Effect of high-alumina ladle furnace slag as cement substitution in masonry mortars

Construction and Building Materials 123 (2016) 404–413

Contents lists available at ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

Effect of high-alumina ladle furnace slag as cement substitution in masonry mortars Tamara Herrero a, Iñigo J. Vegas a, Amaia Santamaría b, José T. San-José b,⇑, Marta Skaf c a

TECNALIA. C/Geldo, Parque Tecnológico de Bizkaia, Edificio 700, 48160 Derio, Spain UPV/EHU, Dept. of Engineering Materials, Alameda Urquijo, s/n, 48013 Bilbao, Spain c University of Burgos, EPS, Dept. of Construction, C/Villadiego s/n, 09001 Burgos, Spain b

h i g h l i g h t s
Ladle furnace slag (LFS) is a non-uniform by-product, and can be high-alumina or high-silica slag.
High-alumina LFS has a better hydraulic reaction in mortars than high-silica LFS.
Partial substitution of cement by LFS in amounts of 20% by weight has no negative effects.
LFS-mortar mixes show useful properties for building sector applications.

a r t i c l e

i n f o

Article history: Received 10 March 2016 Received in revised form 26 May 2016 Accepted 7 July 2016

Keywords: Ladle furnace slag Cement additions Paste Masonry mortar Durability

a b s t r a c t Ladle furnace (white or basic) slag is a significant by-product of the steelmaking industry; nowadays the manufacturing process yields two types of basic slag that are either low or high in alumina. The present research focuses mainly on the composition of the high-alumina slag and the reactivity of its compounds such as calcium aluminates, free calcium oxide, and free magnesium oxide, when aged at room temperature and at water steam temperature (accelerated aging). Additionally, a characterization was performed of pastes and masonry mortars that incorporate high alumina ladle furnace slag as a supplementary cementing material in partial substitution of Portland cement in amounts of 10% and 20% by weight. Different properties are studied such as porosity distribution, volumetric stability, mechanical strength and durability, mainly referring to wetting-drying aging cycles. The study concludes that high-alumina ladle furnace slag can induce slight hydraulic reactivity and its partial addition has no negative effect on the fundamental properties of cement masonry mortars. Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction Nowadays, Spain produces steel in 19 steelmaking factories, approximately 45% of which are situated in the north of the country, where the authors conducted this research. This industrial sector produces at least eight different types of slags [1], which have been studied by many research groups ever since the pioneering papers of Motz, Geiseler and Koros [2–4], to identify reliable applications for each slag type. In this scenario, the construction sector has taken action to respond to this challenge and reuse industrial by-products, as recently done in other sustainable approaches [5–10]; thereby exploiting materials, which would otherwise be of no economic

⇑ Corresponding author. E-mail address: [email protected] (J.T. San-José). http://dx.doi.org/10.1016/j.conbuildmat.2016.07.014 0950-0618/Ó 2016 Elsevier Ltd. All rights reserved.

value, giving them an environmental friendly value with a functional utility [11–18]. One of the slag types is reducing, basic or white Ladle Furnace Slag (LFS), produced in copious amounts as a by-product of the steelmaking industry in the secondary or basic steel-refining process; this steel-refining process yields basic slag which is either low or high in silica or alumina, depending on the saturation method of melting fluxes. Nowadays, the alumina saturation method is applied in approximately 25% of the steelmaking industry. However, its future is promising, in view of the cheaper and cleaner steel that this novel steel-refining process can produce. The present work is focused mainly on the study of the properties of this alumina-rich slag. As has been stated in the literature, in general the main components of LFS are magnesium and calcium oxides accompanying silica and aluminum oxides that act as fluxes. Hence, the compounds of LFS are mainly silicates and aluminates of calcium and

405

T. Herrero et al. / Construction and Building Materials 123 (2016) 404–413

magnesium, as already described and characterized in the scientific literature [19–28]. Obviously, calcium aluminates are preponderant in the high-alumina slag, while in the high-silica slag calcium silicates are more abundant. Free calcium oxide and free magnesium oxide are also always present in these two types of LFS slags. This kind of industrial by-product can be used in various construction and civil engineering applications in which a binder is needed, for various reasons. First, due to the similarity between the oxides that are present (also in similar proportions) in both LFS and Portland clinker and the eventual hydraulic properties that they can have. Secondly, the dusty morphology of the LFS in its original state can facilitate its use as a substitute for dusty binders. Over the past decade, it has already reported in the literature initial studies on the use of LFS in cement matrices [1,18,23,24,27,31] followed by studies on soil stabilization for civil works [19,29–31]. In general, the use of LFS basic slag that is high in silica, as a binder or an aggregate or even as raw material of clinker [32], has been relatively successful; on the contrary, LFS high in alumina has only been studied by itself [22,28] until today. The present work presents a pioneering study on its behavior in hydraulic mixes. For a better comprehension of both LFS types, mainly in terms of hydraulicity, the present research will examine their volume stability [33] and the reactivity of their main components at outdoor temperatures: calcium aluminates, free-CaO and free-MgO. Various analytical techniques are applied to record their chemical and volumetric changes: X-ray diffraction (XRD), thermogravimetric and differential calorimetric scanning analysis (TG-DSC), swelling tests, and SEM (scanning electron microscopy) with EDAX (dispersive energy of X-ray) micro-analysis. This research also studies cement pastes and masonry mortars containing both white slag types in partial substitution of binder (cement). It focuses on analyzing their performance from two Table 1 Chemical composition and other physical properties. Compounds

CEM I 52.5 R

CEM II/A-M 42.5 R

LFS1

LFS2

SiO2 (%) Al2O3 (%) Fe2O3 (total) (%) MnO (%) MgO (%) CaO (%) Na2O (%) K2O (%) TiO2 (%) P2O5 (%) SO3 (%) Loss on ignition (%) Blaine fineness (cm2/g)

18.56 5.05 3.29 – 1.49 63.40 0.21 0.71 – – 1.8 2.72 (C+ CO2 + H2O) 4975

20.83 6.40 3.51 – 1.62 58.29 0.20 0.90 – – 1.2 4.53 (C + CO2 + H2O) 4060

3.77 28.82 3.82 0.29 6.36 53.96 – – 0.14 0.04 2.80 26.2

22.94 4.42 1.01 0.22 5.99 61.64 – – 0.34 0.01 3.43 3% gain 1650

Specific gravity (Mg/m3)

