2025 Volume 66 Issue 10 Pages 1399-1407
Geopolymers are attracting attention as an environmentally friendly alternative to cement. In this study, compressive strength tests were conducted on geopolymer pastes using fly ash and blast furnace slag as active fillers. Sodium hydroxide solution and sodium silicate were used as alkaline activators. Raman spectroscopy was also employed to investigate its potential for analyzing Ca-enriched geopolymers. These results indicated that calcium aluminosilicate hydrate was formed by 7-day in the environment containing a certain amount of soluble Si in NaOH solution. This was reinforced by the formation of magnesium aluminosilicate hydrate and sodium aluminosilicate hydrate by 28-day, which may contribute to higher compressive strength. Moreover, the coexistence of high-polymerized silicates and relatively low-polymerized silicates is thought to enhance the physical entanglement of the silicates and to increase the compressive strength. These findings suggest that Raman spectroscopy is a useful tool for microstructural analysis of Ca-enriched geopolymers.
It is well acknowledged that the amount of cement production has been increasing because of the high-speed development of infrastructures. However, the cement production process releases large amount of green-house gas, carbon dioxide (CO2) and consumes plenty of energy. The cement subsector consumes approximately 12%–15% of total industrial energy use and results in exhaustion of CO2 which occupies 7% of worldwide emission [1]. Several efforts have been made to promote energy conservation in cement industry such as replacing cement with slag, plastic waste, and so on. However, since ordinary portland cement (OPC) is still the most widely used construction material, the CO2 emission due to the calcination of the limestone becomes unavoidable.
Recently, geopolymer materials have attracted large attention to become another alternative for OPC because of its eco-friendly production process using industrial by-product with low CO2 emission. Geopolymer is defined as a solidified framework body obtained from a mixture of raw materials mainly composed of amorphous aluminosilicate binders and at least one kind of an aqueous solution which is the silicate, carbonate or hydroxide of an alkali metal [2]. Fly ash (FA) and blast furnace slag (BFS) are commonly used as binders and the aqueous solution of sodium hydroxide (NaOH) and sodium silicate (Na2SiO3) are commonly used as alkali activators.
There are many reports on the environmental impact reduction effects of geopolymers [3–9]. Davidovits [3] produced geopolymer by mixing calcined earth and stone powder with blast furnace slag fine powder at a ratio of 7:3. It was reported that a 59% and 80% reduction in energy consumption and CO2 emission intensity, respectively, compared to OPCs of similar strength. Furthermore, the CO2 emission intensity of the geopolymer with only FA as active filler was 75–90% lower than that of OPC. Weil et al. [5] reported that the CO2 emissions intensity of a geopolymer with BFS and FA as active fillers was about 30% of that of OPC without accounting for CO2 emissions from the transportation of the materials used, but it was about 60% when CO2 emissions from the transportation of the materials used were accounted for. Habert et al. [9] reported that the greenhouse gas emissions of geopolymers with FA as active filler are about 50% of OPC, but most of the greenhouse gas emissions are due to the use of water glass and sodium hydroxide. Therefore, the CO2 emissions of geopolymer are estimated to be about 40% to 90% lower than those of OPC. These emissions are strongly influenced by CO2 emissions from the alkaline activator and depend on CO2 emissions from transportation and the material used as active filler.
In addition, it is generally known that the strength of concrete-like materials increases as reaction products fill the internal voids. Therefore, it is important for evaluating the mechanical property of hardened body to understand the reaction products. The geopolymer reaction products are known as amorphous which is polymer with silicon (Si) as the core of their structure. The products are generally classified as sodium aluminosilicate hydrate (NASH) or calcium aluminosilicate hydrate (CASH) [10, 11]. Taylor [12] proposed a method to classify Si atom in silica polymers using a notation called Qn structure (n = 0∼4), where Q represents a silicate tetrahedron (SiO4) and n represents the number of oxygen atoms bridging adjacent SiO4 units. When Si is isomorphously replaced by aluminum (Al) and m of the n SiO4 tetrahedra are replaced by AlO4 tetrahedra, the structure is represented as Qn(mAl) shown in Fig. 1. Q0 represents a monomer of Si(OH)4, and it can be seen that the degree of polymerization increases as n increases. This is why the Qn(mAl) structure is believed to affect the strength of hardened cement or geopolymer. However, there are still many details about geopolymer reaction products that remain unknown, and further accumulation of knowledge is needed.
