You are on page 1of 24

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/305496672

Atomistic structure of C-S-H from defective tobermorite structures: variations


of defects and features

Conference Paper · October 2015

CITATIONS READS

0 525

4 authors:

Aslam Kunhi Mohamed Sandra Galmarini


ETH Zurich Empa - Swiss Federal Laboratories for Materials Science and Technology
14 PUBLICATIONS   629 CITATIONS    27 PUBLICATIONS   661 CITATIONS   

SEE PROFILE SEE PROFILE

Paul Bowen Karen L. Scrivener


École Polytechnique Fédérale de Lausanne École Polytechnique Fédérale de Lausanne
231 PUBLICATIONS   6,323 CITATIONS    370 PUBLICATIONS   31,662 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

LC3 - Limestone Calcined Clay Cement View project

material characterization View project

All content following this page was uploaded by Sandra Galmarini on 22 July 2016.

The user has requested enhancement of the downloaded file.


1

2 Atomistic structure of C-S-H from defective tobermorite structures:


3 variations of defects and features
4 Aslam K. Mohamed1, Sandra C. Galmarini2, Paul Bowen1, Karen L. Scrivener3
5 1. Powder Technology Laboratory (LTP), Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015
6 Lausanne, Switzerland
7 2. Building Science and Technology, Eidgenössische Materialprüfungs- und Forschungsanstalt (EMPA), CH-
8 8600 Dübendorf, Switzerland
9 3. Laboratory of Construction Materials, Swiss Federal Institute of Technology Lausanne (EPFL),
10 Switzerland

11 Abstract

12 Understanding the underlying mechanisms of cement hydration has been quite difficult due to the complexity of the
13 system and its continued reaction over time making it hard to observe experimentally. Although atomistic
14 simulations might be useful to study how the presence of different species affect the nucleation and growth of
15 hydrates, the atomic structure of C-S-H, the main hydrate phase, is not clearly known or agreed upon and remains
16 an open question. Proposed structures of C-S-H have been mainly based on tobermorite and a model structure is
17 created by introducing defects in the original non-defective tobermorite structure. This work aims at studying the
18 different defects that can be considered and study their structural features towards creating a realistic C-S-H model.

19 Originality

20 In this work we try to arrive at the atomic structure of C-S-H by careful consideration of the defective structures and
21 its polymorphs. This has not been rigorously done before and usually a defective structure matching the Ca/Si ratio
22 needed is simply created by randomly depolymerizing Si chains and/or by addition of Ca in the interlayer space
23 without actually looking at its stability or variations. Our approach has the potential to consider the creation of a
24 model C-S-H with actual knowledge of the features of different defects that could exist.

25 Keywords: Cement, C-S-H, Molecular dynamics, Tobermorite

26
27 1 INTRODUCTION

28

29 The main hydration product when cement is mixed with water is calcium silicate hydrate (C-S-H) [1].
30 This phase has a variable composition as the hydration progresses, it is nano-crystalline [2,3,4,5] and has
31 a complicated layered structure with varying water content in the interlayer space [6]. Due to its nano-
32 structural features and continued change in composition with hydration time, C-S-H is difficult to
33 characterize experimentally [7, 8, 9]. Also a significant part of the C-S-H structure is constituted of
34 surfaces in contact with water/solution and it becomes critical to understand the interactions at these
35 surfaces to better comprehend the final properties and the reactions in cementitious systems. But the first
36 biggest challenge is to determine the atomistic structure of C-S-H itself [10]. 29Si NMR spectroscopy has
37 been used to study the arrangement of silicon layers in C-S-H and it shows the presence of silicate chains
38 composed of mainly dimers, pentamers and octamers with Q0 ~ < 2 %, Q1 ~ 40-65 % and Q2 ~ 33-60 %
39 [6, 11, 12, 13, 14] (where Qn refers to the percentage of Si connected to n Si atoms.). These results
40 indicate that a characteristic feature is the presence of linear three-unit repetition Silicon chains, so called
41 “dreierketten”. Most of the models for the structure of C-S-H are based on the natural minerals
42 tobermorite and jennite, which have the same “dreierketten” chains observed from NMR results, but
43 infinite in one direction.

