Synthesis and Microstructural Characterization of Fully-Reacted Potassium-Poly(sialate-Siloxo)...

(ProQuest: ... denotes formula omitted.)

INTRODUCTION

In recent years, there has been significant development worldwide in a new type of inorganic binder: geopolymeric cement. Geopolymeric cement is one type of three-dimensional (3D) CaO-free aluminosilicate gel binder, which was first introduced into the inorganic binder world by Glukhovsky in the former Soviet Union in the 1950s.1 In France, Davidovits2 also conducted extensive research in the late 1970s. Geopolymeric cement can be synthesized by mixing aluminosilicate reactive materials with no CaO component (such as metakaolin, dehydrated clay, and Class F fly ash) and strongly alkaline solutions (such as NaOH or KOH), and then curing it at room temperature. Under a strongly alkaline solution, aluminosilicate reactive materials are rapidly dissolved into solution to form free SiO^sub 4^ and AlO^sub 4^ tetrahedral units. With the development of reaction, mixture water is gradually split out and these SiO^sub 4^ and AlO^sub 4^ tetrahedral units are linked alternatively to yield polymeric precursors (-SiO^sub 4^-AlO^sub 4^-, -SiO^sub 4^-AlO^sub 4^-SiO^sub 4^-, or -SiO^sub 4^-AlO^sub 4^-SiO^sub 4^- SiO^sub 4^-) by sharing all oxygen atoms between two tetrahedral units, thereby forming monolithic-like geopolymeric products.3-7 According to the molecular structure, geopolymeric cement can be expressed in the following empirical formula3

R^sub n^{-(SiO^sub 2^)^sub z^-AlO^sub 2^-}^sub n^ × wH2O

where R is a cation such as potassium (K) or sodium (Na); n is degree of polycondensation; z is 1, 2, and 3; and w is binding water amount.

Geopolymeric cement made with reasonable mixturedesign and formulation can exhibit superior properties to portland cement; the production of geopolymeric cement requires much lower calcining temperature (1112 to 1472 °F [600 to 800 °C]) and emits 80 to 90% less CO2 than portland cement.3,8 Reasonable strength can be gained in a short period at room temperature. In most cases, 70% of the final compressive strength is developed in the first 12 hours.3,8-10 Low permeability, comparable with natural granite, is another property of geopolymeric cement.3,11,12 It is also reported that resistance to fire and acid attacks for geopolymeric cement are substantially superior to those for portland cement.13,14 Apart from the high early strength, low permeability, and good fire and acid resistance, geopolymeric cement also can attain higher unconfined compressive strength and shrink much less than portland cement.3,12 Other documented properties include good resistance to freezing-and-thawing cycles as well as excellent solidification of heavy metal ions.3,8-18 These properties make geopolymeric cement a strong candidate for substituting portland cement applied in the fields of civil, bridge, pavement, hydraulic, underground, and military engineering.19-20

In this study, a fully-reacted potassium-poly(sialatesiloxo) (K-PSS) geopolymeric cement matrix is attempted to synthesize at room temperature by optimizing three key molar ratios: SiO^sub 2^/Al^sub 2^O^sub 3^, K^sub 2^O/Al^sub 2^O^sub 3^, and H2O/K^sub 2^O. The compressive strength and microstructure of the hardened geopolymeric cement matrixes are evaluated as a function of the three ratios. The influencing extent of each ratio on the compressive strength will be quantitatively determined on basis of the statistical variance analysis. The microstructural changes as a function of ratios will also be investigated by using X-ray powder diffraction (XRD) and Fourier transform infrared spectroscopy (FTIR) techniques. Based on the macroscopic and microscopic experiments, an almost fully-reacted K-PSS geopolymeric cement matrix with the highest strength and optimum microstructure can be obtained by properly adjusting the three molar ratios. Subsequently, the coordination status of two main construction elements (Al and Si), micrographics and chemical compositions of the fully-reacted K-PSS geopolymeric cement matrix are further characterized and examined by using an environment scanning electron microscope equipped with energy dispersion X-ray analysis (ESEM-EDXA), highly sensitive magic angle spinning-nuclear magnetic resonance spectroscopy (MAS-NMR), and transmission electron microscopyelectron diffraction spectroscopy (TEM-EDS) techniques. Finally, a 3D molecular model is proposed for the fullyreacted K-PSS geopolymeric cement matrix based on the aforementioned microanalysis.

