Changes in portlandite morphology with solvent composition: Atomistic simulations and experiment

TL;DR: In this article, a new analysis tool was developed to quantify the experimentally observed changes in morphology of portlandite, allowing the calculation of the relative surface energies of the crystal facets.

About: This article is published in Cement and Concrete Research.The article was published on 2011-12-01 and is currently open access. It has received 2498 citations till now. The article focuses on the topics: Portlandite.

Summary

1. Introduction

• The purpose of this work is to understand changes in portlandite (CH) morphology with the changing chemical environment of the growing crystal as encountered in cementitious systems.
• If these changes in morphology can be understood, they may be controlled and modified.
• Not much experimental work has been done to study the growth mechanisms.
• The addition of hydroxides led to more but smaller particles.
• The authors restricted their study to the [00.1] surface and did not attempt to estimate the surface energy.

2.1. Materials

• The following chemical products were used as received: calcium chloride CaCl2, 2H2O, calcium nitrate Ca(NO3)2, 4H2O, sodium chloride NaCl, sodium sulfate Na2SO4 (Merck, quality grade GR for analysis), and a sodium hydroxide 50 wt.% solution (Acros Organics, extra pure).
• Stock solutions at 1.0 M were prepared by diluting the desired amount of reagent with ultrapure water.
• These solutions were then filtered at 0.2 μm to eliminate dust.
• The stock solutions were stored in closed vessels to avoid further contamination, and used within a few days to avoid carbonation.

2.2. Precipitation procedure

• The reactive solutions were prepared just before the precipitation by diluting the stock solutions with ultrapure water.
• The nominal concentrations of the different ions after mixing are summarized in Table 1.
• A high stirring rate (200 to 500 rpm with a magnetic stirrer) was maintained during 5 min.
• Then the precipitation mixture was placed on rolling bars (60 rpm) for mild stirring during one hour.
• The precipitate was finally collected by filtration using 0.2 μm filters, washed with 100 ml ultrapure water, and dried for 24 h at 60 °C.

2.3. Characterization methods

• The powder morphologies were analyzed by scanning electron microscopy (SEM, Philips XL 30 FEG microscope).
• The SEM samples were prepared by dispersing the powder in ethanol.
• The suspension was sonicated for 10 min in an ultrasonic bath, and one drop of the suspension was then deposited on an aluminum support and dried in air.
• The phase identification of the precipitates was made with X-ray powder diffraction (XRD, Philips X'Pert diffractometer, Cu–Kα radiation).
• The particle size distribution (PSD) was collected using a laser diffraction method (Malvern Mastersizer S).

2.5. Atomistic simulation

• All atomistic simulations done on portlandite were done with classical potentials.
• Both static energy minimization and molecular dynamics calculations were carried out.
• The same potential set was used for bothmethods.
• It has been chosen for its good performance for both the water and the inorganic components when compared to other force fields [14].
• The following subsections describe the potential and the computational methods used in more detail.

2.5.1. Potential

• The potential used for the inorganic materials was the potential developed by Freeman et al. [6].
• The earlier forcefield [7,15,16] has been shown repeatedly to give good results for the description of inorganic oxide crystals [7,17– 19].
• Aij⋅e − rij ρij− Cij r6ij ð3Þ where q is the charge of the respective ion and Aij, ρij and Cij are constants evaluated by fitting resulting properties to experimental data such as the lattice and elastic constants.
• In addition a Morse potential (Eq. (4), D,α, and r0 are fitted constants) describes the bond between the oxygen and the hydrogen of the hydroxyl group.
• The potential used to describe the water was the Tip4P/2005 potential [20], which is a rigid molecule water force field that yields very good results for water calculations.

2.5.2. Energy minimization

• The energy minimization calculations were made with the METADISE code [21], which implements an energy minimization technique based on classical potential models.
• The total energy of the system is calculated by summing the interactions between all pairs of atoms up to a cutoff radius of 15 Å. Energy minimization is a method whereby the atomic coordinates iteratively converge toward the atomic configuration with minimum total energy.
• Thus several initial configurations have to be tested in order to increase the chances of finding a physically meaningful minimum (e.g. different possible cuts for the same surface).
• For bulk calculations, periodic boundary conditions have been applied in all directions.
• The bulk region has to be deep enough to make long-range coulombic interaction between its top and bottom atoms negligible (≥80 Å).

