Third International Conference on Sustainable Construction Materials and

Technologies http://www.claisse.info/Proceedings.htm

Effect of different MgO on the hydration of MgO-activated

granulated ground blastfurnace slag paste
Fei Jin1*, Adel Abdollahzadeh1 and Abir Al-Tabbaa1
1
Department of Engineering, University of Cambridge, Trumpington Road, Cambridge CB2
1PZ, United Kingdom
*
Geotechnical research office, Department of Engineering, University of Cambridge,
Trumpington Road, Cambridge CB2 1PZ, United Kingdom
Email addresses: fj232@cam.ac.uk (F. Jin); aa651@cam.ac.uk (A. Abdollahzadeh);
aa22@cam.ac.uk (A. Al-Tabbaa)

ABSTRACT
Reactive MgO recently emerged as a potential activator for granulated ground blast-furnace
slag (GGBS), which is one of the most widely used by-products in the cement industry.
However, it is known that the characteristics of reactive MgO vary significantly, which may
affect the activation process. In this study, seven commercial reactive MgOs, whose
reactivities vary widely, were chosen to activate GGBS. The unconfined compressive
strength of the MgO-GGBS paste at up to 90 days was measured. It was found that both the
reactivity and composition of the MgO samples have effects on the strength development.
The major hydration products were identified as hydrotalcite-like phases (Ht) and calcium
silicate hydrate (CSH) using X-ray diffraction (XRD). The microstructural characteristic was
examined by scanning electron microscopy (SEM). It is concluded that the characteristics of
MgO (e.g., reactivity, composition) significantly influence the performance of MgO-GGBS
blends.

Key words: MgO, granulated ground blast-furnace slag, reactivity, hydration, microstructure

INTRODUCTION
Portland cement (PC) production generates a substantial amount of CO2 from the
decomposition of limestone at ~1450°C, the combustion of fuels in the kiln, as well as from
power generation (Hbert et al., 2010). It is estimated that the cement production accounts for
5–8% of man-made CO2 emissions (Scrivener and Kirkpatrick 2008). As a result, various
attempts to reduce cement clinker production have been made, one of which is the extensive
usage of industrial by-products. Alkali-activated binders based on ground granulated blast-
furnace slag (GGBS) are sustainable alternatives to PC due to their low initial capital cost of
the raw material and the saving of energy and resources resulting in lower CO 2 emission
(Yang et al., 2012). It is estimated that more than 80% of CO2 can be reduced by substituting
alkali-activated binders for PC (Juenger et al., 2011). However, several issues concerning the
use of strong alkali exist, which prevents the widespread use of alkali-activated binders

1
including over-rapid setting, the difficulty of handling the caustic alkali and uneconomical
efficiency (Yang et al. 2012). In this sense, alkaline-earth hydroxides such as Ca(OH)2,
Sr(OH)2 and Ba(OH)2 were studied and found to be able to facilitate the hydration of GGBS
(Roy et al. 1992; Yang et al. 2010). Recently, reactive MgO emerged as an effective
activator for GGBS. Li (2012) and Yi et al. (2012) reported that reactive MgO increased the
hydration of GGBS significantly and slag activated by 10-20% MgO outperformed
corresponding lime-activated slag with strength values of up to 4 times higher. The effect of
MgO in the alkali-activated slag system has been investigated by many researchers, either in
the form of internal MgO from slag composition (Ben Haha et al. 2011), or external MgO
addition (Shen et al. 2011). It should be noted that since the GGBS is usually produced
between 1400-1600°C (Shi et al., 2006), the internal MgO is categorised as dead burned
MgO (Shand 2006); whereas reactive grade (calcined under 1000°C) is often selected as the
MgO added externally. Ben Haha et al. (2011) investigated the effect of MgO content
(internal) on the performance of alkali activated cement and revealed that the main hydration
products are calcium silicate hydrate (CSH) gel and hydrotalcite (Mg6Al2CO3(OH)16·4H2O)-
like phases (Ht). Higher content of MgO produced more Ht, which is more voluminous,
resulting in higher strength. Reactive MgO was used to replace 40% of slag, activated with
K2CO3 and sodium silicate, and it was found that the strength was only slightly lower than
the control (Gu 1991). In the work by Shen et al. (2011), light-burnt dolomite (mainly
contains reactive MgO and CaO) was added in sodium silicate-activated slag and fly ash
blends, and the results indicated that hydration of reactive MgO decreased the shrinkage of
the paste, and no interaction between MgO and other components was observed by the
microstructural analysis. Although it is known that the characteristics of reactive MgO vary
significantly (Shand 2006), which may affect the activation process, none of these studies
investigated or presented the effect of the characteristics of MgO on the performance of
MgO-GGBS blends.

