Relationship between Microstructure, Durability and Performance of CEM X Composite Cements

Cement production is a major contributor to anthropogenic CO2 emissions. Composite slag cement is one way of reducing the clinker factor, and hence emissions. Synergies between limestone and slag have been suggested to enhance the performance of such cement. The synergistic effect on mechanical performance has been ascribed to the carboalumination formation, which in turn prevents the conversion of ettringite to mono sulphoaluminate. Whilst the benefits of the synergy are documented, causation factors are not fully understood. In order to maximise the synergy, a clear understanding of the relationship between composition and the carboalumination-ettringite balance and how this impacts on performance is essential. Consequently, this study focused on the kinetics of reaction, microstructure and the implications on strength development and the freeze-thaw durability.
The QXRD/PONKCS method proposed elsewhere has been extended to monitor the residual slag contents in ternary blended systems. The slag phase was modelled from the x-ray diffraction pattern of a 100 % slag specimen. The model was then calibrated on a 50:50 mixture of slag and corundum using the internal standard method. The measurement accuracy was found to be ± 2 % following implementation of the calibrated slag phase on simplified slag-corundum mixes of varying proportions. In hydrated cement systems, however, the accuracy established from the comparison between the results from QXRD/PONKCS and SEM/IA analysis was ± 6 %. The investigated hydration stopping methods except freeze-drying had minimal impact on the results as long as the mass attenuation coefficient (MAC) was accurately quantified. For implementing the QXRD/PONKCS method, freshly ground non-hydration stopped sample preparation is recommended. This method presented the least challenge in calculating the MAC and also preserve the other hydrates particularly the AFt/AFm phases. However, if hydration stopping cannot be avoided, a double solvent exchange regime using isopropanol and diethyl ether is recommended. The residual slag content is overestimated from freeze-dried samples.
Multiple techniques including the QXRD/PONKCS method, SEM/IA, isothermal conduction calorimetry, chemical shrinkage and thermal analysis were subsequently used to investigate the kinetics of hydration. These revealed that limestone accelerates slag and clinker hydration. Alite hydration is accelerated by the supplementary cementitious materials including slag and quartz. However, while belite hydration is retarded in the binary slag mix, this is not the case in limestone ternary blends. Calcium carbonate also reacts with additionally dissolved aluminates instead of the latter converting ettringite to mono sulphoaluminate. Therefore, the synergy between limestone and slag is not only due to preserved ettringite but the additional hydrates and pore structure modification. The carboaluminate content is higher in the presence of limestone but the home to mono carboaluminate balance shifts with the limestone content. The aluminates available for incorporation into other hydrates and in the pore solution are lowered by carboaluminate formation.
The pore structure and pore solution chemistry were analysed in order to further investigate the mechanisms for the accelerated dissolution of slag and the major clinker phases. The overall pore volumes as measured by MIP were similar in the slag containing mixes with or without limestone. Evaluation of the free water content at the end of experiment also indicated excess free water thus indicating that the pore structure and capillary pore water are not the limiting factors for slag hydration in the investigated mixes. The identified drivers behind the role of limestone in the synergy are the nucleation and dilution effects in the case of clinker, while for slag hydration pore solution effects are important. The incorporation of alumina into carboaluminates and C-S-H lowers the alumina in the pore solution and consequently promote slag dissolution.
The freeze-thaw resistance of the ternary blends at the investigated 0.5 water/binder ratio is low compared to the CEM I 42.5 R concrete. Decalcification through carbonation during conditioning and leaching during freeze-thaw is identified to be the dominant microstructural difference between degraded and non-degraded samples. A chemo-mechanical degradation mechanism, which is similar to the glue-spall mechanism, is hypothesised for surface scaling during freeze-thaw. By this, freeze-thaw damage in concrete is caused by decalcification and spalling of successive layers under induced stresses of ice growth.