Thermodynamics and Cement F. P. Glasser University of Aberdeen APDIC meeting, London 24 June 2012
Nature of Portland cement (1) Contains (>92-95%) of four oxides: CaO, Al2O3, Fe2O3 and SiO2. These are supplied as limestone, shale, etc. After pretreatment the batch is heated in a rotary kiln to >1350 C; partial fusion occurs Equilibrium is closely attained at peak temperature
Nature of Portland cement (2) Equilibrium assemblages were studied in period from ~1910 onwards Four solid phases form: Phase notation Approx. composition Alite Ca3SiO5 Belite Ca2SiO4 Aluminate Ca3Al2O6 Ferrite Ca2(Fe,Al)2 O5 N.B. Despite chemical analyses showing ~64-68 wt% CaO, free lime is essentially absent after clinkering
Portland cement (3) The clinker is cooled rapidly and --except for rapid polymorphic transformations-- the high temperature solids are effectively quenched. The indurated clinker is interground with 2-5% calcium sulfate (gypsum, anhydrite) to 3000-5000 cm²/g (thus adding a fifth component. sulfate) To use, the powder is reacted with water (thus adding a sixth component)
Considering only the anhydrous clinker- Non- equilibrium is indicated by: The presence of unconsumed reactants (residual silica( quartz), free CaO, etc. Preservation to ambient of phases with lower limits of thermal stability, e.g. alite. Preservation of polymorphs stable only at high temperature, e.g., α or α belite. Liquid phase freezes independently of reaction with solids Some of the metastable features are encouraged as they improve reactivity!
The Bogue calculation (1) Was a pioneering approach to applying equilibrium phase relationships to predict the mineralogy. It could be used without requiring knowledge of phase diagrams Need to know the phase compositoin arose because many of the clinker properties were clearly functions of clinker mineralogy, e.g., heat of hydration.
The Bogue calculation (2) Bogue proposed a simple test of equilibrium be applied : measure the clinker free lime, content which is nominally zero. Free lime is readily determined by a selective dissolution method: the free lime thus determined could be subtracted from the chemical total and the calculation progressed sequentially. The abstract gives coefficents for the calculation.
The Bogue calculation (3) In recent decades, instrumental methods have been used directly to determine phase compositions. Comparing the two methods, It is generally found that Bogue under-predicts clinker alite. We know the reason for this- Bogue used data for the system at the eutectic with minimum enthalpy. During cooling, the liquid should crystallise and resorb some earlier formed alite, but in fact resorption is a slow process
Bogue calculation (4) Barry and Glasser (2000)* showed that calculation using the minimum enthalpy accounted qualitatively for the differences. Despite differences, Bogue is still used as a tool with which to proportion raw materials. With fine tuning for the quenched state and a more modern database on which to base calculations, Bogue is consistent with experiment when applied to equilibrated cements. *Advances in Cement Research
Thoughts on hydration (1) If thermodynamics correctly predicts the phase composition of cement clinker, why not apply it to hydration? True, more components (water, sulfate, possibly carbonate) have to be brought into the calculations but this can be supported. Yet collectively cement science has not generally used thermodynamics even in the absence of good experimental methods. Why?.
Cement Hydration Why are thermodynamics not applied to hydration? o It is generally believed that because C-S-H, the gellike binder, is metastable it is not amenable to thermodynamic treatments o More components have to be added, e.g sulfate, water...and no data exist for the substances coexisting with C-S-H, mainly portlandite, AFt (ettringite), AFm (hydrocalumite) and an aqueous phase
Phase distribution (1) The phase distribution (4 solids plus an aqueous phase) is amenable to an analytical solution True, the C-S-H phase is unstable with respect to crystalline lime- silica- water phases but it is persistent at ambient temperature. C-S-H thermodynamic properties have been known since at least 1950 and are reproducible.
Experimental A combination of approaches has defined the composition range of the other solid phases, where necessary (Ca(OH)2 has essentially fixed composition) In the past decade a harmonised thermodynamic database has emerged. When interfaced with calculation, the database reproduces well the observed features and enables predictions about the hydrate phase assemblages which, when tested, are experimentally verified.
