Construction and Building Materials 63 (2014) 150160

Contents lists available at ScienceDirect

Construction and Building Materials

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

Nano-mechanical behavior of a green ultra-high performance concrete

Sujing Zhao , Wei Sun
Jiangsu Key Laboratory of Construction Materials, School of Material Science and Engineering, Southeast University, Nanjing 211189, China

h i g h l i g h t s
 The UHPC contained high volumes of y ash and used river sand as aggregate.
 The nano-mechanical behavior of the UHPC was investigated by nanoindentation.
 The hydration products of the UHPC are dominated by high-stiffness hydrate phases.
 The mechanical bond at the interfacial zone is strong and efcient.
 High-stiffness phases such as mullite and hematite in y ash were detected.

a r t i c l e

i n f o

Article history:
Received 11 December 2013
Received in revised form 12 March 2014
Accepted 2 April 2014

Keywords:
Ultra-high performance concrete
Nanoindentation
Microstructure
Fly ash
Elastic modulus
Hardness
Interfacial zone

a b s t r a c t
As a revolutionary construction material, ultra-high performance concrete (UHPC) has been extensively
studied in the last two decades and one research interest has been focused on preparing UHPC with lowpriced raw materials or industrial by-products. In this study, a green UHPC which used high volumes of
y ash and river sand as part of raw materials was prepared, and the objective is to investigate the nanomechanical behavior of the UHPC using nanoindentation. The results show that the hydration products,
which account for about half of the paste by volume, are mainly high-stiffness hydrate phases, and signicant quantities of unreacted cement and y ash have higher mechanical properties than the hydration
products and can function as micro-aggregates to strengthen the UHPC paste. Moreover, the mechanical
properties of the paste near aggregate or ber surfaces are similar to those of the bulk paste, which indicates that the UHPC has a strong and efcient bond at the interfacial zone. The experimental ndings at
the nano-scale could help to understand the macro-performance of the green UHPC.
2014 Elsevier Ltd. All rights reserved.

1. Introduction
Among the recent advancements in concrete technology, one
remarkable achievement is the development and increasing use
of ultra-high performance concrete (UHPC). UHPC is a highly compact, dense material that exhibits attractive owability, excellent
mechanical properties and exceptional durability. UHPC was rst
developed in 1995 by Richard and Cheyrezy [1] and in the past
few years it has been applied to coupling beams in high-rise buildings, precast members, infrastructure rehabilitations, blast resistant structures and special facilities like nuclear waste storage
containers. It is regarded as an innovative and promising material
for the future, yet the high initial cost affects its broader application.
In this regard, the use of industrial by-products such as y ash and
slag to partially substitute for cement in UHPC production have
been extensively studied, and the replacement of costly quartz sand
Corresponding author. Tel.: +86 25 52090667.
E-mail address: zsj907@gmail.com (S. Zhao).
http://dx.doi.org/10.1016/j.conbuildmat.2014.04.029
0950-0618/ 2014 Elsevier Ltd. All rights reserved.

by natural sand or industrial wastes like glass cullet and iron ore
tailings have also been reported [26]. The incorporation of these
materials into UHPC not only fullls the economic requirement,
but also brings other advantages such as lowering the shrinkage
and improving the durability of the material. More importantly,
the substitution contributes to the recycling of industrial waste
and the reduction of cement consumption, which reduces CO2
emission and makes the material environmentally friendly.
The superior performance of UHPC originates from its engineered microstructure, which is obtained by eliminating coarse
aggregate to improve homogeneity, maximizing the packing
density with an optimized gradation of granular constituents and
enhancing the ductility with high-strength bers [1]. The
microstructure of UHPC has been characterized by multiple experimental techniques including secondary electron microscopy
(SEM), X-ray diffraction (XRD), mercury intrusion porosimetry
and thermogravimetric analysis [79]. While the experimental
results reveal that UHPC has a much denser microstructure and
contains much less calcium hydroxide (CH) crystals as compared

S. Zhao, W. Sun / Construction and Building Materials 63 (2014) 150160

151

with ordinary concrete, the underlying mechanisms that govern

the macro-performance are still not very clear. The main binding
phase, calcium silicate hydrates (CSH), is reported to be a porous
material with two different packing densities, known as low
density (LD) CSH and high density (HD) CSH [10,11]. Their differences in microstructure could signicantly affect the material
performance, yet their properties cannot be studied in depth by
the aforementioned conventional methods. Furthermore, the very
low water-to-binder ratio (w/b) of UHPC implies that a large quantity of cementitious materials remain unreacted after curing, which
raises the question of what role the unreacted particles play in the
material.
Recent applications of nanoindentation in cement-based materials can help unveil the mystery of the materials at the micro- or
nano-scale. Nanoindentation has been proved to be a reliable tool
to determine the intrinsic mechanical properties of various phases
in cement-based materials [1215]. In this respect, the mechanical
properties of cement grains and hydration products have been
studied intensively but few research works have been reported
for mineral admixtures like y ash. Another noticeable application
of nanoindentation is to quantitatively characterize the interfacial
transition zone (ITZ), whose thickness is reported to be several tens
of micrometers [1518]. Compared with microindentation, the relatively small interaction volume during penetration makes it more
appropriate and advantageous to get an accurate prole of the
mechanical properties at the interfacial zone. By virtue of the nanoindentation technique, this paper investigates the nano-mechanical behavior of a green UHPC formulae which contains high
volumes of y ash and uses natural river sand as ne aggregate.
The experimental results are considered to help understand the
macro-performance of the UHPC material.
2. Materials and sample preparation
2.1. Materials
The UHPC studied here is applied to produce cover slabs which are ancillary elements of the rapidly-growing high-speed railway in China. The cover slabs are used
to cover the electric cable trenches and provide a pathway for pedestrian and track
maintenance. Fig. 1a shows a picture of the cover slabs along the rail, and Fig. 1b
shows the dimensions of a typical slab. The slabs are designed to have adequate
mechanical properties to bear the applied loads and a high durability to provide a
long service life. The required properties for the UHPC are summarized in Table 1.
Conventional slabs are made of ordinary reinforced concrete which frequently deteriorates due to a poor durability. In contrast, though the initial cost of using UHPC
slabs is higher, the long service life and low maintenance cost could contribute to a
lower life cycle cost. Moreover, the signicant material savings (dead weight
reduced by 40%) could possibly lead to a lower embodied energy.
The mix design of the UHPC is given in Table 2. Ordinary Portland cement, type
F y ash and silica fume were used as the cementitious materials, and their physical
properties and chemical composition can be found at [6]. In particular, the mineralogical composition of the y ash was examined by XRD and is given in Fig. 2. The
XRD pattern displays a series of sharp diffraction peaks, identied as mullite,
quartz, and minor quantities of hematite and lime. In addition to those crystalline
phases, a pronounced broad hump caused by diffuse scattering of X-rays can be
observed in the background between 16 and 35 2h, which indicates the presence
of amorphous vitreous materials. Natural river sand with an apparent density of
2.63 g/cm3 and a maximum particle size of 5 mm was used as ne aggregate. The
steel ber was brass-coated with a length of 13 mm and an aspect ratio of 65.
The mixing water was potable tap water, and a polycarboxylate-based superplasticizer was used. Compared with reactive powder concrete [1], the UHPC used a
large amount of y ash to partially replace cement, and employed natural river sand
as substitute for costly quartz sand. The cost of the material is lowered and the
energy consumption is reduced. Therefore, the UHPC material promotes green
benets with less CO2 emission.

