Use of Infrared Spectroscopy to Study Geopolymerization of

Design of a novel nanocomposite with C-S-H@LA for thermal energy storage: A theoretical and experimental study. Jinyang Jiang , Qi Zheng , Yiru Yan , ...
0 downloads 0 Views 178KB Size
Download PDF
8726

Langmuir 2003, 19, 8726-8734

Use of Infrared Spectroscopy to Study Geopolymerization of Heterogeneous Amorphous Aluminosilicates W. K. W. Lee and J. S. J. van Deventer* Department of Chemical and Biomolecular Engineering, The University of Melbourne, Parkville, Victoria, 3010, Australia Received June 25, 2002. In Final Form: July 4, 2003 Geopolymerization is a general term to describe all the chemical processes that are involved in reacting aluminosilicates with aqueous alkaline solutions to produce a new class of inorganic binders called geopolymers. In the present work, a novel analytical procedure is developed to study geopolymerization of amorphous aluminosilicates in real time. This procedure involves first conducting a series of welldesigned leaching experiments, within which an aluminosilicate is alkali-activated with aqueous alkaline solutions of varying alkalinities and soluble silicate dosages. The leached solutions are diluted and analyzed using inductively coupled plasma equipped with optical emission spectroscopy (ICP-OES), and the activated solid particles are separated, washed, desiccated, and analyzed by Fourier transform infrared (FTIR) spectroscopy. By comparing the results obtained from the ICP-OES and the FTIR analyses, a linear calibration curve can be constructed to correlate the extent of the alkali-activation of the solid and its T-O-Si (T ) Al and Si) asymmetric stretching vibration frequency. Using this calibration curve as a basis, the extent of alkali-activation of the aluminosilicate of interest within a geopolymer can then be approximated in real time with the aid of IR spectral deconvolution. This novel analytical procedure can be used to study the roles of alkalis and soluble silicates in alkali-activation and geopolymerization of a highly heterogeneous and amorphous aluminosilicate such as fly ash. It can also be used to understand the reaction mechanism of geopolymerization and to determine the reaction conditions (such as the SiO2/R2O ratio of the activating solution, where R ) Na or K) that are critical in controlling the various reaction pathways, which in turn affect the products formed and the macroscopic (compressive) strengths of the products.

Introduction Apart from Portland cement, alkali-activation of aluminosilicate powders can also produce inorganic binders of excellent physical and chemical properties suitable for construction purposes. Aluminosilicate powders are reacted with aqueous alkaline solutions to activate and promote solid dissolution, which is then followed by polycondensation of the dissolved species to form an X-ray amorphous aluminosilicate gel (the principal binding phase responsible for holding the unreacted solid particles together) and the solid-state transformation of the gel.1-3 The resultant products of such chemical processes are termed geopolymers. Within the geopolymer research community, the aluminosilicate gel is widely perceived as a three-dimensional framework of SiO4 and AlO4 tetrahedra interlinked by shared O atoms.1-3 The negatively charged and tetrahedrally coordinated Al3+ atoms inside the network are charge-balanced by alkali metal cations from the activating solution. The properties of the aluminosilicate gel, and hence the macroscopic characteristics of the geopolymer, obviously are affected by the nature of the solid raw materials and the activating solutions as well as the curing temperature, humidity, pressure, and possible contamination. A wide range of aluminosilicates of vastly different mineralogy, microstructure, and composition has already been proven as potential solid raw materials for the synthesis of geopolymers.4-6 In general, greater physical * To whom correspondence should be addressed. Tel: +61-383446619. Fax: +61-3-83447707. E-mail: [email protected]. (1) Davidovits, J. J. Mater. Educ. 1994, 16, 91. (2) Palomo, A.; Glasser, F. P. Br. Ceram. Trans. J. 1992, 91, 107. (3) Van Jaarsveld, J. G. S.; Van Deventer, J. S. J. Miner. Eng. 1999, 12, 75. (4) Palomo, A.; Grutzeck, M. W.; Blanco, M. T. Cem. Concr. Res. 1999, 29, 1323. (5) Roy, D. M. Cem. Concr. Res. 1999, 29, 249.

strengths are achieved if amorphous solids, or solids containing high proportions of amorphous phase(s), are used as the raw materials, when compared with crystalline materials. On the other hand, changing the nature of the activating solution and/or the curing conditions has also been investigated extensively.7 In particular, addition of soluble silicates to the activating solution has been described as one of the essential conditions for synthesizing mechanically strong geopolymers.1,8 When using crystalline solids such as stilbite and kaolinite, X-ray diffraction (XRD) has been used with some success to show that addition of soluble silicates could induce greater dissolution of the crystalline solids and thus could produce greater strengths.8 Unfortunately, if solids of high amorphous contents and heterogeneity, such as fly ash, are used instead, quantification of the solid reactivity, that is, the extent of alkali-activation, will be more difficult than for the crystalline counterparts owing to the difficulty to separate each phase within a geopolymer and the lack of suitable analytical technique(s) to study heterogeneous amorphous materials. As amorphous aluminosilicates, which are often heterogeneous, are critical in producing mechanically strong and chemically stable geopolymers, it is of vital importance to develop an analytical technique/ procedure so that the effects of the various reaction variables, such as alkalis and soluble silicates, can be studied. As stated above, the term geopolymerization has been used to encompass all the chemical processes that are involved in producing geopolymers.8 This consists of the OH--promoted hydrolytic reaction, or alkali-activation of (6) Xu, H.; Van Deventer, J. S. J. Int. J. Miner. Process. 2000, 59, 247. (7) Granizo, M. L.; Blanco-Varela, M. T.; Palomo, A. J. Mater. Sci. 2000, 35, 1. (8) Xu, H.; Van Deventer, J. S. J.; Lukey, G. C. Ind. Eng. Chem. Res. 2001, 40, 3749.