RESEARCH SIGNIFICANCE

Geopolymeric cement is a new type of cementitious material, which is quite different from traditional portland cement or alkaline activated slag cement. At present, there are mainly three types of commonly used geopolymeric cement: poly-sialate (PS), poly-sialate-siloxo (PSS), and (poly-sialate-disiloxo) PSDS geopolymeric cement. K-PSS geopolymeric cement shows a great advantage over the other two types such as easy preparation, high strength, good fire resistance, and excellent durability, which makes K-PSS geopolymeric cement a strong candidate for partially substituting portland cement.21 A fully-reacted K-PSS geopolymeric cement matrix, however, is not investigated systematically until now. The microstructure of the fullyreacted K-PSS geopolymeric cement matrix is also not quantitatively characterized. All these limit the commercial application of geopolyemric cement in the fields of civil, bridge, pavement, hydraulic, underground, and military engineering.

MATERIALS AND METHODS

Materials

Metakaolin used in this study is obtained by calcining pure kaolin at 1292 °F (700 °C) for 12 hours. The 27MAS-NMR resonance spectrum of calcined kaolin, that is, metakaolin shows that 95% or greater of six-fold coordinated Al in the origin kaolin has been transformed into a four-fold coordinated one. The chemical compositions and physical properties of metakaolin are listed in Table 1. Analytical grade KOH pellet and potassium silicate solution with the molar ratio of SiO^sub 2^/K^sub 2^O of 3.25 and the solid content of 40% are used as alkaline reagents. Silica fume with 95% or greater of SiO^sub 2^ content as silicon additive and analytical grade NaAlO^sub 2^ as aluminium additive is used to compensate the shortage of silicon or aluminium in metakaoline when preparing the geopolymeric cement matrixes with different SiO^sub 2^/Al^sub 2^O^sub 3^ ratios. Distilled water is used throughout the experiment.

Methods

Specimen preparation for mechanical test-According to Table 2, various reactants were first weighted out. For the formula with high SiO^sub 2^/Al^sub 2^O^sub 3^, silica fume was used to compensate the shortage of silicon in metakaoline. For the one with low SiO^sub 2^/Al^sub 2^O^sub 3^, NaAlO^sub 2^ was used to compensate the shortage of aluminium in metakaoline. After that, KOH, potassium silicate solution, and water were mixed in a plastic beaker and cooled down to room temperature, and then metakaolin powder was slowly added into the previous premix alkaline solution and mixed for 3 minutes. After that, the fresh geopolymeric cement matrix was rapidly poured into cubic steel mold of 1.575 x 1.575 x 1.575 in. (40 x 40 x 40 mm). All samples were vibrated for 2 minutes on the vibration table and set at room temperature. To prevent the evaporation of mixture water, the specimens were covered by plastic film during the setting and hardening process. The mold was removed after 24 hours. The demolded specimens were cured at 68 °F (20 °C) and 95% relative humidity (RH) for 28 days. At least six specimens were made for each formula.

Herein it is important to point out that premixing the KOH and potassium silicate solution, rather than directly adding KOH pellets into the geopolymeric cement matrix, was a key to ensure a long enough pot life to complete the previous process. Otherwise, the setting process of geopolymeric cement would become too rapid to complete the previous preparation.

Compressive test

Compressive strength testing was performed according ASTM C39-96. At least six samples of each formula were tested. The average value was served as the ultimate compressive strength. A closed-loop servo-hydraulicallycontrolled materials testing machine was used to conduct a compressive test. The loading was displacement controlled at a constant rate of 0.0512 in./minute (1.3 mm/minute) for all the compression tests.

Sample preparation for microanalysis

The fragments from different formulas were collected after compressive tests. Some of them were used to conduct ESEM-EDXA analysis. The others were further finely crushed and then immersed in an ethanol for 3 days to stop the further geopolymerization reaction, subsequently ovendried at 140 °F (60 °C) for 6 hours. After that, these fractions were further ground into very fine particles with agate mortar, and the coarse particles were removed with a sieve 80 µm in diameter. The fine powders were placed in desiccators for 24 hours.

FTIR analysis

A Fourier transform infrared spectroscopy (FTIR) was performed using KBr pellet techniques (0.0000066 lb [3 mg]) of previously prepared fine powders homogenously ground with 0.00066 lb (300 mg) of KBr powder together until the mixture had the consistency of fine flour, and then pressed into a disk). The resolution and number of scans used in FTIR collection were 0.79 in.^sup -1^ (2.0 cm -1) and 16, respectively.

