Patent Application
20220402820
The present invention relates to the field of cementitious compositions. Particularly, the invention concerns an alkaline-activated fly ash cementitious composition and the use of this composition as a binder in concrete production.
Global utilization and consumption of concrete is only second to water. It is the crucial building material to give rise to society's infrastructures all around the world. Approximately 25 billion tonnes of concrete are produced annually on a global scale. This is equivalent to over 1.7 billion truck loads per year or about 6.4 million truck loads a day or over 3.8 tonnes per person around the globe each year (WBSCD, 2009; Madlool et al., 2011).
With such high volumes of concrete produce, it significantly increases the production of cement. Cement acts as an adhesive to bind the constituents in the concrete mix. It comprises of approximately 5-20% of concrete.
Cement production contributes 5-7% of all anthropogenic CO2 emissions worldwide. One ton of cement manufacture produces approximately 0.7-0.9 ton of CO2 emissions. In southern Africa, it was reported that CO2 emissions were 753 kg per ton of cement manufactured.
The production of clinker, an intermediate product in cement manufacture, is the most energy-intensive step that accounts for roughly 80% of the cement production (Zhu, 2011). Approximately 50% of carbon emissions, during the fabrication process, originate from decomposition of raw materials to form clinker. Combustion of fossil fuels in pyro-procession units release 40% of emissions. The last 10% are due to transportation of raw materials and electricity consumed by electrical motors and facilities (Ali et al., 2011; Turner & Collins, 2013; Benhelal et al., 2013).
A large portion of CO2 is released during calcination of limestone at 900° C. It converts carbonates to oxides and CO2 forms as a by-product in the chemical reaction. The simplified stoichiometric relationship can be expressed as:
CaCO3+heat→CaO+CO2↑ (1)
About 64-67% of clinker comprises of calcium oxide (CaO) while the rest are iron oxides and aluminium oxide. This amounts to roughly 0.5 kg of CO2 that is produced per kg of clinker and emissions are dependent on clinker to cement ratio. Furthermore, it is estimated that 0.65-0.9 tonne of CO2 is produced per ton of cement depending on the type of fuel used, modern technology and equipment (Turner & Collins, 2013; Ali et al., 2011; Gibbs et al., 2000; Gao et al., 2014).
Since CO2 emissions contribute to 65% of global warming, it has become imperative that alternative methods be established to reduce the carbon footprint. To further justify the need for substitutes, ordinary Portland cement (OPC) is subjected to certain limitations. These limitations include durability issues due to its intrinsic properties, high permeability that can cause carbonation and corrosion problems and alkali-silica reactions (Torgal et al., 2008).
An alternative to OPC cements are blended cements, which are expected to significantly reduce cement use. Cement is partially replaced by supplementary cementitious binder materials based on waste by-products such as fly ash (FA) and granulated blast furnace slag (GBFS). Another alternative cementitious binders are alkali-activated aluminosilicate materials (Turner & Collins, 2013; Rashad, 2014).
Alkali-activated materials (AAM) are one of those alternative binders that have shown similar and at times better mechanical properties than Portland cement with lower carbon emissions. Since AAM utilize source materials rich in Al and Si, it provides an opportunity to use by-products from different processes such as FA, silica fume (SF) and GBFS.
Fly ash is a fine powder that is created as a by-product through the burning of pulverized coal in thermal power plants for producing electricity. In South Africa, Eskom is the entity responsible for generating electricity for the country and are the largest coal consumers. The company reports that the production of FA is approximately 35 million tonnes yearly, of which only 7% gets recycled. The rest gets discarded in massive landfills, which causes numerous environmental problems that is quickly becoming a cause of concern. South Africa has implemented a plan known as The South African National Development Plan 2030 (NDP), which places a specific attention to reduce the waste-to-landfill problem in the country. Since AAM requires by-products such as FA for its production, it can aid in recycling the FA whilst providing an adequate building material. This will promote a drive towards the strategic initiative set out by the NDP and also towards a global reduction in CO2 emissions.
However, alkali-activated materials have a slow rate of adaptation into the industry. This is primarily caused by the fact that AAM's require controlled environments to produce and is highly variable due to the binder's chemical composition. Furthermore, to achieve sufficient strength and mechanical properties, it is often required that FA based AAM's be cured under elevated temperature conditions, which is something not readily available on site and impedes its range of application.
Thus, it is imperative that newer and detailed methods be developed to further drive its acceptance. There are numerous studies into unary and binary blends. Unary blends such as alkali-activated fly ash (AAFA) and alkali-activated slag (AAS) have a distinct set of advantages and limitations. GBFS based AAM can attain relatively high strength under ambient conditions but due to its high reactivity, it is prone to rapid setting, shrinkage and microcracks. GBFS is also not available abundance and is more expensive than FA. Thus, FA based AAM prove to be a more efficient type of binder to use. However, AAFA have low reactivity under ambient conditions, which restricts its application to the pre-cast industry.
