Академический Документы
Профессиональный Документы
Культура Документы
Interfacial transition zone of cement composites with steel furnace slag aggregates
PII: S0958-9465(16)30745-4
DOI: 10.1016/j.cemconcomp.2017.11.012
Reference: CECO 2942
Please cite this article as: A.S. Brand, J.R. Roesler, Interfacial transition zone of cement composites
with steel furnace slag aggregates, Cement and Concrete Composites (2017), doi: 10.1016/
j.cemconcomp.2017.11.012.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
PT
6 Champaign, 205 N. Mathews Ave., Urbana, Illinois, USA, 61801
7
RI
8 Abstract
9 As previous studies of mortar and concrete with steel furnace slag (SFS) aggregates have shown
SC
10 increases or decreases in the bulk mechanical properties, this study investigated the
U
11 microstructural cause of these opposing trends through characterization of the interfacial
AN
12 transition zone (ITZ) with quantitative image analysis of backscatter electron micrographs. Three
13 SFS types – basic oxygen furnace (BOF), electric arc furnace (EAF), EAF/ladle metallurgy
M
14 furnace (EAF/LMF) – were examined as aggregates in a portland cement mortar. The ITZ size
15 for all SFS mortar mixtures was similar, with the ITZ of BOF and EAF/LMF being slightly more
D
16 porous than mortar mixtures with EAF or dolomite. Microstructural examinations of the SFS
TE
17 particle revealed that BOF and EAF/LMF aggregates have different outer and interior
18 compositions, with the outer composition consisting of a porous layer, which likely contributes
EP
19 to the reduced strength relative to EAF. The imaging results demonstrated that the type of SFS
C
20 and its spatial composition greatly influences the bulk properties of mortar and concrete, mainly
AC
21 as a function of porosity content in the ITZ and the outer layer and interior porosity of the SFS
22 aggregate.
*
Corresponding author; abrand2@illinois.edu
1
Present address: Materials and Structural Systems Division, National Institute of Standards and Technology, 100
Bureau Drive, Gaithersburg, MD, USA, 20899
1
ACCEPTED MANUSCRIPT
24 (SEM), steel furnace slag (SFS), basic oxygen furnace (BOF) slag, electric arc furnace (EAF)
26
PT
27 1. Introduction
RI
28 A by-product slag is generated during steel production, which can be broadly termed steel
29 furnace slag (SFS), with modern steel productions typically utilizing basic oxygen furnace
SC
30 (BOF) or electric arc furnace (EAF) processes [1]. SFS consists of various silicate and oxide
32
U
wüstite (FeO) [2]. In addition, free calcium oxide (CaO) and free magnesium oxide (MgO) are
AN
33 often present, which expand by 92% and 120%, respectively, when reacted with water [3]. The
34 free CaO content, which can be around 1-10% for BOF slag and 0-4% for EAF slag [4], and its
M
35 potentially expansive nature are the main reasons why SFS aggregates are not often considered
D
36 for use as an aggregate in bound (i.e., concrete) applications, but the hard and durable
TE
41 literature as to whether SFS aggregates will increase, decrease, or have little effect on concrete
AC
42 mechanical and durability properties, as can be seen in Table 1. The observed discrepancy is
43 likely a function of the SFS source type, as the mineralogy and properties of the SFS are
44 dependent on the production process. One study [6] examined two different SFS aggregates, one
45 BOF slag and one EAF slag, and found that the EAF slag increased and the BOF slag decreased
2
ACCEPTED MANUSCRIPT
46 the compressive strength while both slag aggregates decreased the split tensile strength,
47 increased the fracture properties, and increased the drying shrinkage relative to concrete with
48 dolomite aggregates.
49
PT
50 Table 1. Effects of SFS Aggregates on Concrete Properties*
No. of Studies Proportion (%) of Studies that Showed Effect
Concrete Property
RI
Surveyed Increase Decrease Similar or No Effect
Compressive Strength 39 72 13 15
Split Tensile Strength 21 71 14 14
Flexural Strength 19 68 26 5
SC
Modulus of Elasticity 16 81 6 13
Dynamic Modulus 2 100 0 0
Drying Shrinkage 5 20 60 20
Water Absorption 7 29 57 14
U
Abrasion Resistance 4 100 0 0
*After a literature review in Ref. [7]
51
AN
52 The objective of this study is to examine the microstructure of three different SFS
M
53 aggregates and their effects on the interfacial transition zone (ITZ) in a cement mortar in order to
54 evaluate any explanations for these mechanical property discrepancies in the literature. A few
D
55 studies have examined the ITZ of cementitious composites with SFS aggregates through
TE
56 indentation testing, but all of these studies have shown that SFS aggregates will increase
57 concrete strength and improve the ITZ properties [8–12]. In another study that found higher
EP
58 strengths with EAF aggregates, qualitative image analysis showed that the ITZ with limestone
59 aggregates is more porous than the ITZ with EAF slag aggregates [13]. Therefore, as the
C
60 previous ITZ studies have all been in instances when the strength was shown to increase with the
AC
61 use of SFS aggregates, it is uncertain if the ITZ characteristics are different from natural
62 aggregates in the instances were strength is decreased by the use of SFS aggregates.
63 The variability in the effect of the SFS aggregates on concrete properties may be the
64 result of the properties of the SFS particle. In particular, it has been found that the exterior of the
3
ACCEPTED MANUSCRIPT
65 SFS aggregate may be of different composition than the interior [14]. For example, the exterior
66 of a weathered BOF slag can contain calcite, calcium silicate hydrate, and calcium
67 carboaluminate hydrate phases, which has been argued to be the source of concrete strength
68 reduction as the result of a poorer paste-aggregate bond [15]. However, it has also been argued
PT
69 that an increase in concrete strength with SFS aggregates is because of the rough surface texture
RI
70 of the particles as well as a chemical reaction occurring between the slag and the cement [8–10].
