Acomprehensive understanding ofthe chemical and physical aspectsof raw material transformationinto clinker is an important foun-

dation to increasing production, reducingcosts, and improving quality at the cementplant. Clinker formation relies on numerouskiln feed properties and pyroprocessingconditions. Understanding the influentialproperties of the feed (chemistry, fineness,uniformity, and mineralogy) can lead toimprovements in its burnability and there-fore in the efficiency of plant operations.

When the raw materials available makeit hard to achieve burnability goals, theuse of fluxes and/or mineralisers may behelpful. Fluxes and mineralisers indirectlyaffect burnability promoting earlier clinkerphase formation; their effect on clinkerformation and cement quality is brieflydescribed here. Fluoride-containing com-pounds have proven to be the most effec-tive mineralisers in cement clinkering reac-tions. However, an excessive amount offluoride may delay cement setting time.

Transforming raw materialsPlant management is constantly underpressure to maintain or increase productionrates and reduce costs, without sacrificingquality. These three are challenging goals,especially achieving all three simultane-ously. Understanding the process of clinkermanufacture – how raw materials are trans-formed into clinker – is the key.

This transformation involves bothchemical and physical processes. Themicrostructure of clinker is a function ofkiln feed properties and the pyroprocessingconditions. Knowing what happens where,why, and how, can lead to optimising pro-duction, costs, and quality. In turn, theclinker microstructure and compositiondirectly affect how the phases hydrate andcontribute to cement performance in freshand hardened concrete.

Given its relationship to both plantoperation efficiency and cement perfor-mance, the importance of clinker formationcannot be overestimated.

ClinkerStated simply, the chemical aspect ofclinker formation is the combination of silica with calcium to produce hydrauliccompounds. However, the actual transfor-mation of the raw materials into clinkerinvolves a combination of chemical andphysical processes as the material passesthrough the kiln system. The generalprocess of clinker formation is described inFigure 1. The transformation concludeswith the primary clinker phases: • alite: impure tricalcium silicate, gener-ally termed C3S• belite: impure dicalcium silicate, nor-mally termed C2S• aluminate: tricalcium aluminate, C3A• ferrite: nominally tetracalcium alumino-ferrite, C4AF

Improving pyroprocessingefficiencyThe amount of energy needed to formclinker depends on various properties ofthe raw materials. Using a burnabilityequation such as that shown below canidentify the specific factors that influenceburnability at a particular plant, and

provide the information needed to makeappropriate improvements. This approachcan also be used to anticipate changescaused by the introduction of new rawmaterials. Replacing raw materials withothers because they are less expensive, orto correct the chemical composition maychange how difficult it is to burn the newkiln feed.

Raw material properties and burnabilityThe raw materials used are designed andproportioned to provide the appropriateamounts of the various clinker phases.Specific properties of the materials arecritical to plant efficiency and energy con-sumption, based on their burnability.

Burnability is the ease with which theraw materials are transformed into thedesired clinker phases, and is commonlymeasured by the amount of free (un-reacted) lime remaining in the clinker (lowfree lime indicates an easy-to-burn mix).Ordinarily, the burnability is a measure ofthe ease of formation of alite from beliteand free lime, as formation of the otherphases is normally much easier to achieve.The lower the temperature at which thetargeted free lime can be obtained, thebetter is the burnability of the kiln feed.

The burnability of a raw mix is deter-mined by its chemical composition, themineralogy of its component materials, and

79INTERNATIONAL CEMENT REVIEW / SEPTEMBER 2002

RAW MATERIALS

Burningthe mix

This paper, this year’s prizewinning presentation at the IEEE/PCA Meetingin Florida, discusses clinker formation, its relation to kiln feed properties,and the importance of optimising burnability by careful mix control, goodmix homogeneity, and tailoring the burning process to the raw mix. Theburnability and kiln feed are discussed in relation to specific fuel consumption. Emphasis is placed on the effects of changes in raw materialburnability and clinker formation on kiln operations, finish mill productiv-ity, and the properties of the resulting cement.

by Linda M Hills, Senior Scientist; Vagn Johansen, Senior PrincipalScientist; and F MacGregor Miller, Senior Principal ScientistConstruction Technology Laboratories, Inc

raw mat 79 80/82 84 8/30/02 10:18 AM Page 1 paul Pauls Hard Drive:Desktop Folder:ICR Sept 2002:ORIGINALS:Raw Materials 79 8

