International Journal of Mineral Processing 149 (2016) 42–49

Contents lists available at ScienceDirect

International Journal of Mineral Processing

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

Lightweight aggregates produced by granulation of peat-wood fly ash

with alkali activator
Juho Yliniemi a,⁎, Henk Nugteren b, Mirja Illikainen a, Minna Tiainen c, Rick Weststrate b, Jouko Niinimäki a
a
Fibre and Particle Engineering, P.O. Box 4300, 90014, University of Oulu, Finland
b
Delft University of Technology, Department of Chemical Engineering, Product and Process Engineering Group, Julianalaan 136, 2628 BL Delft, The Netherlands
c
Laboratory of Inorganic Chemistry, Center for Molecular Materials, P.O. Box 3000, 90014, University of Oulu, Finland

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

Article history: This study presents a new method to produce lightweight aggregates (LWAs) by granulating peat-wood ash
Received 25 September 2014 using alkali activators. Peat-wood ash was granulated with potassium silicate and sodium aluminate in a high-
Received in revised form 14 December 2015 shear granulator to produce spherical granules. Blast furnace slag, coal fly ash and metakaolin were studied as
Accepted 15 February 2016
co-binders. A suitable liquid/solid ratio, granulation growth type, microstructure, strength, and crystal structure
Available online 17 February 2016
of granules have been determined. The granulation proceeded by induction-type growth behaviour and suitable
Keywords:
liquid/solid ratio was between 0.34 and 0.43. In the BSE images it was observed that the precursor particles were
Alkali activation embedded in a dense alumino-silicate matrix. The X-ray diffraction analysis supported the success of the alkali
Geopolymer activation. All co-binders increased the strength of the granules, but the blast furnace slag produced the strongest
Granulation granules. The study shows that by simultaneous granulation and alkali activation it is possible to increase the
Agglomeration utilisation of ash and produce valuable products.
Lightweight aggregate © 2016 Elsevier B.V. All rights reserved.
Fly ash

1. Introduction et al., 2012; González-Corrochano et al., 2009a; Huang et al., 2007),

The European Union has set a target to decrease CO2 emissions and
increase the use of renewable energy sources up to 20% by 2020
(European Commission Europe 2020 targets). Biomass, such as wood,
is considered as carbon neutral because it binds the same amount of
CO2 when growing as is released in combustion. In Finland, wood is
often co-combusted with peat in biomass burning facilities, such as
power plants and paper mills. As a by-product, 600,000 tons of peat-
wood ash is generated just in Finland every year (Emilsson, 2006).
One way to utilize ash and other waste materials is to produce light-
weight aggregates (LWAs) that can be used in lightweight concrete or in
civil engineering. Artificial LWAs made from waste by granulation, ag-
glomeration or pelletization are ecologically sound since natural aggre-
gates are saved and damaging activities of aggregate mining are
prevented. LWAs are also cost efficient because waste materials are
turned into products that can be sold. Different waste materials such
as ash (Anagnostopoulos and Stivanakis, 2009; Arslan and Baykal,
2006; Cheeseman et al., 2005a; Cheeseman and Virdi, 2005b; Gesoğlu
Abbreviations: AAM, alkali activated material; BSE, back-scattered electron; C, coal fly
ash; KSil, POTASSIUM silicate; M, metakaolin; Na-Alu, sodium aluminate; P, peat-wood fly
ash; S, ground granulated blast furnace slag.
⁎ Corresponding author.
E-mail address: juho.yliniemi@oulu.fi (J. Yliniemi).
sludge (Huang et al., 2007; González-Corrochano et al., 2009a, b), and
slag (Gesoğlu et al., 2012) have been studied as LWA raw materials. In
order to produce strong LWAs, either ordinary Portland cement or
high sintering temperature is required. Both options increase the costs
of the LWAs and cause CO2 emissions.
An interesting option to bypass the usage of cement and high-
temperature sintering is alkali activation. Alkali activation of coal fly
ash has been intensively researched during the last few decades and
has been comprehensively reviewed in a state-of-the-art report by
Provis and van Deventer (Provis and van Deventer, 2014a). Results
(Provis and van Deventer, 2014a) show that alkali-activated materials
(AAMs) offer excellent properties compared to ordinary Portland ce-
ment, such as lower CO2 emissions, excellent compressive strength,
and thermal resistance. In addition, immobilisation of heavy metals in
AAMs has been reported (Ogundiran et al., 2013; van Jaarsveld et al.,
1997, 1998; Zhang et al., 2008) which means that this way even the
fly ash fractions that contain toxic elements could be utilized.
The objective of this study was to make granules from peat-wood
ash using an alkali activator as the liquid binding phase to glue the pri-
mary particles together in a permanent alumino-silicate matrix. The in-
tention was to produce spherical granules using a high shear granulator
and appropriate granulation conditions. To the best of our knowledge,
no previous studies have used the approach described herein to pro-
duce LWAs. We studied the effect of different alkali activators and co-

http://dx.doi.org/10.1016/j.minpro.2016.02.006
0301-7516/© 2016 Elsevier B.V. All rights reserved.
 J. Yliniemi et al. / International Journal of Mineral Processing 149 (2016) 42–49 43

binders on the strength of the granules. The paper aims for a proof of
principle and focuses on the granulation mechanism and the extent of
the alkali activation reaction.