Hence this paper presents an investigation of the hydration performance of MgO-activated

GGBS using seven different commercially available reactive MgO samples. The unconfined
compressive strength (UCS) of the pastes was measured at different curing times. X-ray
diffraction (XRD) was used to identify the hydrated products and scanning electron
microscopy (SEM) was employed to study the microstructure of the blends. The effect of the
characteristics of the MgO samples on the slag activation process was discussed based on the
experimental results obtained.

MATERIALS AND EXPERIMENTAL METHODS

The sources and chemical compositions of the seven commercial reactive MgOs studied are
shown in Table 1: five MgO samples were obtained from calcined magnesite, one
synthesised from seawater and one synthesised from chemicals. GGBS was obtained from
Hanson, UK, whose composition is also presented in Table 1. The reactivity test was used to
measure the time duration required for the neutralisation of an acidic solution (in this case
0.25M acetic acid) by a certain MgO sample mass (in this case 5.0g) and Phenolphthalein
was adopted as the pH indicator (Shand 2006). The time from adding the MgO to the change
of the solution colour is recorded as the reactivity and the shorter the time, the more reactive
the MgO is. The test was performed in duplicate for each sample and the average value was
reported here.

To prepare the MgO-GGBS pastes, raw materials were hand mixed for 5 minutes in a bench-
scale food mixer and cast into ϕ3.4×6.8cm mini cylindrical moulds. The fresh cement paste
were demoulded after 24 hours and then transferred into a water tank and cured for 7, 28 and

2
90 days. In this study, the MgO:GGBS ratio was fixed at 1:9 according to a preliminary
study, and the water to cement ratio was fixed as 0.4 in order to focus on the influence of the
MgO on the performance of the MgO-GGBS blend.

The unconfined compressive strength (UCS) of the samples, in triplicate, was determined
according to British Standard (BS EN 196-1:2005) at each curing age. Some sliced
specimens were collected after crushing and part of the samples were ground to powder. The
powder and sliced specimens were stored in acetone to arrest the hydration and vacuum
dried for at least 7 days before XRD and microstructure analysis. Once dried, the samples
were further ground to pass through a 75µm sieve. XRD measurement was carried out on the
Siemens D5000 X-ray diffractometer with a CuKα source operating at 40kV and 40mA
using a scanning range from 5 to 55 (2ϴ), with scanning speed of 1s/step and resolution of
0.05°/step. The sliced specimen was gold coated and the morphology was examined on the
JEOL 820 scanning electron microscopy (SEM).

Table 1. Source, label and chemical compositions of the raw materials used in
this study

SourceSupplier Label MgO CaO Fe2O3 Al2O3 SO3 Cl SiO2 LOI

RBH*,
CS 87.9 2.9 0.3 0.13 - - 8.7 1.6
Greece
Styromag,
K10 80.0 6.9 2.8 - - - 6.0 7
Austria
RBH*,
Magnesite CG 88.3 2.6 0.2 0.18 - - 8.8 1.9
Greece
RBH*,
NOR 91.6 1.7 0.1 0.14 - - 6.5 2.3
Greece
RBH*,
CR 88.9 2.6 0.2 0.15 - - 8.2 2.2
Greece
ICL
Industrial
Seawater DSP 99.2 0.2 0.01 0.004 0.125 0.1 0.02 0.8
Products,
Israel#
Chemical Intermag
IM 94.5 0.3 0.02 0.01 0.2 0.2 0.02 6
reaction Ltd, UK
Hanson,
- GGBS 8.5 39.5 0.5 12.5 1 - 36.5 -
UK
*
Richard Baker Harrison Ltd, UK; # Also contains traces of B2O3 and Na

RESULTS AND DISCUSSION

Characterisation of MgO
Table 1 shows the effect of the different origins on the chemical compositions of the
different MgO samples. The MgO content ranged from 80% to 99.2%; where the synthetic
MgO samples generally have higher purity, with MgO content of over 94% for both samples.
The major impurities of calcined MgO are CaO and SiO2, which originate from common
minerals in rocks, while the major impurities of the synthetic MgO are CaO, Cl, and SO3

3
derived from seawater and chemical reagents. It is noteworthy that K10 has a much higher
CaO content than other MgO samples.