2.2. Sample preparation

In this study, both the UHPC paste and UHPC were prepared and tested to investigate the nano-mechanical behavior of the cementitious phases and the nanomechanical behavior of the interfacial zone, respectively. The rationale behind
using the paste sample was to avoid possible disturbance by indentation response

Fig. 1. Sidewalk along the high-speed rail (a) and dimensions of a typical cover slab
(b).

from the aggregate. Moreover, to ensure that small sand particles slightly beneath
the surface would not affect the results at the interfacial zone, sand particles smaller than 0.3 mm was removed by a standard sand sieve.
A mortar mixer with two different revolving speeds (140 rpm and 285 rpm) was
used to prepare the material. The dry materials were rst mixed at low speed for
2 min, which helped to break apart the agglomerates and homogenize the material.
Then water and superplasticizer were gradually added and the fresh material was
mixed at the high speed for another 5 min to obtain good owability. Afterwards,
the mixture was cast into a PVC tube with a diameter of 15 mm and a length of
20 mm. For preparing the UHPC specimen, bers were not added during mixing
but were manually added in parallel with the length direction of the tube after casting, which facilitated locating a round cross section of the ber after trimming the
specimen. The specimens were compacted after molding and cured for 1 day at
20 C and relative humidity of greater than 95%. Then the specimens were demolded and cured in a steam chamber at 80 C for 2 days. After curing, the specimens were soaked in ethanol for 7 days to arrest hydration. Then the specimens
were oven-dried at 50 C for 2 days to remove the ethanol before being stored in
a desiccator for further treatment.
To do a nanoindentation test, proper surface preparation is critical in order to
obtain as at a surface as possible to get repeatable results [22]. In this study, the
cylinder specimens were initially sliced using a diamond saw into samples that
were about 10 mm thick. Then the samples were vacuum impregnated with a
low-viscosity epoxy resin. The low permeability of the samples should have given
a low penetration depth of the epoxy. After the epoxy solidied, the samples were
rst ground by successively ner-grained silicon carbide paper. The exposed air
voids on the surface of the samples were checked to ensure that no epoxy was left
after the grinding procedure, to verify that the epoxy-penetrated layer was
removed. Then the samples were polished by water-based diamond suspensions
of four successively ner particle sizes (9 lm, 3 lm, 1 lm and 0.05 lm) on a Buehler TexMet cloth. The polishing time for using each of the rst three diamond suspensions lasted 15 min, while the nal polishing period was 2 h. After each
polishing step, the sample was cleaned ultrasonically in an ethanol bath for 5 min
to remove any debris left on the surface. The polished samples were kept in a desiccator until testing.

152

S. Zhao, W. Sun / Construction and Building Materials 63 (2014) 150160

Table 1
Property requirements of UHPC.

a
b
c

Test

Compressive strengtha

Flexural strengtha

Youngs modulusa

Rapid chloride penetrationb

Freezethawc

Specimen size (mm)

Performance

100  100  100

P130 MPa

100  100  400

P18 MPa

100  100  300

P48 GPa

100  100  50
<40 C

100  100  400

>500 cycles

Test according to GB/T 50081 2002 [19].

Test according to ASTM C1202 10 [20].
Test according to ASTM C666 Procedure A [21].

Table 2
Mix proportion of UHPC.

a
b

Cement

Fly ash

Silica fume

Sand to binder ratioa

Water to binder ratiob

Superplasticizer (% solid by weight of binder)

Fiber (% by volume)

0.5

0.35

0.15

1.2

0.16

0.98

Content of sand is given in oven-dry state.

Water in superplasticizer is included.

1 1  v 2 1  v 2i

Er
E
Ei

Fig. 2. XRD pattern of y ash.

3. Nanoindentation program
A nanoindentation test consists of establishing contact between
a sample and a diamond tip of known geometry and then continuously measuring the change in indentation depth h as a function of
increasing indentation load P. By applying a continuum scale
model to analyze the Ph curve, two mechanical properties, the
indentation (reduced) modulus Er and the indentation hardness
H, can be derived:

p
Er

p S

2b
Pmax
Ac

p
Ac

where S is the contact stiffness determined by the initial slope of

 
the unloading branch of the Ph curve, i.e., ,S dP
, b is the
dh hhmax
corrected factor for a non-symmetrical indenter (for Berkovich tip,
b = 1.034), and Ac is the projected contact area of the indenter on
the sample surface at peak load. Using the Oliver and Pharr method
[23], Ac can be determined as a function of the contact depth hc,
which can be calculated from the maximum indentation depth hmax.
For the tested material, the relationship between the elastic modulus E and the reduced modulus Er is given by:

where Ei and mi are the elastic modulus and Poissons ratio of the
diamond indenter (Ei = 1141 GPa, mi = 0.07), and m is the Poissons
ratio of the tested material. In this study, m was assumed to be
0.24 for all cementitious phases [14] and 0.3 for steel ber [24].
Therefore, the elastic modulus E for each indentation performed
in the UHPC can be determined using Eq. (3).
The use of nanoindentation in heterogeneous materials should
satisfy the scale separation requirements so that the relations (1)
and (2) can be used to correlate indentation quantities and constitutive properties [25]. For cement-based materials, to ensure that
the experimental results do not depend on any characteristic
length, the indentation depth should be much larger than the characteristic size of the elementary heterogeneity (e.g. globules of
CSH with a cross section of about 5 nm [11]), and much smaller
than the characteristic length of the microstructure (e.g. unhydrated cement with particle size of about several micrometers). In
addition, the indentation depth should be much larger than the
surface roughness which is usually described by root-mean-square
(RMS) roughness. For a well-polished cement paste sample, the
RMS roughness is normally below 50 nm [22]. Therefore, given
the multi-phase heterogeneous nature of cement-based materials
at the micro-scale, the scale separation requirements should be fullled in this study through a load-driven indentation test operated
to a maximum indentation load of 2 mN, which gives a mean
indentation depth of about 200 nm for all the phases in the UHPC
paste. A trapezoidal loading protocol was prescribed. Load was linearly increased from the initial contact of the diamond tip on the
sample to 2 mN at a rate of 12 mN/min; then the load was held
constant for 5 s to eliminate the creep effect before unloading
the sample at the same constant rate.
Nanoindentation tests were carried out on a CSM Nanoindentation Tester using a diamond Berkovich tip. A representative area
was rst determined under an optical microscope before indentation, and then the system moved the stages to bring the sample
into contact with the indenter tip at the optical position to run
the test. For the paste sample, a 20  30 grid containing 600 indentations was performed on a representative area. The grid spacing
was set as 20 lm to avoid potential interferences between neighboring indentations. For the UHPC sample, an aggregatematrix
ber (AMF) zone and an aggregatematrixaggregate (AMA)
zone were selected to study the mechanical properties at the interfacial zone. A rectangular indentation grid was performed at each
of these two zones. The indentations started from the aggregate,
then moved into the matrix, and nally nished on the ber or

S. Zhao, W. Sun / Construction and Building Materials 63 (2014) 150160

153

the other aggregate. The grid spacing in the direction parallel to the
aggregatematrix interface was set as 15 lm, while in the direction perpendicular to the interface it was 10 lm.
4. Results and discussion
4.1. Polishing and microstructure of the UHPC
Surface nish is of critical importance to obtain reliable nanoindentation results. As stated in ASTM E2546 [26], the surface nish
of the sample will directly affect the test results. The test should be
performed on a at specimen with a polished or otherwise suitably
prepared surface. Some prior studies found that the presence of
signicant surface roughness increased the variability in measured
mechanical properties, and resulted in an overall reduction in the
value of the measured properties [27,28]. For hardened cement
paste, its high heterogeneity at the micro-scale implies that a perfect surface polish cannot be obtained due to the differential rates
of abrasion of hard and soft phases. However, the obtained material properties were reported to converge to stable values when
the indentation depth of the softest phase (the LD CSH phase)
was greater than 5 times the RMS roughness [22]. Fig. 3a shows
a backscattered electron (BSE) image of the UHPC after polishing.
It is easy to discern inclusions of air voids, bers and aggregate
within the paste. Unhydrated cement particles and unreacted y
ash can also be readily distinguished. In addition, micro-cracks,
which could be formed during sample preparation, are visible in
the paste. On the whole, the picture shows a relatively at surface
with sharp edges between different constituents. The aggregate
and ber have very smooth surfaces while the paste exhibits relatively high roughness. No scratch is observed on the surface of the
sample, which indicates a good surface polishing. Through the RMS
roughness cannot be directly evaluated from the image, the surface
nish should satisfy the requirement for nanoindentation as the
sample preparation procedure (especially the polishing) is similar
to those adopted in other researches [17,22,25].
Fig. 3bc shows BSE photos of the UHPC paste at two different
length scales. Similar to the paste shown in Fig. 3a, unreacted
cement and y ash can be readily found but silica fume particles
are indistinguishable probably due to their extremely small particle size and high pozzolanic reactivity. However, a large silica fume
agglomerate with a particle size of near 100 lm can be discerned,
which indicates that silica fume was not fully dispersed after mixing. Further research should be carried out to nd an effective
method of dispersing silica fume. The majority of the unreacted
y ash particles are spherical in shape and are scattered throughout the paste. The particle size of the y ash ranges from several
hundreds of nanometers to about 20 lm. Note that there are four
different types of y ash particles: hollow cenospheres, lled precipitators, plerospheres which contain a series of smaller spheres,
and Fe-rich spheres which are bright in the image [29,30]. No CH
crystals are observed owing to their pozzolanic reaction with mineral admixtures.
4.2. Validation of the experimental data
After the test was completed, each indentation Ph curve was
examined to determine the validity of the experimental data.
Fig. 4 shows the typical Ph curves obtained from this study.
Fig. 4a presents an irregular curve that does not provide any information that is of value in determining the mechanical properties.
The tiny force and displacement values shown in the plot may
imply that there was no real contact between the indenter tip
and the sample, which can be explained by the presence of a
cenosphere y ash or a large air void. Another type of irregular

Fig. 3. BSE images of the polished UHPC sample (a) and UHPC paste sample (b and
c).

curve reported in some researches [17,31], which has a jump in

displacement due to fracture cracking, was not observed. Fig. 4b
shows a Ph curve with a false contact point that needs to be corrected. In addition, there are several Ph curves showing a slight
displacement jump at the end portion of the unloading curve
(Fig. 4c). These curves are considered valid because the initial slope

154

S. Zhao, W. Sun / Construction and Building Materials 63 (2014) 150160

Fig. 4. Different indentation loaddepth (Ph) curves: (a) irregular Ph curve shows no contact between the indenter tip and the sample; (b) Ph curve shows a bad contact
point that needs to be corrected; (c) Ph curve shows a slight displacement jump at the end of the unloading curve; (d) typical Ph curves for phases present in the UHPC.

