The Science of Cement and Concrete

1 Introduction
1.1 1.2 1.3 1.! Is concrete a scientifically interesting material? The purpose of this monograph What is (and is not) co ered "asic definitions and terminology

Concrete "asics
2.1 2.2 2.3 Introduction The components of concrete The concrete construction process

3
3.1

#anufacture and composition of portland cement

$ er ie%

This chapter will give an overview of the process of manufacturing portland cement and will then devote somewhat more attention to the final properties of the cement, including the composition and structure of the four main cement minerals, the oxide and mineral composition of various types of portland cement, and the particle size distribution (fineness) of cement. Finally, as this monograph will focus on the basic science of cement, the process of synthesizing pure minerals for research purposes is also covered. 3.2 &ote on cement chemistry notation and mineral names

At the high temperatures present in a cement iln, all of the constituents of the raw ingredients are either driven off as a gas or converted to an oxide form. To ma e the formulas of cement minerals, compounds, and reactions shorter and easier to read, it is traditional to use a shorthand notation that leaves out the oxygen. This system is listed in Table !." .
Ta'le 3.!( #ement chemistry notation based on oxides. $)ide form #a$ %i$& Al&$! Fe&$! %$! &otation C S * +

'

(&$

$ther abbreviations are also in use, such as )g$*) and +a&$ * +, but only those listed above in Table !." will be used in this monograph. ,hile this system has definite advantages, it does ta e some getting used to, since, for example, # normally stands for carbon and ( normally stands for hydrogen. -nfortunately the confusion does not end there, as we also have to deal with chemical and mineral names for solid phases. The chemical name of a solid compound is simply the word version of the chemical formula, for example .calcium hydroxide/. The mineral name (in this case .portlandite/) is meant to represent either a particular crystal structure or a potentially impure form of the phase. The various ways of describing some of the main solid phases associated with cement chemistry are summarized in Table !.". 0n this monograph we will use primarily the chemical name and the cement notation. 0n the case of the cement minerals, this is actually a bit inaccurate. +ames such as alite and belite indicate a particular crystal structure as well as the fact that these phases contain a variety of impurities when found in cement.
Ta'le 3.!( %ummary of the different ways to represent some cement minerals and products.

Chemical &ame Tricalcium %ilicate 2icalcium %ilicate Tricalcium Aluminate Tetracalcium Aluminoferrite #alcium hydroxide #alcium sulfate dihydrate #alcium oxide

Chemical +ormula #a!%i$1 #a&%i$" #a!Al&$4 #a&AlFe$1 #a($()& #a%$".&(&$ #a$

$)ide +ormula !#a$.%i$& &#a$.%i$& !#a$.Al&$! "#a$.Al&$!.Fe&$! #a$.(&$ #a$.%$!.&(&$ #a$

Cement &otation #!% #&% #!A #"AF #( # (& #

#ineral &ame Alite 3elite Aluminate Ferrite 5ortlandite 6ypsum 7ime

3.3

-a% materials for cement manufacture

The first step in the manufacture of portland cement is to combine a variety of raw ingredients so that the resulting cement will have the desired chemical composition. These ingredients are ground into small particles to ma e them more reactive, blended together, and then the resulting raw mix is fed into a cement iln which heats them to extremely high temperatures. %ince the final composition and properties of portland cement are specified within rather strict bounds, it might be supposed that the re8uirements for the raw mix would be similarly strict. As it turns out, this is not the case. ,hile it is important to have the correct proportions of calcium, silicon, aluminum, and iron, the overall chemical composition and structure of the individual raw ingredients can vary considerably. The reason for this is that at the very high

&

temperatures in the iln, many chemical components in the raw ingredients are burned off and replaced with oxygen from the air. Table !." lists 9ust some of the many possible raw ingredients that can be used to provide each of the main cement elements.
Ta'le 3.!: ;xamples of raw materials for portland cement manufacture (adapted from ref. <'=, Table &.').

Calcium
7imestone )arl #alcite Aragonite %hale %ea %hells #ement iln dust

Silicon
#lay )arl %and %hale Fly ash >ice hull ash %lag

*luminum
#lay %hale Fly ash Aluminum ore refuse

Iron
#lay 0ron ore )ill scale %hale 3last furnace dust

The ingredients listed above include both naturally occurring materials such as limestone and clay, and industrial byproduct materials such as slag and fly ash. From Table !." it may seem as if 9ust about any material that contains one of the main cement elements can be tossed into the iln, but this is not 8uite true. )aterials that contain more than minor (or in some cases trace) amounts of metallic elements such as magnesium, sodium, potassium, strontium, and various heavy metals cannot be used, as these will not burn off in the iln and will negatively affect the cement. Another consideration is the reactivity, which is a function of both the chemical structure and the fineness. #lays are ideal because they are made of fine particles already and thus need little processing prior to use, and are the most common source of silica and alumina. #alcium is most often obtained from 8uarried roc , particularly limestone (calcium carbonate) which must be crushed and ground before entering the iln. The most readily abundant source of silica is 8uartz, but pure 8uartz is very unreactive even at the maximum iln temperature and cannot be used. 6rinding and blending prior to entering the iln can be performed with the raw ingredients in the form of a slurry (the wet process) or in dry form (the dry process). The addition of water facilitates grinding. (owever, the water must then be removed by evaporation as the first step in the burning process, which re8uires additional energy. The wet process, which was once standard, has now been rendered obsolete by the development of efficient dry grinding e8uipment, and all modern cement plants use the dry process. ,hen it is ready to enter the iln, the dry raw mix has ?1@ of the particles less than AB Cm in size <&=. 3.! "urning in a .iln / formation of cement clin.er

The next step in the process is to heat the blended mixture of raw ingredients (the raw mix) to convert it into a granular material called cement clin er. This re8uires maximum

temperatures that are high enough to partially melt the raw mix. 3ecause the raw ingredients are not completely melted, the mix must be agitated to ensure that the clin er forms with a uniform composition. This is accomplished by using a long cylindrical iln that slopes downward and rotates slowly (see Figure !DB).
+igure 301( A rotary iln for cement manufacture. The inset shows a view inside the hot iln. (0mage courtesy of the 5ortland #ement Association).

To heat the iln, a mixture of fuel and air is in9ected into the iln and burned at the bottom end (see Figure !DB). The hot gases travel up the iln to the top, through a dust collector, and out a smo estac . A variety of fuels can be used, including pulverized coal or co e, natural gas, lignite, and fuel oil. These fuels create varying types and amounts of ash, which tend to have compositions similar to some of the aluminosilicate ingredients in the raw mix. %ince the ash combines with the raw mix inside the iln, this must be ta en into account in order to correctly predict the cement compassion. There is also an increasing trend to use waste products as part of the fuel, for example old tires. 0n the bestDcase scenario, this saves money on fuel, reduces #$& emissions, and provides a safe method of disposal. The burning process This description refers to a standard dryDprocess iln as illustrated in Figure !DB. %uch a iln is typically about '?B m long and 4 m in diameter, has a downward slope of !D"@, and rotates at 'D& revolutions per minute. (After )indess and Eoung, Fig. !.&)
+igure 301( 2iagram of the reactions occurring in a typical dry process cement iln without a preheater.

