Construction and Building Materials 151 (2017) 138–146

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

The soundness of steel slag with different free CaO and MgO contents
Qiang Wang ⇑, Dengquan Wang, Shiyu Zhuang
Department of Civil Engineering, Tsinghua University, Beijing 100084, China

h i g h l i g h t s

 The primary component that leads to the bad soundness of steel slag is free CaO.
 Steel slag with a 4.96% content of free CaO presents bad soundness.
 Steel slag with a 7.68% content of MgO does not present bad soundness.

a r t i c l e i n f o a b s t r a c t

Article history: In this study, five types of steel slags with different free CaO (f-CaO) and MgO contents were used as min-
Received 9 May 2017 eral admixtures to prepare concretes with W/B ratios of 0.5 and 0.35. The concretes were cured for
Received in revised form 13 June 2017 4 years. Compressive strength and permeability to chloride ion were tested to reveal the influence of f-
Accepted 15 June 2017
CaO and MgO on concrete soundness over 4 years. An accelerated autoclave test was also adopted to
investigate the influence of MgO on concrete soundness after 4 years. The results show that the primary
component that leads to the bad soundness of steel slag is f-CaO. Ca(OH)2 primarily with a bulk structure
Keywords:
is produced by the hydration of f-CaO, which is different from the Ca(OH)2 produced by cement
Steel slag
Soundness
hydration. Steel slag with a 4.96% content of CaO presents bad soundness and leads to the soundness-
f-CaO induced failure of concretes at both high and low water-to-binder (W/B) ratios. When the content of
MgO f-CaO is within 2.09%, the steel slag presents satisfactory soundness. The results of the autoclave test
show that when the content of MgO is within 7.68%, steel slag has no negative influences on the compres-
sive strength of concrete.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Currently, in addition to applications in trace element removers

[5], acid-neutralizing agents [6], and soil conditioners [7,8], steel
Steel slag, a by-product of steel production, is one of the pre- slag is used primarily as a filler in the construction industry
dominant industrial residues in China. It is estimated that global [9,10]. However, the mineral components of steel slag contain a
steel slag output in 2016 was approximately 160–240 million tons significant amount of C3S, C2S, C4AF, and C2F [11–13], which can
and that China accounted for approximately half of this amount exhibit cementitious performance when mixed with water. There-
[1]. The massively deposited steel slag leads to not only the occu- fore, steel slag can act as raw material for cement clinker produc-
pation of farmland but also to the pollution of soil and groundwa- tion and a potential mineral admixture for concrete. Economically,
ter due to its leaching of certain potentially toxic elements. Bayless the use of steel slag as raw material for cement clinker production
et al. [2] reported high concentrations of Cr, Pb and Zn in ground- and mineral admixture for concrete should be given priority.
water immediately below steel slag deposits. Roadcap and Kelly [3] Tsakiridis et al. [14] studied the possibility of adding steel slag to
documented extreme pH values (>12) and high concentrations of raw materials for cement clinker production and revealed that
Fe, Ba, Cr and Mn in groundwater near steel slag deposits. Roadcap adding 10.5% steel slag has almost no adverse effects on the quality
et al. [4] reported significant concentrations of Cd, Cr, Cu, Pb, Ni, of the produced cement. Monshi and Asgarani [15] produced a type
and Zn in ground water below steel slag deposits. Thus, high- of cement equivalent to Type I ordinary Portland cement by mixing
efficiency utilization of steel slag is an important research subject steel slag, iron slag and calcined lime. Additionally, increasing
for resource conservation and environment sustainability. attention has been paid to the utilization of steel slag as a mineral
admixture. Zhang et al. [16] and Li et al. [17] studied the mechan-
ical properties of blended cement containing a suitable content of
⇑ Corresponding author.
steel slag and found that the binder has sufficient strength.
E-mail address: [email protected] (Q. Wang).

http://dx.doi.org/10.1016/j.conbuildmat.2017.06.077
0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.
 Q. Wang et al. / Construction and Building Materials 151 (2017) 138–146 139