of the unloading curve can still be acquired in order to calculate the

contact stiffness, S.
Nanoindentation can identify phases with different mechanical
properties, which is reected by a characteristic indentation
response for each phase. Fig. 4d shows the Ph curves of typical
phases in the UHPC. Steel ber, aggregate, cement and y ash,
which are clearly observed under the microscope, should be considered separate phases. Fly ash is a multi-phase material and only
the indentation response of the vitreous phase is presented here.
Two different hydration products, the LD CSH and HD CSH,
are generally regarded as the main hydration phases for cementbased materials. However, a recent study revealed presence of an
ultra-high density (UHD) phase especially in a low w/b material
[25], so in this study the UHD phase is considered a third hydration
product. Moreover, the pozzolanic reaction of y ash gives calcium
aluminum silicate hydrate (CASH), which is hard to be differentiated from CSH by nanoindentation. Therefore, the CASH is
not considered an independent phase here but part of the CSH.
Regarding the indentation responses of the aforementioned
phases, it can be seen that there are signicant differences between
the unreacted raw materials and hydration products. The maximum indentation depths of the three hydration products range
from 200 to 400 nm, which are much larger than those of the unreacted raw materials (generally less than 100 nm). Correspondingly,

the elastic moduli of the hydration products are much lower than
those of the unreacted raw materials. The E values given in Fig. 4d
are essentially in agreement with the reported values in the literature with the exception of the steel ber, which shows a higher
modulus than the reported value of about 300 GPa [16]. On the
one hand, this may be caused by the indentation size effect, which
dictates a decrease in the mechanical properties with an increase
in the indentation depth [32]. The higher maximum load (5 mN)
applied in [16] could lead to a lower modulus value. On the other
hand, this may result from differences in the composition of the
bers. In addition, from a Ph curve we can look at the elastic/plastic behavior of a material at nano-scale by comparing the maximum indentation depth with the residual depth of penetration.
The indentation responses of the hydration products and steel ber
are mostly irreversible plastic deformation while the elastic deformation comprises a signicant portion for the cement, y ash and
natural sand.

4.3. Nano-mechanical behavior of the paste

Of the 600 indentations performed on the paste sample, only 6
of them were excluded due to the irregular Ph curves as displayed
in Fig. 4a. Compared to the reported 515% of discarded data points

155

S. Zhao, W. Sun / Construction and Building Materials 63 (2014) 150160

Fig. 5. Experimental elastic modulus frequency of the UHPC paste. For purpose of
readability, the histogram is truncated at 200 GPa and the inset shows the
frequency of higher modulus values.

Table 3
Nanoindentation results from literature.

Phase

Elastic modulus, GPa

Indentation hardness, GPa

Source

LD CSH

18.2 4.2a
19.1 5.0a
23.4 3.4
22.89 0.76
19.7 2.5a

0.45 0.14
0.66 0.29
0.73 0.15
0.93 0.11
0.55 0.03

[31]
[38]
[39]
[15]
[35]

HD CSH

29.1 4.0a
32.2 3.0a
31.4 2.1
31.16 2.51
34.2 5.0a

0.83 0.18
1.29 0.11
1.27 0.18
1.22 0.07
1.36 0.35

[31]
[38]
[39]
[15]
[35]

UHD phase

41.45 1.75
48.0 3.3
39.146.1 9.0a

1.43 0.29
1.151.71 0.48

[15]
[40]
[25]

Quartz

76.3 15.1a

5.14 3.08

[35]

Fly ash

79.15 14.34
120.4 20.7

Cement

125145 25
122.20 7.85
126.3 20.16a
141.1 34.8a

Mullite

223.87 4.65

[34]
[41]
810.8 1
6.67 1.23
8.94 1.65
9.12 0.90

[12]
[15]
[22]
[35]
[42]

Reduced modulus Er is given.

[16,31], the low number (1%) may indicate a good surface nish of
the sample.
Fig. 5 displays the frequency histogram of the elastic modulus
results of the UHPC paste. The bin size of the histogram is 2 GPa.
It is observed that most of the experimental modulus values fall
into the range of 10110 GPa, and the rest of them scatter in the
range of 110410 GPa. To acquire the mechanical properties of
different phases, some prior studies employed the deconvolution
technique, which ts a number of Gaussian functions to the

experimental frequency plot [3335]. This method could be inaccurate, however, due to its dependence on assumptions about
the E values of different phases [36,37], and because signicant
error can be introduced during the tting procedure. For example,
it is hard to separate the hydration products into two or three
phases in Fig. 5 by the deconvolution technique because the hydration products part of the histogram does not present an apparent
bimodality or trimodality. Therefore, in this study, separation of
the nanoindentation results into different phases is determined
according to reported values in the literature (Table 3). This is reasonable because the results obtained from nanoindentation are
intrinsic mechanical properties of a material and should be consistent between different studies. In Table 3, the mechanical properties of the main phases in cementitious materials are given and it
can be found that for a given phase, its mechanical properties are
generally stable regardless of different curing methods, w/c, etc.
However, it should be noted that the nanoindentation response
of y ash in prior studies shows only one characteristic modulus
value, which is inconsistent with the XRD ndings. Constantinides
[43] reported that the intrinsic mechanical property could not be
accurately determined if the stiffness mismatch ratio between
the matrix and inclusion was more than ve. Therefore, a highstiffness phase could be ignored or misinterpreted by nanoindentation in a soft matrix. In Fig. 5, we can see that E values higher than
typical values of cement are acquired, which indicates that the
UHPC has a stiff matrix that allows for the measurement of the
high-stiff phases in y ash. For the phases in y ash, the modulus
of mullite is about 230 GPa, which is still lower than the upper
experimental value in Fig. 5. Therefore, these higher modulus values may result from indentation responses of another phase in y
ashhematite. The elastic modulus of hematite is rarely studied by
nanoindentation, but molecular dynamics analysis revealed an
elastic modulus of 359 GPa [44] which is quite close to the highest
experimental E value. Therefore, the modulus values that are
higher than typical values for cement are allocated to mullite
and hematite. These two high-stiffness phases are not further separated due to their low contents in the paste and the variability of
the experimental results. Moreover, the quartz and vitreous phase
in y ash have similar elastic modulus so they cannot be differentiated by nanoindentation analysis. Finally, as summarized in
Table 4, the experimental modulus results are divided into six sections, each of which represents one or two phases. The separation
attempted to comply with the reported values, and it should be
kept in mind that the elastic modulus boundaries between neighboring sections are our best estimates rather than denite values.
The proportions of indents for different phases, which can be
determined from the histogram, are also given in Table 4 and can
be considered the approximate volume fractions of different
phases in the material [31]. It can be concluded that if air voids
and capillary pores are not considered, the volume fraction of the
hydration products is almost equal to that of the unreacted materials in the UHPC paste. For a typical Portland cement, the minimum w/c for a complete hydration is 0.420.44 [45], so the
cementitious materials in the UHPC cannot be fully reacted. The
degree of hydration of the cement in the UHPC can be calculated
as rc = 1  vc,resid qc(w/b + fc/qc + ffa/qfa + fsf/qsf)/fc = 0.55, where
vc,resid is the volume fraction of the unhydrated cement, fc, ffa,
and fsf denote the mass fraction of cement, y ash and silica fume