71 Furthermore, aggregate particle shape, gradation, phase composition, porosity, and surface
SC
72 texture are also known factors that can affect concrete strength [16,17], and these factors will
73 vary depending on the SFS type and source. Pang et al. [18] confirmed this by demonstrating
74
U
that carbonated SFS has a different porosity (i.e., fewer larger pores and more smaller pores) and
AN
75 mineralogy (i.e., more calcium carbonate) relative to an unmodified SFS, and these changes were
77
D
78 2. Experimental Design
TE
79 In this study, three different SFS sources – which were characterized previously for
80 chemical, physical, and mechanical properties [6] – were evaluated primarily for microstructural
EP
81 ITZ properties as well as bulk strength and stiffness. Several pertinent properties of the three SFS
82 sources are summarized in Table 2. When 100% of the coarse aggregate in a concrete mixture
C
83 was replaced by the same SFS1 (BOF) or SFS2 (EAF), the split tensile strength was reduced, the
AC
84 free drying shrinkage was increased, and the freeze/thaw durability was similar to concrete with
86 To evaluate the mechanical properties, coarse mortars were created that consisted of 40%
87 aggregate and 60% cement paste by volume. A high paste content was used to allow enough
4
ACCEPTED MANUSCRIPT
88 space between aggregates to provide clear, distinct ITZs to be imaged by backscatter electron
89 microscopy. The cement paste was Type I portland cement with a water-to-cement ratio of 0.42.
90 The aggregates all had the same gradation of particles passing the 3/8-inch (9.5 mm) sieve and
91 retained on the #4 (4.75 mm) sieve; this was performed to avoid particle size effects on the ITZ
PT
92 (e.g., see Ref. [19]). All aggregates were washed to remove any dust or fines and then were dried
RI
93 at 50% relative humidity and 23°C. As the aggregates had similar absorption contents (BOF
94 2.3%, EAF 1.8%, EAF/LMF 1.7%, and dolomite 1.8% [6]), it is expected that there is minimal
SC
95 effect of aggregate water absorption on the ITZ. Cylindrical specimens were cast that measured 2
96 inches (50 mm) in diameter by 4 inches (100 mm) in height and were tested at ages of 1, 7, and
97
U
28 days for compressive strength [20], split tensile strength [21], and dynamic modulus [22]. At
AN
98 each age, three replicates were tested for each strength test and six replicates were tested for the
99 dynamic modulus test. The specimens were cast, covered and cured for 24 hours at ambient
M
100 laboratory conditions, and then stored in lime saturated water at 23°C until testing.
D
101
TE
102 Table 2. Characteristics and Properties of the SFS Sources (from Brand and Roesler [6])
SFS1 SFS2 SFS3
Source Type BOF EAF EAF/LMF*
Free CaO Content 3.4% 0.1% 0.4%
EP
5
ACCEPTED MANUSCRIPT
104 After 1, 7, and 28 days of curing, samples from the mortar specimens were taken for ITZ
105 analysis. Thin sections (<1 mm thick) were cut and vacuum dried for 24 hours prior to epoxy
106 impregnation. Once the epoxy was set, the samples were polished down to 0.25 µm using
107 successively finer grits of silicon carbide paper and diamond paste. Additional preparation
PT
108 details can be found elsewhere [23,24].
RI
109 At least 50 images at 500x magnification were taken per age using a JEOL JSM-6060LV
110 scanning electron microscope in compositional backscatter electron (BSE) mode at a 12 keV
SC
111 accelerating voltage. Each image was 8-bit greyscale consisting of 960 by 1280 pixels with each
112 pixel measuring approximately 0.2 by 0.2 µm. From greyscale thresholding and image
113
U
processing, it is possible to obtain details about the distribution of detectable porosity, calcium
AN
114 hydroxide (CH), and unhydrated (UH) cement in the ITZ as a function of distance from the
115 aggregate interface [25–28]. This study followed the method developed by Brand [23,24], which
M
116 built upon ITZ analysis concepts developed by Wong and Buenfeld [29,30] and CH morphology
D
117 concepts discussed by Diamond [31] and Gallucci and Scrivener [32].
TE
118 The cumulative greyscale histogram was used to threshold the detectable porosity based
119 on the pore overflow method by Wong and Buenfeld [29]. The CH and UH were thresholded
EP
120 from the peak positions in the greyscale histogram (e.g., see Figure 1 in Ref. [23]). Using the
121 Euclidean distance mapping approach [30], a percentage of a given thresholded phase could be
C
122 determined as a function of distance from the aggregate interface, accounting for the two-
AC
123 dimensional tortuosity of the interface (e.g., see Figure 3 in Ref. [23]). By collating the data from
124 a significant number of images, an estimate of the mean ITZ composition can be generated from
6
ACCEPTED MANUSCRIPT
126 Since it is possible that the CH morphology may influence the strength of concrete [33],
127 it is worth examining the CH morphological properties in the ITZ for comparison between the
128 different aggregates. As CH appears as an elongated particle in the ITZ, the equivalent circular
129 diameter of the CH particles is computed for comparative purposes [31]. The equivalent circular
PT
130 diameter (deq) is determined by 4 / , where s is the pixel size (0.2 µm) and A is the number
RI
131 of pixels that constitutes a given particle.
132 An additional investigation was performed to investigate the SFS interior and exterior
SC
133 composition. As the SFS interior and exterior (edge) compositions may be different [14],
134 particularly with regard to the formation of calcium carbonate or hydrated phases on the SFS
135
U
particle exterior [15], thin sections (<1 mm thick) of aggregate were cut, dried, cast in epoxy,
AN
136 polished down to 0.25 µm, and examined with BSE microscopy and energy dispersive X-ray
M
138
D
141 The mortar split tensile and compressive strengths and dynamic moduli are shown in
EP
142 Table 3 and Table 4, respectively, for ages of 1, 7, and 28 days. The strength, in general, for the
143 SFS aggregate mixtures was not higher than the control mixture except for a few ages and SFS
C
144 types. The split tensile strength was statistically higher than the control at 7 days for mixes with
AC
145 SFS2 and SFS3 while the compressive strength was only statistically higher than the control at
146 28 days for the mix with SFS2. It should be noted, however, that these mortars had a high paste
147 content, so the effects of the aggregates may not be as drastic as may be seen in concrete.