44 In this work, we mainly consider 14 Å tobermorite which is structurally [5,8,15] and compositionally [4]
45 closest to the C-S-H structure observed. The structure of tobermorite itself is complicated, composed of a
46 calcium plane bordered by two parallel “dreierketten” silicon tetrahedral linear chains. This layered
47 structure is repeated, with a repetition distance of 14 Å, in the perpendicular direction with an interlayer
48 space consisting of some Ca ions and water molecules as shown in Figure 1. Tobermorite presents a
49 significant order disorder character, with different polymorphs and stacking sequences [16]. 14 Å
50 tobermorite has an effective calcium to silicon (Ca/Si) ratio of 0.83[17]. This is quite low compared to
51 Ca/Si ratio of C-S-H which varies between 1.2 and 2.1, with an average value of 1.75 [7]. In order to
52 achieve higher Ca/Si ratio, the possible ways one can imagine are by the removal of bridging silica
53 tetrahedra (SiO2), the deprotonation of silanol groups which can be charge compensated by additional
54 calcium ions and the addition of CaO or Ca(OH)2 in the interlayer. Thus C-S-H can essentially be seen as
55 a defective tobermorite structure with varying composition. By removing some silica tetrahedra, the
56 infinite silicon chain is broken leading to the appearance of Q1 in NMR spectra. It is quite challenging to
57 quantify experimentally the kind of defects present in C-S-H and this favors the application of theoretical
58 approach to study the variations of possible defects.
59 Different defective structures can be created with atomistic tools and their stability can be studied using
60 force field potentials in classical molecular dynamics. This modeling approach has the potential
61 advantage of allowing study of the properties and processes happening at the interface from an atomistic
62 point of view and then to extend it to the microscopic/macroscopic experimental observations. Recently,
63 Pellenq et al. [18, 19] and Kovačević et al. [20] have done molecular dynamics modeling of C-S-H. Both
64 of these authors started with 11 Å tobermorite as the base structure and then randomly remove SiO2
65 groups and, in the case of Kovačević et al., addition of some Ca ions in different steps. Pellenq et al.
66 started with dry tobermorite with 4x2x1 crystallographic unit supercell and removed SiO2 groups guided
67 by NMR results. Water adsorption is done in a Grand Canonical Monte Carlo simulation at 300 K, but in
68 addition to the interlayer the water also goes in between the silicate chains. The main drawback of this
69 model is the absence of OH ions and extremely high percentage of monomers, Q0. The authors have tried
70 to take account of these shortcomings in recent work using the ReaxFF potential to split water molecules,
71 making it possible to have hydroxyl group and thus forming Ca-OH. The hydrogen of the split water
72 molecules mainly reacted with non-bonding oxygen of the silicate chains, forming protonated silanol
73 groups. However experimental results indicate that the amount of protonated silanol groups in C-S-H is
74 negligible [21]. In order to remove the monomers they allow for (random) polymerisation of the silicates
75 yet this leads to a silicate chain structure that is in violation of the dreierketten pattern that exist in C-S-H,
76 where we only have dimers, pentamers, octamers and so on.

77 On the other hand Kovačević et al. has compared three different ways of increasing the Ca/Si ratio of
78 tobermorite in order to approach the C-S-H structure and compared them to the early model by Pellenq et
79 al. [18]. The first consisted of removing only bridging tetrahedrons and adding additional Ca(OH)2 units
80 to the structure. The second consisted of removing both dimers and bridging tetrahedra from the silicate
81 chains. The third approach was to randomly remove silicate tetrahedra, similar to the approach taken by
82 Pellenq et al. The authors then relaxed the structures using the ReaxFF force field, similar to the approach
83 in the recent work of Pellenq et al [19].

84 All the models have the final Ca/Si ratio of 1.65 and their total energies and the local order of the
85 structures are calculated to conclude what structure is most probable in C-S-H. The authors conclude that
86 the way of constructing the initial structure has a non-negligible influence and the addition of Ca(OH)2 in
87 the interlayer needs to be included when constructing a C-S-H model.

88 In this work, we go further in the study of different local structures and defects, including a larger range
89 of possible structures, to determine which would need to be included in a model to construct a realistic C-
90 S-H structure, taking into account all the available experimental data, which can be resumed as follows:
91  Average Ca/Si ratio of around 1.6-1.75 [4]
92  Chain length between 3.3 (fresh cement) and closer to 5 (~22 year old cement), with 60 (fresh) to
93 40 (old) % dimers and the rest being longer chains.[10]
94  Small (negligible) amount of protonated silanol groups [21]
95  Around 23 % of the Ca charge compensated by OH ions, corresponding to a ratio of 0.46 .[22]
96  Approximate stoichiometry: Ca1.7SiO3.7 ·1.8 H2O, excluding surface water.
97  “Solid” density (i.e. excluding surface water) of about 2.6 g/cm3 [23].
98  A nanocrystalline character with ordering up to a length scale greater than 3 nm as observed by
99 different authors [5,8,24], with X-ray structure factors indicating a significantly larger scale
100 ordering than that observed in ,for example, silicate glasses [25].

101 None of the currently available models are able to conform to all these experimental observations
102 and no explanation for the nanoparticulate nature of C-S-H, which persists even after prolonged
103 ageing, has come from the current models. It is the opinion of the authors that in order to achieve this,
104 a wider range of possible defect structures needs to be considered. Additionally an idea of the
105 influence of structural details and the order/disorder character of tobermorite and the influence of the
106 defect concentration is needed.

