Quantitative determination of phases in the alkali activation of fly ash. Part I. Potential ash reactivity

Fuel 85 (2006) 625–634 www.fuelfirst.com

Quantitative determination of phases in the alkali activation of fly ash. Part I. Potential ash reactivity A. Ferna´ndez-Jimeneza,*, A.G. de la Torreb, A. Palomoa, G. Lo´pez-Olmob, M.M. Alonsoa, M.A.G. Arandab a Eduardo Torroja Institute (CSIC), c/Serrano Galvache, n84, 28080 Madrid, Spain Department of Inorganic Chemistry, University of Ma´laga, Campus s/n, 29071 Mala´ga, Spain

b

Received 11 May 2005; received in revised form 10 August 2005; accepted 11 August 2005 Available online 9 September 2005

Abstract The various (vitreous and crystalline) components of two type F fly ashes are quantified in this paper using three techniques: chemical analysis with selective solutions, X-ray powder diffraction combined with the Rietveld method and nuclear magnetic resonance. The findings confirm the suitability of the techniques to pursue the objectives while providing further insight into the chemical composition of the vitreous phase of the ash as well as an understanding of the thermal history of these materials. Finally, the paper corroborates the grounds established in prior research for regarding the glassy constituents of an ash to be instrumental in the control of its alkali reactivity during the manufacture of ‘alkaline cements’. q 2005 Elsevier Ltd. All rights reserved. Keywords: Fly ash; NMR-MAS; Rietveld

1. Introduction Fly ash is a by-product of coal-fired steam power plants. With the growing demand for energy, more and more ash is being generated (production is expected to reach 800 Mton by 2010). As only a small portion (30–40%) is used at present, the rest has to be disposed of, with the concomitant problems of space, risk of air pollution, leaching and ensuing pollution of inland and marine water. The use of this by-product in the construction industry has been and continues to be the subject of many research studies [1,2]. Important progress is being made in the search for further ways to recycle fly ash, such as the development of new types of binders known as ‘alkaline cements’, which take their name from the industrial process involved in their manufacture [3,4]. The alkali activation of fly ash (AAFA) is a distinctive procedure in which this grey powder (FA) is mixed with alkaline activators (alkaline solutions) and then cured at * Corresponding author. Tel.: C34 1 30 20440; fax: C34 1 30 26047. E-mail address: [email protected] (A. Ferna´ndez-Jimenez).

0016-2361/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.fuel.2005.08.014

a certain temperature to produce a solid material. The glassy constituent of the fly ash transforms into a compact cement in the process. Previous studies [5–7] have shown that the main reaction product formed in AAFA is an amorphous alumino-silicate gel. Until very recently, this X-ray amorphous material has been difficult to characterise. Today, however, the evidence shows that it is a ‘zeolite precursor’: a short-range order material with a threedimensional structure, in which the Si is found in a variety of environments, but predominantly as Q4(3Al) and Q4(2Al). Palomo et al. [5,8,9] reported finding an Al-rich phase (meta-stable/intermediate phase) after short curing times. This phase evolves with curing time to a more stable Si-rich phase (zeolite precursor). Indeed, it might be hypothesised that the final stage of fly ash alkali activation—in the very long term—may be crystallisation into zeolite. The same authors recently published a paper establishing the fundamentals governing the reactivity of type F fly ashes as potential alkaline binders [10]. They reported that one of the main parameters in the control of the alkali activation process is the composition of the vitreous phase in the fly ash. In the light of those findings, the primary objective pursued in the present analysis (which, given its length,

A. Ferna´ndez-Jimenez et al. / Fuel 85 (2006) 625–634

626

is published in two parts), consisted of identifying and quantifying the proportion of glassy constituent and crystalline phases in the fly ashes from two different thermal power plants (part I), and identifying and quantifying the reaction products of these ashes after alkali activation (part II). Both analytical and instrumentals techniques were used: Selective chemical treatments [11, 12], Rietveld X-ray powder diffraction (XRPD) [13,14] and 29 Si and 27Al MAS NMR.