XRD analysis

XRD was recorded on a spectrometer with the following testing parameters: 40 kV, 30 mA, and CuKa radiation. The XRD patterns were obtained by a scanning rate of 1 degree per minute from 10 to 80 degrees (2?) and steps of 0.05 degrees (2?). The wavelength selected was 6.065 × 10-7 in. (15.40562 nm) (Cu).

ESEM-EDXA analysis

XL30-ESEM equipped with EDXA was used to characterize the microstructure and chemical compositions. The following test parameters were employed: an accelerating voltage of 20 kV, pressure and a RH of 4.2 Torr and 80%, respectively.

TEM-EDS analysis

Geopolymeric cement paste powder was submitted to investigation by TEM fitted with an X-ray microanalysis system at the accelerating voltage of 200 kV. The X-ray counts were obtained by integrating X-ray peaks using an EDS microanalyzer.

MAS-NMR analysis

An NMR spectroscopy is measured using the MAS technique. Powder samples were first placed in a cylinder capsule made with Zr ceramic, and then the capsule was inserted into a sample hole. Compressed air was employed to drive the spinning of the capsule by blowing its bottom. In this study, the spinning speed of the capsule was 4.8 kHz. ^sup 29^Si and ^sup 27^Al spectra were measured and recorded at ^sup 29^Si NMR frequency of 79.3 MHz and ^sup 27^Al frequency of 72.18 MHz. The spectrometer was interfaced with a computer and equipped with a MAS broad-band probe for the solid state experiments. Due to the broad features of the ^sup 29^Si NMR spectrum, it is very difficult to successfully deconvolute the broad unresolved peak of geopolymeirc gels. To address it, Gaussian peak deconvolution, similar to the reports of literature,22 was employed in this study to separate and quantify SiQ^sup 4^(mAl) units with peak positions adopted from extensive studies of geopolymers.3

Chemical shifts were referenced to external TMS in CDCl3 for the ^sup 29^Si nuclei and to external aqueous AlCl3 for the ^sup 27^Al nuclei. The chemical shift values were calculated by taking the midpoint of the signal at half height. The accuracy was 0.5 ppm. Typically, ^sup 29^Si spectra were obtained over a spectra width of 50.0 kHz (acquisition time = 0.04 seconds), with 8192 scans and a relaxation delay of 15 seconds. The ^sup 27^Al spectra were acquired over a spectra width of 41.67 kHz (acquisition time = 0.04 seconds), with 2048 scans and a relaxation delay of 0.5 seconds.

PREPARATION OF FULLY-REACTED KPSS GEOPOLYMERIC MATRIX

Experiential program and results

The main factors influencing the mechanical strength and microstructure of the hardened geopolymeric matrixes include the molar ratio of Si to Al in the molecular chains, the type and amount of alkaline activator, the water content, curing regime (curing temperature and curing age), and the reactivity of aluminosilicate materials.23,24 In this paper, the synthesis of an almost fully-reacted K-PSS type of geopolymeric matrix with high mechanical strength and desirable microstructure at room temperature was attempted. Metakaolin was used as the only reactive alumniosilicate source. KOH and potassium silicate solution were used as alkaline activator. The curing regime was fixed at room temperature and a 28-day curing period. As a result, the effects of the three key molar ratios, SiO^sub 2^/ Al^sub 2^O^sub 3^ (to study the effect of Si-to-Al ratio), K^sub 2^O/Al^sub 2^O^sub 3^ (to study the effect of amount of alkaline activator), and H2O/K^sub 2^O (to study the effect of water content) on mechanical strength and microstructure were mainly systematically investigated.

The main objective of the study was to investigate the effects on mechanical strength and microstructure of three factors: SiO^sub 2^/Al^sub 2^O^sub 3^ (Factor A), K^sub 2^O/Al^sub 2^O^sub 3^ (Factor B), and H2O/K^sub 2^O (Factor C). The three molar ratios were varied within the following ranges, respectively: 3.5 = SiO^sub 2^/Al^sub 2^O^sub 3^ = 4.5, 0.8 = K^sub 2^O/Al^sub 2^O^sub 3^ = 1.2, and 6.0 = H2O/K^sub 2^O = 7.0. The levels for each of the factors were set at three grades: low, intermediate, and high. Thus, the experimental program herein was a three-factor experiment with three levels for each factor, as shown in Table 3. Based on the factor-level table, a total of nine experimental formulas was designed according to a L9(34) orthogonal design principle, as shown in Table 3. The corresponding compressive strength of each formula was also given in Table 3.