107 2 METHODS

108 2.1 SIMULATION


109

110 Classical molecular dynamics with force field potentials have been used in this work. Short range van der
111 Waals interactions were defined using Lennard-Jones and Buckingham potentials for different species.
112 For the bonded interactions in the hydroxyl molecules and silicate chains, Morse potential and Harmonic
113 angle potentials are used. Detailed description of the different species and their interaction potentials can
114 be found elsewhere [26]. For water molecule, the TIP4P/2005 rigid model was adopted. An isolated
115 defective unit in a supercell of 4x3x1 14 Å tobermorite unit cells is used for the molecular dynamics
116 calculation resulting in a simulation cell of 27x18x28 Å. For some of the defects both the B11b [17] and
117 the possible polymorph structure of tobermorite are considered. After pre-equilibration the system is
118 equilibrated for 35 ps and then data is collected over a simulation period of 105 ps using the anisotropic
119 NPT ensemble. Classical molecular dynamics is carried out using the package DL_POLY 2.20 [27]

120 2.2 STRUCTURE BUILDING


121
122 A total number of 11 different defects with Ca/Si ratios ranging from 1.0 to 2.5 are considered here. The
123 methods adopted to create the different defect structures considered in the paper are described below and
124 are summarized in Figure 2. A polymorph of the B11b structure for most of the defects, which might
125 coexist in real structures, are also considered.
126
127 For Ca/Si ratio of 1.00, the protons of two neighboring silanol groups of two bridging silicate tetrahedra
128 on either side of the interlayer are removed from the back bone of 14 Å tobermorite (Ca/Si of 0.83).
129 Figure 2. The protons can be removed from the two different silanol groups that are about equidistant,
130 hence we have two possibilities and both of them are considered.
131
132 For Ca/Si ratio of 1.25, a bridging silicate tetrahedron of a silicate dreierketten chain, equivalent to a SiO2
133 group, is removed from the structure. The hydrogen of the silanol group of the removed bridging silicate
134 tetrahedron is transferred to one of the adjacent paired silicate tetrahedra (Figure 2). The loss of the two
135 oxygen atoms can then be compensated by the addition of two water molecules. In the second variant of
136 this defect, the two oxygen atoms are replaced by one water molecule and a hydroxyl group. The charge
137 of the additional hydroxyl group is then compensated by protonating the free silanol group of the second
138 paired silicate tetrahedron adjacent to the defect.
139
140 For Ca/Si ratio of 1.5, one can start from the previously described Ca/Si = 1.0 defect structure and remove
141 one of the bridging silicate tetrahedron, similar to how the Ca/Si ratio is raised from 0.83 to 1.25. Two
142 water molecules are then added to compensate the loss of the two oxygen atoms as done previously for
143 Ca/Si 1.25 case. In the second variant of this defect, instead of replacing the removed oxygen atoms by
144 two water molecules one can replace them by either one water molecule and a hydroxyl group,
145 protonating one of the free silanol groups, or by two hydroxyl groups, protonating the free silanol groups
146 of both of the paired silicate tetrahedra adjacent to the defect. Only the latter variant is considered here.
147
148 For Ca/Si ratio of 1.75, a bridging silicate tetrahedron is replaced by a calcium and a hydroxide ion from
149 the original 0.83 tobermorite. Stoichiometrically this is equivalent to replacing SiO2+ unit by a Ca2+ion.
150 The removed oxygen is replace by a water molecule. In the second variant of this defect, instead of
151 adding Ca(OH)+ and a water molecule, one can add a Ca(OH)2 and protonate the free silanol group of one
152 of the paired silicate tetrahedra adjacent to the defect.
153
154 For Ca/Si ratio of 2.0, a bridging silicate tetrahedron is replaced by a calcium and two hydroxide ions
155 from the 1.0 Ca/Si ratio defect structure, similar to raising Ca/Si ratio from 0.83 to 1.75.
156 Stoichiometrically this is equivalent to replacing an SiO2 unit by a Ca(OH)2 . No water molecule is added
157 in this case.
158
159 For the last defect, Ca/Si ratio of 2.5, two bridging and two paired silicate tetrahedral from two adjacent
160 14 Å tobermorite units are removed. Stoichiometrically this is equivalent to removing Si4O9(OH)24- and
161 the resulting charge is compensated by adding four hydroxyl groups. Finally, to keep the total number of
162 oxygen atoms in the structure a constant, seven water molecules are added.
163
164