3.13

3.05

2820 (after crushing) 2.75

3.03

perspectives; firstly, their mechanical strength is tested; secondly, the dimensional stability of mixtures and durability issues are examined by exposure to wetting-drying cycles in the laboratory. It could be used in the preparation of layering mortars (façades), rendering and plastering (partitioning), and bonding (masonry) mortars. 2. Materials 2.1. Water, cement, and natural aggregates Clean water from the urban mains supply of the city of Bilbao was used that contained no elements that might negatively affect the quality of the hydraulic mixes. Two Portland cement types were used in present research. Cement type I (CEM I 52.5 R) consisting of 90% Portland clinker, 5% calcium carbonate powder fines, and 5% gypsum, with a unimodal particle-size distribution of around 20 lm (checked by a LS 13320 laser diffractometer). Cement type II (CEM II/A-M (V-L) 42.5 R) consisted of 80% Portland clinker, 15% high-silica fly ash, 2% calcium carbonate, and 3% gypsum, with a bimodal particle-size distribution of around 19 lm and, more frequently, 34 lm. Their chemical composition and other physical properties of both cement types, analyzed by X-ray fluorescence (XRF), are shown in Table 1. A commercial crushed natural limestone aggregate sized between 0 and 4 mm, with a fine fraction (<0.063 lm) of 20%, was also used. Its main mineral compound was calcite (95%), with a specific gravity of 2.68 Mg/m3 and a fineness modulus of 2.4 units. Fig. 1 includes the fine limestone sand morphology, supplied by Morteros Bikain Company. 2.2. Slags High-alumina and high-silica ladle furnace slags, by-products from two different steelmaking industries, were labeled LFS1 and LFS2, respectively. LFS1 is a high-alumina slag weathered on factory grounds over several weeks; its aging level was ‘a priori’ uncertain at the moment of its reception, but its morphology was no longer dusty (original situation), but agglomerated in particles of some centimeters in size (see Fig. 1). LFS2 was a dusty high silica slag taken from the factory immediately after cooling, with no aging at all. Fig. 1 shows the morphology of both LFS types: LFS1 presented an irregular-shaped material, with powder and pieces of low cohesion formed of small aggregated particles, while LFS2 presented a powderydusty appearance with disaggregated particles. Their chemical composition obtained by X-ray fluorescence analysis, on slags in their ‘‘as-received” state, is shown in Table 1, in which the ‘‘loss of ignition” gives an approximate idea of the degree of aging of each slag type. Additionally, Table 2 details their mineralogical analysis by XRD, which indicates the presence-absence of hydration-carbonation

Table 2 XRD analysis of Ladle Furnace Slags. Mineral constituents

LFS1

Tricalcium aluminate (Ca3Al2O6) Calcium-olivine (Ca2SiO4) Brucite (Mg(OH)2) Calcite (CaCO3) Hydrotalcite (Mg6 Al2 (CO3)(OH)164H2O) Mayenite (Ca12Al14O33) Katoite (Ca3Al2(OH)12) Sjögrenite (Mg6 Fe2 (CO3)(OH)164H2O) Periclase (MgO) Jasmundite (Ca22 (SiO4)8O4S2) Fluorite (CaF) Oldhamite (CaS)
























;

Fig. 1. Morphology of limestone and LFS1/LFS2 slag. Black and white stripes are centimeters.

LFS2


















406

T. Herrero et al. / Construction and Building Materials 123 (2016) 404–413

products: calcite, hydrotalcite, sjögrenite, brucite and katoite in LFS1, and nonhydrated or carbonated products in the form of olivine, mayenite, periclase, jasmundite and fluorite in LFS2. The well-known presence of free lime (CaO) in these slags cannot be suitably verified by this XRD technique. Additionally, both basic slags in their ‘‘as received” state were studied under TG–DSC analysis (Fig. 2), seeing how their respective natural moisture disappeared at temperatures of below 140 °C. In LFS1, the first dehydration (4H2O) of hydrotalcites Mg6 (Al,Fe)2CO3(OH)164H2O took place below 300 °C. Katoite Ca3Al2(OH)12 dehydroxilation took place at 370 °C. Brucite Mg(OH)2 decomposition into periclase MgO occurred at 420 °C, and the decomposition of portlandite Ca(OH)2 at 460 °C. Hydrotalcite and Sjögrenite second decomposition (dehydroxilation and decarbonization) took place at 540 °C and finally, calcite CaCO3 decarbonization occurred at 760 °C. No additional reactions were observed, and the total weight loss stood at around 26%. In LFS2, very small amounts (not evidenced by XRD due their scarcity) of brucite Mg(OH)2 decomposition into periclase MgO occurred at 420 °C and decomposition of a very low magnesite MgCO3 content at 680 °C. The oxidation of sulfurs (oldhamite CaS) to sulfates, occurred at around 900 °C showing mass gain and exothermy. Finally, decomposition (not observed in XRD) of very low calcite CaCO3 contents could be mentioned (shoulder-peak at 850 °C).

Approximate stoichiometric calculations of free lime and free periclase contents are shown below for the LFS1 slag in the ‘‘as-received” state. TG–DSC analysis and approximate measurements (see Fig. 2) estimated a free-lime content of 15.5% and a free-periclase content of almost 6%, as presented in the following analysis:

2:19%Mg6 ðFe; AlÞ2 ðCO3 ÞðOHÞ16 4ðH2 OÞ  ð6  44 g=mol  MgOÞ= ð8  18 g=mol  H2 O þ 44 g=mol  CO2 Þ þ 1:25%MgðOHÞ2  ð40 g=mol  MgOÞ=ð18 g=mol  H2 OÞ ¼ 5:9%MgO 10:65%CaCO3  ð56 g=mol  CaOÞ=ð44 g=mol  CO2 Þ þ 0:63%CaðOHÞ2  ð56 g=mol  CaOÞ=ð18g=mol  H2 OÞ ¼ 15:5%CaO These are only approximate results and must be interpreted as the ‘‘hydrated and carbonated fraction” of free lime and free periclase. The real values are higher than those due to the presence of free periclase (verified by XRD, Table 2) and free lime (verified at a later stage, in Section 3, by the increase in calcite content after aging) in the LFS1 in the ‘‘as-received” state. As previously stated in [19], TG, DSC-DTA (or TG-DSC) and XRD analysis are the most useful techniques to determine the amount of both constituents.