10.1021/la026127e CCC: $25.00 © 2003 American Chemical Society Published on Web 09/12/2003

Geopolymerization of Amorphous Aluminosilicates

Langmuir, Vol. 19, No. 21, 2003 8727 Scheme 1

the aluminosilicate raw material, followed by the polymerization of the dissolved species (including the added soluble silicates) to form an aluminosilicate gel (the binding phase) and the subsequent solid-state transformation of the gel.1-3,6,8 With the assumption that Al is always in the fourth coordination, a simplified hydrolytic reaction on an aluminosilicate can be shown in Scheme 1. From Scheme 1, it is clear that during alkali-activation, every bridging oxygen atom (BO) of the original aluminosilicate is replaced by two negatively charged nonbridging oxygen atoms (NBO), which are charge compensated by the alkalis. Microstructurally, this is similar to adding alkalis to synthetic silicate or aluminosilicate glasses, whose alkali and water contents can be artificially controlled. From the literature,9-14 addition of alkalis as network modifiers to an (alumino)silicate is known to generate a greater concentration of NBOs within the structure. The TO4 (T ) Al or Si) units within the silicate network have become more isolated with increasing alkali inclusion and thus lower the molecular vibration force constant of the T-O bond. As a result, an infrared (IR) band attributable to the T-O-Si asymmetric stretching vibration of the TO4 tetrahedra of an (alumino)silicate glass has been found to shift to the lower energy end with increasing alkali content.10-14 The extent of the shift is approximately linear to the alkali content. With the view that a greater extent of alkali-activation of aluminosilicates should give rise to products of greater NBO concentrations, it is possible that similar trends can also be observed in geopolymerization of aluminosilicates especially at the early stages. This work, therefore, used infrared spectroscopy to test the above hypothesis with the aim to develop an analytical procedure and quantitatively measure the extent of alkaliactivation of the amorphous/glassy phase(s) within a typical fly ash. Class F fly ash is chosen as the source aluminosilicate material so as to demonstrate the suitability of this novel procedure to examine even a highly heterogeneous and amorphous aluminosilicate. It will also be shown that this novel procedure can be used to understand the reaction mechanism of geopolymerization and to determine the reaction conditions (such as the SiO2/ R2O ratio of the activating solution) that are critical in controlling the various reaction pathways, and thus the reaction products. Experimental Procedure Materials. Coal-origin fly ash from Gladstone, Australia, was obtained from Queensland Cement Limited (QCL). The fly ash, as determined by fusion analysis using a Siemens SRS3000 sequential X-ray fluorescence (XRF) spectrometer, typically contains 50.01 mass % SiO2, 27.97% Al2O3, 11.95% Fe2O3, 1.30% TiO2, 0.19% MnO, 3.48% CaO, 1.32% MgO, 0.69% K2O, 0.19% Na2O, 1.01% P2O5, 0.50% SO3, and combustibles. The major crystalline constituents of the fly ash are R-quartz (SiO2), mullite (9) Zirl, D. M.; Garofalini, S. H. J. Am. Ceram. Soc. 1992, 75, 2353. (10) Hanna, R.; Su, G.-J. J. Am. Ceram. Soc. 1964, 47, 597. (11) Hanna, R. J. Phys. Chem. 1965, 69, 3846. (12) Sweet, J. R.; White, W. B. Phys. Chem. Glasses 1969, 10, 246. (13) Clark, D. E.; Ethridge, E. C.; Dilmore, M. F.; Hench, L. L. Glass Technol. 1977, 18, 121. (14) McMillan, P. F.; Wolf, G. H. In Structure, Dynamics and Properties of Silicate Melts; Stebbins, J. F., McMillan, P. F., Dingwell, D. B., Eds.; Mineralogical Society of America: Washington, DC, 1995; pp 247-315.

Table 1. Calculated Molar Compositions of the Various Alkali-Activating Solutions system Aa Ba Ca Da Ea Fa Ga Ha Ia Ja Ka La Ma Na Oa Pa Qa Ra Sa Ta Ua Va Wa Xa Ya Za AAa Geo-Ib Geo-IIb Geo-IIIb Geo-IVb

alkali cation (R) [SiO2] (mM) [OH-] (mM) SiO2/R2O Na Na Na Na Na Na Na Na K K K K K K K K Na/Kc Na/Kc Na/Kc Na/Kc Na/Kc Na/Kc Na/Kc Na/Kc Na/Kc Na/Kc Na/Kc Na/Kc Na/Kc Na/Kc Na/Kc

0 14.24 28.48 42.72 71.20 213.60 427.60 569.60 0 14.24 28.48 42.72 71.20 213.60 427.60 569.60 0 14.24 28.48 213.60 569.60 1000.00 2000.00 0 2500.00 0 2500.00 0 2500.00 0 2500.00

∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d ∼600d 1000.00 2000.00 5000.00 5000.00 10000.00 10000.00 5000.00 5000.00 10000.00 10000.00

1.0 1.0 0 0.5 0 0.25 0 0.5 0 0.25

a The solution was prepared for leaching. b The solution was prepared for synthesizing geopolymer. c Molar Na/K ) 0.2. d The solution was adjusted to pH ) 13.95.

