Studies on the compatibility of different superplasticizers with alkaline activators for low calcium geopolymer binders.

The cement industry is coming into focus, as the annual production of around 4 Gt of cement is responsible for the emission of 1.5 Gt of CO2 and thus for over 8 % of anthropogenic CO2 emissions. This leads to the search for alternative binders. Such binders are calcined clays, which are available worldwide but vary greatly in their chemical and mineralogical composition. In many studies, particularly low‐calcium metakaolin is used as calcined clay, which reacts to form a low‐calcium aluminosilicate binder when mixed with a calcium‐free alkaline activator. The adjustment of the properties in the fresh state, especially regarding the consistency of these binders, is almost exclusively achieved by the addition of water, since commercially available superplasticizers are usually ineffective in low calcium geopolymer systems. The objective of this study was to investigate various PCE superplasticizers (MPEG‐, IPEG‐, HPEG‐PCE) with respect to their stability in different alkaline activators (NaOH, KOH, sodium, and potassium silicate solutions). The effectiveness of superplasticizers in low calcium geopolymer binders was verified by rheological tests. Size exclusion chromatography was used to investigate if structural degradation of the superplasticizers occurs. The investigated PCE superplasticizers showed no liquefying effect in the low calcium geopolymer system. This is due to a degradation process, i.e., the hydrolysis of the PEG side chains depending on the alkalinity of the activator.


Introduction
The cement industry is coming into focus, as the annual production of around 4 Gt of cement is responsible for the emission of 1.5 Gt of CO2 and thus for over 8 % of anthropogenic CO2 emissions.This leads to the search for alternative binders [1,2].Cement substitutes such as calcined clays represent a promising way to reduce CO2 emissions.These materials can be dehydroxylated at low temperatures (550 °C -800 °C) and used as supplementary cementitious materials (SCM's) to reduce the amount of Portland cement clinker [3,4].Using alkaline activators like NaOH, KOH or alkaline silicate solutions the calcined clays form alkali activated binders (AAB) especially geopolymers.In particular, the AAB's and geopolymer binders are characterized by high stability to acid attack, high strengths, and durability [5,6].A major problem in the use of calcined clays is their high-water demand resulting from a high fineness and specific surface.Furthermore, these binders showed incompatibility with water-reducing admixtures like superplasticizers (SP).The adjustment of the properties in the fresh state, especially regarding the consistency of these binders, is almost exclusively achieved by the addition of water Although many investigations of SPs have been carried out in Portland cement, only a few studies have been carried out in AAB with various activators [7, 8, 9, 10, 11, 12 13].Studies on NaOH activated slags have shown that competitive adsorption occurs between negatively charged activator and the anionic anchor groups of the superplasticizer.This leads to reduced adsorption of the superplasticizers on the surface of slags, which reduces the efficiency of superplasticizer [14,15].Another reason for the low effectiveness of existing superplasticizers (sulphonated melamine formaldehyde, SMF and sulphonated naphthalene formaldehyde, SNF) is their stability to alkaline activators.Studies have shown that SMF are degraded by up to 65 % at pH = 14.SNF shows a higher stability and was degraded by up to

Abstract
The cement industry is coming into focus, as the annual production of around 4 Gt of cement is responsible for the emission of 1.5 Gt of CO2 and thus for over 8 % of anthropogenic CO2 emissions.This leads to the search for alternative binders.Such binders are calcined clays, which are available worldwide but vary greatly in their chemical and mineralogical composition.In many studies, particularly low-calcium metakaolin is used as calcined clay, which reacts to form a low-calcium aluminosilicate binder when mixed with a calcium-free alkaline activator.The adjustment of the properties in the fresh state, especially regarding the consistency of these binders, is almost exclusively achieved by the addition of water, since commercially available superplasticizers are usually ineffective in low calcium geopolymer systems.The objective of this study was to investigate various PCE superplasticizers (MPEG-, IPEG-, HPEG-PCE) with respect to their stability in different alkaline activators (NaOH, KOH, sodium, and potassium silicate solutions).The effectiveness of superplasticizers in low calcium geopolymer binders was verified by rheological tests.Size exclusion chromatography was used to investigate if structural degradation of the superplasticizers occurs.The investigated PCE superplasticizers showed no liquefying effect in the low calcium geopolymer system.This is due to a degradation process, i.e., the hydrolysis of the PEG side chains depending on the alkalinity of the activator.
10 % [16].Polycarboxylate ether superplasticizers also show insufficient effectiveness in AAB systems [17,18,19].Especially the amide and ester groups, linking the side chains to the backbone of the superplasticizers tend to be hydrolysed in a high pH medium [13].The aim of this work was a systematically investigation of the stability of various polycarboxylate ether-based superplasticizers (methacrylate ester MPEG-, methallyl ether HPEG-and isoprenol ether IPEG-PCE) as a function of the type and alkalinity of the alkaline activator (NaOH, KOH, sodium, and potassium silicate solutions), which are typical used to form low calcium geopolymer binders.