INTERNATIONAL CEMENT REVIEW / SEPTEMBER 200280

RAW MATERIALS

its fineness. There are a number of equa-tions that relate the burnability of a feedto the composition and fineness of theminerals (Fundal 1979, Petersen andJohansen 1979, Christensen 1979, Ludwig1973). One example is the burnabilityequation developed at FLSmidth (Theisen1992), provided below. The importance ofa burnability equation is not to provide anexact value of the free lime of clinker madefrom a given raw mixture; it is more impor-tant that it gives insight into the effect ofraw material properties on the resulting

free lime (clinker burnability) on a relativebasis.

The first part of the equation (firstbracket) represents contribution from thechemical properties of the raw mix. The LSF(lime saturation factor) represents the CaOof the mix, while the SR (silica ratio ormodulus) is related to the amount of liquidphase at the burning zone temperature.Decreasing the SR is equivalent to increas-ing the amount of liquid phase; since theliquid is the transport medium for the reac-tants, more liquid can transport more reac-

tants during a given time and therefore,the burnability is improved. Note the rela-tive magnitude of the coefficients for LSFand SR; the role of the liquid for theclinker reactions is important, not only forthe chemical reactions but also for the for-mation of clinker nodules (the physicalaspect of clinker formation).

The second part of the burnabilityequation (second bracket) represents thecontribution to the burnability from themineralogy and fineness of the raw mix.The percentages of coarse particles repre-sented by C125, Q45, and R45 are determinedby the overall fineness of the raw mix onthe one hand, and the mineralogical prop-erties of the raw materials on the other.Differences in the mineralogy of the mixcomponents leads to differences in grind-ability, which in turn will result in varia-tions in the chemical composition of thedifferent size fractions of the raw mix. Forexample, since quartz is hard to grind, theSiO2 content of the coarser fraction of themix will normally be higher than that ofthe finer fraction. Note that the quartzfineness is more significant than finenessof other materials, as indicated by itslarger coefficient in the equation.

The equation provides a ‘virtual burn-ability test’ to analyse the factors affectingburnability and free lime of the clinker.There is a relation between the free limedetermined by the ‘virtual burnability test’and the free lime of the clinker. Althoughnot identical, it can be assumed that thevirtual burnability free lime will correspondto a certain constant clinker free lime for agiven constant kiln operation. Further, fora given kiln and set of raw materials, if thevirtual burnability free lime of the kilnfeed changes and the kiln process parame-ters remain the same, the clinker free limewill change in a way so the ratio betweenthe virtual free lime value and the clinkerfree lime remains the same as before thechange. For instance if the virtual free limeis 3.2 per cent and the average clinker freelime is 1.4 per cent for a given steadystate of kiln operation, the ratio is 2.3.This defines a burning index and is a mea-sure for how the kiln in that state of oper-ation on average burns the kiln feed.Reasons for variations in burnability couldbe insufficient raw mix control and/orhomogenisation of the kiln feed. Theresulting clinker will vary in free lime con-tent, and the normal reaction from kilnoperators is to burn harder and eventuallyoperate the kiln on the ‘hot side’ to avoid

To 700˚C• Water is lost from clay minerals• Dehydrated clay recrystallises• Some reactive silica may displaceCO2 from CaCO3

700-900˚CAs calcination continues, free limeincreases. Calcination maintains feedtemperature at around 850˚C. Lower-limed aluminate and ferrite form.

Figure 1: a simplified view of the clinkering reactions in a Portland cement kiln (Hills 2000)

900-1150˚CReactive silica combines with CaO tobegin stages of C2S formation.

1150-1200˚C• When calcination is complete, tem-perature increases rapidly.• Small belite crystals form fromcombination of silicates and CaO.

1200-1350˚Cv Above 1250˚C, liquid phase isformed.• Belite and free CaO form alite inthe liquid.

1350-1450˚C• Belite crystals decrease in number,increase in size. • Alite crystals increase in size andnumber.

Cooling• Upon cooling, the C3A and C4AFcrystallise from the liquid phase. • Lamellar structure appears inbelite crystals.