2.5.3. Molecular dynamics

• The molecular dynamics calculations were made with the dl_poly 2.20 code [22].
• The timestep employed for all simulations was 0.2 fs.
• The simulations were done either at constant volume and constant temperature (NVT, thermostat algorithm employed: Nose-Hoover [23] with a relaxation time of 0.5 ps) or at constant pressure and constant temperature (NPT, barostat–thermostat algorithm employed: Nose-Hoover [23] or Nose-Hoover–Melchionna [24]with a relaxation timeof 0.5 ps for both temperature and pressure).

3. Results and discussion

• 1. Identification of crystalline phase of formed particles XRD patterns of the precipitated samples were all similar to the XRD pattern of sample BC presented in Fig.
• In all cases a crystalline material was obtained, which matched the Ca(OH)2 pattern of portlandite (ICDD 04-0733).
• It was concluded that a pure material was obtained with this procedure.

3.2. Carbonation

• Two decomposition steps were observed in the thermogravimetric curves measured on all precipitated samples, as shown with sample B in Fig.
• The first weight loss occurs around 450 °C and corresponds to the dehydration of portlandite into calcium oxide CaO.
• The second weight loss occurs at a higher temperature, around 600 °C, and was attributed to the decomposition of calcium carbonate into CaO.

3.3.1. Qualitative observations

• The SEM images show that particles formed in the presence of either Cl− or NO3− were facetted and regular with three distinct families of facets.
• The size of the particles for the three samples is similar for all three samples, the median volume diameter of the measured size distributions were between 2.65 and 3.72 μm (Table 2), the only trend which might be discernible is a slight decrease in particle size with an increase in chloride concentration from 0.2 to 0.3 mol/l.
• The results are also consistent with results from Berger et al. [27] and Gallucci et al. [4] whose results indicate that the presence of sulfates promotes hexagonal platelet morphology of growing portlandite particles.
• If the concentration of the sulfate ions is decreased to 0.02 mol/l (sample BS02) the shape of most particles shows again the three different families of facets observed for samples B, BC and BN.
• This indicates that silicates have a great influence on portlandite morphology and growth.

3.3.2. Quantitative description of facetted regular particles

• The analysis method described in Section 2. made it possible to completely determine the shape of the particles and to clearly identify the facets present ([00.1], [10.0] and [20.3] surfaces).
• The results of Berger and McGregor [27] indicated that the addition of chlorides and nitrates led to an elongation of the particles in the [00.1] direction, indicating an increase of the [00.1] surface energy relative to the [20.3] and [10.0], not the other way around.
• First of all the particles observed by Berger and McGregor were much larger, of the order of 100 μm, and irregularly shaped.
• It seems that the effect of the sulfate ions dominates the shape of the portlandite particles.
• Since the [20.3] surface does not appear in the morphology only a lower bound for its relative energy can be calculated.

3.4. Atomistic simulation

• To further validate the results, the radial distribution of water molecules around the Ca and hydroxyl ion was compared to values reported in literature by Fulton et al. [32] and by Botti et al. [33].
• Additionally, as Chen et al. pointed out the experimentally measured nearest neighbor peak of the radial distribution function might in fact be a superposition of “true” nearest neighbor water and of H–O….
• Further discussion of the force field and a comparison with other force fields commonly used for cements can be found in [14].

3.4.2. Simulation of portlandite surfaces

• Atomistic simulation was used to calculate the surface energies for both portlandite in vacuum and inwater.
• In vacuum the [20.3] surface seems to have a very high energy, in fact the energy is more then ten times the energy of the [00.1] surface and almost twice the energy of the [10.0] surface.
• The surface energies of both the [10.0] surface and the [20.3] surface was reduced noticeably by the presence of water but the effect is much moremarked for the [20.3] surface.
• In fact the [20.3] surfacewhich has is a high energy surface in vacuum and does not appear in the portlandite-vacuummorphology has a similar energy to the [00.1] and the [10.0] surface in water and becomes part of the calculated portlandite-water equilibrium morphology (Fig. 9).
• The relative surface energies of both the [10.0] and the [20.3] surfaces with respect to the [00.1] surface are similar to the experimentally observed relative surface energies for samples B, BC and BN indicating that Cl− and NO3− ions might only have a very small effect on the equilibrium morphology (Table 3).