The raw mix enters at the upper end of the iln and slowly wor s its way downward to the hottest area at the bottom over a period of 4BDAB minutes, undergoing several different reactions as the temperature increases. 0t is important that the mix move slowly enough to allow each reaction to be completed at the appropriate temperature. 3ecause the initial reactions are endothermic (energy absorbing), it is difficult to heat the mix up to a higher temperature until a given reaction is complete. The general reaction zones (see Figure !DB) are as follows: Dehydration zone (up to ~ 450C): This is simply the evaporation and removal of the free water. ;ven in the .dry process/ there is some adsorbed moisture in the raw mix. Although the temperatures re8uired to do this are not high, this re8uires significant time and energy. 0n the wet process, the dehydration zone would re8uire up to half the length of the iln, while the dry process re8uires a somewhat shorter distance. Calcination zone (450C 00C): The term calcination refers to the process of decomposing a solid material so that one of its constituents is driven off as a gas. At about 4BBF# the bound water is driven out of the clays, and by ABBF# the calcium carbonate is decomposed, releasing carbon dioxide. 3y the end of the calcination zone, the mix consists of oxides of the four main

"

elements which are ready to undergo further reaction into cement minerals. 3ecause calcination does not involve melting, the mix is still a freeDflowing powder at this point. !olid"state reaction zone ( 00 " #$00C): This zone slightly overlaps, and is sometimes included with, the calcination zone. As the temperature continues to increase above G ABBF# there is still no melting, but solidDstate reactions begin to occur. #a$ and reactive silica combine to form small crystals of #&% (dicalcium silicate), one of the four main cement minerals. 0n addition, intermediate calcium aluminates and calcium ferrite compounds form. These play an important role in the clin ering process as fluxing agents, in that they melt at a relatively low temperature of G '!BBF#, allowing a significant increase in the rate of reaction. ,ithout these fluxing agents, the formation of the calcium silicate cement minerals would be slow and difficult. 0n fact, the formation of fluxing agents is the primary reason that portland (calcium silicate) cements contain aluminum and iron at all. The final aluminumD and ironDcontaining cement minerals (#!A and #"AF) in a portland cement contribute little to the final properties. As the mix passes through solidDstate reaction zone it becomes .stic y/ due to the tendency for ad9acent particles to fuse together. Clin%ering zone (#$00C #550C): This is the hottest zone where the formation of the most important cement mineral, #!% (alite), occurs. The zone begins as soon as the intermediate calcium aluminate and ferrite phases melt. The presence of the melt phase causes the mix to agglomerate into relatively large nodules about the size of marbles consisting of many small solid particles bound together by a thin layer of li8uid (see Figure !DB). 0nside the li8uid phase, #!% forms by reaction between #&% crystals and #a$. #rystals of solid #!% grow within the li8uid, while crystals of belite formed earlier decrease in number but grow in size. The clin ering process is complete when all of silica is in the #!% and #&% crystals and the amount of free lime (#a$) is reduced to a minimal level (H'@).
+igure 301: A nodule in the clin ering zone.

Cooling zone: As the clin er moves past the bottom of the iln the temperature drops rapidly and the li8uid phase solidifies, forming the other two cement minerals #!A (aluminate) and #"AF (ferrite). 0n addition, al alis (primarily I) and sulfate dissolved in the li8uid combine to form I&%$" and +a&%$". The nodules formed in the clin ering zone are now hard, and the resulting product is called cement clin er. The rate of cooling from the maximum temperature down to about ''BBF# is important, with rapid cooling giving a more reactive cement. This occurs because in this temperature range the #!% can decompose bac into #&% and #a$, among other reasons. 0t is thus typical to blow air or spray water onto the clin er to cool it more rapidly as it exits the iln.
+igure 301: +odules of portland cement clin er after cooling (5hotograph courtesy of the 5ortland #ement Association).

!uspension preheaters and calciners The chemical reactions that occur in the dehydration and calcination zones are endothermic, meaning that a continuous input of energy to each of the particles of the raw mix is re8uired to complete the reaction. ,hen the raw mix is piled up inside a standard rotary iln, the rate of reaction is limited by the rate at which heat can be transferred into a large mass of particles. To ma e this process more efficient, suspension preheaters are used in modern cement plants to replace the cooler upper end of the rotary iln (see Figure !DB). >aw mix is fed in at the top, while hot gas from the iln heater enters at the bottom. As the hot gas moves upward it creates circulating .cyclones/ that separate the mix particles as they settle down from above. This greatly increases the rate of heating, allowing individual particles of raw mix to be dehydrated and partially calcined within a period of less than a minute. Alternatively, some of the fuel can be burned directly within the preheater to provide even more heating to the suspended particles. The area of the preheater where fuel is burned is called a precalciner. ,ith a precalciner, the particles are nearly completely calcined as they enter the rotary iln. 5reheaters and precalciners save on fuel and increase the rate at which the mix can be moved through the rotary iln.

+igure 301( %uspension preheater.

&rinding and the addition o' gypsu( $nce the nodules of cement clin er have cooled, they are ground bac into a fine powder in a large grinding mill. At the same time, a small amount of calcium sulfate such as gypsum (calcium sulfate dihydrate) is blended into the cement. The calcium sulfate is added to control the rate of early reaction of the cement, as will be discussed in %ection 1.!. At this point the manufacturing process is complete and the cement is ready to be bagged or transported in bul away from the plant. (owever, the cement is normally stored in large silos at the cement plant for a while so that various batches of cement can be blended together to even out small variations in composition that occur over time. #ement manufacturers go to considerable lengths to maintain consistent behavior in their cements over time, with the most important parameters being the time to set, the early strength development, and the wor ability at a given water content. Ce(ent %iln dust As the hot iln gas moves through the iln, it carries with it the smallest particles of the raw mix as well as volatilized inorganic substances such as al alis (sodium and potassium) and chlorides. As the gas cools, the volatiles condense bac round the small particles, and the resulting powder is called cement iln dust (#I2). 0n the old days, the #I2 was simply vented out of the smo estac , after which it would continuously settle out of the air to create a thin coating of grey dust on the surrounding countryside. This is no longer allowed. 0n fact, environmental restrictions even prevent #I2 from being buried in landfills because of the tendency for the al alis and chlorides to leach into groundwater. 0n modern cement plants, the #I2 is removed in the suspension preheater and by and electrostatic precipitators located near the base of the smo estac .

