<<

THE AMERICAN MINERALOGIST, VOL. 47, NOVEMBER-DECEMBER, 1962

PHASE TRANSFORMATIONS IN SILICA-ALUMINA- MAGNESIA MIXTURES AS EXAMINED BY CONTINUOUS X-RAY DIFFRACTION: I. - COMPOSITIONS R. B. Gnen,lF. M. Wnnr, aNp R. E. Gnru, Uniaersityof Illinois, (Jrbana, I llinois.

Ansmacr

The phase transformations of natural and synthetic oxide mixtures of talc-kaolinite compositlons were examined by continuous r-ray difiraction while heated to 1450' C. Dif- fraciion intensity data show the enhanced development of protoenstatite within the sheet structure of talc, thus indicating that the growth of this on firing is largely con- trolled by the structure of the talc. The formation temperatule of , which was used as a measure of the reactivity of various starting materials, indicates that alumina in the form of , rather than as corundum, is most efiective in lowering the develop- ment of this mineral. Additional cordierite formation, through recrystalliza- tion on cooling indicates the presence of an intermediate licluid phase in the formation of this mineral.

INrnotucrlow Generalsta,tement . Continuous high-temperature o-ray diffraction methods make it pos- sible to observehigh-temperature reactions as they occur (Wahl et aI., 1961). Previously, high-temperatrte n-ray diffraction has been used mainly for selected observations at some desired temperature and not for continuousobservations of a sampleas it is heated.The techniqueof continuous scanning was used in the present investigation for the study of high-temperaturereactions in talc, kaolinite, talc:kaolinite mixtures, and mixtures of oxides with compositions corresponding to talc and kaolinite.

Infl,uence of starting materials Reactionrates between are known to be influencedby the of the starting materials. Significantly different results have been ob- tained by using different starting materials which have the same bulk composition.Thus, Fyfe (i960) has discussedthe reactivity of oxidesas opposedto glasses,and Roy (1956)has reported on the reactivity of co- precipitatedgels. Reaction rates in the systemsSiOz and AhOrSiOz with different starting materialshave beeninvestigated by wahl et al. (l96la, 1961b)by meansof continuoushigh-temperature r-ray difiraction.

I Present address: General Dynamics Corporation, Fort Worth, Texas. t273 1274 R. B. GRAF, F, M. WAHL AND R, E. GRITI

Co ntinwo us hi gh- tem p eratur e x-r ay d,if r action In the technique,as usedin the presentstudy, the samplewas packed on a small platinum holder which was then placed in a small platinum- wound furnace.This furnacewas mounted on a -cooledbase which, in turn, was mounted on a Philips recording r-ray diffractometer.The desired20 values were repeatedly scanned while the samplewas heatedat a constant rate. The intensities of the various diagnostic diffraction peakswere plotted againstthe temperaturesat which they occurred.The curvesobtained in this manner depict changesin the amounts (not abso- lute quantities)of crystallinematerials as a function of temperature.Be- causethe samesample was usedthroughout the run, differencesin pack- ing or preferred orientation were minimized and smaller changesin difiraction intensity could be detected. Purpose of the inaestigation A primary objectiveof the investigationwas to determinethe reaction rates of various starting materialswhich had the samebulk composition within the nIgO-AlrOe-SiOzsystem. Cordierite2 was formed from most of thesemixtures, and the formation temperatureof cordieritewas usedas a measureof the reactivity of the starting materials.The transformationsof kaolinite to mullite and and of talc to protoenstatitesand cristobalite,were brieffy investigated. Expnnrunwtar, PnocpounB Furnace and sample holder The furnaceused in this investigationwas similar to that describedby Grim and Kulbicki (1957). In the construction of the present furnace, however,the resistancewinding (PL-137a Rh) was coiledbefore it was wrappedon the alundum core. Five feet of .012-inchwire wereused and coiled in such a manner that four evenly spacedwraps could be made on the alundum centerpiece. The baseplate was groovedso as to insure reproducible aiignment of the platinum sample holders, which were 1X8X 14 mm with a depression0.5 X6X 10 mm. T emperature control and,measur etnent The power suppliedto the furnace was controlledby two variacs,and a constant heating rate of 5.6' C. per minute was obtained. When the temperature of the furnace had reachedthe desiredmaximum (usually '?All of the cordierite which was synthesized in this study was apparently indialite, as discussedby Sorretl (1960) 3 The enstatite in this investigation is apparently the protoenstatite for which a com- plete structure analysis was made by Smith (1959). PHASE TRANSFORMATION 1275

