Sheet1gdΔyd31010.10.3333333333yLsin(thet)vB=vCt_AC=tCDt_BCt0.10.3.docx

Sheet1gdΔyd/31010.10.3333333333yLsin(thet)vB=vCt_AC=tCD
t_BCt0.10.3480102170.28734788561.41421356240.4921607687
0.23570226041.22002379770.20.38873012630.514495755420.3
8873012630.16666666670.94412691930.30.44845413490.66896
473162.44948974280.36616126790.13608276350.86840529920.
40.52068331170.76822127962.82842712470.36817870060.1178
5113020.85420853130.50.60092521260.83205029433.16227766
020.3800584750.10540925530.86552620540.60.68637534270.8
7415727613.46410161510.39627898890.09622504490.8887830
2260.70.77531355660.90286051883.74165738680.4144225280.
08908708060.91793213670.80.86666666670.923076923140.433
33333330.08333333330.950.90.95974533660.93774876074.242
64068710.45242829050.07856742010.983424001111.05409255
340.94868329814.4721359550.47140452080.07453559921.0173
4464081.11.14939597660.95702440444.69041575980.49010409
120.07106690551.05127508781.21.24543611280.96351790964.
89897948560.50844716390.06804138171.08493570961.31.3420
5480930.9686638665.09901951360.52639720470.0653720451.1
1816645441.41.4391355430.97280621475.29150262210.543942
10710.06299407881.150878293
t 0.1 0.2 0.30000000000000004 0.4 0.5 0.6 0.7
0.79999999999999993 0.89999999999999991
0.99999999999999989 1.0999999999999999 1.2
1.3 1.4000000000000001 1.2200237977444093
0.94412691931270665 0.86840529918946907
0.85420853134357533 0.86552620540503789
0.88878302264836029 0.91793213671498253 0.95
0.98342400108749772 1.0173446408320563
1.0512750878335928 1.0849357096237595
1.1181664544043501 1.1508782930286723

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
x
x
x
x
x
x
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
Technical Report Documentation Page
1. Report No.
FHWA/TX-04/0-4240-2
2. Government Accession No.
3. Recipient's Catalog No.
4. Title and Subtitle
HYDRATED LIME STABILIZATION OF SULFATE-BEARING

SOILS IN TEXAS
5. Report Date
August 2004
6. Performing Organization Code
7. Author(s)
Pat Harris, Tom Scullion and Stephen Sebesta
8. Performing Organization Report No.
Report 0-4240-2
10. Work Unit No. (TRAIS)
9. Performing Organization Name and Address
Texas Transportation Institute
The Texas A&M University System
College Station, Texas 77843-3135
11. Contract or Grant No.
Project 0-4240
13. Type of Report and Period Covered
Technical Report:
May 2002 - May 2004
12. Sponsoring Agency Name and Address

Texas Department of Transportation
Research and Technology Implementation Office
P. O. Box 5080
Austin, Texas 78763-5080
14. Sponsoring Agency Code
15. Supplementary Notes
Research performed in cooperation with the Texas Department
of Transportation and the Federal Highway
Administration.
Project Title: Develop Guidelines and Procedures for
Stabilization of Sulfate Soils
16. Abstract
Sulfate-bearing subgrade soils treated with calcium-based
stabilizers often experience heaving problems
(three-dimensional swell) due to chemical reactions with the
sulfate and/or sulfide minerals. This project
focuses on determining the sulfate content at which the
deleterious chemical reactions occur and on
evaluating the effectiveness of alternative construction practices
aimed at reducing swell in high-sulfate soils.
These practices include extended mellowing, double lime
application, and increasing field moisture contents.
To determine what concentrations of sulfate are too high for
stabilization with lime in Texas, a soil from the
Vertisol order that contained no detectable sulfates was selected
for three-dimensional swell measurements.
Two different sulfate compounds were added to the soil: sodium
sulfate (Na2SO4) and gypsum
(CaSO4·2H2O). Samples containing sulfate concentrations of
0, 1000, 2000, 3000, 5000, 7000, and 12,000
parts per million (ppm) were then subjected to a three-

dimensional swell test for a minimum of 45 days.
Results of these systematic swell experiments revealed the
following: (1) sulfate contents up to 3000 ppm
can safely be treated with traditional lime stabilization; (2)
coarse-grained sulfates take longer to swell and
form deleterious reaction products; (3) mellowing effectively
treats sulfate concentrations up to at least 7000
ppm; (4) higher molding moisture contents (2% above optimum)
reduce swell better than optimum moisture;
and (5) single application of lime reduces swell better than
double application. This study, using systematic
laboratory experiments, confirmed empirical field observations
of sulfate limits presented by other
investigators.
17. Key Words
Sulfates, Soils, Stabilization, Ion Chromatography,
Field Testing, Highways, Laboratory Tests
18. Distribution Statement
No restrictions. This document is available to the
public through NTIS:
National Technical Information Service
Springfield, Virginia 22161
http:/www.ntis.gov
19. Security Classif.(of this report)
Unclassified
20. Security Classif.(of this page)
Unclassified
21. No. of Pages