Angular alitecrystals

Round belite crystals

raw mat 79 80/82 84 8/30/02 10:18 AM Page 2 paul Pauls Hard Drive:Desktop Folder:ICR Sept 2002:ORIGINALS:Raw Materials 79 8

82 INTERNATIONAL CEMENT REVIEW / SEPTEMBER 2002

RAW MATERIALS

large variations in the clinker free lime.One penalty that results is increased spe-cific fuel consumption.

An example of the relation between theburnability index and the specific fuel con-sumption is the result of analysis of datafrom a 2000tpd kiln with planetary coolers,as shown in Figure 5. Based on averages ofkiln feed chemical composition and fine-ness, the virtual burnability free lime wasdetermined and together with the averageclinker free lime for the same period, theburning index ratio calculated.

An increasing burnability index meansharder burning, resulting in higher specificfuel consumption. Figure 5 can be used toanalyse the effect of kiln feed homogeneityon the specific fuel consumption. As anexample, consider a kiln with average feedcomposition corresponding to LSF = 98 percent and silica ratio, SR = 2.8 (correspond-ing to about 66 per cent C3S), with virtualburnability free lime of 3.2 per cent and anaverage clinker free lime 1.4per cent. For a steady statekiln operation and a con-stant homogeneous kilnfeed, the burning index is3.2/1.4=2.3. In reality thekiln feed composition willvary between certain limits.In the first example, assumethe variations in LSF and SR

are +/-0.6 per cent and +/-0.07 respec-tively. Using the equation for the virtualburnability test given above, these varia-tions correspond to a variation of the vir-tual burnability test free lime between 3.8-2.6 per cent. For the steady state kilnoperation referred to above, with burningindex = 2.3, this will result in clinker freelime between 1.7-1.1 per cent. In the sec-ond example let the kiln feed compositionvary +/-2 per cent in LSF and +/-0.23 inSR. This corresponds to variations in thevirtual burnability free lime between 4.5and 1.9 per cent, and for constant kilnoperation a variation of the clinker freelime between 2-0.8 per cent. As a result ofthe less homogeneous kiln feed in the lastexample, the clinker free lime will varywithin wider limits. If the operation ischanged to harder burning in order tobring the maximum clinker free lime down

to the average value and narrow the varia-tion in free lime, the burning index willchange from the original 2.3 to4.5/1.4=3.2. From Figure 5 such a changecorresponds to an increase of about six percent in specific fuel consumption.

Once again, an understanding ofburnability and free lime content is impor-tant. There is a relation between the ‘vir-tual burnability’ free lime and the free limeof the clinker from the kiln. This burningindex ratio can be used as a process para-meter. If the kiln is operated in steadystate, the ratio is constant. However, if the‘virtual burnability’ deteriorates (iedecrease in mix fineness or change inchemistry), and the kiln is kept in thesame steady state, then the clinker freelime will increase. The usual operator reac-tion to an increase in clinker free lime isto burn harder. However, if an examinationof the kiln feed were performed to identifythe cause for the increase in ‘virtual burn-ability’ free lime, the correspondingincrease in clinker free lime would beanticipated. The kiln operators wouldtherefore know what to expect, and wouldnot necessarily constantly operate the kilnon the hot side. The examples given abovealso emphasize the importance of good rawmix control and kiln feed homogenisationvis-à-vis the specific fuel consumption.

Most often, the mineralogy and fine-ness change little compared to the chemi-cal composition unless the raw materialsources are changed. Therefore, using thechemical contribution portion of the burn-ability equation is helpful. This calculationwill give an indication of feed burnabilityvery quickly. An example of chemical analy-sis and calculation of its contribution toburnability is below. Note that the trend inburnability is more easily detected whenlooking at the calculated chemical contri-bution instead of the individual factors ofLSF and SR alone.

Fluxes and mineralisersThe use of fluxes and mineralisers can pro-mote clinker phase formation, and there-

fore improve efficiency of thepyroprocessing system. Bydefinition, a flux promotes areaction by increasing theamount of liquid at a giventemperature; for example,Al2O3 and Fe2O3 are fluxes forthe formation of alite. Amineraliser promotes the for-mation of a particular solid

Figure 2: clinker silicate phases as observedunder a microscope (polished section withnital etch)

Alite (C3S) normally hexagonal crystalsobserved in cross-section, 25-50mm in length.