Table 4
Elastic modulus and proportions of constituent phases in UHPC paste obtained from nanoindentation.
Phase

LD CSH

HD CSH

UHD phase

Fly ash (quartz and vitreous phase)

Cement

Fly ash (mullite and hematite)

Elastic modulus, GPa

Frequency, %

1025
4.5

2535
14.3

3550
32.5

5090
29.1

90150
13.3

>150
6.2

156

S. Zhao, W. Sun / Construction and Building Materials 63 (2014) 150160

hydration products accounts for 49.5% of the material by volume

from hardness analysis and the number is 51.3% with elastic modulus analysis. This indicates that the hardness property generally
follows the modulus property, and this may also support the validity of the phase assignments. Hardness is a measure of a materials
resistance to plastic deformation and is closely related to strength.
Therefore, it can be concluded that the unreacted phases have
much higher strength properties than the hydration products,
and thus contribute to the high strength properties of the paste.
Comparing the modulus and hardness proles in Figs. 5 and 6, both
histograms show a compact distribution and a high probability at
lower modulus/hardness values, which mainly represent the
hydration products, and a scattered distribution at higher modulus/hardness values. However, the overall distribution of the hardness is relatively broader than that of the elastic modulus. Fig. 7
plots the experimental indentation hardness H versus the elastic
modulus E. The plot shows a nonlinear scaling between H and E.
The hardness increases faster than the modulus, which is similar
to the relationship between the compressive strength and elastic
modulus of concrete at the macroscopic level. This resemblance
may conrm that the macro-performance of a material stems from
its behavior at micro- or nano-scale. From Fig. 7, It can also be
observed that for a given E value, the H value is scattered, and there
are quite a few data points showing mismatches between their E
and H values. For the hydration products whose E values are considered below 50 GPa, about 15% of the data points show H values
higher than 2 GPa, and these mismatched data points would be
classied as the unreacted phases based on their H values. Likewise, for E values between 50 and 90 GPa, there is about 20% of
the data points show H values lower than 2 GPa. It should also
be mentioned that such mismatches exist between the y ash
and cement phases. Some calculations [43] revealed that the linear
dimension of microvolumes probed by nanoindentation was about
10 times of hmax, so the probed microvolumes in the UHPC paste
can have a linear depth of about 0.54 lm, which shows the possibility of containing different phases. Chen [46] found that a
hydrate-clinker composite phase, for which the mechanical properties were higher than those of the hydration products but lower
than those of the clinker, could be observed as a result of a multiphase response. Therefore, the data points showing mismatches
between E and H could be composite responses from mixtures of
multiple phases. It is possible that some indents landed on the
boundaries of multiple phases. It is also possible that a few indents
was performed on a thin layer of one phase which has another hidden phase underneath.
For an individual indentation test, the result can be affected by
the local surface roughness as well as the composition and structure of the interaction microvolume. Therefore, even though the
results obtained from nanoindentation are the intrinsic mechanical
properties of different phases, it does not imply a constant value
for each phase. The surface roughness can be slightly different at
varied locations of the same phase, and the existence of some
defects such as complicated capillary pores in the interaction volume can affect the indentation response. This can partially explain
why the modulus/hardness values of a phase exhibit some variability and why there are no distinct boundaries between neighboring
phases in Figs. 5 and 6. Therefore, the analysis conducted in this
work, which separates the experimental results based on the

Fig. 6. Experimental indentation hardness frequency of the UHPC paste. For

purpose of readability, the histogram is truncated at 20 GPa and the inset shows the
frequency of higher hardness values.

in the binder, respectively, and qc, qfa and qsf (values are given in
[6]) represent the specic gravity of the cementitious materials.
Almost half of the cement remains unhydrated in the UHPC, which
is quite consistent with some quantitative XRD research results [9].
The degree of reaction of y ash is quite different from that of
cement. Even after heat treatment, the y ash is almost unreacted
due to its low pozzolanic activity which make it much less competitive than the silica fume to react with CH. Therefore, more water is
available for the hydration of cement after mixing and to some
extent the degree of hydration of cement is increased. For a normal
cement paste with a water-to-cement ratio (w/c) of 0.5 [31], statistics from grid nanoindentation revealed that the cement was fully
hydrated and the hydration products consisted of 57% LD CSH,
30% HD CSH and 13% CH. In contrast, more than 90% of the
hydration products in the UHPC paste are HD CSH and the
UHD phase, which indicates that the hydration products of the
UHPC have much higher mechanical properties over those of
ordinary concrete. In addition, as indicated by the nanoindentation
results, the unreacted particles have a much higher stiffness than
the hydration products, and some research reported that the bond
between the hydration products and unreacted particles was
strong [17], so the unreacted particles can function as high-stiffness micro-aggregates to strengthen the UHPC paste.
The frequency histogram of the indentation hardness results of
the UHPC paste is displayed in Fig. 6. The bin size of the histogram
is 0.2 GPa. Similar to the elastic modulus results, the hardness
results are classied into different phases and the statistics are
shown in Table 5. The hardness values for the hydration products
and cement can be found in the literature (Table 3) but those for
the phases in y ash are scant. Therefore, H values higher than
the cement are assigned to mullite and hematite, and H values
between the hydration products and cement are considered indentation responses of the vitreous phase and quartz in the y ash. It
can be found that the proportion of each phase presented in Table 5
is quite close to that derived from the elastic modulus analysis. The

Table 5
Indentation hardness and proportions of constituent phases in UHPC paste obtained from nanoindentation.
Phase

LD CSH

HD CSH

UHD phase

Fly ash (quartz and vitreous phase)

Cement

Fly ash (mullite and hematite)