Fig. 2. TG–DSC test result (a) LFS1 (b) LFS2.

407

T. Herrero et al. / Construction and Building Materials 123 (2016) 404–413 Table 3 X-ray diffraction results of LFS1 after steam test.

3. LFS hydration studies When the LFS (containing free-CaO, free-MgO and other compounds) is weathered, several complex chemical reactions occur, commonly giving rise to an extensive swelling. In the present section, the slags were analyzed under accelerated aging in two ways. Firstly, LFS1 was exposed to a hot and humid jet of steam. Secondly, we studied the hydration of both LFS-types in a moist chamber at moderate temperatures. The LFS2 slag type has been analyzed in depth by the authors in previous works [19]. 3.1. Hydration of ladle furnace basic slag under steam conditions In this study the expansiveness of LFS1 and LFS2 was analyzed according to EN 1744-1:2009 [34], based on steam exposure of slag in SBG 4500 N/C equipment from BAG, at a temperature of between 120° and 130 °C, as shown in Fig. 3a. In an attempt to stabilize the LFS volumetric instabilities, the present experiment was extended for a lengthier period than specified in the standard (24–168 h). At 80 days, with expansion continuing in the highalumina slag, the experiment of LFS1was stopped to record a final volumetric expansiveness value of 27%. Swelling of the high-silica LFS2, ended after 19 days, showed an expansiveness of around 8%. The expansion-swelling of slags is represented in Fig. 3b in a semi-logarithmic graphic. In the case of LFS1, expansion continued over time in two distinct stages (different slopes on the logarithmic scale). Firstly, during the initial week (0-to-5 days), the LFS1 expansion rate was ‘‘lower”, obtaining an expansiveness of 5%; a second stage had a ‘‘higher” expansion rate, of between 5 and 80 days, where an additional expansion of 21% was measured. In another work by the authors [19], the reasons for the aforementioned expansion rates have been identified as the hydration of free-CaO, mainly during the first days where expansiveness occurs at a quicker rate, and a second stage of expansiveness (long-term, over several months), governed by slower hydration of periclase (its hydroxides and subsequent carbonates). Hence, in the long term, due to the presence of periclase, it is difficult to accurately predict the total swelling. Table 3 shows the presence and evolution (0, 3 and 80 aged days) of the LFS1 mineral constituents. After an exposure time of 3 days, increases, shown in Table 3, were observed in the respective contents of calcite (carbonation of residual lime and portlandite), boehmite (after partial hydration of calcium aluminates), and hydrotalcite-sjögrenite (iron, aluminum and magnesium hydroxylation, carbonation and hydration). After a

LFS1 mineral constituents

0d

3d

80 d

Tricalcium aluminate (Ca3Al2O6) celite Calcium-olivine (Ca2SiO4) Boehmite (c-AlO(OH)) Brucite (Mg(OH)2) Calcite (CaCO3) Hydrotalcite (Mg6Al2(CO3)(OH)16  4(H2O)) Katoite (Ca3Al2(OH)12) Mayenite (Ca12Al14O33) Periclase (MgO) Sjögrenite (Mg6Fe(CO3)(OH)16  4(H2O)) Aluminum calcium hydrated sulfate (Ca4Al2O6(SO4)14H2O)



















































































full 3 months of exposure, the periclase disappeared, apparently resulting in emerging brucite and a slight increase in sjögrenite. In addition, the presence of hydrated and carbonated compounds including calcite increased. Tricalcium aluminate remained unreacted at these temperatures, although hydrated aluminum calcium sulfate was observed. The carbonation of hydroxides portlandite and brucite caused no expansiveness. However, as previously stated by the authors [19], hydrotalcites and hydrocaluminates could be formed in more complex and very expansive reactions, with subsequent visible swelling. In this context, the appearance of hydrotalcite and sjögrenite clearly contributed to the observed expansion. Calcium aluminates (celite and mayenite) have low swelling in their hydration reactions [35], and can even show long-term contraction, due to subsequent chemical reactions of conversion between calcium aluminate hydrates (as studied in aluminous cements). As shown in Fig. 4, the X-ray diffraction patterns presented the evolution of LFS1 over time of aging (from 0 to 80 days) and the appearance of expansive compounds (hydrotalcite and sjögrenite). Additionally, SEM, Fig. 4, allows us to observe both states of crystalline growth. In contrast to LFS1, LFS2 was not weathered in storage at the steelmaking plant. It underwent the above-mentioned steam test and, at 19 days, it had obtained a total expansiveness of over 8%; the measured expansion law of LFS2 showed time dependency, also according to two stages, as shown in Fig. 3b. The first stage (0–5 days) corresponds mainly to hydration and carbonation of free-CaO and shows a slope similar to that of LFS1 in its first stage; the second stage (5–16 days) must correspond mainly to the

Fig. 3. (a) Apparatus to measure swelling under steam exposure and (b) graphic of curves.

408

T. Herrero et al. / Construction and Building Materials 123 (2016) 404–413

Fig. 4. Changes in LFS1: hydrotalcite (HT) and sjögrenite (SJ).

periclase carbo-hydration showing a lower slope; from 16 days the swelling of this LFS2 slag was practically inexistent. As has been mentioned, the results of the XRD analysis in its aged state were a little different (portlandite, brucite, katoite, calcium-magnesite and calcite as new compounds, calcium-olivine as unaltered compound) than those of LFS1, and they were presented in a detailed form in a previous work, see Ref. [19].

3.2. Reactivity of LFS under moist chamber conditions In contrast to the previous hydration procedure based on steam action, it appears convenient to analyze other weathering mediums, closer to reality. Accordingly, controlled hydration in a moist chamber allowed us to analyze the aging effect, due to the direct contact between the LFS surface and an environment of high humidity at room temperature. The LFS samples, in the asreceived state, inside a siliceous sample holder (providing a good contact surface with the atmosphere), were then exposed in the classical moist chamber at 20 °C and 98% RH. They were analyzed with the XRD technique at different stages: 0 h, 3 days and 90 days. The XRD results for the LFS1 slag are shown in Table 4. They revealed the final disappearance of several compounds, such as tricalcium aluminate and mayenite, with the appearance of katoite and bayerite. Calcium aluminates, calcite, hydrotalcite, aluminum hydroxides and others were found in abundance in the aged sample. Obviously, the results are different from those obtained with LFS1 aged by exposure to a steam stream; the lack of reactivity of calcium aluminates with water under steam conditions is notorious, as against their reactivity at room temperature.

