1Running head SULFATE ATTACKA potential research topic on a .docx

1
Running head: SULFATE ATTACK
A potential research topic on a human-services-related topic2
Sulfate attack
Student’s name
Institutional affiliation
Sulfate Attack Sulfate attack in mortar and concrete
Sulfate attack in motor and concrete may be ‘Internal’ or
‘external’. The internal attack is caused by incorporating
soluble source into concrete during the time of mixing, for
example gypsum in aggregate. The external sulfate attack is
caused when sulfates penetrate in solution, for instance in
ground water, into concrete from the outside (Cohen, 2006).
External sulfate attack
External sulfate attack is the most familiar type and
characteristically takes place where water that contains sulfate
which is dissolved infiltrates into concrete. A practically well
illustrated reaction frontage may often be viewed in sections
that are polished; further front concrete is ordinary, or close to
ordinary. At the back of reaction front, the concrete’s
microstructure and composition would have altered. The
changes can vary in severity or type, but generally include; loss
of tie between aggregate and cement paste; wide cracking, and

expansion.
Changes in the composition of paste take place, with mono-
sulfate stage changing to ettringite and in afterward phases,
formation of gypsum. The required extra calcium is given by
calcium silicate hydrate and calcium hydroxide, contained in
cement paste. The consequence of these alterations is a general
loss of the strength of concrete. These effects are characteristic
of attack from solutions of potassium sulfate or sodium sulfate.
Solution
s that contain magnesium sulfate are normally extra aggressive,
for equivalent concentration. The reason for this is that;
magnesium too plays a role within the reaction, by substituting
the calcium contained in solid stages, with brucite formation,
and ‘magnesium silicate hydrates. The calcium that is displaced
precipitates principally as gypsum (Douglas & Mary, 2013).
There are more sulfate sources that may lead to sulfate attack.
These sources include action of bacteria in sewers; the
anaerobic bacteria create sulfur oxide that liquefy in water and
afterwards oxidizes hence forming sulfuric acid. The other
source is seawater. Regarding masonry, bricks contain sulfates,
and may be released gradually over a lengthy time period,
resulting to mortar sulfate attack, particularly in cases in which
sulfates are intense because of movement of moisture. Another
source is sulfide minerals’ oxidation in clay that is next to
concrete. The result can be sulfuric acid that the concrete reacts

with.
Internal sulfate attack
The internal sulfate attack takes place in situations where a
sulfate source is integrated into concrete in mixing for instance
utilizing aggregate that has high sulfate, cement that has
gypsum that is excessively added or contamination. Appropriate
testing and screening measures should normally evade internal
sulfate attack. “Delayed ettringite formation” is a unique case
regarding the internal sulfate attack. In most country, DEF
continues to be an important problem. It takes place in concrete
that is cured at high temperatures, for instance in situations
where curing has been done by steam.
DEF was initially recognized in concrete that had been steam
cured, for railroad ties. It can also take place in big concrete
pour in which hydration heat has led to elevated temperatures
inside the concrete. DEF leads to concrete expansion because of
formation of ettringite inside the paste. This may lead to grave
damage to structures made of concrete. DEF isn’t regularly
caused by surplus sulfate inside cement, or also from source
further than cement contained in concrete. Even though surplus
sulfate in cement has the likelihood of increasing expansion
because of DEF, it may take place at usual cement sulfate level.
In comprehending DEF, ettringite is damaged through heating
beyond approximately seventy degrees Celsius.
Belated ettringite formation

DEF takes place if ettringite that usually shape in the stage of
hydration becomes decayed, then consequently forms again in
the toughened concrete. Destruction of concrete takes place if
ettringite crystal applies a force that is expansive, inside the
concrete, while growing. In usual concrete, the entire amount of
ettringite that shapes is clearly restricted by sulfate that was
initially contributed by the cement. It takes that ettringite
quantity forms, is comparatively small. The form of ettringite
crystals are widely dispersed all through the paste. If cracking
is caused by expansion, ettringite can consequently form in
cracks. Nevertheless, this doesn’t mean that ettringite within the
cracks was initially the cause of the cracks (Sidney, 2009).
DEF leads to a characteristic type of harm to concrete. As the
paste is expanding, the aggregate is not expanding. Cracks
shape around the places that are not expanding in the paste and
the larger the aggregate, the larger the gap. There are some
conditions that are required in order for DEF to take place.
First; there is high temperature, normally during the curing
process, but not essentially. The second condition is water:
permanent or intermittent saturation after the curing process.
The other is normally linked with ‘Alkali Silica Reaction’. By
performing laboratory experiment, the limestone aggregate that
is coarse as been seen to minimize expansion.
DEF normally takes place in concrete that is steam cured, or
that has attained high temperature in the curing process due to

exothermic result of cement hydration. As the concrete curing
temperature raises, ettringite usually continues to approximately
seventy degrees Celsius. When this temperature is exceeded, it
decomposes. In concrete that is mature, mono-sulfate is
normally the major ‘sulfate containing hydrate’ stage, and this
continues to approximately one hundred degrees Celsius. Also
in experiment, DEF did not take place in concrete that was
exposed to temperature externally, for instance from fire.
Usual sulfate attack normally leads to ettringite formation. This
utilizes aluminum given by cement and evidently, this is
restricted in quantity in usual concrete. Nevertheless, formation
of thaumasite does not engage aluminum. If there is sufficient
supply of carbonate and sulfate, thaumasite may proceed
forming until the point where ‘calcium silicate hydrate’ is fully
decomposed. As a result, while utilization of Portland cement
that is sulfate resisting facilitates a degree of defense against
the usual sulfate attack, but it does not provide any specific
defense against formation of thaumasite. Sulfate may be
distributed form a variety of sources for instance bricks and
ground water. Carbonate may be distributed from atmospheric
carbon dioxide, or from the limestone that exists in mortar or
concrete (Meida, 2008). Grave damage to masonry or concrete
because of formation of thaumasite is not usual, even in damp,
cool climates.
In conclusion, sulfate attack depicts substance breakdown

mechanism in which ‘sulfate ions’ attacks parts of cement
paste.The compounds that are accountable for the attack are
salts that contain sulfate and are water soluble, for instance
alkali earth which include magnesium and calcium, and alkali
which include potassium and sodium, sulfates which have the
capability of reacting chemically with concrete components.
Sulfate tact can also display itself in various forms based on;
the sulfate chemical form, and the atmospheric condition of
exposing concrete.
References
Douglas, C. & Mary A, W. (2013). Cement: Its Chemistry and
Properties. Journal of Chemical Education, Vol. 80, No. 6 June
2013.
Cohen, D. (2006). Does gypsum formation during sulfate attack
on concrete lead to expansion? Cement and Concrete Research
30 (1): 117–123.
Sidney, Y. (2009). A realistic molecular model of cement
hydrates. Proceedings of the National Academy of Sciences 106
(38): 16102–16107.
Meida, L. (2008). Carbon dioxide emissions from the global
cement industry. Annu. Rev. Energy Environ. 26: 303–329.

Chemistry for Everyone
JChemEd.chem.wisc.edu • Vol. 80 No. 6 June 2003 •
Journal of Chemical Education 623
One of the most active areas in scientific research is the
development of new and exciting materials for a wide vari-
ety of applications. In this context, it could be easy to lose
sight of the importance of more common materials that are
vitally important in many areas of our lives. Cement is one
such material, and its rich chemistry links well with a num-
ber of concepts in most undergraduate chemistry curricula.
This paper addresses several important questions con-
cerning cement, including: What is its optimal composition
and why? Why do cement truck barrels roll? What are the
processes involved in cement setting, and how long does it
take? How does cement break down?
A Brief History of Cement
Cements and cement-containing materials comprised
some of the first structural materials exploited by humanity
(1), as cement’s components are common materials: sand,

lime, and water. On a molecular level, cement is a paste of
calcium silicate hydrates polymerized into a densely cross-
linked matrix (2). Its most important property is called
hydraulicity—the ability to set and remain insoluble under
water (3, 4). Cement can be used as a mortar to bind large
stones or bricks. When sand and stones are added to cement,
the aggregate is called concrete. The word cement comes from
the Latin phrase, opus caementum, or chip work, in reference
to the aggregate often used in applications (3).
Cement production dates back to the ancient Romans,
who produced mortars using a mixture of lime, volcanic ash,
and crushed clay. These cements are referred to as Pozzolanic
cements after the Pozzulana region of Italy, which contained
Italy’s chief supply of ash (1, 5 ). Pozzolanic cements derive
their strength from rich aluminate phases present in the vol-
canic ash that promote efficient hydration of the final ce-
ment powders (6). Fine grinding and attention to consistency
are also fundamental to the success of Roman cement, much
of which is still in existence today in structures such as the
Pantheon, the Pont du Gard, and the Basilica of
Constantinople (2, 5 ). An example of a structure made with
Roman cement is shown in Figure 1.
The art of cement production was lost in Europe after

the fall of the Roman Empire (2, 5). At that time, the access
to volcanic ash was limited and the grinding and heating tech-
niques required for cement precursor production were lost.
Cements of this period, if still in existence, are inconsistent
in composition and are composed almost exclusively of un-
reacted starting materials (1, 2, 5). There was no significant
breakthrough in the development of cement chemistry until
1756, when Smeaton was commissioned to rebuild the
Eddystone lighthouse in Cornwall, England. In contrast to
the methods of his contemporaries, Smeaton found superior
results through experimentation by using an impure lime-
stone with noticeable clay deposits. This produced extremely
strong cement “that would equal the best merchantable Port-
land stone in solidity and durability”(5).1
Another major advance came in the early 19th century
when the French engineer Vicat performed the first empiri-
cal study on the composition of cements. Although crude
and incomplete, it was one of the most comprehensive ex-
aminations of cement chemistry for the next 80 years (3, 4,
8–10).
The term Portland cement did not become officially rec-
ognized until 1824 when Aspidin filed the first patent for its

production (2, 5). Cement compositions at this time were
poorly understood but closely guarded secrets. Portland ce-
ment was introduced into the United States by Saylor in 1871
(3, 4).
By the start of the 20th century, cement manufacture
was common but was still regarded as more of an art than a
science. Emphasis was placed on bulk manufacture, not qual-
ity control or consistency (10, 11). Early in the 20th cen-
tury, cement research became more scientific, incorporating
the relatively new Gibbs phase rule and Le Châtelier equi-
librium principles (3). In 1904 the first set of ASTM stan-
dards2 for cement were presented and in 1906 the geophysical
laboratory of the Carnegie Institution began an extensive in-
vestigation of cement chemistry. These advances resulted in
the development of uniformity in the cement industry, al-
lowing a rapid expansion in the application of cement to large
construction projects such as skyscrapers, roads, and dams
(2, 3, 8, 11).
Cement: Its Chemistry and Properties
Douglas C. MacLaren and Mary Anne White*
Department of Chemistry and Institute for Research in
Materials, Dalhousie University, Halifax, Nova Scotia B3H 4J3,
Canada; *[email protected]