2. Material and methods

2.1. Materials

Peat-wood fly ash (P) was collected from a power plant that uses a
circulating fluidised bed boiler. The proportion of wood in the fuel mix
was approximately 30%. Ground-granulated blast furnace slag
(S) (ORCEM, the Netherlands), coal fly ash (C) and metakaolin
(M) were used as co-binders, and their effects on granulation and alkali
activation were studied. The coal ash was from a pulverised fuel power
plant. Metakaolin was obtained by thermal treatment of kaolin (Sigma
Aldrich) at 750 °C for 2 h.
Potassium silicate (KSil) solution was chosen as an alkali activator
for its known ability to produce strong AAMs, especially with coal fly Fig. 1. A schematic of the granulator. The impeller (14 cm diameter) inside spins in the
ash and slag (Nugteren et al., 2009). The solution was prepared by opposite direction to that of the drum. The volume of the drum was 13 dm3, the tilt
mixing a commercial KSil solution (Kasil 2135; PQ Europe) with potas- angle was 33°, and the drum rotating speed was 44 rpm. The binder was added by
sium hydroxide to obtain K2O/SiO2 = 0.8 (65 wt.% of H2O). Sodium alu- drops from a flask through a tube on to the powder bed.

minate (Na-Alu) was also used as an alkali activator. Na-Alu solution is

available as a waste product. If it could be used as an alkali activator, the The granulation process was carried out using the following
cost of producing AAMs could decrease. The Na-Alu (Sigma-Aldrich) so- method:
lution was dissolved in sodium hydroxide-solution to yield a total molar
ratio of Na2O/Al2O3 = 2.2 (60 wt.% of H2O). To evaluate the success of 1) Dry precursors were weighed and mixed carefully prior to the
the alkali activation, the granulation was also prepared with H2O. granulation.
2) Mixed material was added to the drum, and the drum and impeller
were switched on.
2.2. Granulation
3) Liquid was added by drops on to the powder bed rotating inside the
drum with a spray flask from the hatch of the granulator.
Nine batches with different compositions were prepared (Table 1).
4) Process was continued until the granules stopped growing.
Peat-wood fly ash was granulated using three different liquids: KSil,
5) Each batch was sealed in an air-tight plastic bag and stored at room
Na-Alu and H2O. The effects of slag, coal fly ash and metakaolin as co-
temperature until the analysis of the granules.
binders were studied by adding them in 20 wt.% and 40 wt.% propor-
tions to peat-wood fly ash. KSil was used as an alkali activator with
the co-binders. The materials were blended prior to granulation manu- 2.3. Analysis methods
ally for 30 s and then inside the granulator with an impeller for 30 s.
A high-shear granulator (Eirich R-02) was chosen for this study due The chemical composition of the precursors was determined with X-
to its known ability to spread viscous liquids, to process sticky material ray fluorescence (XRF) from a melt-fused tablet. The particle size distri-
and to produce more compact granules than low-shear granulators butions were measured with a Beckman Coulter LS 13320 and reported
(Reynolds et al., 2007). A schematic of the granulator is presented in as volumetric-based size (d10, d50 and d90).
Fig. 1. The granulator has a rotating drum with an impeller (14 cm diam- The crushing strength of the granules after 28 days was measured
eter) inside that spins in the opposite direction to that of the drum. The with a Zwick, Z100 Roell testing machine. TestXpert II-software was
volume of the drum was 13 dm3, the tilt angle was 33°, and the drum used to determine the crushing force of the granules. The pre-load
rotating speed was 44 rpm. Preliminary experiments were made to de- force was 5 N, and the compression speed was 0.01 mm/s. Similar single
termine a suitable liquid/solid-ratio (w/w) for the material. granule crushing tests for LWAs have been performed by (Arslan and
Baykal, 2006; Cheeseman et al., 2005a; Cheeseman and Virdi, 2005b;
Gesoğlu et al., 2012; González-Corrochano et al., 2009a).
Loose bulk density and the granule size distribution were deter-
Table 1 mined with standards (SFS-EN-1097-3, 1998) and (SFS-EN 933-1,
Proportions of precursors and the liquid used in granulation.
2012) respectively.
Precursors (g) A Siemens 5000 X-ray diffractometer with CuKα radiation (40 mA
Code Binder liquid P S C M
and 40 kV) and a graphite monochromator was used to identify the
main crystalline phases of the powdered samples. The step interval, in-
P100 H2O H2O 300
tegration time and angle interval used were 0.04°/step, 2.5 s/step and
P100 Na-Alu Na-Alu 300
P100 KSil KSil 300 10–70°, respectively. The ICDD database was used for identification of
P80S20 KSil 240 60 crystalline phases (The Powder Diffraction File, 2006).
P60S40 KSil 180 120 The cross-sections of three granules (5–6 mm diameter) from each
P80C20 KSil 240 60
batch were analysed with a field emission electron microscope
P60C40 KSil 180 120
P80M20 KSil 240 60 (FESEM) (Zeiss Sigma). The granules were impregnated in the epoxy
P60M40 KSil 180 120 resin. After curing for 24 h, 2 mm slices were cut and placed in a
Abbreviations: peat-wood fly ash (P), blast furnace slag (S), coal fly ash (C), metakaolin
25 mm diameter plastic mould. The slices were then impregnated in
(M), potassium silicate (KSil) and sodium aluminate (Na-Alu). The numbers in the the epoxy resin and left to cure for 24 h. Hardened samples were then
codes represent weight percentages of the precursors. grinded, polished and coated with carbon to obtain an optimal surface
44 J. Yliniemi et al. / International Journal of Mineral Processing 149 (2016) 42–49
for FESEM analysis. The sample distance was 8.5 mm, the acceleration
voltage was 15 kV, and the beam current was 9.95 nA.
3. Results and discussion
3.1. Precursor properties
The chemical composition of the precursors and their particle sizes
are presented in Table 2. The total amount of SiO2 and Al2O3 in peat-
wood ash, blast furnace slag, coal fly ash and metakaolin was approxi-
mately 56 wt.%, 44 wt.%, 77 wt.%, and 92 wt.%, respectively. Coal fly
ash and peat-wood fly ash had larger particle sizes than blast furnace
slag and metakaolin.
3.2. Granulation process
Table 3 shows the amount of solid precursors, consumed binder liq-
uid, optimum liquid/solid ratio (w/w), and loose bulk density for the dif-
ferent batches. The liquid/solid ratio varied from 0.34 to 0.43 between
different batches. Variation in the density of alkali activators causes
some differences in liquid/solid ratios when calculated as weight-to-
weight. A difference of only a few per grams in the amount of liquid re-