Kourounis et al. [18] found that cement containing 15% or 30% steel 2. Experimental procedures
slag met the strength requirements of class 42.5 of EN 197-1, while
cement containing 45% steel slag met the strength requirements of 2.1. Materials
class 32.5 of EN 197-1. They also concluded that cements contain-
ing steel slag demand less water and can improve the mortar Portland cement with a grade of 42.5 and complying with the
workability. Wang et al. [19] investigated the drying shrinkage, Chinese National Standard GB175-2007 (equivalent to European
permeability to chloride, and carbonation resistance of concrete CEM I 42.5, ENV197-1:2000), crushed limestone 5–25 mm in size,
containing steel slag, and revealed that this concrete performs sim- natural river sand with a fineness of 2.8, polycarboxylic superplas-
ilarly to plain cement concrete. Moreover, researchers have also ticizer, and 5 different steel slags were used in this study. The
reported lower hydration heat [20,21] and longer setting time chemical composition of the cement was SiO2 21.86%, Al2O3
[22,23]. 4.25%, Fe2O3 2.66%, CaO 63.59%, and MgO 2.19%. The properties
Soundness of steel slag refers to its ability to retain its volume of the steel slags are shown in Table 1.
after it has get hardened. The soundness of steel slag is a critical
problem when it is used as a mineral admixture for concrete:
steel slag contains some free CaO and minerals containing MgO 2.2. Test methods
that might react at late ages, produce expansive internal stress,
and affect the volume stability [24,25]. The specific gravity of free The f-CaO contents of the steel slags were determined using the
CaO is 3.34, while that of Ca(OH)2 produced by the reaction of ethylene glycol method according to the Chinese National Standard
free CaO and water is 2.23 [26]. This difference is considered 176-2008 (equivalent to EN 196-2: 2005). The MgO contents of the
one of the reasons for the volume increase. MgO usually exists steel slags were determined by X-ray fluorescence.
in the form of wustite, i.e., Fe (Mn, Mg, Ca)O, in steel slag [26], Concrete samples sized 100  100  100 mm3 were prepared.
and the hydration of MgO occurs in several years [22]. The hydra- The concretes were cured under the conditions of 20 ± 1 °C and
tion of MgO into Mg(OH)2 may also lead to the expansion of 95% relative humidity. The compressive strength of concrete at
concrete. the ages of 28, 90, 360, 540, 730, 1090, and 1460 days was tested
In this study, five steel slags with different f-CaO and MgO according to the Chinese National Standard 50081-2002 (equiva-
contents were used as mineral admixtures in concrete and lent to EN 12390-3-2001). The chloride ion permeability of con-
paste. The samples were cured for 4 years under the conditions crete at the age of 730 and 1460 days was tested according to
of 20 ± 1 °C and 95% relative humidity. Compressive strength ASTM C1202. The micro-morphology of concrete at the age of
and permeability to chloride ions were tested to reveal the 360 days was observed using scanning electron microscopy (SEM).
degree of influence that f-CaO and MgO have on the soundness Five steel slag pastes were prepared by mixing water with steel
of the samples. The microstructures of concrete and hardened slag at a mass ratio of 0.3. The mixed pastes were placed in small
paste were observed to elucidate the soundness failure glass bottles and cured at 20 °C for 4 years. The bottles were mon-
mechanism. itored daily for the existence of cracks.

Table 1
Properties of the steel slags used.

Samples A B C D E
f-CaO content of the mineral compositions (%) 0.35 4.96 0.21 0.51 2.09
MgO content of the chemical compositions (%) 7.68 3.46 6.54 5.98 5.15
Specific surface areas (kg/m3) 453 461 472 442 516

Table 2
Mix proportions of the concretes with W/B of 0.50 and 0.35.

Samples Mix proportions (kg/m3) Note

Cement Steel slag Coarse aggregate Fine aggregate Water
C 400 0 777 1073 200 –
SA-40 240 160 777 1073 200 Steel slag A
SB-40 240 160 777 1073 200 Steel slag B
SC-40 240 160 777 1073 200 Steel slag C
SD-40 240 160 777 1073 200 Steel slag D
SE-40 240 160 777 1073 200 Steel slag E
SA-60 160 240 777 1073 200 Steel slag A
SB-60 160 240 777 1073 200 Steel slag B
SC-60 160 240 777 1073 200 Steel slag C
SD-60 160 240 777 1073 200 Steel slag D
SE-60 160 240 777 1073 200 Steel slag E
LC 400 0 802 1108 140 –
LSA-40 240 160 802 1108 140 Steel slag A
LSB-40 240 160 802 1108 140 Steel slag B
LSC-40 240 160 802 1108 140 Steel slag C
LSD-40 240 160 802 1108 140 Steel slag D
LSE-40 240 160 802 1108 140 Steel slag E
LSA-60 160 240 802 1108 140 Steel slag A
LSB-60 160 240 802 1108 140 Steel slag B
LSC-60 160 240 802 1108 140 Steel slag C
LSD-60 160 240 802 1108 140 Steel slag D
LSE-60 160 240 802 1108 140 Steel slag E
140 Q. Wang et al. / Construction and Building Materials 151 (2017) 138–146