46
22. Price
Form DOT F 1700.7 (8-72)
Reproduction of completed page authorize
HYDRATED LIME STABILIZATION OF SULFATE-BEARING
SOILS IN
TEXAS
by
Pat Harris
Associate Research Scientist
Tom Scullion
Research Engineer

and
Stephen Sebesta
Assistant Transportation Researcher
Report 0-4240-2
Project Number 0-4240
Project Title: Develop Guidelines and Procedures for
Stabilization of Sulfate Soils
Performed in Cooperation with the
Texas Department of Transportation
And the
Federal Highway Administration
August 2004
TEXAS TRANSPORTATION INSTITUTE
The Texas A&M University System
College Station, Texas 77843-3135

v
DISCLAIMER
The contents of this report reflect the views of the authors, who
are responsible for the
facts and the accuracy of the data presented herein. The
contents do not necessarily reflect the
official view or policies of the Federal Highway Administration
(FHWA) or the Texas
Department of Transportation (TxDOT). This report does not
constitute a standard,
specification, or regulation. The engineer in charge was Tom
Scullion, P.E. (# 62683).
vi

ACKNOWLEDGMENTS
Dr. German Claros, P.E., and Mr. Robert E. Boykin, P.E., from
the Texas Department of
Transportation are program coordinator and project director,
respectively, of this important
project and have been active in providing direction to the
research team. The project advisors,
Mr. Richard Williammee, P.E., Mr. Mike Arellano, P.E., Mr.
Maurice Pittman, P.E., and Ms.
Claudia Izzo from TxDOT, along with Mr. Jim Cravens, P.E.,
from the Federal Highway
Administration have also been active in assisting the
researchers. Both TxDOT and the FHWA
provided funds for this project. Ms. Jore’ VonHolt, Mr. Marvin
Zeig, and Ms. Jessica Claros
from the Texas Transportation Institute (TTI) assisted the
researchers with much of the
laboratory testing in this study.
vii

TABLE OF CONTENTS
Page
List of Figures
...............................................................................................
............................... viii
List of Tables
...............................................................................................
.................................. ix
Chapter 1. Introduction
...............................................................................................
....................1
Chapter 2. Background
...............................................................................................
.....................5
Previous Investigations
...............................................................................................
.........5
Chapter 3. Traditional Lime Stabilization
....................................................................................11
Methods..................................................................................
............................................11
Results
...............................................................................................
................................15

Discussion and Interpretation
............................................................................................1
8
Chapter 4. Modified Lime Stabilization for Higher Sulfate
Contents ...........................................23
Methods..................................................................................
............................................23
Results....................................................................................
............................................24
Discussion and Interpretation
............................................................................................2
5
Chapter 5. Conclusions
...............................................................................................
...................31
Chapter 6. Recommendations
...............................................................................................
.........33
References..............................................................................
........................................................35
viii
LIST OF FIGURES

Figure Page
1. Vertical Heaves Generated during Construction of U.S. 67
near Midlothian. ....................1
2. Map Showing Major Metropolitan Areas Constructed on
Vertisols ...................................3
3. Map of Counties in Texas with Potential Sulfate Problems.
...............................................5
4. Variation in Grain Size for Gypsum Present in Texas
Soils................................................9
5. Gypsum Size Fractions Used in Samples Molded for 3-D
Swell Tests ............................12
6. Sample Preparation for 3-D Swell Measurement
..............................................................13
7. 3-D Swell for Selected Concentrations of Gypsum
...........................................................16
8. 3-D Swell for Samples Containing 7000 ppm Sulfates
.....................................................17
9. XRD Patterns for Lime Treated and Untreated Sulfate-Rich
Soils ...................................19
10. SEM Images of Unstabilized and Stabilized Sulfate-Rich
Soils .......................................20
11. 3-D Swell with Sulfate Dissolved in Molding Water and
One-Day Mellowing ...............24