148 Therefore, for comparison, the concrete strength results for concrete with SFS [6] are also shown
7
ACCEPTED MANUSCRIPT
149 in Table 3, which demonstrated that concrete with SFS1 (BOF) had lower strengths than the
150 control; concrete with SFS2 (EAF) produced a statistically higher compressive strength than the
152
PT
153 Table 3. Strengths for Mortar Mixtures with Dolomite and SFS Aggregates
Split Tensile Strength (MPa) Compressive Strength (MPa)
RI
Age Mix
Average Standard Deviation Average Standard Deviation
Control 2.7 0.1 22.8 1.7
1 Day SFS1 (BOF) 3.0 0.3 22.2 0.8
SC
Mortar SFS2 (EAF) 3.1 0.4 24.0 0.7
SFS3 (EAF/LMF) 2.6 0.3 20.6 0.7
Control 3.0 0.9 34.8 3.8
U
7 Days SFS1 (BOF) 3.0 0.3 34.8 1.3
Mortar SFS2 (EAF) 4.9 * 0.3 31.7 7.3
AN
SFS3 (EAF/LMF) 4.6* 0.5 36.1 3.1
Control 4.5 0.4 41.1 2.2
28 Days SFS1 (BOF) 4.3 0.7 38.5 1.1
M
28 Days
SFS1 (BOF) 3.7* 0.2 40.1* 2.0
Concrete**
SFS2 (EAF) 4.4* 0.5 48.3 0.5
TE
*Means are statistically different from the control with 95% confidence
**Concrete results from Brand and Roesler [6]
154
EP
155 At all ages, the dynamic modulus was statistically higher for all SFS aggregate mixtures
156 relative to the control. This increase in dynamic modulus can be attributed, in part, to the
C
157 increase in SFS density, as the computed dynamic modulus is directly proportional to the
AC
158 specimen mass [22]. The apparent specific gravity for these aggregates are 3.5 (BOF), 3.9
159 (EAF), 3.7 (EAF/LMF), and 2.8 (dolomite) [6]. In addition, SFS aggregates have a high stiffness
160 [34,35], largely as a result of the high iron content [36], which may also contribute the greater
161 dynamic modulus values seen for SFS mortars. The BOF and EAF slag aggregates in this study
8
ACCEPTED MANUSCRIPT
162 were also shown to produce a statistically greater concrete stiffness relative to dolomite concrete
163 [7], which further supports that the SFS aggregate have a higher modulus than dolomite.
164
165 Table 4. Dynamic Moduli for Mortar Mixtures with Dolomite and SFS Aggregates
PT
Longitudinal Dynamic Transverse Dynamic
Modulus (GPa) Modulus (GPa)
Age Mix
Standard Standard
Average Average
Deviation Deviation
RI
Control 21.9 1.0 21.3 1.0
1 SFS1 (BOF) 27.6* 1.1 27.7* 1.3
Day SFS2 (EAF) 1.3 1.2
SC
30.4* 30.5*
SFS3 (EAF/LMF) 27.7* 1.6 26.8* 2.3
Control 29.6 0.5 29.2 1.2
7 SFS1 (BOF) 36.8* 1.9 37.1* 1.8
U
Days SFS2 (EAF) 39.1* 1.2 38.1* 1.5
SFS3 (EAF/LMF) 38.1* 1.4 36.9* 1.4
AN
Control 32.9 1.0 33.0 1.5
28 SFS1 (BOF) 40.4* 1.8 40.9* 2.2
Days SFS2 (EAF) 43.6* 0.5 43.9* 1.7
M
168 Examples of the BSE images of the SFS ITZ are shown in the Appendix. From image
169 processing, the detectable porosity, UH, and CH are reported in Figure 1, Figure 2, and Figure 3
EP
170 for mortars at 1, 7, and 28 days, respectively. In addition, the total C-S-H content can be
C
171 estimated by subtracting the detectable porosity, UH, and CH from 100% [37], but this estimated
AC
172 content includes what can otherwise be termed “other hydration products,” since it will also
174 At 1 day, the ITZ characteristics are practically the same for all mortars (Figure 1). This
175 is in agreement with Table 3 that showed that the 1-day strengths were not drastically different
176 between mixtures with and without SFS. Therefore, the statistical difference in modulus values
9
ACCEPTED MANUSCRIPT
177 at 1 day (Table 4) are a result of the stiffness and density of the SFS and not because of the ITZ
178 properties.
179 At 7 days, the ITZ characteristics begin to deviate slightly relative to the dolomite (Figure
180 2). Near the interface, within approximately 20 µm, the porosity is greater for the BOF and
PT
181 EAF/LMF slags relative to the dolomite while the EAF slag porosity is similar to dolomite.
RI
182 While the UH and C-S-H contents and distributions are similar for all mixtures, the CH content
183 appears to be slightly greater for dolomite near the interface (within approximately 10 µm)
SC
184 relative to the SFS mortars.
185 At 28 days, the ITZ characteristics are more noticeably different (Figure 3), particularly
186
U
the BOF and EAF/LMF slags relative to the dolomite, as there is greater porosity and less C-S-H
AN
187 within the ITZ (within about 40 µm). The UH contents and distributions are similar for all
188 mortars, which suggests that the absorption capacity of the aggregate did not influence hydration.
M
189 As at 7 days, the CH is greater for dolomite near the interface (within 10 µm) relative to the SFS
D
190 mortars.
TE
192 interfaces [25,31,38,39]. Monteiro and Mehta [40] suggested that, for carbonate mineral
EP
193 aggregates, CH reacts to form a calcium carbonate hydrate phase, which may increase the bulk
194 mechanical strength. Since the amount of CH at the interface is more prominent for dolomite, it
C
195 is possible that CH is more energetically or chemically attracted to dolomite (i.e., for reaction
AC
196 [40]), which may not be the situation for the SFS.