Who uses thermodynamics? Arguably the first serious users have been the nuclear industry where the high activity of Ca and OH condition low solubility of many radionuclide species. Experimental simulations may suffer from contamination and sluggish kinetic effects and lack a systematic basis for extrapolation. Thermodynamic approaches provide a systematic framework to integrate results, embrace a range of conditions: results can be verified experimentally
Who does not use thermodynamics? Most of the cement industry- the design of cement matrices is still done mainly on the basis of experience,. i.e, empirically.
Why the reluctance to use thermodynamics? Many reasons, but mainly because it allegedly does not correctly depict the phases formed. For example, AFm, calcium mosulfoaluminatehydrate, -comprising some 10% of hydrate mass- is entirely metastable in the C- A-H system! This conclusion is not, however, correct. AFm is stable below 8 C and its stability to higher temperatures is enhanced by partial replacement of hydroxide by carbonate, or sulfate and/or mixtures.
Impact of carbonate on AFm mineralogy: hydroxide activity is buffered by Ca(OH)2 and maximun carbonate activity by CaCO3, calcite. 1.6 molar bulk CO2/Al2O3 1.4 1.2 1.0 0.8 0.6 0.4 0.2 monocarbonate+calcite+ portlandite monocarbonate+ hemicarbonate+ portlandite Hc+ C 4 AH x +CH hemicarbonate+ C 3 AH 6 + portlandite monocarbonate+ C 3 AH 6 + portlandite calcite+ C 3 AH 6 + portlandite 0.0 Hydroxy- AFm stable 0 10 20 30 40 50 60 70 80 90 100 temperature [ C]
From calculation: Hydroxy AFm, C4AH12, is stable, but only at low temperature, <8 C in the presence of CaCO3. Hemi- and mono-carbonate (AFm phases) are stable across a wide range of temperatures and carbonate activities So the problem has been poor characterisation data and inadequate/ incomplete thermodynamic data
Changing nature of cement But the cement industry is undergoing a transition. To lower specific CO2 emissions (presently ca 750kg CO2/tonne of cement) clinker is being diluted. EU specifications (since 1995) permit adding up to 5% limestone (itself >75% CaCO3) to Portland cement
Addressing the challenge We could of course make appropriate test samples and age them for 10, 20, 50 and 100 years (or more) But the number of variables is too large to be sustained and common sense suggests that we develop a predictive capability. Hence need for developing (i) thermodynamic models and (ii) links between thermodynamic descriptions and physical/ engineering properties
Practical example: limestone in cement (1) European regulation EN197-1 permits a cement to be marketed as Portland cement where the clinker content is interground with up to 5% of limestone (itself at least 75% CaCO3). But argument has raged in North America abut allowing this: the evidence, it is alleged, is too empirical
Limestone in cement (2) But what happens at low CaCO3 contents? Can mineralogical reactions occur between cement, water and solid CaCO3? And if so, are reactions rapid or slow? What are the implications, if any for the evolution of engineering properties? The literature is in disagreement about the scope for reaction.