Indentation Hardness, GPa

Frequency, %

0.20.6
3.9

0.61.1
12.8

1.12
32.8

28
30.0

815
13.6

>15
6.9

S. Zhao, W. Sun / Construction and Building Materials 63 (2014) 150160

157

Fig. 7. Plot of indentation hardness H versus elastic modulus E (E values higher than 300 GPa are not present). The inset shows an enlarged view of the hydration products.

reported values in the literature, cannot acquire accurate results

but should give approximate contents of different phases. However, since most of the y ash constituents show lower mechanical
properties than the cement, the possible composite responses from
the unreacted cement and the hydrated phases can be mistaken as
part of the y ash. This may lead to an underestimation of the
cement content and an overestimation of the y ash content.
From the experimental indentation results shown above, we
can see that the hydration products of the UHPC is dominated by
HD CSH and the UHD phase, both of which exhibit high stiffness
and hardness. Prior research [47] suggested that HD CSH formed
in a restricted space while LD CSH was prone to form in open
areas. This may shed some light on the underlying mechanisms
that control the formation of high-stiffness hydration products in
the UHPC. UHPC is essentially designed with a maximized packing
density of raw materials and an extremely low w/b, so the spaces
between the cementitious particles can be very conned. In this
restricted reaction space, the hydration products can pack more
efciently and thus develop higher mechanical properties. Since
fundamental properties of concrete, such as strength, shrinkage,
fracture behavior and durability, are basically governed by the pore
structure of the material and the properties of the hydration products, mainly the CSH, the high mechanical properties of the
hydration products in the UHPC paste should contribute to the
superior performance of the UHPC at macro-scale. In addition,
though the degree of reaction of cementitious materials in the
UHPC is low, these unreacted materials can act as micro-aggregates to strengthen the paste due to their much higher stiffness
and hardness than those of the hydration products.

used to characterize the properties of ITZ, nanoindentation, which

exhibits micro-structural gradients in terms of mechanical properties, proves to be a more sensitive and appropriate experimental
tool [1618].
The nano-mechanical behavior of the UHPC at the interfacial
zone was investigated on an AMF zone and an AMA zone.
Fig. 8 shows the BSE image of the AMF zone. The indents left
in the matrix of the UHPC can hardly be found probably due to
the high heterogeneous nature of the matrix. However, the indents
left in the aggregate and steel ber can be observed from the SEM
image as shown in Fig. 9, which helps to determine the position of
the whole indentation area. The size of the indentation grid at the
AMF zone is 190 lm  60 lm which contains 20 rows and 5
lines. The numbers shown in Fig. 8 demonstrate the sequences of
the indentations. The proles of the elastic modulus E and

4.4. Nano-mechanical behavior at the interfacial zone

Concrete is a composite material and its properties are governed by the matrix, aggregate and ITZ. The ITZ is generally considered a region decient in content of cementitious materials due to
the so-called wall effect, and therefore has a higher porosity compared to the bulk matrix [48]. The effective thickness of the ITZ varies with different mix proportions, mixing procedures, curing, etc.
Compared with electron microscopy which has been primarily

Fig. 8. BSE image of a representative aggregatematrixber zone, where the

rectangle depicts the prole of the indented area (190 lm  60 lm) and the
numbers show the sequences of the indentations.

158

S. Zhao, W. Sun / Construction and Building Materials 63 (2014) 150160

Fig. 9. SEM images of indents left on the aggregate (a) and ber (b).

indentation hardness H at the AMF zone are shown in Fig. 10,

where the E and H values are elucidated as a function of the distance from the starting indentation row. Note that the direction
of the indentation rows is almost parallel to the aggregatematrix
interface and bermatrix interface, so the mechanical behavior at
the interfacial zone could be evaluated through comparing the
mechanical properties of different rows. We can see that the values
representing the ber and aggregate are relatively concentrated,
but due to the heterogeneous nature of the paste and the signicant differences in mechanical properties between the hydrated
and unreacted phases, there is a noticeable scatter of E and H for
each row. The rows which contain a high content of the unreacted
phases generally give higher average E and H values than the rows
which contain a high portion of the hydration products. Therefore,
it seems that it makes little senses to evaluate the mechanical
properties at the interfacial zone based on the average values of
different rows unless a much larger number of intents is performed. However, if the data points which represent the unreacted
phases are excluded, the mechanical properties of the hydration
products across the whole zone appear to be stable and uniform,
and there is no obvious trough near the aggregate or ber surface.
There is only one indent whose distance from the aggregate
boundary is about 40 lm representing LD CSH (E = 16.5 GPa,
H = 0.51 GPa), and all the other hydration products at this zone
are recognized as HD CSH and the UHD phase. Formation of
these high-stiffness products implies a compact microstructure at

Fig. 10. Distributions of Elastic modulus E (a) and indentation hardness H (b) at the
aggregatematrixber zone in the UHPC.

the interfacial zone. Therefore, it can be concluded that the hydration products of the paste at the AMF zone show similar
mechanical properties to those of the UHPC paste. In addition,
the image in Fig. 8 shows a tight bond between the aggregate
and matrix as well as between the ber and matrix.
Fig. 11 shows the BSE image of the AMA zone, where a
160 lm  60 lm indentation grid containing 17 rows and 5 lines
was performed. As shown in the gure, the indentation row of
the rectangular grid is almost parallel to the bottom aggregate
matrix interface. The proles of the elastic modulus and indentation hardness at the AMA zone are shown in Fig. 12. Similar to
the analysis for the AMF zone, only the hydration products are
evaluated here. The hydration products at this zone on average
show higher mechanical properties than those at the AMF zone.
This could be caused by the method of adding the ber. Manually
placing the ber into the UHPC mortar after mixing might slightly
weaken the microstructure of the paste around the ber and thus
results in a lower mechanical performance. In Fig. 12a, there is only
one indent representing LD CSH with an E value of 24.2 GPa, and
in Fig. 12b, all the hydration products can be classied as HD
CSH and UHD. Again, there is almost no change for the mechanical properties of the hydration products as the aggregate surface is
approached, which indicates an efcient bond at the interfacial
zone. Moreover, the aggregate here shows comparable mechanical
properties to that in the AMF zone. There is a small defect on the
top aggregate, which is reected by a lower E and H values for
indent #71 compared with the other indents on the aggregate.

S. Zhao, W. Sun / Construction and Building Materials 63 (2014) 150160

Fig. 11. BSE image of a representative aggregatematrixaggregate zone, where the

rectangle depicts the prole of the indented area (160 lm  60 lm) and the
numbers show the sequences of the indentations.

