Vous êtes sur la page 1sur 7

Chemical Physics Letters 392 (2004) 242248 www.elsevier.

com/locate/cplett

Accelerating eects of colloidal nano-silica for benecial calciumsilicatehydrate formation in cement


J. Bjrnstrm o o
a b

a,*

, A. Martinelli b, A. Matic b, L. Brjesson b, I. Panas o

Department of Inorganic Chemistry, University of Gothenburg, SE-41296 Gteborg, Sweden o Department of Applied Physics, Chalmers University of Technology, SE-41296 Gteborg, Sweden o Received 19 May 2004 Available online 11 June 2004

Abstract The hydration process of Ca3 SiO5 (C3 S) cement is investigated, and accelerating eects of adding colloidal silica (CS) are established. CS accelerates dissolution of the C3 S phase, and renders a more rapid formation of the calciumsilicatehydrate (CSH) binding phase. The role of water is demonstrated. Water evaporation correlates with the appearance of a sharp IR band at 3650 cm1 signifying non-hydrogen bonded OH groups. Incomplete hydration results in Ca(OH)2 precipitation. Suppression of water evaporation causes the 3650 cm1 band to vanish. The Ca2 ions become incorporated in the CSH network, the formation of which is accelerated by CS. 2004 Elsevier B.V. All rights reserved.

1. Introduction A recent contribution to the development of building materials comprises adding synthetic colloidal silica to concrete and cement mortars, whereby the resulting product displays improved aging properties with regard to strength gain, sulphate attack and alkali-silica reactions [1]. Colloidal silica (CS) denotes small particles consisting of an amorphous SiO2 core with a hydroxylated surface, which are insoluble in water. The size of the particles can be varied between 1 and 500 nm, hence they are small enough to remain suspended in a uid medium without settling. Parameters such as specic surface area, size and size distribution can be controlled by the synthesis technique. Due to the high specic surface area for the nano-meter sized CS particles they constitute a highly reactive siliceous material [2]. However, it has not yet been established whether the more rapid hydration of cement in the presence of nano-silica is due to its chemical reactivity upon dissolution (pozzolanic activity) or to a considerable surface activity.

Corresponding author. Fax: +46-31-772-2853. E-mail address: jok@chem.gu.se (J. Bjrnstr m). o o

One main objective of the present work is the removal of this ambiguity. Calciumsilicatehydrate (CSH) is the principal hydration product and primary binding phase in Portland cement. The dashes indicate that no particular composition is implied; rather it suggests a generic name for any amorphous or poorly crystalline calcium silicate hydrate. A wide variety of structural probes, including X-ray diraction, NMR spectroscopy, X-ray absorption spectroscopy, TEM and SEM, provide strong evidence that the average structure of CSH is generally similar to those of tobermorite and/or jennite [310]. These have similar structural characteristics where silicate anion layers have a net negative charge and are held together by Ca2 cations in the interlayer region, which also contains H2 O molecules. The fundamental dierence between the structures is the degree of polymerisation of the silicate anion network. While the present investigation is mainly concerned with the role of nano-meter sized colloidal silica during the hydration process of C3 S (Alite, Ca3 SiO5 ) clinker phase, a stable sample preparation procedure to monitor the hydration process had to be developed. By employing the complementary diuse reectance FTIR (DR-FTIR) spectroscopy, dierential scanning calorimetry (DSC)

0009-2614/$ - see front matter 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.cplett.2004.05.071

J. Bjrnstrm et al. / Chemical Physics Letters 392 (2004) 242248 o o

243

and quantum chemistry we show that CS constitutes highly ecient nucleation sites for monomeric silica units released from the clinker phase during hydration. With the introduction of reactive polymerisation sites DRFTIR spectra tell of increased consumption of C3 S clinker phase and also increased formation rate of CS H binding phase. Quantum chemistry was employed to elucidate vibrational characteristics for silicate structures with varying degree of SiOSi connectivity. The dehydration and dehydroxylation characteristics of hydrated C3 S pastes were investigated utilizing DSC and observed to be similar for those of clay minerals. Being a bulk technique, demonstration of consistency of the DSC behaviour with the surface biased DR-FTIR spectroscopy was an essential element making the observed evidences, of both CS induced acceleration of clinker dissolution and CSH polymerisation, conclusive.

