Cement and Concrete Research 38 (2008) 1119–1125

Contents lists available at ScienceDirect

Cement and Concrete Research

j o u r n a l h o m e p a g e : h t t p : / / e e s . e l s e v i e r. c o m / C E M C O N / d e f a u l t . a s p

The influence of mineralogical, chemical and physical properties on grindability of

commercial clinkers with high MgO level
Vládia Cristina G. de Souza a,⁎, Jair Carlos Koppe a, João F.C.L. Costa a, André Luís Marin Vargas b,
Eduardo Blando b, Roberto Hübler b
a
Department of Mining Engineering, Universidade Federal do Rio Grande do Sul, DEMIN/UFGS, Av. bento Gonçalves, 9500, Prédio 75, Setor 4, CEP 91509-900, Porto Alegre, Brazil
b
GEPSI, faculdade de física da Pontifícia, Instituto de física da PUCRS, Universidade Católica do Rio Grande do Sul, Av. Ipiranga 6681, P.O., CEP 90619-900, Porto Alegre, Brazil

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

Article history: This research investigates various methods able to identify possible mineralogical, physical and chemical
Received 8 June 2006 influences on the grindability of commercial clinkers with high MgO level. The aim of the study is to evaluate
Accepted 27 February 2008 the hardness and elastic modulus of the clinker mineral phases and their fracture strength during the
comminution processes, comparing samples from clinkers with low MgO level (0.5%) and clinkers with
Keywords: elevated MgO levels (N 5.0%). The study of the influence of mineralogical, chemical and physical properties was
Clinker carried out using several analytical techniques, such as: optical microscopy, X-ray diffraction with Rietveld
Comminution
refinement (XRD) and X-ray fluorescence (XRF). These techniques were useful in qualifying the different
Nanohardness
clinker samples. The drop weight test (DWT) and the Bond ball mill grindability test were performed to
Trace metals
characterize the mechanical properties of clinkers. Nanoindentation tests were also carried out. Results from
the Bond ball mill grindability test were found to be related to the hardness of the mineral phase and to
mineralogical characteristics, such as type and amount of inclusions in silicates, belite and alite crystals shape,
or microcracked alites. In contrast, the results obtained by the DWT were associated to the macro
characteristics of clinkers, such as porosity, as well as to the hardness and mineralogical characteristics of
belite crystals in clusters. Hardness instrumented tests helped to determine the Vickers hardness and elastic
modulus from the mineral phases in commercial clinkers and produced different values for the pure phases
compared to previous publications.
© 2008 Elsevier Ltd. All rights reserved.

1. Introduction carried out using several analytical techniques, such as: optical mi-
croscopy, X-ray diffraction with Rietveld refinement (XRD) and X-ray
The knowledge of the mechanical properties of clinker minerals fluorescence (XRF). The drop weight test (DWT), commonly used for
directly assists in controlling and optimizing the manufacturing pro- assessing the breakage characteristics of rocks, and the Bond ball mill
cesses and grinding of clinkers, as well as the quality control of grindability test were performed to characterize the mechanical
the concrete and cements produced as a result. Previous studies properties of clinkers. The first test supplies the impact parameter
in modeling and simulation of concrete and cement microstructures “A.b”, meaning the higher the impact parameter the lower the
have demonstrated what can be achieved in the field of quality breakage strength. The Bond ball mill test provides the so-called
control, via the use of typical applications [1]. Clinker breakage standard work index (WI), defined as the specific power required
strength, its hardness and elastic modulus provide an indication (kWh/t) to reduce a material from a nominal infinite size to a P80 (80%
of energy consumption for the material used during grinding. The of the material passing through a sieve aperture) size of 100 μm [2,3].
cement strength properties, the hardness and elastic modulus of the Instrumented indentation, also known as depth-sensing indentation
clinker depend on the raw materials, the kiln operating conditions and or nanoindentation, is increasingly being used to probe the mechanical
the type of fuels used. response of materials including metals and ceramics as well as
The aim of this study is to evaluate the hardness and elastic polymeric and biological materials. The additional levels of control,
modulus of the clinker mineral phases and their fracture strength sensitivity, and data acquisition offered by instrumented indentation
during the comminution processes by instruments indentation. The systems have resulted in numerous advances in materials science,
influence of mineralogical, chemical and physical properties was particularly regarding fundamental mechanisms of mechanical behavior
at micrometer and even sub-micrometer length scales [4].
In this case study, the elastic modulus and the Vickers hardness of
⁎ Corresponding author. the major four mineral phases contained in commercial clinkers were
E-mail address: [email protected] (V.C.G. de Souza). measured by nanoindentation as well as the matrix inclusions