Approximately !B million tons of #I2 are produced worldwide each year <!=, and finding a way to use or dispose of this material has become a rather urgent problem for the cement industry. %ome #I2 can be returned to the iln as raw mix, but this is limited by the tendency for the recycled #I2 to increase the al ali and chloride contents of the cement clin er to unacceptable levels. (igh al ali contents can promote deleterious reactions with aggregate particles that damage the concrete, while chloride ions corrode reinforcing steel. -p to '1@ of the total input of raw ilnstoc can be lost as #I2. 3ecause #I2 contains a high proportion of soluble al ali chlorides and sulfates, an obvious use is as an activator for blended cements, pozzolans, and hydaulic slags. At present, the drawbac s of #I2 as a cement additive seem to outweigh the advantages, although research in this area is ongoing. This is expanded a bit in the following optional section.

3.4.1 The use of cement kiln dust as an additive or activator This section contains a short discussion of the possibility of using cement iln dust (#I2) as a cement additive or as an activator for reactive byproduct materials such as slag and fly ash. The best results have been achieved with ternary blends of #I2, slag, and $5# <;rror: >eference source not found,"=. The $5# offers early strengths while the slag, which is activated by the #I2, improves durability. These blends can absorb 'BD'1@ by weight of #I2 while still generating strengths e8ual to that of a neat $5# paste. (owever, the advantages of such an approach seem to be outweighed by the drawbac s associated with the #I2. These drawbac s include: At levels higher than 'BD'1@, #I2 reduces the strength of concrete. The high al ali content of #I2 presents potential problems with al aliDsilica reaction between the pore fluid and the aggregate. This issue affects all al aliD activated cementitious materials, and is an important reason why al aliDactivated slag (AA%) concrete, which has excellent early and ultimate strength and durability, has seen limited application <1=. The chloride in #I2 presents increased ris for corrosion of reinforcing steel, ma ing #I2 concretes inappropriate for reinforced concrete. #oncrete containing #I2 has a lower wor ability at a given waterJcement ratio. <;rror: >eference source not found=. ,hile this can be addressed by using water reducing admixtures, this adds cost. #I2 is a variable product. The #I2 from different ilns and from different times can have very different compositions. This ma es it difficult to design a concrete mixture with #I2. This problem is compounded by the fact that careful design of the mix has been shown to be needed to generate good properties with #I2 concrete <;rror: >eference source not found,;rror: >eference source not found=.

Table !.".'.' summarizes the reported oxide composition of #I2 used in recent published studies (six varieties from four publications).

Ta'le 3.!.1.1. #hemical oxide composition of six varieties of #I2 (wt@) reported in refs <;rror: >eference source not found,;rror: >eference source not found,4, K=.

#omponent %i$& Al&$! Fe&$! #a$ )g$ %$! +a&$ I&$ #lD 7oss on ignition

>ange (wt@) '&.B L !".! '.' L 4.A '.? L ".1 !".! L 1K.B B.? L &." '.K L 'B.4 B.! L !.A '.& L ?.& B." L ?.' ?.B L &'.B

Average (wt@) 'K.! ".B &.41 "4.! '.1 4.A '.4 ".B ".& '4.1

The variation in the composition shown in Table !.".'.' is even greater than is suggested by the oxide composition, because the mineral composition cannot be assumed from the oxide composition as it can with a cement. For example, 2yer et al. <;rror: >eference source not found= used two types of #I2 with nearly identical #a$ contents. They performed MDray and >ietfeld analysis to determine the minerals, and found that one contained A@ calcite (a form of calcium carbonate) and &K@ calcium hydroxide, while the other contained 1!@ calcite and ?@ calcium hydroxide. %ince calcium hydroxide is consumed in the pozzolanic reaction while calcite is not, the results obtained with the two #I2s in a blended cement will clearly differ greatly. $ther minerals reported by 2yer et al <;rror: >eference source not found= include free lime, anhydrite, 8uartz, thenardite, FriedelNs salt, halite, and syngenite. $ne way to improve the cementing properties of #I2 is to calcine it at temperatures above about 4BBF# to decompose the calcite. Amin et al. <?= calcined #I2 for & hrs at ?BB, ABB, and 'BBBF#, and found that calcite MDray pea s disappeared in all cases, that the chloride was nearly fully removed at 'BBBF#, the al ali content was unchanged, and that there was an increase in the amount of D #&%. The hydraulic properties of a #I2J slag cement mixture improved with the #I2 calcining temperature. 0t is not clear whether it would be economically viable to perform an additional calcining step on #I2, however. $ne way to reduce the negative effects of #I2 variability is through precasting, where the composition and reactivity of the material can be measured and controlled at the production facility to ensure consistent results. Another advantage of precasting is that elevated temperature curing can be used to accelerate the hydration process. A logical place to locate a manufacturing plant would be ad9acent to a cement plant where the #I2 is produced, because excess heat from the ilns could be reused for the elevated temperature curing. 3.2 3roperties of the ma4or cement minerals

About ABDA1@ of a 5ortland cement is comprised of the four main cement minerals, which are #!%, #&%, #!A, and #"AF, with the remainder consisting of calcium sulfate, al ali sulfates, ?

unreacted (free) #a$, )g$, and other minor constituents left over from the clin ering and grinding steps. The four cement minerals play very different roles in the hydration process that converts the dry cement into hardened cement paste. The #!% and the #&% contribute virtually all of the beneficial properties by generating the main hydration product, #D%D( gel. (owever, the #!% hydrates much more 8uic ly than the #&% and thus is responsible for the early strength development. The #!A and #"AF minerals also hydrate, but the products that are formed contribute little to the properties of the cement paste. As was discussed in the previous section, these minerals are present because pure calcium silicate cements would be virtually impossible to produce economically. The crystal structures of the cement minerals are 8uite complex, and since these structures do not play an important role in the properties of cement paste and concrete we will only present the most important features here. )ore detailed information can be found in the boo by Taylor <;rror: >eference source not found=. The hydration reactions of the cement minerals are covered in %ection1.!. Tricalciu( !ilicate (C$!) #!% is the most abundant mineral in portland cement, occupying "BLKB wt@ of the cement, and it is also the most important. The hydration of #!% gives cement paste most of its strength, particularly at early times. 5ure #!% can form with three different crystal structures. At temperatures below A?BF# the e8uilibrium structure is triclinic. At temperatures between A?BF# L 'BKBF# the structure is monoclinic, and above 'BKBF# it is rhombohedral. 0n addition, the triclinic and monoclinic structures each have three polymorphs, so there are a total of seven possible structures. (owever, all of these structures are rather similar and there are no significant differences in the reactivity. The most important feature of the structure is an aw ward and asymmetric pac ing of the calcium and oxygen ions that leaves large .holes/ in the crystal lattice. ;ssentially, the ions do not fit together very well, causing the crystal structure to have a high internal energy. As a result, #!% is highly reactive. The #!% that forms in a cement clin er contains about !D"@ of oxides other than #a$ and %i$&. %trictly spea ing, this mineral should therefore be called alite rather than #!%. (owever, as discussed in %ection !.&, we will avoid using mineral names in this monograph. 0n a typical clin er the #!% would contain about ' wt@ each of )g$, Al&$!, and Fe&$!, along with much smaller amounts of +a&$, I&$, 5&$1, and %$! <;rror: >eference source not found=. These amounts can vary considerably with the composition of the raw materials used to ma e the cement, however. $f the three ma9or impurities, )g and Fe replace #a, while Al replaces %i. $ne effect of the impurities is to .stabilize/ the monoclinic structure, meaning that the structural transformation from monoclinic to triclinic that would normally occur on cooling is prevented. )ost cements thus contain one of the monoclinic polymorphs of #!%. Dicalciu( !ilicate (C)!) As with #!%, #&% can form with a variety of different structures. There is a high temperature structure with three polymorphs, a structure in that is in e8uilibrium at intermediate temperatures, and a low temperature structure. An important aspect of #&% is that D#&% has a very stable crystal structure that is completely unreactive in water. Fortunately,