12 3-D Swell with 7000 ppm Sulfates Dissolved in Molding
Water .....................................25
13. Effect of Lime Treatment on Sulfate
Content....................................................................27
14. Density Comparison for Samples Molded at Different
Moisture Contents ......................29
ix
LIST OF TABLES
Table Page
1. Sample Matrix for the Two Phases of Sulfate-Heave
Experiments ..................................14

1
CHAPTER 1 INTRODUCTION
Sulfate-bearing subgrade soils have caused tens of millions of
dollars in damage to Texas
highways over the last decade. Many subgrade soils treated
with calcium-based stabilizers
experience heaving problems (Figure 1) due to chemical
reactions with sulfate and/or sulfide
minerals. Field observations indicate that the reactions can be
very rapid and occur overnight
following a single rainfall event. In other cases the reaction is
delayed and it may take years for
the problem to manifest itself in terms of excessive pavement
roughness.
Figure 1. Vertical Hezaves Generated during Construction of
U.S. 67
near Midlothian.
Sulfate problems in cement and concrete research have been
reported for more than

70 years.1 In 1962, Sherwood2 reported problems with sulfates
in lime and cement stabilization
2
of soils. However, reports of sulfate-induced heave in subgrade
soils received little attention
until the mid 1980s. Formation of ettringite was determined to
be the cause of heaving in a case
study from the southern United States.3 Mitchell’s Terzaghi
lecture was the first time sulfate-
induced heave received national recognition.4 He used a
parking lot in Las Vegas that
experienced heave 2 years after construction as an example to
stress the importance of
physicochemical and biological changes in soil mechanics:
Mitchell reported ettringite and
thaumasite were the cause of failure.
Hunter explained many of the physicochemical details
concerning sulfate heave.5
Hunter’s experiments determined that four ingredients (lime,
clay minerals, sulfate ions, and
water) are needed to generate sulfate heave at 77ºF (25ºC), with
sulfate ions being the key

ingredient.
Previous studies have focused on mechanisms of sulfate heave,
although few studies
examined swell caused by lime stabilization of sulfate-rich
soils. Mitchell and Dermatas
systematically added sulfates ranging from 3000 to 62,000 ppm
to artificial kaolinite- and
montmorillonite-rich soils.6 This study focused on extremely
high sulfate concentrations and
extended curing times, generally 30 days. In 1999, another
study evaluated the effects of ground
granulated blastfurnace slag as a stabilizer in an artificial
kaolinite and sulfate-rich soil and a
natural sulfate-bearing Kimmeridge Clay.7 This study
examined extremely high sulfate
concentrations (11,200 ppm sulfate was the lowest) as well.
At the time of this report, the Texas Department of
Transportation (TxDOT) Dallas and
Fort Worth Districts do not recommend using calcium-based
stabilizers for subgrade
stabilization if the sulfate levels are greater than 2000 ppm in a
soil. This limit is based on
empirical field observations and experience.

This research focuses on adding very low sulfate concentrations
(0, 1000, 2000, 3000,
5000, 7000, 10,000, and 12,000 ppm) to a natural soil to
measure three-dimensional (3-D) swell.
With an understanding of the sulfate-heave mechanism, this
study will identify the sulfate
concentrations that cause unacceptable 3-D swell with lime
stabilization.
The primary objectives of this study are as follows:
1. assess the 3-D swell potential of lime stabilized, sulfate-
bearing, subgrade soils;
2. determine the sulfate level safe for traditional lime
stabilization; and
3. assess the effectiveness of mellowing, double lime
application, and increased moisture
3
content in reducing swell in high-sulfate soils.
A sulfate-deficient soil of the Vertisol order with a Plasticity
Index (PI) of 24 was chosen
as the soil to add selected concentrations of sulfate since that is
what underlies a large portion of
the Texas Coastal Plain, where most of the sulfate problems

have been reported (Figure 2).8
Figure 2. Map Showing Major Metropolitan Areas Constructed
on Vertisols.
4
5
CHAPTER 2 BACKGROUND
PREVIOUS INVESTIGATIONS
The most severe heaves reported in Texas were observed at Joe
Pool Lake near Dallas.9
Burkart et al. identified certain geologic formations that possess
high levels of sulfates and
determined that gypsum was the most common sulfate in Dallas
area soils.10 The most severe
heaves in the Dallas/Fort Worth area are associated with the