159

Fig. 13. BSE image of an aggregatematrix interfacial zone in the UHPC.

18 GPa, which was about 85% of that of the paste matrix. Wang
[16] studied the ITZ of ber reinforced mortar by nanoindentation
and found that the low w/b (0.3) mortar had efcient interfacial
bond while the bond was poor for the high w/b (0.5) mortar. For
the UHPC in this study, the mechanical properties of the hydration
products at the interfacial zone is much higher compared to the
ordinary mortar, and the paste in the vicinity of the aggregate or
ber shows almost the same mechanical properties as those of
the bulk paste, which indicates that there is no transition of
mechanical properties at the interfacial zone in the UHPC. This is
mainly due to the use of silica fume, which densies the packing
at the interfacial zone and reacts with CH to form compact
CSH of high Si/Ca ratio around the aggregate and ber [4951].
Moreover, compared with ordinary concrete, the superior waterretention ability and high homogeneity of the fresh UHPC ensures
that there is no bleeding water gathering around the aggregate or
ber; otherwise a weak ITZ could be formed. Fig. 13 shows the
microstructure of an aggregatematrix interfacial zone. It can be
observed that the paste is densely packed against the aggregate
surface with the hydration products lling the interstices left by
the unreacted particles. Therefore, it is concluded that the UHPC
has a strong interfacial bond that efciently connect the matrix
with the aggregate and ber.
5. Conclusions
In this study, nanoindentation was employed to study the nanomechanical behavior of a green UHPC, which used high volumes of
y ash to partially replace cement and utilized river sand as ne
aggregate. Based on the experimental results, the following conclusions can be drawn:

Fig. 12. Distributions of elastic modulus E (a) and indentation hardness H (b) at the
aggregatematrixaggregate zone in the UHPC.

For a mortar with a w/c of 0.5, Mondal [17] found that the
Youngs modulus increased with an increase of distance from the
aggregate surface, and the average modulus of the ITZ was

1. While the hydration products of ordinary concrete are

mainly composed of LD CSH, the hydration products of
the UHPC are dominated by HD CSH and the UHD phase,
both of which have much higher stiffness and hardness
than LD CSH.
2. Due to the multi-phase nature of y ash, different indentation responses have been observed for y ash. The vitreous
phase and quartz show lower mechanical performances
than cement while the mechanical properties of mullite
and hematite are higher than those of cement. Noticeably,
most y ash and about half of cement remain unreacted
after curing. Theses unreacted particles show higher

160

S. Zhao, W. Sun / Construction and Building Materials 63 (2014) 150160

mechanical behavior than the hydration products and thus

can function as micro-aggregates to strengthen the UHPC
paste.
3. At the AMF zone and AMA zone, the mechanical properties of the hydration products do not change as the surface of the aggregate or ber is approached, which
indicates that there is no mechanical gradient for the
hydration products across the interfacial zone. Therefore,
the UHPC has a strong and efcient bond at the interfacial
zone.

Acknowledgment
Financial support from National Basic Research Program of
China (No. 2009CB623203) is gratefully acknowledged.
References
[1] Richard P, Cheyrezy M. Composition of reactive powder concretes. Cem Concr
Res 1995;25:150111.
[2] Yazc H, Yardmc MY, Aydn S, Karabulut AS. Mechanical properties of
reactive powder concrete containing mineral admixtures under different
curing regimes. Constr Build Mater 2009;23:122331.
[3] Yazc H, Yardmc MY, Yigiter Y, Aydn S, Trkel S. Mechanical properties of
reactive powder concrete containing high volumes of ground granulated blast
furnace slag. Cem Concr Compos 2010;32:63948.
[4] Tuan NV, Ye G, Breugel KV, Fraaij ALA, Dai BD. The study of using rice husk ash
to produce ultra high performance concrete. Constr Build Mater
2011;25:20305.
[5] Yang SL, Millard SG, Soutsos MN, Barnett SJ, Le TT. Inuence of aggregate and
curing regime on the mechanical properties of ultra-high performance bre
reinforced concrete (UHPFRC). Constr Build Mater 2009;23:22918.
[6] Zhao S, Fan J, Sun W. Utilization of iron ore tailings as ne aggregate in ultrahigh performance concrete. Constr Build Mater 2014;50:5408.
[7] Reda MM, Shrive NG, Gillott JE. Microstructural investigation of innovative
UHPC. Cem Concr Res 1999;29:3239.
[8] Cheyrezy M, Maret V, Frouin L. Microstructural analysis of RPC (reactive
powder concrete). Cem Concr Res 1995;25:1491500.
[9] Korpa A, Kowald T, Trettin R. Phase development in normal and ultra high
performance cementitious systems by quantitative X-ray analysis and
thermoanalytical methods. Cem Concr Res 2009;39:6976.
[10] Jennings HM. A model for the microstructure of calcium silicate hydrate in
cement paste. Cem Concr Res 2000;30:1016.
[11] Jennings HM. Renements to colloid model of CSH in cement: CM-II. Cem
Concr Res 2008;38:27589.
[12] Velez K, Maximilien S, Damidot D, Fantozzi G, Sorrentino F. Determination by
nanoindentation of elastic modulus and hardness of pure constituents of
Portland cement clinker. Cem Concr Res 2001;31:55561.
[13] Constantinides G, Ulm F-J, Vliet KV. On the use of nanoindentation for
cementitious materials. Mater Struct 2003;36:1916.
[14] Constantinides G, Ulm F-J. The effect of two types of CSH on the elasticity of
cement-based materials: results from nanoindentation and micromechanical
modeling. Cem Concr Res 2004;34:6780.
[15] Mondal P, Shah SP, Marks L. A reliable technique to determine the local
mechanical properties at the nanoscale for cementitious materials. Cem Concr
Res 2007;37:14404.
[16] Wang X, Jacobsen S, He J, Zhang Z, Lee S, Lein H. Application of
nanoindentation testing to study of the interfacial transition zone in steel
ber reinforced mortar. Cem Concr Res 2009;39:7015.
[17] Mondal P, Shah SP, Marks L. Nanoscale characterization of cementitious
materials. ACI Mater J 2008;105:1749.
[18] Li W, Xiao J, Sun Z, Kawashima S, Shah SP. Interfacial transition zones in
recycled aggregate concrete with different mixing approaches. Constr Build
Mater 2012;35:104555.
[19] GB/T 50081-2002. Standard for test method of mechanical properties on
ordinary concrete; 2002 [in Chinese].
[20] ASTM C1202 10. Standard test method for electrical indication of concretes
ability to resist chloride ion penetration. Annual book of ASTM standards, vol.
04.02. West Conshohocken, PA: ASTM International; 2010.
[21] ASTM C 666 03. Standard test method for resistance of concrete to rapid
freezing and thawing. Annual Book of ASTM standards, vol. 04.02. West
Conshohocken, PA: ASTM International; 2004.