the structure is easily stabilized by the other oxide components of the clin er and thus the form is never present in portland cement. The crystal structure of #&% is irregular, but considerably less so than that of #!%, and this accounts for the lower reactivity of #&%. The #&% in cement contains slightly higher levels of impurities than #!%. According to Taylor <;rror: >eference source not found=, the overall substitution of oxides is "D4@, with significant amounts of Al&$!, Fe&$!, and I&$. Tricalciu( *lu(inate (C$*) Tricalcium aluminate (#!A) comprises anywhere from zero to '"@ of a portland cement. 7i e #!%, it is highly reactive, releasing a significant amount of exothermic heat during the early hydration period. -nfortunately, the hydration products of formed from #!A contribute little to the strength or other engineering properties of cement paste. 0n certain environmental conditions (i.e., the presence of sulfate ions), #!A and its products can actually harm the concrete by participating in expansive reactions that lead to stress and crac ing. 5ure #!A forms only with a cubic crystal structure. The structure is characterized by #aO& atoms and rings of six Al$" tetrahedra. As with #!%, the bonds are distorted from their e8uilibrium positions, leading to a high internal energy and thus a high reactivity. %ignificant amounts of #a$ and the Al&$! in the #!A structure can be replaced by other oxides, and at high levels of substitution this can lead to other crystal structures. The #!A in portland cement clin er, which typically contains about '!@ oxide substitution, is primarily cubic, with smaller amounts of orthorhombic #!A. The #!A and #"AF minerals form by simultaneous precipitation as the li8uid phase formed during the clin ering process cools, and thus they are closely intermixed. This ma es it difficult to ascertain the exact compositions of the two phases. The cubic form generally contains G"@ substitution of %i$&, G1@ substitution of Fe&$!, and about '@ each of +a&$, I&$, and )g$. The orthorhombic form has similar levels, but with a greater (G1@) substitution of I&$. Tetracalciu( *lu(ino'errite (C4*+) A stable compound with any composition between #&A and #&F can be formed, and the cement mineral termed #"AF is an approximation that simply the represents the midpoint of this compositional series. The crystal structure is complex, and is believed to be related to that of the mineral perovs ite. The actual composition of #"AF in cement clin er is generally higher in aluminum than in iron, and there is considerable substitution of %i$& and )g$. Taylor <;rror: >eference source not found= reports a typical composition (in normal chemical notation) to be #a&AlFeB.4)gB.&%iB.'1TiB.1$1. (owever, the composition will vary somewhat depending on the overall composition of the cement clin er.

3.5

#ineral and o)ide composition of portland cement

,hen considering the composition of a portland cement, the most important factors are the relative amounts of each of the cement minerals, the amount of calcium sulfate, and the total amount of al alis (+a and I). (owever, when performing a direct chemical analysis of the cement (see %ection ), what is actually determined is the relative concentrations of each of the elements in the cement. These values are then normally converted to the weight fraction of each element in oxide form, since the cement always contains an amount of oxygen sufficient to

'B

charge balance the other elements. This, along with a few other measurements, is then reported as the oxide composition of the cement (see Table !."). 0t should be noted that only a very small fraction of a cement consists of independent oxides of the form shown in Table !.".

Ta'le 3.!. $xide composition of a 5ortland cement. >anges in column & are for $5# Types 0DP, as given in ref. <;rror: >eference source not found=. Palues in column ! are averaged results from '&1 laboratories, from a study conducted by the #ement and #oncrete >esearch 7aboratory (##>7) at +0%T in &BBB (proficiency cement Q'!1).

$)ide #a$ %i$& Al&$! Fe&$! )g$ 5&$1 Ti$& +a&$ I&$ %$! 7oss on 0gnition 0nsoluble residue Free #a$

-ange 6)

(%t

Cement 7132 (%t6) 4!.?' &'."1 "."1 !.BK &."& B.'' B.&& B.&B B.?! &."4 B.?' B.'4 B.4"

4B.& L 44.! '?.4 L &!." &." L 4.! '.! L 4.' B.4 L ".? DD DD B.B1 L '.&B (+a&$ e8uiv) '.K L ".4 DD DD DD

The loss on ignition reported in Table !." is the weight lost when the cement was heated to 'BBBF#. At this temperature any water or #$& present in the cement specimen is driven off. The insoluble residue is the mass of material that is not dissolved by acid. This material generally consists of mineral crystals such as 8uartz, and a rule of thumb used by cement chemists is that the insoluble reside is &J! %i$& and 'J! Al&$!. The free #a$ (often called the .free lime/) is the amount of calcium oxide present as #a$, that is, not bound into the cement minerals. The free #a$ is already counted in the weight fraction of #a$, so the total amount of cement mass accounted for in Table !." is the sum of all the rows except for the last, which is AA.A@. The remaining B.'@ is li ely in the form of other trace elements that were not tested for. As was noted above, the mineral composition of the cement is more useful than the oxide composition. (owever, the mineral composition is more difficult to determine accurately. 0t can

''

be measured directly using techni8ues such as light microscopy, powder MDray diffraction, or %;) MDray analysis, or it can be calculated from the oxide composition. According to a summary by Taylor <;rror: >eference source not found= powder M>2 analysis is difficult because of the dominance of the #!% pea s and the extensive amount of pea overlap, and accurate results are obtained only when very careful procedures and analysis are followed. )icroscopy, which involves a pointDcount method, tends to be operator dependent and unreliable for separating the #!A and #"AF phases. 5robably the best experimental method for directly measuring the mineral composition of cement is 8uantitative %;) imaging <A,'B=. ,ith this method, the cement is dispersed in a hardened resin which is then polished to evenly expose all of the minerals within the cement particles. A location on the specimen is selected at a relatively low magnification (&1BD1BBM), and then the %;) is programmed to automatically scan over the selected image ta ing thousands of xDray measurements in a regular grid pattern. The result is a set of .dot map/ images representing the intensity of the xDray signal for each of elements of interest (typically #a, %i, Al, Fe, %, I, and )g). A computer program then determines which mineral (if any) is present at each pixel from the relative elemental intensities at that location. An added advantage of this techni8ue is that the mineral information can be combined with the original %;) bac scattered image to create a colorDcoded image of the cement showing the composition distribution (see Figure !DB). %uch an image is particularly useful as a realistic starting microstructure for a digital model of cement hydration <;rror: >eference source not found=.