0008-8846/$ – see front matter © 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cemconres.2008.02.011
1120 V.C.G. de Souza et al. / Cement and Concrete Research 38 (2008) 1119–1125

Table 2
Clinker nodules size distribution

Fraction Interval Weight %

size (mm)
A B C

1 N 19.05 17.82 16.87 21.17

2 19.05–9.52 22.53 23.49 22.55
3 9.52–4.75 27.05 17.58 21.31
4 4.75–2.38 28.36 22.9 25.65
5 b 2.38 4.24 12.77 9.32
Total 100.00 100.00 100.00

were randomly chosen and mounted in twenty (20) polishing sections

with 30 mm diameter. Thirty (30) nodules from the 4.75–2.38 mm
fraction were mounted in three (3) polishing sections with 30 mm
diameter (ten nodules by polishing section).
The specimens were cut using a diamond saw, removing the top
Fig. 1. Strength ranking of clinkers according to impact parameter by DWT tests.
and the bottom of the nodules. After, the specimens (“coins” of
clinker) were impregnated with an epoxi resin of low viscosity to
permeate the material's pore system. To speed the infiltration, the
(aluminates and ferrites) in the silicates. The influence of these parameters
specimens were completely immersed in epoxy, and a vacuum applied
and the crystal's mineralogical properties on clinker grindability was
to remove remaining air. Upon release of the vacuum, the epoxy was
investigated.
forced into the pore system. The epoxy was cured at ambient
temperature for 24 h, and then was ready for grinding and polishing.
2. Sample selection and rock breakage tests
Following some procedures suggested by Campbell [7], abrasive
papers of 100, 220, 320, 400 and 600 (silicon carbide paper) were used
Clinker samples coming from different cement plants from Brazil
with propylene glycol as lubricant in the grinding stages. Subsequent
(involving eleven kilns) were used in the comminution study. Five very
polishing stages of 6 μm to 0.25 μm with diamond polishing pastes
narrow size fractions (−25.4 + 22.2 mm, −22.2 + 19.0 mm, −19.0 +
were performed. Approximately 6 to 10 min for grinding or polishing
15.9 mm, −15.9 + 13.5 mm, −13.5 + 12.7 mm) and three inputs of specific
each paper was normally sufficient, after which the sample was
energy levels (0.25; 0.5 and 1.0 kWh/ton) were selected to carry out the
thoroughly cleaned with isopropyl alcohol in a sonic cleaner. The
DWT tests.
polished surfaces were protected with an acrylic spray after the
The commercial clinkers showed a large range in impact parameter
microscopic analyses, which could be removed by gentle rubbing with
(42–106) according to the DWT results appearing in Fig. 1. Besides, Fig. 1,
an acetone- or xylene-soaked rag to be re-analyzed. The polished
shows that the coarse nodules of commercial clinkers have generally
section surfaces were etched by Nital (1.5 mL of nitric acid, HNO3 in
medium (51 b A.b b 63) to very high (Ab b 50) breakage strength. These
100 mL of isopropyl) superimposed on a 20-second potassium
intervals represent, in relative frequency, 36% and 45%, respectively, of
hydroxide etch which turns C3A light brown and colors the silicates.
the clinkers tested. Clinkers with low or very low breakage strength
The reflected light optical microscopy was used to characterize the
represent a minor group (relative frequency of 9%). From these results,
different clinker samples.
three clinkers (A, B and C) were selected: (i) two high strength clinkers
One hundred of belite and alite crystals were measured in each
(A and C) and (ii) one very low strength clinker (B). After, clinkers A, B
polishing section and the average value was adopted as the crystal size
and C were submitted to the ball mill tests to determine the work index
described in this work. Description of the porosity, crystals shape,
values.
inclusions and microcracks in crystals represents the statistical result
Table 1 shows the WI results for clinkers A, B and C. Although clinker
of detailed analyses of hundred crystals contained in twenty three
B has a high WI value (heavy grindability, classification according
(23) polishing sections of each clinker, totaling 69 polishing sections in
to Opoczky [5]) it showed very low breakage strength by DWT. On the
this study.
other hand, clinkers A and C have high WI but they have also high
X-ray Diffraction (XRD) with Rietveld refinement was applied to
breakage strength by DWT.
accurately quantify the phases present in the samples by using the
same procedures adopted by Gobbo [9].
3. Sample preparation and experimental procedures
For the nanoindentation essays at least three additional polished
sections were prepared for each clinker A, B and C. An extra sample for
A mass of approximately 100 kg of clinker was homogenized and
each clinker was cut in a 10 mm edge cubical form. These samples were
quartered to 5 kg aliquot used to determine nodule size distribution by
polished again and analyzed by optical microscope to check the quality
screening. Three size fractions of nodules were selected and prepared
of polishing. This procedure was adopted to avoid the influence of the
(Table 2) following procedures suggested by Campbell and Centurione
preparation, the substrate and setting conditions of the samples on the
[7,8]. The size fractions chosen were then: 19.05 × 9.52 mm,
measurements of hardness and elastic modulus. Twenty indentations
9.52 × 4.75 mm and 4.75 × 2.38 mm. Ten (10) nodules from the fraction
in each mineral phase were carried out and the Vickers hardness and
19.05–9.52 mm and ten (10) nodules from the fraction 9.52–4.75 mm
elastic modulus average value were determined.
The procedures for nanoindentation tests followed ISO 14577 [10],
Table 1 which has been prepared to enable the user to evaluate the inden-
Bond ball mill grindability test results tation of materials by considering the force and displacement during
plastic and elastic deformation. By monitoring the complete cycle of
Clinker symbol WI (kWh/t) Classification
increasing and removal of the test force, hardness values equivalent
Clinker A 22.0 Hard grindability to traditional hardness values can be determined. More significantly,
Clinker B 19.8 Hard grindability additional properties of the material, such as its indentation modulus
Clinker C 18.5 Hard grindability
and elastoplastic hardness, can also be determined. All these values
 V.C.G. de Souza et al. / Cement and Concrete Research 38 (2008) 1119–1125 1121