Products of Chemistry
edited by
George B. Kauffman
California State University
Fresno, CA 93740
Figure 1. Roman aqueduct in Segovia, Spain, from the first cen-
tury C.E. Courtesy Stephen L. Sass. Reproduced, with
permission,
from ref 1.
http://jchemed.chem.wisc.edu/
http://jchemed.chem.wisc.edu/Journal/Issues/2002/
http://jchemed.chem.wisc.edu/Journal/
624 Journal of Chemical Education • Vol. 80 No. 6 June 2003
• JChemEd.chem.wisc.edu
More recent advances in materials-characterization tech-
niques, such as X-ray crystallography, electron microscopy,

nuclear magnetic resonance spectroscopy, Mössbauer spec-
troscopy, infrared spectroscopy, and thermal analysis, have
allowed the systematic examination of cement’s chemistry and
the complex processes surrounding its production and hy-
dration (2, 12). Scientific research has led to a better under-
standing of the properties of cement, cement production, and
cement corrosion. In fact, breakthroughs in cement research
have provided us with cements of increasing quality and
strength.
Cement is prepared in a two-step process. The first step
is the high-temperature mixing and processing of limestone,
sand, and clay starting materials to produce a cement pow-
der. The second step involves the hydration, mixing, and set-
ting of the cement powder into a final cement product (2, 6,
13). The dry portion of Portland cement is composed of
about 63% calcium oxide, 20% silica, 6% alumina, 3%
iron(III) oxide, and small amounts of other matter includ-
ing possibly impurities (7). Calcium silicates and calcium alu-
minates dominate the structure.
The cement literature uses abbreviations for the many
calcium oxide, silicate, aluminate, and ferrate compounds
important to cement. We have used the same abbreviations
here and present the correspondence between the chemical

formulas and abbreviations in Table 1 (14).
Cement Formation
Preparation of Cement Precursors: Clinkers
The raw materials for cement production are blended
in the required proportions, ground, and heated to high tem-
peratures, usually with rotation. Heating first releases H2O
and CO2 and then causes other reactions between the solids,
including partial melting. Cooling results in clinkers, a term
from the coal industry in the 19th century to describe stony,
heavily burnt materials that were left after the burning of coal
(7). Ironically, Aspiden and Vicat both dismissed the hard
glassy clinker material (which was expensive to grind) as be-
ing useless to cement manufacture (8, 11), although we now
know that clinkers are essential for good cement production.
After heating, cement clinkers are reground for use in the
production of cement. Commercial cement manufacture in-
corporates a wide variety of minerals, including: calcium ox-
ide, silica, alumina, iron oxide, magnesium oxide, titanium
dioxide, and many others (5, 14). Of these, three are most
important to the final cement product: calcium oxide, silica,
and alumina. Consideration of all the possible phases pro-
duced by these multicomponent systems is simplified by con-

sidering a ternary system of primary importance—the calcium
oxide�silica�alumina system (14).
High-quality cement powders require the presence of two
major components, tricalcium silicate, ‘C3S’, and dicalcium
snoitaiverbbAdna,ealumroF,snoitisopmoC,semaNtnenopmoCtne
meCnommoC.1elbaT
emaNtnenopmoC noitisopmoC alumroFlaciripmE noitaiverbbA
)emil(edixomuiclaC OaC OaC ’C‘
)acilis(edixoidnociliS OiS 2 OiS 2 ’S‘
)animula(edixomunimulA lA 2O3 lA 2O3 ’A‘
edixo)III(norI eF 2O3 eF 2O3 ’F‘
etacilismuiclaciD OaC2 � OiS 2 aC 2 OiS 4 C‘ 2 ’S
etacilismuiclacirT OaC3 � OiS 2 aC 3 OiS 5 C‘ 3 ’S
etanimulamuiclacirT OaC3 � lA 2O3 aC 3 lA 2O6 C‘ 3 ’A

)etirellimnworB(etarrefonimulamuiclacarteT OaC4 � lA 2O3�
eF 2O3 aC 4 lA 2 eF 2O 01 C‘ 4 ’FA
legetardyhetacilismuiclaC )OaC( x� OiS 2�yH2 htiwO x 5.1<
)HO(aChtiwnoitulosdilosni 2
)elbairav( ’HSC‘
)etinotsalloW(etacilismuiclaC OaC � OiS 2 OiSaC 3 ’SC‘
)etiniknaR(etacilismuiclaC OaC3 � OiS2 2 aC 3 iS 2O7 C‘ 3S2’
)etinelheG(etacilismunimulamuiclaC OaC2 � lA 2O3� OiS 2
aC 2 lA 2 OiS 7 C‘ 2 ’SA
)etilluM(etacilismuinimulA lA3 2O3� OiS2 2 lA 6 iS 2O 31 A‘
3S2’
)etihtronA(etacilismunimulamuiclaC OaC � lA 2O3� OiS2 2
lAaC 2 iS 2O8 SAC‘ 2’
etacilismuinimulA lA2 2O3� OiS2 2 lA 4 iS 2O 01 A‘ 2S2’
etanimulamuiclaC OaC � lA 2O3 lAaC 2O4 ’AC‘

etanimulaidmuiclaC OaC � lA2 2O3 lAaC 4O7 AC‘ 2’
etanimulatpesmuiclacacedoD OaC21 � lA7 2O3 aC 21 lA 41 O
33 C‘ 21 A7’
etanimulaxehmuiclaC OaC � lA6 2O3 lAaC 21 O 91 AC‘ 6’
silicate, ‘C2S’, in the clinkers. These materials react vigorously
with water to produce the cement paste formed in the final
product. Of the two, tricalcium silicate is the more desirable
clinker material because it hydrates and sets much faster than
dicalcium silicate (hours for ‘C3S’, days for ‘C2S’) (2, 15).
The binary phase diagram of SiO2 and CaO is shown
in Figure 2 (16, 17). Most important is the 0–30 mass %

SiO2 region. ‘C3S’ is formed at less than 30 mass % SiO2
but is not stable below about 1250 �C or above about 2200
�C. In the low end of this temperature range ‘C3S’ will form,
but extremely slowly because it involves a reaction between
two solid phases. For example, forming ‘C3S’ at temperatures
of 1200–1400 �C would require heating for days and is not
economical. At the other end, production from the melt at
2200 �C is also impractical because of the very high tempera-
ture.
Therefore, the temperature of the ‘C3S’ production for
the clinker is lowered by fluxing3 the reaction mixture with a
third component, alumina (7, 14, 15). The binary phase dia-
gram of CaO and Al2O3 is shown in Figure 3 (16). Com-
parison with the ternary CaO�SiO2�Al2O3 phase diagram (7,
14, 16), Figure 4, shows that the addition of Al2O3 lowers
the preparation temperature of ‘C3S’.
For this discussion, the important region of the
CaO�SiO2�Al2O3 phase diagram is the ‘C3S’�‘C2S’�‘C3A’
phase field, the region close to the CaO vertex in Figure 4. A
three-dimensional view of the ternary phase diagram in this
region is shown in Figure 5. As the temperature of the sys-
Figure 3. The binary phase diagram of calcium oxide and

alumina
(Al2O3). The temperature of the liquidus of the binary system
de-
creases significantly as Al2O3 is added to the mixture (13).
Al 2O3
1400
1800
2200
1000
CaO + ‘C 3A’
‘C 3A’ + L
‘C 3A’
+ ‘C 12A 7’
‘C12 A 7’
+ ‘CA’

‘CA’ +
‘CA 2’
‘CA 2’
+
‘CA 6’
‘C
A
6 ’ +
A
l
2 O
3
Al 2O3
+ L
‘CA 6’+ L
‘CA 2’+ L

‘CA’ + L
‘C12A7’ + L
CaO + L
‘C3 A’ ‘C12 A7’ ‘CA’ ‘CA2 ’ ‘CA6 ’
40 8020 600 100
mass % Al2O3
T
e
m
p
e
ra
tu
re
/
°

C
(CaO) ( )
Figure 2. The binary phase diagram of calcium oxide and silicon
dioxide. The region of interest is 0–30 mass % SiO2 where
tricalcium silicate (‘C3S’) is formed (14).
1500
2500
1000
2000
20 40 60 80
(CaO) SiO2( )
‘C3S’ ‘C2S’ ‘C3S2’ ‘CS’
α-‘C2 S’ + L
CaO + ‘C 3S’

α -‘CS’ + Tridymite
β-‘CS’ + Tridymite
Tridymite + L
Cristabolite + L
Two
Liquids
α -‘CS’+ L
‘C 3S2’ + L
‘C
3 S
2 ’+
α
-‘C
S

’
‘C 3S2’+
β -‘CS’
‘C
3 S
2 ’+
β
-‘C
2 S
’
‘C 3S’ +
α-‘C 2S’
C3S2+
α-‘C2S’
‘C 3S’ +
β-‘C 2S’
CaO + β-‘C 2S’

CaO + L
‘C 3S’ + L
Cristabolite + L
0 100
mass % SiO2
T
e
m
p
e
ra
tu
re
/
°
C

Figure 5. Three-dimensional view of the CaO-rich portion of the
ternary phase diagram of calcium oxide, silica, and alumina em-
phasizing the tricalcium silicate primary phase field. The
composi-
tion of the liquid will follow the minimum path along the
liquidus,
which deeply slopes into the tricalcium silicate phase field as
the
temperature of the system is lowered from 2150 �C to 1450 �C
(5).
Figure 4. The ternary phase diagram of calcium oxide, silica,
and
alumina. The region nearest the CaO vertex represents the
primary
phase field for the formation of tricalcium silicate.
Temperatures
are presented in �C (13, 16).
x
x
x

x
x
x
xxx x x x x x x
Lime
(CaO)
‘C3S’
�-‘C2S’
G
ehlenite
A
S
’)
‘C3A’
‘C2AS’

(‘C
2
‘CA’
‘CA2’ ‘CA6’
Corundum
(�-Al2O3)
‘CAS2’
(‘CAS2’)
Anorthite
‘C3S2’
� -‘CS’
Tri
dym
ite

C
ris
ta
bo
lit
e
‘A2S2’
‘A
3 S
2 ’
‘CA6’‘CA2’‘CA’‘C12A7’ ‘C3A’
‘C3S’
‘C2S’
‘C3S2’
‘CS’

Two
liquids
SiO2
Al2O3CaO
2570
1470
1455
1350
1335
1335
1542 1415 1605 1789 1860
1840
1850
1512
1380

1590 1552
1500
1475
1405
1380
1385
1553 1547
1512
1265
1318
1310
1307
1315
1545
1170

1345
1368
1470
1598
1698
1723
1698
1470
1436
1544
1460
1464
2130
2050
2150

M
ullite
(
)
tem decreases from about 2100 �C, the composition of the
liquid goes into the ‘C3S’�‘C2S’�‘C3A’ phase field (5). Add-
ing 20 mass % alumina to a silica�lime system lowers the
liquidus into the region of stable ‘C3S’ formation, from the
reaction of ‘C2S’ and CaO in the liquid phase. This, of course,
is much faster than the solid–solid reaction (5, 7, 15, 18).
Therefore, heating a composition in the ‘C3S’ phase field to
1450–1500 �C results in a liquid phase that can be quenched

to form the final ‘C3S’-rich cement clinker. Although nei-
ther Smeaton nor the Romans fully realized the chemistry, it
was the addition of rich aluminate matter in the form of vol-
canic ash or clay impurities that allowed their production of
strong cement precursors (5).
The total process of cement-clinker formation is sum-
marized in Figure 6, which shows the main components as a
function of temperature (18). Calcium carbonate (limestone),
quartz, clay (primarily Al2O3), and water are combined and
heated. (Iron oxide, clay, and other minor components are
neglected in this discussion.) As the temperature rises, first
water is lost, and then above 700 �C, the limestone decom-
poses forming CaO and carbon dioxide. CaO reacts with
silica to form ‘C2S’ and with the aluminate phases to form a
calcium aluminate phase (an Ettringite phase4), which melts
at about 1450 �C (18). The formation of this liquid phase is
associated with the rapid production of tricalcium silicate.
The final mixture at 1500 �C is primarily tricalcium silicate
with smaller portions of dicalcium silicate, aluminate, and
aluminoferrate phases.
The minor components present in cement paste (e.g.,
iron oxide) have only subtle effects on the properties of the
final cement properties (5, 14). One of the reasons for using

them is that they also help to flux the system to a lower tem-
perature. Table 2 lists a group of multicomponent clinker ma-
terials in the ‘C3S’ phase field. It is apparent that adding small
amounts of other minerals can lower the temperature at which
a liquid phase is formed (5).
After quenching, the resulting clinker is milled and
ground into a fine powder. At this stage various other mate-
rials can be added to the cement powder prior to packaging.
Cement Hydration
Cement hydration is a familiar process. The cement pow-
der is mixed with water and then is poured for the desired
application. The final cement product generally contains
about 30–40 mass % water after hydration, and this value
varies little with the composition of the cement clinker. Al-
though it might appear simple, cement hydration consists of
a complex series of chemical reactions, which are still not
completely understood (13). Cement hydration rates can be
affected by a variety of factors, including: the phase compo-
sition of the clinker, the presence of foreign ions, the spe-
cific surface of the mixture, the initial water:cement ratio,
the curing temperature, and the presence of additives (13,
18).