sulted in granules with totally different forms. Granulation did not start
at all if there were a few grams less than the optimal amount of liquid.
In cases where the amount of liquid exceeded the required amount by
a few grams, all the material combined into one big wet ball and finally
into a slurry. This phenomenon has been acknowledged in granulation
theory (Iveson and Litster, 1998; Iveson et al., 2001).
Standard (SFS-EN 13055-1, 2002) requirement for LWA is loose bulk
density of less than 1.20 g/cm3 or a particle density of less than
2.00 g/cm3. In this study, the loose bulk density was less than
1.20 g/cm3 for all batches, so it can be stated that all alkali-activated
granules were LWAs. Variations in loose bulk density between batches
are due to different densities of solid precursors and liquids.
The alkali activation reaction changes the conditions of granule for-
mation during granulation. The alkali activator dissolves reactive
alumino-silicates from the precursor particles, which then form a
silicate-aluminate gel (Krivenko, 1997; Lee and van Deventer, 2002;
Davidovits, 1988; Chen et al., 2011a, b; Fernández-Jiménez et al.,
2005). The gel hardens and binds the undissolved particles together
permanently, thereby increasing the strength of the granules.
In contrast to previous studies on alkali activation of peat-wood ash
(Tyni et al., 2012, 2014), the amount of liquid required for a proper
working consistency was lower for granule samples than for block-
type samples. This was expected because AAM does not have to be a
‘workable’ paste for granulation. The price of alkali activators is one re-
strictive factor in commercialising AAMs. The low liquid/solid ratio de-
creases the costs of alkali-activated granule product compared to block-
or brick-type product. However, maximum strength granules might not
be produced in the absence of sufficient liquid and activator for the
alkali-activation reaction. The narrow suitable liquid/solid ratio also re-
quires a carefully designed granulation process when manufacturing
granules on an industrial scale.
In general, three separate phases can be observed in the granulation
process: 1) wetting and nucleation, 2) growth and compaction and
Table 2
The chemical composition and particle sizes of precursors.
Chemical composition, %
Sample CaO SiO2 Al2O3 Fe2O3 Na2O
P 11.6 44.9 10.7 20.1 1.5
S 39.5 34.5 9.9 0.5 0.4
C 4.9 54.3 22.9 8.0 1.1
M 0.1 59.5 32.8 1.4 0.1
Abbreviations: peat-wood fly ash (P), blast furnace slag (S), coal fly ash (C) and metakaolin (M).
3) breakage and attrition (Iveson and Litster, 1998; Iveson et al., 2001;
Ennis and Litster, 1997; Butensky and Hyman, 1971). When liquid is
added to the dry material, it forms a layer on the surface of the particles.
As the moistened particles roll in the drum and collide, they start to
form bigger agglomerates. The granules grow until binding forces can
no longer hold the particles together due to their large size or drying
out. There are two main types of granule growth behaviour: steady
growth and induction growth (Iveson and Litster, 1998; Iveson et al.,
2001; Kapur, 1978; Hoornaert et al., 1998; Sherington, 1968; Capes,
1965). In the steady growth regime, the granule size grows linearly
with time. In the induction growth regime, there is little or no growth
for a long period, but this is followed by rapid growth when excess liq-
uid is squeezed to the surface due to the consolidation, enabling the co-
alescence of particles.
In the present study, the granule size growth was observed every
30 s during granulation. Although the exact diameters cannot be con-
firmed because the average size of the granules was estimated only vi-
sually with a ruler, growth behaviour was evident. It can be seen from
Fig. 2 that no granules formed for a long period (up to 10 min), despite
90% of the total liquid mass being added to all the samples after the first
7 min. Rapid growth occurred after 7–12 min of granulation. This indi-
cates induction-type growth behaviour and highlights that granulation
will not start before particles have been covered with liquid. Induction
growth suggests that the precursor particles are strong and do not de-
form enough to coalesce during impact (Iveson et al., 2001). However,
they start to consolidate over time when the alkali activator dissolves
the surface of the particles. Induction-type growth results in stronger
granules than steady-type growth, but it is more difficult to control
(Iveson and Litster, 1998; Iveson et al., 2001; Wauters, 2001).
After 8–14 min of granulation, the granules stopped increasing in
size now that all liquid had been consumed even though the granulator
was kept on for 5–10 min more. No more liquid was added since the tar-
get granule size was obtained. The total granulation time was between
10 and 20 min for all batches and the granules were surface dry. One
should be aware of the fact that due to the reaction taking place during
the granulation, liquid to solid ratio as well as the viscosity of the liquid
changes in time due to the reaction. This makes it difficult to determine
the true end point. For this reason it was chosen to take the time at the
end of the consolidation time as the final end point. Since in the 5–
10 min that the granules are in this consolidation time did not show
any change in size, the above criterion was taken.
The scale-up is not straightforward for granulation processes and
should be studied separately, especially when the material has a hetero-
geneous composition and alkali activation reaction may change the
granulation conditions. However, after the experiments made in this
study other waste materials (ashes, mine tailings and slags) have been
successfully granulated with the same granulation parameters, but
with different high shear granulator and different alkali activators [Un-
published results], which gives confidence that the findings are applica-
ble more broadly.
The granule size distribution varied from 0.125 mm to 16 mm (Ap-
pendix). The granules were mostly spherical, and the diameters of near-
ly 90% of all alkali activated granules were between 1 and 8 mm in
diameters for all batches. Therefore, the size distribution was narrow
in contrast to the possible application in civil engineering or as concrete
Particle size
K2O MgO P2O5 TiO2 SO3 b10% (μm) b50% (μm) b90% (μm)
2.1 3.0 2.6 0.4 2.0 2.4 17.4 135.6
0.3 8.1 0.0 1.1 3.4 1.0 10.3 31.9
1.7 1.8 0.7 1.2 0.8 3.0 30.2 133.7
0.6 0.1 0.0 1.9 0.0 1.0 8.0 53.1
 J. Yliniemi et al. / International Journal of Mineral Processing 149 (2016) 42–49 45