Fig. 2. Clinker A observed by optical microscope: (a) nodule of clinker and a large band of belites, type I belites (200×); (b) belites dispersed and subhedral alite crystals (200×);
(c) round belites, type I belite (1000×) and (d) amoeboid or cannibalistic alite with inclusion of round belite (1000×).

Fig. 3. The two types of belites in clinker B measured by HIT and observed by optical microscope: (a) ragged belites, type I belite (1000×); (b) ragged belites and belites inclusion in
alites, type I belite (1000×); (c) coarser belites, type II belite (500×) and (d) zoom on belite, type I belite (1000×).

can be calculated without the need to measure the indentation opti- clusters and to estimate the average size of clusters and crystals.
cally. ISO 14577 specifies the method of instrumented indentation test From the microscopy it was also possible to interpret the burning
for determination of hardness and other materials parameters. conditions [7].
The resolutions of load and displacement for the Fisherscope H100V Clinker A (Figs. 2 and 5a) depicts sub to anhedral (amoeboid) alite
equipment used were 0.02 mN and 2 nm, respectively. Hardness and crystals of 60 μm (ranging from 15 to 120 μm), strongly zoned,
elastic modulus were obtained using a Berkovich indenter, i.e., a three- twinned and cannibalistic alite crystals, without microcracks and with
sided pyramid diamond, with the same nominal area to depth inclusions of belite. The belite crystals also show a large range in size
relationships as the standard Vickers pyramid. The hardness instrument distribution (8–63 μm), with an average of 22 μm. The majority of
tests (HIT) were made with a maximum load of 45 mN and a total time these crystals were rounded off without microcracking at the edges.
load–unload cycle of 120 s. To prevent the load dependence on the The belites appear in clusters but also dispersed (Fig. 2a and b). The
hardness measurements, the indentation data was interpreted using the sample also shows a glass interstitial phase and porosity of around
Doerner–Nix [11], Oliver–Pharr [12] and indentation size effect (ISE) 23% (regular and isolated).
theories [13]. The elastic modulus and hardness were measured at Clinker B (Figs. 3 and 5b) is constituted by subhedral alite crystals,
penetration depths of about 300 and 500 nm. with an average size of 29 μm (ranging from 11 to 77 μm), with
abundant belite inclusions, and many alites are microcracked. There
4. Experimental results are several types of belites in this sample: ragged, some only with
fingers and others with disorganized cores (denominated type I) and
4.1. Optical microscopy rounded off coarser belites (denominated type II). Both belite types
have different nucleus coloration from the edges, but the type II belite
Clinker microscopy was used to distinguish between different is present in minor amount (minor than 2% by quantitative optical
types of belites (C2S) and alites (C3S), specifically to count the belite microscopy). The interstitial phase is comprised of semi-crystallized to

