0·003201 to 0·003553 cm³ g⁻¹).

For the caesium ion, the matrices without zeolite (ZBC-0-Cs and ZBC-0-Ns-Cs) (Fig. 8a) had a higher micropore volume before leaching than the zeolite containing matrices (ZBC-5-Cs and ZBC-5-Ns-Cs): 0·002066 and 0·003255 cm³ g⁻¹ versus 0·000428 and 0·000679 cm³ g⁻¹ respectively. Leaching induced several effects in these matrices. First, in the Z free specimens, this parameter dipped slightly to values of 0·001745 and 0·003073 cm³ g⁻¹. Second, in the zeolite containing matrices, micropore volume was found to rise one order of magnitude after leaching, from 0·000428 and 0·000679 cm³ g⁻¹ to 0·002868 and 0·00725 cm³ g⁻¹ respectively. This behaviour was associated with the differential pore size distribution after 90 days of leaching observed in caesium matrices (Fig. 8b). The peaks on the curves for matrices ZBC-5-Cs and ZBC-5-Ns-Cs, higher for the latter, centred at an average pore diameter of ∼4 nm denoted the high resistance of this matrix to leaching at 40°C.

These results are consistent with the rest of the findings presented here, according to which matrices ZBC-5-Cs and ZBC-5-Ns-Cs exhibited the lowest CFL values, D_e values of 2·9×10⁻¹¹ and 2·5×10⁻¹¹ and the highest L values, at 9·6 for the two matrices. More research is underway to explain the mechanism governing retention of these ions in zeolite and zeolite+nanosilica containing fly ash BC. The respective findings will be published in a future paper.