2. Experimental Pure triclinic C3 S were supplied by Construction Technology Laboratories (Skokie, Illinois, USA). The colloidal silica sol used (Cembinder 50TM ) contained 15 wt% of solid material. The particle size was 5 nm with a 500 m2 /g specic surface area, supplied by EKA Chemicals AB, Bohus, Sweden. The C3 S pastes under investigation were hydrated in sealed containers prior DR-FTIR and DSC measurements. C3 S pastes were mixed with a water-to-C3 S ratio (w/c) of 0.4. Where CS was present it was diluted in the mixing water to ensure a homogeneous particle distribution. The dosages of CS used were 1 and 5 wt% of the sol, with respect to the weight of C3 S clinker phase. All spectroscopic measurements were carried out in a DR mode. The typical probe area in the present DRFTIR experiments had a diameter of 57 mm, and the observed spectra, therefore, provide information about the state of the sample surface on a macroscopic scale. It is also important to stress that no direct mechanical inuence on a sample surface (e.g. scraping or pressing with covering windows) was performed during the measuring procedure. This warrants that the obtained spectra reect only the changes due to the hydration process. Spectra were recorded with a Bruker IFS-66 FTIR spectrometer, with a sample compartment under constant N2 (g) ow, equipped with a standard DR unit (Graseby Specac). The spectral resolution was set to 2 cm1 and all spectra were collected as an average over 64 scans. The C3 S pastes were cast in the DR-FTIR sample holder and covered with Para FilmTM during hydration to prevent evaporation of water, and to obtain smooth sample surfaces. After achieving desired hydration times the covering lm was removed and the samples were allowed to dry for 7 h under the N2 (g) ow in the sample compartment, forcing residual water

not participating in the hydration products to evaporate. The eectiveness of the N2 (g) purging to halt the hydration process, as seen by dissolution of clinker phase, was monitored by recording spectra at 30 min intervals during this drying procedure. It was found that no further dissolution of clinker phase takes place during N2 (g) purging; however 7 h were needed to ensure that all uncoordinated spectator water was evaporated. An alternative drying procedure was employed in order to test for any systematic error introduced by the N2 (g) purging procedure. C3 S pastes were allowed to hydrate for 4 h and were subsequently dried in 200 C for 2 h prior DR-FTIR analysis. From the acquired spectra, consistency was found between the two drying procedures. DSC experiments were performed on a TA Instrument DSC Q10 analyser. Samples ($10 mg) were loaded into sealed alumina pans and heated from 35 to 550 C at a 10 C/min heating rate. An empty alumina pan was used as reference and the heat ow between the sample and reference pan was recorded. The computational procedure for the quantum chemical calculations employed follows that outlined in [11]. The B3LYP hybrid density functional [1214], as implemented in the GA U S S I A N 98 program package [15], was employed throughout. Vibrational frequencies were computed from analytical Hessians, and validation of possible stationary states on the potential energy surfaces were made by checking for any imaginary frequencies.