Table 4 X-ray diffraction results for LFS1 after moist chamber aging. LFS1 mineral constituents

0d

3d

Tricalciumaluminate (Ca3Al2O6) Celite Bayerite (Al(OH)3) Calcium-olivine (Ca2SiO4) Brucite (Mg(OH)2) Calcite (CaCO3) Hydrotalcite (Mg6Al2(CO3)(OH)16  4(H2O)) Katoite Ca3Al2(OH)12 Mayenite (Ca12Al14O33) Periclase (MgO) Sjögrenite (Mg6Fe(CO3)(OH)16  4(H2O))























































90 d

























Besides, the brucite disappeared increasing the hydrotalcite, but the periclase remained in its original proportion; obviously, the chemical conditions necessary for hydration or carbonation of free-periclase were not present in this test, meaning that the slag retains the potential for long-term expansivity. SEM imaging and EDX micro-analysis performed after 90 days (Fig. 5), revealed equidimensional needle-like hydrotalcite crystal compounds named HT (expansiveness trends), polyhedral crystals of calcite CT, and a mass with a high aluminum content AG, mainly consisting in aluminum hydroxide gel and hydrated calcium aluminates (katoite-bayerite). The silica-saturated slag LFS2 kept in the moist chamber was studied in a similar way to LFS1. As shown in Table 5, after 90 days, the hydrated LFS2 had a large amount of calcite and hydrated calcium tricarboaluminate, at the expense of mayenite and

409

T. Herrero et al. / Construction and Building Materials 123 (2016) 404–413

3.3. Discussion

Fig. 5. SEM analysis of LFS1 in the moist chamber.

Table 5 X-ray results of LFS2 after moist chamber aging. LFS2 mineral constituents

0d

3d

90 d

Calcium-olivine Ca2SiO4 Calcite CaCO3 Fluorite CaF Jasmundite Ca22(SiO4)8O4S2 Mayenite Ca12Al14O33 Oldhamite CaS Periclase MgO Calcium tricarboaluminate hydrated Ca6Al2O6CO9  32H2O

























































tricalciumaluminate (Celite) that disappeared. Hence, mayenite and Celite hydration released calcium and aluminum that, in combination with atmospheric CO2 and water, led to new phases. Hence, after 3 months in the moist chamber, the LFS2 was mainly composed of calcite, tricarboaluminate, olivine, periclase and other minor components, although no portlandite appeared. The presence of periclase reveals a potential long-term expansivity in the aged LFS2 slag. SEM imaging (Fig. 6) revealed equidimensional crystals of calcite CT and hexagonal-acicular prisms of hydrated tricarboaluminate TCA (crystalline structures of expansiveness).

By no means is it evident to the authors that the chemical reactions that took place in the steam aging at 130 °C in the LFS slag were the same reactions noted after long-term exposure to room temperature; it has been proven that the hydratable calcium aluminates (mayenite, celite) remained unhydrated in that steam test. The accelerated aging of slag cannot be considered to be independent of the aging temperature, and the temperature reached in the steam test is excessive, so it may not be representative of a slow natural aging of the LFS slag. A limit value in the range of 70– 80 °C should be established, in the opinion of the authors, for this kind of accelerated aging; this statement agrees with the test conditions proposed by the D-4792 ASTM standard [36]. Moreover, aging at room temperature in the moist chamber was also insufficient for both kinds of slags, taking into account that the periclase remains not-hydrated or carbonated after the test. This fact leads us to consider these room temperature aged slags as unstable, and the risk remains that they may develop posterior swelling under appropriate conditions. Finally, a TG-DSC test on the aged slags might indicate an approximation to the initial content of free-CaO and free-MgO, based on stoichiometric calculations taking in mind all new compounds containing them, revealed by X-ray diffraction. These calculations are in fact a little complicated and inaccurate, and in the opinion of the authors, somewhat pointless.

4. Design of cement mixes As presented in Fig. 1, the LFS2 slag type was of an acceptable grain morphology and size for use in hydraulic mixes; however, LFS1 had to be gently crushed to sizes of less than 1 mm. The study of LFS in cement matrices was done by casting cement pastes and non-structural (masonry) mortars, in order to analyze their volumetric instability and durability. 4.1. Cement pastes 10% and 20% of CEM I and II were replaced (by weight) with both types of LFS, and by natural fines (smaller than 0.063 mm, ASTM No. 250 sieve) from limestone crushing. Several trial (10  10  60 mm) specimens were manufactured with a water/ cement ratio (w/c) of 0.4. The labels of Tables 6 and 7 refer to I/II (cements), P (paste) and M (mortar), plus a figure referring to the fines (% partial substitution of cement by LFS or limestone powder), plus an alphanumeric character referring to the filler type: L1 = LF1, L2 = LFS2 and F = limestone fines. The specimens were tested in compression and analyzed by X-ray diffraction.

Table 6 Cement paste mixes. Mechanical behavior. Paste mixes

Fig. 6. SEM analysis of LFS2 under moist chamber conditions.

IP0 (ref.) IP10L1 IP20L1 IP10L2 IP20L2 IP10F IP20F IIP0 (ref.) IIP10L1 IIP20L1 IIP10L2 IIP20L2

Compressive strength (MPa) 7d

28 d

65.5 55.3 44.3 54.4 42.9 50.5 44.4 40.1 38.4 35.9 25.0 23.4

78.7 67.6 55.8 63.8 55.5 67.5 55.4 61.1 57.8 52.0 42.4 39.1

410

T. Herrero et al. / Construction and Building Materials 123 (2016) 404–413

Table 7 XRD results on cement pastes after 28 days. Mineral compounds (after 28 days)

IP0

IP20L1

IP20L2

IP20F

IIP0

IIP20L1

IIP20L2

Alite (Ca3SiO5) Tricalcium aluminate (celite) (Ca3Al2O6) Belite (Ca2SiO4) Bultfonteinite (Ca2SiO2F(OH)3) Brucite (Mg(OH)2) Calcite (CaCO3) Ettringite (Ca6Al2(SO4)3(OH)12 * 26H2O) Calcium semicarboaluminate hydroxide hydrated (Ca4Al2O6(CO3)0.5(OH) * 11.5H2O) Katoite (Ca3Al2(OH)12) Periclase (MgO) Portlandite (Ca(OH)2) Gypsum (CaSO4 * 2H2O)





















