Belite (C2S) normally 25-40mm roundedcrystals with multidirectional lamellae.

Figure 3: phases in clinker interstitial (polished section with nital etch)

Tricalcium aluminate (C3A): observed inetched and polished clinker as blue to grayangular crystals in the interstitial. Highalumina ratio will produce greater amountof aluminate than ferrite.

Ferrite (C4AF) appears more reflective tanC3A in an etched polished surface

CaO1400˚C =[0.343(LSF-93) + 2.74(SR-2.3)] + [0.83Q45 + 0.10C125 +0.39R45]Where:CaO1400˚C = is the free lime after burning for 30 minutes at 1400˚CLSF = %CaO/(2.8% SiO2 + 1.18% Al2O3 + 0.65% Fe2O3)SR = % SiO2/(% Al2O3 + % Fe2O3)Q45 = % quartz grains coarser than 45µmC125 = % calcite grains coarser than 125µmR45 = % other acid-insoluble minerals, (eg feldspar) coarser than 45µm

Figure 4: clinker photomicrograph demon-strates reaction between belite and free CaOto form alite (polished section with nitaletch)

Either properties of the raw materials, mixhomogeneity, or the pyroprocessing condi-tions did not allow the complete transfor-mation to alite to take place.

Belite crystals

Free CaO crystals

raw mat 79 80/82 84 8/30/02 10:18 AM Page 3 paul Pauls Hard Drive:Desktop Folder:ICR Sept 2002:ORIGINALS:Raw Materials 79 8

INTERNATIONAL CEMENT REVIEW / SEPTEMBER 2002 83

RAW MATERIALS

phase through its incorporation in one ormore of the solid phases (normally eitheralite or belite). A general view demonstrat-ing the process of clinker formation usinga flux or mineraliser, contrasted with usingneither is outlined in Figure 6.

Fluxes and mineralisers have beenstudied in laboratory experiments and inpractice (Flint, 1939, Johansen andChristensen, 1979, Surana and Joshi 1990,Erhard, 1994, Altun, 1999). A series of dif-ferent compounds have been used; how-ever, the ones with the most practicalapplication are fluorides.

Fluoride-containing mineralisers havebeen used since the late 1800s (Bhatty,1996). Fluoride-containing compoundssuch as CaF2, NaF, BaF2, and MgF2, are alleffective primarily as mineralisers,although CaF2 has enjoyed the greatestuse. Alkali- and alkaline-earth fluorosili-cate salts like Na2SiF6 and MgSiF6 can alsoachieve similar mineralising effects (Lea,1971). The effect of 5-10 per cent

fluorspar on the clinkering of two cementcompositions was reported (Klemm, 1976)and as expected, these massive additionsresulted in both a reduction in clinkeringtemperature and a consequent retardationof cement setting. However, there was lit-tle effect on the 28-day strength ofhydrated cements.

Although fluoride-containing com-pounds have been proven tobe the most effective miner-aliser in cement clinkeringreactions, a number of otherpossible fluxes and miner-alisers have been studied.These have included gypsumand sulpate mineral tailings,fluoro-gypsum, phosphorus-containing calcareous tail-

ings in combination with gypsum/fluorsparand others; all showed some effect eitheras fluxes or mineralisers but the studiesdid not result in practical use.

The benefits of added fluoride inachieving desired free lime levels at lowertemperatures are illustrated by a study inwhich the same rather hard-burning rawmix was burned with and without the addi-tion of 0.25 per cent fluoride. In this par-ticular case, the fluoride was added as a

fluoride salt. The mixes wereburned at 1450˚C (2640˚F) for 60minutes. The resulting free limecontent of the control clinkerwas 5.05 per cent, while that ofthe clinker to which fluoride wasadded was 2.31 per cent. A fur-ther demonstration of the bene-fits to burnability of added fluo-ride was carried out using theso-called ‘sandwich’ technique.In this technique, two compactsare prepared. The first is calcu-lated to be an alite/free limemixture, while the second is cal-culated to be an alite/belite

mixture after clinkering. The compacts hadthe compositions shown in the tablebelow.