2. Experimental methods Characterisation of raw materials. Two different type F fly ashes, identified as L and M, were used in this study. Table 1 gives the chemical composition and amount of reactive silica, in percent, found with analyses conducted to Spanish standards UNE 80-215-88 and UNE 80-225-93. Table 2 shows the particle size distribution of each fly ash in terms of the percentage of powder retained in different size sieves. Selective chemical treatment with 1% HF. After characterisation, the ashes were treated with a 1% HF solution. This type of chemical treatment dissolves the vitreous phase of the ash, but not the crystalline phases (mullite, quartz, hematite, etc.). The procedure described by D.M. Roy [12] was followed: 1 g of ash was added to 100 ml of 1% HF in a plastic beaker, stirred for 6 h at ambient temperature and then filtered. The filter paper containing the insoluble residue was then washed to a neutral pH, placed in a platinum crucible and oven-dried for 1–2 h at 100 8C. Thereafter, it was calcined at 1000 8C and the vitreous phase was determined by subtracting the final from the initial mass. Ash L was subjected to a second HF treatment to verify the effectiveness of the method. Techniques. Both the initial ash and the residue obtained after the 1% HF treatment were characterised mineralogically and microstructurally using XRPD (Rietveld quantitative analysis) and 27Al and 29Si MAS (Magic Angle Spinning) NMR. 2.1. XRPD Sample preparation. Standard a-Al2O3 was synthesised as follows: 6 g of g-Al2O3 (99.997% from Alfa) was ground in an agate ball mill at 200 rpm for 30 min. The resulting powder was placed in a Pt crucible and heated at 1200 8C for

4 h. The oxide was allowed to cool to 150 8C over w5 h and ground at room temperature in an agate mortar for 5 min. The sample was subjected to a second thermal treatment at 1300 8C for 6 h and then cooled as above. This standard was then ground in an agate mortar for 5 min and sieved (! 35 mm) prior to weighing. Each of the mixtures used for analysis was carefully ground in an agate mortar for 10 min, adding acetone to facilitate particle dispersion, and then heated at 60 8C. Around 30% (wt) of corundum was added, with the exact quantity recorded for each measurement. The samples were gently loaded (vertically) onto an aluminium X-ray sample holder. X-ray data collection. Laboratory XRPD measurements were performed for all mixtures on a Siemens D5000 automated diffractometer with Cu Ka 1,2 radiation ˚ ) and a secondary curved graphite monochroma(1.5418 A tor. The readings were taken in vertical Bragg-Brentano (q/2q) geometry (flat reflection mode) between 18 and 708 (2q) at 0.038 steps, measuring 15 s per step. The samples were rotated at 15 rpm during acquisition to improve powder averaging, which is essential to have accurate intensities and hence good phase analysis. The D5000 diffractometer optic consisted of a system of primary Soller foils between the X-ray tube and the 2-mm fixed aperture slit. One 2-mm scattered-radiation slit was placed downstream of the sample, followed by a system of secondary Soller slits and a 0.2-mm detector slit. The X-ray tube operated at 40 kV and 30 mA. X-ray data analysis. Rietveld refinement was used on the powder patterns [15], applying GSAS software [16] with a pseudo-Voigt function [17] and including a correction for asymmetry due to axial divergence [18]. The crystal structures used to interpret the powder patterns were taken from the Inorganic Crystal Structure Database (ICSD). The collection codes for each structure were: corundum 73725; a-quartz 63532; mullite 66263; maghemite 87119; calcium oxide 61550; calcite 80869; aphthitalite 26018 and albite 68193. Neither the positional nor the thermal vibration parameters were refined. The parameters optimised were: background coefficients, cell parameters, zero-shift error, peak shape parameters (including anisotropic terms if needed), and phase fractions. 29 Si and 27Al MAS NMR spectroscopic characterisation was conducted with a Bruker apparatus, model MSL-400. The resonance frequencies used in this study were 79.5 and 104.3 MHz and the spinning rates were 4 and 12 kHz. The measurements were made at ambient temperature with

Table 1 Elemental chemical composition of the fly ashes (expressed as oxides wt%) Oxide%

SiO2

Al2O3

Fe2O3

CaO

MgO

Na2O

K2O

SO3

Reactive SiO2

FA La FA Mb

51.51 59.89

27.47 27.67

7.23 3.02

4.39 3.45

1.86 1.22

0.70 0.94

3.46 1.01

0.15 0.51

42.17 45.07

a b

Loss on ignitionZ1.89 and insoluble residueZ0.40. Loss of ignitionZ2.09 and insoluble residueZ0.20.

A. Ferna´ndez-Jimenez et al. / Fuel 85 (2006) 625–634

627

Table 2 Particle size distribution (%) of the fly ashes Sieve mesh size FA L FA M

210 mm

125 mm

90 mm

45 mm

Total

O45 mm

!45 mm

0.40 0.5

3.45 0.6

2.03 0.8

15.62 8.0

100 100

21.5 9.9

78.50 90.1

tetramethylsilane (TMS) and AlðH2 OÞ3C as external 6 standards. The error in the chemical shift values was estimated to be lower than 1 ppm. Magnetic materials were removed from the samples prior to NMR spectra recording by exposing the sample to a strong magnetic field.