;

In the cement I, see Table 6, the addition of natural limestone fines and both LF slags caused a similar decrease in the strength of the pastes at all ages and under all the proportions that were analyzed. In the cement II mixes, containing fly ash as an initial addition of cement, the use of LFS2 was more detrimental than the use of LFS1, the useful hydraulicity of which is reflected in its higher compressive strength. That weakened strength detected in the pastes containing LFS2 and fly ash points to a none-too positive interaction (at least in the short term) between these two additions in the presence of Portland clinker. Further research is planned by authors on this question. Additionally, the hydraulic reaction of slags was verified by the X-ray diffraction analysis of the pastes at both ages. As most significate trends, Table 7 shows the results of this analysis on mixes with the maximum of cement substitution (20%) after 28 days of aging in a moist chamber. In the case of the type I cement with LFS1, XRD analysis revealed the presence of the slag compounds, as unreactive celite and katoite. The presence of gypsum after 28 days is highly worrying in view of the subsequent formation of secondary ettringite; in the same way, the presence of periclase is also disturbing. Additionally, a moderate proportion of hydrated calcium carboaluminate hydroxide appeared in pastes IP0, IP10L1, IP20L1, IP10F and IP20F; the use of LFS2 slag was not favorable to this carbo-aluminate presence. Its influence on strength was low, but its presence is evidence of the hydraulicity of calcium carbonate fines. The addition of LFS2 adds to the mixtures the presence of issues from fluorine-spate, added to gypsum and periclase. The addition of limestone filler never provokes the detection of both dangerous components. In the case of type II cement, the XRD results also showed an absence of the calcium carbo-aluminate in pastes. It appears












































































;

















evident that the presence of high amounts of reactive silica in the fly ash inhibited the formation of this compound. However, the worrying presence of free gypsum continued in the LFS mixes. The presence of free-MgO (periclase or brucite) could be tolerated in low proportions in all these hydraulic mixes, as has been previously stated [1,27,31]; however, the presence of free calcium sulfate in pastes after 28 days was highly worrying and requires further investigation. 4.2. Masonry mortars The tests that yielded the results detailed above for the pastes were applied to masonry mortars with the same partial substitutions of cement by natural fines and LFS. The mixes in the study were referenced to a standardized mortar denoted ‘‘M10” in European standard EN 998-2 [34], widely used in commercial masonry mortars. Limestone sand 0–4 mm was used in large amounts as aggregate, as described in Section 2.1. The proportioning of the manufactured mortars was based on their workability, mainly the amount of water, and the proportion binder/aggregate was chosen, as specified in EN 998-2 [34], at 1/9 in weight. The total volume of mixtures was sufficient to fill moulds of 40  40  160 mm and the specimens were prepared in triplicate. Both reference mortars (known as IM0 and IIM0 in Table 8) were prepared with a workability of 175 ± 10 m spreading in shaking table EN 1015-3/A1:2007 [34] and a minimum value of 10 MPa (after 28d) in compressive strength. The fresh density was evaluated through EN 1015-6 [34] and the higher water demand from the CEM I mixes may be underlined; in cement II mixes, the presence of fly ash favored their workability. Mixes IIM10L2 and IIM20L2 showed too low mechanical responses in the corresponding pastes and were not studied.

Table 8 LFS masonry mortars. Mix design, fresh state and mechanical strength. Mixes

Water/cem/sand/LF (g/mixture)

Fresh density (Mg/m3)

Spreading (mm)

IM0 (ref.) IM10L1 IM20L1 IM10L2 IM20L2 IM10F IM20F IIM0 (ref.) IIM10L1 IIM20L1 IIM10F IIM20F

472/350/3150/0 472/315/3150/35 472/280/3150/70 472/315/3150/35 472/280/3150/70 472/315/3150/35 472/280/3150/70 455/350/3150/0 455/315/3150/35 455/280/3150/70 455/315/3150/35 455/280/3150/70

2.18 2.17 2.18 2.14 2.14 2.19 2.17 2.21 2.21 2.18 2.19 2.21

175 178 181 178 177 179 174 175 180 178 178 176

Compressive strength (MPa)

Flexural strength (MPa)

7d

28 d

45 d

7d

28 d

45 d

11.4 9.3 7.6 7.4 6.1 9.7 7.7 10.6 6.6 6.7 7.7 6.6

17.6 15.0 11.8 12.2 10.2 15.1 12.5 17.7 13.7 11.4 13.9 11.7

18.2 15.8 11.9 12.1 10.4 15.5 12.9 17.4 13.4 10.9 13.5 11.9

3.0 2.6 2.4 2.2 1.7 2.9 2.4 2.6 2.2 1.7 2.1 1.9

4.9 4.4 3.3 3.7 2.9 3.8 3.7 4.3 3.7 2.9 3.4 3.3

5.4 4.5 3.8 3.5 3.3 4.1 3.4 5.1 3.5 2.7 3.5 3.1

T. Herrero et al. / Construction and Building Materials 123 (2016) 404–413

The strength results under compression (each final value is the average of six trials) and flexural (three trials average) loads (Table 8) were performed according to EN 1015-11 [34], by testing prismatic samples cured for 24 h in a moist room (98%RH) and then held under water (room temperature) until those trial specimens were tested at 7, 28 and 45 days, respectively. The main parameter that deserves further discussion is compressive strength. Flexural strength tests were also performed, to identify eventual excessive scatter in the test results, which was not observed. All the mixes had compressive strengths of over 10 MPa (threshold) at 28 days. As observed in the pastes, the partial substitution of cement by additions (natural limestone or LF slag) implied a slight linearly dependent reduction in terms of compressive strength. The strength gain of almost all mixes in the interval of 28–45 days was, in general, low or negligible, and this is the reason why they no further long-term strength tests were performed. In the CEM II mixes, both the compression and flexural strengths even underwent a slight reduction between 28 and 45 days, also observed in some of the CEM I mixtures under flexural loads. In view of the mortar bearing capacity, it could be stated that the alumina-saturated slag appeared to be a slightly better suited addition than the silica-saturated LFS, in terms of the mechanical strength. Considering also the workability, LFS1 may be said to have a promising future as an addition to masonry mortars; the addition of LFS2 type slag in masonry mortar has been already analyzed and justified [1,31]. In fact, a maximum partial substitution of 20% ladle furnace slag (of any type) by weight of cement in these masonry mortars appears advisable, to achieve both a costeffective solution and a reliable level of performance.