(3Al2O3‚2SiO2), hematite (Fe2O3), and magnetite (Fe3O4), with minor proportions of lime (CaO) and gypsum (CaSO4‚2H2O). By using the method of known addition,15 the weight fractions of the crystalline R-quartz and mullite were determined to be 4.2 and 12.3 mass %, respectively. Using a simple mass balance, the amorphous/glassy aluminosilicate content of the fly ash was approximately 70 mass %, assuming all the Fe was in the crystalline phases and lime and gypsum were negligible. As will be shown below, the reactivity of the crystalline aluminosilicate phases within the fly ash was negligible as compared to that of the amorphous counterpart(s). The Brunauer-Emmett-Teller (BET) surface area of the fly ash, as determined by nitrogen adsorption on a Micromeritics ASAP2000 instrument, is 4.13 m2/g, and the mean particle size (d50) is 14.1 µm. According to the ASTM definition, the Gladstone fly ash should be classified as Class F due to the relatively low calcium content. Sodium silicate solution (Vitrosol N(N40), molar ratio SiO2/Na2O ) Rm ) 3.32, [SiO2] ) 6.63 M) and potassium silicate solution (KASIL 2236(K32), molar ratio SiO2/K2O ) Rm ) 3.50, [SiO2] ) 5.30 M) were obtained from PQ Australia. Laboratory-grade reagents (NaOH and KOH) were obtained from Ajax Chemicals Australia. Leaching Experiments. Leaching was performed to accurately measure the extent of fly ash dissolution as a result of alkali-activation. In each leaching experiment, 100 g of the predried fly ash was added to 1000 mL of the leaching solution (or the activating solution). All leaching tests were run in parallel at 20 ( 0.5 °C to minimize experimental errors. The activating solutions were prepared by dissolving calculated amounts of the silicate solution (Vitrosol N(N40) or KASIL 2236(K32)) and ROH (R ) Na and/or K) in distilled/deionized water. The final compositions of the activating solutions are presented in Table 1. A glass electrode was used to measure and adjust the solution pH to ∼13.95 for the leaching experiments of low alkali (15) White, S. C.; Case, E. D. J. Mater. Sci. 1990, 25, 5215.

8728

Langmuir, Vol. 19, No. 21, 2003

Lee and van Deventer

Figure 1. Schematic drawing of the procedures used for separating the alkali-activated fly ash from the activating solutions. concentration. The pH was corrected by taking into account the alkali cations present in the solutions. At the designated time interval, 10 mL of the suspensions was collected until 168 h of leaching. The alkali-activated solids and the respective solutions were separated using the procedures described previously,16 which are briefly summarized in Figure 1. The compositions of the leached solutions, or the clear solutions of the leaching suspension, were determined by inductively coupled plasma equipped with optical emission spectroscopy (ICP-OES) using a Perkin-Elmer 3000 instrument. Note that the solutions analyzed in this work should include all soluble silicates of monomeric and oligomeric silicates and may also contain colloidal polysilicates. This paper does not intend to differentiate the speciation of soluble silicates in the solutions. Geopolymer Synthesis. The geopolymers were synthesized by first preparing the activating solutions, whose compositions are shown in Table 1, using calculated amounts of the sodium silicate solution (Vitrosol N(N40)), NaOH, KOH, and distilled/ deionized water. After 500 g of the fly ash was placed in a mixing bowl, 250 g of the activating solution was added. The contents of the mixing bowl were mixed in a mechanical mixer for 5 min to produce a homogeneous paste. These pastes were then transferred to alkali-resistant plastic containers and cured at 20 ( 3 °C and a humidity of 50 ( 5% under atmospheric pressure. Compressive strength tests were conducted on an ELE compression machine using a pace rate of 1 kN/s, conforming to ASTM C39. The cylindrical sample dimension was 50 mm diameter × 100 mm. All the values collected were the averages of three separate tests, with a standard deviation of less than 5%. Characterization. The alkali-activated fly ash and geopolymers were analyzed by powder XRD analysis (Phillips PW 1800) using Cu KR radiation, which was generated at 30 mA and 40 kV with an averaged wavelength of 1.54184 Å. The samples were step-scanned at 0.05° 2θ and integrated at the rate of 2 s step-1. Fourier transform infrared (FTIR) spectra of the alkali-activated fly ash and geopolymers were collected in a Bio-Rad FTS 165 FTIR spectrometer in absorbance mode. Pellets of samples were prepared using the normal KBr procedure, dried in an oven at 60 °C overnight, and pressed again before spectra were taken. All spectra were obtained with a sensitivity of 8 cm-1, and 64 scans per spectrum were taken. Spectral decomposition (or deconvolution) of the complex bands collected over the 8001350 spectral region was performed by using the CurveFit function of the BioRad Win-IR program (version 4.14, level II, Galactic Industries Corp.). The deconvolution procedure used in this work was in accordance with that described previously.17 Briefly, the starting calculation parameters for the curve fitting process, such as the number of component bands, band positions, width at half-maximum, relative intensities, and shape, were predetermined by combined procedures of the secondary derivative method and Fourier self-deconvolution.18 The accurate (16) Lee, W. K. W.; Van Deventer, J. S. J. Colloids Surf., A 2002, 211, 49.

Figure 2. The silicon and calcium leaching characteristics of systems Q and U. [Si] ) concentration of Si in the solution at time t. [Si]0 ) concentration of Si at the start of leaching. The notations are the same for Ca. positions of the band maxima were determined by the second derivative of the spectra at the inflection points. Similar information and other calculation parameters were obtained by the deconvolution process. The actual curve fitting process was performed using the Levenberg-Marquardt algorithm from the BioRad Win-IR trn program to iteratively minimize the weighted difference between the actual (or the curve fitted) and the measured data, with the aim to minimize the number of component bands.