Optimization of mixture proportions of K-PSS geopolymeric matrix

Based on the compressive strength, XRD and IR spectra of each geopolymeric formula, the effects of the three molar ratios (SiO^sub 2^/Al^sub 2^O^sub 3^, K^sub 2^O/Al^sub 2^O^sub 3^, and H2O/K^sub 2^O) were quantitatively determined by using the gradation analysis and variance analysis.25 Subsequently, an almost fullyreacted K-PSS geopolymeric cement matrix with the highest strength and optimum microstructure can be obtained by properly selecting the three molar ratios.

Compressive test

The gradation analysis was first employed and illustrated in Table 2. It can be seen that SiO^sub 2^/Al^sub 2^O^sub 3^ exhibits the most significant influencing extent on the compressive strength among the three molar ratios. The corresponding gradation range was 26.5. Compared with SiO^sub 2^/Al^sub 2^O^sub 3^, K^sub 2^O/Al^sub 2^O^sub 3^ and H2O/K^sub 2^O have relatively little influences on the compressive strength, whose gradation ranges are 3.9 and 4.2, respectively. Comparing the difference among I1, II2, and III3 shows that the hardened geopolymeric cement matrix had the highest compressive strength at SiO^sub 2^/Al^sub 2^O^sub 3^ = 4.5, K^sub 2^O/Al^sub 2^O^sub 3^ = 0.8, and H2O/K^sub 2^O = 5.0.

The significance and influencing extent of the three molar ratios on the compressive strength can be easily observed through the previous gradation analysis. It is well known, however, that the gradation analysis cannot quantitatively distinguish whether the strength difference among different levels is caused by testing errors or by different level in the case of the same factor. To solve the problem, a variance analysis was employed, as shown in Table 4.

Careful observation of Table 4 shows that the sum of square of S^sub erro^ is close to SB and greater than SC. This means that the difference in the experimental results between Factors B and C probability was caused by the testing error rather than the factors. Therefore, the sum of the square of Factors B and C can sum up to an error term, that is, S^sub collect^ = (SB + SC + S^sub erro^). The observed mean square SA is to be tested against the error mean square S^sub collect^/6 = 44.5) with 6 degrees of freedom df. The proper test statistic is the F statistic with 2 and 6 df. At the 10% significance level (a = 0.1), the critical region of F is F = 3.18, which is usually taken as the upper tail of the F distribution, rejecting H0 if F = F1-a, where a is the area above F1-a. A significant F will indicate that the difference between Factor A and the testing error has something in it besides the estimate of variance. It probably implies that there is a real difference in the means, and H0 should be rejected.

Comparing each mean square with the error mean square Scollec indicates that Factor A, that is, the molar ratio of SiO^sub 2^/ Al^sub 2^O^sub 3^ has a considerable impact on the compressive strength at a = 0.1, whereas the effects of other factors such as K^sub 2^O/Al^sub 2^O^sub 3^ and H2O/K^sub 2^O are negligible. A comparison of the values of I1, II2, and III3 and for Factor A shows that the compressive strength increases with an increase in SiO^sub 2^/ Al^sub 2^O^sub 3^ (note that the maximum difference in I1, II2, and III3 reaches 26.5). The geopolymeric cement matrix made with SiO^sub 2^/Al^sub 2^O^sub 3^ = 4.5 has the highest compressive strength. Considering that there is little effect of the other two factors, K^sub 2^O/Al^sub 2^O^sub 3^ = 0.8 and H2O/K^sub 2^O = 5.0 are selected to reduce the amount of expensive KOH in this study. The optimum mixture proportions determined according to the variance analysis is consistent with that according to the gradation analysis. The conclusion obtained by the variance analysis can be used as feedback to design a better experiment.

As a result, the optimum mixture of K-PSS geopolymeric cement matrix is made on the basis of compressive strength with SiO^sub 2^/Al^sub 2^O^sub 3^ = 4.5, K^sub 2^O/Al^sub 2^O^sub 3^ = 0.8, and H2O/K^sub 2^O = 5.0, that is, Formula K-PSS7.

FTIR analysis

The previous mechanical test shows that the compressive strength of the hardened geopolymeric cement matrixes made with different formulas has a considerable difference. To elucidate in a microscopic scale the strength difference among different geopolymeric formulas, the FTIR technique was employed to characterize the microstructure of these hardened geopolymeric cement pastes and elucidate the difference of the compressive strength for different formulas.

