Cement and Concrete Research: M.D.A. Thomas, R.D. Hooton, A. Scott, H. Zibara
Cement and Concrete Research: M.D.A. Thomas, R.D. Hooton, A. Scott, H. Zibara
Cement and Concrete Research: M.D.A. Thomas, R.D. Hooton, A. Scott, H. Zibara
a r t i c l e i n f o a b s t r a c t
Article history: This paper reports the results of a study to determine the effects of supplementary cementitious materials
Received 12 February 2010 (SCMs) on the chloride binding of portland cement pastes. The results show that SCMs with significant
Accepted 12 January 2011 quantities of alumina increase the binding capacity of cement paste. Pastes with metakaolin (45% Al2O3)
showed the greatest chloride binding capacity and pastes with silica fume (b 0.5% Al2O3) showed the least
Keywords:
binding. The chemical binding in solutions of high chloride concentration is mainly attributed to the
Chloride binding
Fly ash (D)
formation of Friedel's salt, C3A·CaCl2·10H2O. When pastes originally exposed to high chloride concentrations
Granulated blast-furnace slag are subsequently exposed to chloride-free solution, a portion of the bound chloride is released, but a
Silica fume (D) significant portion remains irreversibly bound. There is some evidence that Friedel's salt may partially convert
Metakaolin (D) to Kuzel's salt, C3A·(0.5CaCl2)(0.5CaSO4)·12H2O, under these conditions. The binding relationships were best
described by the Freundlich isotherm and binding coefficients are given for all the binders tested.
© 2011 Elsevier Ltd. All rights reserved.
0008-8846/$ – see front matter © 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cemconres.2011.01.001
2 M.D.A. Thomas et al. / Cement and Concrete Research 42 (2012) 1–7
metakaolin, blast furnace slag and three fly ashes; the fly ashes were Table 2
selected to have varying calcium contents to cover the range Bogue cement composition.
Portland Silica Meta- Slag Fly ash Fly ash Fly ash Table 3
cement fume kaolin (SG) (FA) (FA1) (FA2) Details of paste mixtures investigated.
(OPC) (SF) (MK)
Mix W/CM Portland Silica Meta- Fly ash Slag
SiO2 21.26 94.48 52.01 36.18 53.89 43.07 40.82
cement fume kaolin (%) (%)
Al2O3 4.09 0.24 44.72 10.02 24.65 20.78 11.09
(%) (%) (%)
TiO2 0.2 0.01 1.6 0.67 1.14 1.10 0.47
P2O5 0.07 0.14 0.09 0.01 0.71 0.67 0.10 OPC 0.3, 0.5 100
Fe2O3 2.89 0.63 0.58 0.50 8.63 5.41 6.21 8SF 0.3, 0.5 92 8
CaO 63.58 0.44 0.00 35.49 4.37 17.72 29.89 8MK 0.3, 0.5 92 8
SrO 0.12 0.02 0.02 0.04 0.16 0.54 0.30 25FA 0.3, 0.5 75 25
MgO 2.47 0.38 0.00 0.66 0.83 4.18 4.41 25FA1 0.5 75 25
Mn2O3 0.06 0.03 0.01 13.58 0.05 0.17 0.06 25FA2 0.5 75 25
Na2O3 0.17 0.16 0.32 0.43 0.6 1.46 1.14 40FA 0.3 60 40
K2O 0.62 1.01 0.19 0.50 2.14 0.75 1.71 25SL 0.3, 0.5 75 25
Na2Oe 0.58 0.82 0.45 0.77 2.07 1.95 2.26 40SL 0.3 60 40
SO3a 2.79 0.36 0.12 1.51 0.61 2.17 2.17 25FA6SF 0.3, 0.5 69 6 25
LOIb 0.99 2.87 0.9 1.72 1.53 2.57 1.30 40FA5SF 0.3 55 5 40
a
25SL6SF 0.3, 0.5 69 6 25
SO3 — based on sulfur determined by LECO.
b
40SL5SF 0.3 55 5 40
LOI — loss on ignition at 1000 °C.
M.D.A. Thomas et al. / Cement and Concrete Research 42 (2012) 1–7 3
3.1 where α and β are binding constants. The Langmuir isotherm provides
a reasonable fit to the experimental data, but in all cases the
OPC-0.30
Chloride Concentration (Mol.)
Table 4
“Best-fit” binding coefficients for Freundlich isotherms.
25
15 OPC
w/cm = 0.30
8SF w/cm = 0.30 25FA/6SF
20 8MK OPC
OPC 25SG/6SF w/cm = 0.50
w/cm =
8SF 10 25FA/6SF
15 0.50
8MK
10
5
0 0
0 1 2 3 0 1 2 3
Free Chloride, Cf (Mol.) Free Chloride, Cf (Mol.)