3. Results and discussion The DR-FTIR spectrum of pure un-hydrated C3 S (Fig. 1a) show a broad group of bands in the region from 817 to 953 cm1 attributed to anti-symmetrical stretching vibrations of SiO4 tetrahedrons [16,17]. The sharp band at 980 cm1 is characteristic for the SiOH vibration (refer to quantum chemical calculations). The band for SiO stretching vibrations is present at higher vibrational frequencies ranging from 1000 to 1200 cm1 [9,16]. The gradually decreasing IR absorption intensities for bands attributed to C3 S clinker phase during the hydration process can be seen in Fig. 1b and c, depicting the change in the whole spectral region and the 750 1400 cm1 region respectively, as a function of hydration time. These samples contain no CS and are denoted reference or Ref. in gure legends. Except from the gradual absorption decrease for the clinker bands (817953 cm1 ), and thus consumption of the C3 S clinker phase, as the hydration process progresses, spectra presented in Fig. 1c, also indicate that the band present at 1030 cm1 gradually broadens to include vibrations of higher frequencies. Yu et al. [18]

244

J. Bjrnstrm et al. / Chemical Physics Letters 392 (2004) 242248 o o

Fig. 1. DR-FTIR spectra of: (a) un-hydrated clinker phase; (b) reference samples hydrated for dierent times; 4, 8, 12 and 24 h; (c) reference samples in the region 7001400 cm1 ; (d) reference samples in the region 24004000 cm1 .

performed FTIR experiments on 1.4 and 1.1 nm tobermorite, where the latter contains a substantial concentration of Q3 sites (where Qn indicates n bridging oxygen/tetrahedron), and concluded that 1.1 nm tobermorite gave rise to a 1200 cm1 absorption band. The shift of the SiO stretching vibration to higher frequencies is also described as being a ngerprint evidence for the degree of polymerization with the formation of CSH phase as a result of cement hydration [19]. To test this qualitatively, quantum chemical model calculations were employed on silicates with varying connectivity. Fig. 2 displays geometry optimized structures for silicates with varying connectivity, similar to those found in the suggested layered structure of CSH binding phase, presented along with the calculated maximum vibrational frequency of SiO for each structure. The results obtained conrm that SiO vibrational frequencies indeed increase with increasing degree of connectivity, and add that strains in ring structures contribute further to the up-shift of SiOSi stretching vibration frequencies. The broad band at 25003600 cm1 (Fig. 1b and d) is due to stretching vibrations of OH groups in hydroxyls with a wide range of hydrogen bond strengths. As the

Fig. 2. Calculated vibrational frequencies for silicate structures with dierent connectivity.

J. Bjrnstrm et al. / Chemical Physics Letters 392 (2004) 242248 o o

245

Fig. 3. DR-FTIR spectra for reference, 1 and 5 wt% CS with increased hydration times: (a) hydration for 4 h, spectra in the 7001400 cm1 region, (b) hydration for 4 h, spectra in the 24004000 cm1 region, (c) hydration for 8 h, spectra in the 7001400 cm1 region, (d) hydration for 8 h, spectra in the 24004000 cm1 region, (e) hydration for 12 h, spectra in the 7001400 cm1 region, (f) hydration for 12 h, spectra in the 24004000 cm1 region, (g) hydration for 24 h, spectra in the 7001400 cm1 region, (h) hydration for 24 h, spectra in the 24004000 cm1 region.