Results and Discussion Alkali-Activation. Figure 2 shows a typical leaching curve of the Gladstone fly ash. It is clear that the addition of soluble silicates to the activating solution at pH ∼ 13.95 (system U) could induce significant changes to the fly ash dissolution/reprecipitation characteristics as compared to the system without soluble silicates (system R). [Si]0 denotes the initial soluble silicate concentration in the activating solution. A positive ([Si] - [Si]0) value means the overall process is dissolution and vice versa. The effects of soluble silicates on the extent of apparent Al dissolution can be divided into three major intervals as shown in Table 2. At 0 e [Si]0 e 14.24 mM (systems A, B, I, J, Q, and R), the presence of soluble silicates increased the maximum apparent dissolution of Al. At 14.24 < [Si]0 < 213.60 mM (systems C-E, K-M, and S-T), the maximum apparent Al dissolution decreased with increasing soluble silicate dosage. If the soluble silicate dosage was increased still further ([Si]0 g 213.60 mM, systems F-H, N-P, and T-U), the maximum apparent Al dissolution was found to increase again with increasing soluble silicate dosage. The maximum apparent Al dissolution was observed at the end of each leaching experiment at 168 h. No Mg dissolution was observed at all except when the activating solutions were highly concentrated in alkali and soluble silicate (systems W and Y). The dissolution characteristics of Si and Ca were found to be very different from those of Al. See Table 2. The (17) Handke, M.; Mozgawa, W.; Nocun´, M. J. Mol. Struct. 1994, 325, 129. (18) Griffiths, P. R.; Pariente, G. L. Trends Anal. Chem. 1956, 5, 209.

Geopolymerization of Amorphous Aluminosilicates

Langmuir, Vol. 19, No. 21, 2003 8729

Table 2. Maximum Apparent Dissolution Observed from Gladstone Fly Ash in the Activating Solutions at 20 °C maximum apparent dissolution (mM)a system

Al

A 7.0 (168)a B 9.1 (168) C 7.3 (168) D 1.7 (168) E 0.6 (97) F 1.1 (26) G 2.3 (10) H 3.5 (8) I 6.9 (168) J 9.6 (168) K 7.7 (168) L 2.2 (168) M 0.9 (168) N 0.9 (48) O 1.9 (48) P 3.7 (8) Q 6.9 (168) R 9.6 (168) S 7.8 (168) T 0.9 (48) U 3.7 (8) V 13.4 (48) W 52.3 (72) X 53.0 (168) Y 164.6 (120) Z 78.2 (168) AA 123.5 (168) theoretical max 375.4 dissolutionc

Si 4.6 (168) 0b 0b 0b 0b 0.7 (26) 20.5 (10) 154.9 (8) 4.5 (168) 0b 0b 0b 0b 6.05 (8) 16.8 (8) 167.3 (8) 4.5 (168) 0b 0b 6.1 (8) 154.9 (8) 533.8 (48) 623.1 (48) 81.2 (168) 691.4 (2) 119.3 (168) 692.2 (2) 704.6

Ca

Mg

1.1 (2) 0 0.2 (168) 0 0.2 (168) 0 0.1 (168) 0 0 0 2.3 (26) 0 15.8 (10) 0 30.2 (8) 0 0.8 (2) 0 0.1 (24) 0 0.1 (48) 0 0.1 (168) 0 0 0 3.8 (8) 0 11.1 (8) 0 23.3 (8) 0 0.8 (2) 0 0.1 (24) 0 0.1 (48) 0 3.8 (8) 0 23.3 (8) 0 60.9 (8) 2.0 (8) 58.6 (8) 2.3 (8) 0 0 49.2 (120) 9.3 (120) 0 0 53.2 (168) 8.5 (168) 62.1 32.8

a The parenthesis denotes the time of leaching in hours when the value of the maximum apparent dissolution was taken. The maximum apparent dissolution was obtained by subtracting the initial concentration from the elemental concentration at the designated time, i.e., ([Si] - [Si]0). b No apparent dissolution but only precipitation was observed. c Values were obtained by assuming R-quartz and mullite were insoluble due to the fact that the intensities of the XRD peaks for these crystalline phases were relatively unchanged before and after the alkali-activation.

presence of soluble silicates in the initial activating solution evidently decreased the extents of Si and Ca dissolutions when less than 213.60 mM soluble silicates were used (systems B-E, J-M, and R-S). In these systems, the time when the maximum apparent dissolution of Si and Ca was observed was at the end of the leaching experiment at 168 h. If, however, more than ∼200 mM soluble silicates were used (systems F-H, N-P, and T-U), both the maximum apparent dissolutions of Si and Ca were found to increase with increasing soluble silicate dosage. At [Si]0 ) 569.60 mM (systems H, P, and U), the maximum apparent dissolutions of Si and Ca, which occurred at approximately 12 h, were significantly greater than those without soluble silicates (systems A, I, and Q). See Figure 2. After the maximum dissolution was achieved, Si (including the dissolved and the added soluble silicates) and Ca then precipitated out as secondary phase(s), the identification of which is out of the scope of this paper. From the above, it can be concluded that addition of soluble silicates to an activating solution at pH ∼ 13.95 could be effective in promoting fly ash dissolution if the soluble silicate dosage was over ∼200 mM. Similar trends could also be observed when the alkali concentration of the activating solution was increased. The maximum apparent dissolutions of Si and Ca for system Y ([MOH]0 ) 5 M, [Si]0 ) 2.5 M) and system AA ([MOH]0 ) 10 M, [Si]0 ) 2.5 M) were significantly greater than those of system X ([MOH]0 ) 5 M, [Si]0 ) 0 M) and system Z ([MOH]0 ) 10 M, [Si]0 ) 0 M), respectively. See Table 2.

Figure 3. The XRD diffractograms of the untreated fly ash (Gdf) and the various alkali-activated solids (systems Q and U, Geo-II, and Geo-IV) after 168 h of activation.

Figure 4. The FTIR spectra of the untreated fly ash (Gdf) and the various alkali-activated solids (systems Q and U, Geo-II, and Geo-IV) after 168 h of activation.