The concrete was placed under autoclave conditions (216 °C tively, with a W/B ratio of 0.5. The strengths of concretes
and 2 MPa) for 3 h at the age of 1460 days. The compressive containing 40% steel slag A, C, D and E are 27.89%–39.44% lower
strength was tested before and after autoclaved curing. than that of plain cement concrete at the age of 28 days, while
Two cement replacements (40% and 60%) were adopted in this the strengths of concrete containing 60% steel slag are 51.99%–
study. Table 2 shows the mix proportions of the concretes with 59.56% lower. This result indicates that the loss rate of compressive
water-to-binder (W/B) ratios of 0.50 and 0.35. Superplasticizer is strength for the corresponding steel slag replacement is lower at
added to the concretes with W/B of 0.35. The dosage of superplas- 28 days. Thus, the soundness problem of steel slags A, C, D and E
ticizer is 1.2% (superplasticizer/cementitious materials in mass) for can be ignored at 28 days, primarily because of the low reaction
concretes containing 40% steel slag and 1.0% for concretes contain- degree of expansive components in steel slags (free CaO and
ing 60% steel slag. MgO) at 28 days. At the age of 1460 days, the strengths of con-
cretes containing 40% steel slag A, C, D and E are 26.32%–32.55%
lower than that of plain cement concrete, while the strengths of
3. Results and discussion concretes containing 60% steel slag are 36.90%–47.00% lower. This
result indicates that the loss rate of compressive strength of steel
3.1. Compressive strength of concrete slag concrete decreases with age because the increase of steel
slag’s reaction degree can contribute to the strength of concrete
Figs. 1 and 2 show the influence of steel slags on the compres-
sive strength of concrete containing 40% and 60% steel slag, respec-

Fig. 1. Compressive strengths of the plain cement concrete and the concretes
containing 40% steel slag with a W/B ratio of 0.50. Fig. 3. The cracks on the surface of the concrete containing 60% steel slag B with a
W/B ratio of 0.50.

Fig. 2. Compressive strengths of the plain cement concrete and the concretes Fig. 4. Compressive strengths of the plain cement concrete and the concretes
containing 60% steel slag with a W/B ratio of 0.50. containing 40% steel slag with a W/B ratio of 0.35.
 Q. Wang et al. / Construction and Building Materials 151 (2017) 138–146 141

at late ages. Moreover, the results reveal that steel slags A, C, D and Figs. 4 and 5 show the influence of steel slags on the compres-
E provide satisfactory soundness and cause no loss to the strength sive strength of concrete containing 40% and 60% steel slag, respec-
of concrete within 1460 days. tively, with a W/B ratio of 0.35 The strength results of concretes
For steel slag B, the concrete with 40% content exhibits stable containing steel slag A, C, D and E are similar to those with a W/
strength development before 1090 days, but it is cracked between B ratio of 0.5: the concretes with 40% and 60% content can exhibit
1090 days and 1460 days, and its compressive strength falls to stable strength development and remain intact. For steel slag B,
0 MPa at 1460 days. When the content of steel slag B is 60%, the however, the concrete with 40% steel slag content exhibits stable
concrete is cracked between 90 days and 360 days, and its com- strength development before 730 days, but its strength starts to
pressive strength falls to 0 MPa at 360 days. This result indicates decrease at 730 days, and the concrete is cracked between
that steel slag B presents bad soundness, and that the increase of 1090 days and 1460 days. When the content of steel slag B is
slag content causes earlier and more severe damage to the 60%, the strength of the concrete develops stably before 730 days,
microstructure of concrete. Fig. 3 shows the surface of concrete but it starts to decrease at 730 days, and the concrete is cracked
containing 60% steel slag B with the W/B of 0.50 at the age of between 730 days and 1090 days. This result indicates that the
360 days. Many cracks are visible on the surface of the concrete concrete containing steel slag B still cracks due to volume expan-
specimen, and the concrete has lost its rigidity. sion even though the concrete strength is increased at the lower
W/B ratio. This phenomenon occurs because the microstructure
of the concrete is much denser at a lower W/B ratio, which can sig-
nificantly increase the internal expansion stress arising from the
hydration of expansive components in steel slag B. Consequently,
the expansion stress is greater than the resistance stress of con-
crete, and the concrete cracks.
In conclusion, steel slag B can result in microstructural failure
and strength loss of the concrete, although the ages of
soundness-induced failure vary because the concretes have differ-
ent water-to-binder ratios and steel slag contents. By contrast, the
concretes containing steel slag A, C, D and E remain intact at
1460 days.