QuickTime and a TIFF (Uncompressed) decompressor are needed to see this picture.

+igure 301( %;)JMDray image of a portland cement with each solid phase assigned a different color. 0mage is &14 Cm by &BB Cm. #olors correspond to the following phases: redD#!%, a8uaD#&%, greenD#!A, orangeD#"AF, pale greenD calcium sulfates, yellowDI&%$", and whiteDfree #a$. 0mage ta en from ref. <;rror: >eference source not found=.

'&

3ecause this techni8ue re8uires a significant amount of time on a goodD8uality %;) instrument, it is clearly not ideal for routine analysis. The most attractive option in terms of time and expense is to calculate the mineral composition from the measured oxide composition. 0f this is done carefully, reasonably accurate results can be obtained. This is discussed below. The ,ogue calculation o' ce(ent (ineral co(position A simple estimate of the phase composition of a portland cement can be obtained from the oxide composition if one assumes that the four main cement minerals occur in their pure form. ,ith this assumption, all of the Fe&$! is assigned to #"AF and the remaining Al&$! is assigned to #!A. This leaves a set of two linear e8uations to be solved for the amounts of #&% and #!%. This method is named after the cement chemist >.(. 3ogue. A standardized version of this simple method is given in A%T) # '1B <''=. There are two sets of e8uations, based on the ratio of AJF in the cement (both inputs and outputs are in weight percent): -' *.+ / 0014 #!% * ".BK'# L K.4BB% L 4.K'?A L '."!BF L &.?1& #&% * &.?4K% L B.K1""#!% #!A * &.41BA L '.4A&F #"AF * !.B"!F -' *.+ 2 0014 #!% * ".BK'# L K.4BB% L "."KAA L &.?1AF L &.?1& #&% * &.?4K% L B.K1""#!% #!A * B #"AF * &.'BBA O '.KB&F

!.' (a)

!.' (b)

Applying ;8s. !.'(a) to the oxide composition given in column ! of Table !." results in: 11."@ #!%, 'A.K@ #&%, 4.4@ #!A, and A.!@ #"AF. ,hile this method is simple, it is not very accurate. 0n particular, there is a tendency to overestimate the amount of #&% and underestimate the amount of #!%. $ne study found that the 3ogue calculation gave #!% contents that were on average ?@ lower than of those obtained from light microscopy (a techni8ue that is reasonably accurate for estimating #!%). Also, the sum of the four minerals tends to be much less than 'BB@. ,hile there are several sources of error in the 3ogue calculation, the most significant is the assumption that the minerals occur in their pure form. 0n fact, the actual composition of each of the cement minerals differs rather significantly from its assumed formula. This was discussed in %ection !.1. 3y ta ing into account the nown tendency for other oxides to substitute into the cement minerals, a more accurate (but also more complex) calculation can be devised, as described below. 3odi'ied ,ogue calculation based on actual (ineral co(positions To accurately calculate the mineral composition of a cement from the oxide composition it is necessary to estimate the true composition of the cement minerals in the clin er and to account for phases other than the main four cement minerals such as gypsum. A more

'!

complete description of this useful techni8ue is given in ref. <'&=. To perform this type of calculation re8uires an accurate and detailed oxide composition, such as that shown in Table !.". The process can be divided into the following steps: #) -' the o4ide co(position is 'or a ce(ent (i0e05 clin%er ground 6ith gypsu()5 subtract the contribution o' gypsu( 'ro( the o4ide co(position0 As an example, assume that & wt@ gypsum (# (&) was added to the clin er to ma e the cement. 6ypsum consists of !!@ # and "K@ by weight, so we would subtract B.44@ from the total # content and B.A"@ from the total content (the remaining &B wt@ of gypsum is bound water, which would be lost on ignition). 0f the amount of gypsum added to the cement is not nown, then a reasonable estimate must be made. A typical clin er content of B.1@ could be assumed, so the remaining would be in gypsum. #learly this will add a possible source of error to the calculation. )) 7sti(ate the a(ount o' C5 !5 *5 and + in the clin%er that is present in phases other than the (ain ce(ent (inerals5 and subtract these a(ounts 'ro( the o8erall o4ide co(positions0 The content of the clin er (having subtracted the gypsum in step ', if the oxide composition is for a cement) can be distributed in a few different phases, including I&%$", +a&%$", #a&I(%$")!, and as impurities in the cement minerals. The only concern at this point is the amount of # bound into phases such as #a&I(%$")!. 0f the ratio J(+OI) in the clin er is less than about ', this will be a small correction that can probably be safely ignored. 0f not, a procedure for estimating this amount is given in ref. <;rror: >eference source not found=. The insoluble residue of the clin er or cement should be considered 4K@ % and !!@ A, and these amounts subtracted from the total amounts of % and A. 0f the insoluble residue is not nown, a value of B.! wt@ can be assumed <;rror: >eference source not found=. The amount of free #a$ should be subtracted from the # content. Finally, if there is any #$& in the cement or clin er, it should be assumed to be in the form of calcium carbonate (#a#$!)R this means that "?@ of the #$& content should be subtracted from the total amount of #. After these corrections, all of the remaining #, %, A, and F can be assumed to reside in the main cement minerals. $) 7sti(ate the true co(positions o' the 'our (ain ce(ent (inerals0 These compositions can be measured directly using an xDray microanalysis techni8ue similar to that used to generate Figure !DB. (owever, this obviously negates all of the advantages of performing the calculation. As noted by Taylor <;rror: >eference source not found=, modern $5#s are processed in such a way that the compositions of the minerals in the clin er are fairly constant, and thus average compositional values give a good approximation. Table !." lists the typical compositions reported by Taylor <;rror: >eference source not found=, along with the theoretical compositions of the pure minerals.
Ta'le 3.!. Typical compositions of the cement minerals as found in portland cement clin er and cement (after <;rror: >eference source not found=, and also given in <;rror: >eference source not found=). The numbers in parentheses are the values for the pure minerals. All values are wt@.