The rate of hydration of ‘C3S’ in a Portland cement clin-
ker is shown in Figure 7. Immediately upon contact with
water ‘C3S’ undergoes an intense, short-lived reaction, the
pre-induction period (I). The rate (dα/dt, where α is the de-
gree of hydration or the fraction of cement precursor mate-
rial that has been hydrated) is as high as 5 day�1. This process
begins with the dissolution of ‘C3S’. Oxygen ions
on the sur-
face of the ‘C3S’ lattice react with protons in the water and
form hydroxide ions, which in turn combine with Ca2� to
form Ca(OH)2 (13):
OH�(aq)O2�(lattice) + H�(aq) (1)
Ca(OH)2(aq)2 OH
�(aq) + Ca2�(aq) (2)
reknilC.2elbaT stnenopmoC C‘ehtni 3 esahPtnemeC’S
noitamroFdiuqiLfoserutarepmeTriehTdnadleiF )5(
stnenopmoC
foerutarepmeT

/noitamroFdiuqiL �C
OiS–OaC 2 5602
OiS–OaC 2 lA– 2O3 5541
OiS–OaC 2 lA– 2O3 aN– 2O 0341
OiS–OaC 2 lA– 2O3 OgM– 5731
OiS–OaC 2 lA– 2O3 eF– 2O3 0431
OiS–OaC 2 lA– 2O3 aN– 2 OgM–O 5631
OiS–OaC 2 lA– 2O3 aN– 2 eF–O 2O3 5131
OiS–OaC 2 lA– 2O3 eF–OgM– 2O3 0031
OiS–OaC 2 lA– 2O3 aN– 2 eF–OgM–O 2O3 0821
Figure 6. A schematic view of the components of cement-
clinker
formation, their reactions, and the products formed as the
tempera-
ture of the mixture is raised. Calcium carbonate decomposes to

form calcium oxide and carbon dioxide. Calcium oxide reacts
with
silica to form dicalcium silicate at temperatures below 1250
�C,
which converts to tricalcium silicate at temperatures above 1250
�C. Formation of a liquid aluminate, Ettringite, phase at about
1450
�C facilitates the conversion of dicalcium silicate to tricalcium
sili-
cate (18).
At the same time, silicate material from the ‘C3S’ lattice sur-
face enters the liquid phase (13):
HnSiO4

(4-n)�(aq)SiO4
4�(lattice) + n H�(aq) (3)
The dissolved components combine to form the calcium
silicate hydrate ‘CSH’ gel, an amorphous two-component
solid solution composed of Ca(OH)2 and a calcium silicate
hydrate of low Ca:Si ratio, hydrated as in this example (13,
19, 20):
3 CaO�2SiO2�3 H2O(s) + 3 Ca(OH)2(aq)
2 (3CaO�SiO2)(s) + 6 H2O(l)
(4)
However, the reaction would not likely be of this exact sto-
ichiometry.
Most cement powders have gypsum (CaSO4) added prior
to packaging. Gypsum acts to slow down the pre-induction
period to avoid rapid setting of the cement (3, 8). It reacts
with tricalcium aluminate (‘C3A’) to form various aluminate
and sulfoaluminate phases, collectively referred to as Ettringite
phases (7, 13, 15, 19). Some examples are:

3CaO�Al2O3�3CaSO4�32 H2O(s)
3CaO�Al2O3(s) + 3 CaSO4(s) + 32 H2O(l)
(5)
3CaO�Al2O3�3CaSO3�12 H2O(s)
3CaO�Al2O3(s) + 3CaSO4(s) + 12 H2O(s)
(6)
‘C3A’ and ‘C4AF’ can also hydrate independently of calcium
sulfate:
3CaO�Al2O3�6 H2O(s)3CaO�Al2O3(s) + 6H2O(l) (7)
3CaO�Al2O3�6 H2O(s) + 3CaO�Fe2O3�6 H2O(s)
4 CaO�Al2O3�Fe2O3(s) + 2 Ca(OH)2(aq) + 10 H2O(l)
(8)
During the pre-induction period about 5–25% of the ‘C3A’
and ‘C4AF’ undergoes hydration, causing a saturation of
Ettringite in the solution (13).
After a few minutes of hydration an induction period

(II in Figure 7) begins where the reaction slows signifi-
cantly, dα/dt = 0.01 day�1. The exact reason for this in-
duction period is not known. Several theories have been
proposed that involve some sort of mixture saturation from
the intense burst of hydration in the pre-induction period
(13). One theory states that the ‘CSH’ layer quickly cov-
ers the surface of dissolving ‘C3S’, slowing the reaction.
As time passes, the ‘CSH’ becomes more permeable and
the reaction accelerates. Another theory states that the so-
lution may become supersaturated with Ca(OH)2 because
the surfaces of Ca(OH)2 crystal nuclei are poisoned by sili-
cate ions. The high concentration of aqueous Ca(OH)2
limits the rate of dissolution of the silicate species to neg-
ligible rates. Eventually the level of aqueous Ca(OH)2 be-
comes too high and calcium hydroxide cr ystallizes,
allowing the hydration reactions to continue. Another
theory speculates that two types of ‘CSH’ are formed. The
rate of “first-stage” ‘CSH’ is dependent on the concentra-
tion of aqueous Ca(OH)2. As the concentration of aque-
ous Ca(OH)2 decreases, the production of “first-stage”
‘CSH’ stops, causing induction. Hydration resumes later
when the thermodynamic barrier for the nucleation of
“second-stage” ‘CSH’ is overcome (13).
At any rate, an induction period occurs and varies in

time depending on the type of cement and the desired
application, usually lasting several hours. This property of ce-
ment hydration is what makes it easy to use as a construc-
tion material—it is a semi-solid that can be easily poured into
desired shapes for application. Aqueous gels are often semi-
solid owing to interaction between water molecules and the
surfaces of the particles. Mixing of the system provides energy
to overcome these interactions and allows the gel to become
more fluid. In the case of cement mixtures, constant mixing
is required to keep the material in a fluid state (15). This is
why wet cement is often stored in large rotating drums until
it is poured. During this induction time, so long as it is
continuously mixed, the cement can be held ready for
pouring.
Following induction, the reaction rate accelerates to ap-
proximately dα/dt = 1 day�1. At this point the hydration pro-
cesses are limited by the nucleation and growth of the
hydration products. This acceleration stage (III in Figure 7)
is characterized by rapid hydration of ‘C3S’, followed slowly
by the hydration of ‘C2S’ (13):
Figure 7. A graphic representation of the rate of consumption of
tricalcium silicate (‘C3S’) as a function of hydration time: (A)
changes

in hydration rates in the first few hours as a result of (I) pre-
induc-
tion, (II) induction, (III) acceleration, and (IV) deceleration
processes.
(B) an expanded view showing the length of time required for
com-
plete cement hydration (13).
Hydration Time / h
F
ra
ct
io
n
of
'C
3
S
'
H

yd
ra
te
d
50 10 15
0.00
I
II
III
IV
0.05
0.10
0.15
0.20

0.25
Hydration Time / days
500 100 150
IV
F
ra
ct
io
n
of
'C
3
S
'
H
yd
ra

te
d
0.0
0.4
0.6
0.8
1.0
0.2
A
B

3CaO�2SiO2�3H2O(s) + Ca(OH)2(aq)
2(2CaO�SiO2)(s) + 4H2O(aq)
(9)
During this process, calcium hydroxide reaches its maximum
concentration in the solution and then begins to precipitate
out as crystalline calcium hydroxide, referred to as Portlandite
by cement chemists (7, 13, 15). As the solution becomes con-
centrated with solid product the rate of hydration slows and
becomes diffusion controlled. The reactions slow to nearly
negligible rates but continue for weeks as the ‘CSH’ gel con-
tinues to form.
Calcium Silicate Hydrate (‘CSH’) Gel Formation:
NMR Studies
In its final form, cement is a suspension of calcium hy-
droxide, Ettringite, and unreacted clinker materials in a solid
solution of mineral glue called ‘CSH’ gel (13, 15). The for-

mation of ‘CSH’ gel is vital to the understanding of cement
hydration processes.
One of the most powerful tools for studying the reac-
tions of cement hydration is solid-state nuclear magnetic reso-
nance spectroscopy (21–23). Cements are rich in several
NMR active isotopes: 1H, 29Si, 27Al, and 23Na. 29Si magic
angle spinning (MAS) NMR can be used to examine the sili-
con–oxygen bonding in a cement sample as a function of hy-
dration time. This facilitates the understanding of ‘CSH’
formation (6, 22).
Various forms of Si–O bonding are shown in Figure 8
(24, 25). The basic tetrahedral unit, (SiO4)
4�, is referred to
in this field as a Q0 unit, where the superscript on Q refers
to the number of (SiO4)
4� units attached to the central
(SiO4)
4� unit. Q1 represents a dimer and Q2 corresponds to
silicon atoms within a polymeric chain of (SiO4)

4� units. Q3
and Q4 correspond to silicon centers from which increasingly
complex degrees of chain branching occur, as shown in Fig-
ure 8 (25).
29Si NMR is especially useful for examining Si–O bond-
ing because an increase in the number of (SiO4)
4� units
bonded to each Si center produces an increase in the average
electron density around the central Si atom. This leads to a
more negative chemical shift, relative to tetramethylsilane
(TMS), for successively increasing n values in Qn (see Figure
8 for typical values).
In the pre-induction period of cement hydration, 29Si
MAS NMR shows the presence of monomeric (SiO4)
4� units,
Q0. 1H NMR shows that in the first few minutes protona-
tion of the (SiO4)
4� units also occurs, an indication that the
surface hydroxylation mentioned previously (eqs 1–3) is prob-

ably the first step of the reaction (13, 24). As the reaction
continues, signals corresponding to Q1 units become pre-
dominant, indicating a dimerization of (SiO4)
4� units. As time
passes the intensities of the Q1 signals decrease, and signals
corresponding to polymerization of the dimers, Q2, increase.
Crystallographic and NMR studies have shown that the pri-
mary species formed are pentamer (Si5O16)
12� and octamer
(Si8O25)
18� units (13).
It is interesting to note that Q3 and Q4 signals are not
observed for silicon in the hydration of Portland cement, in-
dicating that polymerization takes place predominantly in a
linear fashion without branching. 29Si MAS NMR spectra
of pure ‘C2S’ and pure ‘C3S’ in comparison with a Portland
cement (PC) sample that had been hydrated for 28 days are
shown in Figure 9. Broad Q1 and Q2 signals in the �75 to
�88 ppm region of the cement sample show the presence of
dimer and linear polymer units. However, the signal corre-

sponding to ‘C2S’ in the hydrated cement sample remains
essentially unchanged after 28 days, which shows the slow
hydration rates of ‘C2S’ relative to ‘C3S’. This is why the pro-
duction of ‘C3S’ in the cement clinker is so vital for effective
cement hydration (24).
Data from NMR experiments such as these, combined
with X-ray crystallography and microscopy, can be used to
postulate a general structure for the cement paste. ‘CSH’ gel
has structural features similar to that of two naturally occurring
minerals: Tobermorite and Jennite (13, 20). In fact, ‘CSH’
gel is often referred to as Tobermorite gel in the cement litera-
ture. These minerals, shown schematically in Figure 10, are
characterized by linear Q2 type O�Si�O bonding and are
formed as multiple layers separated by layers of Ca2� or
Ca(OH)2.
Figure 8. Various arrangements of silicon–oxygen bonding are
ex-
pressed using a superscripted Q, where the superscript refers to
the number of (SiO4)4� units bound to the central (SiO4)4�
unit in
the cluster. The average 29Si NMR signals (relative to TMS)
are also
shown for each unit. For simplicity, charges are omitted.

