Mechanical behaviour of Portland cement mortars with incorporation of

Al-containing salt slags

D.A. Pereira
a
, Barroso de Aguiar
b
, F. Castro
c
, M.F. Almeida
d
, J.A. Labrincha
e,
*
a
Dep. Eng. Meca nica, Instituto Superior de Engenharia, 4200 Porto, Portugal
b
Dep. Eng. Civil, Universidade do Minho, 4810 Guimaraes, Portugal
c
Dep. Eng. Meca nica, Universidade do Minho, 4810 Guimaraes, Portugal
d
Dep. Eng. Metalurgica, Universidade do Porto, LEPAE, 4000 Porto, Portugal
e
Dep. Eng. Cera mica e do Vidro, Universidade de Aveiro, UIMC, 3810-193 Aveiro, Portugal
Received 19 November 1999; accepted 28 March 2000
Abstract
Recovery of a salt cake slag produced from aluminum scrap re-melted in rotary furnaces, at the lowest cost, while complying with all
environmental laws is the main objective of this study. The characterisation of the slag residue involved determinations of the water leaching
conditions, toxicological analysis, chemical analysis of the leaching waste, and mineralogical composition. A two-phase study was made to
determine the effect of waste additions on mechanical properties of cement mortars, involving partial replacements of either sand or cement.
SEM characterisation of fracture surfaces was also made. Partial substitution of cement, which is the most expensive component of the
mortar, by salt slag has both environmental and economical advantages. However, only small amounts (up to 10 wt.%) can be replaced by
washed slag without significant damage to mortar characteristics. Sand substitution by slag is less problematic and 3050 wt.% replacement
levels are easily achieved. However, the direct economical impact is less relevant. The incorporation of unwashed slag in impracticable, as a
result of noxious gases released and significant volumetric expansion effects. D 2000 Elsevier Science Ltd. All rights reserved.
Keywords: Waste management; Mortar; Portland cement; Salt slag; Mechanical properties
1. Introduction
During the recovery of aluminum through scrap recy-
cling, a great quantity of oxide layer forms on the surface
of molten metal, which has to be removed from the melt to
refine the quality of the final product. In order to promote
an easier separation and to avoid the formation of addi-
tional oxidation phases, it is common to use a protective
salt cover. The molten salt flux promotes the coalescence of
suspended metal droplets and helps separate the clean metal
from oxide contamination. The scum formed on the surface
of molten metal is made of a mixture of oxides, fluxes,
gases, and some free metal. Its common name is aluminum
salt slag.
Fluxing techniques and production practices strongly
change throughout different industries giving several types
of dross, normally called white and black. White dross is
produced from a melting process of strongly pure scraps,
having high aluminum metal content. In this case, fluxing
inside the furnace is practically absent and the color of dross
skimmed is gray or metallic white. Black dross is produced
by secondary aluminum smelters, by using rotary furnaces
to melt old casting and a low-grade aluminum scrap. Fluxes
are composed of mixtures of sodium and potassium chlor-
ides, and contain small amounts of fluorides. At high
molten-metal temperatures, the flux melts and becomes
dark. This slag is highly salt-rich (5070%) and is the
object of our study.
The Environmental Protection Agency of the US (EPA)
classifies these residues as toxic and hazardous wastes [1].
Even the poorer aluminum-rich residue is very prone to
leaching chlorides when in contact with rainfall or ground-
water. Moreover, salt slag evolves gases when in contact
with water, namely hydrogen, ammonia, methane, and
* Corresponding author. Tel.: +351-234-370250; fax: +351-234-
425300.
E-mail address: [email protected] (J.A. Labrincha).
0008-8846/00/$ see front matter D 2000 Elsevier Science Ltd. All rights reserved.