Pairs of compacts were made, with andwithout added fluoride. The individualcompacts were burned at 1450˚C for 30minutes, cooled rapidly to room tempera-ture, and the free lime levels determined.The low lime compacts had negligible freelime levels (<0.12 per cent), and the high

lime compacts had free lime levels around13 per cent. The high lime and low limecompact cylinders were placed in intimatemutual contact, and the cylinders re-burned at 1500˚C for one hour. Figure 7demonstrates this ‘sandwich’ technique.

The resulting clinkers were cut, trans-verse to the joint, etched with Nitaletchant, and examined under reflected lightmicroscopy. The results, shown in Figures 8and 9, illustrate the broadening of theregion near the joint where no belite or freelime exists – ie, where alite formation hasoccurred between the high and low limecompacts. The much broader width in thefluoride-doped clinker (about twice as wide)shows how much faster alite forms in thepresence of 0.25 per cent fluoride.

With regard to fluxes and mineralisersin general, Bhatty (1996) reports a surveyregarding their use in the cement industry.The fluxes and mineralisers referred to werefluoride-based (primarily fluorspar), andother non-fluoride-based materials. Thelatter were industrial by-products or waste-derived materials primarily used as correc-tive materials in the mix. They containediron oxide and alumina and resulted indecreasing of the silica ratio of the rawmix. This silica ratio reduction is equiva-lent to an increase of the clinker melt con-tent, and hence to an improvement of theburnability. The study indicates thatcement plants outside North America usefluxes and mineralisers with better resultsthan do the North American plants. Inmost cases improvements in burnability,kiln operation, energy savings, and cementquality were noted. With the fluoride min-eraliser, setting time was reported toincrease in some cases. Christensen (1980)suggested a mechanism for the effect ofmineralisers that implies increased settingtime, which is in line with observationswith using fluoride compounds.

One must be aware of side-effects ofusing some materials as fluxes or miner-alisers. In North America, the most com-monly reported problem when using fluo-ride-based fluxes/mineralisers is the

blockage of preheaters causedby fines and volatiles. To pre-vent these blockages, it isrecommended to tightly con-trol kiln temperature and thethermal profile to preventoverheating of the load andgeneration of additionalvolatiles and fines. In otherwords, the burning zone tem-

LSF SR Chemical contribution[0.343(LSF-93) + 2.74(SR-2.3)]

95.2 2.2 0.4895.6 2.3 0.8994.9 2.8 2.0295.9 3.1 3.1995.8 3.2 3.43

Alite AliteBelite Free lime

<0.12% 13%free lime free lime

Figure 5: Burning Index Ratio (virtual burn-ability free lime/clinker free lime). Each pointrepresents the average over a 4-week period.No corrections were made for kiln downtimeduring the 4-week periods, which mightexplain the outliers

Reagent High lime compact, % Low lime compact, %Calcium carbonate 78.9 74.3Quartz (SiO2) 10.73 15.33Alumina (Al2O3) 3.86 3.86Ferric oxide (Fe2O3) 1.93 1.93Magnesium hydroxide 3.05 3.05Sodium carbonate (Na2CO3) 0.51 0.51Potassium carbonate (K2CO3) 1.02 1.02

Alite AliteAlite

Belite Free lime

Figure 7: diagrams demonstrate the clinker ‘sandwich’ before(left) and after (right) re-burning

raw mat 79 80/82 84 8/30/02 10:18 AM Page 4 paul Pauls Hard Drive:Desktop Folder:ICR Sept 2002:ORIGINALS:Raw Materials 79 8

84 INTERNATIONAL CEMENT REVIEW / SEPTEMBER 2002

RAW MATERIALS

perature reduction that has nowbecome possible because of theuse of mineralisers must beimplemented.

Fluoride can also be aretarder in cement hydration ifpresent at too high a level. Inthe example provided below, aparticular plant produced twoType I cements with differingfluoride levels; the fluoride ratiovaried by a factor of four basedon the raw materials. The high-fluoride cement had an initialsetting time of 2.4 times longerthan the low-fluoride cement.The conduction calorimeter,which measures the hydrationprofile of cement pastes at con-stant temperature, gave a verygraphic picture of the differencein hydration of the two cements,as shown in Figure 10.