197 Comparing the 7- and 28-day ITZ trends to the strength findings in Table 3, the bulk
198 mortar strengths are statistically similar for BOF and the control despite that the BOF slag ITZ
199 appears to be more porous than dolomite. The EAF slag mortar ITZ is very similar to dolomite
10
ACCEPTED MANUSCRIPT
200 mortar, but the strengths can be statistically greater for EAF slag at later ages, and thus there
201 must be additional factors other than the ITZ causing the strength increase. Likewise, the
202 EAF/LMF slag ITZ trends were in between the trends for the BOF and EAF slags, but it also did
203 not significantly change the strength properties relative to the dolomite control.
PT
204
RI
U SC
AN
M
D
TE
EP
205
C
206 Figure 1. Detectable porosity, UH, and CH and estimated C-S-H at 1 day for mortars with
AC
11
ACCEPTED MANUSCRIPT
PT
RI
U SC
AN
M
209
D
210 Figure 2. Detectable porosity, UH, and CH and estimated C-S-H at 7 days for mortars with
211 dolomite and three SFS sources.
TE
212
C EP
AC
12
ACCEPTED MANUSCRIPT
PT
RI
U SC
AN
M
213
D
214 Figure 3. Detectable porosity, UH, and CH and estimated C-S-H at 28 days for mortars with
215 dolomite and three SFS sources.
TE
216
217 Using a statistical z-test to determine the size of the ITZ [23], i.e., when the ITZ porosity
EP
218 is no longer statistically different (95% confidence) than the bulk porosity, Table 5 shows that
219 the mean size of the ITZ may be slightly larger for mortars with SFS aggregates. Dolomite
C
220 mortar was found to have an ITZ thickness of around 40 µm, while the ITZ thickness for mortar
AC
222 The standard error, defined as the standard deviation normalized by the square root of the
223 sample size, was found to be ≤1.4% for all phase measurements in the ITZ (Table 5). Scrivener
224 [41] reported that the standard error of such measurements could be on the order 5-10% for data
225 collected from 50-100 images, while Crumbie [38] and Brand and Roesler [23] found standard
13
ACCEPTED MANUSCRIPT
226 error values of ≤2% and ≤1.5%, respectively, for phase content measurements in the ITZ of
227 concrete or mortar samples when averaging data from 50 or more images. Therefore, the
228 variability in the measurements for this study are consistent with past literature.
229
PT
230 Table 5. Size of the ITZ and Average Standard Error within the ITZ at 28 days
No. of Size of the Average Standard Error (%)
Mix
RI
Images ITZ* (µm) Porosity CH UH
Dolomite 58 40 0.9 1.0 0.8
SFS1 (BOF) 54 44 1.4 0.9 0.9
SC
SFS2 (EAF) 53 41 1.3 1.0 0.8
SFS3 (EAF/LMF) 57 46 1.3 0.8 1.0
*Determined based on porosity distribution
231
234 was conducted to evaluate if the SFS aggregate is affecting CH growth or nucleation, as other
M
235 recycled materials have been shown to affect CH growth [23]. The mean equivalent CH particle
D
236 size is shown in Table 6, indicating that the mean equivalent circular diameter varies very little
TE
237 as a function of distance from the interface. Immediately near the interface (within 1 µm), the
238 particles are slightly smaller for SFS compared to dolomite, which agrees with Figure 3 that
EP
239 there is more CH near the interface for dolomite. The mean CH particle size is roughly
240 equivalent in all mixtures when considering all particles within 5, 40, and 100 µm of the
C
241 interface, although the EAF slag mortar may exhibit a slightly smaller mean particle size. Figure
AC
242 4 indicates that all mortars (with and without SFS) have a similar amount of CH particles
243 throughout the ITZ and bulk matrix, and all mortar samples contained the most CH particles
244 within ~25 µm of the interface. Therefore, the SFS does not appear to significantly affect CH
245 nucleation or growth within the ITZ, although there is some evidence (e.g., CH trends near the
14
ACCEPTED MANUSCRIPT
246 interface in Figure 2 and Figure 3) from the BSE image processing to suggest that CH
248
PT
Dolomite SFS1 SFS2 SFS3
Mortar (BOF) (EAF) (EAF/LMF)
Within 1 µm of
4.7 4.4 4.2 4.4
RI
the Interface
Within 5 µm of
4.8 4.9 4.3 4.6
the Interface
SC
Within 40 µm
4.8 4.9 4.4 4.7
of the Interface
Within 100 µm
4.8 4.9 4.4 4.7
U
of the Interface
250
AN
2.5%
Percent of Total CH Particles
2.0%
1.5%
D
1.0%
TE
0.5%
0.0%
EP
0 20 40 60 80 100
Distance from the Aggregate Interface (μm)
251
AC
252 Figure 4. Histograms of CH occurrence as a function of distance from the aggregate interface
253 (within 100 µm of the aggregate interface) for each of the mortars (with and without SFS) at 28
254 days.
255
258 inherent microstructural characteristic of the SFS particle that is affecting the ITZ properties and
15
ACCEPTED MANUSCRIPT
259 the bulk mortar/concrete mechanical behavior. Therefore, polished sections of plain dolomite
260 and SFS particles were qualitatively examined with BSE microscopy and EDXS.
261 Examination of the EAF (SFS2) slag (Figure 5) indicates a matrix rich in calcium and
262 silicon (likely to be larnite, based on X-ray diffraction data in Table 2) with embedded particles
PT
263 rich in iron (wüstite and magnetite), which is similar to what was shown by Arribas et al. [11].
RI
264 The iron-rich phase also appears to contain magnesium, which is expected since a
265 magnesiowüstite phase is common in EAF slags rather than pure wüstite [42]. The aluminum-
SC
266 containing phases (i.e., mayenite) are also evident in the matrix and appear to preferentially
267 occur along the edge of the particle. In Figure 6, the edge of the EAF particle appears to contain
268
U
less of the iron-rich phases compared to the interior of the EAF particle; with image processing,
AN
269 for Figure 6a it is estimated that, within 50 µm of the particle edge, 12% of the area consists of
270 the iron-rich phase compared to 41% of the area in the interior (assumed to be the region greater
M
271 than 50 µm from the particle edge) whereas Figure 6b in total consists of approximately 41% of
D
273
274
EP
275
276
C
277
AC
278
16
ACCEPTED MANUSCRIPT
PT
RI
(a) BSE (b) Ca
U SC
AN
M
D
(c) Si (d) Fe
TE
C EP
AC
(e) Al (f) Mg
279 Figure 5. Compositional BSE (a) and EDXS maps (b-f) of the edge of an EAF slag particle. The
280 black phase along the right hand side in the BSE micrograph is epoxy. Note: EDXS images have
281 been contrast-enhanced to better indicate the elemental density.