Fig. 4. Clinker C observed by optical microscope: (a) amoeboid belite (500×); (b) subhedral alite crystals with inclusions of matrix and MgO at the edges (1000×); (c) round type I
belites in clusters (200×) and (d) round belites around pores (nests).
1122 V.C.G. de Souza et al. / Cement and Concrete Research 38 (2008) 1119–1125

Fig. 5. Zoom on alites of the clinkers tested: (a) clinker A with cannibalistic, zoned and rounded (anhedral or amoeboid) alites (500×); (b) clinker B with microcracked alites (1000×)
and (c) clinker C with abundant inclusions of matrix and MgO in alites (1000×).

crystallized minerals, as well as crystals of periclase developed in the 4.2. X-ray Diffraction (XRD) and fluorescence (XRF)
matrix. Clinker B showed average porosity around 40%, as well as
irregular and interconnected pores. Its characteristics were typical of The clinkers mineral composition determined by XRD is presented in
clinkers showing a slow first cooling followed by a second cooling Table 4. The alite content in clinker A is higher than in clinkers B and C.
tending to slow. The belite content in clinker C is significantly lower than in clinkers A and
Clinker C (Figs. 4 and 5c) shows subhedral alite crystals of 24 μm B. Clinker A is the ferrite richest one. Clinker A presents only C3A cubic and
(ranging from 3 to 54 μm), cannibalistic, with matrix (abundant) and clinkers B and C present the two types of aluminates: C3A orthorhombic
MgO inclusions. The belite crystals were almost rounded off (without and cubic. The largest differences are in the minor phases: periclase,
fingers) in nests and clusters, and some were amoeboid and dispersed free lime and arcanite. Clinker B has the lower content of arcanite, and
in the matrix. The interstitial phase of clinker C is semi-crystallized two types of sulfates were detected in clinker A: arcanite and calcium
and the porosity is low (30%), with regular and interconnected pores. langbeinite. Clinker A has the lowest free lime content, followed by
The Fig. 5 shows the alites from clinkers A, B and C respectively. The clinkers C and B with 0.8 and 1.7% free lime content respectively.
shape of the alite crystals, the type and amount of inclusions in alites The clinkers chemical ratios were estimated from oxides quantified
are very different in these three samples. Clinker A shows alite crystals by XRF (Table 5). The free lime was quantified by ethylene glycol
with the largest size, rounded shape and zoning; clinker B shows alites method according to the NBRNM13 [14]. It was observed that clinker A
microcracked with subrectilinear edges and a large size distribution; presents a higher lime saturation factor (LSF) than B and C, but a lower
clinker C shows alites subhedral with abundant inclusions of matrix silica ratio (SR). The alumina ratio (AR) can influence the grindability
and MgO. [6]; normally clinkers with a high AR have high grindability. AR values
The description of belite clusters in clinkers A, B and C found in the for the studied clinkers are 1.52 (clinker A), 1.57 (clinker B) and 1.46
coarse nodules tested by DWT (19.05 × 9.52 mm) is shown in Table 3. (clinker C). TiO2 and Mn2O3 contents were determined by XRF, and
Regarding Table 3, clinker C presents an average size of belite clinker A showed the highest values.
cluster larger than that found in clinkers B and A. Clinker B shows the
highest maximum size of belite cluster (919 μm) and very high average 4.3. Nanoindentation
counting of belite clusters by nodule (36 clusters by nodule). Clinker A
shows the lowest average counting of belite clusters by nodule (13 Small differences were observed in the hardness and elastic modu-
clusters by nodule), but has a great number of dispersed and very lus values measured by the nanoindentation between the two types
rounded belites (60% from belites are dispersed). of samples tested: (i) round sections of 30 mm in diameter, and (ii)
Clinker A and, particularly, clinkers B and C, although featuring large 10 mm edge cubical pieces. Vickers hardness and elastic modulus
(N150 μm) average size of belite cluster, produce blended cements
according to the Brazilian mechanical and physical exigencies [6].
Table 4
X-ray Diffraction results for studied clinkers (wt.%)