about 1475oC.), the power was turned off and the furnacewas allowedto cool.Because the furnacehas a small heat capacity,a rapid and uniform coolingrate was obtained. Temperatureswere measuredby using Pt-Pt l07o Rh thermocoupleswith a cold junction and a Minneapolis-Honey- well Model 2732 potentiometer. The accuracy of the thermocouples was checked several times during the investigation by using the inversion of BaSOaat 1149" C. The agreementwas usually within about 5o C. and never greaterthan 10" C. At least two runs were made for each sample.In the first run, 20 values oI 2" to 50' (the maximum range with this furnace) were repeatedly scannedas the sampleswere heated. In the secondrun, only the diag- nostic difiraction peaks for each phase were measured, usually at tem- perature intervals of about 25" C. AII of the data were obtained using nickel-filteredcopper radiation generatedat 45 kv and 18 ma.

Starting materials The following materials were used as reactants: MgO, light powder, Allied Dye and Chemical Corporationl "silicic acid," Mallinckrodt Chemical Works; talc, Baker Chemical Company; kaolin, Mclntyre, ; and synthetic , AI(OH)3, Merck and Company. The kaolin was shown by x-ray diffraction to be substantially pure well- crystallized kaolinite. A small amount of was present and analysesof similar kaolinsindicate that oneor two per cent K2O may be present.The "silicic acid" and synthetic gibbsite were of reagent grade. When the MgO was fired in a continuous run, it was at first characterizedby very broad difiraction peaks which correspondedto those of periclase.These diffraction peaksbecame sharper as the firing was continued. The mixtures were made according to mole per cent calculations in ratiosof talc to kaoliniteas followsl 4:1,3:2, l:1,2:3, and 1:4. Oneof group of mixes was composedof natural talc and kaolinite in those pro- portions. Another group was composedof kaolinite and MgO-"silicic acid" which were mixed in proportions corresponding to talc. A third group consistedof talc and syntheticgibbsite-"silicic acid" corresponding to the composition of kaolinite. The mixtures of oxides correspondingto the natural mineralswill be referredto as TC and KC, meaningrespec- tively talc compositionand kaolin composition.Thus, 4K: TC refersto a mixture in which MgO and "silicic acid" corresponding to the com- position of talc are mixed with four times as much naturally occurring kaolinite (mole per cent). In this investigation,talc and kaolinite were comparedto the mixtures of oxidesas if they were anhydrous.However, the water content was consideredin the weighing and mixing of the, samples. 1276 R. B. GRAF, F. M. WAHL AND R. E. GRIIT

I I

F a z lal F z CrisL(lll)

P-En.(l2l)

800 900 rooo lloo 1200 1500 14000c

Frc. 1 lntensity of difiraction by high-temperature phases in heated talc.

The materials which were used in the various mixes were dried, weighed and mixed in the desired proportions, mixed further as a paste with distilled water, dried at 100' C. for about 8 hours, and then re- ground. An attempt was made to pack the sameamount into the sample holder each time. ExpBnnrBNrAL RESULTS When talc wasfired in a continuousrun (Fig. 1), it beganto decompose at about 830" C. and protoenstatitewas formed immediatqly. When the temperature reached approximately 1220o C., cristobalitea also was formed. The most prominent feature of the formation of protoenstatite was the enhancedintensity of the (310) as compared to the (l2t) rc- flection.Such intensity relationshipshave beenused by Kulbicki (1959) to explain the structural relationships of the transformations of , attapulgite, and to enstatite. He consideredthe formation of enstatite to be favored by the fibrous nature of sepiolite and attapulgite, and stated that enstatite would form with less intensity and rapidity from a mineral with a layer-type structure.In the presentinvestigation, however, protoenstatiteformed rapidly and in considerableabundance from talc. Using the (310) and (121) reflectionsit is possibleto compare the intensitiesof reflectionsfrom planesparallel to the c axis (310) with those from planes which intersect the c axis (l2l). The intensity differ-

a The cristobalite formed at elevated is beta cristobalite in every case. PHASE TMNSFORMATION 1277 encesindicate that protoenstatite is preferentially grown with the c axis in the plane of the silica sheets of the original talc. The sheets must therefore be partitioned into chains to afford the protoenstatite nuclea- tions, with consequent expulsion of one-fourth of the silica, which later forms cristobalite. Therefore, it is considered that the protoenstatite has grown perferentially within the talc and its development is in- fluencedby the talc structure. Protoenstatite was also synthesized from a mixture of MgO-"silicic acid" with a MgO: SiOzratio equivalent to talc (Fig. 2), the mineral first appearingat 750oC. The (310) of the newly formed protoenstatiteagain