3.4.3. Comparison to other results in literature

• Previous simulation work on portlandite-water interfaces is scarce, however Kalinichev et al. did look at the [00.1] surface of portlandite in water.
• The density profile of the water hydrogen and the water oxygen over the surface was then calculated (Fig. 10).
• By comparing the density profile with the one reported in [9] both similarities and differences can be distinguished.
• Probably as a consequence of the more diffuse first water layer the second water layer is farther from the surface compared to the results of Kalinichev et al.

4. Conclusion

• The morphology of portlandite particles formed by coprecipitation in the presence of different ions has been observed.
• The addition of sulfates leads to the formation of particles with a hexagonal platelet shape.
• As the concentration of the silicate ions was much smaller than the concentration of the other ions, their effect upon the growth of portlandite particles seems to be the most marked.

Figures

Table 3 Relative surface energies measured for samples B, BC, BN and BS as well as calculated via atomistic simulation for particles in water and in vacuum.

Fig. 6. Radial distribution function of O–O in water. Line: experimental results reported by Soper et al. [26]. Diamonds: current results.

Table 1 Nominal concentration of the different ions present in the precipitation mixture.

Fig. 8. Calculated equilibrium morphology of portlandite in vacuum. Dark gray: [00.1] facet, light gray: [10.0] surface.

Fig. 7. Radial distribution function of water hydrogen atoms around the hydroxyl oxygen atoms in aqueous solution. Line: experimental results reported by Botti et al. [30]. Diamonds: current results.

Fig. 9. Calculated equilibrium morphology of portlandite in water. Dark gray: [00.1] facet, light gray: [20.3] facet.

Fig. 1. Schematic view of parameters used for the fitting between the calculated and the experimentally observed shape.

Fig. 10. Density profile of the number density of water hydrogen (black line) and water oxygen (gray line) relative to the bulk density at the [00.1] surface of portlandite. d is the distance from the portlandite oxygen layer closest to the surface.

Fig. 4. Thermogravimetric curve obtained for sample B. Heating rate 10 °C/min in flowing air 30 ml/min.

Fig. 3. XRD pattern of a precipitated portlandite (sample BC). Lines correspond to the peaks of ICDD 04-0733 (pure portlandite). Similar patterns were recorded for all samples.

Fig. 2. Schematic view of portlandite-water surface energy calculations.

Table 2 Characteristic values of the measured volume size distribution of the particles. 10% of the measured particles have a size lower than D(v10), 50% lower than D(v,50) and 90% lower than D(v,90).

Fig. 5. SEM images of portlandite particles for the different samples. a) B, b) BC, c) BN, d) BS, e) BS02 and f) BSi.

Frequently asked questions

Q1. What are the contributions mentioned in the paper "Changes in portlandite morphology with solvent composition: atomistic simulations and experiment" ?

In this paper, the authors used the Wulff shape of portlandite in vacuum and in water to measure the relative surface energy of the [ 10.0 ] and [ 20.3 ] facets. 

Q2. How many ps were used for the simulations?

The final simulations were carried out for 10+20 ps where the first 10 ps were used for equilibration of the system and the following 20 ps were used for all calculations. 

Q3. What were the different batches of portlandite used to observe their influence on the morphology?

nitrate, sulfate and silicate salts were then added to different batches to observe their influence on the morphology of portlandite. 

Q4. What are the main techniques used to study portlandite?

To understand the changes in portlandite morphology and to characterize portlandite surfaces, atomistic simulation techniques, namely classical energy minimization and classical molecular dynamics, have been employed. 

Q5. What is the energy of the portlandite surface in vacuum?