Table 3
Amount of solid precursors, consumed binder liquid, optimum liquid/solid ratio (w/w), and loose bulk density (g/cm3) for the different batches.

Code P100 H2O P100 KSil P100 Na-Alu P80 S20 P60 S40 P80 C20 P60 C40 P80 M20 P60 M40

Solid (g) 300.0 300.0 300.0 300.0 300.0 300.0 300.0 300.0 300.0
Liquid (g) 130.2 121.7 115.0 115.1 117.0 115.7 102.2 113.9 116.5
Liquid/solid ratio 0.43 0.41 0.38 0.38 0.39 0.39 0.34 0.38 0.39
Loose bulk density 0.93 1.14 0.97 1.03 1.08 0.90 1.00 0.96 1.04

Abbreviations: peat-wood fly ash (P), blast furnace slag (S), coal fly ash (C), metakaolin (M), potassium silicate (KSil) and sodium aluminate (Na-Alu).

aggregate. There was hardly any un-granulated material left in any of
the batches, except when H2O was used as liquid (P100 H2O).
3.3. Strength
The strength of the granules was measured after 28 days of curing in
sealed plastic bags at room temperature. Due to the spherical shape of
the granules, the compressive strength could not be measured. There-
fore, only the force needed to break the granules is given. As the force
required to break the granules depends on their size, 15 granules be-
tween 5 and 6 mm diameters were chosen from each batch.
Fig. 3 shows the forces required to break the samples from all
batches. For P100 H2O, only 10 N was required. The fact that all samples
granulated with alkali activators were stronger than those granulated
with water gives an indication that the alkali-activation reaction oc-
curred. Viscosity of the liquid has a major effect on the granulation pro-
cess, but does not per sé have an additional effect on the final granules
strength. If that relation is found this is due to the granulation process
(one may assume the dissolution is not much affected by the viscosity,
thus resulting in a similar strength of the solid bridge content). If the vis-
cosity is increased due to the use of polymer systems, the polymers may
give addition strength upon drying. In this case there we no polymers
added, they were formed, or at least there were solid bridges formed
(See Section 3.5) by the reaction and not only by the drying out. So it
is clear that in this case the extra strength is due to the reactions of
the activators. The KSil activator yielded higher strength granules than
the Na-Alu activator for peat-wood fly ash. Slag as co-binder produced
the strongest granules, with the breaking force being almost 500 N.
There was not much difference in the strength of the granules with
20 wt.% and 40 wt.% slag content. Coal fly ash increased the strength
of the granules slightly compared to P100 KSil. Metakaolin increased
the strength of the granules more than coal fly ash but less than slag.
With regard to metakaolin, 40 wt.% produced stronger granules than
20 wt.%. The co-binders provide more reactive material to the precursor
mixture and thus they help the binding of the fly ash particles together
by increasing the amount of alumino-silicate gel. Nonetheless, by com-
paring the strength between the granules prepared with H2O and alkali
activators it can be stated that the peat-wood fly ash is also reactive.
Cold-bonded fly ash-slag-cement granules were prepared by
(Gesoğlu et al., 2012). In contrast to their results, the strengths of the
alkali-activated granules were similar or even higher. In both studies,
the fly ash had a similar chemical composition and the measured gran-
ule size was nearly the same.
González-Corrochano et al. produced sludge-fly ash-used motor oil
(González-Corrochano et al., 2009a) and sludge-clay-rich sediment
pellets (González-Corrochano et al., 2009b) by sintering in 1150 °C–
1225 °C. They calculated the compressive strength of the granules by
the following equation:
ð2:8  FÞ
S¼ ð1Þ
π  X2
where X is the granule diameter and F is the fracture load force. As
calculated by Eq. (1), the compressive strength of alkali-activated gran-
ules varied from 3.5 MPa (P100 Na-Alu) to 14.3 MPa (P60S40) whereas
granules prepared by sintering were between 1.41 MPa and 5.64 MPa in
(González-Corrochano et al., 2009a) and 0.5 MPa–13.31 MPa in
(González-Corrochano et al., 2009b).