The IR spectra of various geopolymeric cement matrixes is presented in Fig. 1. The chemical shifts of main IR bands and corresponding species were determined based on the references.26,27 By carefully comparing the IR spectra, the following phenomena can be observed:

1. The strong band at 427.6 in.^sup -1^ (1086 cm^sup -1^) shifted toward the low wavenumber after geopolymerization reaction. The shift was approximately 30.7 in.^sup -1^ (78 cm^sup -1^). This demonstrates that an obvious change in the microstructure took place during hydration reaction, resulting in a formation of new products with different microstructure from metakaolin. Similar results were also seen in the XRD spectra, as shown in Fig. 2. According to References 26 and 27, a 427.6 in.^sup -1^ (1086 cm^sup -1^) peak was caused by symmetrical vibration of the Si-O bond, whereas a 396.9 in.^sup -1^ (1008 cm^sup -1^) peak was ascribed to asymmetrical vibration of the Si-O bond. The large shift toward the low wavenumber may be attributed to the partial replacement of SiO^sub 4^ tetrahedron by AlO^sub 4^ tetrahedron, resulting in a change in the local chemical environment of Si-O bond. In addition, this band at 396.9 in.^sup -1^ (1008 cm^sup -1^) was very strong but a lack of sharp features is indicative of the general disorder in the Si(Al)-O- network, reflecting the wide distribution of the SiQ^sup n^(mAl) units in the polymeric molecular chains of geopolymeric cement products. SiQ^sup n^(mAl) units are the conventional notation used to described the structural units in aluminosilicates, where n represents the degree of condensation of SiO^sub 4^ tetrahedra. In this way, if n = 0, 1, 2, 3, and 4, then silicon is respectively in isolated mono-group (SiQ0), in dislicates and chain end groups (SiQ1), in middle groups in chains (SiQ2), in sheet sites (SiQ3), and in 3D cross-linked sites (SiQ^sup 4^). The letter m represents the number of aluminum atoms in the first coordination sphere of silicon. In general, if there is no aluminum atom in the second coordination sphere of silicon, the notation becomes SiQ^sup n^(nSi). The stretching modes of the Si-O bonds of the SiQ^sup n^ units are IR active in the 334.6 to 472.4 in.^sup -1^ (850 to 1200 cm^sup -1^) region with the absorption bands of the SiQ^sup n^ unit with n = 4, 3, 2, 1, and 0 centered at approximately 472.4, 433.1, 374.0, 354.3, and 334.6 in.^sup -1^ (1200, 1100, 950, 900, and 850 cm^sup -1^), respectively.15 These values shift to lower wave numbers when the degree of silicon substitution by aluminum in the second coordination sphere increases, as a consequence of the weaker Al-O bond.

2. The intensities of the bands at 359.8 and 314.2 in.^sup -1^ (914 and 798 cm^sup -1^) in IR spectrum of metakaolin caused by six-fold coordinated Al(VI)-OH stretching vibration and six coordinated Al(VI)-O stretching vibration, respectively, are considerably reduced or disappeared after geopolymerization reaction of geopolymeric cement. In addition, a similar trend is also found for the Si-O symmetrically stretching band at approximately 274.4 in.^sup -1^ (697 cm^sup -1^). This shows that sixcoordinated Al(VI) maybe changes into a four-coordinated one, and participates in the framework structure of geopolymeric products during the process of hydration reaction.

3. A weak band at approximately 330.7 in.^sup -1^(840 cm^sup -1^) can be observed in the K-PSDS1, K-PSDS2, K-PSDS3, and K-PSDS9 formulas. The band cannot be found in the IR spectrum of metakaolin. The new band is assigned to the bending vibration of Si-OH. This can be caused by some bond breakage sites in the network structure of geopolymeric products for some geopolymeric cement mixtures. The existence of Si-OH will cause a decrease in the degree of condensation, thus a reduction in mechanical strength.