Eagle Ford Formation shown in
Figure 3 below. Since the inception of this project, other areas
with high sulfate concentrations
have been identified around Texas. Counties known to have
problematic sulfate concentrations
are identified in Figure 3.
Figure 3. Map of Counties in Texas with Potential Sulfate
Problems.
6
Researchers at Louisiana State University investigated the
possibility of anhydrite
(CaSO4) converting to gypsum in a humid environment as the
heave mechanism.11 They
determined that heave was actually due to formation of
ettringite in the cement-stabilized soil.
It is also important to recognize that gypsum is not the only
problematic mineral in soils.
Pyrite (FeS2) is a sulfide mineral that alters to gypsum
(CaSO4·2H2O) under the right conditions

and creates similar problems. Dubbe et al. reported five case
histories where pyritic shales
oxidized to sulfates, causing heave and concrete
deterioration.12 Pyrite-derived sulfate was
documented as the cause of heave in Portland cement-stabilized
minestone.13 Oxidation of
pyrite-bearing Eagle Ford shale in north Texas is the source of
sulfates in many soils in that
region.10
Sulfate-Induced Heave Chemical Reactions
The literature on sulfate-induced heave in soils reports many
mechanisms by which heave
may occur. A detailed explanation on possible heave
mechanisms is given in Mitchell and
Dermatas;6 however, only two will be discussed briefly as
follows:
1. Sulfide minerals oxidize and react with other soil minerals to
form sulfate
minerals.12 This transformation involves an increase in
volume due to variations
in atomic packing as well as the addition of water to the mineral
structure.

2. The formation of the mineral ettringite, which only occurs
under special
circumstances, causes expansion of up to 250 percent when
completely formed.14
Hunter5 performed extensive experiments to elucidate factors
controlling the
formation of this expansive mineral.
The first mechanism of sulfate-induced heave mentioned above
is oxidation of sulfides.
Pyrite and marcasite (both minerals are FeS2 but the atoms are
arranged differently) form under a
reducing (oxygen deficient) environment and are not stable in
an oxygen-rich environment.
They are abundant in many coals, carbonaceous shales, and
limestones. Often these rocks are
exposed to the atmosphere during road construction. Upon
exposure, O2(g) from the atmosphere
serves as an oxidizing agent for pyrite and marcasite.
The following reaction (Rxn) illustrates what takes place:
4FeS2 + 15O2 + 14H2O → 4Fe(OH)3 + 8H2SO4 Rxn. 1
7

The iron and sulfur are oxidized by surface water that is
enriched in atmospheric oxygen.
The iron generally precipitates as a ferric hydroxide and the
sulfate will either remain in solution
or precipitate as gypsum if there is sufficient calcium present.
The source of calcium is often
limestone (CaCO3), which is very soluble in acids. Looking at
the right side of Rxn. 1, there are
8 moles of sulfuric acid (H2SO4) released in the weathering of
4 moles of pyrite or marcasite.
This will make the surrounding environment very acidic and
promote the dissolution of
limestone (Rxn. 2) which will supply Ca2+ for the formation of
gypsum (Rxn. 3).
CaCO3 + H2SO4 → Ca2+ + (SO4)2- + H2O + CO2 Rxn. 2
Ca2+ + (SO4)2- + 2H2O → CaSO4·2H2O Rxn. 3
The mineral transformation of pyrite to ferric hydroxide and
gypsum results in an
increase in volume. The oxidation of pyrite and formation of
gypsum alone is responsible for

distress experienced in some construction projects.12 In other
projects, where traditional
calcium-based stabilization is performed, other deleterious
reactions may occur. Heave caused
by calcium-based stabilizers in sulfate and clay-rich
environments is mainly due to the formation
of hydrous calcium-hydroxide-sulfate minerals.
The second mechanism mentioned on page 6 is the formation of
ettringite,
Ca6[Al(OH)6]2·(SO4)3·26H2O, which requires unique
conditions to form. At standard
temperature (25ºC) the pH has to be above 10, and a source of
water is critical for the 26 moles
of water in the mineral structure; additionally, a source of
aluminum, sulfur, and calcium are also
required to form ettringite. When sulfur-bearing clay-rich soils
are stabilized with lime or
cement, then all of the above criteria are met. Lime and cement
both raise the pH to above 12,
which causes dissolution of clay minerals and releases
aluminum into the system. Water may be
supplied from a number of sources: during the stabilization
process, as precipitation after
stabilization, or from the groundwater or adjacent reservoirs.