165 3 RESULTS
166 The dimensions of the crystalline unit cell of an original 14 Å tobermorite calculated by molecular
167 dynamics are shown in Table 1. Values from the XRD structure [BMK05] are also included in the table
168 for comparison. It can be seen that the results obtained in our work are in very good agreement with
169 experimental results. This again validates the applicability of the force field potentials used in this work
170 for cementitious systems as well as portlandite systems [26]. In the case of defective structures, our
171 results show that there is very limited structural relaxation for these defects which indicates that they are
172 quite stable. The introduction of these locally higher Ca/Si defects will allow us to get closer to a C-S-H
173 which matches all the experimental structural properties available, with fewer dimers, longer average
174 chain lengths, fewer protonated silanol groups and more hydroxyl groups. It has been observed that
175 polymorphism does not play a major role in all the defect structures that we have considered here. The
176 resulting structure and their features are discussed for each defect in the following paragraphs.
177
178 For the Ca/Si ratio 1.0 defect structure, we can see that both the bridging silicate tetrahedra that were
179 deprotonated (dp sites in) have rotated compared to the initial bulk position. Also both the original and
180 additional interlayer calcium ions are shared between them, one on either side of the silicate chain.
181 Locally the water positions appear to have changed slightly but within the nearest neighbor waters. The
182 density of the structure is slightly increased from 2.22 to 2.34 g/cm3 due to the replacement of two
183 hydrogen atoms with a calcium and the bound water content is decreased from 5.3 %mol to 0 %mol (Table
184 2).
185
186 For the Ca/Si ratio 1.25 defect structure, one of the silicate tetrahedron adjacent to the defect, the one
187 sharing the charge of the interlayer calcium with the defect site, has rotated compared to the bulk structure
188 (rt site in Figure 4). The interlayer calcium ion has moved towards the original position of the removed
189 silicate tetrahedron (Ca site in Figure 4) and one of the added water appears to occupy the original
190 position of the hydroxyl group in the non-defective structure (wt site). No clearly defined site occupancy
191 is seen for the other additional water molecule and it seems to cause a certain amount of positional
192 disorder of the surrounding water molecules. The density of this structure is 2.1 g/cm3 and it contains
193 about the same amount of chemically bound water but significantly higher amount of physically bound
194 water compared to the non-defective structure. In the case of the second variant of this defect, the amount
195 of chemically bound water is increased and the amount of physically bound water is decreased, making
196 the total water content constant. Also, 20 % of the charge of the calcium ions is compensated by
197 hydroxyls of Ca-OH groups. Here the structure of the second variant appears to be closer to the original
198 14 Å tobermorite structure than the first with the interlayer calcium essentially occupying the same
199 position as in the non-defective bulk structure (Ca site in Figure 5). The hydroxyl groups (h and p sites)
200 are clustered around the vacated bridging silicate site and the positional disorder of the water structure is
201 much less pronounced.
202
203 For the Ca/Si ratio 1.5 defect structure, positional disorder in the interlayer has increased (Figure 6).
204 One of the two interlayer calcium atoms (Ca site), that are shared between the defect site and the silicate
205 chain on the other side of the interlayer, is found to be occupying a position closer to one of the free
206 silanol groups of the paired silicate tetrahedra next to the defect site. The other calcium seems to occupy
207 similar position as in the Ca/Si 1.0 defect structure. The positions of the water molecules appear to have
208 become quite ill defined. One of the additional water molecules (wt site) is situated between the two free,
209 deprotonated silanol groups next to the defect site, donating a hydrogen bond to each of them. The density
210 of this defect structure is 2.20 g/cm3 (Table 2) and the water content is very high 52 %mol and no
211 chemically bound water. In the case of second variant considered for this defect, the percentage of
212 chemically bound water is higher (19 %mol) and that physically bound is smaller (33 %mol). Three of the
213 four hydroxyl groups (p and h1 sites in Figure 7) are clustered around the defect position, the hydrogen
214 atoms more or less pointing towards the original position of the removed silicon ion. The fourth hydroxyl
215 group (h2 site) has in fact moved away from the original defect position. The position of the interlayer
216 calcium ions seem to correspond to the position in the Ca/Si 1.0 defect structure whereas the positions of
217 the water molecules appear to be better defined compared to the first variant of Ca/Si 1.5 defect. In this
218 variant 33 % of charge of calcium ions is compensated by hydroxyls. This is higher compared to what is
219 seen experimentally (23 %)
220
221 For the Ca/Si 1.75 defect structure, the additional calcium ion (Ca site in Figure 8) occupies a similar
222 position in the structure as the removed silicate tetrahedron but slightly displaced towards the interlayer
223 hydroxyl group (h site). The second calcium ion retains its position in the non-defective structure. The
224 additional water molecule (wt site) is found to adopt a position similar to the one occupied by the
225 hydroxyl group of the bridging silicate tetrahedron removed. For the other water molecules the positional
226 disorder seems to have significantly increased compared to the non-defective structure. The density of the
227 structure is higher 2.3 g/cm3 (Table 2) due to the additional calcium in the interlayer. The amount of
228 chemically bound water is similar to the original 14 Å tobermorite but the total water content is higher.
229 The second variant considered has higher chemically bound water and lower physically bound water,
230 keeping the total water constant. The additional calcium ion (Ca site in Figure 9) occupies more or less
231 exactly the position of the removed silicon ion, only slightly displaced towards the center of the interlayer.
232 This is possibly to accommodate the larger Ca-O nearest neighbor distance compared to Si-O. One of the
233 interlayer hydroxyl (h1 site) is shared between the two interlayer calcium ions whereas the second
234 hydroxyl (h2 site) has moved towards the defective calcium silicate sheet and is occupying a position
235 similar to that of the hydroxyl of the removed bridging silicate tetrahedron. As in the case of other defects,
236 the positional disorder of water molecules appears to be lower in the second variant. It can be observed
237 that more chemically bound water and less physically bound water seem to decrease the positional
238 disorder. The percentage of the charge of calcium ions that is compensated by the hydroxyl group of the
239 Ca-OH are 14 and 29 % for the first and second variants of this defect respectively.
240
241 For the Ca/Si ratio of 2.0 defect structure, the additional calcium ion (Ca site in Figure 10) lodges a
242 similar position as the silicon ion it replaces whereas the other two interlayer calcium ions have not
243 significantly moved from their positions in the Ca/Si 1.0 defect structure. The two interlayer hydroxyl
244 groups (h sites) are shared between the interlayer calcium ions and are positioned approximately in the
245 middle of the interlayer. A water molecule (wt site) donates a hydrogen bond to each of the dangling
246 oxygen atoms of the paired tetrahedra positioned on either side of the defect site. The other interlayer
247 water molecules have slightly increased positional disorder compared to the Ca/Si 1.0 defect structure.
248 Here we find a relatively high density of 2.4 g/cm3 due to the large concentration of calcium ions in the
249 interlayer and slightly higher percentage of chemically bound water. Calcium ion charge compensation by
250 hydroxyl group is about 25 %.
251
252 For Ca/Si ratio of 2.5 defect structure, surprisingly little disorder is observed from the molecular
253 dynamics simulation. The positions of the calcium ions adjacent to the removed silicate chain do not
254 appear to have changed, leaving the calcium sheet intact. The added hydroxyl groups (h sites in Figure 11)
255 remain stably bound to the calcium layer throughout the simulation. The position of the interlayer calcium
256 does not change significantly either. Due to the removal of a large number of silicates which are replaced
257 by water, the density of this defective unit is very low (2.0 g/cm3) and the water content is very high
258 (Table 2). Also 40 % of the charge of the calcium ions are compensated by the hydroxyl groups which is
259 quite high. This time, the hydroxyl groups are part of the calcium-silicate layer, though not present as
260 silanol groups as in the non-defective structure.