Figure 3 shows that regardless of the activating solutions used, the alkali-activating process employed in this work did not cause any observable changes to the XRD diffractograms of the fly ash. No new crystalline peaks could be identified, and the peak intensity remained unchanged before and after the activation. Hence, it can be assumed that the action of the alkali-activation was only involved with the amorphous/glassy phase(s) of the fly ash throughout the duration of the present work. Furthermore, the products of alkali-activation and/or geopolymerization were X-ray amorphous, as also reported by others.1-3,6,8 The IR spectra of some representative samples of the alkali-activated fly ash are presented in Figure 4. The major IR vibrational bands were assigned and summarized in Table 3 by referring to numerous previous studies.17,19-27 (19) Ghosh, S. N. J. Mater. Sci. 1978, 13, 1877. (20) Farmer, V. C. The Infrared Spectra of Minerals; Mineralogical Society: London, 1974. (21) Gadsden, J. A. Infrared Spectra of Minerals and Related Inorganic Compounds; Butterworth: London, 1975. (22) Vempati, R. K.; Rao, A.; Hess, T. R.; Cocke, D. L.; Lauer, H. V., Jr. Cem. Concr. Res. 1994, 24, 1153. (23) Mollah, M. Y. A.; Hess, T. R.; Cocke, D. L. Cem. Concr. Res. 1994, 24, 109. (24) Uchino, T.; Sakka, T.; Iwasaki, M. J. Am. Ceram. Soc. 1991, 74, 306. (25) Uchino, T.; Sakka, T.; Iwasaki, M. J. Am. Ceram. Soc. 1989, 72, 2173. (26) Poe, B. T.; McMillan, P. F.; Angell, C. A.; Sato, R. K. Chem. Geol. 1992, 96, 333. (27) Sitarz, M.; Mozgawa, W.; Handke, M. J. Mol. Struct. 1997, 404, 193.

8730

Langmuir, Vol. 19, No. 21, 2003

Lee and van Deventer

Table 3. Characteristic IR Vibrational Bands of the Gladstone Class F Fly Ash wavenumber (cm-1)a 950-1250 (s) 1165 (sh) 1115-1140 (sh) 1077 (s) 950-980 (sh) 882 (s) 798 (m) 727 (sh) 620 (sh) 561 (s) 466 (s)

assignmentb asymmetric stretching (Si-O-Si and Al-O-Si) asymmetric stretching (Si-O-Si) asymmetric stretching (Si-O-Si and Al-O-Si) asymmetric stretching (Si-O-Si and Al-O-Si) Si-O stretching (Si-O-R+) Si-O stretching and OH bending (Si-OH) symmetric stretching (Si-O-Si) symmetric stretching (Si-O-Si and Al-O-Si) symmetric stretching (Si-O-Si and Al-O-Si) symmetric stretching (Al-O-Si) bending (Si-O-Si and O-Si-O)

references 19-21 20-24 21, 22 21, 22 24 25 21, 23 17, 20, 26, 27 20, 21, 26 21, 26 17, 19-27

a The abbreviations in parentheses are as follows: s ) strong, m ) medium, and sh ) shoulder. b R ) Na and K.

Figure 6. The FTIR spectral deconvolution of the system-Uactivated fly ash after 168 h of activation.

Figure 5. The FTIR spectral deconvolution of the untreated fly ash.

Figure 7. The FTIR spectral deconvolution of Geo-II after 48 h of activation.

It can be seen that with limited apparent fly ash dissolution, such as system Q, the IR spectrum of the alkali-activated fly ash at the end of the leaching experiment was almost identical to that of the unreacted fly ash (Gdf). On the other hand, significant spectral differences could be observed from the solids that had demonstrated significant apparent dissolution such as system U (Table 2). The vibrational band with a maximum at 1077 cm-1 was found to shift toward a lower wavenumber in the system-U-activated fly ash. This indicates that the original silicate and/or aluminosilicate structures in the fly ash had been significantly depolymerized.23,28,29 The significant reduction in intensity of the spectral shoulder at ∼560 cm-1, which is attributable to silicate and/or aluminosilicate glasses with long-range structural order such as rings of tetrahedra and/or octahedra,17,26,27 further suggests that the system U activating solution was effective in destroying the glassy content(s) of the fly ash. The results of the IR analysis on the activated fly ash are consistent with those obtained from the ICP-OES analysis. Figure 5 includes the result of the spectral deconvolution of the unreacted fly ash. It is clear that one principal band

and several satellite bands should comprise the more complex original band over the 800-1350 cm-1 region. The 1077 cm-1 principal band should be assigned to the T-O-Si (T ) Si or Al) asymmetric stretching vibration within the TO4 tetrahedra typical of an amorphous aluminosilicate glass.22 The three satellite bands at 1203, 1167, and 1117 cm-1 could be the spectral shoulders of the principal band at 1077 cm-1, which are typical in silicate and aluminosilicate glasses24,26 and could be attributed to different structural units within the same silicate structure.25 They could also be a result of the combination of the spectral shoulders of the 1077 cm-1 principal band and the bands attributable to R-quartz and mullite. The low abundance of R-quartz (4.2 wt %) and mullite (12.3 wt %) in the Gladstone fly ash, as determined by quantitative XRD, and the low sensitivity used during spectrum acquisition of this work may altogether prevent these satellite bands from being completely resolved. By comparison of the result of spectral deconvolution of the unreacted fly ash (Figure 5) with those of the various alkali-activated fly ash and geopolymers (Figures 6-8), it is clear that changes to the satellite bands were relatively small. Only the position of the principal band located originally at 1077 cm-1 was found to shift significantly due to alkali-activation. From previous research,14 it was found that as the concentration of alkali metal oxide increased in a silicate

(28) Ortego, J. D.; Barroeta, Y.; Cartledge, F. K.; Akhter, H. Environ. Sci. Technol. 1991, 25, 1171. (29) Yu, P.; Kirkpatrick, R. M.; Poe, B.; McMillan, P. F.; Cong, X. J. Am. Ceram. Soc. 1999, 82, 742.