Calcium is released by the lime
and cement during stabilization, and the sulfur is supplied from
the sulfide- and/or sulfate-
bearing soils or water.
The following is an abbreviated geochemical reaction model
from Hunter:5
8
Ca(OH)2 → Ca2+ + 2(OH)- Rxn. 4
(Ionization of lime; pH rises to 12.3)
Al4Si4O10(OH)8 + 4(OH)- + 10H2O → 4Al(OH)4- + 4H4SiO4
Rxn. 5
(Dissolution of kaolinite at pH > 10.5)
CaSO4·2H2O → Ca2+ + SO42- + 2H2O Rxn. 6
(Dissolution of gypsum)
6Ca2+ + 2Al(OH)4- + 4(OH)- + 3(SO4)2- + 26H2O →
Ca6[Al(OH)6]2·(SO4)3·26H2O Rxn. 7
(Formation of ettringite)

Rxns. 4 and 5 occur in any lime-stabilized kaolinite-bearing
soil. Addition of lime to the
soil causes the pH to rise to approximately 12.3, releasing large
amounts of calcium to the soil.
Clay minerals are unstable at a pH above 10.5, so the clays start
breaking down into aluminum
hydroxide and silicic acid. Sulfate ions (Rxn. 6) are supplied
by the dissolution of gypsum. The
only other elemental requirement for the formation of ettringite
is water. Ettringite only forms in
a high pH ≈ 10-12 environment. Once the pH drops below 10
ettringite stops forming. In this
example, kaolinite is the aluminum source and gypsum is the
sulfur source; aluminum may be
derived from dissolution of any clay mineral and sulfur may be
derived from any sulfur-bearing
mineral as previously discussed. Gypsum is used as the sulfur-
bearing mineral in this example,
and it appears to be the dominant sulfur-bearing mineral
responsible for sulfate-induced heave in
Texas soils.10
Sulfate-Induced Heave Reaction Rates

The speed at which the aforementioned reactions proceed is
controlled by a number of
factors, namely the temperature, concentrations of reactants and
products, and the rate of mass
transfer into and out of a fixed reaction site.15 An important
aspect of the rate of mass transfer
involves the grain size of the sulfate minerals. For example, a
soil containing large gypsum
9
(CaSO4·2H2O) crystals will dissolve more slowly than a soil
containing small crystals. Texas
soils have a range of grain sizes that influence how rapidly
sulfate heave reactions occur (Figure
4).
Figure 4. Variation in Grain Size for Gypsum Present in Texas
Soils.
10

The top image in Figure 4 shows gypsum crystals in excess of 6
inches long. The bottom
image is a scanning electron microscope (SEM) image of
gypsum crystals smaller than 1 µm.
One can imagine that it will take longer for the gypsum crystals
in the top image of Figure 4 to
dissolve than those in the bottom image because of the lower
total surface area. Therefore, the
size of the sulfate minerals plays a key role in determining how
rapidly the reactions proceed in
forming ettringite and generating sulfate-induced heave.
11
CHAPTER 3 TRADITIONAL LIME STABILIZATION
This chapter focuses on determining the sulfate content
considered to be too high for
traditional lime stabilization. Traditional lime stabilization is
defined as lime mixed into the soil
and immediately compacted without allowing the lime/soil
mixture to sit/mellow for an extended
period of time before compaction. The researchers developed
the following laboratory testing

program to identify the maximum sulfate concentration for lime
stabilization without special
construction procedures.
METHODS
A soil from the Vertisol order in College Station, Texas, was
selected for swell
measurements to determine what concentrations of sulfate are
too high for stabilization with lime
in Texas. Vertisols are present over large parts of the Texas
Coastal Plain (Figure 2) and have
high shrink/swell potential due to smectitic clay minerals.8
This soil was selected because it is
typical of lime-stabilized soils in Texas and does not contain
detectable sulfates greater than
100 ppm.
Samples were also selected from the Eagle Ford Formation
(Figure 3) in Fort Worth,
Texas, for comparison to the swell generated with the College
Station soil. Construction projects
on soils from the Eagle Ford Formation have generated a large
percentage of the sulfate-induced
heave problems in the Dallas/Fort Worth area. Samples from
this particular location did not

contain detectable sulfates as well.
Soil Processing
The soils were dried in a 140ºF (60ºC) oven to a constant
weight and pulverized to pass a
#4 sieve as outlined in American Society for Testing and
Materials (ASTM) D 698. The
engineering properties of the two soils were determined as
follows:
1. For the College Station soil, a plasticity index of 24 was
determined by ASTM D 4318;
an optimum lime content of 6% determined by the Eades and
Grim Test16 or ASTM D
6276; and the optimum moisture content determined by
modified Proctor (ASTM D
1557) using 6% lime is 22%.
12
2. The Fort Worth soil was determined to have a plasticity index
of 42, an optimum lime
content of 3 to 4% and an optimum moisture content of 19%.