Si
O
O
O
O OSi
O
O
O Si
O
O
O
OSi

O
O
O OSi
O
O
O Si
O
O
O
Si
O
O
O

O
OSi
O
O
O OSi
O
O
O Si
O
O
O
Si

O
O
O
O
Si
O
O
O
O
Q0= -70 ppm Q1 = -80 ppm
Q2 = -88 ppm
OSi
O

O
O Si
O
O
O
Si
O
O
O
O
Q3 = -98 ppm
Q4 = -110 ppm

Cement Degradation
Crumbling cement, rust stains, and cracks in reinforced
concrete are commonly observed. These are a few examples
of a serious problem that costs North Americans nearly a bil-
lion dollars a year—cement corrosion (26). Corrosion in ce-
ment and concrete materials is a twofold problem because
the cement material and the steel reinforcement are both sus-
ceptible to corrosion, and the weakening of one generally ac-
celerates the degradation of the other. Although cement
corrosion is complicated, the action of water is a common
factor (27).
Cement is a porous material containing a dual network
of pores. The capillary pore system, with a distribution of di-

ameters that range from 50 to 1000 nm, extends throughout
the system, acting as channels between various components
of the system. The cement gel itself contains a network of gel
pores, with diameters on the order of 10–50 nm (19, 28).
Physical properties of cement such as its elastic modu-
lus, fire resistance, and durability are directly related to the
amount of water present (29). Cement is generally 30–40
mass % water, which is present in three forms:
1. Chemically bound water: Water of hydration chemi-
cally bound to the cement precursor materials in the
form of hydrates. This comprises more than 90% of
the water in the system.
2. Physically bound water: Water adsorbed on the sur-
faces of the capillaries. This water is most predomi-
nant in the small gel pores of the system.
3. Free water: Water within larger pores that is free to
flow in and out of the system. The amount of free wa-
ter depends on the pore structure and volume, the rela-
tive humidity, and the presence of water in direct
contact with the cement surface, such as in water-bear-
ing cement pipes and marine structures (19, 27, 30).

Figure 9. 29Si NMR examination of cement hydration: (A) pure
dicalcium silicate, (B) pure tricalcium silicate, (C) Portland
cement
sample hydrated at 40% by mass of water for 28 days. (A) and
(B) show Q0 29Si NMR signals (~ �70 ppm). The addition of
Q1
and Q2 29Si signals (�80 to �90 ppm) is seen upon hydration.
The
slow hydration rate of dicalcium silicate is shown by a large
peak
of unreacted pure material at about �70 ppm (22).
Figure 10. Cement paste is believed to closely resemble the
miner-
als Tobermorite and Jennite. These minerals are characterized
by
layers of polymerized silicon oxide cross-linked with calcium
oxide
or calcium hydroxide (13).
Tobermorite
Jennite

[Ca4(Si3O9H)2]Ca2 . 8H2O
[Ca8(Si3O9H)2(OH)8]Ca2 .6H2O
Si
O O
O O
Si
O O
CaCaCaCa
Si
O O
O
Si
O O
OO
Si

O
OO
Si
OO
OO
Si
Si
HO
Ca
OH
Si
O
OO
Si
O
O

OO
Si
Ca Ca Ca
OO
−
−
−
− −
− −
HO
O
O
−
O

Si
O O
O
Si
O O
Ca
OH OH OH OH
Ca Ca CaCaCaCa
OHOHOHOH
Ca
OO
Si
OO
OO
Si

CaCaCa
OHOHOHOH
Ca
OH
Si
O
OO
Si
OO
OO
Si
O OH
Si
O

O
Si
O O
O
Si
O O
Ca
OH OH OH OH
Ca Ca Ca
OH
Si
O
O
OH
Si

O
O OH
Si
O
O
8 H2O Ca
2+ 8 H2O Ca
2+
6 H2O Ca
2+ 6 H2O Ca
2+
B
A

Corrosion of cement due to water can be discussed in terms
of physical and chemical corrosion.
Physical Corrosion of Cement
Physical corrosion of cements is attributable to the physi-
cal properties of water, especially its volume change during
freezing and its ability to dissolve cement components. The
most significant problem concerning degradation of cements
is the free water in the system. When cement is hydrated,
most of the water used in the process is taken up as hydrates.
If too much water is present, the remaining water is able to
move through the cement causing various problems.
Drying of a cement or concrete paste is an important
factor in the physical corrosion of cement. As a cement paste

hydrates over a period of several months, its porosity de-
creases. Initially, the drying process takes place through cap-
illary flow of water in the larger pore system. As porosity
decreases, the drying process slows and becomes diffusive (13,
28).
Higher water:cement ratios in the hydration reactions
result in larger pore sizes as the cement gel forms, and these
pores contain a larger volume of water. Larger pore sizes also
lead to faster drying rates, which is a serious problem in ar-
eas with low humidity. When cement is exposed to low hu-
midity, free water in the large pores (> 50-nm diameter)
evaporates quickly. This water removal is not serious if the
cement is in contact with water periodically because large
pores also quickly fill with water. However, if cement is ex-
posed to an extended period of low humidity and high tem-
peratures, adsorbed water in the gel pores of the cement will
evaporate. This process leads to drying shrinkage. Drying
shrinkage is destructive because partially filled gel pores (5–
50-nm diameter) contain water menisci that exert consider-
able tensile stress on the walls of the pores. This stress leads
to microcracking and eventually weakens the material. The
use of aggregates minimizes the effect of drying shrinkage
because aggregates increase the elastic modulus and compres-
sive strength of the finished product (28).

Cements in maritime climates at midlatitudes are par-
ticularly susceptible to stress owing to a process known as
freeze–thaw cycling (6, 28, 30). Freeze–thaw cycles occur in
winter when ambient temperatures hover near 0 �C. In these
climates freeze–thaw cycles can occur on nearly a daily basis
in a typical winter season. Freeze–thaw cycles are damaging
to cements because of the 9% volume increase of water upon
freezing (31). When water in the capillary pores freezes, it
expands and exerts stress on the pore walls. This leads to
microcracks, which can in turn fill with water during the sub-
sequent thaw period. Stress exerted in the microcracks dur-
ing further freezing will extend the cracks until macroscopic
cracking is observed. While freeze–thaw degradation gener-
ally is most serious at the surfaces of the cement structure,
extensive cracking will allow the penetration of water deeper
into the structure leading to the eventual failure of the sys-
tem (6, 30).
Crystalline calcium hydroxide makes up about 10% of
the volume of most common cements (5, 13, 15), and seri-
ous physical corrosion of cements results from the leaching
of calcium hydroxide (15). With a room-temperature solu-
bility of 1.7 g�L (15), calcium hydroxide can be easily dis-
solved in free water within cement pastes. This is especially

problematic with pure water, for example, rain water, melted
snow, and condensation within pipes (32). Removal of cal-
cium hydroxide leaves void volumes within the cement, caus-
ing a loss of strength and allowing the deeper penetration of
leaching waters (15, 27, 30).
Calcium hydroxide leaching can be observed in a spec-
tacular effect: opaque white material appears to ooze out of
concrete walls or hang in a stalactite formation from con-
crete ceilings. In this case, water containing dissolved calcium
hydroxide has leached out of the concrete and evaporated,
leaving behind a layer of calcium hydroxide that reacts with
carbon dioxide to form calcium carbonate (15)
CaCO3(s) + H2O(l)Ca(OH)2(s) + CO2(g) (10)
in a process known as efflorescence.5 Efflorescence is often a
sign of water seepage problems in the concrete or cement
structure.
Chemical Corrosion of Cement
Water also carries chemical agents into cement pastes that
react to destroy various components of the cement. A seri-

ous problem is the action of acidic waters from acid precipi-
tation, industrial effluent, or the decay of organic matter (6,
15, 32). Acids also lower the pH of the pore water within
cement pastes, which otherwise has a pH of 11–13 owing to
the large amount of calcium hydroxide present (5, 15, 30).
Lowering the pH will also increase the rate of the corrosion
of the iron in iron-reinforced cement.
The conversion of calcium hydroxide to calcium carbon-
ate through the action of carbon dioxide in the atmosphere
is a problem for all types of cement. This can take place di-
rectly on the surface (efflorescence), or as the CO2 diffuses
into the cement (5, 15, 30):
H2CO3(aq)CO2(g) + H2O(l) (11)
CaCO3(s) + 2H2O(l)H2CO3(aq) + Ca(OH)3(s) (12)
Ca(HCO3)2(aq)H2CO3(aq) + CaCO3(s) (13)
2CaCO3(s) + 2H2O(l)Ca(HCO3)2(aq) + Ca(OH)2(s) (14)
This process, called carbonation, depletes the cement of cal-
cium hydroxide and leaves CaCO3 deposits inside the cement.