Various patent documents teach of the use of AAFA in this field. In particular, Chinese patents CN105523723; CN106746826; CN104829200; CN102303036; CN101880151 and CN101830653 and US patent U.S. Pat. No. 5,565,028 disclose AAFA to produce alkali-activated cementitious compositions. None of these prior art documents teach of a cementitious composition whereby the materials are combined in one step in situ.
In view of the foregoing discussion, there is thus a clear need in the art for an improved concrete composition that does not suffer from the disadvantages and shortcomings associated with conventional compositions and the teachings of the prior art.
The present invention thus aims to solve said disadvantages and shortcomings in the prior art by providing a novel cementitious composition.
According to a first aspect of the invention, there is provided a cementitious composition which includes:
It will be appreciated that any suitable source of fly ash may be used, including but not limited to unclassified fly ash (UFA), classified fly ash (CFA) and a combination thereof.
In an embodiment, the source of amorphous (reactive) silica present in the composition may be from 1 to 20 wt % of fly ash mass.
In an embodiment, the Na2CO3 present in the composition may be between 3 and 15 Na2Oeq wt % of fly ash mass.
In an embodiment, the source of calcium present in the composition may be between 3.5 and 20 CaO wt % of fly ash mass.
In an embodiment, the composition may be cured at ambient temperature in situ on, for instance, a construction site. It also can be cured at elevated temperature at a precast production site.
In an embodiment, the present invention provides for the composition, as disclosed herein, for use as a one-part “just add water” cementitious composition.
According to a second aspect of the invention, there is provided the use of the cementitious composition, as defined and described in accordance with the first aspect of the present invention, in the manufacture of concrete, mortar, grout and the like.
According to a third aspect of the invention, there is provided a process for the preparation of a dry mixed ready-to-use cementitious composition, the process including the steps consisting of:
According to a fourth aspect of the invention, there is provided a concrete, mortar or grout product or the like prepared according to the process as defined and described in accordance with the third aspect of the present invention.
These and other aspects of the present invention will now be described in more detail herein and below.
The invention will now be described in more detail, by way of example only, with reference to the accompanying figures in which:
The invention will now be described with reference to the following non-limiting examples.
1.1. Experimental Setup
The experimental setup dealt with performing a detailed analysis of the novel composition of the present invention by mathematical design of experiments (MDE). Central composite designs (CCD), also known as Box-Wilson designs, are highly efficient and flexible. MDE allows for modelling and analysis of the response of interest as well as providing satisfactory information on the experimental variables and their effects and error with minimal number of runs.
CCD included an embedded factorial or fractional factorial design with centre points that is augmented with a group of axial/star points that allows curvature estimation. This type of design can be used to estimate the first- and second-order terms/effects as well as model a response variable. Using a 2-level CCD example, it will contain a centre point (circle), four factorial points at its corners (the +1 and −1 coded variable levels shown by triangles) and four axial/star points (diamond) that lie a distance +α and −α on each axis. This is shown in
There are three kinds of CCD (inscribed, circumscribed, faced) where each type uses a different a value. For this experimental design, a face-centred CCD (α=±1) was chosen as the region of interest encompasses the extremities and a 3k level CCD was chosen. A visual representation of the box is shown in
As mentioned earlier, a 3-level CCD with 3 centre point runs was chosen for the experimental program. Table 1 states the chosen control and response variables and the constant parameters. Table 2 gives the coded values of the control variables and Table 3 states the values of the constant parameters.
The choice of the response variable is dependent on the system being studied. Mechanical strength provides a satisfactory indication of the degree of reaction of the system's final product, alkali aluminosilicate gel. For this experimentation, the average of three compressive strength tests done on different days in accordance to SANS 50196-1. Control variables calcium hydroxide, sodium carbonate and silica fume were chosen to be the significant factors to affect the strength gain.
Unlike GBFS, which has latent hydraulic properties due to its high calcium content, fly ash is shown to have pozzolanic activity and will exhibit cementitious properties in the presence of lime, hence the addition of Ca(OH)2 as a control variable. However, the activation of fly ash with Ca(OH)2 has low strength development. Shi & Day (2000) used different chemical activators to improve the strength development of FA-Ca(OH)2 binder system. The authors incorporated Na2SO4 to increase the pH of the solution as the addition of Na2SO4 with Ca(OH)2 gives the chemical reaction:
Na2SO4+Ca(OH)2+2H2O↔CaSO4.2H2O↓+2NaOH (2)
The formation of NaOH is responsible for increasing the pH of the solution, promoting an increase in the degree of early activation. However, the use of Na2SO4 produces durability problems such as sulfate attack; and thus Jeon et al. (2015) incorporated the use of Na2CO3. The authors found that the same pH increasing effect was present and there was a noticeable improvement in the compressive strength. Jeon et al. (2015) state the strength was approximately 4-5 times higher than samples containing no Na2CO3. Thus, to further quantify the effect, this experimentation used different amounts of Ca(OH)2 and Na2CO3 to see the variability in the results.