3-D Swell Samples
The sulfate compounds, sodium sulfate (Na2SO4) and gypsum
(CaSO4·2H2O), were
added to soil samples at concentrations of 0, 1000, 2000, 3000,
5000, 7000, and 12,000 ppm by
four different techniques to represent scenarios observed in the
field, as follows:
1. Sodium sulfate was added to the mixing water of some
samples and dissolved.
2. Sodium sulfate was added to the water bath of other samples
and dissolved to
represent sulfates being added via an external water source
(water truck) and
groundwater, respectively.
3. Fine-grained (F.G.) gypsum passing the #200 sieve was added
directly to the soil
in a solid state.
4. Coarse-grained (C.G.) gypsum passing the #10 sieve was
retained on the
#40 sieve.
The fraction sizes in 3 and 4 above (shown in Figure 5 below)
were chosen because they
are representative of the more reactive sulfates found in natural

soils in Texas. Grain size is an
important issue because the larger the grains, the longer it takes
for them to dissolve and react.
Figure 5. Gypsum Size Fractions Used in Samples Molded for
3-D Swell Tests.
13
Each sample was weighed and mixed separately using one of the
four methods of sulfate
application stated above. All samples were molded in duplicate
at the density determined by
modified Proctor in one lift with a Superpave Gyratory
Compactor. The sample size was
restricted to 4 inches in diameter by 4.5 inches tall (10.16 cm ×
11.43 cm) due to the constraint
of molding in one lift. The samples were then air dried for 3
days, placed in a 3-D swell test
modeled after unpublished data by Tom Petry and wrapped in a
paper membrane saturated with
distilled water. Each sample utilized porous stones placed on
the bottom and top of each sample,

with a latex membrane placed over the sample. Duplicate
samples were placed in a distilled
water bath at 100 percent humidity and 77ºF (25º ± 2ºC) (Figure
6). The distilled water level
was maintained just below the top of the porous stone located
on the bottom of the sample to
allow water to be drawn up into the sample by capillary action.
Figure 6. Sample Preparation for 3-D Swell Measurement.
Three-dimensional swell was measured by determining the
height to the nearest 0.01 inch
in three places 120º apart. The circumference was measured
with a clear plastic tape to the
nearest 0.0197 inch near the top, middle, and base of each
sample. The three height and
circumference measurements were averaged and the volume was
calculated.
For swell testing, the National Lime Association recommends
placing samples in a water
bath for 3-D swell measurement immediately after molding;14
however, these tests were
conducted by air drying for 3 days after compaction then
placing the samples in the swell test.

14
The National Lime Association procedures were not used nor
the seven-day moist cure
since neither technique is representative of field conditions.
The typical scenario with heave in
Texas has been that the subgrade is stabilized with lime,
compacted to density, usually baked in
the summer sun for a couple of days, and then saturated with
water by a thunderstorm.
Generally, subgrade heaves are observed the day after the
storm.
Testing Plan
The researchers performed the testing in two phases as shown in
Table 1. Phase I
determined what sulfate content was too high for traditional
lime stabilization (i.e., no
mellowing) and is discussed further in this chapter. Phase II
identified sulfate levels too high for
modified lime stabilization techniques, which are discussed in
Chapter 4.
After swell testing for at least 45 days, the samples were dried

in a 140ºF (60ºC) oven
until a constant weight was reached. The samples were then
measured again for volume change.
To determine if deleterious reaction products were actually
contributing to the swell of
samples, X-Ray Diffraction (XRD) was performed on selected
samples with a Rigaku X-ray
diffractometer using Cukα radiation at a scan speed of 0.75º per
minute with a step of
0.02 degrees. A bulk sample analysis was performed on
selected samples to identify reaction
products. A side-loading random powder mount reduced
preferred orientation of minerals.17
Table 1. Sample Matrix for the Two Phases of Sulfate-Heave
Experiments.
Sulfate content
(ppm)*
Sulfate Application
Method
Moisture content
(%dry wt.)
Mellow
Time (days)