261 4 DISCUSSION

262 Now it is important to compare the structural details of our results to experimental observations reported
263 in literature. We have observed a decrease of the silicate chain length, enrichment of the interlayer with
264 calcium and the appearance of Ca-OH groups, as seen in experimental results with increase in Ca/Si ratio
265 [3]. However this is not surprising since our methodology of creating defects was guided by experimental
266 observations reported in literature, for example the removal of a single paired silicate tetrahedron was
267 never considered here. Another structural feature is the evolution of the protonated silanol groups with
268 increasing Ca/Si ratio which can be discussed with the help of precipitation enthalpies of these defects
269 [26]. This will be discussed in a future journal paper and is beyond the scope of this paper. The small
270 structural changes that we observed in the structures are mainly due to an increased structural disorder of
271 the interlayer with increasing Ca/Si ratio. Thus we can sum up that increasing Ca/Si ratio is unlikely to
272 change the nearest neighbor distances (interatomic distances) which is again consistent with experimental
273 observations. However further analysis, like calculation of radial distribution functions of structures with
274 higher defect concentrations should be done to confirm this point.
275 We observe that both the water content and the density of the defect structures considered increase with
276 increasing Ca/Si ratio as seen in experimental observations [28,23]. This is due to the increase of the
277 calcium concentration in the interlayer and the replacement of oxygen atoms of the silicate tetrahedra
278 removed with water. For instance, let us consider a structure with an overall calcium to silicon ratio of 1.7
279 composed of 26 % non-defective 14 Å tobermorite units and 74 % defective tobermorite units with a
280 Ca/Si ratio of 2.0. The resulting structure would have an average density of 2.4 g/cm3, a water content of
281 43 %mol and 19 % of the charge of the calcium ions compensated by hydroxyl groups. Muller et al. [23]
282 reported an overall water content for Ca/Si ratio 1.7 C-S-H to be 40-47 %mol, depending on whether or not
283 a correction for the surface water is applied, which would be in good agreement with our results. The
284 percentage of the calcium charge that is compensated by hydroxyl ions is a bit smaller than the
285 experimental value (23 %) [22].This could be compensated by having a certain percentage of defect with
286 higher compensation percentage (for e.g. Ca/Si 1.75 has 29 % charge compensation). The overall density
287 is a bit lower than that reported by Muller et al. [23] (2.5 -2.7 g/cm3). But these experimental results are
288 greatly influenced by the high surface area of C-S-H. Our simulations are at the moment for isolated
289 defects in bulk tobermorite and due to the low density of defects, the interlayer distance is fixed to that of
290 14 Å tobermorite. This paper has tried to imply that it is quite important to understand the different kinds
291 of defects that could exist in C-S-H and before building a realistic model it is worth studying the
292 structural features of these defects and their energetics [26].This knowledge can then be extended to
293 systematically create a model C-S-H structure from defective tobermorite structures by carefully
294 considering the different defects and their arrangements to form bulk structures and surfaces.
295 5 CONCLUSION