Fig. 3. Chloride binding isotherms for pastes with silica fume and metakaolin Fig. 5. Chloride binding relationships for ternary cement blends (w/cm = 0.30 and 0.50,
(w/cm = 0.30 and 0.50, T = 23 °C). T = 23 °C).
the peaks corresponded to ettringite. Fig. 9 shows a plot of the peak heights shown in Fig. 11 using the relationship shown in Fig. 9,
maximum peak intensity for Friedel's salt (at d = 7.9 Å) against the the results yield 14.0 mg/g and 11.8 mg/g, respectively for the 8MK
bound chloride content, Cb, for the pastes after reaching equilibrium paste after reaching equilibrium with the 200 mL and 1000 mL of
with the 3-molar chloride solution. There is a good correlation chloride-free solution. This compares with 19.2 mg/g and 13.1 mg/g
between these two parameters suggesting that much of the binding calculated based on the solution equilibrium concentration (Table 5).
that occurs at this high concentration of chloride can be attributed to Fig. 11 shows the presence of a second peak (marked K? in the
the formation of Friedel's salt. figures) after the 8MK has reached equilibrium with 1000 mL of
The “desorption isotherm” for the paste with 8% metakaolin is chloride-free solution and this peak coincides with the maximum
shown in Fig. 10 and the results from these tests for all three binders (2-theta) intensity peak of Kuzel's salt, C3A·(0.5CaCl2)(0.5CaSO4)·
tested are shown in Table 5. After adding 200 mL of chloride-free, 12H2O. Kuzel's salt was found at lower chloride concentrations in
saturated Ca(OH)2 solution and allowing equilibrium to be reached, other parts of this research [16], increasing the likelihood that this
all three binders retained about 80% of the bound chlorides. The peak was indeed Kuzel's salt in this instance.
chloride concentrations at equilibrium were in the range of 0.12 to
0.14 M. At these chloride concentrations the bound chlorides from the 4. Discussion
three pastes during sorption were in that range of 3 to 6% (Fig. 2).
With a 1000-mL addition, the equilibrium chloride concentration was The chloride binding capacity of the cement pastes was not only
less than 0.04 M, but still about 40% of the chlorides initially bound influenced by the composition of the binder but also by the w/cm. The
after reaching equilibrium with the 3-molar chloride solution impact of w/cm was particularly pronounced for binders containing
remained bound. Fig. 11 shows the results of XRD analysis of the SCM especially for SCMs that contained large quantities of alumina
8MK paste once equilibrium had been reached after the addition of (e.g. 8MK and, to a lesser extent, 25FA and 25SG). This may indicate
200 mL and 1000 mL of chloride-free solution. Although the main that the hydration and pozzolanic reactions are stifled at the lower
peak corresponding to Friedel's salt diminishes as the paste reaches w/cm and that much of the alumina present does not become available
equilibrium with the lower chloride concentrations, significant for the formation of Friedel's salt.
Friedel's salt still remains even when the free chloride falls below For the 0.50-w/cm pastes there is clearly a link between the
0.04 M. If the bound chloride concentration is estimated from the alumina content of the SCM and the extent to which the incorporation
15 20
Bound Chloride,Cb (mg/g sample)
15
10
OPC 10
w/cm = 0.30 25SG OPC
5 25FA 25FA
OPC 5 25FA1
w/cm = 0.50 25SG
25FA2
25FA
0 0
0 1 2 3 0 1 2 3
Free Chloride, Cf (Mol.) Free Chloride, Cf (Mol.)
Fig. 4. Chloride binding relationships for fly ash and slag (w/cm = 0.30 and 0.50, Fig. 6. Chloride binding isotherms for pastes with different fly ashes (w/cm = 0.50,
T = 23 °C). T = 23 °C).
M.D.A. Thomas et al. / Cement and Concrete Research 42 (2012) 1–7 5
25 1,000
25FA-2m
Bound Chloride,Cb (mg/g sample)
25FA-9m 8MK
5 200
8SF
0 0
0 1 2 3 8 12 16 20 24
Free Chloride, Cf (Mol.) Bound Chloride,Cb (mg/g sample)
Fig. 7. Chloride binding isotherms for pastes with different fly ashes (w/cm = 0.50, Fig. 9. Relationship between the maximum peak intensity for Friedel's salt (d = 7.9 Å)
T = 23 °C). and the bound chloride of cement pastes after immersion in NaCl solution, Ci = 3 M.