282
17
ACCEPTED MANUSCRIPT
PT
RI
(a) (b)
SC
283 Figure 6. Compositional BSE micrographs of the edge (a) and interior (b) of an EAF slag
284 particle.
285
U
286 In contrast to the EAF slag, the BOF (SFS1) slag exhibits a clear compositional
AN
287 difference along the edge of the particle relative to the interior, as shown in Figure 7, which is
288 consistent with the BSE micrographs of BOF slag shown by Coomarasamy and Walzak [14].
M
289 This outer layer with different composition did not exist in some sections of the particle and
290 ranged up to 1 mm thickness, but more commonly was on the order of a few tens to hundreds of
D
291 microns thick. EDXS mapping indicates that this outer layer is rich in calcium and contains less
TE
292 silicon and iron than the interior. This layer is also more porous; by the porosity thresholding
EP
293 technique used previously, the BOF interior (Figure 7a) has 0.3% porosity, while the estimated
294 porosities of the edge and interior of the particle in Figure 7b are 24% and 0.1%, respectively,
C
295 and the estimated porosities of the edge and interior of the particle in Figure 7c are 10% and
AC
296 0.1%, respectively. The greater porosity of this outer layer is also the source of the slightly
297 higher water absorption of the BOF slag relative to other SFS and dolomite: 2.3% for BOF
298 compared to 1.7-1.8% for EAF, EAF/LMF, and dolomite [6]. Ultimately, this outer layer is
299 likely to be composed of calcium carbonate and other reaction products (e.g., calcium hydroxide
300 or some form of a calcium silicate hydrate, since both calcium and silicon are present in the outer
18
ACCEPTED MANUSCRIPT
301 layer), as suggested in the literature [14,15,18]. This BOF slag was found to contain about 1.3%
302 calcium hydroxide, 3.0% calcium carbonate, and 3.4% free calcium oxide [6,7]. The interior of
303 the BOF slag appears to be composed of a calcium- and iron-based matrix (possibly
304 srebrodolskite) with areas of calcium silicates (larnite and tricalcium silicate) and iron- and iron-
PT
305 magnesium-based phases (wüstite and magnesioferrite), which can be deduced from the EDXS
RI
306 maps in Figure 8.
307
U SC
AN
M
D
(a) (b)
TE
C EP
AC
(c) (d)
308 Figure 7. Compositional BSE micrographs of the interior (a) and edge (b-d) of a single BOF slag
309 particle.
310
19
ACCEPTED MANUSCRIPT
PT
RI
(a) BSE (b) Ca
U SC
AN
M
(c) Si (d) Fe
D
TE
C EP
AC
(e) C (f) Mg
311 Figure 8. Compositional BSE (a) and EDXS maps (b-f) of the edge of a BOF slag particle. The
312 black phase along the left hand side in the BSE micrograph is epoxy. Note: EDXS images have
313 been contrast-enhanced to better indicate the elemental density.
314
315
20
ACCEPTED MANUSCRIPT
316 The EAF/LMF slag (SFS3) appears to exhibit different microstructural behavior, as
317 shown in Figure 9. Both the interior and the edge of the particle can consist of a microstructure
318 that is similar to the EAF slag (Figure 9a,d), though not all sections of the edge appeared to
319 exhibit the same microstructure (Figure 9e). However, sections of the interior of the particle also
PT
320 exhibited large circular pores with some smaller pores (Figure 9b) as well as substantial porosity
RI
321 (Figure 9c). The large circular pores were also found to exist along the edge of the particle
322 (Figure 9f), while other sections of the particle were found to be significantly porous at the
SC
323 interior as well as at the edge (Figure 9c,g,h). While not consistent throughout the entire particle,
324 sections of the edges also showed evidence of the outer layer of different composition (Figure
325
U
9e-h), similar to what was seen in the BOF slag. The estimated porosities by image analysis are:
AN
326 0.8% for Figure 9a, 3.8% for Figure 9b, 17% for Figure 9c, 0.1% (interior) and 24% (edge) for
327 Figure 9e, 16% (interior) and 44% (edge) for Figure 9f, and 23% (interior) and 34% (edge) for
M
328 Figure 9g. In examining the EDXS elemental maps (Figure 10), the solid composition contains a
D
329 calcium silicate (larnite) matrix with an iron-rich phase (wüstite) and a calcium aluminate phase
TE
330 (mayenite). The outer layer is rich in calcium and aluminum along with some silicon and carbon,
331 suggesting the presence of calcium carbonate, calcium hydroxide, calcium carboaluminate
EP
332 hydrate, and/or calcium silicate hydrate phases [14,15,18]. This EAF/LMF slag was found to
333 contain about 0.2% calcium hydroxide, 0.9% calcium carbonate, and 0.4% free calcium oxide
C
334 [6,7].
AC
335
336
21
ACCEPTED MANUSCRIPT
PT
RI
SC
(a) (b) (c)
U
AN
M
D
TE
(d) (e) (f)
Figure 9. Compositional BSE micrographs
of an EAF/LMF slag particle showing the
EP
solid interior (a), interior with some pores
(b), porous interior (c), edge with
composition similar to the solid interior
C
22
ACCEPTED MANUSCRIPT
PT
RI
(a) BSE (b) Ca
U SC
AN
M
(c) Si (d) Fe
D
TE
C EP
AC
(e) C (f) Al
337 Figure 10. Compositional BSE (a) and EDXS maps (b-f) of the edge of an EAF/LMF slag
338 particle. The black phase along the left hand side in the BSE micrograph is epoxy. Note: EDXS
339 images have been contrast-enhanced to better indicate the elemental density.