246

J. Bjrnstrm et al. / Chemical Physics Letters 392 (2004) 242248 o o

clinker phase hydrates, more H2 O molecules become coordinated in the produced calcium silicate hydrate structure, participating with SiO groups to form an intricate network of hydrogen bonds. The intensity of the sharp band at 3650 cm1 , characteristic for nonhydrogen bonded OH groups, i.e., hydroxyl groups neither interacting with other hydroxyl groups nor with water molecules, decreases with the formation of CSH binding phase. These characteristics are discussed in more detail below. Comparing the reference samples with the samples including CS after dierent hydration times, interesting features regarding the eects of CS are revealed. In Fig. 3, DR-FTIR spectra at dierent hydration times and two dierent CS dosages (1 and 5 wt%) are presented. The spectra obtained show that when CS is present the bands attributed to pure C3 S (817953 cm1 ) decrease more rapidly compared to the reference sample, indicative that CS accelerate the dissolution of C3 S. From spectra presented in Fig. 3a and c there are signs that an increase in the amount of CS from 1 to 5 wt% further increase the dissolution rate of C3 S. This observation provides rm proof for CS not being dissolved in the hydration process but rather that its surface provides highly reactive sites for condensation of SiO4 monomers released from C3 S. After 12 h (Fig. 3e) of hydration the dierence between 1 wt% CS and 5 wt% CS become less, still they appear to dier from the reference sample. Fig. 3g show spectra obtained after 24 h of hydration. Here the dierence between the samples containing CS and the reference is seen to be further reduced. Fig. 3a,c,e and g indicate that the band present at 1030 cm1 exhibits some broadening to include vibrational modes at higher frequencies as the CS amount is increased. As mentioned previously, the broadening of this band is consistent with increased connectivity in the silicate network, but because of interference from the intrinsic CS SiOSi vibrational bands at this frequency window, the apparent broadening in the presence of CS becomes less conclusive than that which was observed for the reference sample. On examining the spectra in the 24004000 cm1 region (Fig. 3b,d,f and h one can follow the decrease of non-hydrogen bonded OH vibrations at 3650 cm1 , and possibly the evolution of a hydrogen bonded network as the hydration process progresses. The intensity decrease of the 3650 cm1 absorption band as the hydration process progresses is particularly informative. This sharp band, characteristic for non-hydrogen bonded hydroxyl groups; diminish as the hydration process is allowed to proceed. The dierences between the different samples are most pronounced after 48 h of hydration (Fig. 3b and d). After 12 h (Fig. 3f) only a small absorption shoulder is detected for the reference and the sample containing 1 wt% CS, while it is absent in the 5

wt% sample. Finally, after 24 h (Fig. 3h) only the spectrum for the reference sample seems to include a small shoulder at 3650 cm1 . This behaviour is interpreted as follows. Monomeric silica (Q0 ), released from the clinker phase, undergo polymerisation into Q2 and Q3 structures. In layered clay minerals, such structures are stabilized by Ca2 ions and water molecules. In congurations of this nature it is reasonable to expect substantial inter-layer hydrogen bonding between SiOH groups and OH coordinated to Ca2 ions. In as much as perfect cement produces C SH, i.e. analogues to clay minerals, in inter-grain voids, the segregation of e.g. Ca(OH)2 is indicative of incomplete hydration. Consistently, suppression of the 3650 cm1 absorption intensity, due to non-hydrogen bonded OH groups, is understood to result from longer hydration times during which water is not allowed to evaporate. It is conceivable that Ca(OH)2 is formed as a consequence of water deciency. By not allowing the cement paste to dehydrate, Ca(OH)2 precipitation becomes inhibited while Ca2 ions released from the clinker phase are allowed to participate in building the CSH clay structure. Interestingly, we are able to correlate a dry sample by the diminishing of HOH water bending absorptions at $1640 cm1 (see Fig. 4). This result in conjunction with the fairly robust broad band at 25003600 cm1 , due to OH stretch vibrations, implies the hydrogen bonding network of CSH to result from Q23 SiOH OH Ca2 complexes, and eectively excludes the presence of any appreciable amounts of molecular water in the matrix. Having thus established a stable sample preparation procedure and demonstrated the power of DR-FTIR in discriminating between samples subjected to dierent hydration times, it may be asked to what extent DR-

Fig. 4. The evolution of DR-FTIR spectra as a function of time during N2 (g) purging. Rapid dehydration the rst 3 h, no dierence between the spectra recorded after 7 and 12 h. Note the absence of the 1640 cm1 bending vibration of molecular water in the dry sample.

J. Bjrnstrm et al. / Chemical Physics Letters 392 (2004) 242248 o o

247

Fig. 5. Dierential scanning calorimetry curves of reference (Ref.) and samples containing 5 wt% CS at dierent hydration times: (a) 1 h hydration; (b) 4 h hydration; (c) 8 h hydration and (d) 24 h hydration.