of the SCM increases binding, especially at the higher chloride fume (25FA and 25SG). Nilsson et al. [17] suggested that silica fume
concentrations used in this study. Metakaolin, which comprises influences binding in three ways: (i) dilution of the C3A which may
approximately 45% Al2O3, most of which is assumed to be available, reduce binding, (ii) reduction of the pH of the pore solution which
produces the largest increase in the chloride binding capacity, should increase chloride binding, and (iii) and increase in the amount
whereas silica fume, which contains a negligible amount of alumina of C–S–H which should increase physical binding of chlorides. The
(b0.5% Al2O3), actually reduces the binding capacity compared to the incorporation of silica fume is also known to reduce the C/S ratio of the
control (OPC). Fly ash and slag (10 to 25% Al2O3) increase binding but C–S–H and it has been suggested that this will lead to a reduction in the
not to the same extent as metakaolin. Fig. 12 shows the relationship amount of chloride that can be physically adsorbed onto the surface of
between the bound chloride after immersion in NaCl solution the C–S–H [14,18]. Reductions in the C/S ratio of the C–S–H by slag and
(Ci = 3 M) and the alumina content of the binder (w/cm = 0.50, fly ash also occur, but these SCMs have higher alumina contents and
T = 23 °C). For binders without fly ash, the relationship between these the subsequent formation of C–A–S–H may result in increased binding,
two parameters is clear, however, binders with fly ash do not follow in addition to increased formation of Friedel's salt [19]. Experiments by
the same relationship. Regarding fly ash, there appears to be no clear Zibara [16,20] on lime–silica fume pastes confirmed this phenomenon
criteria, such as total Al2O3 or CaO content, that determine its effect on as the chloride binding was observed to decrease as the lime/silica
binding. However, much of the alumina in fly ash is present in fume ratio decreased. Other workers [21,22] have also observed a
crystalline components such as mullite, which would not be available reduction in binding due to the presence of silica fume.
to form calcium chloro-aluminates. Consequently, the total alumina The desorption tests indicate that chloride binding is not a
content, as reported in Table 1, does not give a reliable indication of completely irreversible process and that a portion of the chlorides that
how a fly ash will behave with regards to chloride binding. What is are bound may be released if the free chloride concentration in the pore
required is some measure of the alumina that is “available” and this solution decreases. However, there is significant hysteresis in the
could be determined using Rietveld quantitative XRD analysis. absorption–desorption process and a significant quantity of chlorides
The binding capacity of paste containing portland cement plus remains bound, mainly as Freidel's salt, even when the free chloride
silica fume (8SF) was lower than that of the control paste (OPC). In concentration approaches zero. In other words, a portion of the
addition, the incorporation of 5% silica fume into ternary mixes with chemically-bound chlorides is irreversibly bound. The XRD pattern for
25% fly ash or slag (25FA/5SF and 25SG/5FA) reduced the binding
compared with binary mixes with fly ash and slag, but without silica
25
Bound Chloride,Cb (mg/g sample)
20
15
10
0
0 1 2 3
Free Chloride, Cf (Mol.)
Fig. 8. X-ray diffraction patterns for pastes after immersion in NaCl solution (Ci = 3 M) Fig. 10. Chloride “desorption isotherm” for paste with metakaolin (w/cm = 0.50,
(w/cm = 0.50, T = 23 °C) F = Friedel's salt. T = 22 °C).
6 M.D.A. Thomas et al. / Cement and Concrete Research 42 (2012) 1–7
Table 5 25
[17] L.-O. Nilsson, E. Poulson, P. Sandberg, H.E. Sorensen, O. Klinghoffer, in: J.M. [20] H. Zibara, R.D. Hooton, M.D.A. Thomas, K. Stanish, Influence of the C/S and C/A
Fredriksen (Ed.), HETEK: Chloride Penetration into Concrete, State-of-the-Art, ratios of hydration products on the chloride ion binding capacity of lime–SF and
Transport Processes, Corrosion Initiation, Test Methods and Prediction Models, lime–MK mixtures, Cement and Concrete Research 38 (2008) 422–426.
Report, vol. 53, The Road Directorate, Denmark, 19968, 151 pp. [21] C.L. Page, O. Vennesland, Pore solution composition and chloride binding capacity
[18] G.K. Glass, N.M. Hassanein, N.R. Buenfeld, Neural network modelling of chloride of silica fume cement pastes, Material and Construction 16 (19) (1983) 19–25.
binding, Magazine of Concrete Research 49 (181) (1997) 323–335. [22] Rasheeduzzafar, S. Ehtesham Hussain, A.S. Al-Gahtani, Pore solution composition
[19] S.-U. Hong, F.P. Glasser, Alkali sorption by C–S–H and C–A–S–H gels Part II. Role of and reinforcement corrosion characteristics of microsilica blended cement
alumina, Cement and Concrete Research 32 (2002) 1101–1111. concrete, Cement and Concrete Research 21 (6) (1991) 1035–1047.