Table 3 A B C
Description of belite clusters in the clinkers analyzed by optical microscopy
Phase/sample
C3S 70.2 66.6 66.9
Clinkers A B C
C2S 13.3 12.3 7.9
Distribution (in percentage) Regular 15 55 80 C3A-cubic 2.6 3.0 1.3
and shape (between parenthesis) clusters (round) (ragged) (round) C3A-orthorhombic – 1.9 1.6
of the belite crystals 15 C4AF 10.1 7.6 8.2
(round) Periclase 0.6 6.4 11.2
Irregular 25 28 15 Free lime 0.2 1.7 0.8
clusters (round) (ragged) (round) K2SO4a 1.7 0.5 2.1
Dispersed 60 2 5 K2Ca2(SO4)3b 1.3 – –
crystals (round) (ragged) (round) Total 100.0 100.0 100.0
Average size of belite cluster (µm) 170 262 300
Minimum and maximum size of 50–400 159–919 130–578 Relations
belite cluster (µm) C3S + C2S 83.5 78.9 74.8
Average counting of belite clusters by nodule 13 36 20 C3A + C4AF 12.7 12.5 11.1
Average size of belite crystals (µm) 25 20 22 C3S/C2S 5.28 5.41 8.47
Minimum and maximum size of 3–53 8–72 8–63 a
Arcanite.
belite crystals (µm) b
Calcium langbeinite.
 V.C.G. de Souza et al. / Cement and Concrete Research 38 (2008) 1119–1125 1123

Table 5
X-ray fluorescence results for clinkers A, B and C (wt.%)

A B C

SiO2 21.50 20.87 19.85

Al2O3 4.70 4.05 3.75
Fe2O3 3.10 2.58 2.56
CaO 64.85 61.58 57.65
MgO 1.50 7.48 12.01
Na2O 0.10 0.40 0.12
K2O 1.10 1.10 1.36
TiO2 0.25 0.12 0.12
SO3 2.20 0.28 0.97
Mn2O3 0.10 0.05 –
LSFa 95.70 94.90 93.48
SR 2.76 3.15 3.15
AR 1.52 1.57 1.46
Free limeb 0.50 1.01 1.21
Alkalies equivalent 0.82 1.12 1.01
a
Ignoring SO3.
b
Determined by ethylene glycol method. Fig. 6. Load–displacement curves of mineral phases of samples from clinker A.

average values for the mineral phases for clinkers A, B and C are shown free lime grains in the vicinity of the indentation performed on the
respectively in Tables 6 and 7 (the round sections were identified as crystal could also influence the measurements.
A, B and C and the cubical pieces were identified as A', B' and C'). The standard deviation of the Vickers hardness values were high
The indentations number on inclusions of aluminate or ferrite inside for the two types of belites found in the samples from clinker B
alites from clinker A had to be reduced of twenty to five measures by (around 10.0% for type I belite and around 18.4% for type II belite).
polishing section, because the inclusions more common in this case The average values of elastic modulus for alites and type I belites
were of type I belites. ranged from 125 to 122 GPa, respectively (Table 7). These values are
The hardest phases were the ferrite (C4AF) and the type II belite, evidence of the larger plasticity of the silicates compared against the
followed by the aluminate (C3A), alite (C3S) and type I belite. Therefore, aluminate and ferrite phases. The last two phases show, respectively,
the type I belite shows the lowest plastic deformation strength (the an average elastic modulus of 139 and 142 GPa.
lowest Vickers hardness). The elastic modulus of the alite in clinker A was smaller than the
Several factors influence the HIT measurements such as: distinc- elastic modulus of the alite in clinkers C and B. This indicates that the
tions between Vickers hardness measured from an indentation on a alite in clinker A is more plastic than the alite in B and C. This is consistent
matrix inclusion (C3A or C4AF) inside silicate crystals and Vickers hard- with the Bond Index results.
ness measured from an indentation outside these inclusions. The The type II belite presents lower significantly fracture strength
mineral inclusions of aluminates or ferrites inside silicates showed an than the type I belite (Tables 6 and 7).
intermediate Vickers hardness value, which is between the value The aluminate of clinker A is more plastic than the aluminates of
obtained for silicates and the value obtained for matrix phases. clinkers B and C (clinkers are based on their elastic modulus). The values
Measurement deviations derive from several factors, such as crystal of elastic modulus show, thus, a large difference in the mechanical
orientation, crystal chemical composition, substrate and parallelism behaviors of the matrix mineral phases (aluminates and ferrites) for all
problems in the polished section. Inclusions, pores, oxides, sulfates or three tested clinkers.

Table 6
Average values Vickers hardness (GPa) and standard deviation, S.D. (%), of the minerals phases from clinkers A, B and C

Clinker C3S S.D. Type I C2S S.D. Type II C2S S.D. C3 A S.D. C4AF S.D. Inc.⁎ S.D.