PII: S0008- 8846( 00) 00272- 6
Cement and Concrete Research 30 (2000) 11311138
small amounts of hydrogen sulfide and phosphine. Some
of these gases are explosive, poisonous, and might pro-
duce extremely unpleasant odors [2]. The correct disposal
of these wastes causes environmental concerns. The treat-
ment of salt cake to minimize land-filling problems is
generally done in Europe, US and Canada. Typically, a dry
separation process of the aluminum, followed by leaching
and filtration steps to separate the soluble flux salts from
the insoluble oxides, and crystallization to regenerate the
salt fluxes in usable form are the main steps of that
treatment. Thus, the waste material is separated into a
suitable melted metallic fraction, a reusable salt flux and
an inert oxide mainly composed of Al
2
O
3
and a spinel
which has several possible industrial applications including
the construction sector, to improve the mechanical strength
of mortar and concrete.
2. Experimental
2.1. Characterisation of the residue
A salt slag from a Portuguese aluminum-alloy produ-
cing industry (Alpor, Braga) was used in this work. The
adaptability of the residue to the washing by water process
was studied first by doing leaching tests in several con-
ditions and by measuring the compositions of the insolu-
ble part and of the solution [3]. The correct sampling of
the slag was obtained by following normalised methods
[4]. Grain size distribution of the washed residue was
made by sieving and the following result was obtained:
>1.40 mm = 11%; 0.51.40 mm = 7%; 0.180.50 mm =
19%; <0.18 mm = 60%. Chemical composition of each
sieved fraction was estimated by X-ray fluorescence (see
Table 1). Mineralogical determinations were also obtained
by X-ray diffraction.
2.2. Incorporation in cement mortars
These tests are contained in a more embracing study
about the economical viability of the development of a
correct technology for treatment of salt slag containing
insoluble aluminum oxides. In this case, the search for an
alternative way to obtain an alumina-rich product that could
be economically interesting to be incorporated in construc-
tion materials, like in cement mortars, is the central goal of
the work.
Mortar preparation was conducted according to the
EN196-1 procedure [5]. Two sets of experiments were done,
one involving the substitution of sand by the residue and
another where cement by slag substitution was promoted.
Flexural and compression strengths were measured on
prismatic samples, having dimensions of 40 40 160
mm. These measurements were carried out after curing
periods of 3, 7, 28, and 90 days in a standard testing
machine (Autograph AG-A, Shimadzu, Japan). According
to standard procedures [5], the arithmetic mean of six tests
was determined.
Table 1
Chemical analysis (XRF) of different sieved fractions of the washed residue (% or ppm)
Retained fractions by sieving (mm)
Component 1.00 0.71 0.50 0.355 0.25 0.18 0.125 0.09 0.063 <0.063
NaCl 0.9 0.8 0.7 0.8 0.6 0.5 0.3 0.4 0.3 0.3
KCl 2.0 2.3 1.8 1.9 1.4 0.9 0.7 0.5 0.5 0.6
MgO 9.7 9.6 10.0 9.9 10.1 10.2 10.2 10.4 10.4 10.0
Al
2
O
3
60.6 62.4 62.2 61.4 60.4 61.8 65.3 70.9 69.7 69.6
SiO
2
8.4 6.6 6.5 5.7 5.8 4.3 4.1 4.3 4.3 4.4
P
2
O
5
0.33 0.40 0.35 0.40 0.36 0.40 0.34 0.34 0.34 0.32
S* 1980 2070 1780 1960 1730 1780 1470 1820 1830 1610
CaO 7.1 7.3 7.1 7.0 6.6 5.8 6.3 5.1 5.7 6.0
TiO
2
4.9 6.4 6.6 8.4 10.4 12.1 9.2 4.7 5.2 5.1
V
2
O
5
* 290 390 470 370 430 440 400 240 290 330
Cr
2
O
3
* 1590 1220 1790 2370 2150 1500 1410 1240 1310 1180
MnO* 1570 1360 1430 1250 1300 1110 910 890 900 960
Fe
2
O
3
2.6 2.8 3.0 2.7 2.6 2.2 2.0 1.9 1.9 2.0
Ni* 170 170 190 250 150 130 190 210 170 240
Cu* 23,800 8100 9500 8300 8300 6300 5700 4700 5100 6000
Zn* 1270 1480 1300 960 910 690 550 490 540 590
Sr* 150 160 160 170 160 130 110 90 100 100
ZrO
2
* 460 240 640 690 450 460 490 300 330 370
Sn* 270 350 220 170 150 170 160 150 150 170
BaO* 860 1050 1170 1470 1670 1720 1980 1680 1860 1810
PbO* 1730 2210 2030 2190 2170 2290 2190 2080 2070 2150
Other* 157 142 190 233 99 304 234 200 176 121
* Results in parts per million (ppm).