The reactivity values (as the time increases, the reactivity decreases) of all the seven MgO
powders tested are plotted in Figure 1. It is found that the synthetic ones (i.e., IM and DSP),
had the highest reactivities, which were quite similar, while the MgO samples calcined from
magnesite showed much lower reactivities and a much wider range. This confirms that the
reactivity of the calcined MgOs is strongly affected by the thermal conditions of the
calcination process (Birchal et al. 2000; Mo et al. 2010; Sun et al. 1991). Higher reactivity
suggests more brucite formed at the same curing age, which will participate in the reaction
with GGBS.
240
220
200
180
Reactivity (s)

160
140
120
100
80
60
40
20
IM
CG

NOR

CR
K10
CS

DSP

Figure 1. The reactivity of the different commercial MgOs tested

UCS Results
Figure 2 shows the UCS values of all MgO-GGBS pastes at curing ages of 7, 28 and 90 days,
which clearly indicates that the MgO successfully activates the GGBS and the strength
continues to increase with time. It is found that using different MgO samples, the UCS of
MgO-GGBS pastes can achieve between 23~30MPa after 90 days’ curing. The effect of
reactivity on the UCS of MgO-GGBS pastes is shown in Figure 3. At the early age (7 days),
the strength increased with reactivity (i.e., decrease of reactivity value). However, in the
long term, the UCS of the less reactive MgOs increased more remarkably than those of the
most reactive ones (IM and DSP), which will be explained later in this paper. In addition,
K10 exhibited higher strengths than expected at all ages, which is attributed to its higher
CaO content. This phenomenon is consistent with Lu et al. (1957) that the combined use of
CaO and MgO is more beneficial for strength development. There are two roles that CaO
plays in the system. Firstly, CaO increased the pH of the system since the equilibrium pH of
Ca(OH)2 is 12.5, which is much higher than Mg(OH)2 (10.5) (Zhang et al. 2011). The
increase of pH facilitates the dissolution of GGBS, resulting in higher hydration degree
(Song et al. 2000). In addition, CaO, when it hydrates, participates in the reaction with
GGBS to form more CSH, which contributes to strength.

4
 32
30 CS
28 K10
26 CG
24 NOR
22 CR
UCS (MPa) 20 DSP
18 IM
16
14
12
10
8
6
4
2
0
0 10 20 30 40 50 60 70 80 90 100
Curing time (d)

Figure 2. Strength development of MgO-GGBS pastes

32
30 7d
28 28d
26 90d
24
22
UCS (MPa)

20
18
K10
16
14
12
10
8
6
4
2
0
0 20 40 60 80 100 120 140 160 180 200 220 240
Reactivity (s)

Figure 3. Effect of MgO reactivity on the UCS of MgO-GGBS pastes

XRD Results

To study the hydration products of the MgO-GGBS blends and the effect of MgO, the highly
reactive and less reactive MgO samples (i.e., DSP and CS, respectively) activated GGBS
were selected for XRD and SEM analysis. Figure 4 shows the XRD curves of the pastes at
different curing ages. The broad and diffuse peak at 25-35° 2ϴ reflects the short range order
of the CaO-Al2O3-MgO-SiO2 glass structure. The curves showed the presence of Ht and
CSH as the major hydration products, which is consistent with (Ben Haha et al. 2011). The
most obvious difference between the two reactive MgO-activated GGBS was the intensity of
the MgO and Ht. As expected, higher reactivity resulted in less MgO remained in the blends
at every curing time. When DSP (higher reactivity) was used, Ht was observed at 7 days and
its intensity increased with curing time, while CS activated GGBS showed no sign of Ht in
the short term (7 days). It should be noted that in both systems, there was no brucite detected
indicating that the hydrated MgO was all consumed in the hydration. The fate of Mg ion was