Geopolymerization of Amorphous Aluminosilicates

Figure 8. The FTIR spectral deconvolution of Geo-IV after 48 h of activation.

structure, a fully polymerized tetrahedral framework would become increasingly “depolymerized”, a result of an increased number of NBOs in the structure. The bonding between the TO4 tetrahedral units in this case should become more ionic at a higher NBO concentration. Consequently, the T-O bonds in the tetrahedra should possess smaller molecular vibration force constants.11 Therefore, the T-O-Si asymmetric stretching vibration band should appear at a lower wavenumber with an increasing alkali metal oxide inclusion and/or a greater extent of silicate depolymerization. Numerous previous studies have experimentally proved this point.10-14 The materials used for such studies include binary and ternary alkali and alkaline earth silicate glasses and calcium silicate hydrates. It has also been suggested that a similar trend should apply to aluminosilicates and other mineral silicates. As pointed out earlier, the effect of alkali-activation on aluminosilicates, from a structural point of view, is to introduce alkali cations to the existing aluminosilicate structure while the framework is being destroyed through OH--promoted hydrolytic reactions (see Scheme 1).8 The product of alkali-activation, therefore, should exhibit a decreased Al/R (R ) Na or K) ratio. From the literature,9,30 an aluminosilicate with Al/R ) 1 (the amount of alkali equals the amount of the negatively charged and tetrahedrally coordinated Al) was reported to contain no NBOs. NBOs, however, were generated if Al/R was less than 1. Hence, it can be anticipated that with increasing extent of alkali-activation, the NBO concentration of the aluminosilicate (assume that the Al is in fourth coordination) should also increase. By considering the results of previous research,10-14 it is anticipated that an increase in the extent of alkali-activation should move the T-O-Si asymmetric stretching vibration band to a lower energy field. Figure 9 shows that the frequency shift of the T-O-Si asymmetric stretching maximum of the IR principal band of the fly ash was approximately linear to the extent of alkali-activation of the glassy content within the fly ash. The greater the extent of activation, the lower the wavenumber of this principal band maximum. This is consistent with the hypothesis raised earlier and indicates that the assignment of the T-O-Si asymmetric stretching vibration “principal” band to the glass phase(s) of the fly ash is correct. The values of the principal band maxima (30) Onorato, P. I. K.; Alexander, M. N.; Struck, C. W.; Tasker, G. W.; Uhlmann, D. R. J. Am. Ceram. Soc. 1985, 68, C148.

Langmuir, Vol. 19, No. 21, 2003 8731

Figure 9. The calibration curve of the position of the T-O-Si asymmetric stretching principal band versus the extent of alkaliactivation for systems A-H, I-P, Q-U, V-Y, and AA after 168 h of activation.

Figure 10. The position of the T-O-Si stretching vibration principal band versus the soluble silicate dosage of systems A-H and I-P after 168 h of activation.

were obtained by performing a spectral deconvolution process similar to that used for obtaining Figures 5-8, and the extent of activation was calculated as [100 × (maximum apparent Si dissolution)/(theoretical maximum Si dissolution)], where the values were obtained from the leaching experiment. See Table 2. From Figure 9, the exact position of the principal band at a certain extent of activation was unaffected by the type of alkali cation in the starting activating solution. Therefore, Figure 9 can be used as a calibration curve for estimating the extent of alkali-activation of the amorphous/ glassy content within the fly ash, regardless of the type of the alkali(s) present in the solution. This is despite the fact that at the same alkali concentration and soluble silicate dosage, the Na-activating solutions (systems A-H) were more effective in destroying the amorphous/glassy content of the fly ash than the K-counterparts (systems I-P). See Table 2 and Figure 10. This implies that less NaOH, or sodium silicate solution, was required to achieve the same extent of activation as the K-counterpart(s). Table 2 shows that almost complete Si dissolution was possible in systems Y and AA. The time taken to achieve maximum apparent Si dissolution was similar (within 2 h) for both systems even when the alkali concentration used was significantly different. This suggests that these systems (solid/liquid ratio ) 0.1) could be saturated with the activator, the hydroxyl ions. See Table 1 for the compositions of the activating solutions. On the other

8732

Langmuir, Vol. 19, No. 21, 2003

Figure 11. The position of the T-O-Si stretching vibration principal band and the extent of activation versus the duration of activation for system-U-activated fly ash and Geo-II and Geo-IV.

hand, if the solid/liquid ratio was increased to 2 (Geo-II and Geo-IV), which is typical for the synthesis of geopolymers, while fixing the total alkali concentration and the Na/K ratio in the activating solutions, very different activation characteristics could be observed. See Figure 11. Geo-IV attained complete dissolution in ∼48 h. On the other hand, Geo-II had reached 80% activation in ∼5 h but the activation was terminated thereafter. The obvious differences between Geo-II and Geo-IV perhaps could be explained by considering the relative concentrations of the solid and hydroxyl contents in these systems. Geo-IV could be saturated, or even oversaturated, with the activator, while Geo-II was undersaturated. The fact that the activation curves (Figure 11) for both Geo-II and GeoIV were almost the same from the start of the activation up to ∼5 h suggests that (1) the alkali concentration within the concentration range 5 M e [OH-] e 10 M did not affect the kinetics of the alkali-activation process(es) and (2) the alkali-activation process(es) could be the same in both geopolymeric systems until the activator was run out, as in the case of Geo-II. It had been stated previously that the first step of sodium silicate glass corrosion was the diffusion of Na+ ion from the glass into the solution through an ion-exchange process (the Si-O-Na+ bond was replaced by Si-O-H+) before the silicate structural backbone was destroyed.13 This was reflected in the IR analysis. The Si-O asymmetric stretching vibration band was found to first shift to a higher frequency field before shifting to the lower energy end as the corrosion continued. As the first step of alkaliactivation was also the diffusion of Ca from the fly ash into the activating solution (see Figure 2), similar processes could also occur in the alkali-activation as demonstrated by system U in Figure 11. The T-O-Si asymmetric stretching principal band was found to shift to a higher energy field at the start of the activation before moving to the lower energy end, as in system U and in the other systems activated with relatively greater dosages of soluble silicates (Geo-II, Geo-IV, and systems F-H, N-P, T-W, Y, and AA). On the other hand, if no or little dosages of soluble silicates were used (Geo-I, Geo-III, and systems A-E, I-M, Q-S, X, and Z), the shift of the T-O-Si asymmetric stretching principal band to a lower frequency