In Jeon et al. (2015) investigation, the system was thermally activated. However, the aim of this research is to produce a binder than can be cured at ambient conditions on site. Thus, curing conditions were limited to ambient temperature of 25±1° C. and relative humidity of 85±5%.
1.2. Materials
Fly ash, sodium carbonate, hydrated lime and silica fume were procured from local distributors in South Africa. Table 4 represents the chemical composition of fly ash and silica fume.
1.3. Sample Preparation and Testing Procedures
1.3.1. Design Matrix and Sample Preparation
The binder components (FA, SF, Ca(OH)2 and Na2CO3) were intermixed in a ball mill for approximately 5 minutes to limit the effect of grinding on performance of the binders.
The design matrix of the face-centred CCD with 3 centre points is shown in Table 5.
The replication of the centre point runs provides an estimate of the experimental error. The face-centred CCD was performed in random order to eliminate the possibility that one run depends on the conditions of the previous run or have an influence of subsequent runs.
Binder pastes were mixed in a Hobart mixer. The dry binder was added to a damp bowl and then water was added with the mixer running. Mixing was done for 5 minutes and then mixer was stopped and all dry material was scrapped off the bowl surface. The mixer was run again for another 5 minutes until the paste was completely mixed.
Prismatic moulds were used to cast 40×40×160 mm samples. The pastes were placed in the moulds, compacted on a vibrating table until there were no air bubbles, covered in film and left to set overnight at 25±1° C. The samples were then demoulded the following day and cured in a room with an ambient temperature of 25±1° C. and relative humidity of 85±5%. The samples were first weighed in air then weighed in water before strength tests were conducted. Compressive strength tests were carried out on three halves of the prism samples at 3, 7 and 28 days. The test procedure was done in accordance to SANS 50196-1.
1.3.2. Statistical Analysis Procedure
The first stage of statistical analysis deals with interpreting the variable effects. This aids in providing necessary information on which variables and interactions may prove to be important terms to include in the model.
Thereafter, the initial model is formed which is a full model with all factors and interactions included. For this investigation, a full quadratic model was chosen to include possibility of curvature. Then analysis of variance (ANOVA) is applied to test the significance of the model. This is quantified by looking at the p-value, which is the converted F-statistic that is derived from the mean squares. If the p-value is less than 0.05, the model (and its terms) is deemed significant. If it is greater than 0.1, then the model terms are statistically insignificant. The next stage deals with model refinement where insignificant terms in the model are removed and ANOVA tests run until only significant terms remain in the model.
Thereafter, residual plots are analysed to confirm model assumptions and check its adequacy. This is an important part of the statistical analysis as it examines whether the model in question provides an adequate representation of the true system and it does not violate the least squares regression assumptions. The least squares regression assumptions are:
1. Normality assumption;
2. Constant variance assumption;
3. Independence assumption.
The normality assumption is checked with the normality probability residual plot. If the graph follows a straight line then the assumption holds true and the model follows a normal distribution. If there is a s-shape or any abnormal shape then the normality assumption is violated.
The constant variance assumption is validated by the residuals versus predicted response graph. This plot should show a random scatter to satisfy the assumption. A presence of any sort of trend such as a megaphone shape indicates the assumption does not hold true and suggests an inequality of variance.
Plotting the residuals versus experimental run validates the independence of assumption, which again should show a random scatter with no noticeable trends. If all the assumptions hold true then the model can be deemed adequate and analysis of response surfaces and contour plots can be performed (Montgomery, 2011).
1.3.3. Characterization Techniques
Characterization techniques of X-Ray Diffraction (XRD), Fourier-Transform Infrared Spectroscopy (FTIR) and Scanning Electron Microscopy (SEM) were performed. These techniques were used to evaluate the microstructural and mineralogical changes over time for selected samples at 3 and 28 days. This provides information on the different compressive strength values observed during the study.
The samples required to be prepared before the characterisation techniques could be conducted. After being tested for strength at 3 and 28 days, representative samples for selected binders were collected, then they were dehydrated to prevent further material evolution. This was achieved by placing the samples in acetone to stop the hydration and then transferred to a vacuum chamber to filter out the acetone. Thereafter, the samples were placed in a desiccator at 25° C. to completely dry the sample until testing date (Ismail, et al., 2013). Samples for XRD and FTIR require the material to be in powdered form. This was attained by hand milling the dried samples using a pestle and mortar.