340
342 single-phase microstructure with some porosity. The white interstitial phase seen in the
23
ACCEPTED MANUSCRIPT
343 microstructure was reasoned to be silica, based on evidence from EDXS. In contrast to some of
344 the SFS types, the dolomite does not exhibit a weakened edge, i.e., higher porosity or a different
346
PT
RI
U SC
AN
(a) (b)
347 Figure 11. Compositional BSE micrographs of the interior (a) and edge (b) of a dolomite
348 aggregate.
M
349
351 The SEM analysis demonstrated that, depending on the SFS type, the ITZ porosity can be
TE
352 different from dolomite aggregates. EAF slag aggregates can produce mortar and concrete bulk
EP
353 strengths that are statistically greater than the control (dolomite), which can be reasoned to be a
354 result of the similar ITZ composition to dolomite (e.g., porosity distribution), the angularity and
C
355 rough texture of the EAF particle [43], which provides better mechanical interlock, and the
AC
356 higher modulus (stiffness) of the aggregate. In general, SFS is denser and stiffer than carbonate
357 aggregates, which, along with the higher density, is the cause for the statistically greater dynamic
358 modulus values, given that the ITZ composition was not necessarily the same for all SFS types.
359 This BOF slag was found to reduce the concrete strength at 28 days relative to dolomite
360 [6], which can be argued to be a result of: 1) the more porous ITZ, allowing for easier crack
24
ACCEPTED MANUSCRIPT
361 initiation and propagation, and 2) the outer layer on the BOF particles, which was also more
362 porous, thus also allowing for easier crack initiation and propagation. Therefore, the BOF slag
363 can be thought of as containing two ITZs, one surrounding the SFS particle that is likely to be
364 composed of reaction products (e.g., calcium carbonate, which was identified by X-ray
PT
365 diffraction) and the other formed with the cement paste. Either or both of these ITZs can be
RI
366 sources of the failure of the bulk mortar or concrete, thus yielding the reduction in strength. The
367 EAF/LMF slag exhibited some evidence of a similar porous outer layer, but the layer was not
SC
368 present everywhere. Rather the EAF/LMF slag microstructure was rather porous in some areas,
369 which is an additional source of weakness, potentially allowing for easier crack propagation
370
U
through the particle. Comparatively, EAF slag did not exhibit such an outer layer and was found
AN
371 to lead to statistically higher strengths relative to dolomite. Since the EAF/LMF slag was not
372 porous throughout the entire particle and the outer layer was not present along all edges, it is
M
373 possible that the EAF/LMF slag experiences similar bonding and mechanical interlock to the
D
374 EAF slag, thus potentially yielding higher strengths relative to dolomite.
TE
375 An additional consideration concerns the possibility of chemical reaction of the cement
376 paste with the SFS surface. While SFS mineral composition may have some similarities to
EP
377 portland cement clinker, the dicalcium silicate (larnite) that is present has been argued to be
378 nonreactive [44]. Studies have demonstrated that EAF slag is relatively inert [45], but BOF slag
C
379 may be somewhat (weakly) reactive [46,47], primarily as a result of the higher free CaO content,
AC
380 although more reactive phases such as tricalcium silicate can be present (Table 2). Other studies
381 have shown that weakly reactive SFS has to be reprocessed in order to enhance its reactivity
382 [48,49]. Regardless, a number of studies have argued that the increase in concrete strength with
383 SFS aggregates is because of such a chemical reaction [8–10], although the reacted outer layer
25
ACCEPTED MANUSCRIPT
384 has also been argued to be the cause of the decrease in concrete strength with SFS aggregates
385 [15]. Therefore, it can be argued that SFS aggregates have the potential to increase mortar or
386 concrete strength because of the rough angularity and high stiffness of the particles in addition to
387 the potential for a chemical reaction. This potential for increase in strength is counteracted by
PT
388 weaknesses, such as a greater ITZ porosity, a porous outer layer on the SFS aggregate, and/or a
RI
389 porous interior of the SFS aggregate. These weaknesses counteracted any strength increasing
390 factors such that the BOF slag aggregates produced lower concrete strengths relative to dolomite.
SC
391 The lack of such weaknesses is why the EAF slag aggregates produced higher concrete strengths
393
U
Based on the findings presented herein and those presented in the literature, it can be
AN
394 concluded that a number of factors affect the mechanical behavior of cementitious composites
395 with SFS aggregates. An increase in performance may be attributed to the roughness, angularity,
M
396 density, and stiffness of the SFS aggregate as well as potential for chemical reaction. A decrease
D
397 in performance may be attributed to a more porous ITZ, a larger ITZ width, the presence of a
TE
398 porous outer layer on the SFS aggregate, the composition of the outer layer on the SFS
400
401 4. Conclusions
C
402 Steel furnace slag (SFS) aggregates are a waste by-product consisting of dense, stiff
AC
403 particles with significant variability in chemical composition, morphology, and properties
404 depending on the steel production process. In this study, SFS aggregates from three different
405 processes – basic oxygen furnace (BOF), electric arc furnace (EAF), and EAF/ladle metallurgy
406 furnace (EAF/LMF) slags – were evaluated for their effect on microstructure development in
26
ACCEPTED MANUSCRIPT
407 cementitious composites. Relative to a dolomite aggregate source, all three SFS aggregate types
408 statistically increased the dynamic modulus in a high paste mortar. The EAF slag resulted in
409 statistically greater mortar compressive strengths at 28 days, while concrete with BOF slag
410 yielded statistically lower compressive strengths at 28 days. The reported trends are consistent
PT
411 with the variability in the literature, where SFS aggregates have been shown to increase,
RI
412 decrease, or have no effect on concrete properties.
413 Backscatter electron microscopy and quantitative image analysis were employed to
SC
414 observe changes in the composition of the SFS aggregates and its influence on the interfacial
415 transition zone (ITZ) of SFS mortars. The ITZ was more porous for the BOF and EAF/LMF slag
416
U
mortars while the EAF slag mortar had a similar porosity to the dolomite mortar. The unhydrated
AN
417 cement distributions were similar for all mortars. The dolomite mortar had a greater amount of
418 calcium hydroxide immediately near the aggregate interface, while the SFS mortars had calcium
M
419 hydroxide contents that were relatively consistent along the length of the ITZ.