'"

C #!% #&% #!A #"AF K'.4 (K!.K) 4!.1 (41.') 14.4 (4&.!) "K.1 ("4.&)

S &1.& (&4.!) !'.1 (!".A) !.K (B) !.4 (B)

* '.B (B) &.' (B) !'.! (!K.K) &'.A (&'.B)

+ B.K (B) B.A (B) 1.' (B) &'." (!&.A)

4 ) !et up and sol8e a syste( o' 'our e9uations and 'our un%no6ns to 'ind the (ineral co(position o' the ce(ent0 $nce the total amount of #, %, A, and F residing in the cement minerals has been calculated by ad9usting the total oxide composition of the cement or clin er (steps ' and &) and the ratio of the oxides within each of the main cement minerals has been estimated (step !), a system of four e8uations in four un nowns can be set up and solved for the amount (in wt@) of each cement mineral. -sing the cement oxide composition for proficiency cement Q'!1 given in Table !." and the mineral oxide compositions given in Table !." results in the following set of e8uations: B.K'4C$! O B.4!1C)! O B.144C$* O B."K1C4*+ * 4&.1& (#) B.&1&C$! O B.!'1C)! O B.B!KC$* O B.B!4C4*+ * &'.!" (%) B.B'BC$! O B.B&'C)! O B.!'!C$* O B.&'AC4*+ * "."B B.BBKC$! O B.BBAC)! O B.B1'C$* O B.&'"C4*+ * !.BK (A) (F) !.&

;8s. !.& can be solved using a computer program or spreadsheet or a programmable calculator (or by hand if you have the patienceS). Table !." compares the mineral compositions for cement Q'!1 calculated using the standard 3ogue calculation (;8s. !.'(a)) and the modified calculation of ;8s. !.&, along with an experimental determination using %;)JMDray microanalysis <;rror: >eference source not found=.
Ta'le 3.!. The calculated and measured mineral composition of cement Q'!1 (see Table !." for the oxide composition). )ineral %tandard 3ogue )odified 3ogue %;)JMDray (ref. <;rror: >eference source not found=a) 4&.' '4." 4.& ?.4 A!.!

#!% #&% #!A #"AF Total

11." 'A.K 4.4 A.! A'.B

4".K '"." !."K 'B.? A!.!

'1

Palues reported in ref. <;rror: >eference source not found= were converted from vol@ to wt@ and renormalized to a total of A!.!@

#omparison of the values shown in Table !." indicates that the modified 3ogue gives much higher #!% content and a lower #&% content than the standard 3ogue, and that these values are in better agreement with the %;)JMDray analysis. For the #!A and #"AF phases the agreement is not as good. 0t is difficult to draw any conclusions on this, because it is difficult to distinguish the #!A and #"AF experimentally. 0n any case, the #!% content of the cement is the most important aspect of the mineral composition, and this is certainly more accurate with the modified 3ogue. According to the modified 3ogue calculation, the total weight percent occupied by the cement minerals is A!.!@. From Table !." we can see that the free lime, loss on ignition, and insoluble reside account for '.4@. $ther phases and their estimated amounts are: &@ calcium sulfate (i.e. added gypsum), '.&@ I&%$", B.?@ )g$ (the mineral periclase), and B.&@ +a&%$".

3.8

The fineness and particle si9e distri'ution of 3ortland cement

The final step in the production of cement is to grind the clin er pellets (and added gypsum) into a fine powder. The resulting individual cement particles have a variety of angular shapes, and a wide range of sizes (see Figure !DB below). )ost of the cement particles will consist of regions of each of the four main cement minerals.

+igure 301( %;) micrograph of a portland cement showing the wide range of particle sizes and shapes.

The size of a cement particle has an important effect on the rate at which it will hydrate when exposed to water. As it reacts, a layer of hydration product forms around the outside of the particle, separating the unreacted core of the particle from the surrounding water. As this layer grows thic er, the rate of hydration slows down. Therefore, a small particle will react much more 8uic ly than a large particle. A particle that has a diameter of ' Cm will react completely in about ' day, whereas a particle with a diameter of 'B Cm will react completely in about ' month. 5articles larger than about 1B Cm will probably never become fully reacted, even if there is a sufficient source of water. #learly, the particle size distribution is critical for controlling the rate at which a cement sets and gains strength. There must be a certain amount of small particles to ensure that the cement sets in a reasonable amount of time, but if there are too many small particles the cement will set too 8uic ly, leaving no time for mixing and placing. Fortunately, the grinding process has a natural tendency to produce a wide range of particle sizes, so this is not a problem. Figure !DB shows the particle size distribution for a typical portland cement. About 'B wt@ of the cement is made of particles larger than 1B Cm, and only a few wt@ is particles larger than AB Cm. $n the fine end, less than 'B@ of the cement is particles smaller than & Cm. %uch a distribution is traditionally measured by a sedimentation process, in which the size of the particles is determined from the rate at which they sin in a column of li8uid. A faster and more precise method is laser particle analysis. A small amount of powder is dispersed in a solvent

'4

such as methanol and then pumped through a clear tube in front of a laser. The particle size distribution is calculated from the diffraction pattern generated by the scattering of the laser light off of the particles.

+igure 301( The particle size distribution of a typical portland cement. 7eft: the cumulative distribution, where the y axis value is the wt@ of the particles that are smaller than the xDaxis value. >ight: a histogram showing the wt@ of particles in different size ranges.

The average particle size, defined as the size for which 1B wt@ of the cement consists of larger particles, is sometimes used to characterize the overall fineness of the cement (in Figure !DB the average particle size is ). (owever, this is only one aspect of the particle size distribution, the other being its breadth. 0f two cements have the same average particle size but different distributions, the cement with the broader distribution will have more fine particles and thus will exhibit a faster rate of early hydration. A better parameter for describing the fineness of the cement (at least in terms of nowing how reactive it will be at early times) is the specific surface area, because most of the surface area comes from the smallest particles. The most common method for characterizing the surface area of a cement is the 3laine air permeability test, which is described by an A%T) standard <'!=. This test is based on the fact that the rate at which air can pass through a porous bed of particles under a given pressure gradient is a function of the surface area of the powder. A chamber of nown cross sectional area and volume is filled with a nown mass of cement, and then the time re8uired to pass a nown volume of air through the powder is measured. ,hile the surface area can in theory be calculated explicitly from this data, in practice the surface area is determined through an empirical e8uation developed by measuring powders of a nown surface area using the same instrument. The resulting value, called the 3laine fineness, is today expressed in units of m&J g, although in previous times it was expressed in cm&Jg. The 3laine fineness of $5# usually ranges from !BB L 1BB m&J g (!BBB L 1BBB cm&Jg). 0t should be noted that the specific surface area is a notoriously difficult parameter to pin down because most materials have features at many different length scales, so that the more sensitive the measurement, the more surface area will be found. Thus different measuring techni8ues give different values. ,hen the surface area of a cement is measured using gas sorption, a techni8ue that measures how many gas molecules can fit on the surface of a solid, the resulting value is &D! times higher than the 3laine value <;rror: >eference source not found=. This occurs because the cement particles have a rough, irregular surface with internal crac s and pores. The specific surface area of a hydrated cement paste is much higher than that of the dry cement powder, because the main hydration product, the #D%D( gel, has a system of very small internal pores. This extremely important aspect of the microstructure of cement paste will be discussed in detail in #hapter K. As a final note, the specific surface area of a cement powder tells little about the size of the largest particles or the breadth of the particle size distribution. These parameters affect the wor ability of the cement at a given waterJcement ratio and the degree of hydration and strength development over long periods of time, among other factors. +o single parameter can substitute for a full particle size distribution.