X-Ray diffraction analysis (XRD) allows for crystalline phase determination. The samples were prepared according to the standardized PANalytical back-loading system, which provides nearly random distribution of the particles. The samples were analyzed using a PANalyticalX′Pert Pro powder diffractometerin θ-θ configuration with an X′Celeratordetector and variable divergence- and fixed receiving slits with Fe filtered Co-Kα radiation (λ=1.789 Å). The phases were identified using X′PertHighscore Plus software. The data was recorded in the angular range 5°<2Θ<90°. The relative phase amounts (wt %) were estimated using the Rietveld method (X′PertHighscore Plus software).
Microstructure evolution can be further studied with Fourier-Transform Infrared Spectroscopy (FTIR). This characterisation technique provides information on the vibrations generated by the chemical bonds in a material. Lee and Van Deventer (2002) comment that it is necessary to follow the progressive change of chemical bonds within a structure and then correlate that with the physical characteristics observed. Criado et al. (2012) discovered a correlation between the pendulum movement of the main band of the FTIR spectra and compressive strength development of alkali-activated fly ash, over time. The FTIR of the selected samples was recorded on a solid state by a VERTEX 70v spectrometer equipped with the Golden Gate diamond ATR cell (Bruker). FTIR was recorded on the 4000-400 cm−1 spectral region, with 32 acquisitions at a 4 cm−1 resolution. Chemical bonds that correspond to the spectra bands were identified according to literature.
The SEM analysis was conducted on Zeiss Ultra plus SEM at an accelerating voltage of 1 kV. The SEM images were captured for fractured samples.
1.4. Results and Discussion
1.4.1. Face-Centred CCD Results and Analysis
The computer software, Minitab® 17.1.0, was used to assist in the design, analysis and interpretation of the 3k face centred central composite design. The model was analysed first as a full quadratic model which includes first- and second-order terms as well as two-factor interaction terms. This was done to incorporate any presence of curvature in the system.
It is often required that the impact of different factors be expressed as effect sizes and represented graphically. This helps in understanding which factor did or did not significantly influence the response, which was the 28-day compressive strength. The main factor effects can be calculated as the difference between the factor average and the grand mean. This is represented graphically in
The graph in
From
However, if there are significant interaction terms then it is not possible to fully interpret the main effects without looking at the interaction plots.
An interaction plot shows the relationship of how one factor and the response depends on the value of another factor. For interaction plots, analysis is conducted by looking at both lines of an interaction and compared. If both lines are parallel then the interaction effect is concluded to be zero. However, the more different and steeper the slopes become, the greater the influence it has on the results. Furthermore, if the lines cross each other, it is known as disordinal interactions and if they do not then they are called ordinal interactions (Stevens, 2000). Each criterion alters how the factors are interpreted.
From
As there are interaction effects, it is possible to deduce simple main effects and check whether it is significant or not. This is evaluated by looking at the average of the circle, diamond and square points at each factor level on the x-axis and comparing them. This is shown by the purple line in
From
Evaluating the simple main effect of Na2CO3 by taking average of the square, diamond and star points at each factor level (line in
From
Evaluating the simple main effect of SF by taking averages of the points at each factor level (line in
The possible significance of the interactions and simple main effects will be checked via analysis of variance (ANOVA).
The analysis of variance (ANOVA) was computed first to test the adequacy of the model as shown in Table 8.
As seen from Table 7, the F-value of the model is 52.41 implying that the model is significant and the probability that this value was due to noise is less than 0.01%. Furthermore, the lack-of-fit P-value is 0.286 which indicates that it is insignificant as p>0.05. The insignificance of the lack-of-fit implies that the model fits the experimental data and the factors under consideration have a considerable impact on the response. However, it can be noticed that there is an insignificant term in the model (Na2CO3*Na2CO3). This was removed and the reduced model ANOVA results are shown in Table 8.
The reduced model has a F-value of 65.34 which is greater than the initial model and still implies the model is significant. Moreover, the lack-of-fit is still insignificant (p=0.324>0.05) which again shows the adequacy of the model. Table 9 gives a comparison of the summary statistics for ANOVA of the initial model vs. reduced model.
The R2 of the reduced model indicates it can explain 98.49% of the model's variability. A large R2 is usually a good indicator of the model's fit. However, the value of R2 can be artificial increased as more predictors get included in the model without any real improvement to the system. Hence it is beneficial to compare the adjusted R2 to the original R2. The adjusted R2 is a modified version of the normal R2 and will only increase if the new term improves the model, if it has a strong correlation to the response. It decreases when the predictor does not have a strong correlation to the response. The adjust R2 will always be less than R2. From Table 9, the adjusted R2 is 96.99% which is relatively high and further emphasises the model's adequacy and that non-significant terms have not been included.