D.A. Pereira et al. / Cement and Concrete Research 30 (2000) 11311138 1132
The composition of non-slag-containing mortars was also
normalised [6] and were obtained by mixing 1350 g of
standard sand (silica-rich) with 450 g of cement (Portland
type I, class 42.5) and 225 ml of water. Several mixtures
containing different weight percentages of previously
washed residue were prepared (see Table 2). The high water
demand of the slag led to the preparation of mixtures that
contain relatively higher amount of water to get enough
workability. Different tests involving coarse (up to 2 mm)
and fine particle fractions (<180 mm) were made.
The curing process was also conducted in normalised
conditions with mortar samples immersed in water and
placed inside a humidity-controlled chamber.
3. Discussion of results
As an attempt to define optimal washing conditions, the
following laboratory-scale sequence was tested: (i) a sample
of milled and dry salt slag was stirred for 1 h in a glass
beaker with deionised water corresponding to the solid/
liquid ratio of 1:2; (ii) a clear liquid sample was extracted
and its density and amount of dissolved salts were deter-
mined by pycnometry and evaporation methods, respec-
tively; (iii) these salts were redissolved into deionised water
that was added to the slag sample in the beaker to keep
constant the solid/liquid ratio (1:3); (iv) steps (ii) and (iii)
were repeated for 1:4, 1:5, 1:6, 1:7, and 1:8 solid/liquid
ratios. Fig. 1 shows the density evolution of the liquid in
contact with salt slag as well as the percentage of salt slag
dissolved. Once the 1:5 washing step was reached, almost
no more salt slag was dissolved.
To achieve optimal levels of washing efficiency when
using cold water, another sequential test was done, starting
from a suspension made by the solid/liquid ratio of 1:4.
The salt slag was then added to reach final ratios of 3:4
and both liquid density and amount of dissolved salt were
determined (Fig. 2).
Leaching treatment was conducted by dissolving 50
60% of the as-received material, which proved to be
very effective in the removal of alkaline salts (almost
complete). Leached waste is very dark and is mainly
composed of aluminum oxides, hydroxides and silicates,
and residual chlorides. Saturated brine contains about
300 g/l of sodium/potassium chlorides and its density
is about 1.2. Leaching tests of washed slags allowed
their classification as non-toxic materials, while the as-
received residues were highly toxic. The inertness char-
acter of the treated slag allows its use for different
applications [7,8].
Salt slag is initially composed of a heterogeneous mass
in which two main mineral groups were found: (i) water-
soluble compounds, mainly alkaline salts (NaCl and KCl)
but also traces of AlN, AlC
3
and Al
2
S
3
; (ii) non-soluble
compounds in water, mainly corundum (a-Al
2
O
3
), Al
2
S,
a spinel of the system MgOAl
2
O
3
, quartz, and b-
CaOSiO
2
. The main crystalline phases present in the
fine fraction are given in Fig. 3. TGA analysis showed
that the material is non-reactive above nearly 600C,
while under this temperature, strong decomposition reac-
tions are observed due to the presence of hydroxides and
organic matter.