5
probably attributed to the formation of magnesium silicate hydrates (MSH) first and Ht at
later age. The existence of MSH by the reaction of MgO and silica fume at ambient
temperature was confirmed in numerous papers (Ono and Wada 2007; Wei et al. 2006, 2011;
Zhang et al. 2011). However in this study, due to the low crystallinity of MSH and its
miscibility with CSH (Brew and Glasser 2005), it is hard to be distinguished by XRD. As
suggested by (Ben Haha et al. 2011), Ht is more voluminous than CSH gels, resulting in a
matrix with lower porosity in an adequate quantity; however, it is found that when too much
Ht is present in the blend, the over-expansion causes volume instability and may be
detrimental to the strength as presented in Figure 3. Although not detected here, ettringite is
believed to be formed due to sulfate (S) in MgO-GGBS blends, which is confirmed in an
early study using thermogravimetric analysis. Since both the highly reactive MgO contains S
as well as slag itself, their low strength at later ages was also attributed to the expansion
caused by formation of ettringite.

CSH
MgO
Ht
90 day
Intensity (a.u.)

28 day

7 day

5 10 15 20 25 30 35 40 45 50 55
Two Theta (degree)
(a)
CSH
MgO
Ht
Intensity (a.u.)

90 day

28 day

7 day

5 10 15 20 25 30 35 40 45 50 55
Two Theta (degree)
(b)

Figure 4. XRD curves of (a) DSP-GGBS blends; (b) CS-GGBS blends

6
SEM Results
Figures 5 and 6 show the microstructure of CS-GGBS and DSP-GGBS blends at 28 and 90
days. At 28 days, micro-cracks were observed in DSP-GGBS blend with brucite aggregates
dispersed on the surface of CSH gel. While for CS-GGBS blend, the irregular unhydrated
slag particles were identified and no cracks were formed at this stage. This explains the UCS
results that although DSP is more reactive than CS and produced more hydration products
accordingly in the short term, the higher expansion destroyed the integrity of the matrix
resulting in lower UCS. As mentioned before, the cracks could be attributed to the formation
of ettringite and Ht. At 90 days, both blends were found to contain many cracks in the matrix
with brucite particles on the surface, but the crack propagation is more significant in 83CS-
GGBS than in DSP-GGBS blend. It is concluded that MgO with lower reactivity is more
detrimental to the cement soundness in the long term due to the latent hydration of less
reactive MgO.

(a) (b)

(c) (d)

Figure 5. SEM images of DSP-GGBS blends (a) (b) at 28d; (c) (d) at 90d

7
 (a) (b)

(c) (d)

Figure 6. SEM images of CS-GGBS blends (a) (b) at 28d; (c) (d) at 90d

CONCLUSIONS
By investigating the hydration of different MgO activated GGBS pastes, the following
conclusions can be drawn:

1. The characteristics of reactive MgO vary significantly in terms of their reactivity and
chemical composition;
2. The strength of MgO-GGBS blend is affected by the reactivity of the MgO. More
reactive MgO resulted in higher early strength.
3. The CaO content in the reactive MgO samples is beneficial to strength development;
4. The major hydration products are hydrotalcite-like phases (Ht) and CSH. More
reactive MgO reacts faster than less reactive ones, leading to formation of Ht in a
shorter time. However, it also causes cracks resulting in strength decrease earlier
than less reactive MgO samples. In longer term, due to the latent hydration of less
reactive MgO, the cement soundness is also destroyed.

ACKNOWLEDGEMENT
The authors are grateful to Cambridge Overseas Trust (COT) and China Scholarship Council
(CSC) for their financial help of the PhD studentship for the first author.