The predicted R2 is used, in conjunction with the adjusted R2, to determine if there are too many predictors in the model and if it has been over-fit and is modelling random noise. An over-fit model will usually have a high R2 value and low predicted R2. From Table 9, the predicted R2 is 88.12% which shows the models high predictive capability and confirms that the model has not been over-fit. It is also within the acceptable margin of 0.2 from the adjusted R2.
To further confirm the adequacy of the model, it is required to assess the residual plots. A residual is the difference between the observed value and its fitted value. Analysis of residual plots assists in confirming the assumption that the errors are approximately normally distributed with constant variance and whether additional terms would need to be added to the model (Montgomery, 2011).
The graph in
Response=deterministic+stochastic
where the deterministic part is the fit and stochastic part is the error. The error part is most commonly assumed to be normally distributed. Therefore, a graph of a normal probability plot is generated from the residuals of the fitted model.
The graph in
Another assumption required to be validated is where the errors are said to independent of each other. A plot of residual vs. order plot is required to validate this assumption, which is a plot of the residuals versus the order in which the data was collected. This is shown in
With the model statistics and assumptions satisfied after analysis of the residuals, the final model can be used to navigate the design space.
The final empirical model equation is given in Eq. 9 and Eq. 10 as coded and actual units, respectively. Coded variables help with interpretation of the model and variable effects as the magnitude of the coefficients are given on a common scale.
Y=30.200+3.981·A+3.593·B+1.256·C−6.295·B2+5.830·C2+5.735·A·B+4.047·A C−3.952·B·C (3)
where,
Y—Compressive strength (MPa)
A—Na2CO3 (wt %), −1≤A≤1
B—Ca(OH)2 (wt %), −1≤B≤1
C—Silica Fume (wt %), −1≤C≤1
Y=23.82−7.106·A+6.41·B−4.471·C−0.2518·B2+0.2332·C2+0.3829·A·B+0.2698·A·C−0.1581·B·C (4)
where,
Y—Compressive strength (MPa)
A—Na2CO3 (wt %), 6≤A≤12
B—Ca(OH)2 (wt %), 10≤B≤20
C—Silica Fume (wt %), 5≤C≤15
1.4.2. Compressive Strength
The surface and contour plots are given in
The first trend of high early strength is in correlation with Jeon et al. (2015) where the author's samples gained considerable strength at 3 days (28 MPa) but did not develop as aggressively and only increased in strength by 29% at 28 days (36 MPa). The second trend was noticed by other authors (Abdalqader et al., 2016; Wang et al., 1994; Fernández-Jiménezet al., 1999; Li and Sun, 2000) and that is due to the lower pH of Na2CO3. It can demonstrate a lower strength in the beginning but a higher strength at later stages due to the effect of CO32− ions, which causes the formation of carbonated compounds that improves mechanical strength.
Huang and Cheng (1986) mentions that in a FA-Ca(OH)2—H2O system, the hydration rate is low. The system's hydration rate increased by 1.5% in 7 days and was less than 20% after 180 days. The reasoning stated was that unlike slag, FA has a lower content of CaO; higher aluminium and silicon content, which requires a higher degree of polymerization. Thus, the hydration activation is much less than that of slag. Furthermore, in a system of FA-Ca(OH)2, the pH is less than 13. Fraay and Bejen (1989) demonstrated that a pH value of 13.3 is required for appropriate dissolution of alumina and silica species. Therefore, it is required that other additives be added to the system to increase the alkalinity (Li et al., 2000).
Sodium carbonate, Na2CO3, compared to other conventional activators such as NaOH and Na2SiO3 has a lower pH. Thus Na2CO3, having a pH of 12.5, does not contain sufficient dissolution power in a FA based system.
However, by combining Na2CO3 with Ca(OH)2 the following chemical balanced equation is formed:
Na2CO3+Ca(OH)2→2NaOH(aq)+CaCO3 (5)
From Equation 11 it can be noticed there is formation of NaOH which increases the alkalinity of the mixture, although not a dominant factor, which promotes dissolution of the silica and alumina species in the FA. Furthermore, the CaCO3 can act as filler material to reduce porosity and promote higher mechanical strength. However, considering the Gibbs energy of the chemical equation, it can be argued that such a reaction does not occur under ambient conditions. But the complexity of the system could cause the reaction to occur or other reactions due to the inclusion of SF which has a high reactivity due to its surface area. Jeon et al. (2015) had a similar blend (FA+Ca(OH)2+Na2CO3) but was cured thermally at 60° C., which attained 36 MPa at 28-days. The highest strength sample in this study attained 44.2 MPa at 28 days under ambient conditions.