296 An attempt towards a systematic approach to creating a realistic C-S-H model from the structure of 14 Å
297 tobermorite is put forth in this paper. By carefully studying a wider range of possible defects that increase
298 the Ca/Si ratio of a C-S-H model (defective tobermorite) structure we can conclude that it is important to
299 take into account the full range of different variations, structural features and details of the defects. In
300 literature often a random distribution of the defective structures are seen, while in this paper we have
301 attempted to convey the message that it is quite vital to understand the kind of defects that exists in a C-S-
302 H structure before making a final structure. Our methodology has been guided by the experimental
303 evidence, for instance, negligible silicate monomers are found in C-S-H and only bridging silicon
304 tetrahedra or in one instance a silicate dimer are removed in our defective structures leading to creation of
305 zero monomers in the final structures and a silicate chain structure consistent with the dreierketten motive
306 only allowing chain lengths of 3n-1.

307 It has been shown that by the combination of non-defective tobermorite structures with different amounts
308 of defective structures, we can build a model with the required Ca/Si ratio, density, water content and
309 charge compensation of calcium ions by Ca-OH groups. Once we take into account the stability of these
310 defects (soon to be published), the authors aim to build a model/models for C-S-H structure. This model
311 can then be used to create C-S-H surfaces and study the interactions with the help of molecular dynamics
312 and other atomistic simulation tools.

313 6 REFERENCES

314 1. H. F. W. Taylor, 1997. Cement Chemistry. Academic Press.