Based on the previous analysis of IR spectra, the following is concluded: during geopolymerizaiton reaction of geopolymeric cement, SiO^sub 4^ and AlO^sub 4^ tetrahedaron will be released from the surface of metakaolin particles under the attack of strongly alkaline solution. The VI coordinated Al introduced by the metakaolin will also be leached out in strongly alkaline solution, and the coordination state will transform from VI to IV. As the dissolution proceeds, more and more SiO^sub 4^ and AlO^sub 4^ tetrahedarons go into the alkaline solution. After undergoing structural reorientation to a certain extent, the SiO^sub 4^ and AlO^sub 4^ tetrahedarons will produce a polycondensation reaction and thus a formation of geopolymeric products with a framework structure. It is the bonding of AlO^sub 4^ (IV) with SiO^sub 4^ that causes the great changes in the original Si-O chemical environment, resulting in different chemical shifts and a reduction or disappearance of some characteristic bands in IR spectra. When the mixture proportion of geopolymeric cement matrix is not designed properly, incomplete poycondensation reaction will occur, thus some bond breakage in the framework structure. The breakage bonds will be saturated by OH. As a result, the Si-OH bending band at 330.7 in.^sup -1^ (840 cm^sup -1^) will appear in some geopolymeric cement formulas such as K-PSS1, K-PSS2, K-PSS3, and K-PSS9. From Fig. 1, it is clear that K-PSS7 had a prominent band at 396.9 in.^sup -1^ (1008 cm^sup -1^), which is the characteristic band of K-PSS geopolymeric products. The Si-OH bending band at 330.7 in.^sup -1^(840 cm^sup -1^) cannot be seen. Thus, it is reasonable to assume that K-PSS7 is the optimum one of the nine formulas. The geopolymerization degree of K-PSS7 was the largest and the compressive strength was also the highest.

The previous macroscopic and microscopic experimental results reveal that an almost fully-reacted K-PSS geopolymeric cement matrix can obtain when the molar ratio SiO^sub 2^/Al^sub 2^O^sub 3^ = 4.5, K^sub 2^O/Al^sub 2^O^sub 3^ = 0.8, and H2O/K^sub 2^O = 5.0, that is formula K-PSS7. SiO^sub 2^/Al^sub 2^O^sub 3^ had a very important influence on strength gain and desirable microstructure of K-PSS geopolymeric cement matrix. Comparatively, K^sub 2^O/Al^sub 2^O^sub 3^ and H2O/K^sub 2^O exhibit little impact. The sodium content, however, was used to satisfy the charge-balance requirements within the structure, without proving an excess that can form sodium carbonate and may destroy the polycondensation process. The water content for a critical polycondensation reaction is less obvious, whereas it is clearly necessary to provide sufficient water to facilitate initial mixing and act as a carrier for ionic transport. The excess water perhaps dilutes the polycondensation reaction or leaches more soluble components and transports them away from the reaction zone.

MICROSTRUCTURAL CHARACTERISTICS OF FULLY-REACTED K-PSS GEOPOLYMERIC CEMENT MATRIX

XRD analysis

As seen in Fig. 2, there is a large diffuse halo peak at approximately 20 to 40 degrees (2^sub max^CuK) in X-ray diffractogram of the fully-reacted K-PSS geopolymeric cement matrix, that is, formula K-PSS7. This indicates that K-PSS geopolymeric products are mainly X-ray amorphous materials consisting of randomly arranged Si-O-Al polytetrahedra with a lack of periodically repeating Si-O-Al atomic ordering.

ESEM-EDXA analysis

Figure 3 shows an ESEM micrograph of the hardened K-PSS7 geopolymeric cement matrix. It can be clearly seen that the microstructure of geopolymeric products is sponge-like. No crystal with a regular shape is observed in the bulk geopolymeric cement matrix.

EDXA is also performed on the whole region shown in the ESEM micrograph to determine its chemical composition after polycondensation reaction. The oxide molar ratios of SiO^sub 2^/Al^sub 2^O^sub 3^ and K^sub 2^O/Al^sub 2^O^sub 3^ are 4.28 and 1.06, respectively, which are close to the theoretical values of K-PSS geopolymeric matrix (SiO^sub 2^/Al^sub 2^O^sub 3^ = 4.0 and K^sub 2^O/Al^sub 2^O^sub 3^ = 1.0).

In addition, X-ray mappings of three types of main elements (K, Si, and Al) of the whole region shown in the previous ESEM micrograph are also collected and displayed in Fig. 4. It can be seen from Fig. 4 that K, Si, and Al are uniformly distributed across the bulk paste. This means that the sponge-like geopolymeric products are generated evenly and homogeneously during the whole process of geopolymerization reaction.