3.2. Chloride permeability of concrete

Table 3 shows the charge passed and permeability to chloride

ion of the concretes containing 40% and 60% steel slag with W/B
ratios of 0.5 and 0.35 at the ages of 730 days and 1460 days. At a
W/B ratio of 0.5, plain cement concrete exhibits low permeability
at 730 days and 1460 days. For steel slags A, C, D and E, the con-
cretes with 40% content exhibit moderate permeability, while the
concretes with 60% content exhibit high permeability at 730 days.
Fig. 5. Compressive strengths of the plain cement concrete and the concretes This result can be attributed to the low reaction degree of steel slag
containing 60% steel slag with a W/B ratio of 0.35. and the resulting high porosity in the concrete containing steel

Table 3
Permeability to chloride ion of concrete containing steel slags with the W/B of 0.50 and 0.35 at the ages of 730 days and 1460 days.

Samples Ages: days

730 1460
Charge passed: Coulombs Permeability grade Charge passed: Coulombs Permeability grade
C 1281 Low 1068 Low
SA-40 2542 Moderate 2145 Moderate
SB-40 2087 Moderate Damaged (untested) Damaged (untested)
SC-40 2365 Moderate 2133 Moderate
SD-40 2394 Moderate 2014 Moderate
SE-40 2221 Moderate 2154 Moderate
SA-60 4431 High 3689 Moderate
SB-60 Damaged (untested) Damaged (untested) Damaged (untested) Damaged (untested)
SC-60 4161 High 3548 Moderate
SD-60 4287 High 3154 Moderate
SE-60 4529 High 4026 High
LC 546 Very low 618 Very low
LSA-40 795 Very low 632 Very low
LSB-40 841 Very low Damaged (untested) Damaged (untested)
LSC-40 762 Very low 751 Very low
LSD-40 687 Very low 648 Very low
LSE-40 754 Very low 602 Very low
LSA-60 1124 Low 1082 Low
LSB-60 1265 Low Damaged (untested) Damaged (untested)
LSC-60 1354 Low 1125 Low
LSD-60 1211 Low 1095 Low
LSE-60 1314 Low 1214 Low
142 Q. Wang et al. / Construction and Building Materials 151 (2017) 138–146