3. Results Table 3 shows the insoluble residue content, in percent, found after treating the ash samples with 1% HF as described in the preceding section; the (soluble) vitreous phase content was computed by subtracting the mass of the residue from the initial mass of the material. The results for sample LHF2(R) were obtained by re-calculating the values found for sample LHF2 (second treatment) with respect to original ash L. A comparison of the values for samples LHF(R) and MHF shows that ash L has of the order of 25% more glassy phase than ash M. Rietveld XRPD phase quantification was conducted on these same samples, both for the initial ash and the insoluble residue obtained after HF treatment, to determine the phases present and compare these results to the findings from chemical analysis. The vitreous constituent was quantified by mixing all samples with (approximately) 30% wt of corundum (Al2O3) as described above. Rietveld quantitative analysis of the amorphous content relates the overall amorphous/vitreous content, A (wt%), to the overestimation, in Rietveld refinement, of an internal crystalline standard, Al2O3. Assuming a standard with no amorphous phases, A is obtained from Eq. (1) Table 3 1% HF-insoluble residue in the fly ashes Sample

1% HF treatment

Name

Characteristics

L L HF L HF2 LHF2(R) M M HF

Raw material First HF treatment Second HF treatment Recalculation Raw material First HF treatment

a

Insoluble residue (%)

Soluble phase (vitreous component) (%)

35.1a 67.85b 23.81

64.90 32.15 76.18

45.72b

54.28

Insoluble residue obtained without calcing the sample, to subject it to a second treatment. b Insoluble residue obtained after calcinations at 1000 8C.

1K AZ

  WS RS

100KWS

!104 %

(1)

where WS (%) is the concentration by weight of the internal standard and RS (wt%) its Rietveld-quantified concentration. Fig. 1(a) and (b) contains the Rietveld plots for initial fly ashes L and M, respectively, which show that both ashes present a pattern (halo) typical of vitreous/amorphous materials at 2q angles of between 20 and 308. The signals corresponding to the crystalline phases found in the material—quartz, mullite and Fe2O3 in the form of maghemite—can also be seen in these diffractograms. When the ash is treated with 1% HF, most of the halo associated with its glassy constituent disappears (see Fig. 2, traces (a) and (b), for ash L (LHF) and M (MHF), respectively). At the same time, these diffractograms evidence an increase in the intensity of the peaks associated with the crystalline phases, compared to the traces in Fig. 1. The Rietveld plots for ashes M and L in Fig. 2 show that after the acid treatment the halo disappears almost completely from the former, whereas a significant portion of the vitreous phase of ash L remains undissolved (Fig. 2(a)—LHF). This is why the ash L HF-insoluble residue was subjected to a second treatment under identical conditions. The diffractogram in Fig. 3 shows the pattern for the insoluble residue obtained after this second treatment (sample LHF2). Table 4 shows the results of Rietveld quantification of the diffractograms in Figs. 1–3. The values given in the table for the vitreous phase are higher for both ashes L and M than the results found with the chemical treatment procedure (Table 3). Nonetheless, these values, like the data in Table 3, show a higher glassy content in ash L than in ash M. It may likewise be deduced from these results that ash M has twice the mullite content and a third more quartz than ash L. The rather obvious inference of these findings is that the differences between vitreous and crystalline phase content of ashes M and L are the chief explanation for the differences in their reactivity in alkaline media, and therefore in the cementitious properties of the resulting product. In short, these results indicate that while the ashes are very similar in terms of both their physical and chemical properties (see Tables 1 and 2), their considerable mineralogical differences have a visible impact on their reactivity. LHF and MHF, which have a clearly higher mullite content than the initial ash, were examined to ascertain

Al2O3; Ma

Al2O3

(a) L

Al2O3; SiO2 Al2O3 CaO

Mu;SiO S 2 Mu Al2O3

Mu

Ma;SiO2

SiO2 Al2O3

Mu

Mu;SiO2 Mu

SiO2

Ma Mu CaO Mu

-2.0

Counts

0.0

2.0

Ma;CaO a;