Structures of SiO4 tetrahedra incorporated with Al.
The microstructure and the reaction degree of geopolymers are usually studied by X-Ray Diffraction (XRD) [13, 14], Nuclear Magnetic Resonance (NMR) [15–18], Fourier Transform Infra-Red spectroscopy (FTIR) [19–23].
Since the cement or geopolymer reaction products are amorphous, it is not possible to obtain much information using XRD, a common method for analyzing crystal structures. On the other hand, NMR is a powerful approach to provide the Qn(mAl) structure. Most research using NMR focuses on the degree of polycondensation. However, the strength of hardened geopolymer depends on not only the Qn(mAl) structure, but also any other bonds that may exist in the products. Then, in addition to these methods mentioned above, FTIR is usually applied to study the vibration of bonds between silicate and other metal elements related to the gel formation of geopolymer materials. It is pointed out that FTIR technology has difficulty in applying FA/BFS based geopolymer since the characteristic bands for both calcium silicate hydrate (CSH) gel and sodium aluminosilicate hydrate (NASH) gel are hard to distinguish. Based on this statement, FTIR may not be the most suitable approach to study FA/BFS based geopolymer.
Geopolymer research using Raman spectroscopy has attracted attention recently [24–27]. Like FTIR, Raman spectroscopy is a technique to analyze functional groups through molecular vibrations and can be applied to amorphous as well as crystalline materials. It is also possible to distinguish between CSH and sodium silicate hydrate [28], and to evaluate the Qn(mAl) structure, which represents the polymerization characteristics of silicates. Therefore, Raman spectroscopy can provide important information for analysis of the microstructure of FA/BFS-based geopolymers. However, Raman spectroscopy has not been widely adopted in the field of cement chemistry. Therefore, the main objective of this paper is to apply Raman spectroscopy to FA/BFS based geopolymer, especially Ca-enriched geopolymer system with high BFS replacement. Furthermore, this paper focuses on not only high BFS replaced geopolymer, but also low BFS replaced geopolymer, and the effect of the amount of soluble Si regarding compressive strength of hardened body.
FA and BFS are used as binders in this research, with the density of 2.40 g/cm3 and 2.88 g/cm3 respectively. The chemical composition is described in Table 1. XRD patterns of FA and BFS are shown in Fig. 2. Figure 3 describes the Raman spectra of FA and BFS. The crystalline phase in FA is mainly composed of quartz and mullite. BFS is nearly amorphous except for gypsum which is recognized as a small peak at around 1005 cm−1 in Raman spectrum.
;%0A%09%09%09newWindow.document.open();%0A%09%09%09newWindow.document.write('<img src=%22./Graphics/MT-Z2025003tab01.jpg%22>');%0A%09%09%09newWindow.document.close();%0A%09%09)
XRD patterns of FA and BFS, (a) Fly ash, (b) Blast furnace slag. (online color)
Raman spectra of FA and BFS, (a) Fly ash, (b) Blast furnace slag.
The aqueous solutions of sodium hydroxide (SH) and sodium silicate (SS) are used as alkali activators in this research. The chemical composition is described in Table 2. Purified water is used to prepare alkali activator solution as well as additional water in geopolymer paste.