'K

3.:

Types of 3ortland cement

The A%T) has designated five types of portland cement, designated Types 0DP. 5hysically and chemically, these cement types differ primarily in their content of #!A and in their fineness. 0n terms of performance, they differ primarily in the rate of early hydration and in their ability to resist sulfate attac . The general characteristics of these types are listed in Table !.". The oxide and mineral compositions of a typical Type 0 portland cement were given in Tables !." and !.4.
Ta'le 3.!. 6eneral features of the main types of portland cement. Classification Type I 6eneral purpose Characteristics Fairly high #!% content for good early strength development 7ow #!A content (H?@) 6round more finely, may have slightly more #!% 7ow content of #!% (H1B@) and #!A Pery low #!A content (H1@) +o #"AF, low )g$ *pplications 6eneral construction (most buildings, bridges, pavements, precast units, etc) %tructures exposed to soil or water containing sulfate ions >apid construction, cold weather concreting )assive structures such as dams. +ow rare. %tructures exposed to high levels of sulfate ions 2ecorative (otherwise has properties similar to Type 0)

Type II Type III Type I; Type ; White

)oderate sulfate resistance (igh early strength 7ow heat of hydration (slow reacting) (igh sulfate resistance ,hite color

The differences between these cement types are rather subtle. All five types contain about K1 wt@ calcium silicate minerals, and the properties of mature concretes made with all five are 8uite similar. Thus these five types are often described by the term .ordinary portland cement/, or $5#. Types 00 and P $5# are designed to be resistant to sulfate attac . %ulfate attac is an important phenomenon that can cause severe damage to concrete structures. 0t is a chemical reaction between the hydration products of #!A and sulfate ions that enter the concrete from the outside environment. The products generated by this reaction have a larger volume than the reactants, and this creates stresses which force the concrete to expand and crac . Although hydration products of #"AF are similar to those of #!A, they are less vulnerable to expansion, so the designations for Type 00 and Type P cement focus on eeping the #!A content low. There is actually little difference between a Type 0 and Type 00 cement, and it is common to see cements meeting both designations labeled as .Type 0J00/. The phenomenon of sulfate attac will be discussed in much more detail in %ections 1.! and '&.!, but it should be noted here that the most effective way to prevent sulfate attac is to eep the sulfate ions from entering the concrete in the first place. This can be done by using mix designs that give a low permeability

'?

(mainly by eeping the wJc ratio low) and, if practical, by putting physical barriers such as sheets of plastic between the concrete and the soil. Type 000 cement is designed to develop early strength more 8uic ly than a Type 0 cement. This is useful for maintaining a rapid pace of construction, since it allows castDinDplace concrete to bear loads sooner and it reduces the time that precast concrete elements must remain in their forms. These advantages are particularly important in cold weather, which significantly reduces the rate of hydration (and thus strength gain) of all portland cements. The downsides of rapidD reacting cements are a shorter period of wor ability, greater heat of hydration, and a slightly lower ultimate strength. Type 0P cement is designed to release heat more slowly than a Type 0 cement, meaning of course that it also gains strength more slowly. A slower rate of heat release limits the increase in the core temperature of a concrete element. The maximum temperature scales with the size of the structure, and Type 000 concrete was developed because of the problem of excessive temperature rise in the interior of very large concrete structures such as dams. Type 0P cement is rarely used today, because similar properties can be obtained by using a blended cement. ,hite portland cement (,5#) is made with raw ingredients that are low in iron and magnesium, the elements that give cement its grey color. These elements contribute essentially nothing to the properties of cement paste, so white portland cement actually has 8uite good properties. 0t tends to be significantly more expensive than $5#, however, so it is typically confined to architectural applications. ,5# is sometimes used for basic cements research because the lac of iron improves the resolution of nuclear magnetic resonance (+)>) measurements.

3.<

Synthetic minerals used for research purposes

The hydration of portland cement is a complicated process because all of the cement minerals are hydrating at the same time, forming a wide variety of hydration products. 7i e all scientists who study complex systems, cement chemists often prefer to study simpler systems consisting of one or perhaps two minerals. The advantage of this is that the results are much easier to analyze and interpret. The disadvantage, of course, is that the behavior of the simple system may or may not be characteristic of the full system. Fortunately, the pure mineral #!% is in many respects a good model system for portland cement. The development of strength, the early hydration inetics, and the main hydration products in a #!% paste are all very similar to those of portland cement. 5ure cement minerals cannot be obtained by separating them out from a cement, but must be manufactured separately. This is not particularly difficult, but it can be time consuming and labor intensive, and the amount of material formed is generally 8uite small. The starting materials should be reagent grade #a#$!, %i$&, Al&$!, and Fe&$! with low impurity levels and a small particle size to enhance reactivity. %ilica fume and colloidal silica are good sources of %i$&, but 8uartz will also wor as long as the particles are smaller than about "B Cm. These reactants should be combined in the correct mole ratio for the mineral being synthesized. The formation of the pure minerals is by solidDstate reaction, which re8uires close contact between the reactants. Therefore the reactants should be well mixed in the form of slurry made with a solvent such as ethanol. The solvent should then be fully evaporated and the resulting wellDmixed powder pressed into a pellet to bring the reactants closer together.

'A

To fire the pellets, they should be placed in a platinum crucible and heated in a highD temperature electric furnace. The heating rate should be slow, with a ' hour hold near ABBF# to fully decompose the #a#$!. The hold time at the highest temperature should be about & hoursR longer times do not noticeably increase the conversion. The conversion of the reactants into minerals does not occur as readily as it does during the clin ering process, because of the lac of a li8uid phase. -sually it is necessary to grind and repress the pellet and fire it a second or even a third time before all of the starting material reacts. The degree of reaction can be evaluated by performing MDray diffraction scans to see if there are pea s corresponding to unreacted material or intermediate compounds. 0n some cases it will be desirable to include impurity components in order to stabilize a highDtemperature polymorph or to more closely simulate the composition of the minerals as found in cement clin er. As reported in ref. <'"=, there are two issues to watch out for. %ome components, including boron, sodium, potassium, sulfur, uranium, and fluorine will volatilize at high temperatures, meaning that they may end up condensed on the inside of the furnace rather than in the solid product. #overing the crucible will help but the composition of the product should be measured independently. Also, phosphorous, iron, nic el, cobalt, and other pDbloc elements and transition metals in their elemental form can damage the platinum crucible <;rror: >eference source not found=. This may not be a problem if the conditions are oxidizing and the concentrations are low, but should be ept in mind if you are using someone elseNs crucible. (ere are some additional instructions and suggestions for ma ing each of the main four cement minerals, ta en from references ;rror: >eference source not found and '1. Tricalciu( silicate (C$!) To ma e pure triclinic #!%, fire pellets at '"1BD'1BBF#. 6rind up and refire at least once, using M>2 to ensure that no unreacted #a$ remains. To form the monoclinic form of #!% found in cement re8uires stabilizing impurities. $ne possibility is to add up to ' wt@ of )g$ and Al&$! in place of %i$&. Dicalciu( silicate (C)!) -se of pure reactants will result in the unreactive gamma form of #&%. To stabilize the beta form found in cement re8uires impurities such as )g$ and Al&$! in amounts generally H ' wt@. 3&$! is an excellent stabilizer but is not found in cement. The firing procedure is the same as for #!%, but lower temperatures (down to G'!1BF#) may be used. >egrind as needed, chec ing for free #a$. >apid cooling by removing the crucible from the furnace will help stabilize the beta form. %ee the section below on the 5echini process for ma ing a highly reactive form of D#&% using a different method. Tricalciu( alu(inate (C$*) The firing temperature should be G '"&1F#, and multiple grindings may be needed. 0mpurities are not needed for stabilization. Tetracalciu( alu(ino'errite (C4*+) +ote that the stoichiometry of the cement mineral tends to be rather different than the formula suggests (see %ection !.1). The ratio of A to F can be varied, but #J(AOF) should be e8ual to &. 0mpurities are not needed for stabilization. The firing temperature should be G &B