Fig. 2. Granule size growth vs. granulation time. The granules began to increase in size
between 6 and 12 min after 90% of the liquid had been added. The growth was rapid
which indicates induction-type growth. After 8–14 min size growth stopped.
Fig. 3. Forces required to break granules of 5–6 mm diameter. From each batch, 15
granules were measured, and the average was calculated. The error bars represent the
confidence interval for means at 95 % confidence level. Samples granulated with alkali
activators were stronger than those granulated with H2O. Using slag as a co-binder
produced the strongest granules.
46 J. Yliniemi et al. / International Journal of Mineral Processing 149 (2016) 42–49

Arslan and Baykal (Arslan and Baykal, 2006) calculated the compres- Fig. 4a present the findings for peat-wood ash and its granulation
sive strength of fly ash-cement pellets by a slightly different equation: products with H2O, Na-Alu and KSil as the binder liquids. It can be
seen that the X-ray amorphous halo is increased in P100 KSil and in
F P100 Na-Alu. However, in P100 H2O, the halo is nearly the same size
S¼π : ð2Þ
d
2 as in the peat-wood ash diffractogram. The amorphous halo is consid-
4 ered an indicator of the success of alkali activation (Criado et al., 2007;
van Jaarsveld and van Deventer, 1999; Williams and van Riessen,
In this equation, F is the fracture load force and d is the mean diam- 2010; Guo et al., 2010; Provis et al., 2005) the variations in the size of
eter of granule. Calculating by Eq. (2) and comparing the results, it is the amorphous halo correlate with the granule breaking force results.
clear that the alkali-activated granules were much stronger (from KSil produced stronger granules than Na-Alu. With H2O as the binder
4.95 MPa (P100 Na-Sil) to 17.5 MPa (P60S40)) than the fly ash- liquid, the breaking force was minimal.
cement granules (Arslan and Baykal, 2006) (from 0.871 MPa (10 wt.% The X-ray diffractogram of slag (Fig. 4b) shows only X-ray amor-
cement and 90 wt.% fly ash) to 1.423 MPa (30 wt.% cement and phous material (a large halo between 20° and 40° 2Θ). Since the XRF-
70 wt.% fly ash). analysis results in Table 2 show that slag contains mainly calcium, silica
and alumina, it can be presumed that slag consists mostly of amorphous
3.4. X-ray diffraction calcium silicates that are known to be reactive in the alkali-activation
reaction (Provis and van Deventer, 2014a; Williams and van Riessen,
Fig. 4a–d show the X-ray diffractograms for the precursors and reac- 2010). This likely explains the increase in breaking forces of P80S20
tion products. Peat-wood ash is a very heterogeneous material, and and P60S40 (Fig. 3) compared to P100 KSil. However, too straightfor-
some differences are possible between diffractograms from two sam- ward assumptions should not be made, since the correlation between
ples taken from the same ash batch. This must be taken into account the amorphous halo of the AAM and the strength of the AAM is not nec-
when comparing the signals found in different X-ray diffractograms. essarily linear since the halo may contain also unreacted material.
The identified phases in peat-wood ash were quartz (Q), maghemite Coal fly ash contained quartz, mullite and amorphous material
(M) and albite (A). X-ray amorphous material was observed between (Fig. 4c). Again, it can be seen that the size of the amorphous halo has
20° and 40° 2Θ (shown in grey). For clarity, only the phases not found grown in P80C20 and in P60C40 compared to the peat-wood ash
in the peat-wood ash diffractogram are marked to the other X-ray diffractogram. In addition, the highest point of the amorphous halo of
diffractograms. the coal fly ash is at 25° 2Θ, whereas the highest point in P80C20 and

Fig. 4. a–d. Diffractograms for precursors and reaction products. a) Peat-wood ash, P100 H2O, P100 Na-Alu and P100 KSil; b) slag, peat-wood ash, P80S20 and P60S40; c) coal ash, peat-
wood ash, P80C20 and P60C40; and d) metakaolin, peat-wood ash, P80M20 and P60M40. Abbreviations: quartz (Q), maghemite (M) albite (A), Mu, mullite; An, anatase. For clarity, only
the phases that were not found in the peat-wood ash diffractogram are marked into other diffractograms. The X-ray amorphous halo is shown in grey to highlight the change between the
precursors and products.
 J. Yliniemi et al. / International Journal of Mineral Processing 149 (2016) 42–49 47

P60C40 is at 30° 2Θ. This shift from 25° to 30° 2Θ indicates the success of
alkali activation and is discussed in more detail in (Provis et al., 2005).
The shift was not as clearly observable in the granules containing slag
since the slag had a high content of X-ray amorphous material in the al-
ready prior to alkali activation. The diffraction signals of maghemite are
clearly observable only in the peat-wood ash and coal ash, which corre-
lates with the XRF-analysis results (Table 2). These signals are explained
by only the peat-wood ash and coal ash containing a significant amount
of iron.
Quartz, anatase, muscovite and amorphous material were identified
in the metakaolin diffractogram (Fig. 4d). As shown in the figure, there
is a shift in the centre of the amorphous hump from 25° to 30° 2Θ.