Materials
Three PCE superplasticizers (HPEG-, IPEG-and MPEG-PCE) were provided by MBCC Group (Master Builders Solutions Deutschland GmbH) for investigation.The aqueous PCE samples were frozen in liquid nitrogen and vacuum dried until mass consistency was reached.The dried samples were then ground by hand with a mortar and pestle.White powders were obtained from grinding HPEG-PCE and IPEG-PCE samples, while small yellow brown flakes were obtained from the MPEG-PCE sample.The PCE were characterized by their molecular weight distribution and their polymer content using size exclusion chromatography (SEC).The method and the calculation of the polymer content are explained in detail further below.The elution curves for all three PCE are shown in Figure 1.The weight and number average molar mass, PDI and polymer content are given in Table 1.All samples show a large initial peak, corresponding to the PCE polymers.This peak is followed by two smaller peaks, where the third peak is smaller than the second.In case of IPEG-PCE, a shoulder on the right-hand side of the second peak is formed, instead of a distinct third peak.All peaks are not baseline separated from each other.The second and third peak correspond to (macro-)monomers.These (macro-)monomers are most likely leftover reactants from PCE synthesis.Macromonomers are most likely methoxy polyethylene glycol (MPEG), hydroxy polyethylene glycol (HPEG) and isopropoxy polyethylene glycol (IPEG), while monomers are most likely carboxylic acid derivatives such as methacrylic acid (MAA) or maleic anhydrite.The SEC results are shown in Table 1.While the number average and weight average molar mass is different for all the samples, HPEG-and MPEG-PCE appear to be structurally more similar to one another than to IPEG-PCE.The PDI is differing only slightly between samples.The calculated polymer contents confirm that the samples do contain polymeric PCE as well as (macro-)monomers.
For pH neutralization of the alkaline solutions prepared as described above, concentrated hydrochloric acid (37 wt.-%, Carl Roth GmbH + Co. KG) was diluted with ultrapure water to 1, 4, 6.6, 7.8 and 8 M solutions.
A metakaolin (Metaver O, Newchem GmbH) was used for the preparation of pastes to investigate the influence of PCE's on rheological properties of low calcium geopolymer pastes.The metakaolin is mainly composed of 52.0 % SiO2 and 41.4 % Al2O3 and is free of CaO.The amorphous content is approximately 72 % with the crystalline phases mainly consisting of kaolinite with small amounts of quartz, calcite and anatase.The specific surface area determined according to BET is ≈ 11,5 m²/g while the particle size distribution has a d10 percentile of 0,75 µm, a d90 percentile of 20,22 µm and an average particle size of 8,19 µm.The pozzolanic reactivity determined by the modified Chapelle test described in the French norm NF P 18-513, Annexe A is high with 1275 ± 30 mg/g.The norm requires a value of ≥ 700 mg/g for a material to be classified as "pozzolanic".