;%0A%09%09%09newWindow.document.open();%0A%09%09%09newWindow.document.write('<img src=%22./Graphics/MT-Z2025003tab02.jpg%22>');%0A%09%09%09newWindow.document.close();%0A%09%09)
The experiment was designed to study the relationship between the amount of soluble Si and compressive strength under different BFS replacement ratios. The Si/Alkali (Si/A) ratio was adjusted by SS replacement ratio, which referred to the mass ratio of SS solution and all alkali solutions (SS and SH), as 0 wt%, 20 wt%, 50 wt%, 80 wt% and 100 wt%. The BFS replacement ratio, which referred to the mass ratio of BFS and all binders (FA and BFS), was selected as 20 wt% and 60 wt%. The mixture proportions are shown in Table 3. The concentration of SH was kept constant as 5 mol/L; The Water/Binder (W/B) ratio was kept constant as 40 wt%; The Alkali/Water (A/W) ratio was kept constant as 8 mol% to eliminate all the other possible influences. Three specimens were prepared for each proportion. The casting and curing process in the experiment was as follows. Alkali activators were prepared by mixing alkali aqueous solution including SH and SS with water. Binders with a certain amount of FA and BFS were pre-mixed in mortar mixer. Then, alkali activators were poured into mortar mixer, and the slurry was mixed for 30 seconds at low speed. After resting for 90 seconds, the materials were mixed again for 60 seconds at high speed. The whole mixing time was kept as 3 minutes. The fresh geopolymer paste was casted into cylinders with Φ5 × 10 cm and the molds were put on the shaking table for 30 seconds to reduce the air bubbles inside the slurry. After initial setting, all the cylinder specimens were wrapped with plastic films to prevent water evaporation. Both 7-day and 28-day compressive strengths were measured on each. The mean compressive strengths of each three specimens were reported and the variations were expressed using the standard deviation.
;%0A%09%09%09newWindow.document.open();%0A%09%09%09newWindow.document.write('<img src=%22./Graphics/MT-Z2025003tab03.jpg%22>');%0A%09%09%09newWindow.document.close();%0A%09%09)
The geopolymer samples after the compression test were wrapped with plastic film and the fracture surfaces were used to conduct Raman spectroscopy with the excitation wavelength of 532 nm. Raman spectra were taken at nine points, and the average spectra were adopted as the final results. Based on the previous research [29–39], the band assignments in Raman spectroscopy related to geopolymers can be summarized in Table 4.
;%0A%09%09%09newWindow.document.open();%0A%09%09%09newWindow.document.write('<img src=%22./Graphics/MT-Z2025003tab04.jpg%22>');%0A%09%09%09newWindow.document.close();%0A%09%09)
The test results of compressive strength are shown in Fig. 4. The experiment results which 100 wt% SS replacement should be neglected because of its bad workability which results in low compressive strength. It can be noted that the 28-day compressive strength is lower than 7-day compressive strength for 50 wt% SS replacement under 60 wt% BFS replacement. This may be caused only one of the three data in 50 wt% SS replacement under 60 wt% BFS replacement could be obtained due to the work. It can be concluded that the 7-day compressive strength reaches maximum when 50 wt% SS replacement and 28-day compressive strength reaches maximum when 80 wt% SS replacement. When comparing the same BFS replacement ratio and same material age, with further decrease of soluble Si when the SS replacement ratio is low, the decrease of the core elements of the geopolymer reaction products and the compressive strength also decreases. In other words, a certain amount of soluble Si is necessary to obtain sufficient strength. On the other hand, if SH ratio is high, dissolving the binders at least in the period from 7 to 28-day will increase the amount of the elements leached from the binders. This is why some extent concentration of SH is thought to be necessary. The maximum compressive strength may vary with the balance between soluble Si and SH concentrations, and in this experiment, the maximum strength was obtained at 80 wt% SS replacement ratio at 28-day.
Results of Compression test. (online color)
I. Garcia-Lodeiro et al. [40] proposed the reaction model for NASH gel formation. The prerequisite for polymerization reaction is that the paste reaches a Si rich phase. However, Si is more difficult to dissolve because Si-O bonds are stronger than Al-O bonds [41, 42]. With the dissolution of Al-O, part of alkali activator is consumed which results in low dissolution of Si-O especially in low alkaline environment. The soluble Si in sodium silicate solution can raise the Si concentration of the paste, where the polymerization reaction is more likely to happen. J. Li et al. [43] described the different reaction process of alkali activated slag using both sodium silicate and sodium hydroxide. When sodium silicate is used as alkali activator, the dissolved Ca from slag can react with the soluble Si from sodium silicate to form CSH gel. The dissolved Al and Si from slag furtherly forms C(A)SH gel. When sodium hydroxide is used as an alkali activator, the dissolved Ca results in the crystallization of Ca(OH)2. The dissolved Al and Si from slag forms C(A)SH gel which covers the surface of slag particles due to its low solubility. Based on the difference stated above, it has been pointed out that sodium silicate-activated slag has smaller porosity, higher compressive strength and bigger shrinkage than sodium hydroxide-activated slag. Also based on the reports of above-mentioned previous research, soluble Si is important for both FA based geopolymer and alkali activated slag which also agrees with the results of the experiment for FA/BFS based geopolymers. In addition, some extent of high alkali environment is necessary to dissolve Si-O bonds because part of alkali activator is consumed with the dissolution of Al-O bonds. Both the large amount of soluble Si in alkali activator and some extent high alkali environment could contribute to higher compressive strength.