F U' z t! F z

800 900 looo lloo 1200 1300 1400"c

Frc. 2. Intensity of difiraction by high-temperature phases in a heated talc composition mixture prepared from MgO-silicic acid.

had a higher initial intensity than the (121),but the differencewas much smaller than in talc. At 1275" C. the intensities were equal, and at 1400'C. the (121) was considerablymore intensethan the (310), which is the proper relationship when no preferential growth has occurred. Protoenstatite which was formed in the firing of MgO-"silicic acid" : kao- Iinite mixtures had the same relative difiraction intensities that were ob- servedwhen the MgO-"silicic acid" was fired separately(Fig.2).In the talc:kaolinite mixtures (Fig. 3), protoenstatiteformed at the sametem- perature and had the same intensity relationships that were recorded when the talc was fired separately. When kaolinite was fired in a continuous run, mullite was detected first at 1180'C. and cristobalite at 1350oC. (Fig. a). The difiraction 1278 R. B. GRAF, F. M. WAHL AND R. E. GRIM

1 I

F (^ z lrl F 1

800 900 tooo ltoo t200 t300 t400"c

Frc. 3. Intensity of difiraction by high-temperature phasesin a heated mixture of talc-kaolinite in the ratio of 1:1. intensitiesof the (120) and the (210) peaks of mulrite had a relatively uniform increasefrom 1180oc. to about 1300oc., at which temperature the intensities remained relatively constant as the temperature was further increasedup to about 1435' c. The cristobalitedifiraction inten- sity also increasedin amount up to a temperature of about 1435. C. Fusion beganat highertemperatures and the diffractionintensity of both decreased.

F U' z IJ Fz

looo iloo t200 t300 t400 t5000c

Frc. 4. Intensity of diffraction by high-temperature phases in heated kaolinite. PHASE TRANSFORMATION r279

When talc was mixed with kaolinite and fired, protoenstatite,mullite, and cristobalitewere formed. Thesephases then reactedtogether to form cordierite.Figure 3 illustratesthe generalpattern of the reactionsfor all of the talc:kaolinite mixes studied. In the talc: kaolinite mixtures, cordieritewas formed at about 1195'C. regardlessof the talc:kaolinite ratio. However, the temperatureat which cordieritebegins to fuse does depend to some extent upon the talc:kaolinite ratio. The fusion tem- perature was found to be relatively higher in those mixes containing more kaolinite than talc. Thus, the cordieritein the 4T: K mix beganto fuseat about 1295" C., whereasin the T:4K mix the diffraction intensity of cordieritebegan to decreaseat 13500C. Further, cristobaliteis formed at temperatures50 to 100 degreeslower in the mixtures than when either talc or kaolin are fired separately.On the other hand mullite is formed at a higher temperature (usually about 50o higher) in mixtures than when kaoliniteis firedalone. In all mixtures the (111) of cristobalite,the (110) of mullite, and the (310) of protoenstatitedecrease in intensity as cordieriteis formed. The mixtures always contain more silicathan doescordierite. The amount of cordierite that can form is dependentupon the abundanceof the MgO and AlzOsphases which are presentin the mix and on the relative reactiv- ity of thesecomponents. In the mixture with 4 talc:.1kaolinite the muliite is used up and its diffraction lines disappear at 1290"C., whereasin the 1 talc:4 kaolinite mix, the mullite persiststo 1420"C. In both casesthe cordieriteintensity increasesas the mullite intensity decreasesup to the temperature of the disappearanceof the mullite. Similarly, protoensta- tite disappearsat 1315oC. in l talc:4 kaolinite mixture and is still pres- ent up to the fusion point in 4 talc:l kaolinite mixture with comparable variation in Lhedevelopment of cordierite. When kaolinite is mixed with MgO-"silicic acid" (Fig. 5) in propor- tions correspondingto talc, cordierite is formed at temperaturesabout 25oC. lower than in the natural talc:kaolinite mixeswith the correspond- ing compositions.Mullite is also formed at lower temperatures,below 1000"C. in the talc composition:kaolinitemixtures with high proportions of kaolinite. Protoenstatite is formed at temperatures ranging from 810' C. to 840o C., which is a higher temperature than for MgO and "silicic acid" alone. The diffraction intensity of the (310) reflection of protoenstatitein the TC:K mixes is much less than in the talc:kaolin mixes becauseof the previously mentionedlack of preferentialgrowth. The disappearanceof the protoenstatiteand mullite with the develop- ment of cordieritefollow the samereaction sequence as in the talc:kaolin mrxtures. The resultsobtained by firing the talc:synthetic gibbsite-"silicicacid" mixtures (Fig. 6) difier greatly from the talc:kaolin and the talc com- R. B. GRAF, F. M. WAHL AND R. E. GRIM