Another problem, particularly in marine environments,
is the action of corrosive sulfates such as ammonium sulfate
and magnesium sulfate on cement. These salts react with cal-
cium hydroxide to form calcium sulfate (12, 15):
CaSO4(s) + 2 NH3(aq) + 2 H2O(l)
(NH4)2(SO4)(aq) + Ca(OH)2(s)
(15)
CaSO4(s) + Mg(OH)2(aq)
Mg(SO4)(aq) + Ca(OH)2(s)
(16)
Reactions that deplete cement pastes of calcium hydroxide
are particularly destructive because the products are usually
materials with significantly larger volumes. For example, the
volume of calcium sulfate, 74.2 mL�mol, formed in eqs 15

and 16 is more than twice the volume of the calcium hy-
droxide removed, 33.2 mL�mol (5). This volume change
leads to stresses and cracks that further accelerate the pro-
cesses discussed above.
Behavior of Water in Cement
Understanding the behavior of water in porous cement
is central to the understanding of cement corrosion. Various
theoretical and statistical-mechanical approaches have been
used to try to describe the movement and distribution of
water in the pores of cement (27, 33–35). However, for many
years examination of water in cement pastes was hindered
by the absence of viable experimental techniques for observ-
ing its presence. Recent developments in nuclear magnetic
resonance imaging have provided valuable experimental data
(6).
Magnetic resonance imaging (MRI) is a common tech-
nique used in imaging materials, especially biological mate-

rials. MRI is typically used to measure the spatial distribution
of water in a material (21, 36, 37). It is based on the prin-
ciple that the nuclear magnetic resonance frequency of a
nucleus, such as 1H, in a magnetic field gradient is propor-
tional to its spatial position in the magnetic field gradient
B G z
z
kzz=
− +
= +ν
γ α
ν
( )( ( ))1
2
0
0 (17)
where ν is the observed NMR frequency, γ is the magneto-
gyric ratio of the nucleus, α is the chemical shielding of the

nucleus, B0 is the magnetic field associated with a static field
measurement, Gz(z) is the magnetic field gradient (dB/dz),
ν0 is the NMR frequency in the static field (B0), k is a scal-
ing constant for the signal, and z is the position of the nucleus
in the field (21). Figure 11(A) shows two nuclei in a mag-
netic field gradient Gz = dBz�dz. In Figure 11(B) an inter-
ferogram is produced when a 90� radio frequency pulse is
applied. Fourier transformation of the signal in (B) produces
two peaks separated by ∆ν = γGzdz , shown in Figure 11(C).
The width of the individual peaks is proportional to (πT2)�1,
which gives a resolution, dz,
dz
G Tz
∝
π
1
2
(18)
where T2 is the spin–spin relaxation time of the nucleus (22,
23). The shift in the NMR frequency of a nucleus is propor-
tional to the position of the nucleus in the sample, while the

signal intensity corresponds to the amount of that nucleus
present.
Water in gel pores is tightly confined and is susceptible
to the effects of various paramagnetic species in the sample
including iron and aluminum (6). These factors combine to
make the spin–spin relaxation times short, dramatically de-
creasing the effective resolution (6, 37). For example, using
conventional MRI techniques, a field gradient of about 10
T�m is necessary for a resolution of 10 mm in a cement paste
(6). This is much greater than the normal field gradients used
in MRI, but is on the order of the stray fields associated with
the superconducting magnets of high-resolution NMR in-
struments.6 In 1988, a MRI technique known as stray field
imaging (STRAFI) was developed (38). In this experiment a
sample is moved through a stationary field gradient of a su-
perconducting magnet. STRAFI experiments have allowed
the detailed examination of water in solid cement samples
(35).
Imaging techniques such as STRAFI are useful for ex-
amining the effectiveness of waterproof coatings. One of the
easiest ways to prevent cement corrosion is to prohibit the
movement of water in and out of the material by establish-

ing a waterproof barrier on the exposed surfaces (6, 30). A
wide variety of surface coatings are used in the waterproof-
ing of cements; for example, a common class of waterproof-
ing agents is silanes (22). The rate and depth of surface water
absorption into the cement surface can be compared for a
series of coatings and treatments. The depth and durability
of the surface treatment can also be examined for various ap-
plications (6, 22). A STRAFI image of a Portland cement
sample coated with methyltrialkoxysilane is shown in Figure
12. The images show penetration of the silane coating as it
is repeatedly applied to the surface. After 24 hours the coat-
ing penetrates to a depth of about 2.5 mm. A comparison of
the water penetration in treated and untreated Portland ce-
ment is shown in Figure 13. The treated sample shows water
on the surface (intense surface signal) and the silane coating
Figure 11. Schematic of conventional MRI experiment. (A) Two
nu-
clei situated in a magnetic field gradient, Gz, are separated by
dz. (B) A 90� RF pulse is used to obtain an interferogram of
the
nuclei in the sample. (C) Fourier transformation of (B) gives
two
peaks separated by a frequency proportional to their separation
in the sample (21).

A
B
C
0
ν
M
t
Gz =
9
0
o
p
u
ls

e
∆ν = γGz dz
z
dBz
dz
dz
1
width ~
πT2

penetrating to about 2–3 mm. The untreated sample shows
little surface water but significant water penetration to 8–9
mm after 24 hours (22).
While STRAFI is a powerful technique for examining
the water content of a cement paste, it is limited to relatively
small sample sizes (∼ 2-cm diameter) because of magnetic
field constraints. Studies of concrete are limited to those with
very small aggregates such as fine gravels and sands (28).
Another way of alleviating the problem of short T2 while
avoiding enormous field gradients is through the use of single
point imaging (SPI; ref 21, 39). SPI is an MRI technique
that uses an oscillating field gradient in which signals are mea-
sured at a constant encoding time, tp, following a radio fre-
quency (RF) pulse. A recent variation on SPI developed by
Balcom et al. (40) has proven useful in the examination of
water in cement samples. This technique, called SPRITE
(single point ramped imaging with T1 enhancement), uses a
ramped magnetic field gradient that is much easier to con-
trol than the oscillating gradients used in conventional SPI
(28, 29, 40). While conventional MRI, including STRAFI,
measures all resonance frequencies simultaneously and

deconvolutes using a Fourier transform, SPRITE uses a pro-
cess called position encoding where only one frequency, cor-
responding to a particular encoding time, tp, is measured. The
spatial position, z, of the analyte nucleus is encoded in re-
ciprocal space such that the signal, S(k), is proportional to k
(40):
k ==
1
2 πγG tz,max p
(19)
With a constant encoding time, k is inversely proportional
to the maximum field gradient, Gz,max. A schematic SPRITE
imaging sequence and resulting image are shown in Figure
14. Following an RF pulse a single frequency is measured
after a desired encoding time, tp. Next, the field gradient is
ramped and the sequence is repeated. Each sequence gives
the nuclear density at a particular point in the sample.
Through repetition of the sequences at varying Gz,max an im-
Figure 13. One-dimensional STRAFI of Portland cement
samples in
contact with water for 24 hours. Sample A was treated with

methyl-
trialkoxysilane. Sample B was untreated and shows deep
penetra-
tion of water into the cement surface (22).
Figure 14. Schematic of the SPRITE imaging sequence: (A) the
sig-
nal is measured after a time, tp, has elapsed from an RF pulse;
(B)
the ramping of the field gradient that accompanies the measure-
ment of signal; (C) each successive sequence will show the
density
of the analyte nucleus at a particular position, dictated by the
gra-
dient used during that sequence (40).
z
t p t p t p t p t p
RF RF RF RF RF
Gz
Time

Time
z1 z2 z3 z4 z5
S
ig
n
a
l
z1 z2 z3 z4 z5
A
B
C
Figure 12. One-dimensional STRAFI image of a Portland
cement
sample coated with methyltrialkoxysilane. The coating is
applied
every 30 minutes during the analysis. The signals show the

ingress
of the polymer coating into the cement to a depth of about 2.5
mm
after 24 hours (22).
age is produced. SPRITE and other SPI techniques take
longer than conventional MRI techniques, but are less sus-
ceptible to noise and magnetic inhomogeneities in the sample
because only one frequency is analyzed at a time (21, 28, 29,
39, 40). SPRITE is also useful because signal resolution de-
pends only on the size of Gz,max and the ramping sequence
used, not on the T2 of the analyte nucleus (40).
SPRITE can be used to examine the behavior of pro-
tons in a concrete sample as a function of physical param-

eters such as temperature. From eq 19 it can be shown that
keeping Gz,max and tp constant results in repeated measure-
ments of protons in a defined position, z. The intensity of
the signal can be observed as a function of temperature,
shown in Figure 15. Through an adjustment of parameters
such as Gz,max, RF flip angles, and tp, the experiment can be
tailored to be sensitive to a nucleus of defined T2. This al-
lows the ability to differentiate between the protons of free
liquid water (T2 ≈ 200 µs) and ice (T2 ≈ 10 µs) in a cement
sample. By following the appearance and disappearance of
free water at various regions of a cement sample in the freeze–
thaw cycle, characteristics of the material can be examined
(29, 40).
In cement gels, MRI has shown that water freezes in two
steps. The first step occurs between 0 and �2 �C, where free
bulk water and water in the capillary pores freeze (29). This
freezing-point variation is due to a freezing-point depression
phenomenon caused by vapor pressure lowering in the cap-
illaries and related to the pore size of the capillaries by the
Kelvin equation (29),
=
2 0γ

ρ
T
MT
r H
∆
∆
(20)
where ∆T is the freezing-point depression, γ is the surface
tension of the liquid, M is the molecular weight, T0 is the
normal freezing point, r is the pore radius, ρ is the density
of the absorbate, and ∆H is the molar enthalpy of fusion (41).
Information on the freezing-point depression of water in a
cement sample is valuable for the determination of pore dis-
tributions in these materials.
As ice forms in a cement sample the internal pressure of
the closed system increases owing to the volume expansion
of water. The resulting pressure increase once freezing begins
in the gel pores forces the migration of water from the gel
pores to larger pore regions where ice will form immediately.
This results in a secondary freezing point at about �40 to �45

�C (29). Figure 16 shows a measurement of these two freez-
ing phenomena for a cement sample measured using an SPI
technique (29), in which the evaporable water content of a
concrete sample is measured as a function of temperature as
the sample is slowly cooled (2 K�hr). The first freezing event
is seen between 0 and �1.6 �C. As the sample temperature is
lowered the amount of evaporable water decreases slowly at
freezing temperatures corresponding to the respective pore
sizes present. The large change at 0 to �1.6 �C shows that the
majority of evaporable water present is contained in large
pores. The second major freezing event, associated with the
desorption and freezing of water from the gel pores, is shown
at �45 �C. Other studies have shown that evaporable water
can still exist in cement samples at temperatures as low as
�90 �C (29).
Corrosion of Steel Reinforcement
Reinforced concrete is often used in bridge decks, roads,
and sidewalks. One of the most serious threats to concrete
in cold climates is the use of deicing salts in the winter to
ensure safe conditions for motor vehicles and pedestrians
Figure 16. Magnetization signal for evaporable water in a
sample

of Portland cement mixed with 14-mm diameter graded quartz
ag-
gregate measured using SPRITE as a function of temperature.
Freez-
ing of water is associated with a decrease in signal intensity
(29).
0-10 10-20-30-40-50
0.0
0.2
0.4
0.6
0.8
1.0
- center
M
a

g
n
e
ti
z
a
ti
o
n
(
a
rb
.
u
)
- drying face
T / °C

Figure 15. Schematic of SPRITE used for temperature
dependent
measurements. (A) The normal SPRITE sequence from Figure
14 is
used again, but (B) the field gradient is kept constant. (C) This
al-
lows for a repeated measurement of the signal density for
analyte
nuclei at a particular position in the sample as a function of
tem-
perature (40).
t p t p t p t p t p
RF RF RF RF RF
Gz
Time
Time
T1 T2 T3 T4 T5

Temperature
S
ig
n
a
l
a
t
z
1
T1 T2 T3 T4 T5
A
B
C

(6, 15, 26, 42, 43). Chloride ions, when transported by wa-
ter, attack steel reinforcement (rebar) of these structures caus-
ing them to weaken from within. High pH is important for
minimizing the rate of steel rebar corrosion because it allows
the formation of a passive oxide layer on the surface of the
metal (6, 30). Low pH aqueous states caused by the leaching
of calcium hydroxide from the cement, combined with chlo-
ride ion ingress, causes extensive rebar corrosion in short pe-
riods of time. Chloride ions set up redox reactions along the
rebar, as shown by the following equations (26):
2 Fe2�(aq) + 4 e�2 Fe(s)
4 OH�(aq)O2(g) + 2 H2O(l) + 4e
�
2 FeCl2 (aq)2 Fe

2�(aq) + 4 Cl�(aq)
2 Fe(OH)2 + 4 Cl
�(aq)2 FeCl2(aq) + 4 OH
�(aq)
Fe2O3(s) + 2 H2O(l)2 Fe(OH)2 +
1/2 O2(g)
Fe2O3(s)2 Fe(s) +
3/2O2(g)
(21)
(22)
(23)
(25)
(24)
(26)