TEM-EDS analysis

Figure 5 presents the TEM micrograph of one particle randomly selected in K-PSS geopolymeric products. It can be clearly seen from Fig. 5 that the shape of geopolymeric products is sponge-like, similar to the ESEM micrograph. To investigate the microstructure nature of the sponge-like products, electronic diffraction is further performed to determine whether the sponge-like products are amorphous or not. Figure 6 depicts the electronic diffraction pattern of the sponge-like products. From Fig. 6, the diffraction pattern shows a cloudy ring shape, which is the typical characteristic of amorphous materials. If the geopolymeric products are crystal materials, the diffraction pattern should be a clear dotted ring shape, as shown in Fig. 7. The TEM finding is consistent with one by using XRD and ESEM techniques. It is worth noting, however, that the analysis of TEM-EDS conducted herein is only at a semi-quantitative level, as the analysis covers only a very little area of the K-PSS geopolymeric bulk matrix. Considering other practical techniques cannot exactly distinguish the crystalline phase from the amorphous gel, such a semi-quantitative method is satisfactory for providing some insight into the structural compositions of geopolymeric cement matrix.

MAS-NMR analysis

MAS-NMR spectroscopy can provide useful structural data for aluminosilicates (zeolites, clays, ceramics, portland cements, and geopolymeric cement). In particular, ^sup 29^Si and ^sup 27^Al MAS-NMR studies represent a very powerful tool.28

^sup 27^Al MAS-NMR

Earlier investigations29,30 showed that in aluminate anions, four-coordinated Al (with respect to oxygen) resonates at 60 to 80 ppm, and that in aluminosilicates, four-coordinated Al resonates at 50 to 20 ppm, whereas six-coordinated Al resonates at 0 to 10 ppm from [Al(H2O)6]3+.

The ^sup 27^Al MAS-NMR spectroscopy of the fully-reacted K-PSS geopolymeric cement matrix is shown in Fig. 8. It can be seen that there is a predominant resonance at 40 ppm that assigned to the four-coordinated Al and its belongs of AlQ4(4Si) type with respect to the reference.3 In addition to the resonance at 40 ppm, a very small resonance at -12.3 ppm that assigned to the six-coordinated Al can also be observed. The six-coordinated Al is caused by traces of kaolin in metakaolin due to the uncompleted calcinations. The absence of any other resonance excludes any residual singular building units of low molecular weight such as AlQ4(0Si), AlQ4(1Si), and AlQ4(2Si). Thus, K-PSS geopolymeric products are mainly 3D framework aluminosilicates with polymeric building units.

Although ^sup 27^Al MAS-NMR is a powerful tool in determining the coordination environment of Al, it cannot differentiate the various molecular configurations proposed for geopolymeric cement: poly(sialate) (Si-O-Al-O-)^sub n^, poly(sialate-siloxo) (Si-O-Al-O-Si-O-)^sub n^, or poly(sialate-disiloxo) (Si-O-Al- O-Si-O-Si-O-)^sub n^ polymeric building units. This differentiation can be achieved by using ^sup 29^Si MAS-NMR spectroscopy. Therefore, the combination of the ^sup 27^Al MAS-NMR and ^sup 29^Si MAS-NMR spectra make it possible to better understand the structural nature of K-PSS.

^sup 29^Si MAS-NMR

The ^sup 29^Si MAS-NMR spectrum of the K-PSS geopolymeric cement matrix is illustrated in Fig. 9. It can be seen that K-PSS gives a broad resonance in the range of -70 to -110 ppm associated with a strong resonance at -82 ppm, a very strong resonance at -88 ppm and a small resonance at -103.8 ppm. Broad resonances are generally found in zeolitic gels before crystallization of the zeolites.3 The ^sup 29^Si broad resonances indicate that the Si and Al tetrahedras in K-PSS geopolymeric cement matrix are not regularly ordered along the polymeric chains.

A previous study has shown that the chemical shift of ^sup 29^Si in an amorphous or highly disordered state as geopolymer would be increase by approximately 5 ppm when comparing with the zeolite with the same chemical compositions.2,3 The three main resonances in ^sup 29^Si MAS-NMR spectrum of K-PSS geopolymeric cement matrix, namely, -82 ppm, -88 ppm, and -103.8 ppm, corresponding to the ordered ^sup 29^Si chemical shifts of -87 ppm, -93 ppm, and -108.8 ppm in zeolite, can be assigned to SiQ^sup 4^(4Al), SiQ^sup 4^(2Al), and SiQ^sup 4^(4Si) units, respectively.30

Molecular structural model of K-PSS geopolymeric cement matrix

On the basis of the previous XRD, FTIR, ESEM-EDXA, TEM-EDS, and MAS-NMR spectra, it can be observed that the fully-reacted K-PSS geopolymeric cement paste shows structural characteristics similar to glassy or gel substances in having a wide range of Si endowments, but predominantly the network molecular chains of Si partially replaced by fourcoordinated Al tetrahedral, that is, SiQ^sup 4^(4Al) and SiQ^sup 4^(2Al).