tI

F U) z tu - z

800 900 rooo iloo 1200 1300 14000c

Frc. 5. Intensity of diffraction by high-temperature phases in a heated mixture of kaolin : talc composition in the ratio of 1 : 1. position:kaolin mixes, becausecorundum is formed from the AI(OH)3, and it is much lessreactive than mullite. Not only is a higher tempera- lure necessaryto form cordierite,but smaller amounts are formed. The cordierite begins to fuse at higher temperaturesin those mixes which have

70

t 60 I I 50 F

20 Cor.(lOl) to Mu.(llO)\ tooo iloo 1200 1300 1400'c

Frc. 6. Intensity of diffraction by high-temperature phaseq in a heated mixture of talc:kaolin comoosition in the ratio of 1:1. PHASE TRANSFORMATION I28I

a higher AlzOacontent. No cordierite was formed in the 4T:KC mix, while in the other mixesit forms at 1300' C. regardlessof the MgO:Al2O3 ratio. In the 4T:KC sample the diffraction intensities of corundum, protoenstatite, and cristobalite, begin to decreaseat approximately 1200oC., just as in the other mixes,but apparently more AlzO: must be present to form cordierite at the rate of firing used in the experiments. Mullite is formedin the 1:1,2:3, and 1:4 talc:kaolinite composition mix- tures in amounts which are proportional to the AlzOgcontent of the sam- ples.The formation of protoenstatitefrom talc is the sameas in talc: kao- linite mixes. The disappearanceof the protoenstatite also follows the same pattern as the reaction proceeds, but with greater variability in this group of mixes (i.e., protoenstatitepersists to a higher temperature in the mixes which contain greater amounts of this mineral). Also cristobaliteis usually formed at a higher temperaturein talc:synthetic gibbsite-"silicic acid" mixtures than in the other mixtures. The trans- formation to cristobalite takes place more readily in natural kaolinite than in synthetic mixes.

DrscussroNon RBsurrs The formation temperatures of cordierite within each group of mixes appear to be dependentlargely on the nature of the reactantsand inde- pendentof the MgO-AlzOsratio of the sample.Thus, whenevercorundum was a reactant, cordierite was formed at about 1300' C. When mullite was the sourceof alumina, cordieritewas first detectedat temperatures ranging from 1150oC. ts 1225"C. In certain of the samples,a glassy phase could be observed after the sampleswere cooled.The sampleswith the highest MgO contentsfused more easily and when cooled,they consistedof a colorlesstransparent bead of glass.The amounts of glassdecreased as the AlzOecontent of the sample increased. The fired samples which contained more AIrOa con- sistedof the glassand mullite, and in somecases, corundum. Some indication of the composition of the glassy material is given by the fact that when the sample was cooled and then r-rayed, the diffrac- tion intensity of cordierite was considerablyenhanced as compared to the intensity measuredat the higher temperature. When a 2 talc:3 kaolinite mixture was fired to the temperature at which the cordierite began to fuse (about 1400' C.) and then quickly cooled(from 1400' C. to 1000' C. in about 70 secondsand more slowly thereafter), the increasein the diffraction intensity of cordierite was about 60 per cent. When the sample was cooled quickly from a temperature of only 1300' C and then r-rayed, the increasein the diffraction intensity of cordierite was much smaller, usually about 10 per cent or less. This indicates that cordierite - 1282 R, B. GRAF, F, M, WAHL AND R. E. GRIM lized from the glassymaterial on cooling-to a greaterdegree if the cool- ing is slow. The mechanismwhich may be operative in the reactionsin this in- vestigation is that of differential melting. In this mechanismadjacent grains of enstatite (from the talc) and mullite (from the kaolinite) react with cristobalite (from both talc and kaolinite) to form a Iiquid which has a compositioncorresponding to that which has the lowest melting point in the phasediagram. It is possiblethat becausethis point is adjacent to the stability field of cordierite, cordierite is formed from the liquid. As the temperatureis increased,more material can react and the com- position of the liquid can changetoward that of the original bulk com- position of the sample. Therefore, the reaction rates which were meas- ured were probably the tendenciesof the materialsto react to form the initial liquid. This would explain why the nature of the reactants was apparently more important than variations in the MgO:AlrOa ratios of the samples. In a considerationof the talc:4 kaolinite mix, the intensitiesof mullite and protoenstatitebegan to decreaseat 1200"C. and cordieritebegan to form at 1215'C. As the temperaturewas increased,cristobalite began to decreasein intensityat about 1275'C.However, cordierite continued to in- creaseto 1350oC., i.e.,even af ter themullite, protoenstatiteand cristobalite beganto decrease.Because of the increasein temperature,the composi- tion of the liquid could changetoward that of the original bulk composi- tion of the mix. At about 1300' C. the enstatitewas usedup, at 1350oC. the cordierite began to decreasein intensity, and at 1415'C. mullite and a liquid remained.The protoenstatitewhich was reactedby 1310' C. in this mix was observedto persistto successivelyhigher temperaturesin the mixes with higher contentsof protoenstatite.Conversely, it was ob- served that in the mixes with more protoenstatite than mullite, the formation of a liquid phaseused up the mullite first and that enstatite was presentuntil fusion.