In vacuum the [20.3] surface seems to have a very high energy, in fact the energy is more then ten times the energy of the [00.1] surface and almost twice the energy of the [10.0] surface. 

Q6. What is the effect of the addition of nitrates on the portlandite particles?

contained silicates whereas the current BS samples contained chlorides, both of which are ions which seem to influence the shape of the portlandite particles. 

Q7. What is the energy of the portlandite-water surface?

The portlandite-water surface energies calculated by molecular dynamics are: [00.1]: 0.110 J/m2, [10.0]: 0.128 J/m2, [20.3]: 0.078 J/m2. 

Q8. What is the potential used to describe the water?

Vij rij= D 1−e−a rij−r0ð Þ h i2 −D ð4ÞThe potential used to describe the water was the Tip4P/2005 potential [20], which is a rigid molecule water force field that yields very good results for water calculations. 

Q9. What is the energy of the portlandite-water equilibrium morphology?

The mean surface energy of the calculated equilibriummorphology is 0.081 J/m2, which is consistent with the portlandite-water surface energy estimation of 0.114 J/m2 of Harutyunyan et al. [5]. 

Q10. What was the first NPT simulation for portlandite?

For portlandite the simulation box was first equilibrated (NPT simulation with an anisotropic Noose-Hoover– Melchionna thermostat which allows the cell to relax for 5+15 ps). 

Q11. What is the effect of the silicate ions on the growth of portlandite particles?

As the concentration of the silicate ions was much smaller than the concentration of the other ions, their effect upon the growth of portlandite particles seems to be the most marked. 

Q12. What is the next step in the atomistic simulations?

The next step in the atomistic simulations is to make the link with the effect of the different ions in solution, by incorporating nitrate, chloride and sulfate ions and observe their influence on the interfacial energies. 

Q13. What is the weighting parameter for the morphology of a portlandite shape?

The final fitting function which was minimized to fit the calculated to the experimentally observed shape takes into account the relative projection area and position of the different facets as well as their shape (Eqs. (1) and (2)).1 Nfaces∑ Nfacesi" α Θi−Θ exp i !2 + βr2i atot − r exp i 2 a exptot!2 + γai atot− a exp ia exptot!2 + δF shapei #ð1ÞFshapei = 1Nicorner ∑Nicornerj κ ψij−ψ i; exp j 2 + λl i j 2 ai − li; expj 2 a expi2 643 75 ð2Þwhere Nfaces is the number of facets of the shape, ri is the position of the centroid of facet i relative to the centroid of the total shape, Θi is the angle between ri and r1, ai is the area of facet i, atot is the total areaNicorner is the number of corners of facet i, lij is edge j of facet i andΨij is the angle between edge j and edge j+1 of facet i (see Fig. 1).α, β, γ, δ, κ and λ are weighting parameters which were set to α=β=γ=δ=0.25 and κ=λ=0.5. 

Q14. How many ps were used to calculate the portlandite-water surface energy?

To calculate the portlandite-water surface energy γ, the energy of the original portlandite-water slab Eport_H2O was compared to half the energy of a pure portlandite slab Eport,slab plus half the energy of a pure water slab EH2O,slab both twice the size of the original portlandite or water slab respectively (Eq. (7) and Fig. 2).γ = 1= Asurf ⋅ Eport H2O−0:5⋅Eport;slab−0:5⋅EH2O;slabð7Þ3.1. 

Q15. What is the common method used to study portlandite?

Apart from several articles using portlandite for the atomistic potential development and validation [6–8] there is, to their knowledge, only one group who have used atomistic simulation to studyportlandite and portlandite surfaces: Kalinichev et al. used classical molecular dynamics to look at the structure of the water above the surface and to estimate the extent of chloride binding at portlandite surfaces [9,10]. 

Q16. What is the energy minimization technique used in the simulations?

Vij rij = 4ε σ rij!12 − σrij!6" # ð5ÞThe exact parameters of the force field used in the simulations can be found in the Appendix A.The energy minimization calculations were made with the METADISE code [21], which implements an energy minimization technique based on classical potential models.