Percent lime
(Initial/final)
PHASE I
0 1) Na2SO4 in 22 0 6/0
1000 water bath
2000 2) Na2SO4 in
3000 molding water
5000 3) Gypsum (F.G.)
7000 4) Gypsum (C.G.)
12,000
PHASE II
5000 Na2SO4 in 22 1 6/0
7000 molding water 24 2 3/3
10,000 3
*Two samples of each sulfate content were constructed.
15
A JEOL 6400 SEM with a Princeton Gammatech Energy
Dispersive Spectrometer (EDS)
was used to observe crystal habit (shape) and location of
reaction products. The SEM was
operated at a beam current of 15 kV and 10 mm (0.4 in.)
working distance to maximize EDS

results.
RESULTS
As previously stated in the Testing Plan on page 14, Phase I
identifies the concentration
of sulfates too high for “traditional” lime stabilization (i.e., no
mellowing).
Traditional Lime Stabilization 3-D Swell Results
Figure 7 (top graph) shows three-dimensional swell progressing
with time for the Fort
Worth soil with fine-grained gypsum. The unstabilized control
sample was molded at 17%
optimum moisture and the lime-stabilized samples were molded
at 19% optimum moisture. All
samples were molded at 104 lb/ft3 density. Negative swells
from day 0 to day 3 are a result of air
drying the sample for 3 days before placing it in the swell test.
The bottom curve is a control
with no sulfates and stabilized with 6% lime which still swelled
more than 3%. The top curve is
the unstabilized control; it swelled by about 30%. The curves
between the two controls reveal a
consistent trend of increasing swell with increasing sulfate

level.
Figure 7 (bottom graph) shows three-dimensional swell
progressing with time for the
College Station soil with coarse-grained gypsum. All of the
samples were molded at 22%
moisture and 104 lb/ft3 density. The bottom curve illustrates
how lime treatment of the soil with
no sulfates results in greatly reduced swell compared to the
same soil with no stabilizer (top
curve). This is exactly what would be expected for lime
treatment and indicates that lime is
doing its job. The curves between the two control samples show
that increasing amounts of
sulfate, from 3000 to 12,000 ppm in the form of coarse-grained
gypsum, result in progressively
more swell.
The trends observed for the two soils shown in Figure 7 are
similar. The Fort Worth soil
experienced larger swells, no doubt due in part to its higher PI
value. Based on field experience,
TxDOT and the National Lime Association have ranked soils
containing less than 3000 ppm
sulfates as soils with a low risk of lime-induced sulfate heave
problems and soils containing

16
more than 8000 ppm sulfates are ranked as high-risk soils. The
laboratory results presented in
Figure 7 appear to support these views.
Fort Worth Soil (PI = 42)
-5.00%
0.00%
5.00%
10.00%
15.00%
20.00%
25.00%
30.00%
0 10 20 30 40 50 60
Days
S
w
el

l
0 ppm 6% lime
0 ppm no lime
10K ppm 6% lime
7K ppm 6% lime
5K ppm 6% lime
3K ppm 6% lime
College Station Soil (PI = 24)
-5.00%
0.00%
5.00%
10.00%
15.00%
20.00%
0 10 20 30 40 50 60
Days
S
w

el
l
0 ppm 6% lime
3K ppm 6% lime
5K ppm 6% lime
7K ppm 6% lime
12K ppm 6% lime
0 ppm no lime
Figure 7. 3-D Swell for Selected Concentrations of Gypsum.
17
The method by which sulfate is introduced to the stabilized
layer has an impact on swell
results as illustrated in Figure 8. The sulfates were introduced
either in the molding water or as
fine-grained or coarse-grained crystals with the water added
prior to compaction. The three
upper curves show samples containing 7000 ppm sulfate and the
lower curve is the control (6 %
lime stabilization, no sulfates). Figure 8 illustrates the
following points:

1. The samples with dissolved sulfate reach equilibrium the
fastest and produce the
greatest swell (top curve).
2. The fine-grained sulfate (F.G., 6% lime) reaches equilibrium,
or plateaus, much more
rapidly than the coarse-grained samples (C.G., 6% lime).
3. The coarse-grained gypsum (C.G., 6% lime) resulted in the
lowest swell.
The variations in the individual swell curves in Figures 7 and 8
are due to different
researchers performing the 3-D swell measurements. A single
researcher was assigned the task
of performing all subsequent swell measurements and the results
became more consistent. In
general, the repeatability between duplicate samples was very
good. The difference in swell
measurements ranges from a minimum of 0.01% at the lowest
sulfate content to a maximum of
2.76% at the highest sulfate content.
-4.00%
-2.00%
0.00%

2.00%
4.00%
6.00%
8.00%
10.00%
12.00%
14.00%
0 10 20 30 40 50 60
Days
S
w
el
l
0 ppm 6% lime
C.G., 6% Lime
F.G., 6% Lime
Molding H2O, 6% Lime
Figure 8. 3-D Swell for Samples Containing 7000 ppm Sulfates.