[22] Miller M, Bobko C, Vandamme M, Ulm F-J. Surface roughness criteria for
cement paste nanoindentation. Cem Concr Res 2008;38:46776.
[23] Oliver WC, Pharr GM. An improved technique for determining hardness and
elastic modulus using load and displacement sensing indentation
experiments. J Mater Res 1992;7:156483.
[24] Howatson AM, Lund PG, Todd JD. Engineering tables and data. 2nd
ed. London: Chapman & Hall; 1991.
[25] Vandamme M, Ulm F-J, Fonollosa P. Nanogranular packing of CSH at
substochiometric conditions. Cem Concr Res 2010;40:1426.
[26] ASTM E 2546 07. Standard practice for instrumented indentation testing.
Annual book of ASTM standards, vol. 03.01. Philadelphia, PA: ASTM
International; 2007.
[27] Kim JU, Lee JJ, Lee YH, Jang J, Kown D. Surface roughness effect in instrumented
indentation: a simple contact depth model and its verication. J Mater Res
2006;21:29758.
[28] Donnely E, Baker SP, Boskey AL, Van der Meulen MCH. Effects of surface
roughness and maximum load on the mechanical properties of cancellous
bone measured by nanoindentation. J Biomed Mater Res A 2006;77:42635.
[29] Giere R, Carleton LE, Lumpkin GR. Micro and nanochemistry of y ash from a
coal red power plant. Am Miner 2003;88:185365.
[30] Matsunaga T, Kim J, Hardcastle S, Rohatgi P. Crystallinity and selected
properties of y ash particles. Mater Sci Eng A 2002;325:33343.
[31] Constantinides G, Ulm F-J. The nanogranular nature of CSH. J Mech Phys
Solids 2007;55:6490.
[32] Voyiadjis GZ, Peters R. Size effects in nanoindentation: an experimental and
analytical study. Acta Mech 2010;211:13153.
[33] Ulm F-J, Vandamme M, Bobko C, Ortega JA. Statistical indentation techniques
for hydrated nanocomposites: concrete, bone, and shale. J Am Ceram Soc
2007;90:267792.
[34] Nemecek J, milauer V, Kopecky L. Nanoindentation characteristics of alkaliactivated aluminosilicate materials. Cem Concr Compos 2011;33:16370.
[35] Sorelli L, Constantinides G, Ulm F-J, Toutlemonde F. The nano-mechanical
signature of ultra high performance concrete by statistical nanoindentation
techniques. Cem Concr Res 2008;38:144756.
[36] Trtik P, Mnch B, Lura P. A critical examination of nanoindentation on model
materials and hardened cement pastes based on virtual experiments. Cem
Concr Compos 2009;31:70514.
[37] Lura P, Mnch B, Trtik P. Validity of recent approaches for statistical
nanoindentation of cement pastes. Cem Concr Compos 2011;33:45765.
[38] Dejong MJ, Ulm F-J. The nanogranular behavior of CSH at elevated
temperatures (up to 700 C). Cem Concr Res 2007;37:112.
[39] Zhu W, Hughes JJ, Bicanic N, Pearce CJ. Nanoindentation mapping of
mechanical properties of cement paste and natural rocks. Mater Charact
2007;58:118998.
[40] Emmy M, Foley EM, Kim JJ, Taha MR. Synthesis and nano-mechanical
characterization of calciumsilicatehydrate (CSH) made with 1.5 CaO/
SiO2 mixture. Cem Concr Res 2012;42:122532.
[41] Sakulich AR, Li VC. Nanoscale characterization of engineered cementitious
composites (ECC). Cem Concr Res 2011;41:16975.
[42] Nath S, Dey A, Mukhopadhyay AK, Basu B. Nanoindentation response of novel
hydroxyapatitemullite composites. Mater Sci Eng A 2009;513514:197201.
[43] Constantinides G, Chandran KSR, Ulm F-J, Vliet KJV. Grid indentation analysis
of composite microstructure and mechanics: principles and validation. Mater
Sci Eng A 2006;430:189202.
[44] Chicot D, Mendoza J, Zaoui A, Louis G, Lepingle V, Roudet F, et al. Mechanical
properties of magnetite (Fe3O4), hematite (a-Fe2O3) and goethite (a-FeOOH)
by instrumented indentation and molecular dynamics analysis. Mater Chem
Phys 2011;129:86270.
[45] Taylor HFW. Cement chemistry. 2nd ed. London: Thomas Telford; 1997.
[46] Chen JJ, Sorelli L, Vandamme M, Ulm F-J, Chanvillard G. A coupled
nanoindentation/SEM-EDX study on low water/cement ratio Portland
cement paste: evidence for CSH/CH nanocomposites. J Am Ceram Soc
2010;93:148493.
[47] Smilauer V, Bittnar Z. Microstructure-based micromechanical prediction of
elastic properties in hydrating cement paste. Cem Concr Res
2006;36:170818.
[48] Escadeillas G, Maso JC. Approach of the initial state in cement paste, mortar,
and concrete. In: Mindess S, editor. Advances in cement materials, ceramic
transactions, vol. 16. American Ceramic Society; 1991. p. 16984.
[49] Scrivener KL, Crumbie AK, Laugesen P. The interfacial transition zone (ITZ)
between cement paste and aggregate in concrete. Interface Sci
2004;12:41121.
[50] Scrivener KL, Bentur A, Pratt PL. Quantitative characterization of the transition
zone in high strength concretes. Adv Cem Res 1988;1:2307.
[51] Chan Y, Chu S. Effect of silica fume on steel ber bond characteristics in
reactive powder concrete. Cem Concr Res 2004;34:116772.