411

discussed partial compensation of the mortar shrinkage by means of LFS expansiveness. Hence, it appears advisable to analyze this expansiveness and, to do so, following the specifications in standards ASTM C596-07 and ASTM C490-93 [36], four specimens (25  25  285 mm) were cast from each three mixtures (IM0ref., IM10L1 and IM10L2). The specimens were cured for 24 h in a moist room (20 ± 1 °C and 95%RH) and, once demolded, were kept submerged in water throughout the rest of the experiment, to avoid the influence of dry shrinkage of Portland cement mixes. In view of the expansiveness of the compounds that form when the LFS is wet, it was decided to apply a partial LFS substitution of only 10%. The length variation was measured in a rigid frame equipped with a 0.01 mm precision apparatus. The measured values were constant from an age of 125 days (almost stabilized as from day 75), as seen in Fig. 7. Each point represents one sample, and a potential function fit was performed. A higher asymptotic expansiveness value in the aluminasaturated slag was noted, in the set of values from both LFS mortars (IM10L1 and IM10L2), of 0.12 mm/m and 0.09 mm/m, respectively. In contrast, the IM0 reference mortar without slag, presented null expansiveness. Therefore, considering an average expansiveness in both LFS mortars of, approximately, 0.10 mm/m, the presence of LFS in masonry mortars could partially compensate the expected drying shrinkage of mortars, from LFS contents of 10%. However, it should be remarked that an excessive amount of LFS may be detrimental for other reasons already commented, even though it may partially compensate shrinkage contraction. 5.2. Wetting-drying cycles

5. Durability studies in masonry mortars Durability studies were based on CEM I mortar mixes, to avoid unknown influences from LFS and any active addition from blended cements. Two weathering studies guided experimental work: dimensional variations, and wetting-drying cycles. 5.1. Dimensional variations In masonry mortars, acceptable maximum drying shrinkage stands at around 0.5 mm/m, while in structural mortars a value close to 1 mm/m represents a limit. Some research works [1] have

These cycles simulate rainy days or humid conditions (water saturation) and spontaneous drying at lab scale. The weathering effect on masonry mortars was measured by testing the porosity and mechanical performance evolution at different control ages. Guided by ASTM D559 [36] on wetting and drying compacted soil-cement mixtures, but in application to LFS mortars, aluminasaturated LFS1 was used in partial substitution of CEM I, as this slag could be considered more dangerous than LFS2. By doing so, we are testing the most unfavorable possible situation with regard to slags, and if the results of durability with LFS1 are favorable, it may be understood that durability with LFS2 will be almost

Fig. 7. Evolution of specimen length over time in submerged specimens.

412

T. Herrero et al. / Construction and Building Materials 123 (2016) 404–413

Table 9 Physical and mechanical properties under wetting-drying cycles. Mortar mixes

IM0 (ref.) IM10L1 IM20L1

Weight variation (%)

Compressive strength (MPa)

Flexural strength (MPa)

0–360 d

0d

90 d

18 0d

36 0d

0d

90 d

180 d

360 d

+0.08 +0.09 +0.13

17.2 14.1 10.7

19.2 15.0 12.8

18.2 14.6 12.3

18.0 14.4 11.0

4.5 3.8 2.9

4.9 4.2 4.1

5.2 4.3 3.5

4.6 3.8 2.8

certain; durability tests of hydraulic mixtures that include LFS2 silica-saturated slag have been extensively performed by these and other authors [37,38]. The tested mortar mixes were IM0-ref., M10L1 and IM20L1; a total of twelve prismatic specimens (40x40x160 mm) were cast of each three mixtures and were held in a moist chamber over 28 days. They were then tested in 30 wetting-drying cycles (24 h/cycle) distributed in three daily stages. Firstly, 16 h submerged at room temperature water. Secondly, after wiping the specimens with a cloth (to stop dripping), they were dried for 6 h in a stove at 60 °C. Thirdly, each whole specimen was kept 2 h in air at room temperature (20 ± 2 °C), in order to avoid thermal shocks. After these wetting-drying cycles had been performed over 30 days, the specimens were exposed to weathering (under a roof, avoiding direct exposure to the rain) for one year, from day 30 to day 360. Once this period was over, weight variations, the results of a visual inspection, changes in compressive and flexural strength, and porosity evolution were all evaluated. Once the specimens had been dried in a stove (60 °C) up to constant weight, a visual inspection was performed. None of the specimens appeared to be damaged on their surface, presenting a chalky and uniform texture, without cracks, flaws, peeling or mass loss. Table 9 includes, among other results, the weight variation from 0 days (28 d after casting) until 360 days of weathering exposure. The higher LFS1 content (lower clinker content), the slightly higher long-term weight; the hydraulicity of the aluminasaturated slag (after one year exposure) all contributed to a gain in weight. The compressive and flexural strengths were tested at different aging periods: 0 d, 90 d, 180 d and 360 d, as presented in Table 9. Neither the compressive strength nor the flexural strength was