Chemical attack of chloride ions is destructive because it not
only reduces the amount of hydroxide ion and iron, but it
also acts in a catalytic manner. Transport of chloride ions
throughout the system is also increased as cracks form as a
result of the other decay processes. Furthermore, patching
can create localized corrosion cells between the rebar in the
existing chloride-contaminated concrete and in the new
chloride-free patch, accelerating the concrete corrosion (43).
Waterproof coatings will stop the introduction of new
chloride ions, but will not remove the chloride ions already
present in the system. Coatings also become ineffective if the
concrete surface is cracked or damaged (26, 43).
A particularly interesting approach to treating chloride
ion ingress is electrochemical chloride extraction (ECE), in
which chloride ions are effectively pulled from the concrete.
A dc circuit, shown in Figure 17, is set up using rebar as the
anode and an electrolyte gel packed on the concrete surface
as the cathode. When a dc potential of 10,000–30,000 V is
applied, water hydrolyzes at the anode, replenishing the hy-
droxide content of the system. The negatively charged steel
rebar repels chloride ions to the surface of the concrete and
into the electrolyte gel. After 4–8 weeks the process is com-

plete, at a fraction of the cost of replacement (43). After seal-
ing with a waterproof coating, the concrete is effectively
protected against further rebar corrosion.
Structures with extremely problematic chloride ion prob-
lems, such as ocean piers, can be cathodically protected by
constant maintenance of the rebar at a negative potential of
about 10,000–30,000 V (43).
Concluding Remarks
The study of cement offers an opportunity to explore
the chemistry of earth materials, their preparation, and re-
sulting properties. Furthermore, examination of cement deg-
radation comprises an extensive part of modern cement
chemistry. Recent innovations in research techniques have
made the study of cement preparation and degradation be-
havior more accessible. Improvement of corrosion resistance
in cement and concrete structures would significantly
lengthen the lifetime of applications using these materials,
potentially saving billions of dollars worldwide.
Acknowledgments
This work was supported by the Natural Sciences and

Engineering Research Council of Canada and the Izaak
Walton Killam Trusts.
Notes
1. Portland stone, a gray stone quarried from the Dorset
region of England, was a commonly used building material in
Europe in the 16th–19th centuries (2, 7).
2. Founded in 1898, ASTM International is a nonprofit orga-
nization that provides a global forum for the development and
pub-
lication of voluntary consensus standards for materials,
products,
systems, and services. See http://www.astm.org (accessed Mar
2003).
3. Fluxing is a process that promotes fusing of materials, in
this case by lowering of the melting point of a mixture by
adding
another component (7).
4. Ettringite is a collective term referring to the various alu-
minate and sulfoaluminate phases present in the clinker
material.

5. Efflorescence is the “blossoming” to a powdery substance
on exposure to air.
6. The stray field near a 9.4 T (400 MHz) NMR magnet is
on the order of 60 T�m.
Figure 17. Schematic of electrochemical chloride extraction
(ECE).
A dc voltage of 10–30 000 V is applied between the steel rebar
(anode) and an electrolyte gel (cathode) on the surface of the
con-
crete. Hydrolysis of water takes place at the rebar and chloride
ions are repelled from the concrete into the electrolyte gel (43).
Cl
��
Cl �
+
�
�

H2O H + OH
H + OH
�
��
steel rebar
concrete sample
electrolyte gel
DC power supply
2H O
http://www.astm.org

Literature Cited
1. Sass, S. L. The Substance of Civilization; Arcade: New York,
1998.
2. Blezard, R. G. The History of Calcerous Cements. In Lea’s
Chemistry of Cement and Concrete, 4th ed.; Hewlett, P. C., Ed.;
Arnold: London, 1998.
3. Ryan, J. F. J. Chem. Educ. 1935, 6, 1855.
4. Hall, C. J. Chem. Educ. 1976, 53, 222.
5. Bogue, R. H. Chemistry of Portland Cement; Rhinehold: New
York, 1947; British Patent #5022, October 21, 1824, see
http://www.buildbyte.com/grasim/ceoscorner2.html (accessed
Mar
2003).
6. Hewlett, P. C.; Hunter, G.; Jones, R. Chemistry in Britain
1999, 35, 40.
7. West, A. R. Solid State Chemistry and Its Applications;

Wiley:
Chichester, England, 1984.
8. Bates, P. H. J. Chem. Educ. 1926, 3, 519.
9. Fisk, H. G. J. Chem. Educ. 1934, 11, 195.
10. Bogue, R. H. J. Chem. Educ. 1942, 19, 32.
11. Ryan, J. F. J. Chem. Educ. 1935, 6, 2128.
12. Daugherty, K. E.; Robertson, L. D. J. Chem. Educ. 1972, 49,
522.
13. Odler, I. Hydration, Setting and Hardening of Portland Ce-
ment. In Lea’s Chemistry of Cement and Concrete, 4th ed.;
Hewlett, P. C., Ed.; Arnold: London, 1998.
14. MacPhee, D. E.; Lachowski, E. E. Cement Compositions
and
Their Phase Relations. In Lea’s Chemistry of Cement and Con-
crete 4th ed.; Hewlett, P. C., Ed.; Arnold: London, 1998.
15. Czerin, W. P. Cement Chemistry and Physics for Civil
Engineers;
Chemical Pub. Co.: New York, 1962.

16. Glasser, F. P. Applications of the Phase Rule to Cement
Chem-
istry. In Phase Diagrams: Materials Science and Technology
Vol-
ume II, The use of Phase Diagrams in Metal, Refractory,
Ceramic,
and Cement Technology; Alper, A. M. Ed.; Academic Press:
New
York, 1970.
17. National Physics Laboratory: CaO-SiO2 Phase Diagram.
http://www.npl.co.uk/npl/cmmt/mtdata/dgox1.html (accessed
Mar 2003).
18. Jackson, P. J. Portland Cement: Classification and Manufac-
ture. In Lea’s Chemistry of Cement and Concrete, 4th ed.;
Hewlett, P. C., Ed.; Arnold: London, 1998.
19. Consolati, G.; Dotelli, G.; Quasso, F. J. Appl. Phys. 1999,
86,
4225.
20. Richardson, I. G.; Graves, G. W. Cement and Concrete Res.
1992, 22, 1001.

21. Strange, J. H.; Halse, M. R. Imaging Techniques for Solids
and Quasi-Solids. In Encylopedia of Magnetic Resonance; Grant
D. M., Harris, R. K., Eds.; Wiley: Chichester, England, 1996;
Vol. 4, p 2472.
22. Black, S.; Lane, D. M.; McDonald, P. J.; Hannant, D. J.;
Mulheron, M.; Hunter, G.; Jones, M. R. J. Mater. Sci. Lett.
1995, 14, 1175.
23. Bohris, A. J.; Newling, B.; McDonald, P. J.; Raoof, A.;
Tran,
N. L. J. Mater Chem. 1998, 33, 859.
24. Johansson, K.; Larsson, C.; Antzutkin, O. N.; Forsling, W.;
Kota, H. R.; Ronin, V. Cement and Concrete Res. 1999, 29,
1575.
25. Lippma, E.; Magi, M.; Samoson, A.; Englehardt, G.; Grim-
mer, A.-R. J. Am. Chem. Soc. 1980, 102, 4889.
26. Luma, C. Chem. Eng. 1998, November, 149.
27. Adenot, F.; Buil, M. Cement and Concrete Res. 1992, 22,
489.
28. Beyea, S. D. ; Balcom, B. J.; Bremner, T. W.; Prado, P. J.;

Cross,
A. R.; Armstrong, R. L.; Grattan-Bellew, P. E. Solid State
Nuclear Magnetic Resonance 1998, 13, 93.
29. Prado, P. J.; Balcom, B. J.; Beyea, S. D.; Armstrong, R. L.;
Grattan-Bellew, P. E. Cement and Concrete Res. 1998, 28, 261.
30. Eglinton, M. Resistance of Concrete to Destructive
Agencies.
In Lea’s Chemistry of Cement and Concrete, 4th ed; Hewlett, P.
C., Ed.; Arnold: London, 1998.
31. Laidler, K. J.; Meiser, J. H. Physical Chemistry, 2nd ed.;
Houghton Mifflin: Boston, 1995.
32. Djuric, M.; Ranogajec, J.; Omorjan, R.; Miletic, S. Cement
and Concrete Res. 1996, 26, 1295.
33. Schmidt-Döhl, F.; Rostásy, F. S. Cement and Concrete Res.
1999, 29, 1039.
34. Schmidt-Döhl, F.; Rostásy, F. S. Cement and Concrete Res.
1999, 29, 1047.

35. Nagesh, M.; Bhattacharjee, B. ACI Mater. J. 1998, March-
April, 113.
36. Andrew, E. R. Imaging: A Historical Overview. In
Encylopedia
of Magnetic Resonance; Grant D. M., Harris, R. K., Eds.;
Wiley:
Chichester, England, 1996; p 2462.
37. Canet, D. Nuclear Magnetic Resonance–Concepts and
Methods;
John Wiley & Sons: Chichester, England, 1986.
38. Samoilenko, A. A.; Artemov, D. Y.; Sibel’dina, L. A. JETP
Lett. 1988, 47, 417.
39. Crooks, L. E. Image Formation Methods. In Encylopedia of
Magnetic Resonance; Grant D. M., Harris, R. K., Eds.; Wiley:
Chichester, England, 1996; Vol. 4, p 2439.
40. Balcom, B. J.; MacGregor, R. P.; Beyea, S. D.; Green, D.
P.;
Armstrong, R. L.; Bremner, T. W. J. Magn. Reson., Ser. A
1996,
123, 131.

41. Overloop, K.; Van Gerven, L. J. Magn. Reson., Ser. A 1993,
101, 179.
42. Sandberg, P. Cement and Concrete Res. 1999, 29, 473.
43. Whitmore, D.; Abbott, S.; Veilivasakis, E. Civil
Engineering
1999, January, 46.
http://www.buildbyte.com/grasim/ceoscorner2.html
http://www.npl.co.uk/npl/cmmt/mtdata/dgox1.html
In the mid-1990’s, several cases of premature deterioration of
concrete pavements and precast members gained notoriety
because of uncertainty over the cause of their distress. Because
of the unexplained and complex nature of several of these cases,
considerable debate and controversy have arisen in the research
and consulting community. To a great extent, this has led to a
misperception that the problems are more prevalent than actual

case studies would indicate. However, irrespective of the fact
that cases of premature deterioration are limited, it is essential
to
address those that have occurred and provide practical,
technically sound solutions so that users can confidently specify
concrete in their structures.
Central to the debate has been the effect of a compound known
as ettringite. The objectives of this paper are:
• to define ettringite and its form and presence in concrete,
• to respond to questions about the observed problems and the
various deterioration mechanisms that have been proposed, and
• to provide some recommendations on designing for durable
concrete.
Because many of the questions raised relate to cement
character-
istics, a brief primer on cement manufacture and chemistry is
included in Appendix A.
What Is Ettringite?
Ettringite is the mineral name for calcium sulfoaluminate