'!KBF# and more than one grinding may be needed. +or(ation o' 'ine po6ders (:echini process) As discussed above, the formation of solid compounds by solidDstate reaction is limited by the need for the individual reactants to be in close proximity. 0f the reactants can be mixed on an atomic scale prior to firing, it is possible to form solid compounds at much lower temperatures than those listed above. This can be accomplished by forming intermediate chemical compounds. An important side effect of a lower firing temperature is a higher surface area product, since there is less opportunity for finely divided material to sinter. A high surface area product is particularly useful for investigating #&%, because its reactivity with water is limited by its relatively low solubility. The solubility of a powder is proportional to its surface area. Another advantage of this method is that pure D#&% can be formed, i.e. with no stabilizing impurities. The transformation to the more stable gamma phase is prevented by the high surface area. As summarized by (ong and Eoung <'4=, reactive #&% can be formed by a number of methods, including sol gel and spray drying, evaporative decomposition of solutions, thermal decomposition of the mineral hillebrandite or of calcium silicate hydrate phases, and by the 5echini process <'K==. The latter process, summarized briefly below, is particularly useful because the surface area can be accurately controlled by varying the firing temperature <;rror: >eference source not found,;rror: >eference source not found=. Table !." shows the surface area (measured by gas sorption) and the e8uivalent particle diameter (calculated from the surface area assuming all particles are the same size) of #&% formed at different calcination temperatures.
Ta'le 3.!. ;ffect of firing temperature on the surface area of dicalcium silicate (#&%) powder, after <;rror: >eference source not found=. Firing Temperature (F#) KBB ?BB ABB 'BBB '&BB '!BB 3;T %urface Area (m&Jg) "B !' '4 4 '.& '.' ;8uiv. 2iam . (Cm) B.B"1 B.B1A B.''" B.!BB '.4B '.K1

The surface area is a strong function of the firing temperature, particularly in the range of ?BB L '&BBF#. (ong and Eoung <;rror: >eference source not found= found that the #&% fired at KBBF# could be fully hydrated in less than " days, whereas #&% formed at '!BBF# or higher re8uires more than &? days. To form #&% by the 5echini process, the source of silica is a colloidal silica slurry

&'

consisting of nanometerDsized %i$& particles in water (this is commercially available). The correct stoichiometric amount of #a(+$!)& is dissolved in the slurry, causing the reactants to be mixed on the scale of nanometers. This solution is stabilized by combining it with a resin composed of citric acid monohydrate and ethylene glycol. As the mixture of reactants and resin is dried in a 5yrex bea er over a hot plate, it forms a crusty, papery solid that can be easily ground into a fine powder before firing. For more details see refs. <;rror: >eference source not found= and <;rror: >eference source not found=.

-eferences

&&

' . %.(. Iosmat a, 3. Ier hoff, and ,.#. 5anarese, 2esign and #ontrol of #oncrete )ixtures, 5ortland #ement Association, % o ie, 07. '"th ;d., &BB&. &. (.F.,. Taylor, #ement #hemistry, Thomas Telford, 7ondon, &nd ;d., 'AAK. ! . T.2. 2yer, T.;. (alliday, and >.I. 2hir, .An investigation of the hydration chemistry of ternary blends containing cement iln dust,/ ;0 3ater0 !ci0 3!, pp. "AK1D"A?! ('AAA). " . I.T. ,ang, ).%. IonstaD6doutos, and %.>. %hah, .(ydration, rheology, and strength of ordinary 5ortland cement ($5#)Dcement iln dust (#I2)Dslag binders,/ *C- 3ater0 ;. <<, pp. 'K!D'KA (&BB&). 1 . %.D2. ,ang, .Al aliDactivated slag cement and concrete: a review of properties and problems,/ *d80 Ce(0 <es. 8, pp. A!D'B& ('AA1). 4 . T.). %alem and %.). >agai, .;lectrical conductivity of granulated slag cement iln dustD silica fume pastes at different porosities,/ Ce(0 Concr0 <es. 31, pp. K?'DK?K (&BB'). K . Eoussef, .-tilization of cement iln dust in the manufacture of wall tiles,/ -nd0 Cera(. 22, pp. 'D? (&BB&). ? . A.). Amin, ;ebied, and (. ;ldidamony, .Activation of 6ranulated %lag with #alcined #ement Iiln 2ust,/ !ilicates -ndustriels 51, pp. 'BAD''1 ('AA1). A . 2.5. 3entz and 5.;. %tutzman, .%;) Analysis and #omputer )odeling of (ydration of 5ortland #ement,/ pp. 4BDK! in 5etrography of #ementitious )aterials, %.). 2ehayes, 2. %tar , ;ds. A%T), 5hiladelphia, 'AA". 'B . 2.5. 3entz, M. Feng, #.DT. (aec er, and 5.;. %tutzman, .Analysis of ##>7 5roficiency #ements '!1 and '!4 using #;)(E2!2,/ +0%T 0nternal >eport 1545, August &BBB. '' . A%T) #'1B '& . (.F.,. Taylor, .)odification of the 3ogue calculation,/ *d80 Ce(0 <es0 2, pp. K!DKK ('A?A). '! . A%T) # &B" 3laine fineness '" . %. 3arnett and 2. 3rew, .2##Uaberdeen: 2atabase of cementitious material synthesis,/ http:..6660abdn0ac0u%.che(istry.research.ce(ent.dcc.inde40ht(. '1 . >. 3erliner, #. 3all, and 5.3. ,est, .+eutron powder diffraction investigation of model cement compounds,/ Ce(0 Concr0 <es0 28, pp. 11'D1K1 ('AAK). '4 . %.D(. (ong and T.F. Eoung, .(ydration Iinetics and 5hase %tability of 2icalcium %ilicate %ynthesized by the 5echini 5rocess,/ ;0 *(0 Cera(0 !oc0 :2, pp. '4?'D'4?4 ('AAA). 'K . 0. +ettleship, T.7. %hull, and ,.). Iriven, .#hemical preparation and phase stability of #a&%i$" and %r&%i$" powders,/ ;0 7uropean Cera(0 !oc0 11, pp. &A'D&A? ('AA!).