Raman spectra of geopolymer paste samples are shown in Fig. 5 to Fig. 8. In low wavenumber region (300–600 cm−1), a strong peak around 460 cm−1 which may relate to quartz located in the core of FA are recognized. It can be seen that, except for some spectra, the larger the dosage of FA, the material age, and the dasage of SH, the larger the quartz-derived peak becomes. As dosage of SH gets larger, pH of the alkali solution gets higher because OH− increases in the alkali solution. As a result, it becomes easier to dissolve the binders (FA and BFS). And the dissolution reaction proceeds with the material age. In the system with large dosage of FA, the peak derived from quartz, which is included in the unreacted portion of FA particles, are considered to be particularly strong. On the other hand, in some samples shown in Fig. 6 and Fig. 8, small dosage of FA and large dosage of SH caused to have dissolved even quartz, resulting in the disappearance of quartz-derived peak.
Raman spectra of 7-day geopolymer sample with 20 wt% BFS replacement. (online color)
Raman spectra of 7-day geopolymer sample with 60 wt% BFS replacement. (online color)
Raman spectra of 28-day geopolymer sample with 20 wt% BFS replacement. (online color)
Raman spectra of 28-day geopolymer sample with 60 wt% BFS replacement. (online color)
In medium and high wavenumber region (more than 600 cm−1), especially for geopolymer samples with 60 wt% BFS replacement, Raman band related to O-Si-O bending vibration appears at around 670 cm−1, which proves the formation of C(A)SH gel and magnesium aluminosilicate hydrate (M(A)SH) gel. The hydration product of C(A)SH gel incorporated with Mg can improve early mechanical properties by decreasing the porosity [44, 45]. The trapezoidal peaks in the region from 800 to 1200 cm−1 usually relate to different types of SiO4 tetrahedra incorporated with Al. Therefore, peak fitting was conducted to the spectra by the following method. Some points were picked up from the raw data of Raman spectra in the range from 250 to 1800 cm−1. Then, a quadratic curve was approximated to the extracted data by the least-squares method, and the approximation was subtracted from the raw data spectra to remove background of the spectra. If necessary, the data was corrected so that the minimum value was greater than or equal to zero in the range of 500 to 1200 cm−1. The corrected spectra were fitted using the Gaussian curves with center values shown in Table 4, and the area occupied by each Gaussian curve relative to the total area of the spectra was calculated. An example of curve fitting is shown in Fig. 9. The area ratios of each Gaussian curve are compared between the two material ages of the same proportion. Figure 10 shows the fitting results of the highest strength of 80 wt% SS replacement at the age of 28-day, when the reaction had progressed to a certain extent. Figure 10(a) shows that the proportion of Q4(mAl) for m = 1∼3 was large at 7-day, while the proportion of Q2, Q3, and Q4(4Al) increased at 28-day in the 80 wt% SS replacement. Nailia R. Rakhimova et al. [46] pointed out that Q4(3Al) and Q4(2Al) units have a large content percentage in low calcium geopolymer matrix which contributes to NASH gel. G. Fang et al. [47] studied the microstructure of alkali activated fly ash/slag paste and conclude that Q2 is related to C(A)SH gel; Q3 and Q4(4Al) are related to highly cross-linked NCASH gel; and Q4(1Al), Q4(2Al) and Q4(3Al) are related to NASH gel. Therefore, it can be inferred that NASH was mainly generated at 7-day, and CASH and NCASH were generated until 28-day in this experiment sample with the 20 wt% BFS replacement and 80 wt% SS replacement. On the other hand, Fig. 10(b) shows that the ratio of NASH at 7-day was relatively small, and the ratio increased at 28-day in the 80 wt% SS replacement. In addition, judged from O-Si-O(C(A)SH) and O-Si-O(M(A)SH) peaks, C(A)SH was already produced at 7-day, and M(A)SH was produced between 7 and 28-day. These results suggest that the 60 wt% BFS replacement mainly produced C(A)SH at 7-day, and NASH and M(A)SH were produced until 28-day, thereby increasing the strength. Aughenbaugh et al. [48] mentioned that fly ash works as a network modifier to improve the polymerization of aluminosilicate gel. The results in Fig. 10 suggest that the formation of NASH and M(A)SH acted as a modifier. Here, NASH is considered to be Na-bonded at the end of the polymerized chain with relatively high Si polymerization compared to C(A)SH and M(A)SH. Therefore, the order of the reactions of N(A)SH, C(A)SH, and M(A)SH may change following the concentration of Si, Na, Ca, and Mg which depends on FA/BFS ratio and the content of alkali activator. When Fig. 4 is considered with focusing on the BFS replacement ratio, it can be seen that the higher the BFS replacement ratio, the greater the compressive strength. This is similar to previous studies [49–51]. And in the period up to 28-day, C(A)SH and M(A)SH, which are produced by Ca and Mg in the BFS, are considered to have a significant effect on the strength enhancement. This is the reason why the higher BFS replacement ratio has an effect to the greater compressive strength.