slag, which is also reflected in the compressive strength results. At the W/B of 0.35, plain cement concrete and concretes con-
Note that the permeability of the concretes containing steel slag taining 40% steel slag A, C, D and E exhibit very low permeability,
A, C, D and E changes when the age changes from 730 days to and concretes containing 60% steel slag A, C, D and E exhibit low
1460 days. Concretes with 40% content exhibit the same perme- permeability at the age of 730 days and 1460 days. This character-
ability (moderate) at the two ages, but the charge passed within istic indicates that for concretes containing steel slag A, C, D and E,
the concretes decreases. The charge passed of concretes with 60% soundness-induced failure does not occur at low W/B ratios, which
content all decrease significantly from 730 days to 1460 days. Con- is consistent with the compressive strength results. The concrete
cretes containing steel slag A, C and D exhibit lower permeability containing steel slag exhibits better resistance to chloride ion per-
(moderate) at 1460 days than do concretes at 730 days (high per- meability. This improvement probably occurs for two reasons: the
meability), but the concrete containing steel slag E exhibits the ‘‘boundary effect” of reduced hydration products occurs because
same permeability (high) at the two ages. Therefore, for concretes the steel slag replacement becomes weaker, and the hydration
containing steel slag A, C, D and E, increased permeability induced products of steel slag, particularly at late ages, can block already-
by the soundness is not found at a W/B ratio of 0.5. Permeability formed pores because of the relatively slow reaction rate of steel
tests were not conducted on concretes containing steel slag B at slag. For steel slag B, concretes with 40% and 60% content exhibit
certain ages because they had cracked. The test was conducted very low and low permeability at the age of 730 days, respectively,
only on the concrete with 40% content at the age of 730 days. This identical to the case for concretes containing the other four steel
concrete exhibited moderate permeability, similar to the concretes slags. This result indicates that steel slag B has not yet caused dam-
containing other types of steel slag. This similarity indicates that age to the microstructure of the concrete at 730 days. At the age of
steel slag B had not yet caused damage to the microstructure of 1460 days, however, concretes containing steel slag B with a W/B
the concrete at 730 days, which is consistent with the compressive ratio of 0.35 have been cracked.
strength results. From the compressive strength and chloride ion permeability
tests, it can be concluded that steel slags A, C, D and E provide sat-
isfactory soundness and cause no damage to the macroscopic prop-
erties of concrete. Steel slag B, however, results in bad soundness
and the soundness-induced failure of the concrete, which is stron-
ger for concrete with a high steel slag replacement. Additionally,
the soundness-induced failure of concrete containing steel slag B
occurs even at a low W/B ratio.

3.3. Morphologies of the microstructure of concrete

The morphologies of the microstructure of the concrete con-

taining 60% steel slag B at the age of 360 days were observed. There
are certain large crystalline components with an irregular or bulk
structure in the hydration products of the concrete SB60
(Figs. 6 and 8). The Energy Dispersive X-ray Spectroscopy (EDX)
results (Figs. 7 and 9) show that the primary elements of the large
crystalline components are Ca and O, and the fitting results reveal
that the Ca/O atomic ratio is approximately 0.5, indicating that the
huge crystalline components are Ca(OH)2. The crystalline Ca(OH)2
produced by the hydration of C2S and C3S normally has a regular
layered or flake structure [27,28]. The large crystalline Ca(OH)2 in
Fig. 6. Morphology of the crystalline components in the hydration products of Figs. 6 and 8, whose morphology is different from that of Ca
concrete SB60 (irregular structure).
(OH)2 produced by the hydration of C2S and C3S, is concluded to

Fig. 7. EDX results of the crystalline components in Fig. 6.

 Q. Wang et al. / Construction and Building Materials 151 (2017) 138–146 143

Fig. 8. Morphology of the crystalline components in the hydration products of Fig. 10. Ettringite in the hydration products of concrete SB60.
concrete SB60 (bulk structure).
in small glass bottles and cured at 20 °C for 4 years. The glass bot-
tle containing steel slag paste B had cracked at the age of 60 days,
be produced by the reaction of free CaO, which is a prominent while the other four glass bottles were intact at the age of
cause of the increased expansive internal stress. 4 years. Fig. 11 shows the SEM image of a visible crack in the
Fig. 10 indicates that many ettringites exist in the hydration steel slag hardened paste B that was placed in the cracked glass
products of concrete SB60. The expansive pressure caused by the bottle. Magnified SEM micrographs were taken along the crack,
formation of ettringite can cause expansion and cracking of con- and the bulk crystalline product was observed in the crack. The
crete. In general, many ettringites are not commonly found in typical EDX results (Fig. 12) show that the primary elements of
dense structures due to restricted space. However, as cracks form the huge crystalline product are Ca and O, and the fitting results
due to the bad soundness of steel slag in the concrete, more space show that the Ca/O atomic ratio is approximately 0.5. These
becomes available for the ettringite to grow. This increase further results indicate that the large crystalline product is Ca(OH)2. Thus,
promotes crack propagation. It indicates that the crack propagation free CaO reacts with water to form bulk Ca(OH)2, which produces
and the growth of ettringite can be seen as an interactive process. expansive internal stress and causes cracking in the hardened
Therefore, the influence of excessive growth of ettringite on the paste or concrete.
damage of the concrete should not be neglected.