3.5. Field emission scanning electron spectroscopy

Figs. 5 and 6 show back-scattered electron (BSE) images of cross-

sections of the P100 H2O, P100 KSil, P100 Na-Alu, P80S20, P80C20 and
P80M20 samples. Angular and spherical particles from peat-wood ash
can be seen in all the BSE-images. The peat-wood ash particles seem
to be bound together mainly by physical interlocking in P100 H2O. In
P80S20, angular blast furnace slag particles are observable (marked
with arrows). More round particles originating from the coal fly ash
can be seen in P80C20.
The angular shape and small particle size of slag can enhance granu-
lation (Linkson et al., 1973; Newitt and Conway-Jones, 1958; Rumpf,
1962). On the other hand, metakaolin has plate-shaped particles
(Provis et al., 2010), which may inhibit the granulation process and
thus yield in lower strength.
There is a distinct difference between the samples granulated with
H2O and those granulated with KSil or Na-Alu (Figs. 5 and 6). P100
H2O shows scattered and loose material, with large pores. The alkali-
activated granules are denser than the P100 H2O granules, and all parti-
cles are embedded in a grey matrix. There are fewer pores in P80S20
and P80M20 than in P100 KSil, P100 Na-Alu and P80C20 (Fig. 5). The
Fig. 5. BSE images of P100 H2O (×110), P100 KSil (×114), P100 Na-Alu (×110), P80S20
(× 110), P80C20 (×103) and P80M20 (1 ×10) samples. The images show the general
appearance of the samples. The boundaries of the granules can be seen on the left in the
images.
Fig. 6. BSE-images with × 1000 magnification of P100 H2O, P100 KSil, P100 Na-Alu,
P80S20, P80C20 and P80M20 samples. The images show the general appearance of the
samples. Arrows show the slag particles. Alkali-activated samples have a denser matrix
in contrast to P100 H2O samples.
low porosity of P80S20 and P80M20 correlates with the breaking force
results, samples with slag and metakaolin also had highest strengths.
So to conclude, the highest strength obtained with slag as co-binder
can be attributed to the combined effect of more suitable particle
shape, higher amount of X-ray amorphous material and lower porosity
of the granules.
In ordinary Portland cement concrete and concrete-type AAMs, the
porosity of the product is an important factor affecting its applications
(Provis et al., 2010; Bernal, 2014). If the porosity is (too) high, there is
a danger of corrosion of steel reinforcements when harmful ions and
H2O ingress the material. With granule-type products, low porosity is
not required because there is no steel present to corrode. However, if
granules are used in civil engineering construction and concrete appli-
cations, possible reactions inside the granules due to a change of pH or
H2O penetration must be considered. In addition, if the fly ash contains
hazardous components, these could leach out if the porosity is high
(Provis and van Deventer, 2014b, c).
Based on these results, a descriptive diagram modelling the
granulation-alkali activation process is shown in a schematic way in
Fig. 7. The upper left corner (Fig. 7a) shows the addition of the alkali ac-
tivator to the dry ash precursor particles. During the granulation, the
particles are wetted (Fig. 7b). Alkali-activator dissolves reactive materi-
al from the surfaces of appropriate particles (Fig. 7c), and an alumino-
silicate gel starts to form (Fig. 7d), as shown in the studies by
Fernández-Jiménez and Chen (Fernández-Jiménez et al., 2005; Chen
et al., 2011a, b). With alumino-silicate gel on the surface, particles coa-
lesce during granulation and bind together strongly (Fig. 7e). This pro-
cess results in spherical granules that are formed of precursor particles
bound in an alumino-silicate matrix (Fig. 7f). Since the precursors take
part in the formation of the alumino-silicate gel there is no clear border
between the precursor particles and the binding matrix. However, as
occurs in OPC concrete between the cement and the aggregates, also
here could have formed an interfacial transition zone with different
chemical composition than the cementing geopolymeric matrix.
48 J. Yliniemi et al. / International Journal of Mineral Processing 149 (2016) 42–49
Fig. 7. Descriptive model of the granulation-alkali activation process. The model shows (a) the addition of the alkali activator to the precursor particles. During the granulation, the particles
are wetted by the alkali activator (b). The reactive material dissolves (c) and forms an alumino-silicate gel (d), which binds the particles together (e). The process results in spherical
granules (f).
4. Conclusion
In the present study, a new method to produce LWAs is presented. It
was shown that by granulating peat-wood fly ash with an alkali-
activator it is possible to produce strong LWAs. The granulation resulted
in a relatively narrow size distribution with spherical granules. The
granulation mechanism was induction type. The most suitable liquid/
solid ratio was between 0.34–0.43, regardless of the composition of
the precursor or the binder liquid. The use of KSil rather than Na-Alu