A 5.61 11.30 5.45 7.91 – – 7.85 5.92 12.75 3.63 7.69 6.63
A' 5.59 11.11 5.53 6.72 – – 8.76 6.04 12.86 3.11 7.49 3.37
B 5.86 5.56 5.85 10.38 12.56 19.58 8.10 5.00 10.04 5.00 6.51 5.72
B' 5.87 5.00 5.60 9.07 11.21 17.25 7.65 8.20 9.93 5.58 6.28 6.86
C 5.36 5.24 4.75 7.91 – – 9.16 8.74 10.65 8.13 7.44 7.52
C' 5.83 6.56 4.77 7.18 – – 9.02 6.35 10.46 8.67 6.57 7.76
Average global 5.69 7.46 5.33 8.20 11.89 18.42 8.42 6.71 11.12 5.69 7.00 6.31

Note: Inc.⁎ C3A or C4AF inclusion in alites. S.D. (Standard Deviation).

Table 7
Average values of elastic modulus (GPa) of the minerals phases from clinkers A, B and C (GPa)

Clinker C3S S.D. Type I C2S S.D. Type II C2S S.D. C3A S.D. C4AF S.D. Inc.⁎ S.D.

A 112.79 10.06 117.48 11.33 – – 124.39 2.47 135.92 2.41 124.19 2.80
A' 107.24 4.75 122.60 5.86 – – 130.19 4.80 136.32 2.06 125.66 2.15
B 138.75 2.66 122.23 8.09 161.89 13.99 138.50 5.45 127.46 4.39 143.16 2.65
B' 130.81 6.28 124.52 9.92 162.60 9.09 144.48 6.35 128.24 4.17 140.90 4.48
C 129.32 4.86 117.27 9.02 – – 149.11 4.31 165.13 7.73 130.56 4.55
C' 129.58 6.02 126.99 9.35 – – 149.21 5.78 161.07 6.77 128.58 4.57
Average global 124.75 5.77 121.85 8.93 162.25 11.54 139.31 4.86 142.36 4.59 132.18 3.53

Note: Inc.⁎ C3A or C4AF inclusion. S.D. (Standard Deviation).

1124 V.C.G. de Souza et al. / Cement and Concrete Research 38 (2008) 1119–1125

Fig. 7. Load–displacement curves of the belite phases of samples from clinker B. Fig. 9. Load–displacement curves of the mineral phases of samples from clinker C.

Figs. 6–9 show the mechanical behavior of the mineral phases In Fig. 9, the belite crystals (type I) from clinker C shows a dis-
in commercial clinkers through load–displacement curves. All of these placement curve to the right of the alite curve, evidencing a more
anhydrous phases exhibit elastoplastic behavior. The matrix phase shows plastic behavior.
a non-pure elastic behavior while the silicates generally present a plastic The average Vickers hardness weighted, considering only the
behavior. These load–displacement curves shown in Figs. 6–9 represent major four phases from XRD analysis (Table 4), produced results of
the average curves by mineral phase and by clinker, calculated from approximately 6.2 GPa for clinker A, 6.0 GPa for clinker B (considering
all indentations made on polishing sections (round section or cubical 10.5% of type I belite and 1.8% of type II belite) and 5.2 GPa for clinker
piece). C. The average elastic modulus was 110.2, 121.6 and 114.0 GPa for
Figs. 6 and 9 show that the load–displacement curves from one clinkers A, B and C, respectively.
indentation made inside and outside one mineral inclusion (in alite
crystals) are different. For example, when the indentation is made inside 5. Discussion
one inclusion with lower hardness, such as type I belite, the curve is
displaced to the right of alite load–displacement curve made outside The crystals shape and size, as well as the “cannibalism” (crystals
inclusion (Fig. 6). On the other hand, when the indentation is made union), observed in alites or belites are one reflex of the clinker
inside one inclusion with higher hardness, such as one inclusion of C3A formation process and can be explained by Ostwald ripening [15]:
or C4AF, the curve is displaced to the left of alite load–displacement mineral grain size and shapes spontaneously adjust toward a state of
curve made outside inclusion (Fig. 9). lower surface free energy. According to Best [15], large and small
In Fig. 7, the differences in the mechanical behavior of the two crystals of the same phase dispersed in a communicating fluid or melt
types of belites (types I and II) in clinker B can be observed. The type I at fixed pressure and temperature are relatively more and less stable,
belite is more plastic than type II belite, the last with a higher Vickers respectively, because of their differences in surface free energy per
hardness. Morphological differences between these two belite types volume. Consequently, smaller crystals tend to be consumed at the
were observed by optical microscopy (Fig. 3). expense of larger, more stable grains. The result of this equilibrating
In Fig. 8, the ferrite of clinker B shows a sharp curve of unloading “ripening” is in an overall increase in average grain size of the system.
(near the pure elastic recovery) while the other phases present a be- Clinker A, with rounded shape and big size crystals, is a perfect
havior of plastic deformation. example of this.
Clinker C shows some similarities with clinker A due the presence
of some amoeboid belite crystals, cannibalistic alites and presence
of belite clusters with size bigger than 150 μm. Both clinkers were
thus probably submitted to high temperatures for long time. Ac-
cording to Hills [16], clinkers that show these characteristics present
hard grindability. Observations on industrial practice also lead the
authors to believe that clinkers submitted to a longer burning time are
actually very hard to grind (high energy costs). On the other hand, this
theory can very well explain why clinkers A and C show high WI and
high fracture strength by DWT, but doesn't explain the behavior of
clinker B.
It is necessary to remember that in the Bond tests the material must
be previously crushed until it passes through a 3 mm sieve. The belite
clusters with high hardness (as type II belites from clinker B) or made by
ragged belites (as type I belites from clinker B) will have higher proba-
bility to be crushed prior to Bond ball mill test being conducted, due to
the maximum grain size admitted for the WI experiment (3 mm), than
rounded belites dispersed or in clusters. This explains better why clinker
B features WI values higher or similar to those of clinkers A and C, but
Fig. 8. Load–displacement curves of the mineral phases of samples from clinker B. lower breakage strength by DWT (coarser nodules).
 V.C.G. de Souza et al. / Cement and Concrete Research 38 (2008) 1119–1125 1125