Al2O3

Al2O3

Mu SiO2

4.0

Mu;Ma

SiO2

Mu

Al2O3

8.0

X10E 3

6.0

Al2O3; Mu

A. Ferna´ndez-Jimenez et al. / Fuel 85 (2006) 625–634

628

20.0

30.0

40.0

50.0

60.0

70.0

Mu

CaCO3 Mu

Al2O3;Mu

Al2O3

Al2O3 Al2O3

Al2O3 CaCO3 Mu Mu

Al2O3

Mu;Ma

Al2O3 SiO2

(b) M

-2.0

Counts

0.0

2.0

SiO2

4.0

Mu

6.0

X10E 3

8.0

Al2O3

2-Theta, deg

20.0

30.0

40.0

50.0

60.0

70.0

2-Theta, deg Fig. 1. Rietveld plots (18–708/2q) showing the observed (crosses), calculated (solid line), and difference (bottom line) powder patterns for (a) ash L and (b) ash M. The marks denote the Bragg peaks of the different phases; the main diffraction peaks of each phase are highlighted.

whether Rietveld refinement revealed differences in the degree of crystallinity of that phase that might be indicative of different crystallisation temperatures or Si/Al ratios in its chemical composition. Table 5 lists the values of parameters a, b and c and unit cell volume (V) for mullite in both samples. The significance of these data merits some discussion, although the following is by no means intended as an in-depth structural review of mullite, a complex issue falling outside the bounds of the present study. Based on the papers by Cameron [19], Okada [20] and Fisher [21] (who sustain that parameters a and b change linearly with composition and the value of parameter c depends on the initial formation

temperature), the mullite in both types of ash may be said to present orthorhombic polymorphism, inasmuch as parameters a and b are similar to one another and a doublet is found on their respective diffractograms at 2q positionZ 26.28. With respect to the formation temperature, according to Okada [20], parameter c for mullite crystals formed at a temperature below 1200 8C is approximately 0.0005 nm smaller than for mullite crystals with the same composition but formed at temperatures of over 1200 8C. Assuming the mullite composition to be the same in ashes M and L, the parameter c values found in the present analysis (see Table 5) might suggest that ash L was formed at a higher

Mu;Al2O3 Mu;Af He

Mu

Al2O3 Mu;Af;He SiO2:Af

SiO2

SiO2:Mu

i 2:Af SiO Mu

He SiO i 2;Af

Al2O3

(a) LHF

Mu;He;Al2O3

Al2O3 Mu;Al2O3

Mu;Al2O3

629

Counts

-0.5

0.0

Mu He

Mu Af

SiO i 2

Mu

Mu

Al2O3

1.0 0.5

X10E4

SiO2

A. Ferna´ndez-Jimenez et al. / Fuel 85 (2006) 625–634

20.0

30.0

40.0

50.0

60.0

70.0

SiO2

2-Theta, deg

Counts

-0.5

Al2O3;Mu

Al2O3

Mu

Al2O3 Mu Mu

Albite

Al2O3

Al2O3

Al2O3

Al2O3

Mu;Ma Al2O3

SiO2

0.0

Crist

0.5

Mu

1.0

X10E4

(b) MHF

20.0

30.0

40.0

50.0

60.0

70.0

2-Theta, deg Fig. 2. Rietveld plots (18–708/2q) showing the observed (crosses), calculated (line), and difference (bottom line) powder patterns for the 1% HF-insoluble insoluble residues. (a) Ash L (LHF), (b) ash M (MHF). The marks denote the Bragg peaks of the different phases; the main diffraction peaks of each phase are highlighted.

temperature than ash M. This would substantiate the findings reported here, to the effect that ash L has a higher vitreous phase content than ash M. Finally, Fig. 4 contains the MAS NMR spectra for 29Si and 27Al in the initial ashes and the insoluble residues obtained after treating the samples with HF solution as described in the section on experimental procedure. Generally speaking, the wide signals on the silicon spectra for the ashes (L and M) are indicative of matrix heterogeneity in terms of the distribution of the Si atoms. Table 6 gives the spectral deconvolution values obtained assuming constant

bandwidth. These results show that the wide signal corresponds to the envelope of a series of signals. Hence, according to the literature, the peaks detected at K84, K94, K98 and K103 ppm correspond primarily to the vitreous phase content of the ash [5]; the peak appearing at around K87/88 ppm corresponds to the crystalline mullite present in the ash [22,23]; and finally, the peaks at values of over K108 ppm are attributed to the presence of different crystalline phases of quartz [Q4 (0Al) signals] [24]. A large portion of the 29Si NMR spectra for the initial ashes is missing in the spectra for the HF-insoluble residues

A. Ferna´ndez-Jimenez et al. / Fuel 85 (2006) 625–634

1.5

SiO2

630

All2O3;M u Al2O3

Al2O3

SiO2

Al2O3

Al2O3

Al2O3

Al2O3 Mag

Mag;Ma Mu Mu

Mu;Ma

Al2O3

Counts

0.0

0.5

SiO2

Mu

1.0

X10E 4

LHF2

20.0

30.0

40.0

50.0

60.0

70.0

2-Theta, deg Fig. 3. Rietveld plot (18–708/2q) for the HF-insoluble residue (LHF2) obtained after subjecting ash L to a second treatment, as in Fig. 1.