affected by the 30 wet-dry cycles, as shown comparing the results after 0 days (before the wetting-drying cycles) and after 90 days. Subsequently, both their compressive and their flexural strength dropped slightly after weathering of between 90 and 360 days in all mixes. The presence of alumina-saturated LFS1, in partial substitution of cement, had an outstanding detrimental influence after one year in proportions of 20% (with values of 11.0 MPa and 2.8 MPa in Table 9). These results are hardly surprising if we consider the aforementioned problems associated with the use of this high-alumina LFS slag, mainly in the sections on expansiveness and cement pastes of this article. The occurrence of deleterious chemical reactions among the worrying compounds (calcium sulfate, free magnesium oxide, free calcium oxide, aluminates. . .) in an uncontrollable way during the weathering, throughout the life of the constructive elements, is a real risk which must be studied in the future. The results obtained in Table 8 for the long-term compressive strength are somewhat different among the three mixes (from 18 to 11 MPa as the LFS1 content increases), indicating that important differences must exist among them in the internal structure. In the opinion of the authors, the chemical evolution of the mortar compounds over time is attributed mainly to the presence of slag, and the role of its components is not a too positive factor in the improvement of its strength. As is well known, the mechanical strength of ceramic materials is dependent on their pore structure, especially in the case of capillary-continuous porosity. In the present research, the MIP (mercury intrusion porosimetry) technique using an Autopore IV 9500 apparatus (Micromeritics) at pressure 33,000 psi was employed to study the evolution of pore structure on the wetting-drying samples. Table 10 shows the main average results of three mixes at four aging stages, and Fig. 8 displays the form of the curves which reflects the internal pore structure. The higher LFS1 content produces the higher porosity (24.1%) and pore-size (0.67 lm) at 0 days, although this singularity was reduced throughout the aging process. Besides, the form of the curves (Fig. 8) is fairly similar in the three studied mixes at 0 days, and subsequently their evolution along time conserves that similarity. Briefly, the presence of LFS in the mixes produces no drastic changes in the pore structure. In the short term, the porosity of the LFS is usually considered to add to the porosity of the mortar matrix; hence, the initial values of IM20L1 are slightly higher than those of IM0. In contrast to the reference mortar, the amount of LFS mortar pores over aging was

Table 10 Porosity evolution during wetting-drying cycles. Mortar mixes

IM0 (ref.) IM10L1 IM20L1

Average Øpore (lm)

Total porosity (%)

Pore size distribution

0d

90d

180d

360d

0d

90d

180d

360d

0.35 0.48 0.67

0.31 0.33 0.52

0.34 0.52 0.50

0.38 0.44 0.44

21.8 23.1 24.1

21.9 22.6 23.1

21.6 22.9 22.2

22.6 22.3 22.0

Fig. 8. MIP results for mortars IM0, IM10L1 and IM20L1.

Bimode asymmetry to small sizes Bi/unimode asymmetry to small sizes Bi/unimode asymmetry to small sizes

T. Herrero et al. / Construction and Building Materials 123 (2016) 404–413

reduced (by 2.1 units in IM20L1) and the average pore diameter (0.23 lm) was also reduced, due to new hydration products or new phase precipitations. Finally, after 360 days, the porosity values (size, distribution) obtained in the three mixes were fairly similar. 6. Conclusions Ladle furnace slags are mainly composed of aluminum, silicon calcium and magnesium oxides, as well as oxides of titanium and iron. This composition is completed with a huge variety of chemical compounds depending on the production conditions of the basic refining steelworks. Its use must take into account its previous weathering-aging in the factory or in the stocking. With regard to the casting cement mixes (pastes and masonry mortars), the alumina saturated slags may be said to show similar or better physical-mechanical performance than the silica saturated slags. Hence, LFS could be said to induce slight hydraulic reactivity and, even a content of below 20% by weight of LFS (in substitution of cement) in mixes had low or non-negative effects on the mechanical performance and workability of the masonry mortars. Durability of mixes appears as non-problematic, but it should be studied in detail to evaluate the importance of the presence of free gypsum. The use of ladle furnace slag in masonry mortars is an affordable application for this by-product. The LFS-mortar mixes show useful properties for building sector applications; they constitute a contribution to global sustainability by partially substituting the consumption of cement and proportionally reducing greenhouse gas emissions. Acknowledgements Our thanks to the Basque Government for financial support to Research Group IT781-13, to the Vice-Rectorate of Investigation of the University of the Basque Country (UPV/EHU) for Grant PIF 2013 and, likewise, to the Spanish Ministry MINECO and FEDER Funds for financial support through Project BlueCons: BIA201455576-C2-2-R. Part of this research work was also conducted within the FISSAC project framework (European Union’s Horizon 2020 under Grant agreement No. 642154). We are also grateful to assistance with materials from two companies: Arcelor Mittal (Sestao-Vizcaya) and Morteros y Revocos Bikain. References [1] A. Santamaría, E. Rojí, M. Skaf, I. Marcos, J.J. González, The use of steelmaking slags and fly ash in structural mortars, Constr. Build. Mater. 106 (2016) 364– 373, http://dx.doi.org/10.1016/j.conbuildmat.2015.12.121. [2] J. Geiseler, Use of steelworks slag in Europe, Waste Manage. 16 (1–3) (1996) 59–63, http://dx.doi.org/10.1016/S0956-053X(96)00070-0. [3] H. Motz, J. Geiseler, Products of steel slags an opportunity to save natural resources, Waste Manage. 21 (3) (2001) 285–293, http://dx.doi.org/10.1016/ S0956-053X(00)00102-1. [4] P.J. Koros, Dusts, scale, slags, sludges... Not wastes, but sources of profits, Metall. Mater. Trans. B. 34 (6) (2003) 769–779, http://dx.doi.org/10.1007/ s11663-003-0083-0. [5] J.O. Akinmusuru, Potential beneficial uses of steel slag wastes for civil engineering purposes, Resour. Conserv. Recycl. 5 (1) (1991) 73–80, http://dx. doi.org/10.1016/0921-3449(91)90041-L. [6] M. Tüfekçi, A. Demirbasß, H. Genç, Evaluation of steel furnace slags as cement additives, Cem. Concr. Res. 27 (11) (1997) 1713–1717, http://dx.doi.org/ 10.1016/S0008-8846(97)00158-0. [7] C. Garcia, J.T. San Jose, J. Urreta, Reuse and valorization in civil works of electric arc furnace (EAF) slag produced in C. A. P. V, in: REWAS’99: Global Symposium on Recycling, Waste Treatment and Clean Technology, ISBN 84-923445-63, 1999, pp. 417–424. [8] T.R. Naik, Greener concrete using recycled materials, Concr. Int. 24 (7) (2002) 45–49. [9] P.K. Mehta, Global concrete industry sustainability, Concr. Int. 31 (02) (2009) 45–48.