3.4. Morphologies of steel slag paste 3.5. Autoclave test of concrete

Five steel slag pastes were prepared by mixing water with The results above show that steel slag B presents bad sound-
steel slag at a mass ratio of 0.3. The mixed pastes were placed ness, while steel slags A, C, D and E present qualified soundness.

Fig. 9. EDX results of the crystalline components in Fig. 8.

144 Q. Wang et al. / Construction and Building Materials 151 (2017) 138–146

Fig. 11. Crack in hardened steel slag B paste and bulk crystalline product along the crack (a–g).

Table 1 shows that the free CaO (4.96%) content of steel slag B is After being over-burned, the activity of MgO is relatively low,
the highest among the five steel slags; the MgO (3.46%) content and the hydration reactions of MgO that cause expansion proceed
is the lowest. Analyses of the morphologies of the concrete and at a relatively slow rate [22,29]. As a result, no damage to concrete
hardened paste have confirmed that the bad soundness of steel caused by MgO can be observed at the age of 1460 days. Steel slag
slag B originates from the expansive component of free CaO. A, in which the MgO content is up to 7.68%, and steel slag E, whose
 Q. Wang et al. / Construction and Building Materials 151 (2017) 138–146 145

Fig. 12. Typical EDX results of the crystalline products in Fig. 11.

Fig. 13. Compressive strengths of concrete with a W/B ratio of 0.50 before and after Fig. 14. Compressive strengths of concrete with a W/B ratio of 0.35 before and after
autoclave curing. autoclave curing.

MgO and CaO contents are 5.15% and 2.09%, respectively, both pre- curing. This decrease occurs because the microstructure becomes
sent satisfactory soundness. To further investigate the soundness weaker during autoclave curing due to expansive stresses on the
of steel slags A, C, D and E at a longer age, an autoclave test was pores when low-reactivity components (e.g., inert MgO) partly
adopted to accelerate the reaction of MgO. This method is currently hydrate to form expansive products [30–32]. Moreover, the high
the standard test method for the evaluation of potential expansion temperature of autoclave curing (216 °C) can change the packing
failure due to the delayed hydration of MgO. Changes in concrete of C-S-H at the nanoscale with decreased bound water and
strength before and after autoclave curing were observed to evalu- increased apparent density and thus make the structure much
ate damage caused by the hydration of MgO. coarser and more porous [33–35].
Fig. 13 shows the strength of plain cement concrete and con- The focus here is on the relationship between the strength loss
cretes containing steel slag A, C, D and E with a W/B ratio of 0.5 rates of concretes containing steel slag after autoclave curing and
before and after autoclaved curing. Fig. 14 shows the results at a the loss rate of plain cement concrete. At a W/B ratio of 0.5, the
W/B ratio of 0.35. Note that the autoclave test was not conducted compressive strengths of concretes containing 40% and 60% steel
on concrete containing steel slag B because it had cracked. The slag after autoclave curing are 14.6%–20.6% and 12.7%–20.2%
results show that the compressive strength of all the concretes, lower, respectively, than those of concretes before autoclave cur-
including the plain cement concrete, decreases after autoclaved ing. This result is close to or even smaller than the strength loss
146 Q. Wang et al. / Construction and Building Materials 151 (2017) 138–146