(see Fig. 4); and indeed, the part that has disappeared is the area associated with the vitreous phase of the ash. The most reactive environments in the ash are usually the ones around Si atoms surrounded by 2, 3 or even 4 aluminia; these environments correspond to the signals appearing on the spectrum between K84 and K103 ppm. The signals located at values higher than K108 ppm, on the contrary, which are associated with the free silica component (such as quartz), show a visible increase in intensity. This serves as further confirmation that the acid treatment with 1% HF primarily affects the glassy constituent of the ash, and not the crystalline phases. The results given in Fig. 4(b) for the 27Al MAS NMR spectra show that both initial ashes (L and M) present two wide components: one centred at C51.6 ppm, associated with tetrahedral aluminium (AlT), and another smaller signal at around C1 ppm associated with octahedral

aluminium (AlO). The octahedral component is attributed to the aluminium in the mullite present in the ash. This greater intensity of this peak in ash M than ash L is particularly visible in the 27Al spectra for the HF-insoluble residues (samples LHF and MHF). The immediate deduction from these spectra is that the AlT/AlO ratio is higher for ash L than ash M, corroborating the XRD findings, which showed that the latter had a higher mullite content.

4. Discussion A previous paper [10] showed that the vitreous phase content of a fly ash is one of the factors determining its reactivity in alkaline media and therefore in the manufacture of alkaline cement. The results of the present study not only

Table 4 XRPD Rietveld quantification on ashes L and M and their 1% HF-insoluble residues Sample

Quantification by Rietveld XRD (% by mass)

Name

Characteristics

Vitreous

Mullite

Quartz

Fe2O3 Maghemite

Total

L L HFb

Raw material First HF treatment (IRZ35.1%) Second HF treatment (IRZ24.12%) Re-calculation Raw material First HF treatment (IRZ45.72)

80.6 41.8

11.7 38.1

6.3 16.4

1.1 3.7

99.7C0.3aZ100 100

22.7

48.0

25.1

4.2

100

5.47 65.7 28.7

11.57 22.6 45.1

6.05 10.8 23.0

1.01 0.5 0.3

24.1 99.6C0.4cZ100 97.1C2.5dC0.4dZ100

L HF2 LHF2(R) M M HFb a b c d

CaO was also identified in this sample. IRZ1% HF-insoluble residue. Albite was also identified in this sample. Albite: 2.5% and CaO: 0.4%—were also identified in this sample.

A. Ferna´ndez-Jimenez et al. / Fuel 85 (2006) 625–634 Table 5 Mullite unit cell parameters obtained from the Rietveld refinements Sample

˚) a (A

˚) b (A

˚) c (A

˚) V (A

LHF MHF

7.5749(6) 7.5610(8)

7.6971(6) 7.6978(8)

2.8886(2) 2.8883(2)

168.42(2) 168.11(3)

confirm prior research, but also evidence the effectiveness of the methodology used to determine and quantify the vitreous constituents of fly ash. The data obtained with the various techniques for the insoluble residue had to be reformulated with Eqs. (2) and (3) for comparison to the data for the initial ash: VT ð%Þ Z SRHF C VR VR Z

VXRPD !IRHF 100

(2) (3)

where VT is the percentage of total vitreous phase in the ash; SRHF is the soluble residue (in 1% HFZdissolved vitreous phase) (see Table 3); VR is the residual vitreous phaseZ (vitreous phase not dissolved after 1% HF treatment); VXRPD is the % of any phase quantified by XRPD Rietveld analysis (see Table 4); IRHFZ1% HF-insoluble residue (see Table 3). Table 7 gives the amount of total vitreous phase found in the various samples as determined by the different methods