315
316 2. Ake Grudemo, January 1984. Variation with solid-phase concentration of composition, structure and
317 strength of cement pastes at high age. Cement and Concrete Research, 14(1):123–132.
318
319 3. I. G. Richardson, September 2004. Tobermorite/jennite- and tobermorite/calcium hydroxide based models
320 for the structure of c-s-h: applicability to hardened pastes of tricalciumsilicate, [beta]-dicalcium silicate,
321 portland cement, and blends of portland cement with blast-furnace slag, metakaolin, or silica fume. Cement
322 and Concrete Research, 34(9):1733–1777.
323
324 4. I. G. Richardson, 2008. The calcium silicate hydrates. Cement and Concrete Research, 38(2):137–158.
325
326 5. André Nonat, September 2004. The structure and stoichiometry of C-S-H, Cement and Concrete Research,
327 34 (9): 1521-1528.
328
329 6. Xiandong Cong and R. James Kirkpatrick, 1996. 29SiMAS NMR study of the structure of calcium silicate
330 hydrate. Advanced Cement Based Materials, 3(3-4):144– 156.
331
332 7. I. G. Richardson and G. W. Groves, January 1993. Microstructure and microanalysis of hardened ordinary
333 portland cement pastes. Journal of Materials Science, 28(1):265–277.
334
335 8. G. Renaudin, J. Russias, F. Leroux, F. Frizon, and C. Cau-dit Coumes, 2009. Structural characterization of
336 CSH and CASH samples–Part i: Long-range order investigated by rietveld analyses. Journal of Solid State
337 Chemistry, 182(12):3312–3319.
338
339 9. Krassimir Garbev, Peter Stemmermann, Leon Black, Chris Breen, Jack Yarwood, and Biliana Gasharova,
340 2007. Structural features of c-s-h (i) and its carbonation in air - a raman spectroscopic study. Part 1: Fresh
341 phases. Journal of the American Ceramic Society, 90(3):900–907.
342
343 10. K.Mohan and H.F.W. Taylor, January 1982. A trimethylsilylation study of tricalcium silicate pastes.
344 Cement and Concrete Research, 12(1):25–31.
345
346 11. J. Skibsted and C. Hall, 2008. Characterization of cement minerals, cements and their reaction products at
347 the atomic and nano scale. Cement and Concrete Research, 38(2):205–225.
348
349 12. F. Brunet, Ph. Bertani, Th. Charpentier, A. Nonat, and J. Virlet, October 2004. Application of 29Si
350 homonuclear and 1H-29Si heteronuclear NMR correlation to structural studies of calcium silicate hydrates.
351 The Journal of Physical Chemistry B, 108(40):15494–15502.
352
353 13. I. Klur, J.-F. Jacquinot, F. Brunet, T. Charpentier, J. Virlet, C. Schneider, and P. Tekely, November 2000.
354 NMR cross-polarization when TIS>T1rho; examples from silica gel and calcium silicate hydrates. The
355 Journal of Physical Chemistry B, 104(44):10162–10167.
356
357 14. A. R. Brough, C. M. Dobson, I. G. Richardson, and G. W. Groves, January 1994. In situ solid-state NMR
358 studies of Ca3SiO5: hydration at room temperature and at elevated temperatures using 29Si enrichment.
359 Journal of Materials Science, 29(15):3926–3940.
360
361 15. Pawel Rejmak, Jorge S. Dolado, Malcolm J. Stott, and Andrés Ayuela, 2012. 29
Si NMR in Cement: A
362 Theoretical Study on Calcium Silicate Hydrates. The Journal of Physical Chemistry C, 116 (17), 9755-
363 9761
364
365 16. Elena Bonaccorsi, StefanoMerlino, and Anthony R. Kampf, 2005. The crystal structure of tobermorite 14 a;
366 (plombierite), a c-s-h phase. Journal of the American Ceramic Society, 88(3):505–512.
367 17. S. A. Hamid, 1981. The crystal structure of the 11 a natural tobermorite. Zeitschrift für Kristallographie,
368 154:189–198.
369
370 18. Roland J.-M. Pellenq, Akihiro Kushima, Rouzbeh Shahsavari, Krystyn J. Van Vliet, Markus J. Buehler,
371 Sidney Yip, and Franz-Josef Ulm, 2009. A realistic molecular model of cement hydrates PNAS 106 (38):
372 16102-16107.
373
374 19. Qomi, M. J. A.; Krakowiak, K. J.; Bauchy, M.; Stewart, K. L.; Shahsavari, R.; Jagannathan, D.; Brommer,
375 D. B.; Baronnet, A.; Buehler, M. J.; Yip, S.; Ulm, F. J.; Van Vlient, K. J.; Pellenq, R. J. M, . 2015.
376 Combinatorial Molecular Optimization of Cement Hydrates Nat. Commun, 5, 4960.
377
378 20. Goran Kovačević, Björn Persson, Luc Nicoleau, André Nonat, Valera Veryazov, January 2015. Atomistic
379 modeling of crystal structure of Ca1.67SiHx, Cement and Concrete Research, 67: 197-203.
380
381 21. P. Yu, R. J Kirkpatrick, B. Poe, P. FMcMillan, and X. Cong, 1999. Structure of calcium silicate hydrate
382 (CSH): near-, mid-, and far-infrared spectroscopy. J. Am. Ceram. Soc., 82(3):742–748.
383
384 22. Jeffrey J. Thomas, Jeffrey J. Chen, Hamlin M. Jennings, and Dan A. Neumann, October 2003. Ca-OH
385 bonding in the c-s-h gel phase of tricalcium silicate and white Portland cement pastes measured by inelastic
386 neutron scattering. Chemistry of Materials, 15(20):3813–3817.
387
388 23. Arnaud C. A.Muller, Karen L. Scrivener, AgataM. Gajewicz, and Peter J.McDonald, January 2013.
389 Densification of C–S–H measured by 1H NMR relaxometry. The Journal of Physical Chemistry C,
390 117(1):403–412.
391
392 24. CaglaMeral, C.J. Benmore, and Paulo J.M.Monteiro, July 2011. The study of disorder and nanocrystallinity
393 in C–S–H, supplementary cementitious materials and geopolymers using pair distribution function analysis.
394 Cement and Concrete Research, 41(7):696–710.
395
396 25. Skinner, L., Chae, S., Benmore, C., Wenk, H. and Monteiro, P. (2010). Nanostructure of Calcium Silicate
397 Hydrates in Cements. Phys. Rev. Lett., 104(19).
398
399 26. S. C. Galmarini, 2013. Atomistic Simulation of Cementitious Systems. Thesis (PhD), EPFL Switzerland, n°
400 5754
401
402 27. W. Smith and T. R. Forester, June 1996. DL_POLY_2.0: a general-purpose parallel molecular dynamics
403 simulation package. Journal of Molecular Graphics, 14(3):136–141.
404
405
406 28. A. J Allen, J. J Thomas, and H. M Jennings, 2007. Composition and density of nanoscale calcium-silicate-
407 hydrate in cement. Nature Materials, 6(4):311–316.
408

409
410 7 TABLES AND FIGURES

411
412 Table 1: Unit cell structure parameters calculated from MD compared with experimental XRD values

a b c α β γ
[Å] [Å] [Å] [o] [o] [o]
MD 6.86 ± 0.34 7.27 ± 0.36 28.00 ± 1.40 90 ± 10 93 ± 10 123 ± 10
[BMK05] 6.74 7.43 27.99 90 90 123
413
414
415
416 Table 2: Density, percentage of chemically bound and physically bound water and the percentage of the charge of the calcium
417 ions that is compensated by OH for different structures considered with their stoichiometry. The percentage of chemically and
418 physically bound water are calculated from their respective stoichiometric formulas obtained and comparing it with the same
419 written in format. If p is the coefficient of physically bound water, for instance 3.5 in Ca/Si ratio 0.83 , then percentage of
420 physically bound water is and that of chemically bound water is, . is the ratio of the charges of the OH- ions with respect to the
421 Ca2+ which corresponds to half the atomic ratio between the two species and can be calculated by looking at the structures (see
422 the corresponding structures of each Ca/Si ratio defect in Figure 2).