D
420 In examining the microstructure of SFS aggregates, BOF slag and some sections of the
TE
421 EAF/LMF slag were found to contain a more porous outer layer with different composition than
422 the interior of the particle. This outer layer was not seen in the EAF slag. Thus, mortar with SFS
EP
423 aggregates that have this outer layer can be thought of as consisting of two ITZs, the one
424 surrounding the SFS particle and the other formed with the cement paste, both of which can
C
425 allow for easier crack initiation and propagation, and therefore, can result in a reduction in
AC
427 From the findings in this study in conjunction with previous studies, the mechanical
428 properties of cementitious composites are affected by the roughness, angularity, density, and
429 stiffness of the SFS aggregates. In addition, the potential for chemical reaction between the SFS
27
ACCEPTED MANUSCRIPT
430 aggregate and the cement paste may increase bonding and improve mechanical performance.
431 However, some SFS aggregates may result in a larger or more porous ITZ and the SFS aggregate
432 may have a porous outer layer or a porous interior, all of which may negatively affect the
PT
434
RI
435 Acknowledgements
436 The SEM and EDXS analyses were carried out in part at the Frederick Seitz Materials Research
SC
437 Laboratory Central Research Facilities, University of Illinois. The authors would like to thank
438 the National Slag Association, Tube City IMS, South Shore Slag, and Edw. C. Levy Co. for
441 References
M
442 [1] C. Shi, Steel slag - its production, processing, characteristics, and cementitious properties,
443 J. Mater. Civ. Eng. 16 (2004) 230–236.
D
444 [2] H. Motz, J. Geiseler, Products of steel slags: an opportunity to save natural resources, in:
445 G.R. Woolley, J.J.J.M. Goumans, P.J. Wainwright (Eds.), Waste Materials in
TE
449 [4] N. Balcázar, M. Kühn, J.M. Baena, A. Formoso, J. Piret, Summary report on RTD in iron
450 and steel slags: development and perspectives, Proceedings No. EUR 19066 EN,
451 European Commission, Luxembourg, 1999.
452 [5] R.J. Collins, S.K. Ciesielski, Recycling and use of waste materials and by-products in
C
455 [6] A.S. Brand, J.R. Roesler, Steel furnace slag aggregate expansion and hardened concrete
456 properties, Cem. Concr. Compos. 60 (2015) 1–9. doi:10.1016/j.cemconcomp.2015.04.006.
457 [7] A.S. Brand, J.R. Roesler, Concrete with Steel Furnace Slag and Fractionated Reclaimed
458 Asphalt Pavement, Report No. ICT-14-015, Urbana, 2014.
459 [8] D.G. Montgomery, G. Wang, Instant-chilled steel slag aggregate in concrete - strength
460 related properties, Cem. Concr. Res. Res. 21 (1991) 1083–1091.
461 [9] Y. Wang, The effect of bond characteristics between steel slag fine aggregate and cement
462 paste on mechanical properties of concrete and mortar, Mater. Res. Soc. Symp. Proc. 113
463 (1988) 301–306.
28
ACCEPTED MANUSCRIPT
464 [10] G. Wang, Properties and utilization of steel slag in engineering applications, PhD
465 Dissertation, University of Wollongong, Wollongong, Australia, 1992.
466 [11] I. Arribas, A. Santamaría, E. Ruiz, V. Ortega-López, J.M. Manso, Electric arc furnace slag
467 and its use in hydraulic concrete, Constr. Build. Mater. 90 (2015) 68–79.
468 doi:10.1016/j.conbuildmat.2015.05.003.
469 [12] B. Pang, Z. Zhou, X. Cheng, P. Du, H. Xu, ITZ properties of concrete with carbonated
470 steel slag aggregate in salty freeze-thaw environment, Constr. Build. Mater. 114 (2016)
PT
471 162–171. doi:10.1016/j.conbuildmat.2016.03.168.
472 [13] J.T. San-José, I. Vegas, I. Arribas, I. Marcos, The performance of steel-making slag
473 concretes in the hardened state, Mater. Des. 60 (2014) 612–619.
RI
474 [14] A. Coomarasamy, T.L. Walzak, Effects of moisture on surface chemistry of steel slags
475 and steel slag-asphalt paving mixes, Transp. Res. Rec. 1492 (1995) 85–95.
476 [15] M. Kawamura, K. Torii, S. Hasaba, N. Nicho, K. Oda, Applicability of basic oxygen
SC
477 furnace slag as a concrete aggregate, in: V.M. Malhotra (Ed.), Fly Ash, Silica Fume, Slag
478 and Other Mineral By-Products in Concrete, ACI Special Publication 79, American
479 Concrete Institute, Detroit, 1983: pp. 1123–1141.
480 [16] S. Mindess, J.F. Young, D. Darwin, Concrete, 2nd ed., Pearson Education, Upper Saddle
U
481 River, 2003.
482 [17] P.K. Mehta, P.J.M. Monteiro, Concrete: microstructure, properties, and materials, 4th ed.,
AN
483 McGraw-Hill, New York, 2014.
484 [18] B. Pang, Z. Zhou, H. Xu, Utilization of carbonated and granulated steel slag aggregate in
485 concrete, Constr. Build. Mater. 84 (2015) 454–467.
486 doi:10.1016/j.conbuildmat.2015.03.008.
M
487 [19] A. Elsharief, M.D. Cohen, J. Olek, Influence of aggregate size, water cement ratio and age
488 on the microstructure of the interfacial transition zone, Cem. Concr. Res. 33 (2003) 1837–
489 1849. doi:10.1016/S0008-8846(03)00205-9.
D
490 [20] ASTM C39, Standard test method for compressive strength of cylindrical concrete
491 specimens, ASTM International, West Conshohocken, 2015.
TE
492 [21] ASTM C496, Standard test method for splitting tensile strength of cylindrical concrete
493 specimens, ASTM International, West Conshohocken, 2011.