631

used. These results show good reproducibility for Rietveld quantification (the error is G2%). Such reproducibility is also observed in the quantification of the crystalline phases present in the insoluble residues (initially, these phases are regarded as stable and unaltered by the 1% HF acid treatment). The data in Table 7 also indicate that the quantity of vitreous phase found with the chemical treatment are of the order of 10–15% lower than the results obtained with the Rietveld technique. This is interpreted as being due to the fact that the 1% HF treatment does not fully dissolve the entire glassy constituent of the ash and that a poorly crystallised (not XRD-detectable) fraction of mullite or quartz remains insoluble after the treatment. Finally, the results obtained in this study also serve to determine the SiO2 and Al2O3 content in the vitreous phase and therefore deduce the quantities of these compounds available to form the alkaline silico-aluminate gel that is the main product of the alkali activation reaction [5]. The values in Table 8 were found using Eqs. (4) and (5) ½SiO2 vitreous Z ½SiO2 Total ðsee Table 1Þ K½SiO2 quartz K½SiO2 Mullite

(4)

½Al2 O3 vitreous Z ½Al2 O3 Total K½Al2 O3 Mullite

(5)

Fig. 4. (a) 29Si MAS NMR and (b) 27Al MAS NMR spectra of the initial ashes and 1% HF-insoluble resides.

A. Ferna´ndez-Jimenez et al. / Fuel 85 (2006) 625–634

632

Table 6 Results of 29Si MAS NMR spectral deconvolution for the fly ashes Chemical shift (error G1 ppm) L

LHF

M

MHF

Pos. (ppm) Width Integration (%) Pos. (ppm) Width Integration (%) Pos. (ppm) Width Integration (%) Pos. (ppm) Width Integration (%)

84

K88

K94

K98

K103

K108

K113

K118

K84.0 7.15 5.32

K88.3 7.15 11.0 K88.5 7.48 9.34 K89 8.52 14.25 K89.7 4.99 11.21

K93.6 7.15 16.98 K93 7.78 10.39 K95.2 8.52 7.76 K94.6 4.99 9.19

K98.6 7.15 20.66 K97.5 7.78 12.24 K99.4 8.52 9.15 K99.10 4.99 7.08

K103.4 7.15 19.46 K102.8 7.78 33.76 K104 8.52 17.58 K103.2 4.99 12.31

K108 7.15 20.9 K108.6 7.78 23.53 K109 8.52 34.17 K108.2 4.99 31.50

K115 7.15 5.61 K114.5 7.78 10.74 K116 8.52 16.84 K113.2 4.99 13.91

K118.5 4.99 4.29

K86 4.99 10.51

Table 7 Total vitreous (VT) and crystalline (VXRPD) phase content in the ashes Method

Phase

Chemical treatment XRPD Rietveld quantification

Vitreous phase

Sample L

LHF

LHF2

64.94

76.18

M

MHF 54.28

Vitreous phase

80.6

79.57

81.35

65.7

67.40

Quartz Mullite

6.3 11.7

5.75 13.37

6.05 11.57

10.8 22.6

10.51 20.6

In the case of ash M, the sum of SiO2CAl2O3 dissolved in the vitreous phase (54.16%) is almost the same as the value determined by chemical treatment with 1% HF (54.28%, see Table 3). In the case of ash L, there is a small difference, as discussed above, which may be due to the existence of poorly-crystallised and therefore more amorphous phases that are nonetheless unaltered by the acid treatment. Another finding of note in Table 8 is the low vitreous aluminium content in ash M, a factor that will prove to limit the reactivity of this ash. The number of resonances appearing in the NMR spectra, in turn, is identical to the number of physically non-equivalent nuclei per unit cell. Initially, wide spectra are obtained for both types of ash. The presence of Q4(nAl) units (where nZ0,1,2,3,4) may be deduced from signal deconvolution. When the samples are treated with 1% HF (samples LHF and MHF, see Fig. 4), the Si atoms surrounded by the largest number of Al are the first to be

attacked. Normalising as per Eq. (2) yields the percentages of the different units dissolved, which are shown in Table 9. These findings indicate that structural units Q4(4Al), Q4(3Al) and Q4(2Al) (signals K88, K94, K98 ppm) are most readily dissolved, whereas the signals associated with the units Q4(1Al) and Q4(0Al) are much less soluble in the HF medium. Another interesting finding (observed in the 29Si MAS NMR spectra for the 1% HF-insoluble residue) is the greater intensity of the signals appearing in chemical shifts of over K108 ppm (associated with quartz-type phases) compared to the mullite signals (from K87 to K94 ppm). Such higher signal intensity does not necessarily denote a greater amount of quartz than mullite in these samples, and indeed, as the Rietveld XRD quantification showed, the mullite content is higher than the quartz content in both. Rather, this finding may be a reflection of the fact is that in NMR spectra the number of resonances is equal to the number of physically

Table 8 Oxide content in the fly ashes Fly ash L M

Total Vitreous phase Total Vitreous phase

SiO2

Al2O3

SiO2/Al2O3

SiO2CAl2O3

51.51 40.53 59.89 40.05

27.47 20.45 27.67 14.11

1.87 1.98 2.16 2.84

78.98 60.98 87.56 54.16

These calculations were performed assuming that all the mullite present in the system is orthorhombic.