Methods
Three PCE solutions with a solids content of 20 wt.-% were prepared by dissolving 1 g of PCE sample in 5 ml of ultrapure water.A beaker with 99 ml of alkaline solution and a magnetic stirrer was prepared and 1 ml of PCE solution was added.The concentration of PCE in the solution was therefore approximately 2 mg/ml.After set time intervals (5, 10, 15 and 30 min), 20 ml of the solution were filled in a round bottom flask and neutralized with hydrochloric acid (HCl).When alkali silica solutions were neutralized, silica gel (SiO2) was precipitated out of the solution.The samples were frozen in liquid nitrogen and vacuum dried until mass consistency was reached.The dried samples should mainly consist of sodium chloride (NaCl) or potassium chloride (KCl), as well as silica gel (SiO2) in the case of alkali silicate solutions.30 mL of ethanol (≥ 96 %) were added to each sample, which were then shaken for an hour in a horizontal shaker.While PCE are soluble in ethanol, solubility for NaCl (0.65 g/kg) and KCl (0.37 g/kg) is low and SiO2 isn't soluble in ethanol at 25 °C.The samples were centrifuged at 2215 G (Eppendorf Centrifuge 5804R) for five minutes before the supernatant was separated using a syringe.The supernatant was then filtered through a syringe filter with 0.45 µm mesh and filled in a round bottom flask, together with 60 mL of ultrapure water.The samples were then frozen in liquid nitrogen and vacuum dried until mass consistency was reached.
The molecular weight distribution of investigated PCE superplasticizers was determined by size exclusion chromatography (SEC).A SEC column (Shodex OHpak SB-805 HQ, Resonac Europe GmbH) was installed in an AF2000 MultiFlow FFF System (Postnova Analytics GmbH).The system is equipped with a Multi-Angle Light Scattering (MALS), a UV-Absorption (UV) and a Refractive Index (RI) detector and was calibrated using pullulan standards with a molecular weight of 180 -708.000Da (PSS Polymer Standards Service GmbH).
Samples were prepared by dissolution of 10 mg of dry sample in 10 ml of 0.05 % NaN3 solution.The flowrate of the SEC system was set to 0.5 ml/min and 20 µl of each sample was injected via autosampler.The MALS detector was notable to detect the macromonomers with low molecular weight.Therefore, the RI detector data was used for evaluation.
The methods accuracy was determined by analysing three samples with 1 mg/ml of PCE and three samples with 5 mg/ml of PCE for all three types of investigated PCE.The standard deviation and coefficient of variation for the weight average molar mass Mw, number average molar mass Mn and the polydispersity index PDI were calculated from the results of all six samples of each type of PCE.The values were rounded up and are given in Table 2.
A significant factor to consider in regards of the design of experiment was the possible adsorption of PCE on the surface of silica gel precipitated out of the alkali silica solutions during neutralization.A sample with sodium silicate solution and MPEG-PCE was prepared and neutralized with hydrochloric acid.The sample was centrifuged at 2215 G for five minutes and the supernatant was separated from the silica gel.The silica gel was washed five times with ethanol (≥ 96 %, Carl Roth GmbH + Co. KG).The ethanol was evaporated, and the sample was dissolved in 0.05 % NaN3 solution.The supernatant was prepared for SEC as previously explained.The SEC results showed that only small amounts of (macro)monomers were found in the sample washed off the silica gel.It was therefore assumed that PCE do not get adsorbed by silica gel to a significant degree.
The rheological properties of geopolymer pastes were determined by measurement of viscosity and shear stress with a Brookfield DV-III rheometer (AMETEK GmbH) equipped with a spindle of type SC4 29.The samples were pre-sheared of 60 seconds at 120 RPM, followed by the measuring phase, during which the rotational speed is lowered every 30 seconds while a measurement is taken every 5 seconds.The rotational speeds investigated reach from 120 to 1 RPM.The recipe of the pastes was designed in a way that the reference paste has a similar viscosity regardless of what type of alkali silicate solution is used.The powdered PCE superplasticizers were dissolved in ultrapure water, such that 1 g of the solution would contain the desired amount of PCE (0.5, 1.0 and 2.0 wt.-%) related to the amount of binder.The binder is considered as the amount of metakaolin and the amount of solids content in the alkali silica solution.The pastes were obtained by mixing 20 g of metakaolin with 18 g of sodium silicate solution or 16 g of potassium silicate solution, 1 ml of prepared PCE solution and a fixed amount of additional water with an overhead stirrer (2 min, 400 RPM).The water to binder ratio of pastes prepared with sodium silicate solution was 0.43 and with potassium hydroxide solution was 0.40.

Degradation in alkali hydroxide solutions
Exemplary shown in Figure 2 are the elution curves for PCE before and after exposure to 4 M sodium hydroxide solution for 30 minutes.The first peak of all samples, corresponding to the PCE polymers, is much smaller and a bit narrower after the exposure to sodium hydroxide solution.This indicates a degradation of the polymeric structure.
The second peak is only slightly affected, while the third peak has become much larger and a bit wider, reducing the second peak to a shoulder in case of MPEG-PCE.This indicates that the third peak corresponds to cleaved PEG sidechains from PCE and macromonomers, as well as leftover carboxylic acid derivatives from cleaved macromonomers.The lack of further peaks that appear indicates, that the PCE backbones are in the same range of molecular weight as the PEG macromonomers.These observations were made for all samples, regardless of whether they were exposed to sodium hydroxide or potassium hydroxide solution.The reduction of the first and the growth of the third peak were more intense with increasing molarity of the alkali hydroxide solution.
Considering the first peak as the polymeric portion of the sample and all other peaks as the (macro-)monomeric portion, the polymer and (macro-)monomer content of samples can be calculated.Figure 3 shows the polymer content of PCE samples exposed to 1, 4 and 8 M sodium hydroxide solution for 5, 10, 15 and 30 minutes.The polymer content is reduced over time by 1 M sodium hydroxide solution, while the degradation process seems to occur much faster in 4 M and 8 M sodium hydroxide solution.For these samples, no significant reduction can be seen aside of the initial drop after 5 minutes.However, the degree of degradation appears to be related to the molarity of the hydroxide solution.The polymer content is reduced to > 60 % in 8 M sodium hydroxide solution.Exposure to potassium hydroxide solution leads to similar results.Though, the polymer content of PCE samples exposed to 1 M potassium hydroxide solution appears to be time independent.The degree of degradation appears to be related to the molarity of the hydroxide solution as previously seen with 4 M and 8 M sodium hydroxide solution.
The polymer content is reduced to > 70 % in 8 M potassium hydroxide solution.
The number average molar mass (Mn) and the weight average molar mass (Mw) for samples exposed to sodium hydroxide or potassium hydroxide solution for 30 minutes are plotted in Figure 4 and