FTIR really reects what goes on in the bulk of the sample. In order to investigate bulk properties as such, we performed DSC experiments. Two distinct phase transitions are seen during early hydration of C3 S. The rst transition occurs between 80 and 120 C and the second between 460 and 500 C (see Fig. 5). We ascribe the rst endotherm (80120 C) to dehydration of bulk like water. The second endotherm is ascribed to condensation of OH groups in the produced CSH structure. Similar endotherms has previously been observed and described by thermogravimetric studies of dierent clay structures including tobermorite and hillebrandite [2022]. Neither DR-FTIR spectral data nor the DSC plots support the formation of any appreciable amounts of Ca(OH)2 with increased hydration time, as DSC measurements on Ca(OH)2 within our lab gave rise to an endotherm at 530550 C. It is not uncommon that endotherms near 480 C are ascribed as calcinations of Ca(OH)2 when hydrated cement samples are under investigation [2325]. Fig. 5 shows DSC curves for hydrated C3 S in the absence (Ref.) and presence of 5 wt% CS at dierent hydration times. From this it is clear that the production of the phase responsible for the endothermic response at 450500 C evolves more rapidly when CS is present during the C3 S hydration process, although after 24 h the amounts of this phase is fairly independent of the CS additive. The DSC results suggest

that the addition of CS mostly aects the initial silica polymerisation rates rather than the amount of product ultimately formed. Thus, an independent qualitative connection is established between the surface sensitive DR-FTIR technique and the bulk sensitive DSC technique, regarding the accelerating performance of CS in the initial hydration process.

4. Concluding remarks In the present study, we have monitored the hydration process of C3 S pastes. Diuse reectance FTIR signatures of C3 S phase dissolution, silica polymerisation, and hydrogen- and non-hydrogen bonded OH groups have been recorded. DSC has been employed to make a connection between surface characteristics and bulk properties of the hardening paste. Furthermore, the accelerating eects of a 5 nm colloidal silica additive on the rate of C3 S phase dissolution, calciumsilicatehydrate CSH gel formation and removal of non-hydrogen bonded OH groups were demonstrated. The accelerating eects were more pronounced when the concentration of colloidal silica CS was increased from 1 to 5 wt%. However, already the presence of small amounts of CS particles (1 wt% of the sol) seems to have considerable eects on the hydration process of C3 S clinker phase.

248

J. Bjrnstrm et al. / Chemical Physics Letters 392 (2004) 242248 o o

Of conceptual importance is that the observation of increase in C3 S consumption with an increase of the CS dosage is understood, and that this contradicts any scenario which involves the alkaline dissolution of CS, since this would provide an additional source of monomeric silica, which would suppress the C3 S dissolution process. Rather, the accelerated C3 S dissolution upon adding CS particles is taken to be due to the highly reactive and large (500 m2 g1 ) surface of the nanoparticles. The reactivity is due to surface Q23 SiOH groups, which cover the nano-particle and constitute condensation sites for monomeric silica units released from the clinker phase. The surface DR-FTIR and the bulk DSC data independently imply that the accelerating eects from CS, with respect to the rate of C3 S consumption and CSH formation, are primarily present during the initial stages of hydration (412 h) and that CS particles act as condensation centres. The competition between Ca(OH)2 precipitation and CSH formation is known to be detrimental to the quality of any C3 S based cement. Independent proof for the well known essential role of water during cement hydration was provided. With the suppression of water evaporation during the hydration process, DR-FTIR absorption intensities associated with non-hydrogen bonded OH groups was seen to gradually diminish with time. This transformation under humid conditions was understood in terms of Ca2 ions, released during the dissolution of the clinker phase, increasingly preferring to act as counter-ions to the growing silicate anion network, thus contributing to the formation of the calcium silicate hydrate gel. The increasing DSC response at 450 C, with increasing hydration time, would correspond to water condensation and evaporation from the Q23 SiOH OH Ca2 CSH gel, thus producing a ceramic compound with Q23 SiO Ca2 local connectivity. The objective of the present study has been to provide means and monitor the hydration of C3 S, but our goal is to the signicantly more complex task of contributing to a comprehensive conceptual chemical understanding of the performance and workability of concrete. A subsequent study will address the eects on the rate of hydration of the clinker-nano-silica system of additives which improve the workability of the cement.