It was found impossible to make cement mortars
containing non-washed slags due to the volumetric
expansion observed during setting. The chloride content
is significant and high concentrations of noxious gases
are released.
The dependence of the mechanical resistance of samples
as a function of slag amount, the setting time and the
average grain size of the residue is shown in Figs. 46
Fig. 1. Dissolution of salt slag having different densities as a function of the
water added.
Fig. 2. Characteristics of the washing solution of the salt slag.
Table 2
Compositions of tested mortars involving slag fractions under 180 mm
Amount of components (g)
Sand Cement Slag Water
Sand/slag 0 1350 450 0 225
10 1305 450 45 225
20 1260 450 90 230
30 1215 450 135 250
Cement/slag 10 1350 405 45 225
30 1350 315 135 225
50 1350 225 225 250
70 1350 135 315 260
Type of
substitution
Percentage
slag
D.A. Pereira et al. / Cement and Concrete Research 30 (2000) 11311138 1133
for samples in which residue for sand substitution was
promoted. The addition of coarse slag grains promotes
deleterious effects on the mechanical properties of the
mortars. Samples containing 20% and 30% of slag seem
to be weaker than the others (Fig. 4), but the differences
tend to decrease for longer curing periods. After 28 days, the
resistance of all samples tends to be more similar. This
observation is an indication of the retarding action of the
sludge, probably due to its acidic nature and incomplete
correction of pH.
On the contrary, the addition of small amounts (up to
30 wt.%) of washed fine-grained slag did not decrease the
mechanical properties of the mortars (Fig. 5). The good
dispersion of these smaller particles through the cement
matrix might assure the formation of a suitable micro-
structure. Therefore, the slag cannot be considered as a
Fig. 4. Dependence of the mechanical resistance of mortars on the curing period and on added amount of washed coarse salt slag (<2 mm) to partially substitute
the sand fraction: (a) compression and (b) flexural strengths.
Fig. 3. XRD pattern of the powder fraction under 180 mm. (1) a-Al
2
O
3
, (2)
MgAl
2
O
4
, (3) CaF
2
, (4) Al
2
Si
4
O
10
.
D.A. Pereira et al. / Cement and Concrete Research 30 (2000) 11311138 1134
totally inactive additive. Moreover, the previous slag
washing treatment is certainly more efficient by using
finer particles and ageing effects on the materials proper-
ties are less probable.
A direct comparison of samples containing coarse- and
fine-grain residue additions is given in Fig. 6. The benefit
is much more evident on using fine-grained slag fractions,
especially on flexural resistance of the mortars. However,
the sieving operation enhances the treatment cost of the
residue, as it requires a previous (and efficient) drying
step and the use of grinding facilities to recover the
coarser particles.
The typical microstructures (SEM) of fracture surfaces of
mortars with or without slag incorporation (partially repla-
cing the sand) are shown in Fig. 7. The changes are less
visible, but slag-containing samples seem to be more
compact, which is in accordance with improving mechanical
characteristics previously discussed.
Fig. 5. Dependence of the mechanical resistance of mortars on the curing period and on the added amount of washed fine salt slag (<180 mm) to partially
substitute the sand fraction: (a) compression and (b) flexural strengths.
D.A. Pereira et al. / Cement and Concrete Research 30 (2000) 11311138 1135
Fig. 8 shows the mechanical behaviour of samples
where cement for slag substitutions were promoted. Both
compression and flexural strengths decrease with increas-
ing slag contents, irrespective of the curing period. This
deleterious trend was predicted and would also be expected
for any kind of cement for inert substitution, whatever the
nature of the inert material. In fact, the structural support of
these mortars is obviously given by the cement, and its
decreasing amount has a negative impact on the mechan-
ical resistance.