18
DISCUSSION AND INTERPRETATION
The three-dimensional swell experiments show an unmistakable
trend of increasing swell
with increasing sulfate content. As discussed in Chapter 2,
most sulfate-induced heave is
attributed to the formation of ettringite and/or thaumasite. The
question is: Can the swell we
obtained in our experiments be attributed to formation of
ettringite and/or thaumasite?
Factors Causing Swell
The experiments in this report were conducted in the range of
22º to 25ºC; therefore, the
formation of thaumasite is ruled out since it requires
temperatures below 15ºC to form.5
Selected samples were analyzed by XRD and SEM to determine
if deleterious reaction
products actually formed in the lime-stabilized samples. Figure
9 shows partial XRD patterns
for an unstabilized gypsum-bearing sample (solid line) which
contains a sharp peak at 7.51 Å.
This peak confirms the presence of gypsum but is absent in the
lime-stabilized sample which

initially contained gypsum (dashed line). The broad peak at
7.16 Å is the (001) kaolinite peak
and is present in both samples. The two peaks at 9.66 and 5.57
Å in the lime-stabilized sample
are diagnostic of ettringite. The presence of these two peaks
and the absence of the gypsum peak
in the lime-stabilized sample illustrates that gypsum is being
consumed to form the highly
expansive mineral ettringite. These XRD patterns confirm that
ettringite was formed in the swell
tests.
19
0
50
100
150
200
250

300
8 9 10 11 12 13 14 15 16 17
2θ
C
ou
nt
s
Treated
Untreated
9.
66
5.
57
7.
51
7.
16
Figure 9. XRD Patterns for Lime Treated and Untreated
Sulfate-Rich Soil.
To confirm the XRD results, SEM was used to analyze pieces of
the same two samples
represented in Figure 9 and the two images shown in Figure 10

are the result. The top image is
of the unstabilized gypsum-bearing soil and the cornflake-
appearing grains dispersed throughout
the image are smectite clay minerals. The bottom SEM image is
the lime-stabilized sample and
the balls of radiating fibrous crystals evident in the lower image
have the morphology of
ettringite; EDS analyses of the radiating balls show the presence
of calcium with lesser and
subequal amounts of sulfur and aluminum, which confirms the
presence of ettringite identified
by XRD. Note the absence of ettringite in the unstabilized soil.
Thus, based on XRD, SEM, and
EDS analyses, formation of ettringite caused swell in the lime-
stabilized samples.
20
Unstabilized Soil
Stabilized Soil
Figure 10. SEM Images of Unstabilized and Stabilized Sulfate-
Rich Soils.

21
Upper Sulfate Limit for Conventional Lime Stabilization
Upon consulting with TxDOT engineers it was determined that
no criteria were available
for an acceptable level of swell for lime-stabilized soils;
therefore, it was decided to use the no
sulfate swell level as the baseline cutoff. Using the College
Station soil, this was approximately
1-2% 3-D swell (Figure 7, bottom graph on page 16, and Figure
8 on page 17). A 3-D swell of
2-3% correlates with 3000 ppm sulfates (Figure 7 bottom
graph), which implies that no modified
construction techniques are required for sulfate concentrations
of 3000 ppm or less; however,
caution should be exercised for the following reasons.
First, there must be a limited supply of sulfate ions. For
example, suppose there is a body
of water adjacent to a construction site where sulfates at
concentrations less than 3000 ppm have
been detected. A full geotechnical investigation should be
conducted to determine the source of

Sheet1gdΔyd31010.10.3333333333yLsin(thet)vB=vCt_AC=tCDt_BCt0.10.3.docx

Sheet1gdΔyd31010.10.3333333333yLsin(thet)vB=vCt_AC=tCDt_BCt0.10.3.docx

Recommended

Recommended

More Related Content

Similar to Sheet1gdΔyd31010.10.3333333333yLsin(thet)vB=vCt_AC=tCDt_BCt0.10.3.docx

Similar to Sheet1gdΔyd31010.10.3333333333yLsin(thet)vB=vCt_AC=tCDt_BCt0.10.3.docx (20)

More from lesleyryder69361

More from lesleyryder69361 (20)

Recently uploaded

Recently uploaded (20)

Sheet1gdΔyd31010.10.3333333333yLsin(thet)vB=vCt_AC=tCDt_BCt0.10.3.docx