The graphs demonstrate thedelay in the C3S peak for thecement with high fluoride con-tent. The maximum hydrationrate is delayed from about ninehours to about 17 hours. It appears fromthese results that the amount of fluoridepresent in the high-fluoride cement wastoo high for acceptable cement settingbehaviour. The behaviour of cements withoptimum fluoride addition may be consid-ered somewhat analogous to that ofcements hydrated at low temperatures, orto cements hydrated in the presence ofretarding admixtures. Although the earlyhydration is delayed, the ultimate hydra-tion may indeed lead to better latestrengths (Moir, 1982). It is not unusualfor cements experiencing slight retardationin setting time to develop 28-daystrengths higher than those of their faster-setting counterparts.

Operations and performanceChanges in burnability and/or clinker for-mation can have important practicalimpact on what happens in the kiln, themill, and the product. If the mix is hard toburn, the operator will be obliged to

increase the burning zone temperature toachieve the desired free lime level. Hardburning will tend to cause low clinkerporosity, large crystals of alite, and oftencontributes to generation of dust and/orlarge clinker balls, instead of good, nodularclinker. It also slows down the resultingcooling process, both because the maxi-mum temperature is higher, and becausethe low porosity clinker is more difficult tocool. This can often result in generation ofcement with reduced strength potentialand increased water demand. Clinker tem-peratures exiting the cooler may increase,

further increasing fuel consumptionand presenting handling problems.The high temperature conditionsmay lead to reductions in clinkeralkali and sulphate level, colourvariations, and increases in waterdemand attributable to increasedlevels of aluminate. The reducedclinker alkali and sulphate willaffect setting time, decrease earlystrength and increase later-agestrength; and fresh concrete mightdevelop admixture incompatibilityand changes in its rheologicalbehaviour.

ConclusionSpecific properties of cement rawmaterials are critical to plant effi-ciency and energy consumption,based on burnability. This involvesthe chemical properties of the rawmix, along with the mineralogy andfineness of the materials.Determining a ‘virtual burnability’using these parameters identifiesthe critical areas where changescould be made to improve burnabil-

ity and lower fuel consumption. When the quarry and the raw materi-

als available make it hard to achieve burn-ability goals, the judicious use of fluxesand/or mineralisers may be a prudent mea-sure. Fluxes and mineralisers promoteclinker phase formation to occur earlier,thereby improving the efficiency of thepyroprocessing system. Fluoride-containingcompounds appear the most effective min-eraliser in clinkering reactions. An experi-mental study demonstrated quicker aliteformation in the presence of 0.25 per centfluoride.

Changes in raw materials and/or clinkerformation can influence what happensthrough the kiln, the mill, all the waythrough to the cement performance.

This paper demonstrates the impor-tance of optimising burnability and clinkerformation by careful mix control, good mixhomogeneity, and tailoring the burningprocess to the raw mix. Real energy sav-ings, potential production increases in kilnand mill, and product enhancement are therewards for following these principles.

_________________________❒

This paper has been edited and reprintedwith kind permission from Linda Hills —‘Solving raw material challenges’ © 2002IEEE, New York, NY.

Figure 8: clinker ‘sandwich’ after re-burning

Figure 9: clinker ‘sandwich’ with fluoride afterre-burning

Without mineraliser With mineralisers and fluxes, desired clinker or flux, clinker phases phases are formed earlier, and is discharged are still forming from kiln. The result: residence time inin kiln kiln is reduced, and less energy is required

Mineralisers promote earlier silicate Increased liquid phasephase formation (some enhance liquid allows more C3S formation, some form intermediate phases) (alite) to form

Figure 6: simplified clinkering reactions in a Portland cement kiln withno additions, with mineraliser, and with flux

Clay particle Limestone particle

Normal clinkering Clinkering with mineraliser Clinkering with flux

Liquid phase formation occurs earlier (at lower temperature)

Figure 10: calorimeter results of high and lowfluoride cements

raw mat 79 80/82 84 8/30/02 10:18 AM Page 5 paul Pauls Hard Drive:Desktop Folder:ICR Sept 2002:ORIGINALS:Raw Materials 79 8