A. Ferna´ndez-Jimenez et al. / Fuel 85 (2006) 625–634 Table 9 Results obtained from

633

29

Si MAS NMR spectrum normalization Chemical shift K84 K88

L LHFnormalized (IRZ35.1%) LLHFNormalized(SUBTRACTING) (%) Solubilized M MHFNormalizad (IRZ45.72%) MMHFNormalized(SUBTRACTIN-

K94

K98

K103

K108

K115

11.0 3.28 7.72

16.98 3.64 13.34

20.66 4,29 16.37

19.46 11.65 7.61

20.9 8.25 12.65

5.61 3.77 1.84

4.80 K4.80

70.2 14.25 5.12 9.13

78.6 7.76 4.20 3.56

79,2 9.15 3.23 5.92

39.1 17.58 5.62 11.96

60.5 34.15 14.40 19.75

32.8 16.84 6.36 10.49

1.96 K1.96

0

64.1

45.9

64.7

68.0

57.8

62.3

0

5.32 – 100

K118

Total 99.93 34.88 59.53 (RSZ64.9) – 99.73 45.69 60.81K6.76Z54.05

G) (IRZ54.28%)

(%) Solubilized

non-equivalent nuclei per unit cell present in the system, whereas signal intensity provides an indication of the total number of nuclei in that position. For this reason, the signals for Si atoms in environments rich in Al (Q4(4Al)), such as in the case of mullite, are less intense. According to findings reported by Gomes and Francois [23], mullite generates an NMR signal at around K87 ppm that corresponds to tetrahedral SiO4 surrounded by three tetrahedral AlO4 and one polyhedral AlO6. At the same time, this phase generates other signals at K90 and K94 ppm that correspond, respectively, to one or two SiO4 substituting for a tetrahedral AlO4. Consequently, the intensity changes in these signals must be attributed to the extent to which the Si and Al atoms combine in the mullite. The 27Al MAS NMR spectra of the ashes, in turn, essentially comprise two resonances: one at around 53/54 ppm and the other at C2/K2 ppm. The strong resonance at 53/54 ppm is indicative of the existence of a majority of AlO4 as opposed to AlO6 units. In other words, it evidences the existence of –Si–O–Al– bonds. The greater quantity of octahedral aluminium in ash M than ash L, in turn, is seen more clearly when the 27Al MAS NMR spectra for the HF-insoluble residues are compared. These spectra seem to suggest that the AlO content increases after the chemical treatment, when in fact it is the quantity of AlT that declines, since it is partially dissolved by the HF solution. These samples (LHF MHF, see Fig. 4) have another resonance at around 43/44 ppm, attributed to the tetrahedral aluminium units in the mullite. This signal is much sharper on the MHF than the LHF trace, because the former has a higher mullite content. It has been shown [19–21] that the higher the calcination temperature, the higher the Si content in mullite, which is tantamount to saying that its AlT content rises while its AlO content drops. In the light of these observations it is reasonable to assume that ash M was produced at a lower calcination temperature than ash L, providing further support for the hypothesis that ash M has a lower vitreous phase content and the mullite in ash M has a higher AlO content.

5. Conclusions † The three techniques used in the study—chemical analysis with selective HF acid solution, Rietveld X-ray powder diffraction quantification and nuclear magnetic resonance—can be regarded as being highly effective in characterising and quantifying the phases present in fly ash. † Thanks to the complementary nature of these techniques, the information they generate can be used (in addition to obtaining a fairly accurate quantification of the vitreous phase content of fly ash), to deduce information on the thermal history of an ash and the SiO2 and Al2O3 content present in its vitreous phase. † A way to measure the Al2O3 content of the vitreous phase of an ash has been descried. The role played by this oxide is essential in the correct establishment of the chemical reactions taking place in the alkali activation processes.

Acknowledgements Funding for this research was provided by the Directorate General of Scientific Research under project COO-1999AX-038; a post-doctoral contract associated with the study was awarded by the CSIC co-financed by the European social bottom (REF. I3P-PC2004L). The authors wish to thank I. Sobrados and J. Sanz for their help with the MAS NMR studies.