494 [22] ASTM C215, Standard test method for fundamental transverse, longitudinal, and torsional
EP
499 [24] A.S. Brand, Interfacial transition zone composition and bonding in cementitious systems
500 with asphalt-coated particles, University of Illinois at Urbana-Champaign, 2015.
AC
501 http://hdl.handle.net/2142/88969.
502 [25] K.L. Scrivener, A.K. Crumbie, P. Laugesen, The interfacial transition zone (ITZ) between
503 cement paste and aggregate in concrete, Interface Sci. 12 (2004) 411–421.
504 doi:10.1023/B:INTS.0000042339.92990.4c.
505 [26] K.L. Scrivener, E.M. Gartner, Microstructural gradients in cement paste around aggregate
506 particles, Mater. Res. Soc. Symp. Proc. 114 (1988) 77–86.
507 [27] K.L. Scrivener, Backscattered electron imaging of cementitious microstructures:
508 understanding and quantification, Cem. Concr. Compos. 26 (2004) 935–945.
509 doi:10.1016/j.cemconcomp.2004.02.029.
29
ACCEPTED MANUSCRIPT
510 [28] S. Diamond, Considerations in image analysis as applied to investigations of the ITZ in
511 concrete, Cem. Concr. Compos. 23 (2001) 171–178. doi:10.1016/S0958-9465(00)00085-
512 8.
513 [29] H.S. Wong, M.K. Head, N.R. Buenfeld, Pore segmentation of cement-based materials
514 from backscattered electron images, Cem. Concr. Res. 36 (2006) 1083–1090.
515 doi:10.1016/j.cemconres.2005.10.006.
516 [30] H.S. Wong, N.R. Buenfeld, Euclidean distance mapping for computing microstructural
PT
517 gradients at interfaces in composite materials, Cem. Concr. Res. 36 (2006) 1091–1097.
518 doi:10.1016/j.cemconres.2005.10.003.
519 [31] S. Diamond, Calcium hydroxide in cement paste and concrete – a microstructural apraisal,
RI
520 in: J. Skalny, J. Gebauer, I. Odler (Eds.), Material Science of Concrete, Special Volume:
521 Calcium Hydroxide in Concrete, American Ceramic Society, Westerville, 2001: pp. 37–
522 58.
SC
523 [32] E. Gallucci, K. Scrivener, Crystallisation of calcium hydroxide in early age model and
524 ordinary cementitious systems, Cem. Concr. Res. 34 (2007) 492–501.
525 [33] S. Mindess, The strength and fracture of concrete: The role of calcium hydroxide, in: J.
526 Skalny, J. Gebauer, I. Odler (Eds.), Material Science of Concrete, Special Volume:
U
527 Calcium Hydroxide in Concrete, American Ceramic Society, Westerville, 2001: pp. 143–
528 154.
AN
529 [34] Federal Highway Administration, User Guidelines for Waste and Byproduct Materials in
530 Pavement Construction, Report No. FHWA-RD-97-148, Washington DC, 1997.
531 [35] G.C. Wang, The Utilization of Slag in Civil Infrastructure Construction, Elsevier,
532 Amsterdam, 2016.
M
533 [36] R.S. Kalyoncu, Slag—Iron and Steel, in: Miner. Yearb., United States Geological Survey,
534 Reston, 2001.
535 [37] S. Diamond, J. Huang, The ITZ in concrete - a different view based on image analysis and
D
538 [38] A.K. Crumbie, Characerisation of the microstructure of concrete, PhD Dissertation,
539 University of London, 1994.
540 [39] D. Bonen, Calcium hydroxide deposition in the near interfacial zone in plain concrete, J.
EP
30
ACCEPTED MANUSCRIPT
556 as construction material: Expansive compounds, Cem. Concr. Res. 34 (2004) 1881–1888.
557 doi:10.1016/j.cemconres.2004.01.029.
558 [46] P.Y. Mahieux, J.E. Aubert, G. Escadeillas, M. Measson, Quantification of Hydraulic
559 Phase Contained in a Basic Oxygen Furnace Slag, J. Mater. Civ. Eng. 26 (2014) 593–598.
560 doi:10.1061/(ASCE)MT.1943-5533.0000867.
561 [47] E. Belhadj, C. Diliberto, A. Lecomte, Properties of hydraulic paste of basic oxygen
562 furnace slag, Cem. Concr. Compos. 45 (2014) 15–21.
PT
563 doi:10.1016/j.cemconcomp.2013.09.016.
564 [48] J.N. Murphy, T.R. Meadowcroft, P.V. Barr, Enhancement of the cementitious properties
565 of steelmaking slag, Can. Metall. Q. 36 (1997) 315–331. doi:10.1179/cmq.1997.36.5.315.
RI
566 [49] L. Muhmood, S. Vitta, D. Venkateswaran, Cementitious and pozzolanic behavior of
567 electric arc furnace steel slags, Cem. Concr. Res. 39 (2009) 102–109.
568 doi:10.1016/j.cemconres.2008.11.002.
SC
569
570 Appendix
U
571 Examples of BSE images of the ITZ are shown in Figure A.1, detailing the relevant
AN
572 features. The SFS particle is highly visible as a large white feature because of the high iron
573 content. The UH cement particle are also visible as white features. CH is light grey and porosity
M
574 is black. Any intermediate greyscale between porosity and CH is considered C-S-H and other
575 hydration products. The outer layer, as discussed in Section 3.4 for the BOF and EAF/LMF
D
576 slags, is also visible and labeled in Figure A.1. This outer layer, while of similar greyscale
TE
577 characteristics to sections of the ITZ, could be segmented from the rest of the ITZ because it was
EP
578 continuous.
579
C
AC
31
ACCEPTED MANUSCRIPT
PT
RI
SC
(a)
U
AN
M
D
TE
EP
(b)
580 Figure A.1. Example BSE images of the ITZ for (a) BOF slag at 1 day and (b) EAF/LMF slag at
C
581 7 days. The interior (I) and outer layer (OL) of the SFS are detailed as well as the approximate
582 ITZ width. Examples of porosity, UH, and CH are also highlighted.
AC
583
32