Jumping ahead to the conclusions... Yes, CaCO3 is reactive. The nature and extent of reactions are temperature dependent in the range 0-40 C. The mineralogy of the paste is affected. Mainly, sulfate and hydroxide in AFm and AFt phases are replaced by carbonate. Looking in more detail at carbonate/hydroxy interactions-
Finder diagram for AFm (ph buffered by Ca(OH)2) AFm not stable 1.6 1.4 monocarbonate+calcite+ portlandite molar bulk CO2/Al2O3 1.2 1.0 0.8 0.6 0.4 0.2 monocarbonate+ hemicarbonate+ portlandite Hc+ C 4 AH x +CH hemicarbonate+ C 3 AH 6 + portlandite monocarbonate+ C 3 AH 6 + portlandite calcite+ C 3 AH 6 + portlandite 0.0 0 10 20 30 40 50 60 70 80 90 100 temperature [ C]
Sulfate/ carbonate/ hydroxy reactions in the exchangible anion sites AFm: With rising temperature, 0 C to 85 C, order of stability favours sulfate >carbonate >hydroxide AFt : With rising temperature, 0 C to 85 C, substitution favours sulfate>carbonate (OH negligible at ph ~12.5) AFm/AFt; AFt destabilised with respect to AFm especially at temperatures > ~50 C
Finder diagram for representative cement at 40 C 3.5 excess portlandite present in all assemblages VI AFt +gypsum + calcite 3.0 molar bulk SO3/Al2O3-ratio 2.5 2.0 1.5 1.0 VII AFt + monocarboaluminate+ monosulfoaluminate 1) VIIIa M c+m s-ss 2) V AFt + monocarboaluminate + calcite 0.5 III VIII Ms-ss 3) + Hc+ C Mc + Hc + Ms-ss 4) 3 AH 6 0.0 0 0.25 0.5 0.75 1 1.25 1.5 molar bulk CO 2 /Al 2 O 3 -ratio 1) calc. composition Ca 4 Al 2 (SO 4 ) 0.96 (OH) 12.08.6H 2 O 2) calc. composition Ca 4 Al 2 (SO 4 ) x (OH) 14-2x.6H 2 O (0.86 x 0.96) 3) calc. composition Ca 4 Al 2 (SO 4 ) 0.86 (OH) 12.28.6H 2 O 4) calc. composition Ca 4 Al 2 (SO 4 ) 0.9 (OH) 12.2.6H 2 O at 40 C
Supplementary calculations Specific volume of solids is optimised by maximising the AFt content (region VIII, ) Achieving the best space filling corresponds to maximum strength and minimum permeability, other factors being equal. Thus carbonate additions can be used to drive sulfate into AFt (ettringite) (density ca 1.77g/cm3) which has the highest specific volume and best space filling. Conclusion seems to be generic: maximise AFt content!
Progress in acceptance of calcite North American interests had refused to change standards to allow carbonate additions But the calculations illustrated here (and only a sample is shown) were used to make a compelling scientific case for the benefist from addition of calcium carbonate. Calcium carbonate at low addition levels, >5%, behaves as a reactive admixture; it is not a filler Development of a scientific and quantitative explanation of the role of CaCO3 has helped lead to a change in Canadian and US standards Estimated savings of CO2 thus achieved are on the order of 10 million tonnes /year.
Results Application of thermodynamics, supported by focussed experiment, has been accepted as a basis for changing standards Probably the change in standards would eventually have occurred but the work described briefly here has shortened the time required for change by perhaps 5-10 years
Challenges World consensus is that higher replacement levels of cement by reactive waste materials such as fly ash and calcined clay must be used partially to replace Portland cement These materials differ from Portland cement in chemistry, mineralogy, granulometry, reactivity, etc. But how can it be proved that in the long term, >100 years, blends will perform as well as or better than presently- allowed formulations?
Challenges A thermodynamic approach is, I believe, essential to the development of new paradigms Of course links need to be established between on the one hand, composition and microstructure and on the other, engineering properties of the material Working in partnership and with a deep understanding of the material is essential
Summary Applied thermodynamics is a neglected tool useful in the design of new cementitious materials and in assessing performance in their service environment. At the same time, experimental studies are needed to assess relevant kinetic and confirm the validity of calculations and of constitutional models linking to engineering properties
Acknowledgement To the organisers, for giving me a platform To many colleagues for critical discussions To Thomas Matschei, a former student who undertook many of the calculations shown here and to Nanocem, a not- for -profit organisation for financial support Questions?
New types of representation Note that the activity of one component is restrained by bulk ratio of carbonate/ aluminate, as well as limited to a maximum imposed by CaCO3 solubility at a fixed ph The selection of restraints demands some knowledge of cement formulation But the information derived form thermodynamic calulations is generic, not limited to conditions of a particular set of experimental conditions
Role of Sulfate Calcium sulfate is added to cement for two reasons: Gives a period of fluidity of the fresh mix Controls shrinkage during hardening Usually 2-4 wt% calcium sulfate added Optimum sulfate addition determined empirically to optimise the above factors