rate of plain cement concrete (20.6%). Similar to the results at a W/ [5] C. Oh, S. Rhee, M. Oh, J. Park, Removal characteristics of As(III) and As(V) from
acidic aqueous solution by steel making slag, J. Hazard. Mater. 213–214 (2012)
B ratio of 0.5, the strength loss rates of concretes containing 40%
147–155.
and 60% steel slag after autoclave curing are close to or even smal- [6] C.S. Gahan, M.L. Cunha, Å. Sandström, Comparative study on different steel
ler than that of plain cement concrete at a W/B ratio of 0.35. This slags as neutralising agent in bioleaching, Hydrometallurgy. 95 (2009) 190–
similarity indicates that relative to plain cement concrete, no seri- 197.
[7] L.A. Wendling, G.B. Douglas, S. Coleman, Productive use of steelmaking by-
ous strength loss occurs for concretes containing steel slag under product in environmental applications – II: Leachate geochemistry, ecotoxicity
autoclave curing. In conclusion, at a W/B ratio of 0.5 and 0.35, steel and environmental radioactivity, Miner. Eng. 39 (2012) 219–227.
slags with a 5.15%–7.68% MgO content will not lead to soundness- [8] L.A. Wendling, M.T. Binet, Z. Yuan, F. Gissi, D.J. Koppel, Geochemical and
ecotoxicological assessment of iron- and steel-making slags for potential use in
induced failure of concrete in autoclaved curing. environmental applications, Environ. Toxicol. Pharmacol. 32 (2013) 2602–
MgO in steel slag primarily exists in the RO phase (CaO-FeO- 2610.
MnO-MgO solid solution) due to the high basicity condition in [9] W.G. Shen, M.K. Zhou, W. Ma, J.Q. Hu, Z. Cai, Investigation on the application of
steel slag–fly ash–phosphogypsum solidified material as road base material, J.
molten steel slag and the similar radii of Mg++, Fe++, and Mb++ Hazard. Mater. 164 (2009) 99–104.
[26]. The reactivity of the RO phase is relatively low [11,13]. It [10] P. Ahmedzade, B. Sengoz, Evaluation of steel slag coarse aggregate in hot mix
has been reported that the MgO-bearing RO phase represents poor asphalt concrete, J. Hazard. Mater. 165 (2009) 300–305.
[11] Q. Wang, P.Y. Yan, Hydration properties of basic oxygen furnace steel slag,
soundness only when the MgO content is over 70% in the RO phase. Constr. Build. Mater. 24 (2010) 1134–1140.
The influence of MgO on the soundness of steel slag is relatively [12] J.X. Li, Q.J. Yu, J.X. Wei, T.S. Zhang, Structural characteristics and hydration
complex due to its different contents and existent forms in steel kinetics of modified steel slag, Cem. Concr. Res. 41 (2011) 324–329.
[13] Q. Wang, P.Y. Yan, J.W. Feng, A discussion on improving hydration activity of
slag, and further studies are needed to draw scientific conclusions.
steel slag by altering its mineral compositions, J. Hazard. Mater. 186 (2011)
The results of this study show that when the MgO content is within 1070–1075.
7.68%, steel slag presents satisfactory soundness, and soundness- [14] P.E. Tsakiridis, G.D. Papadimitriou, S. Tsivilis, C. Koroneos, Utilization of steel
induced failure of concrete will not occur within 1460 days, even slag for Portland cement clinker production, J. Hazard. Mater. 152 (2008) 805–
811.
under autoclaved curing. [15] A. Monshi, M.K. Asgarani, Producing Portland cement from iron and steel slags
and limestone, Cem. Concr. Res. 29 (1999) 1373–1377.
[16] T.S. Zhang, Q.J. Yu, J.X. Wei, J.X. Li, P.P. Zhang, Preparation of high performance
4. Conclusions blended cements and reclamation of iron concentrate from basic oxygen
furnace steel slag, Resour. Conserv. Recycl. 56 (2011) 48–55.
[17] D.X. Li, X.H. Fu, X.Q. Wu, M.S. Tang, Durability study of steel slag cement, Cem.
(1) Over a time span of 4 years, f-CaO is the primary component Concr. Res. 27 (1997) 983–987.
that leads to the bad soundness of steel slag. The morphol- [18] S. Kourounis, S. Tsivilis, P.E. Tsakiridis, G.D. Papadimitriou, Z. Tsibouki,
Properties and hydration of blended cements with steelmaking slag, Cem.
ogy of Ca(OH)2 produced by f-CaO, which is primarily a bulk
Concr. Res. 37 (2007) 815–822.
structure, is substantially different from that of Ca(OH)2 pro- [19] Q. Wang, P.Y. Yan, J.W. Yang, B. Zhang, Influence of steel slag on mechanical