Besides, clinker A shows: (1) rounder and big belite and alite samples of clinker B show the possibility of the coexistence of belites
crystals and (2) highest C3S and C2S content. As the silicates present with distinct mechanical behaviors in the same clinker. These belites
the lowest Vickers hardness values of the four major components of will be object of further studies.
clinkers, clinker A showed the hardest grindability.
The Vickers hardness weighted values, considering only the major 6. Conclusions
four phases from XRD analysis, indicate that clinker C should have the
highest breakage strength, but in this calculation was not considered The most important evidence obtained from HIT results is that
the Vickers hardness of the matrix inclusions in alites. If the inclusions silicates (especially belites) are more susceptible to plastic deformation
of matrix in the silicates of clinker C, which are abundant, would be and, for this reason, can produce crack deviation that would propagate in
considered, the actual Vickers hardness of this clinker would be the clinkers nodules. Coarse nodules of clinker with high porosity, if they
slightly higher. Thus, the three clinkers, A, B and C, in fact, contain contain these ragged belite clusters or microcracked alites, will present
crystals with high and similar fracture strength. This explains why the very low breakage strength in the first chamber of cement mills.
three clinkers tested showed high and similar work index values. On the other hand, the same clinker, in the second chamber cement
To explain the fracture mechanics of the coarser nodules of clinkers, mill (fine grinding), not necessarily will show easy grindability. This
it is necessary to analyze the DWT tests. DWT test could be more rep- happens with samples from clinker B, in fact, which showed the lowest
resentative than WI as it uses the very last five narrow size fractions fracture strength under impact by DWT, but also very high WI value.
and three inputs of specific energy levels (kWh/t). It is able to reflect The small differences in WI are related to mineralogical and me-
basically the macro characteristics of the coarser nodules of clinkers, chanical properties of the mineral phases of commercial clinkers. The
such as porosity. In contrast, other mineralogical properties of clinkers largest differences among the samples tested in DWT (clinkers A, B
seem to affect the DWT results, such as, belite clusters and microcracks and C) are associated with macro characteristics of clinkers, such as
in alites. porosity and belite clusters.
The ragged belite clusters, many microcracked alites and a higher The Bond and DWT tests together may be used for explaining the
porosity (irregular and interconnected pores) were factors that de- resistance to breakage of clinkers and the influence of their chemical,
finitively may have contributed together to decrease the breakage physical and mineralogical characteristics on grinding, especially in
strength of the coarser nodules of clinker B (impact parameter very multi chamber cement mills. HIT can supply additional information on
high). The coarser belites (type II) in clusters, from clinker B, that show the breakage mechanisms of the industrial clinkers, and was used in
very high Vickers hardness, also helped to decrease the breakage the present work specially to show that silicates really have more easy
strength of these coarser nodules. plastic deformation than aluminates or ferrites.
Therefore, counting belite cluster or nests is not sufficient to evaluate
or explain the fracture mechanics of clinker coarser nodules. It must Acknowledgements
be considered also the type of belites, their shape (round, ragged or
dismembered), the amount and distribution of belites, and specially The authors thank the financial support from CNPq — Conselho
Vickers hardness of the mineral phases. Nacional de Desenvolvimento Científico e Tecnológico, Brazil and
Besides, the periclase and free CaO content (determined by Cimbagé (CIMPOR — Cimentos de Portugal).
ethylene glycol) had negligible effect on fracture strength of the