The maximum chloride contents for different concrete
types are given in Table 3 [9]. The 30% slag-containing
mortars, corresponding to the higher and acceptable incor-
poration limit, were tested and chloride contents (expressed
as percentage of the cement) of about 0.4% were deter-
mined. This confirms the potential of this incorporation
route for non-hazardous slag on mortar fabrication for
normal concrete.
4. Conclusions
The incorporation of washed aluminum-containing salt
slag in cement mortars seems possible within controlled
limits. For samples containing a fixed amount of cement,
the slag for sand substitution (for levels up to 30 wt.%)
did not promote deleterious effects on the mechanical
properties. At the same time, some properties benefit
was found for samples containing fine slag particles
(under 180 mm) and especially for longer curing periods
(28 or 90 days). Slag for cement substitution on samples
containing a fixed amount of sand strongly affects the
mechanical properties, and low tolerable levels are ac-
cepted (up to 10 wt.%). Even this result might be
attractive from an economical point of view, due to the
high cost of the cement. At the same time, this replace-
ment seems to be a helpful solution to partly solve the
environmental problem associated with the production of
such type of slags. The incorporation of salt slag increases
the mortar water demand, making this effect more pro-
Fig. 6. Changes in the mechanical resistance of mortars after curing for 28
days as a function of the amount and average grain size of washed salt slag
used to partially replace the sand: (a) compression and (b) flexural
strengths.
Fig. 7. Scanning electron micrographs obtained on fracture surfaces of (a)
slag-free and (b) 30 wt.% slag-containing mortars.
D.A. Pereira et al. / Cement and Concrete Research 30 (2000) 11311138 1136
nounced with the use of finer particles and for cement by
slag replacing samples.
References
[1] J.S. Viland, A secondary's view of recycling, Recycling of Metals
and Engineered Materials, Minerals, Metals and Materials Society,
1990, p. 427.
[2] I. Alfaro, R. Ballhord, The applications of aluminum-oxide obtained
from the recycling of aluminium, in: I. Alfaro (Ed.), Proceedings of the
Third International Conference on Recycling of Metals, Barcelona,
1997, pp. 405419.
[3] D.A. Pereira, F. Castro, F. Almeida, Characterization of Al-rich salt
scories, in: L.G. Rosa (Ed.), Proceedings of 8th Congress of Portuguese
Materials Society, Marinha Grande, Vol. 1, 1997, pp. 182189.
Fig. 8. Dependence of the mechanical resistance of mortars on the curing period and on the added amount of washed fine salt slag (<180 mm) to partially
substitute the cement: (a) compression and (b) flexural strengths.
Table 3
Maximum limits of chloride content in different concrete types
Type of concrete Cl percentage in cement
Normal 1
Reinforced 0.4
Pre-stressed 0.2
D.A. Pereira et al. / Cement and Concrete Research 30 (2000) 11311138 1137
[4] Association Francaise de Normalisation X 32-210, Dechets, Essai de
lixiviation, December, 1992.
[5] European Committee for Standardisation, EN 196-1, Methods of Test-
ing cement; Determination of Strength, May, 1987.
[6] European Committee for Standardisation, ENV 197-1, Cements: Defi-
nition, Composition, Specifications and Conformity Criteria: Part 1.
Usual Cements, 1992.
[7] D.M.S. Couto, J.A. Labrincha, R.F. Silva, L. Guise, F. Castro, In-
corporation of hydroxide-metal containing sludges in ceramic bricks,
Ceramica Acta, in press.
[8] D.M.S. Couto, J.A. Labrincha, R.F. Silva, F. Castro, Inertization of
metallurgical sludges in clay-based ceramics, in: C.S. Gomes (Ed.),
Proceedings 2nd International Meeting of the Mediterranean Clay
Groups, Vol. 2, 1998, pp. 214219.
[9] European Committee for Standardisation, ENV 206, ConcretePer-
formance, Production, Placing and Compliance Criteria, October, 1993.
D.A. Pereira et al. / Cement and Concrete Research 30 (2000) 11311138 1138