as an activator resulted in higher strength granules. All co-binders in-
creased the strength of the granules, but the strongest granules were
obtained with slag. The peat-wood contained crystalline and amor-
phous material. An amorphous halo between 20° and 40° 2Θ was ob-
served in the X-ray diffractograms of the alkali-activated samples,
supporting the success of the alkali activation. In the BSE images it
was observed that the precursor particles embedded in a dense
alumino-silicate matrix.
The production of the strongest granules with slag as a co-binder is
likely due to it having a suitable particle size distribution, which en-
hanced packing. It was also the most reactive co-binder in alkali
activation.
Granulation of peat-wood ash by alkali activators can increase the
utilisation of ash and generate valuable products that could be used
in civil engineering or lightweight concrete. More generally, the
manufacture of artificial aggregates from waste not only helps save
natural resources, but offers an ecological solution to waste manage-
ment. However, the bioavailability of hazardous components in
granules should be studied before large scale usage becomes
applicable.
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.minpro.2016.02.006.
Acknowledgements
This work was carried out under the auspices of the “GEOPO” project
supported by the Finnish Funding Agency for Technology and Innova-
tion (Tekes) and various companies (Boliden Harjavalta Oy, Ekokem-
Palvelu Oy, Fortum Power and Heat Oy, Helen Oy, Jyväskylän Energia
Oy, Kemira Chemicals Oy, Valmet Power Oy, Metsä Group, Napapiirin
Energia ja Vesi, Stora Enso Oyj, and UPM). The authors would like to
thank the Renlund Foundation for its financial support. Special thanks
to Dr. Gabrie Meesters for his valuable advices about granulation. Mr.
Jarno Karvonen and Mr. Jani Österlund are acknowledged for their con-
tributions to the laboratory analyses.
References
Anagnostopoulos, I.M., Stivanakis, V.E., 2009. Utilization of lignite power generation resi-
dues for the production of lightweight aggregates. J. Hazard. Mater. 163, 329–336.
Arslan, H., Baykal, G., 2006. Utilization of fly ash as engineering pellet aggregates. Environ.
Geol. 50, 761–770.
Bernal, S., 2014. Durability and testing — degradation via mass transport. In: Provis, J., van
Deventer, J. (Eds.), Alkali Activated Materials — State-of-the-art Report, RILEM TC
224-AAM 13. Springer Netherlands, pp. 223–276. http://dx.doi.org/10.1007/978-94-
007-7672-2.
Butensky, M., Hyman, D., 1971. Rotary drum granulation. An experimental study of fac-
tors affecting granule size. Ind. Eng. Chem. Fundam. 10, 212–219.
Capes, C., 1965. Granule formation by the agglomeration of damp powders: Part 1. The
mechanism of granule growth. Trans. Inst. Chem. Eng. 43, 116–124.
Cheeseman, C.R., Makinde, A., Bethanis, S., 2005a. Properties of lightweight aggregate pro-
duced by rapid sintering of incinerator bottom ash. Resour. Conserv. Recycl. 43,
147–162.
Cheeseman, C.R., Virdi, G.S., 2005b. Properties and microstructure of lightweight aggre-
gate produced from sintered sewage sludge ash. Resour. Conserv. Recycl. 45, 18–30.
Chen, C., Gong, W., Lutze, W., Pegg, I.L., 2011a. Kinetics of fly ash geopolymerization.
J. Mater. Sci. 46, 3073–3083.
Chen, C., Gong, W., Lutze, W., Pegg, I.L., Zhai, J., 2011b. Kinetics of fly ash leaching in
strongly alkaline solutions. J. Mater. Sci. 46, 590–597.
Criado, M., Fernández-Jiménez, A., de la Torre, A.G., Aranda, M.A.G., Palomo, A., 2007. An
XRD study of the effect of the SiO2/Na2O ratio on the alkali activation of fly ash.
Cem. Concr. Res. 37, 671–679.
Davidovits, J., 1988. Geopolymer properties and chemistry. 1st European Conference on
Soft Minerallurgy, Geopolymer; France, Compiengne, pp. 25–48.
European Commission Europe 2020 targets: http://ec.europa.eu/europe2020/targets/eu-
targets/ (last visited 23.7.2014)
Emilsson, S., 2006. International handbook: from extraction of forest fuels to ash
recycling. Skogsstyrelsen. Swedish Forest Agency.
Ennis, B., Litster, J., 1997. Particle Size Enlargement. In: Perry, R., Green, D. (Eds.), Perry´s
Chemical Engineers´ Handbook, seventh ed., pp. 20–89 (New York).
Fernández-Jiménez, A., Palomo, A., Criado, M., 2005. Microstructure development of
alkali-activated fly ash cement: a descriptive model. Cem. Concr. Res. 35, 1204–1209.
Gesoğlu, M., Güneyisi, E., Öz, H.Ö., 2012. Properties of lightweight aggregates produced
with cold-bonding pelletization of fly ash and ground granulated blast furnace slag.
Mater. Struct. 45, 1535–1546.
González-Corrochano, B., Alonso-Azcárate, J., Rodas, M., 2009a. Characterization of light-
weight aggregates manufactured from washing aggregate sludge and fly ash. Resour.
Conserv. Recycl. 53, 571–581.
González-Corrochano, B., Alonso-Azcárate, J., Rodas, M., 2009b. Production of lightweight
aggregates from mining and industrial wastes. J. Environ. Manag. 90, 2801–2812.
Guo, X., ShI, H., Dick, W.A., 2010. Compressive strength and microstructural characteris-
tics of class C fly ash geopolymer. Cem. Concr. Compos. 32, 142–147.