413

[10] I. Papayianni, E. Anastasiou, Production of high-strength concrete using high volume of industrial by-products, Constr. Build. Mater. 24 (8) (2010) 1412– 1417, http://dx.doi.org/10.1016/j.conbuildmat.2010.01.016. [11] G. Wang, Y. Wang, Z. Gao, Use of steel slag as a granular material: volume expansion prediction and usability criteria, J. Hazard. Mater. 184 (1) (2010) 555–560. [12] M. Etxeberria, C. Pacheco, J.M. Meneses, I. Berridi, Properties of concrete using metallurgical industrial by-products as aggregates, Constr. Build. Mater. 24 (9) (2010) 1594–1600, http://dx.doi.org/10.1016/j.conbuildmat.2010.02.034. [13] M. Pasetto, N. Baldo, Mix design and performance analysis of asphalt concretes with electric arc furnace slag, Constr. Build. Mater. 25 (8) (2011) 3458–3468, http://dx.doi.org/10.1016/j.conbuildmat.2011.03.037. [14] C. Pellegrino, P. Cavagnis, F. Faleschini, K. Brunelli, Properties of concretes with black/oxidizing electric arc furnace slag aggregate, Cem. Concr. Compos. 37 (2013) 232–240, http://dx.doi.org/10.1016/j.cemconcomp.2012.09.001. [15] C. Pellegrino, V. Gaddo, Mechanical and durability characteristics of concrete containing EAF slag as aggregate, Cem. Concr. Compos. 31 (9) (2009) 663–671, http://dx.doi.org/10.1016/j.cemconcomp.2009.05.006. [16] M.A. González-Ortega, I. Segura, S.H.P. Cavalaro, B. Toralles-Carbonari, A. Aguado, A.C. Andrello, Radiological protection and mechanical properties of concretes with EAF steel slags, Constr. Build. Mater. 51 (2014) 432–438, http:// dx.doi.org/10.1016/j.conbuildmat.2013.10.067. [17] E. Anastasiou, K. Georgiadis Filikas, M. Stefanidou, Utilization of fine recycled aggregates in concrete with fly ash and steel slag, Constr. Build. Mater. 50 (2014) 154–161, http://dx.doi.org/10.1016/j.conbuildmat.2013.09.037. [18] T. Herrero, I. Vegas, J.T. San-José, J.J. González, A. Santamaría, D. Garcia, Ladle furnace slag in cement mixes experimental studies, in: WASCON 2015-9th International Conference on the Environmental and Technical Implications of Construction with Alternative Materials. Santander SPAIN 2015, ISBN 978-84606-8423-7, 2015, pp. 153–159. [19] V. Ortega-López, J.M. Manso, I.I. Cuesta, J.J. González, The long-term accelerated expansion of various ladle-furnace basic slags and their soilstabilization applications, Constr. Build. Mater. 68 (2014) 455–464, http://dx. doi.org/10.1016/j.conbuildmat.2014.07.023. [20] D. Goldring, L. Juckes, Petrology and stability of steel slags, Ironmaking Steelmaking 24 (6) (1997) 447–456. [21] W. Posch, H. Presslinger, H. Hiebler, Mineralogical evaluation of ladle slags at voestalpine Stahl GmbH, Ironmaking Steelmaking 29 (4) (2002) 308–312. [22] I.Z. Yildirim, M. Prezzi, Chemical, mineralogical, and morphological properties of steel slag, Adv. Civ. Eng. 2011 (2011) 597–604, http://dx.doi.org/10.1155/ 2011/463638. Article ID 463638:13 p. [23] I. Papayianni, E. Anastasiou, Effect of granulometry on cementitious properties of ladle furnace slag, Cem. Concr. Compos. 34 (3) (2012) 400–407. [24] A. Mladenovicˇ, B. Mirticˇ, A. Meden, V.Z. Serjun, Calcium aluminate rich secondary stainless steel slag as a supplementary cementitious material, Constr. Build. Mater. 116 (2016) 216–225. [25] C. Shi, Characteristics and cementitious properties of ladle slag fines from steel production, Cem. Concr. Res. 32 (3) (2002) 459–462. [26] J. Setién, D. Hernández, J.J. González, Characterization of ladle furnace basic slag for use as a construction material, Constr. Build. Mater. 23 (5) (2009) 1788–1794, http://dx.doi.org/10.1016/j.conbuildmat.2008.10.003. [27] J.M. Manso, D. Hernández, M.M. Losanez, J.J. González, Design and elaboration of concrete mixtures using steelmaking slags, ACI Mater. J. 108 (6) (2011) 673– 681. [28] D. Adolfsson, R. Robinson, F. Engström, B. Björkman, Influence of mineralogy on the hydraulic properties of ladle slag, Cem. Concr. Res. 41 (8) (2011) 865– 871, http://dx.doi.org/10.1016/j.cemconres.2011.04.003. [29] J.M. Manso, V. Ortega-López, J.A. Polanco, J. Setién, The use of ladle furnace slag in soil stabilization, Constr. Build. Mater. 40 (2013) 126–134, http://dx.doi.org/ 10.1016/j.conbuildmat.2012.09.079. [30] J. Montenegro, M. Celemín-Matachana, J. Cañizal, J. Setién, Ladle furnace slag in the construction of embankments: expansive behavior, J. Mater. Civ. Eng. 25 (8) (2012) 972–979. [31] J.M. Manso, M. Losañez, J.A. Polanco, J.J. Gonzalez, Ladle furnace slag in construction, J. Mater. Civ. Eng. 17 (5) (2005) 513–518. [32] A. Sáez-de-Guinoa Vilaplana, V.J. Ferreira, A.M. López-Sabirón, A. Aranda-Usón, C. Lausín-González, C. Berganza-Conde, et al., Utilization of ladle furnace slag from a steelwork for laboratory scale production of Portland cement, Constr. Build. Mater. 94 (2015) 837–843, http://dx.doi.org/10.1016/ j.conbuildmat.2015.07.075. [33] L. Juckes, The volume stability of modern steelmaking slags, Min. Process. Extractive Metall. 112 (3) (2003) 177–197. [34] CEN European Committee for standardization. Rue de Stassart, 36. Brussels B1050. [35] Y. Iguchi, T. Narushima, C. Izumi, Calorimetric study on hydration of CaObased oxides, J. Alloy. Compd. 321 (2) (2001) 276–281. [36] International A. Annual Book of ASTM Standards. West Conshohocken, 194292959. PA, USA, 2008. [37] J.M. Manso, A. Rodríguez, A. Aragón, J.J. González, The durability of masonry mortars made with ladle furnace slag, Constr. Build. Mater. 25 (2010) 3508– 3519. [38] E. Anastasiou, I. Papayianni, M. Papachristoforou, Behavior of self compacting concrete containing ladle furnace slag and steel fiber reinforcement, Mater. Des. 59 (2014) 454–460.