It can be seen from Fig. 9 that the broad resonance in range of -70 to -110 ppm of the ^sup 29^Si MAS-NMR spectrum is mainly composed of three SiQ^sup 4^(mAl) units: SiQ^sup 4^(4Al), SiQ^sup 4^(2Al), and SiQ^sup 4^(4Si). The three resonances assigned to the three different SiQ^sup 4^(mAl) units are so close, however, that the interaction between resonances occurs, resulting in a lack of spectral resolution. To address it, Gaussian peak deconveolution is adopted in this study to separate and quantify SiQ^sup 4^(mAl) units, as shown in Fig. 9. Quantification of network ordering may be allowed to establish the molecular model for describing speciation of silicon according to the types and relative proportions of SiQ^sup 4^(mAl) units.

According to the types of SiQ^sup 4^(mAl) units and their relative proportions given in Table 5, the molar ratio of Si can be calculated: Al using Eq. (1) proposed in literature.31 The ratio of Si: Al equal to 2.055358:1 for K-PSS synthesized in this study, which is in a good agreement with the theoretical value of K-PSS geopolymeric cement matrix (Si: Al = 2.0)

... (1)

where ISi(mAl) is the intensity of each peak in the deconveluted ^sup 29^Si MAS-NMR spectra.

Some studies3 proposed that structural configuration of K-PSS geopolymeric cement matrix was similar to that of hydrated potassium aluminum silicate, KAlSiO. One novel molecular structural model of K-PSS geopolymeric matrix is proposed based on the structural configuration of hydrated Leucite, as shown in Fig. 10. The model is different from the structural models established by Davidovits3 and Valeria et al.,27 respectively. In the proposed model, SiO^sub 4^ and AlO^sub 4^ tetrahedras are so arranged that the types and relative proportions of the main SiQ^sup 4^(mAl) units are statistically consistent with the decomposed results of ^sup 29^Si MAS-NMR spectrum and the molar ratio of Si:Al (2.055358:1). And all Al atoms are four-coordination and have Si nearest neighbors.

CONCLUSIONS

In this study, three key parameters-SiO^sub 2^/Al^sub 2^O^sub 3^, K^sub 2^O/ Al^sub 2^O^sub 3^, and H2O/K^sub 2^O-that influence the synthesization of the K-PSS geopolymeric cement matrix at room temperature were systematically investigated. A total of nine K-PSS geopolymeric cement matrixes with different SiO^sub 2^/Al^sub 2^O^sub 3^, K^sub 2^O/Al^sub 2^O^sub 3^, and H2O/K^sub 2^O were designated to investigate the effects of the three parameters on mechanical strength and microstructure in accordance with orthogonal design principle. The gradation and variance analysis on experimental results showed that SiO^sub 2^/Al^sub 2^O^sub 3^ had a significant impact on compressive strength. The highest compressive strength (5.04 ksi [34.8MPa]) can be achieved when SiO^sub 2^/ Al^sub 2^O^sub 3^ = 4.5, K^sub 2^O/Al^sub 2^O^sub 3^ = 0.8, and H2O/K^sub 2^O = 5.0. The corresponding microstructure was also proved to be the optimum one among the nine geopolymeric cement mixtures according to IR spectra. As a result, the previous mixture was regarded as a fully-reacted K-PSS geopolymeric cement matrix.

Subsequently, XRD, ESEM-EDXA, TEM-EDS, and MAS-NMR techniques are employed to further characterize the microstructure of the fully-reacted geopolymeric cement matrix. The microscopic analysis shows that the fully-reacted K-PSS geopolymeric cement matrix possesses structural characteristics similar to glassy or gel substances in having a wide range of Si endowments, but predominantly the framework molecular chains of Si partially replaced by four-coordinated Al tetrahedral. A 3D molecular structural model is also proposed based on the decomposition of ^sup 29^Si MAS-NMR spectrum of the fully-reacted K-PSS geopolymeric cement matrix synthesized from the optimum mixture proportion.

ACKNOWLEDGMENTS

The authors gratefully acknowledge the financial support from the Natural Science Foundation of China (No. 50702014), the Jiangsu Province Natural Science Foundation (BK2006555), and the China West Communication Structures Project (2006ZB12). Some of the research was carried out at the Material Laboratory, Department of Civil Engineering, Hong Kong University of Science and Technology.