Degradation in alkali silicate solutions
The degradation of PCE in alkali silicate solutions was generally found to be much stronger in comparison to the degradation alkali hydroxide solutions.The exemplary elution curves of samples exposed to sodium silicate solution are shown in Figure 6.The large first peak has disappeared completely, indicating a vast degradation of PCE polymers to (macro-)monomers.This is confirmed by the very small second and very large third peak, corresponding to the (macro-)monomer content of the sample.The change in polymer content of PCE samples over time confirms the rapid and almost complete degradation of PCE polymers (Figure 7).The effect is immediate for most samples regardless of the type of alkali silica solution.The polymer content of those samples is reduced to ≤ 10 % after only 5 minutes of exposure.A longer exposure doesn't appear to affect the remaining polymers to a significant degree.However, the MPEG-PCE samples polymer content only drops to 20 % after 5 minutes of exposure to potassium silicate solution and remains at around 10 % after 10 minutes and further on.This again indicates a slightly better chemical stability of MPEG-PCE in alkaline media compared to the other PCE samples.
The number average molar mass (Mn) and the weight average molar mass (Mw) for samples exposed to sodium silicate or potassium silicate solution for 30 minutes are plotted in Figure 8 and Figure 9.A drastic reduction of Mn and Mw to average molar masses of 2 kDa or lower can be seen for all samples regardless of the type of alkali silica solution they were exposed to.The amount of alkali hydroxide in the alkali silicate solution is equal to 7.8 M sodium hydroxide or 6.6 M potassium hydroxide solution.However, the degradation process of polymers in alkali silicate solutions was much stronger then in alkali hydroxide solutions with a molarity of 8.This must be related to the silicate ions present in alkali silicate solutions.Further investigation is needed to find an explanation for this effect.

Impact on the rheology of geopolymer pastes
The dynamic viscosity and shear stress over the shear rate of the reference geopolymer pastes are shown Figure 10.
The values represent the average of four individual samples and the standard deviation is plotted.The dynamic viscosity at low shear rates is lower for pastes prepared with potassium silicate solution.Only small differences can be seen with shear rates > 1 s -1 .The shear stress is lower for pastes prepared with potassium silicate solution regardless of shear rate.However, the difference is increasing with increasing shear rate.
A stiffening effect was observed for all pastes when PCE were introduced, regardless of the type of PCE and the added amount (0.5, 1.0, 2.0 wt.-%).The stiffening effect was stronger with increasing amount of added PCE.All pastes were too stiff to be investigated with the rheometer.Pictures of pastes prepared with sodium silicate solution are shown in Figure 11.The polymer content is reduced to less than 10 %.

Figure 1
Figure 1 Elution curves of PCE samples

Figure 5 .
For all samples a decline in number average as well as weight average molar mass can be seen with rising molarity of the respective hydroxide solution.The difference in the decline of Mn and Mw between samples exposed to sodium hydroxide and potassium hydroxide solution is negligible for most HPEG-PCE and all IPEGPCE samples.The HPEG-PCE sample exposed to 4 M alkali hydroxide solutions shows a lower Mw in potassium hydroxide solution, which is considered as an artifact.The MPEG-PCE samples exposed to 1 M and 4 M potassium hydroxide solution show a much smaller decline of Mn and Mw compared to the samples exposed to sodium hydroxide solution.This indicates that the ester bonds are less prone to cleaving then the ether bonds of HPEG-and IPEG-PCE in hydroxide solutions up to 4 M.

Figure 2 Figure 3 Figure 4 Figure 5
Figure 2 Elution curves of reference PCE and PCE samples exposed to 4 M sodium hydroxide solution for 30 minutes

Figure 6
Figure 6 Elution curves of reference PCE and PCE exposed to sodium silicate solution for 30 minutes

Figure 7
Figure 7 Polymer content of PCE samples exposed to sodium silicate solution (left) and potassium silicate solution (right) over time

Figure 8 Figure 9
Figure 8 Number average molar mass and weight average molar mass of the PCE reference and samples exposed to sodium silicate solution

Figure 10 Figure 11
Figure 10 Dynamic viscosity and shear stress over shear rate of reference geopolymer pastes

Table 1
Number average (Mn) and weight average (MW) molar mass, polydispersity index (PDI) and polymer content of PCE samples

Table 2
Standard deviation for Mw, Mn and PDI in regard of the type of PCE