8
REFERENCES
Ben Haha, M., Lothenbach, B., Le Saout, G., and Winnefeld, F. (2011). “Influence of slag
chemistry on the hydration of alkali-activated blast-furnace slag—Part I: Effect of
MgO.” Cement and Concrete Research, 41(9), 955–963.
Birchal, V. S., Rocha, S. D., and Ciminelli, V. S. (2000). “The effect of magnesite
calcination conditions on magnesia hydration.” Minerals engineering, 13(14-15),
1629–1633.
Brew, D. R. M., and Glasser, F. P. (2005). “Synthesis and characterisation of magnesium
silicate hydrate gels.” Cement and Concrete Research, 35(1), 85–98.
British Standards BS EN 196-1:2005. (2005). Methods of testing cement-Part 1:
Determination of strength. Standards Policy and Strategy Committee.
Gu, J. (1991). “Hydration mechanism, properties and application of alkali-slag cement.”
Cement and Concrete Product (in Chinese), 5, 8–11.
Hbert, G., Billard, C., Rossi, P., and Chen, C. (2010). “Cement production technology
improvement compared to factor 4 objectives.” Cement and Concrete Research, 40(5),
820–826.
Juenger, M. C., Winnefeld, F., Provis, J. L., and Ideker, J. H. (2011). “Advances in
alternative cementitious binders.” Cement and Concrete Research, 41(12), 1232–1243.
Li, X. (2012). “Mechanical performance and durability of MgO cement concrete. ” PhD
Thesis. University of Cambridge.
Lu, L. W., Davidson, D. T., Handy, R. L., and Laguros, J. G. (1957). “The calcium-
magnesium ratio in soil-lime stabilization.” Soils, Geology and Foundations, 36, 794–
805.
Mo, L., Deng, M., and Tang, M. (2010). “Effects of calcination condition on expansion
property of MgO-type expansive agent used in cement-based materials.” Cement and
Concrete Research, 40(3), 437–446.
Ono, H., and Wada, S. I. (2007). “Properties of layer silicates formed from MgO-SiO2-H2O
mixtures at 25 ˚ C.” Journal of the Faculty of Agriculture Kyushu University, 52(1),
159–162.
Roy, A., Schilling, P., and Eaton, H. (1992). “Activation of ground blast furnace slag by
alkali metal and alkaline earth hydroxides.” Journal of American Ceramic Society,
75(12), 3233–3240.
Scrivener, K. L., and Kirkpatrick, R. J. (2008). “Innovation in use and research on
cementitious material.” Cement and Concrete Research, 38(2), 128–136.
Shand, M. A. (2006). The Chemistry and Technology of Magnesia. John Wiley & Sons, Ltd.,
Hoboken, New Jersey.
Shen, W., Wang, Y., Zhang, Tao, Zhou, M., Li, J., and Cui, X. (2011). “Magnesia
modification of alkali-activated slag fly ash cement.” Journal of Wuhan University of
Technology-Mater. Sci. Ed., 26(1), 121–125.
Shi, C., Krivenko, P. V., and Roy, D. M. (2006). Alkali-Activated Cements and Concretes.
Taylor & Francis.
Song, S., Sohn, D., and Jennings, H. (2000). “Hydration of alkali-activated ground
granulated blast furnace slag.” Journal of materials science, 35(1), 249 – 257.
Sun, W. H., Cui, C., Zhang, H. H., and Cui, K. H. (1991). “Relationship between crystalline
size and lattice distortion of MgO and its activity.” Jounal of WuHan University of
Technology, 13(4), 21–24.
Wei, J., Chen, Y., and Yongxin, L. (2006). “The reaction mechanism between MgO and
microsilica at room temperature.” Journal of Wuhan University of Technology, 21(2),
88–91.

9
Wei, J., Yu, Q., Zhang, W., and Zhang, H. (2011). “Reaction products of MgO and
microsilica cementitious materials at different temperatures.” Journal of Wuhan
University of Technology-Mater. Sci. Ed., 26(4), 745–748.
Yang, K.-H., Cho, A.-R., Song, J.-K., and Nam, S.-H. (2012). “Hydration products and
strength development of calcium hydroxide-based alkali-activated slag mortars.”
Construction and Building Materials, 29, 410–419.
Yang, K.-H., Sim, J.-I., and Nam, S.-H. (2010). “Enhancement of reactivity of calcium
hydroxide-activated slag mortars by the addition of barium hydroxide.” Construction
and Building Materials, 24(3), 241–251.
Yi, Y., Liska, M., and Al-Tabbaa, A. (2012). “Initial investigation into the use of GGBS-
MgO in soil stabilisation.” Proc 4th Int. Conf. Grouting and Deep Mixing, New
Orleans.
Zhang, T., Cheeseman, C. R., and Vandeperre, L. J. (2011). “Development of low pH
cement systems forming magnesium silicate hydrate (MSH).” Cement and Concrete
Research, 41(4), 439–442.

10