References [1] Malhotra VM. Making concrete greener with fly ash. Concr Int 1999; 21(11):61–6. [2] Malhotra VM, Mehta PK. High-performance, high-volume fly ash concrete: materials, mixture proportioning, properties, construction practice, and case histories. Ottawa, Canada,: Printed by Marquardt printing Ltd.; 20020-9731507-0-X.

634

A. Ferna´ndez-Jimenez et al. / Fuel 85 (2006) 625–634

[3] Krivenko PV. Alkali cements First international conference of alkaline cements and concretes. Ukraine, Kiev 1994 p. 12–129. [4] Palomo A, Grutzeck MW, Blanco MT. Alkali-activated fly ashes a cement for the future. Cem Concr Res 1999;29:1323–9. [5] Palomo A, Alonso S, Ferna´ndez-Jime´nez A, Sobrados I, Sanz J. Alkaline activation of fly ashes. A 29Si NMR study of the reaction products. J Am Ceram Soc 2004;87(6):1141–5. [6] van Jaarsveld JGS, van Deventer JSJ, Lukey GC. The characterisation of source materials in fly ash-based geopolymers. Mater Lett 2003; 57(7):1272–80. [7] Ferna´ndez-Jime´nez A, Palomo A, Criado M. Microstructural development of alkali-activated fly ash cement: a descriptive model. Cem Concr Res 2005;35:1204–9. [8] Ferna´ndez-Jime´nez A, Palomo A. Activation of fly ashes: a general view Eighth CANMET/ACI international conference on fly ash, silica fume, slag and natural pozzolans in concrete Las Vegas, U.S.A. 2004; 23–29: ISBN: 0-87031-146-8. [9] Ferna´ndez-Jime´nez A, Palomo A. Alkali activated fly ashes structural studies trough mid-infrared spectroscopy Microporous Mesoporous Mater 2005; in press. [10] Ferna´ndez-Jime´nez A, Palomo A. Characterisation of fly ashes. Potential reactivity as alkaline cements. Fuel 2003;82:2259–65. [11] Granizo ML, Alonso S, Blanco-Varela MT, Palomo A. Alkaline activation of metakaolin: effect of calcium hydroxide in the products of reaction. J Am Ceram Soc 2002;85(1):225–31. [12] Arjuan P, Silbee MR, Roy DM. Quantitative determination of the crystalline and amorphous phases in low calcium fly ashes. Proceeding of the 10th international congress of the chemistry of cement, vol. 3, 3v020, 4 pp, Gothenburg, Sweden; June 2–6, 1997, pp. 2–6.

[13] De La Torre A, Bruque S, Aranda MAG. Rietveld quantitative amorphous content. J Appl Cryst 2001;34:196–202. [14] Rietveld HM. A profile refinement method for nuclear and magnetic structures. J Appl Cryst 1969;2:65–71. [15] McCusker LB, Von Dreele RB, Cox DE, Louer D, Scardi P. Rietveld refinements guidelines. J Appl Cryst 1999;32:36–50. [16] Larson AC, Von Dreele RB. Los alamos nacional laboratory report no. LA-UR-86-748; 1994. [17] Thompson P, Cox DE, Hasting JB. Rietveld refinement of DebyeScherrer synchrotron X-ray data from Al2O3. J Appl Cryst 1987;20: 79–83. [18] Finger LW, Cox DE, Jephcoat AP. A correction for powder diffraction peak asymmetry due to axial divergence. J Appl Cryst 1994;27: 892–900. [19] Cameron WE. Composition and cell dimensions of mullite. Am Ceram Soc Bull 1977;56(11):1003–10011. [20] Takayuki Ban, Kiyoshi Okada. Structure refinement of mullite by the rietveld method and a new method for estimation of chemical composition. J Am Ceram Soc 1992;75(1):227–30. [21] Fisher RX, Schneider H, Schmucker M. Crystal structure of al-rich mullite. Am Mineral 1994;79:983. [22] Hongping He, Jiugao Guo, Jianxi Zhu, Peng Yuan, Cheng Hu. 29Si and 27Al MAS NMR spectra of mullites from different kaolinites. Spectrochim Acta Part A 2004;60:1061–4. [23] Gomes S, Francois M. Characterisation of mullite in silicoaluminous fly ash by XRD, TEM and 29Si MAS NMR. Cem Concr Res 2000;30: 175–81. [24] Engelhardth G, Michel D. High resolution solid state NMR of silicates and zeolites. London: Wiley; 1987.