(3CaO•Al
2
O
3
•3CaSO
4
•32H
2
O), which is normally found in
portland cement concretes. Calcium sulfate sources, such as
gypsum, are intentionally added to portland cement to regulate
early hydration reactions to prevent flash setting, improve
strength development, and reduce drying shrinkage. Sulfate and
aluminate are also present in supplementary cementitious
materials and admixtures. Gypsum and other sulfate compounds
react with calcium aluminate in the cement to form ettringite
within the first few hours after mixing with water. Most of the
sulfate in the cement is normally consumed to form ettringite at
early ages. The formation of ettringite in the fresh, plastic
concrete is the mechanism that controls stiffening. At this stage

ettringite is uniformly and discretely dispersed throughout the
cement paste at a submicroscopic level (less than a micrometer1
in cross-section).
Ettringite formed at early ages is often referred to as “primary
ettringite.” It is a necessary and beneficial component of
portland
cement systems.
Why Does Microscopic Analysis Often Show
Ettringite in Concrete?
It has been known for many years that if concrete is exposed to
water for long periods of time, primary ettringite can slowly
dissolve and reform in any available voids or microcracks
[Lerch,
1945]. This is why, under microscopic examination, needle-like
crystals of ettringite are often observed lining air voids or
cracks
in older concretes, including those from structures that have
demonstrated a high level of durability.
Ettringite Formation and the Performance of Concrete

©Portland Cement Association, 2001 IS417
All rights reserved
Fig. 1. Portland cements are manufactured by a process that
combines sources of lime (such as limestone), silica and
alumina (such as clay), and iron oxide (such as iron ore).
Appropriately proportioned mixtures of these raw materials
are finely ground and then heated in a rotary kiln at high
temperatures, about 1450 °C (2640 °F), to form cement
compounds. The product of this process is called clinker
(nodules at right in above photo). After cooling, the clinker is
interground with about 5% of one or more of the forms of
calcium sulfate (gypsum shown at left in photo) to form
portland cement. The calcium sulfate controls setting time,
strength development, and drying shrinkage potential.
1A micrometer is one millionth of a meter, which is about
0.00004 in.
Any deterioration of concrete by freeze-thaw action, alkali-
silica
reactivity (ASR), or other means, accelerates the rate at which
ettringite leaves its original location in the paste to go into

solution and recrystallize in larger spaces such as voids or
cracks.
This is because both water and space must be present for the
crystals to form. The space can be provided by cracks that form
due to damage caused by frost action, ASR, drying shrinkage,
thermal gradients, or other mechanisms. In addition,
deterioration
caused by such mechanisms provides greater surface area of
exposure and easier paths for ingress and egress of water.
Ettringite crystals in air voids and cracks are typically 2 to 4
micrometers in cross-section and 20 to 30 micrometers long.
Under conditions of extreme deterioration, and repeated wetting
and drying, ettringite crystals can appear to completely fill
voids
or cracks. However, ettringite, found in this benign state as
large
needle-like crystals, should not be interpreted as causing the
expansion of deteriorating concrete.
To confirm that ettringite does not contribute to expansion of
deteriorating cast-in-place concrete, Lerch investigated effects
of
cement sulfate levels on alkali-silica reaction and freeze-thaw
action [Lerch, 1945]. By using cements of different sulfate

contents (higher sulfate contents having potential to form more
ettringite), he evaluated whether the solution and
recrystallization
of the calcium sulfoaluminate contribute to expansion.
The ASR study used a reactive aggregate with cement alkalies
ranging from 0.53% to 1.05% Na
2
O equivalent. Mortar and con-
crete prisms were tested beyond three years. Concrete prisms
were exposed to field and laboratory conditions. Specimens
with
the 3.5% sulfate cements usually had about the same, and often
less, expansion than those with the 1.5% sulfate cements. This
indicates that expansion in the specimens resulted from ASR,
and
recrystallization of the ettringite, occurring in spaces created by
the ASR, did not contribute to the expansion.
The freeze-thaw study tested concretes with cements having
sulfate contents ranging from 1.7% to 4.0%. The specimens
were
exposed to 160 freeze-thaw cycles, followed by 28 days of

drying in air, followed by one year in water. This testing
regimen
theoretically would disrupt the paste and provide a dry and then
wet environment ideal for the recrystallization of ettringite. The
specimens prepared with the cements with 4.0% sulfate had less
expansion and a smaller decrease in dynamic modulus than
those with lower sulfate content. No abnormal expansion was
observed after the water storage. This indicates that the
expansion in the specimens resulted from frost damage and not
from recrystallization of ettringite. This research did not
investigate the hypothesis that partial or total void filling by
ettringite could reduce the protection the air void system affords
against freeze-thaw damage (see discussion of freeze-thaw
resistance below).
Ettringite formed by dissolution and recrystallization in void
spaces and cracks is often referred to as “secondary ettringite.”
Secondary ettringite is not detrimental to concrete performance.
Can Excess Soluble Sulfates in Cement Cause
Deleterious Expansions in Concrete?
If inordinately high amounts of gypsum (or other sources of
calcium sulfate) are added to cement, and if concrete made with
that cement is in a moist service environment, deleterious

expan-
sions can occur. Sulfates from gypsum or other forms of
calcium
sulfate used to control properties of cement are readily soluble
and react soon after cement comes into contact with mix water.
However, if extremely high levels of gypsum are added,
abnormal expansions can occur from excessive calcium
sulfoalu-
minate formation after hardening and continuing until the
gypsum becomes depleted. This was demonstrated by Lerch in
1945; he showed that gypsum additions to give cement SO
3
con-
tents of up to 5.0% had no significant impact on expansion of
mortar bars stored in a moist room. However, when inordinately
high gypsum levels (up to 7.5% as SO
3
) were added to clinker,
excessive expansions were generated. The rate of length
changes
in the mortar bars showed that most of the expansion took place

2
Portland Cement Association
Fig. 2. Ettringite (white, needle-like crystals) are commonly
found in samples taken from concrete in service, especially in
concrete from moist environments. Secondary ettringite is not
detrimental to concrete performance. [Concrete in photo
exposed to phenolphthalein stain to highlight features.]
Fig. 3. This is a photomicrograph of a mortar sample that had
been
stored in water for six years at 20 °C. To simulate an over
sulfated
system, part of the sand was replaced by gypsum, and thus
significant expansions occurred. Note that the gaps around the
sand grains in the mortar are similar in appearance to those
observed for delayed ettringite formation.
within three months. This implies that if excessive soluble
sulfate
contents in cements were available, it would be reasonable to

expect field problems in concrete made from that cement to
show up relatively early in the life of the structure, probably
within the first six months. Commercially produced portland
cements in the U.S. contain less than 5% SO
3
.
The potential impact of excessive gypsum has thus been known
for
some time, and is the reason ASTM C 150, Standard
Specification
for Portland Cement contains provisions for control of sulfates
in
portland cement. Using requirements of ASTM C 150, one can
optimize the level of sulfates in cements to maximize strength
and/or minimize shrinkage, without excessive expansion
potential.
Cements are optimized prior to production by testing at various
sulfate levels to achieve maximum strengths in mortar cubes
(and/or minimum shrinkage of mortar prisms). ASTM C 150
contains numerical limits on sulfate levels that are a function of
cement type. In cases where the optimum sulfate level exceeds
table limits, the standard includes an option for a performance
test

that measures expansion of mortar bars stored in water. This
provides the opportunity to optimize cements for sulfates, while
protecting against the potential for deleterious expansions.
Can Ettringite Formation Reduce the Freeze-
Thaw Resistance of Concrete?
One hypothesis that has been considered to explain premature
damage observed in pavements is that the protective air
entrainment system has been rendered ineffective because of
ettringite filling of air voids. The theory is that increased
sulfate
levels within the concrete, from excess sulfates in cement or fly
ash, result in an excessive level of ettringite that clogs the
entrained air void system. A related hypothesis is that external
sources of sulfates, as found for example in gypsum in deicing
salts, could increase susceptibility to freeze-thaw damage.
These
hypotheses have appeal. Theoretically, if even one ettringite
crystal were to form in an air void, the volume of air void space
would be reduced. However, the question remains as to whether
there could be sufficient infilling to harm properly air-entrained
concrete, which would have a paste system with air voids well
in excess of those needed to accommodate volumetric expansion
of water during freezing.

A series of laboratory tests were undertaken to test these
hypotheses [Detwiler and Powers-Couche, 1997]. Three cements
produced from the same raw materials were used. Two were
commercial Type I (C
3
A=12%; SO
3
=2.03%) and Type II
(C
3
A=5%; SO
3
=2.72%) cements; the third was made by
intergrinding the Type I cement with additional gypsum
(SO
3
=3.14%) to increase the amount of available sulfate in the
concrete. Concrete prisms made from these cements were

subjected to freezing and thawing under conditions outlined in
ASTM C 666 Procedure A, except that 3% NaCl solutions either
with or without added gypsum (to simulate road salt) were used
instead of water. In addition, freeze-thaw cycles were
interrupted
over weekends to simulate a wetting and drying environment, an
exposure conducive to ettringite crystal growth. Air contents
originally selected for test were 2±0.5%, 4±0.5%, and 6±0.5%.
However, after over 300 cycles, the 6% air specimens were
discontinued, as no deterioration was observed. Thus, the
remaining tests were conducted on marginal to poor air content
concretes. Specimens were tested to destruction in most cases.
Periodically during the testing, companion specimens were eval-
uated petrographically to evaluate ettringite formation in voids.
Conclusions from this study were that performance of the
concretes was dominated by the quality of the air void system.
For non-air-entrained concrete (2%), damage occurred without
formation of ettringite deposits. For marginally air-entrained
concrete (4%), deposition of ettringite appeared to follow the
for-
mation of cracks as freeze-thaw deterioration occurred.
Ettringite
did not cause the cracking, nor did it contribute to the

propagation of existing cracks. Cracks due to frost damage
created space for ettringite crystals to grow. Also, the presence
of
gypsum in the salt solution had no significant effect on the test
results. The governing factors in performance of the specimens
were the volume of air and quality of the air void system in the
hardened concrete.
Subsequent to the above program, a second program was
initiated to test two cement systems at higher sulfate levels
[Taylor, 1999]. One cement (labeled A in figs. 4-6) had a
sulfate
level of 4.04% (clinker sulfate was 1.9%). The other cement had
a sulfate level of 2.78% (clinker sulfate was 0.39%), but prior
to
testing, gypsum was intentionally added to reach a cement
sulfate level of 4%. Concretes were made from each cement and
tested under the modified ASTM C 666 as described above, but
using only the 3% NaCl solution with added gypsum. The fresh
concretes had entrained air contents of 5.5% and 5.9%, with a
good air void distribution as measured on the hardened
concretes. Results exceeded commonly applied performance
requirements at 300 cycles (relative dynamic modulus greater
than 90%), so testing was continued to 900 cycles. Conclusions
were that ettringite was observed to have formed in voids near

the surface, but was not related to the observed deterioration of
the specimens at later cycles. Performance was a function of the
quality of the air void system.
Both studies once again confirm the need for properly air-
entrained
concrete in structures exposed to severe freeze-thaw
environments.
Are There Slowly Soluble Sulfates in Cement
Clinker That Can Result in Late-Forming
Ettringite and Subsequent Deleterious
Expansions in Concrete?
The hypothesis for a deterioration mechanism related to slowly
soluble sulfates is that some portions of the sulfates in clinker
are in
a form that is very slowly soluble or are incorporated in the
silicate
phases such that they are not available to participate in early
hydration reactions. The slowly soluble form has been attributed
to
the potential presence of β-anhydrite,2 CaSO
4
(see Appendix A).