Samples that contained a higher content of sodium carbonate had a higher 28-day strength value than those with a lower content. This improvement could be explained by the fact that Na2O content increases which increases the pH. An increased pH aids in the dissolution of the silicon and alumina species, which is responsible for strength development due to increased gel formation in the samples (Wang et al, 1994; Li and Sun, 2000). Overall both Na2CO3 and Ca(OH)2 and their interaction (line in
Silica fume, on the other hand, had a positive and negative impact on the mechanical strength depending on the quantity. Silica fume's interaction with Na2CO3 was overall positive (line
The following sections deal with the characterization techniques used to evaluate the microstructural and mineralogical changes over time for selected samples at 3 and 28 days. The selected samples were SC12|CH20|SF5, SC12|CH10|SF5 and SC6|CH20|SF5, which gained high strength, low strength and moderate strength, respectively.
1.4.3. X-Ray Diffraction (XRD) Analysis
The graph in
Fly ash consists of stable but un-reactive crystalline phases such as mullite and quartz as well as reactive amorphous phases. Therefore, during reaction, the reactive part undergoes alkali activation whilst the un-reactive phases act as micro-aggregate in the final matrix (Kumar et al. 2017; Duxson et al. 2005). The major peak at approximately 31° is due to quartz that was present in the fly ash.
Portlandite, which is calcium hydroxide, gets consumed and lower amounts are present at 28 days as seen in the XRD results. Sample SC12|CH20|SF5 at 3 days (
Walkley et al. (2016) suggests that increasing the calcium content in precursors enables greater formation of low-Al, high Ca containing C-(N)-A-S-H with lower mean chain length (MCL) and AFm type phases with little evolution of binder chemistry at later stage. Authors mention that in general, increasing the Ca content in precursors impedes the formation of N-A-S-H and AFm type phases and in Al-rich samples, promotes the formation of portlandite and Al-rich reaction products. However, the diffractograms did not show any zeolitic formation and the hydration product for this type of binder system is most likely to be a C-S-H variant.
As the peaks of portlandite decrease/disappear with increasing age, there is progressive consumption which can be facilitated by lime-consuming C-S-H formation and also due to the available silicon in the reaction mixtures, forming additional gels (Jeon et al. 2015). The increase in calcite amount over the time period can also decrease the amount of portlandite due to carbonation. This is proven by the calcite peaks noticed in the 28-day samples in sample SC12|CH20|SF5 and SC6|CH20|SF5 at position 34° in both diffractograms. In sample SC12|CH10|SF5 there was no calcite reflection at 28-day but the portlandite reflection at 3-day completely disappears, which can be most likely attributed to its consumption and formation of C-S-H type gels.
Carbonate salt gaylussite, sodium-calcium carbonate, was identified in the samples which indicate cation exchange reaction between the precursor and activator. In the alkali-activation, the Ca2+ ions provided by calcium hydroxide must react with the CO32− from sodium carbonate to form carbonate salts such as calcite and gaylussite (Eq. 6) such that the pH can be increased through the release of OH− ions. After gels start to precipitate, the gaylussite dissolves as a result of decreasing concentration of gaylussite CO ions in the aqueous phase. The carbonate then re-precipitates as CaCO3 polymorphs such as calcite. Thus, gaylussite is a transient phase that decreases/get consumed as more stable carbonates form at later stages as shown in Eq. 7 (Ke et al., 2016; Yuan et al., 2017; Bernal et al., 2014). This is in correlation with the XRD patterns in
5H2O+2Na++Ca2++2CO32−→Na2Ca(CO3)2.5H2O(gaylussite) (6)
Na2Ca(CO3)2.5H2O→CaCO3+2Na++CO32−+5H2O (7)
In conclusion, the mechanism noticed is the formation of large quantity of gaylussite for all samples which transforms into carbonate phases at later age, noticed by the lower gaylussite peaks and formation of calcite peaks. The portlandite consumption can be tied to the formation of C—S—H type gels. This consumption followed by calcite formation can explain the difference in strength noticed for the three samples. For the high strength sample (SC12|CH20|SF5), portlandite peak intensity was the highest when compared to the other two. Furthermore, unlike the other two samples, portlandite peaks still remained at 28 days showing that more gels could form and the sample can gain more strength. Furthermore, for SC6|CH20|SF5, calcite formation was occurring already at 3rd day and its peak intensity increased at 28 day. This shows the gaylussite transformation was occurring much earlier on when compared to SC12|CH20|SF5 and SC12|CH10|SF5.