Suulranv The formation of protoenstatitefrom talc was characterizedby the en- hancedgrowth of the protoenstatitein the "c" direction indicating that the growth of the mineral was controlledby the structure of the talc. When mixturesof talc, kaolinite, and oxides(corresponding to the com- positions of talc and kaolinite) were fired, cordierite was formed at tem- peraturesranging from 1150oC. to 1300"C. The nature of the reactants was apparently more important in the reactionsthan the mole per cent variations in the compositionsof the mixes. Thus, when the alumina is present in corundum rather than mullite, cordierite forms at the higher PHASE TRANSFORMATION 1283 temperature. It is suggestedthat the crystalline materials reacted to form a liquid which had a compositioncorresponding to the lowest melt- ing point in the system.Such a liquid would have a compositionin or near the stability field of cordierite.When someof the sampleswere cooled and x-rayed,additional cordieritehad recrystallized.At the lower tempera- tures, when cordieritewas first beginningto form, only small amounts of cordierite were formed through recrystallizationupon cooling the sam- ples. The sampleswith the highest l\tlgOcontents could be fused to a clear glass bead. The fired mixtures containing greater amounts of AI:Oa consistedof glassymaterial, mullite, and in some cases,corundum. The cordieritebegan to fuseat highertemperatures in the mixeswhich had the higher contents of AlrOg. This indicates that the composition of the liquid was changingtoward the ALOTSiO2side of the diagram, an area in which the Iiquid and cordieritecan coexistat higher temperatures.

AcrlrowrptGMENTS This work forms part of a seriesof high-temperaturer-ray difiraction studiesmade possible through a researchgrant from the National Science Foundation.

RrnnnBNcrs Fvra, W. S. (1960) Hydrothermal synthesis ancl determination of equilibrium betr,veen minerals in the subJiquidus region. four. Geol.,68, 553-566. Gnru, R. E., lltr G. Kur.srcrr (1957) Etude aux rayons X cles reactions des mineraux argileux a haute temperature. Bull. Soc.Franc. Ceram.2l-27. Kultrcrr, G. (1959) High temperature phasesin sepiolite, attapulgite, and saponite. ,4m. Minerol. 44t 752-764. Rov, R (1956) Aids in hydrothermal experimentation: II, Methods of making mixtures for both "dry" and "wet" phase equilibrium stttdi.es.Jour. Am Cey.9oc.,39,145-146. Sonnelr., C. A. (1960) Reaction sequenceand structural changesin cordierite refractories. Jour Ant.. Ceram.Soc. 43,337 343. Surrn, J. V. (1959) The of protoenstatite, MgSiOs. Acta. Crysl.all.12, 515-519. Weul, F. M., R. E. Gnru.lNl R. B. Gnel (1961a) Phase transformations in silica as ex- amined by continuous o-ray difiraction Am. Minerol. 46, 196-208. --- (1961b) Phase transformations in silica-aiumina mixtures as examined by con- tinuous r-ray diffraction. Am. Mineral,.46, 7064-1076.

Manuscript recehtetl,April 9, 1962.