Hoornaert, F., Wauters, P.A.L., Meesters, G., Pratsinis, S., Scarlett, B., 1998. Agglomeration
behaviour of powders in a Lödige mixer granulator. Powder Technol. 96, 116–128.
Huang, S.-C., et al., 2007. Production of lightweight aggregates from mining residues,
heavy metal sludge, and incinerator fly ash. J. Hazard. Mater. 144, 52–58.
Iveson, S., Litster, J., 1998. Growth regime map for liquid-bound granules. AIChE J 44 (7),
1510–1518.
Iveson, S.M., Wauters, P.A., Forrest, S., Litster, J.D., Meesters, G.M., Scarlett, B., 2001.
Growth regime map for liquid-bound granules: further development and experimen-
tal validation. Powder Technol. 117, 83–97.
van Jaarsveld, J., van Deventer, J., Lorenzen, L., 1997. The potential use of geopolymeric
materials to immobilise toxic metals: Part I. Theory and applications. Miner. Eng. 10
(7), 659–669.
 J. Yliniemi et al. / International Journal of Mineral Processing 149 (2016) 42–49
van Jaarsveld, J., van Deventer, J., Schwartzman, A., 1998. The potential use of
geopolymeric materials to immobilise toxic metals: Part II. Material and leaching
characteristics. Miner. Eng. 12 (1), 75–91.
van Jaarsveld, J.G.S., Van Deventer, J.S.J., 1999. Effect of the alkali metal activator on the
properties of fly ash-based geopolymers. Ind. Eng. Chem. Res. 38, 3932–3941.
Kapur, P., 1978. Balling and granulation. Adv. Chem. Eng. 10, 55–123.
Krivenko, P., 1997. Alkaline cements: terminology, classification, aspects of durability.
Proc of 10th Int Cong Chem Cem. Göteborg, Sweden (4iv046-4iv050).
Lee, W., van Deventer, J., 2002. Factors affecting the setting behaviour of fly ash based
geopolymeric binders. Geopolymers, Melbourne.
Linkson, P., Glastonbury, J., Duffy, G., 1973. The mechanism of granule growth in wet
pelletising. Trans. Inst. Chem. Eng. 51, 251–259.
Newitt, D., Conway-Jones, J., 1958. A contribution to the theory and practice of granula-
tion. Trans. Inst. Chem. Eng. 36, 422–442.
Nugteren, H., Butselaar-Orthlieb, V., Izquierdo, M., 2009. High strength geopolymers pro-
duced from coal combustion fly ash. Glob. NEST J. 11, 155–161.
Ogundiran, M., Nugteren, H., Witkamp, G., 2013. Immobilisation of lead smelting slag
within spent aluminate — fly ash based geopolymers. J. Hazard. Mater. 248-249,
29–36.
Provis, J.L., Lukey, G.C., van Deventer, J.S.J., 2005. Do geopolymers actually contain nano-
crystalline zeolites? A reexamination of existing results. Chem. Mater. 17, 3075–3085.
Provis, J.L., Duxson, P., van Deventer, J.S.J., 2010. The role of particle technology in devel-
oping sustainable construction materials. Adv. Powder Technol. 21, 2–7.
Provis, J., van Deventer, J., 2014a. Alkali Activated Materials - State-of-the-Art Report,
RILEM TC 224-AAM. 13. Springer Netherlands. http://dx.doi.org/10.1007/978-94-
007-7672-2.
Provis, J., van Deventer, J., 2014b. chapter 8 durability and testing-chemical matrix degra-
dation processes. In: Provis, J., van Deventer, J. (Eds.), Alkali Activated Materials —
State-of-the-art Report. RILEM TC 224-AAM. Springer Netherlands;, p. 13 http://dx.
doi.org/10.1007/978-94-007-7672-2.
49
Provis, J., van Deventer, J., 2014c. Chapter 9 durability and testing — degradation via mass
transport. In: Provis, J., van Deventer, J. (Eds.), Alkali Activated Materials — State-of-
the-art Report. RILEM TC 224-AAM. Springer Netherlands, p. 13 http://dx.doi.org/
10.1007/978-94-007-7672-2.
Reynolds, G., Le, P., Nilpawar, A., 2007. High shear granulation. In: Salman, A., Hounslow,
M., Seville, J. (Eds.), Handbook of Powder Technology. Elsevier, Netherlands, p. 11.
Rumpf, H., 1962. The strength of granules and agglomerates. Agglomeration. Interscience
publishers, pp. 379–418.
SFS-EN-1097-3, 1998. Tests for Mechanical and Physical Properties of Aggregates — Part
3: Determination of Loose Bulk Density and Voids.
SFS-EN 13055-1, 2002. Lightweight Aggregates — Part 1: Lightweight Aggregates for Con-
crete, Mortar and Grout.
SFS-EN 933-1, 2012. Tests for geometrical properties of aggregates. Part 1: determination
of particle size distribution. Sieving Method.
Sherington, P., 1968. The granulation of sand as an aid to understanding fertilizer granu-
lation. Chem. Eng. 201–15.
The Powder Diffraction File, 2006. PDF. The International Centre for Diffraction Data.
Tyni, S.K., Yliniemi, Juho, Tiainen, Minna, Laitinen, Risto, 2012. Fly ash as precursor for
geopolymers. Proc of the 21st Int Conf on Fluid Bed Comb. 1, pp. 171–178.
Tyni, S.K., Karppinen, J.A., Tiainen, M.S., Laitinen, R.S., 2014. Preparation and characteriza-
tion of amorphous aluminosilicate polymers from ash formed in combustion of peat
and wood mixtures. J. Non-Cryst. Solids 387, 94–100.
Wauters, P.A.L., 2001. Chapter 2 mechanisms. Modelling and Mechanisms of Granulation,
pp. 7–36.
Williams, R.P., Van Riessen, A., 2010. Determination of the reactive component of fly ashes
for geopolymer production using XRF and XRD. Fuel 89, 3683–3692.
Zhang, J., Provis, J., Feng, D., van Deventer, J., 2008. Geopolymers for immobilization of
Cr6+, Cd2+, and Pb2+. J. Hazard. Mater. 157 (2-3), 587–598.