duced by C3S and C2S. properties and durability of concrete, Constr. Build. Mater. 47 (2013) 1414–
1420.
(2) When the content of f-CaO is 4.96%, the steel slag presents
[20] T.S. Zhang, Q.J. Yu, J.X. Wei, J.X. Li, Investigation on mechanical properties,
bad soundness, which leads to the soundness-induced fail- durability and micro-structural development of steel slag blended cements, J.
ure of concrete, although the failure is delayed at low W/B Therm. Anal. Calorim. 110 (2012) 633–639.
ratios. [21] S.H. Liu, L.H. Li, Influence of fineness on the cementitious properties of steel
slag, J. Therm. Anal. Calorim. 117 (2014) 629–634.
(3) When the content of f-CaO is within 2.09%, steel slag pre- [22] N. Ak, S.Y. Altun, Lmaz, Study on steel furnace slags with high MgO as additive
sents satisfactory soundness over 4 years. in Portland cement, Cem. Concr. Res. 32 (2002) 1247–1249.
(4) When the MgO content (partly in the form of RO phase or [23] A. Rai, J. Prabakar, C.B. Raju, R.K. Morchalle, Metallurgical slag as a component
in blended cement, Constr. Build. Mater. 16 (2002) 489–494.
unreactive minerals) is within 7.68%, steel slag causes no [24] G.R. Qian, D.D. Sun, J.H. Tay, Z.Y. Lai, Hydrothermal reaction and autoclave
damage to the concrete after autoclave curing based on the stability of Mg bearing RO phase in steel slag, Br. Ceram. Trans. 101 (2002)
autoclave test at the age of 4 years. 159–164.
[25] Q.Z. Xiao, Expansion and its inhibition of steel slag, J. Chin. Ceram. Soc. 24
(1996) 635–640.
[26] G. Wang, Y.H. Wang, Z.L. Gao, Use of steel slag as a granular material: Volume
expansion prediction and usability criteria, J. Hazard. Mater. 184 (2010) 555–
Acknowledgements 560.
[27] Q. Wang, M.X. Shi, Z.Q. Zhang, Hydration properties of steel slag under
autoclaved condition, J. Therm. Anal. Calorim. 120 (2015) 1241–1248.
This work was supported by the Tsinghua University Initiative
[28] S. Mindess, J. Young, Concrete, New Jersey: Englewood Cliffs, 1981.
Scientific Research Program [grant number 20161080079]; the [29] P.W. Gao, S.X. Wu, X.L. Lu, D. Min, L.P. Hua, M.T. Zhong-ru, Wu A, Soundness
National Natural Science Foundation of China [grant number evaluation of concrete with MgO, Constr. Build. Mater. 21 (2007) 132–138.
51108245]. [30] M.A.A. Sherir, K.M.A. Hossain, M. Lachemi, Self-healing and expansion
characteristics of cementitious composites with high volume fly ash and
MgO-type expansive agent, Constr. Build. Mater. 127 (2016) 80–92.
[31] W.C. Wang, Feasibility of stabilizing expanding property of furnace slag by
References: autoclave method, Constr. Build. Mater. 68 (2014) 552–557.
[32] M.A. Rahman, P.K. Sarker, F.U.A. Shaikh, A.K. Saha, Soundness and compressive
[1] U.S. Geological Survey Mineral Commodity Summaries 2017: Iron and Steel strength of Portland cement blended with ground granulated ferronickel slag,
Slag, 2017. Constr. Build. Mater. 140 (2017) 194–202.
[2] E.R. Bayless, T.K. Greeman, C.C. Harvey, Hydrology and geochemistry of a slag- [33] E. Gallucci, X. Zhang, K.L. Scrivener, Effect of temperature on the
affected aquifer and chemical characteristics of slag-affected ground water. in: microstructure of calcium silicate hydrate (C-S-H), Cem. Concr. Res. 185–195
US Geological Survey Water-Resources Investigations Report, Northwestern (2013).
Indiana and Northeastern Illinois, 1998, pp. 4197–4198. [34] K.O. Kjellsen, R.J. Detwiler, O.E. Gjorv, Pore structure of cement pastes
[3] G.S. Roadcap, W.R. Kelly, Shallow ground-water chemistry in the Lake Calumet hydrated at different temperatures, Cem. Concr. Res. 20 (1990) 927–933.
Area, Chicago, Illinois, National Symposium on Water Quality, American Water [35] B. Lothenbach, T. Matschei, G. Möschner, F.P. Glasser, Thermodynamic
Resources Association, Herndon, Virginia, 1994. modelling of the effect of temperature on the hydration and porosity of
[4] G.S. Roadcap, W.R. Kelly, C.M. Bethke, Geochemistry of extremely alkaline Portland cement, Cem. Concr. Res. 38 (2008) 1–18.
(pH > 12) ground water in slag-fill aquifers, Ground Water 43 (2005) 806–816.