1.4.4. SEM Analysis
One of the main trends noticed in the experiment was that samples that gained higher strength at 28-day had lower strengths at early age. The opposite occurred for samples that gained lower strengths at 28-day.
For the samples at 3-day, it can already be noticed that there is a large disparity in the amount of unreacted particles. Mix SC12|CH20|SF5, which gained the second highest strength at 28 days but low 3-day strength, has a noticeable amount of unreacted FA and SF particles as well as sodium carbonate crystals. Mix SC12|CH10|SF5, which gained the lowest 28-day strength, showed similar characteristics to the sample SC12|CH20|SF5 SEM image but the amount of unreacted particles were more prominent. However, mix SC6|CH20|SF5, which had one of the highest 3-day strength of all the samples, showed a significantly lower amount of unreacted particles, comparatively. The SEM image of SC6|CH20|SF5 (3D) shows a more developed microstructure with large amorphous area at early age compared to the other two, which can explain the higher strength at 3-day.
Investigating the 28-day samples, SC12|CH20|SF5 had scarce amount of unreacted particles and mostly consisted of a dense amorphous microstructure. This shows that most of the unreacted particles noticed in the 3-day samples had reacted and could be the reasoning behind on why it gained high strength. Mix SC12|CH10|SF5 which had similar characteristic to mix SC12|CH20|SF5 at 3-day did not, however, show a similar trend. There is still copious amounts of unreacted FA and SF particles as well as sodium carbonate crystals. Sample SC6|CH20|SF5 showed the same characteristic as its 3-day counterpart but contained a larger quantity and size of unreacted particles when compared to mix SC12|CH20|SF5. This shows that it did not undergo a large dissolution phase unlike sample SC12|CH20|SF5, which caused the strength value to be lower in comparison.
Comparing the microstructure of mix SC12|CH20|SF5 between 3-day and 28-day, it can be noticed the porosity of sample decreased over time. This can be attributed to the formation of calcite and its formation noticed in the XRD diffractograms. This confirms the theory that calcite acts a pore-filling material as mentioned earlier. Similar trend is noticed in the other two samples but the porosity is slightly higher. Moreover, the reduction in porosity can be also be attributed to the inclusion of SF. The particles of SF are spherical and very fine. This implies it has a large surface area and reacts very readily. Improved strength is noticeable due to the packing effect of fine and spherical SF that acts as a filler material for voids. This creates a more compact microstructure. Furthermore, they also act as nucleation sites for reactions and thus will promote the formation of amorphous gels (Nurrudin et al., 2010; Rashad & Khalil, 2013; Sayed & Zedan, 2013; Songpiriyakij et al., 2011).
Moreover, since no additional adjustments were made to accommodate the increase in SF content, the samples were prone to excessive self-desiccation and cracking due to its large surface area (Khater, 2013; De Silva et al., 2007). Rashad and Khalil (2013) observed similar scenario where a 5% SF replacement created a denser and more homogenous microstructure and had the highest strength. However, with increasing SF, the microstructure deteriorated and caused lower mechanical strength but was still better than with no SF inclusion. Aydin (2013) also had a similar situation where a 20 wt % SF replacement caused a porous structure with disconnected pores. Thus, it can be said that SF could prove to be detrimental at higher levels but beneficial at lower levels.
In conclusion, the SEM analysis correlated well with the XRD data. The moderate strength sample SC6|CH20|SF5 at 3-day had a lower amount of unreacted particles and a more homogenous microstructure, which can be attributed to the gaylussite transformation into carbonate species as it was the only sample to show calcite peaks at 3-day. This could also imply gel formation took place earlier and it did not gain additional strength unlike SC12|CH20|SF5. SC12|CH20|SF5 gained higher strength due to the denser microstructure at 28-day and this can be connected to the largest portlandite peaks noticed at 3- and 28-day in the XRD data for this sample. SC12|CH6|SF5 still had unreacted after 28-days showing full dissolution had not taken place or was taking place at a slow pace. It was also the most porous sample which can again be tied to no calcite peaks noticed in the XRD data at 28-day in
1.4.5. FTIR Analysis
Fourier transform infrared spectroscopy (FTIR) enables interpretation of the degree of polymerization of the reaction products and making conclusions about the types of structures present. The vibration of molecular bonds will differ depending on the absorption of photons at appropriate energies. Thus, this allows relating structures to the energy that is absorbed and define the functional groups of the product. Molecular bonds will exhibit narrow peaks if the structure is highly crystalline. If the structure is amorphous, the IR spectra will show wider/broader bands/peaks (Dakhane et al., 2017).