Lee and van Deventer

Figure 12. The extent of activation and the compressive strength versus the duration of activation for Geo-IV.

field was not observed. It is therefore possible that without significant soluble silicate addition, major structural breakdown was inhibited after the initial ion-exchange process. Polymerization/Solid-State Transformation. Figure 12 shows that Geo-IV had already been significantly activated (∼65% of activation) even before setting, which denotes the state of a paste that is being transformed from fluidity to rigidity. After setting, the compressive strength and the extent of activation were clearly not related. The geopolymer continued to gain strength while no further dissolution of the fly ash was observed. This result is not surprising as the extent of activation measures the degree of “depolymerization” of the original amorphous/ glassy content(s) whereas the compressive strength is an indication of the degree of “polymerization” of the newly formed aluminosilicate gel. A careful examination of the results of the IR spectral deconvolution (Figure 8) reveals that a new band at 1107 cm-1 could be identified in GeoIV. A similar band was also observed at 1102 cm-1 in Geo-II (Figure 7) as well as in the geopolymers synthesized from metakaolin.32 In the present work, this band was found to grow in intensity with age without significant band shift. From previous work,33 this new band can be assigned to the Si-O asymmetric stretching vibration attributable to a highly polymerized sheet and/or framework structure, which in nuclear magnetic resonance (NMR) terminology can be denoted as Q3 and Q4, respectively (meaning the tetrahedral silicon is surrounded by 3 BOs for Q3 and 4 BOs for Q4). Because geopolymers produced from metakaolin have been proven previously to possess a Q4 structure1,32 and the ∼1105 cm-1 new IR band is common in Geo-II, Geo-IV, and the geopolymers synthesized from metakaolin,32 the new phase found in Geo-II and in Geo-IV can be denoted as a Q4 structure. An attempt was thus made to correlate the observed compressive strength of Geo-II and Geo-IV with the relative intensity of this new IR band. The results are summarized in Figure 13, where Inew is the intensity of the newly formed and highly polymerized Q4 T-O-Si asymmetric stretching vibration band at ∼1105 cm-1, and (31) Sˇ imon, I.; McMahon, H. O. J. Am. Ceram. Soc. 1953, 36, 160. (32) Rahier, H.; Simons, W.; Van Mele, B.; Biesemans, M. J. Mater. Sci. 1997, 32, 2237. (33) Furukawa, T.; Fox, K. E.; White, W. B. J. Chem. Phys. 1981, 75, 3226.

Geopolymerization of Amorphous Aluminosilicates

Figure 13. The compressive strength versus the relative intensity of the new phase at ∼1105 cm-1 for Geo-II and GeoIV. The numbers within the figure denote the duration of activation when the alkali-activated fly ash was taken for IR analysis.

Iprincipal is the intensity of the T-O-Si asymmetric stretching vibration principal band that was found to shift with the extent of activation. From Figure 13, the Q4 T-O-Si asymmetric stretching band of both Geo-II and Geo-IV was found to grow in intensity while little physical strength was observed before setting. Both mixtures obtained setting when the Inew/ Iprincipal ratio of ∼0.13 was reached at ∼2 h. After setting, it seems that as the intensity of the Q4 T-O-Si asymmetric stretching band became stronger, the geopolymers also became physically stronger. This is true for Geo-II throughout the entire investigation as well as for the early stage of Geo-IV. In fact, when the Inew/Iprincipal ratio was the same, both of the geopolymers appeared to possess similar compressive strength. Geo-IV had a higher strength than Geo-II at the end of this investigation, possibly because more Q4 phase was formed in Geo-IV after ∼7 h. The Q4 new phase continued to form thereafter until 100% activation was reached at ∼48 h (Figure 11). If it is assumed that the new phase of the new Q4 T-OSi asymmetric stretching band was responsible for the early strength of Geo-II and Geo-IV, the “extra” strength of Geo-IV after 48 h must come from other source(s). Figure 11 shows that after 48 h, and hence after 100% activation was achieved, the T-O-Si asymmetric stretching principal band of Geo-IV, which is attributable to the hydrolyzed glassy content of the fly ash, then began to shift back to a higher energy field. This means that the once-depolymerized species due to alkali-activation was somehow triggered to repolymerize again. The increased polymerization could be responsible for the extra strength observed in Geo-IV after ∼48 h. It is possible that as time went on, excessive water was evaporated from the geopolymer. The solubility limit of the dissolved species was thus exceeded once again and ultimately caused further precipitation to enhance the strength of geopolymers. This, however, does not explain why the repolymerization process did not occur in Geo-II and thoroughly deserves future investigation. Nevertheless, the solution used for activating Geo-IV is far more efficient than GeoII in promoting (1) fly ash dissolution (Figure 11), (2) formation of a highly polymerized Q4 phase (Figure 13), and (3) repolymerization of the once-depolymerized phase. Finally, it is important to note that the Q4 T-O-Si asymmetric stretching band of Geo-II and Geo-IV is

Langmuir, Vol. 19, No. 21, 2003 8733

different in nature from the high-frequency principal band of Geo-I and Geo-III as shown in Figure 4. The principal band stayed at 1082 and 1091 cm-1 for Geo-I and Geo-III, respectively, even after 672 h of activation. Though highly polymerized, these phases did not contribute to significant physical strength of the geopolymers. Both systems only attained a compressive strength of less than 2 MPa at the end of the investigation. The Role of Soluble Silicates in Geopolymerization. From the above, the reactions involved in the alkaliactivated systems can be categorized into two major groups. Those that were activated by solutions containing little or no soluble silicates were clearly different from those activated by large dosages of soluble silicates. It had been shown previously that when a sodium alkaline solution containing no soluble silicates was used, the reacted metakaolin and kaolinite contained both the amorphous aluminosilicates and hydroxysodalite crystallite.32,34 Though no quantitative information was given, it seems the crystalline content was significant. If a potassium alkaline solution was used instead, the crystalline reaction product was illite.35 It is therefore expected that reacting aluminosilicate solids, crystalline or amorphous, with an alkaline solution without soluble silicates should give rise to a reaction mixture of aluminosilicate crystallite(s) as well as amorphous aluminosilicate(s). In this work, only highly polymerized and X-ray amorphous phases were identified in the fly ash activated by solutions containing little or no soluble silicates (Geo-I and GeoIII). See Figures 3 and 4. It is likely that the highly polymerized phase(s) of Geo-I and Geo-III may be precursors to a crystalline phase(s) such as hydroxysodalite and/ or illite. This will have to be investigated in the future. Nevertheless, the analytical procedures developed in this work showed that leaching of alkali and/or alkaline earth cations such as Ca from the initial aluminosilicate was responsible for the formation of the highly polymerized structure with decreased NBO concentration. This structure, however, demonstrated little interparticle strength that is crucial for the macroscopic strength of the hardened paste. On the other hand, when a significant amount of soluble silicate was used (Geo-II and Geo-IV), the early geopolymerization process could be similar to the corrosion of alkali and alkaline earth silicate glasses.13 The initial leaching of alkali and/or alkaline cations from the aluminosilicate solid gave rise to a more condensed structure before the (alumino)silicate backbone was hydrolyzed. The dissolved species then formed a highly polymerized phase, which was believed to be responsible for the early macroscopic (compressive) strength of the paste. Subsequently, the less polymerized phase attributable to the hydrolyzed aluminosilicate solid was found to repolymerize after 100% activation was achieved. This repolymerization can be attributed to the later (ultimate) strength of geopolymer. It had been previously reported that when a metakaolin was reacted with alkaline silicate solutions, whose SiO2/R2O ratios were between 0.2 and 0.8, there was a phase separation within the resultant geopolymers.33 A small quantity of crystalline material(s) (not identified) and a highly polymerized phase were found within the mainly amorphous and less polymerized aluminosilicate gels. Since the SiO2/R2O ratio of the activating solutions used in this work for the synthesis of Geo-II (SiO2/R2O ) 0.5) and Geo-IV (SiO2/R2O ) 0.25) was also within the range of 0.2-0.8, it can be concluded (34) Felsche, J.; Luger, S. Thermochim. Acta 1987, 118, 35. (35) Bauer, A.; Velde, B.; Berger, G. Appl. Geochem. 1998, 13, 619.

8734

Langmuir, Vol. 19, No. 21, 2003

that the phase separation should be common for all geopolymers synthesized by alkali-activating various (amorphous) aluminosilicates with alkaline silicate solutions of 0.2 e SiO2/R2O e 0.8. This work shows that a true geopolymerization process, including OH--promoted solid hydrolysis, polymerization of dissolved species, and solid-state transformation, only occurred if an activating solution containing significant levels of soluble silicates (SiO2/R2O > ∼0.2) was used. Conclusion A new analytical procedure was developed in this work to study the alkali-activation and geopolymerization of a typical heterogeneous amorphous aluminosilicate. Infrared spectra and ICP-OES analysis were conducted on the alkali-activated fly ash and the solution obtained from the leaching experiment (solid/liquid ratio ) 0.1), respectively. The results were then used to construct a linear calibration curve correlating the IR frequency shift of the T-O-Si asymmetric stretching vibration band attributable to the amorphous/glassy content of the fly ash, which is a measure of the degree of polymerization of the (alumino)silicate network, and the extent of activation of the amorphous aluminosilicate. As the type of alkali cation(s) present in the activating solutions was found to exert no effects on the positioning of this IR band, this calibration curve can be used to approximate the extent of activation (OH--promoted hydrolysis) of the amorphous/ glassy aluminosilicate in many different applications.

Lee and van Deventer

From the use of the calibration curve and with the aid of IR deconvolution, it was found that at a solid/liquid ratio of 2, geopolymerization was initiated by ion exchange between the aluminosilicate and the activating solution. The subsequent reactions were then controlled by the concentration of soluble silicates. If an activating solution of SiO2/R2O < ∼0.2 was used, a highly polymerized phase with little interparticle bonding was formed. No apparent hydrolysis of the aluminosilicate was observed. Phase separation, on the other hand, was found in the geopolymers activated by solutions of 0.2 e SiO2/R2O e 0.8, which resulted in the formation of a highly polymerized phase and a less polymerized phase. In these geopolymers, the early macroscopic (compressive) strength was attributed to the highly polymerized phase. The degree of polymerization of the less polymerized phase attributable to the once-hydrolyzed aluminosilicate then controlled the later strength. It was found that without achieving 100% activation (or hydrolysis), repolymerization of the hydrolyzed aluminosilicate did not occur. High concentrations of alkalis and soluble silicates were both required to hydrolyze the aluminosilicate to completion. Acknowledgment. The financial support from the Australian Research Council (ARC), the Particulate Fluids Processing Centre (PFPC), and Defor Pty. Ltd. is gratefully acknowledged. LA026127E