coarse nodules of the tested clinkers. On the other hand, WI values of References
the clinkers can be correlated with these components: clinker with
[1] K. Velez, S. Maximilien, D. Damidot, G. Fantozzi, F. Sorrentino, Determination by
higher free CaO and periclase content showed lower WI. nanoindentation of elastic modulus and hardness of pure constituents of Portland
Regarding the nanoindentation results, there is another interesting cement clinker, Cem. Concr. Res. 31 (2001) 555–561.
point: Vickers hardness and elastic modulus of pure mineral phases [2] F.C. Bond, The third theory of comminution, Trans. Soc. Min. Eng. AIME 193 (1952)
484–494.
such as C3S and C2S were previously determined in the literature by [3] T.J. Napier-Munn, S. Morrell, R.D. Morrison, T. Kojovic, Mineral Comminution
Velez et al. [1]. They concluded that there was little influence of minor Circuits — Their Operation and Optimisation, JKMRC, Austrália, 1996.
elements in the hardness and in the elastic modulus, although the [4] M.R. VanLandingham, Review of instrumented indentation, J. Res. Natl. Inst. Stand.
Technol. 108 (2003) 249–265.
load–displacement curves had shown a large difference between pure [5] L. Opoczky, V. Gável, Effect off certain trace elements on the grindability of cement clinkers
phases and impure phases. Tables 6 and 7 show the Vickers hardness in the connection with the use of wastes, Int. J. Miner. Process. 74S (2004) S129–S136.
and elastic modulus of mineral phases at commercial clinkers. These [6] NBR5736, Pozzolanic Portland Cement — Technical Standards Brazilian Association
(ABNT), 1991.
results are really different from those of the pure phases determined
[7] D.H. Campbell, Microscopical Examination and Interpretation of Portland Cement
in prior literature [1], due to the presence probably of certain trace and Clinker, 2nd ed.Portland Cement Association, Skokie, I.L., USA, 1999.
elements that will be object of further study. According to Opczky [5], [8] S.L. Centurione, Influência das características das matérias-primas no processo de
the incorporation of the trace elements may affect: the micro sym- sinterização do clínquer Portland, Master dissertation, Universidade de São Paulo,
Brazil, 1993.
metry and electrostatic relations of the structure of clinker minerals, [9] L.A. Gobbo, L. Sant', L. Garcez, C3A polymorphs related to industrial clinker alkalies
the chemical bonds (covalent, ionic) between the ions, the coordi- content, Cem. Concr. Res. 34 (2004) 657–664.
nation of ions and certain physical–mechanical properties (hard- [10] ISO 14577-1, Metallic materials — Instrumented indentation test for hardness and
materials parameters — Part 1: test method, Int. Org. for Stand., 2002.
ness, rigidity) of the clinker minerals. The change in the grindability [11] M.F. Doerner, W.D. Nix, A method for interpreting the data from depth-sensing
of clinkers is the macroscopic manifestation of these effects. This indentation instruments, J. Mater. Res. 1 (1986) 601–609.
could even explain differences in Vickers hardness and elastic modu- [12] W.C. Olivier, G.M. Pharr, An improved technique for determining hardness and
elastic modulus using load and displacement sensing indentation experiments,
lus values for the same mineral phase among several commercial J. Mater. Res. 7 (1992) 1564–1583.
clinkers. [13] P.C. Jindal, D.T. Quinto, Load dependence of microhardness of hard coatings, Surf.
The shape of the unloading curve showed that the ferrite presents Coat.Technol. 36 (1988) 683–694.
[14] NBRNM13, Portland Cement, Chemical Analyses, Determination of Free Lime by
more elastic recovery behavior than the aluminates, while the silicate
Ethileno Glycol Method, Technical Standards Brazilian Association (ABNT), 2004.
phases present a behavior next to plastic deformation. The main [15] M.G. Best, Igneous and Metamorphic Petrology, 2nd ed.Blackwell Science Ltd, UK, 2003.
differences of Vickers hardness and elastic modulus are in the matrix [16] L.M. Hills, Under the microscope — an introduction to clinker microscopy, Cem. Am.
(1999) 39–43.
and belite phases. The load–displacement curves from indentation on