3
2 ß-anhydrite is often referred to as “hard burned” anhydrite,
which is a term used in the
gypsum plaster industry. It is commonly misperceived that ß-
anhydrite is “insoluble.”
However, while its rate of solubility is less than other forms, it
is sufficient to act as an
effective set controller [Hansen et al., 1988, and Michaud and
Suderman, 1999]
4
The implication of the hypothesis regarding sulfates
incorporated in
the silicate phases is that some of the silicates, particularly
belite
(C

2
S), do not hydrate until after initial hardening; thus, the
“trapped”
sulfates would only become available to form ettringite after
concrete has hardened, assuming available moisture.
The potential for slowly soluble sulfates in clinker has been
asso-
ciated with increased levels of total clinker sulfate. The concern
expressed is that changes in cement manufacturing technology
(changes in fuels, recycling of cement kiln dust—CKD—or use
of
alternative fuels such as tires) have resulted in increased levels
of
sulfates and the potential for slowly soluble sulfates. Concern
has
also been expressed that moving from wet process production to
dry process and preheater, precalciner systems has increased
potential for sulfates in alite and belite.
Changes have occurred in cement manufacture with movement
toward more energy efficient dry process production and
environmental controls. There are four primary types of kiln

systems: wet, long dry, preheater, and precalciner. The latter
three are dry process systems that do not use a water slurry of
raw materials. The preheater and precalciner are the most
energy
efficient kiln types, and generally are newer. However, dry
process kilns have been used for many years (Appendix A).
There
is no basis to assume that the kiln system in itself would govern
sulfate levels or potential for slowly soluble sulfate.
On average, clinker sulfate levels have increased.3 The change
from natural gas to coal as a fuel source, which the industry
made
as a result of the energy crisis of the early 1970s, led to some
increase. Also, since the original Clean Air Act in the 1970s
mandated collection systems for particulate emissions, there has
been a relative increase in CKD available as a raw material
source.
CKD is composed of fine particles that consist predominantly of
partially calcined forms of the original raw materials introduced
into the kiln [Klemm, 1994]. CKD also contains alkali sulfates,
which are incorporated into the dust as they volatilize off the
kiln
feed during calcination. Relative to cement performance, alkali
sulfates react readily during early hydration reactions. Reuse of

CKD varies from plant to plant because of kiln operating
conditions. In addition, since CKD can contribute to increased
alkali levels of cement, the amount that can be used as a raw
feed
source may be limited.4 It should also be noted that recycling of
CKD as kiln feed is not new or unique. It has been a normal,
environmentally beneficial operating practice in many plants
worldwide. In fact many plants in the U.S. recycle all of their
CKD,
without detriment to the final product. There are no data to
indicate
that CKD use as a raw material contributes to potentially
detrimental slowly soluble sulfates, nor is there any identified
relationship between CKD use as raw feed in cement
manufacture
and observed field problems with concrete made from that
cement.
Questions have also arisen regarding use of alternative fuels
such
as liquid solvents or tires. Liquid solvents are normally lower in
sulfur than the fuel they replace. Published data on sulfur in
tires
indicate a range from about 0.9% to 1.8% by mass, with an

-0.10
-0.06
-0.02
0.02
0.06
0.10
0 200 400 600 800 1000
Cycles
Length
change,
%
A
B
Fig. 5. Length change of concrete samples subjected to

modified ASTM C 666 cycling (after Taylor 1999)
0
20
40
60
80
100
120
0 200 400 600 800 1000
Cycles
Relative
dynamic
modulus,
%

A B
Fig. 6. Relative Dynamic Modulus of concrete samples
subjected to modified ASTM C 666 cycling (after Taylor 1999)
0
2
4
6
8
10
0 200 400 600 800 1000
Cycles
Mass
loss,
%

A B
Fig. 4. Mass loss of concrete samples subjected to modified
ASTM C 666 cycling (after Taylor 1999)
3 See Appendix A.
4 Contrary to a perception often expressed, CKD use as a raw
material is not the
primary source of cement alkalies. Alkali levels in cement are
strongly dependent
on the alkali content of the original raw materials.
5
and subjected to moist storage conditions [Tennis et al., 1999,
and Olek and Zhang, 1999]. These tests were undertaken to
confirm the levels of expansion that could occur at the known
amounts and distribution of clinker sulfate. Results of the
mortar
bar expansion measurements show that, for curing temperatures

below 70 °C (158 °F), no excessive expansions have occurred,
even when excess gypsum was added to give cement sulfate
levels one percent above optimum.
The selective dissolution tests indicate that clinkers
manufactured
in accordance with conventional industrial practices do not
contain slowly soluble sulfates that could cause cements to
induce deleterious expansions in concretes. Review of the
literature on this topic supports this finding [Taylor, 1996].
Analytical methods using selective dissolution are available to
verify the form and distribution of sulfates in clinker.
What About Delayed Ettringite Formation due
to High Temperatures?
It has been known for some time that concrete subjected to
early-
age temperatures high enough to destroy some or all of the
ettringite originally formed can, in the presence of moisture,
undergo deterioration with the reformation of ettringite in the
hardened paste system [Day, 1992; Famy, 1999]. The term
“delayed ettringite formation” (DEF) is commonly used to refer
to
the potentially deleterious reformation of ettringite in moist
concrete, mortar, or paste after destruction of primary ettringite

by
high temperature. Such early-age temperatures may result from
heat
treatment, or in extreme cases, from internal heat of hydration.
The temperature conditions for deleterious expansion due to
DEF
have not been conclusively defined. However, a heat treatment
temperature above about 70 °C (158 °F) is most often cited
[Taylor,
1994]. This temperature is affected by factors such as moisture
con-
ditions during heat treatment, cement characteristics, the
concrete
mix, and interactive effects of other deterioration mechanisms,
such
as alkali-silica reactivity, and freezing and thawing.
Based on laboratory testing of mortar prisms, deleterious
expansions have not been observed at temperatures of 70 °C
(158 °F) or less, irrespective of cement characteristics. At
increasing maximum temperature levels above 70 °C (158 °F),
factors such as cement characteristics begin to have an impact
[Kelham, 1999].

Based on laboratory tests of mortars, Kelham identified
character-
istics of cement that show increased sensitivity to heat
treatment
[Kelham, 1997 and 1999]. When cured at temperatures above
90 °C (194 °F) the following characteristics of cement led to
greater expansions in mortars subjected to extended periods (5
years) of moist curing:
• Higher fineness
• Higher C
3
A
• Higher C
3
S
• Higher alkali (Na
2
O
eq
)

• Higher MgO.
average of 1.3%. Data for coal indicate a range of about 0.3% to
4.0%, with an average of 1.5%. Tires have about a 20% higher
Btu value than coal. Given the fact that tires only replace about
10% to 20% of conventional fuel Btu value, they have little
impact on overall sulfate levels in clinker. It should also be
recognized that there is no identified relationship between the
problems encountered in field concretes and cement sourced
from plants using alternative fuels.
The hypothesis that clinker sulfates are slowly soluble has been
criticized on the basis that the dominant form of sulfate in
clinker,
alkali sulfates, are in fact rapidly soluble, as are the forms of
calcium sulfate added to make portland cement: gypsum,
hemihydrate, and anhydrite. Michaud and Suderman (1999) have
demonstrated that even anhydrite is sufficiently soluble to act
as a
set control agent as it dissolves and reacts quickly. The amount
of
calcium sulfate added to the mill with the clinker to control
cement setting is adjusted to account for sulfates present in
clinker.

To evaluate the hypothesis regarding slowly soluble clinker
sulfates, a study of 33 commercially available clinker samples
was undertaken [Klemm and Miller, 1997]. The SO
3
contents
ranged from 0.03% to 3.00% by mass. The sample population
was skewed toward higher (> 1.5%) SO
3
clinker, since these
were of most interest and were specifically requested for the
voluntary testing program. The samples also represent current
conventional manufacturing practice for all four kiln system
types. The molar SO
3
/Na
2
O

eq
of the clinker samples ranged from
0.06 to 2.54, which is representative of a broad range of
production. This ratio is of interest because it indicates the
balance of sulfates and alkalies in the system (alkali sulfates are
rapidly soluble). The samples were evaluated by selective
dissolution techniques. The objective was to determine the form
and distribution of the sulfates. The selective dissolution
technique works by using a potassium hydroxide/sugar solution
to dissolve non-silicate (interstitial) phases and then evaluating
the sulfates in the residue left after extraction. The residue is
then
further dissolved in salicylic acid/methanol solution to remove
the silicates, and the remaining residue is evaluated for sulfates.
This allows differentiation of sulfates in the silicate phases
from
the interstitial phases. In addition, the form of sulfates in the
phases can be evaluated by X-ray diffraction.
No β-anhydrite was found in the samples. Also, only small
amounts of sulfate were found to be incorporated in the silicate
phases; this distribution did not correlate with the total
measured
sulfate. Thus, the tests indicate that only very minor amounts of

clinker sulfate appear to remain after the first day of hydration.
In
no case did this represent more than 0.85% in the silicates or
0.68% for the clinker as a whole. For clinker sulfate
incorporated
in silicate phases, the amount of SO
3
potentially available for
possible late reactions to form ettringite was balanced by an
equal or greater amount of alumina, thus making the formation
of ettringite unlikely, as three times more SO
3
than alumina is
required to form ettringite. If SO
3
and alumina are balanced,
monosulfoaluminate would be expected to form.
This work has been extended to include expansion tests of
mortar bars made from the cements of the test clinker samples

Cements with high fineness, C
3
A, C
3
S, and Na
2
O
eq
are generally
associated with higher SO
3
levels because more sulfates are
required to control early stiffening. In addition, these are the
same characteristics that provide increased early-age strengths
and are most desirable from the perspective of rapid
construction
cycles. The sulfate (SO

3
) to aluminate (Al
2
O
3
) ratio is important
as, for a given C
3
A content, it influences the amount of primary
ettringite formed.
Another factor that is significant regarding early heat treatment
of
concrete is the preset or delay period prior to application of
heat
[Day, 1992]. Application of heat without allowance for a preset
period increases susceptibility to DEF. The importance of preset
time has long been established, as it also affects strength
[Pfeifer
and Landgren, 1982]. Control of early temperatures has become

more significant with the advent of higher release strength
requirements, availability of more reactive cements, high-range
water reducers that permit very low water-cement ratios, and
use
6
-0.10
0.00
0.10
0.20
0.30
0.40
0.50
0 200 400 600 800 1000
Time, days

Expansion,
%
C 01
C 03
C 09
C 19
C 22
C 03-III
C 09-III
Cured: 90°C,
stored 23°C.
Fig. 8. When cured at 90 °C, some mortar bars show
significant expansions after about 100 days.
0
1

2
3
4
5
0.0 0.2 0.4 0.6 0.8
f'1/f'28
SO3,
%
Fig. 9. Those factors that lead to higher early strengths in
cement (higher fineness, C3A, C3S, and alkalies) are associated
with higher sulfate (SO3) levels to control early stiffening. In
this figure, the required level of cement sulfate is shown to
increase with an increasing ratio of 1–28- day strengths. Both
Type I and Type III cements are included in the figure.
-0.10
0.00

0.10
0.20
0.30
0.40
0.50
0 200 400 600 800 1000
Time, days
Expansion,
%
C 01
C 03
C 09
C 19
C 22

1Running head SULFATE ATTACKA potential research topic on a .docx

1Running head SULFATE ATTACKA potential research topic on a .docx

Recommended

Recommended

More Related Content

Similar to 1Running head SULFATE ATTACKA potential research topic on a .docx

Similar to 1Running head SULFATE ATTACKA potential research topic on a .docx (20)

More from eugeniadean34240

More from eugeniadean34240 (20)

Recently uploaded

Recently uploaded (20)

1Running head SULFATE ATTACKA potential research topic on a .docx