Noticing the IR spectra of the raw blended mix for all samples, there is an absorbance band around 1080 cm−1 wavelength. This is attributed to asymmetric stretching vibration mode of Si—O-T (T=tetrahedral Si or Al) caused by the presence of FA in the blended mixture. For all investigated samples, this main band shifted to a lower wavelength with increasing age. This is known to be a common occurrence in FA based activated materials as it represents the extent of the reaction/formation of gels due to the effect of activators on the original material, fly ash (de Vargas et al., 2014; Jang and Lee, 2016; Siyal et al., 2016).
For SC12|CH20|SF5 the main band in the blended mix appears at 1086 cm−1 and shifted to 954.7 cm−1 at 3-day and then to 965.4 cm−1 at 28-day (
Therefore, for SC12|CH20|SF5, the reversal in band to a higher wavelength is noticed and is in the range of 950 to 970 cm−1 which implies it may have a C-A-S-H type gel. Similarly, for SC6|CH20|SF5, the main band appeared at 1076 cm−1 and moved to 965.6 cm−1 at 3-day and then to 975.2 cm−1 at 28-day (
For SC12|CH10|SF5, the main band appeared at frequency 1079 cm−1 which shifted to 1019 cm−1 at 3-day but then moved to a lower 1003 cm−1 at 28-day (
The series of small bands appearing around 790 cm−1 and 660 cm−1 can be attributed to symmetric stretching of O—Si—O bonds and the strong peaks appearing around 560 cm−1 can be associated with symmetric stretching of Al—O—Si bonds. This is related to the presence of quartz and mullite from the raw fly ash in the raw blended samples (Jang and Lee, 2016).
The present invention considers trying to eliminate the use NaOH, which is caustic in nature. The process of producing alkali-activated binders has its drawbacks that prevent the process from being accepted in the industry: the use of highly alkaline activators like NaOH and Na2SiO3 to attain alkali activation possess problems pertaining to handling and storage (Dakhane, et al., 2017).
The novel composition of the present invention uses a low impact activator sodium carbonate (Na2CO3) and hydrated lime (Ca(OH)2). Sodium carbonate is a naturally occurring mineral and can be obtained from sodium carbonate-rich brines or chemical processes such as the Solvay process. The worldwide total of natural occurring sodium carbonate amounts to 24 billion tonnes (USGS, 2017). Moreover, Na2CO3 is reported to be 2-3 times cheaper than NaOH or sodium silicate. In addition to the low cost of this activator compared to conventional ones, it is safer to handle and has been reported to yield lower drying shrinkage. Thus, the use of Na2CO3 as an activator can contribute to the development of more sustainable AAMs (Abdalqader et al., 2016).
The efflorescence formation of the FA-SF blend reacted with Ca(OH)2 and Na2CO3 (FA|SF|CH|SC) was little to none. The FA|SF|CH|SC blend, produces a dry AAM cement powder that can be mixed together and stored in bags just like OPC cement bags.
The FA|SF|CH|SC blend gained a maximum strength of 44.2 MPa and lowest strength of 14.2 MPa at 28-day.
Sodium carbonate and Ca(OH)2 were the main factors that positively affected the mechanical strength of the FA|SF|CH|SC blend with increasing content.
The main trend noticed in the FA|SF|CH|SC blend was that samples containing a lower quantity of Na2CO3 gained high strength at 3-day but did not undergo much reaction afterward. The opposite of this was also true. This shows Na2CO3 is required to react with Ca(OH)2 to increase the pH of the system and help with the dissolution of the Si and Al precursors in the source materials.
XRD analysis shows crystalline phases of gaylussite and calcite in the FA|SF|CH|SC blend. Gaylussite is a transient phase that decreases/get consumed as more stable carbonates form at later stages.
Calcite acts as an inactive filler material and do not contribute to the formation of the gel structure. Samples that gained high strength at 3-day but did not undergo much reaction afterward was because of early reaction due to the lower Na2CO3 content. This causes a lower quantity of gel formation but an earlier reaction phase. Therefore, the gaylussite had a longer time to transform to its stable carbonate species, such as calcite, which enabled the mixes to gain better strength as seen by the characterization techniques.
In conclusion, the novel composition of the present invention proves to be a viable FA-based AAM system that provides a cost effective and efficient building material that can gain sufficient strength under ambient conditions. Furthermore, this type of blend can be created as a dry material that can be stored and distributed in bags and only requires the addition of water to form the building material.
Whilst only certain embodiments or examples of the instant invention have been shown in the above description, it will be readily understood by a person skilled in the art that other modifications and/or variations of the invention are possible. Such modifications and/or variations are therefore to be considered as following within the spirit and scope of the present invention as defined herein.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2019/07315 | Nov 2019 | ZA | national |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/IB2020/060350 | 11/4/2020 | WO |