Acknowledgements Support from The Knowledge Foundation (KK-stiftelsen, Stockholm), Swedish Research Council, and EKA Chemicals, Bohus, Sweden are gratefully acknowledged.

References
[1] S. Chandra, in: V.M. Malhotra (Ed.), Proceedings of the 6th CANMET/ACI International Conference on Fly ash, Silica Fume, Slag and Natural Pozzolan in Concrete Supplementary papers, Bangkok, 1998, p. 323. [2] J. Bjrnstr m, A. Martinelli, J.R.T. Johnson, A. Matic, I. Panas, o o Chem. Phys. Lett. 380 (2003) 165. [3] R.J. Kirkpatrick, G.E. Brown, N. Xu, X.D. Cong, Adv. Cem. Res. 9 (1997) 31. [4] H.F.W. Taylor, J. Am. Ceram. Soc. 69 (1986) 464. [5] H.F.W. Taylor, Cement Chemistry, Academic Press, London, 1992. [6] X.D. Cong, R.J. Kirkpatrick, Adv. Cem. Bas. Mater. 3 (1996) 133. [7] X.D. Cong, R.J. Kirkpatrick, Adv. Cem. Bas. Mater. 3 (1996) 144. [8] D.E. Macphee, E.E. Lachowski, F.P. Glaseer, Adv. Cem. Res. 1 (1988) 131. [9] R.J. Kirkpatrick, J.L. Yarger, P.F. McMillan, P. Yu, X.D. Cong, Adv. Cem. Bas. Mater. 5 (1997) 93. [10] H.M. Jennings, B.J. Dalgleish, P.L. Pratt, J. Am. Ceram. Soc. 64 (1981) 567. [11] J.R.T. Johnson, I. Panas, Chem. Phys. 276 (2002) 45. [12] A.D. Becke, J. Chem. Phys. 98 (1993) 5648. [13] C. Lee, W. Yang, R.G. Parr, Phys. Rev. B 37 (1988) 785. [14] S.H. Vosko, L. Wilks, M. Nusair, Can. J. Phys. 58 (1980) 1200. [15] M.J. Frisch et al., GA U S S I A N 98, Gaussian, Inc., Pittsburgh, PA, 1998. [16] A.H. Delgado, R.M. Paroli, J.J. Beaudoin, Appl. Spectrosc. 50 (8) (1996) 970. [17] T.L. Hughes, C.M. Methven, T.G.J. Jones, S.E. Pelham, P. Fletcher, C. Hall, Adv. Cem. Bas. Mat. 2 (1995) 91. [18] P. Yu, R.J. Kirkpatrick, B. Poe, P.F. McMillan, X.D. Cong, J. Am. Ceram. Soc. 82 (1999) 742. [19] M.Y.A. Mollah, W. Yu, R. Schennach, D.L. Cocke, Cem. Concr. Res. 30 (2000) 267. [20] S. Shaw, C.M.B. Henderson, B.U. Komanschek, Chem. Geol. 167 (2000) 141. [21] H. Ishida, K. Mabuchi, K. Sasaki, T. Mitsuda, J. Am. Ceram. Soc. 75 (1992) 2427. [22] R.L. Frost, Z. Ding, Thermochim. Acta 397 (2003) 119. [23] W. Sha, E.A. ONeill, Z. Guo, Cem. Concr. Res. 29 (1999) 1487. [24] W. Sha, G.B. Pereira, Cem. Concr. Comp. 23 (2001) 455. [25] J. Zelic, D. Rusic, R. Krstulovic, J. Therm. Anal. Calorim. 67 (2002) 613.

Vous aimerez peut-être aussi