ρ
Formula

0.83 2.22 5.3 36.8 0


1.0 2.34 0.0 36.8 0
-do-
1.25 2.1 4.8 52.4 0

2.09 14.3 42.9 20


1.5 2.20 0.0 52.4 0
2.20 19.0 33.3 33
1.75 2.31 48 42.9 14
2.30 14.3 33.3 29
2.0 2.42 9.5 33.3 25
2.5 1.95 8.7 60.9 40
423

424

425
Figure 1: On the left XRD structure of a 14 Å tobermorite is shown with Ca: green, Si: gray,
O: red, H: white. On the right schematic illustration of the structural unit of the calcium-
silicate backbone of 14 Å tobermorite is shown in a dashed box. Lighter colors are
indicative of atoms shared between two units. The water molecules are not shown in the
schematic view

426

427

428
429

1.25 1.5

1.75 0.83 . 1.0

2.5 2.0

Second variant
1.25 1.5 1.75

430 Figure 2: Schematic view of the defective structure building methodology with Ca/Si ratio of each defective unit shown on top
431 left. Schematic view of the silicate chain of 14 Å tobermorite shown in 2nd row, 2nd column from which each structure is built.
432 The arrow marks indicate the increasing of Ca/Si ratio from the previous structure.

433
Figure 3: Structure of a 14 Å tobermorite defect with the Ca/Si ratio of 1.0. Schematic view (left) and partially minimized
snapshot of molecular dynamics calculations viewed along (middle) and perpendicular to (right) the silicate chains. Ca: green,
Si: gray, O: red, H: white. Labels – dp: deprotonated silanol group, Ca: additional calcium ion.

Figure 4: Structure of a 14 Å tobermorite defect with the Ca/Si ratio of 1.25. Schematic view (left) and partially minimized
snapshot of molecular dynamics calculations viewed along (middle) and perpendicular to (right) the silicate chains. Ca: green, Si:
gray, O: red, H: white. Labels – p: protonated silanol group, wt: water site replacing the hydroxyl group of the original structure,
Ca: interlayer calcium, rt: rotated silicate tetrahedron.

434
Figure 5: Structure of a 14 Å tobermorite defect with the Ca/Si ratio of 1.25, second variant. Schematic view (left) and partially
minimized snapshot of molecular dynamics calculations viewed along (middle) and perpendicular to (right) the silicate chains.
Ca: green, Si: gray, O: red, H: white. Labels – p: protonated silanol group, h: interlayer hydroxyl, Ca: interlayer calcium.

Figure 6: Structure of a 14 Å tobermorite defect with the Ca/Si ratio of 1.5. Schematic view (left) and partially minimized
snapshot of molecular dynamics calculations viewed along (middle) and perpendicular to (right) the silicate chains. Ca: green, Si:
gray, O: red, H: white. Labels – wt: water with hydrogen bonds to the free silanol group, Ca: interlayer calcium.

435

436

437
Figure 7: Structure of a 14 Å tobermorite defect with the Ca/Si ratio of 1.5, second variant. Schematic view (left) and partially
minimized snapshot of molecular dynamics calculations viewed along (middle) and perpendicular to (right) the silicate chains.
Ca: green, Si: gray, O: red, H: white. Labels – p: protonated silanol groups, h1/h2: interlayer hydroxyl.

Figure 8: Structure of a 14 Å tobermorite defect with the Ca/Si ratio of 1.75. Schematic view (left) and partially minimized
snapshot of molecular dynamics calculations viewed along (middle) and perpendicular to (right) the silicate chains. Ca: green, Si:
gray, O: red, H: white. Labels – wt: water site replacing the hydroxyl group of the original structure, Ca: additional interlayer
calcium, h: additional interlayer hydroxyl

438

439
Figure 9: Structure of a 14 Å tobermorite defect with the Ca/Si ratio of 1.75, second variant. Schematic view (left) and partially
minimized snapshot of molecular dynamics calculations viewed along (middle) and perpendicular to (right) the silicate chains.
Ca: green, Si: gray, O: red, H: white. Labels – Ca: additional interlayer calcium, h1/h2: additional interlayer hydroxyl.

Figure 10: Structure of a 14 Å tobermorite defect with the Ca/Si ratio of 2.0. Schematic view (left) and partially minimized
snapshot of molecular dynamics calculations viewed along (middle) and perpendicular to (right) the silicate chains. Ca: green, Si:
gray, O: red, H: white. Labels – wt: water with hydrogen bonds to the free silanol group, Ca: additional interlayer calcium, h:
additional interlayer hydroxyl

440

441
Figure 11: Structure of a 14 Å tobermorite defect with the Ca/Si ratio of 2.5. Schematic view (left) and partially minimized
snapshot of molecular dynamics calculations viewed along (middle) and perpendicular to (right) the silicate chains. Ca: green, Si:
gray, O: red, H: white. Labels –h: additional interlayer hydroxyl

442

443

View publication stats

You might also like