An example of curve fitting. (online color)
Comparison of peak fittings in 80 wt% SS replacement by material age, (a) 20 wt% BFS replacement, (b) 60 wt% BFS replacement.
The 50 wt% SS replacement showed the greatest compressive strength at 7-day regardless of the BFS replacement ratio, while the 80 wt% SS replacement showed the greatest compressive strength at 28-day. Figure 11 shows a graph comparing the area ratio of each peak for the 50 wt% SS replacement ratio and the 80 wt% SS replacement ratio. The value of Q4(mAl) is the sum of Q4(1Al) to Q4(4Al). From the figure, it can be seen that the 50 wt% SS replacement shows less change in the area ratio of each peak from 7-day to 28-day than the 80 wt% SS replacement ratio. This suggests that the strength increase may be greater if a certain type of silicate is formed by the 7-day and then another type of silicate is formed as reinforcement, rather than if each silicate is produced at a constant rate. However, the total amount of product formed in different systems (e.g. the 50 wt% SS replacement ratio and the 80 wt% SS replacement ratio) cannot be discussed by using the intensities of Raman peaks because the intensity of each peak in the Raman spectrum is relative. Since the strength of the hardened body is naturally affected by the total amount of products, it is necessary to study the relationship between the strength of the hardened body and the total amount of products as well as the type of products in the future work.
Comparison of peak fittings in 50 wt% SS replacement and 80 wt% SS replacement, (a) 20 wt% BFS replacement, (b) 60 wt% BFS replacement. (online color)
The integrated area of Q0, Q1, Q2, Q3, and Q4(mAl) at 28-day are calculated, and the results are shown in Fig. 12. The relationship between the compressive strength of hardened body and the ratio of Q3+Q4(mAl) is shown in Fig. 13. With the exception of the 0 wt% SS replacement, there is a negative correlation between the ratio of Q3+Q4(mAl) and compressive strength. This may be due to the fact that highly polymerized chains such as Q3 and Q4 are prone to steric hindrance, which may weaken the entanglement between the polymerized chains. It is considered that the coexistence of a certain number of relatively slightly polymerized chains such as Q0, Q1, and Q2 may facilitate the expression of strength by physical entanglement. Since the 0 wt% SS replacement does not contain SS, but only SH, it is considered that the lack of soluble Si as described above and the too high pH environment prevented polymerization of silicate, resulting in the failure of strength development. This can be seen from the fact that the 0 wt% SS replacement has a much higher percentage of Q0 than the others in Fig. 12.
Integrated area ratio of Q0, Q1, Q2, Q3, and Q4(mAl), (a) 20 wt% BFS replacement, (b) 60 wt% BFS replacement.
Relationship between the compressive strength of hardened body and the ratio of Q3+Q4(mAl) (Circled data are 0 wt% SS replacement).
In this study, microstructure of reaction product and compressive strength of hardened body of geopolymer paste using fly ash and blast furnace slag as active fillers and sodium silicate and sodium hydroxide solution as alkali activators were investigated. The following main conclusions are drawn:
