ClayMinerals (2002)37, 487–496

Identificationofhalloysite(7A Ê ) by ethyleneglycolsolvation: the‘MacEwaneffect’

1, 2 S. H IL L IE R * AN D P. C. R Y AN

1MacaulayInstitute, Craigiebuckler, Aberdeen, AB15 8QH, UK, and 2GeologyDepartment, Middlebury College, Middlebury,VT 05753USA

(Received 14 September 2001; revised 14 January 2002 )

ABSTRACT:X-ray powder diffraction patterns of halloysite (7 A Ê )are characteristically altered following solvation with ethylene glycol. Some effect was first noted in the classic work of MacEwan but its value in the unequivocal identification of halloysite (7 A Ê )seems to have been overlooked subsequently. The response to ethylene glycol solvation involves adecrease in the intensity (peak height) of the peak at ~7.2 A Ê and an increase in the intensity (peak height) of the peak at ~3.58 AÊ thus narrowing the 7.2 A Ê /3.58 AÊ peak height intensity ratio. For pure samples of halloysite, this ratio is narrowed by an average of ~50%. This distinctive change is related to the interstratified nature of halloysite (7 A Ê ),specifically the presence of ‘residual’interlayer water, i.e. halloysite (10 A Ê ),which can be replaced with ethylene glycol so forming 10.9 A Ê layers, aspacing that is almost exactly one and ahalf times the thickness of dehydrated (7.2 A Ê )layers which do not imbibe ethylene glycol. Thus the separation between the 001 peaks in the 7.2 A Ê /10.9 AÊ Ê Ê interstratification is increased and the 002 7.2 (3.58 A) and 00310.9 (3.63 A)peaks become more or less coincident, compared to the 7.2 A Ê /10 AÊ interstratification, i.e. the partially hydrated state. The widespread use of ethylene glycol solvation in studies makes it aparticularly useful and simple test to determine the presence of halloysite. Pure halloysites should be readily identifiable and experiments indicate a‘routine’sensitivity of ~20% halloysite in mixtures with , although this will depend on factors such as ‘crystallinity’and could be improved with careful attention to intensity measurements. It is proposed to call this phenomenon the ‘MacEwan effect’in honour of its discoverer Douglas Maclean Clark MacEwan.

KEYWORDS:halloysite,kaolinite, kaolin, ethylene glycol, X-ray powder diffraction, MacEwan.

Thefundamental feature that distinguishes halloysite example,analysis by XRPD beforeand after heating fromother members of the kaolin subgroup is the shouldbe definitive (Brindley & Brown,1984). The presenceof interlayer water (Churchman & Carr, interlayerwater is, however, very labile so that 1975).In a completelyhydrated state, samples of halloysiteis most commonly observed in a more halloysiteexhibit X-ray powder diffraction (XRPD) dehydratedform. Furthermore, changes in the state patternswith an intense peak at 10 A Ê . This ofhydration have substantial affects on XRPD correspondsto a singlesheet of water molecules patternsof halloysite, and were largely the source ~2.8 AÊ thickbetween the 7.2 A Ê layers.In this state ofmuch early debate over nomenclature (summar- theidentification of halloysiteis straightforward.For izedby Churchman & Carr,1975). This has been resolvedby the general acceptance of the recom- mendationthat the ‘ d-spacing’of the 001 peak, *E-mail: [email protected] whichreflects the state of hydration, is appended to DOI: 10.1180/0009855023730047 thename halloysite, and other names such as

# 2002The Mineralogical Society 488 S. Hillier and P.C. Ryan

‘metahalloysite’have been dropped (Churchman & amongstthe other organic compounds that he Carr,1975; Bailey, 1980; Giese, 1988). Halloysite studied.He also noted that the reaction was not (7 AÊ ),i.e.the largely dehydrated form, poses some observedafter heating his samples (75 8C, 12 h) problemsof identification since its XRPD pattern priorto exposure to ethylene glycol. The purpose of resemblesthat of many , with which it is thecurrent paper is to reaffir mMacEwan’s commonlyadmixed. Indeed, Brindley et al. (1963), observationsregarding halloysite (7 A Ê ). This is usingprepared mixtures of halloysite (7 A Ê ) and doneby illustrating, using modern approaches, that kaolinite,suggested that up to 60% halloysite (7 A Ê ) theintensity and the shape of the basal peaks on couldbe easily overlooked when examined by XRPD patternsof halloysite (7 A Ê )thatwe have XRPD. Indetail, features in XRPD patternssuch examined,are invariably, and often dramatically, asbroader peaks (FWHM >0.3 82y)andlarger basal affectedby ethylene glycol solvation, depending spacings(>7.15 A Ê )thanthose normally observed for uponheating history. The changes observed are kaolinite,as well as relatively intense non-basal explainedby comparison with simulated XRPD peaksin oriented sample mounts, point strongly to patterns.Additionally, the results suggest that the thepresence of halloysite (7 A Ê ).Avarietyof other effectof ethylene glycol on halloysite (7 A Ê ) is techniquesincluding electron microscopy, thermal sufficientlydistinct to form the basis of a routine analysisand IR spectroscopymay also indicate the andreliable test for its presence in asamplewithout presenceof halloysite in a sample.The merits and recourseto other treatments. The utility of such a perspectivesof each of these techniques were testlies in the widespread use of ethylene glycol summarizedbyCh urchman&Carr(1 975). solvationas an aid to the characterizatio nofother Nonetheless,confirma tionof the presenc eof commonclay minerals. Experiments with prepared halloysiteis generally based on empirical tests that mixturesindicate that comparison of the changes relyon differences in the chemical reactivity of betweenXRPD patternsrecorded for the air-dried halloysitewith various organic compounds when andsubsequently the glycolated state is routinely comparedto other kaolins (Wada, 1961; Miller & capableof detecting as little as 20% halloysite Keller,1963; Range et al.,1969;Churchman et al., mixedwith kaolinite. 1984; Theng et al.,1984).The most recent test of thiskind proposed by Churchman et al.(1984)relies MATERIALSANDMETHODS onthe much more rapid formation of a 10A Ê complexby halloysite compared to kaolinite (and Fivekaolins known to contain halloysite were dickiteand ) when treated with formamide. selectedfromthe Macaulay Institut emineral Thistest has been widely adopted (e.g. Singer, 1993; collection,namely; Hal-9, Hal-12, Hal-14, Hal-26 Merriman& Kemp,1995; Kretzschamer et al., andHal-27. Hal-9 is also labelled ‘ Heddle’s 1997),no doubt in part because it consists of a Halloysite’and comes from Hospital Quarry, singlestep and is therefore relatively quick and easy Elgin,Scotland (Heddle, 1882). Hal-12 and Hal-14, toapply, but also because it appears to provide an arefrom an unknown locality in Surinam (possibly accurateindication of the amount of halloysite ‘Brokopondo’, SjerryVan de Gaast pers. comm., present(Churchman, 1990) in comparison to other 2001),whilst Hal-26 is from Hong Kong (Merriman techniques. &Kemp,1995), and Hal-27 is sold by Wards Inhis classical investigations into the formation NaturalScience as ‘ ’(catalogue number ofhalloysite-orga nicintercalates, MacEwan (1946, 49-0616).All samples had been stored at room 1948,1949) demonstrated that a one-layerethylene temperatureand humidity prior to analysis, some for glycolhalloysite complex could be formed from the manyyears. Additionally, numerous samples of fully-hydrated10 A Ê formof halloysite. Later work kaolinitewere also selec ted.Cla yfractions hasalso shown that the one-layer ethylene glycol (<2 mm)were separated from all samples by timed intercalatecan be prepared from less hydrous sedimentationand prepared as orientedspecimens on halloysite(7 A Ê ),viaprior steps involving the 3 cm2 glassslides by a filterpeel transfer method, displacementof other intercalates (e.g. Miller & similarto that described by Drever (1973). The Keller,1963). Previously, MacEwan (1946, 1948) XRPD patternswere recorde droutinelyfrom hadalso noted the direct, but partial, reaction of 2 ­ 4582y using Co-Ka radiationsuccessively on ‘metahalloysite’(halloysi te7 A Ê )withethylene thesame samples, after drying in air, after solvation glycoland the exceptional nature of this reaction byethylene glycol vapour at 60 8Covernight,and Identification of halloysite 489 afterheating to 300 8Cfor1 h.Subsequently, all (notablyHal-27) showed clear evidence of a sampleswere re-exposed to ethylene glycol vapour minor 10 AÊ component(Fig.1). Neverthe less, overnightfollowing heating to 300 8C.Additionally, comparisonof the various patterns recorded for four100 mg mixtures of Hal-9 with a reference eachanda llsp ecimensrevealedcle aran d kaolinite(KGa-2, available from the Clay Minerals consistentresponses of the 7 A Ê peaksto glycolation Society,Source Clay Minerals Repository) were andto heating. Thus, without exception, the peak preparedby weighing out freeze-dried <2 mm size heightintensity of the peak near 7.2 A Ê decreased fractions.Note that freeze drying was verified to andthe peak height intensity of the peak near haveno noticeable affect on theresponse of Hal-9to 3.58 AÊ increasedin the glycolated patterns (Fig. 2), solvationwith ethylene glycol. Mixtures were relativeto their values in the air-dried patterns preparedwith concentrations of 5, 10, 20 and 40% (Fig.1). Furthermo re,in all but one sample, Hal-9,dispersed in de-ionized water using an glycolationresulted in the peak near 3.58 A Ê ultrasonicprobe and oriented mounts prepared and becomingthe most intense peak (height) in the examinedas describe dpreviously.All XRPD patterns.Thus the 7.2 A Ê /3.58 AÊ peak height patternswere obtained by step scanning in 0.02 82y intensityratio was reversed from that observed in steps,counting for 2 sperstep, using fixed 1 8 theair-dried state. This results in a ratherodd divergenceand anti-scatter slits, 0.6 mm receiving lookingdiffraction pattern for a kaolinmineral slits,and tube settings of 40 kV and 40 mA. Peak sincethe 3.58 A Ê peakbecomes the strongest peak heightintensity measurements were made using inthe pattern, whereas normally the 7.2 A Ê is the BrukerDiffrac Plus software by first automatically moreintense peak (assuming a diffractometerwith subtractingthe background using parameters of 0 fixeddivergence slits). Additionally, changes in and1 forcurvature and threshold, and then using the peakshape were apparent. In this respect, the 7.2 A Ê peaksearch algorithm to locate peaks and auto- peaktends to sharpen up, sometimes associated maticallydetermine intensities (peak height). withthe development of a smallpeak at ~10.9 A Ê , othertimes with a morediffuse increase in intensity inthis region of the pattern, best described as a R E S U L T S low-angletail.Similarly ,theincrease in the TheXRPD patternsobtained from the air-dried intensityof the 3.58 A Ê peakupon glycolation is halloysitesamples indicated that all were, by and associatedwith a noticeablynarrower intensity large,dehydr ated7 A Ê forms,althou ghsome spreadof its peak tails to both low and high

7.2 ¯

3.58 ¯

10.0 ¯ Hal-9

Hal-12

Hal-14

Hal-26

Hal-27

4 10 20 30 °2 q Co-Ka

FIG.1. XRPDpatterns of <2 mmoriented samples of halloysite after drying in air. Note Hal-14 contains anotable amount of gibbsite. Patterns are displaced along the vertical axis for clarity. 490 S. Hillier and P.C. Ryan

7.2 ¯ 3.58 ¯

Hal-9

Hal-12

Hal-14

Hal-26

Hal-27

4 10 20 30 °2 q Co-Ka

FIG.2. XRPDpattern of <2 mmoriented samples of halloysite after exposure to ethylene glycol. Note the reversal of the 001/002 peak height ratio compared to that in the air-dried samples (Figure 1). Patterns are displaced along the vertical axis for clarity. angles.Patterns recorded after heating to 300 8C for 2and3 butare most fully appreciated by directly 1halsodemonstrated a noticeableincrease in peak overlayingthe various diffrac tionpattern sas heightintensity compared to both air-dried and illustratedfor one sample in Fig. 4. Attempts to glycolatedtraces, a narrowingof peak widths, and effectchanges in the XRPD patternsafter heating to therestoration of a more‘ normal’7.2 A Ê /3.58 AÊ 3008Cbya secondexposure to ethylene glycol peakintensity ratio (Fig. 3). These changes are vapourshowed only negative or slight responses. A apparentfor all the patterns illustrated in Figures 1, slightresponse was characte rizedby a small

7.2 ¯

3.58 ¯

Hal-9

Hal-12

Hal-14

Hal-26

Hal-27

4 10 20 30 °2 q Co-Ka

FIG.3. XRPDpatterns of <2 mmoriented samples of halloysite after heating to 300 8Cfor 1h. Patterns are displaced along the vertical axis for clarity. Identification of halloysite 491

7.2 ¯

3.58 ¯

Air-dried Glycolated Heated (300 Ê C)

11 12 13 14 15 16 17 25 26 27 28 29 30 31 °2 q Co-Ka

FIG.4. Detailed comparison of air-dried, glycolated and heated XRPDpatterns for sample Hal-9. decreasein the peak height of the 7.2 A Ê peak and a notedthe exceptional nature of the reaction of smallincrease in the peak height of the 3.58 A Ê ethyleneglycol with halloysite (7 A Ê )amongstthe peak,relative to the heated trace. In contrast to organiccompounds that he studied. This included theseobservations for halloysite, examination of glycerolwhich was stated to have ‘ provedless numeroussamples of kaolinite showed no obvious effective’thanethylene glycol, and we have changesin peak intensities, widths or shapes, confirmedthis by exposure of some samples to regardlessof treatment. Measured peak height glycerolvapour for periods up to several days at intensitiesand ratios for all samples of halloysite, 1108C. alongwith those for the prepared mixtures of Hal-9 andthe pure reference kaolinite KGa-2 are listed in D I S C U S S I O N Table1. The percent agechange of the ratio betweenair-dried and glycolated states is also Eversince the pioneering work of MacEwan (1946, listedsince this paramete risindepend entof 1948,1949) it has been known that the fully experimentalfactors such as fixed vs.variableslit hydrated10 A Ê formof halloysite can form a one- systemsand is therefore a mostuseful measure of layercomplex with ethylene glycol with a resultant thechange. As mentioned previously, MacEwan spacingof ~10.9 A Ê .Furthermore,MacEwan (1946,

TABLE 1. Intensity data (peak height, counts per second) for air-dried and glycolated states, peak height ratios, and percentage change of peak height intensity ratios, for samples of halloysite, reference kaolinite (KGa-2), and prepared mixtures of halloysite (Hal-9) and kaolinite (KGa-2).

Sample Halloysite Air-dried Glycolated Air-dried Glycolated Change % 7.2 3.58 7.2 3.58 7.2/3.58 7.2/3.58 %

Hal-9 100 3966 2003 2888 3657 1.98 0.79 60 Hal-12 100 2977 2166 2378 2795 1.37 0.85 38 Hal-14 100 766 453 411 793 1.69 0.52 69 Hal-26 100 1510 819 1187 1573 1.84 0.75 59 Hal-27 100 730 491 654 663 1.49 0.99 34 KGa-2 0 20342 12923 21159 13436 1.57 1.57 0 5% Hal-9 5 24244 16998 25102 17971 1.43 1.40 2 10% Hal-9 10 21708 17997 22280 18920 1.21 1.18 2 20% Hal9 20 15099 10969 15715 12050 1.38 1.30 5 40% Hal-9 40 9276 6659 9486 7676 1.39 1.24 11 492 S. Hillier and P.C. Ryan

1948)also recorded some susceptibility of more importantlyto ethylene glycol solvatio nafter dehydratedforms of halloysite to treatment with heatingsuggests little reason to doubt MacEwan’ s ethyleneglycol. In the first edition (Brindley, 1951) andBrindley ’sexplanationthat the entry of ofthe Mineralogical Society Monograph on the ethyleneglycol into halloysite depends on the X-rayidentification and crystal structures of clay presenceof ‘ residual’waterlayers. Although minerals,Brindley commented on the work of wateris very easily and irrevers iblylost or MacEwanand noted, in accordance with earlier removedfromhalloysitei nthel aboratory, commentsby MacEwan (1948), that it ‘‘seems accordingto the work of Brindley & Goodyear probablethat the entry of glycol into the structure (1948)comple teremoval of interla yerwater dependson the silicate layers being already kept requiresheating to temperatures of ~400 8C, at partlyopen by residual water layers ’’. In later whichthe layers themselves are on the point of editionsof the monograph (Brindley & Brown, decomposition.Samples of halloysite (7 A Ê ) invari- 1984),however ,theseobserv ationswere not ablycontain some residual interlayer water, indeed recordedin the text or remarked upon in tables. itspresence is implied by definition (Churchman & ThusTable 5.8 in Brindley & Brown(1984) for Carr,1975). Furthermore, the presence of this water example,although concerned mainly with changes isprobably the key to the use of the wide variety of in d-spacings,contains no remarks on the response empiricaltests for halloysite based on chemical ofhalloysite (7 A Ê )tosolvation with ethylene reactivity(Churchman & Theng,1984), especially glycol,or to heating at 300 ­ 3508C.Thisconveys sincemany are affected by the heating history of theimpression that halloysite (7 A Ê )isunaffected by thesample prior to their applica tion.Whilst suchtreatments. Our results, however, reaffirm the compoundswhich form strong bonds are observationsmade by MacEwan and demonstrate capableof re-expanding dehydrated layers, polar thatchanges in the XRPD patternsbefore and after organiccompounds cannot. Presumably then, the treatmentare substantial. XRPD responseof halloysite (7 A Ê )toethylene Althoughchanges in peak shape and intensity are glycolso lvationi saresultsimp lyof th e moresubtle than changes in position, the question replacementof the ‘ residual’water layers with arisesas to why the affect of ethylene glycol on layersof ethylen eglycol.Itfollows that if halloysiteappears to have been largely forgotten. interlayerwater were to be lost completely, there Certainly,examination of a numberof excellent shouldbe no response to ethylene glycol solvation, monographsand books with chapters on kaolin aseffectively shown by samples heated prior to (Gieseking,1975; Brindley & Brown,1984; Dixon glycolation(MacEwan, 1948). &Weed,1989; Moore & Reynolds,1997) suggests Perhapsthemostpe culiarfea tureo fthe thatMacEwan’ s observationshave somehow missed diffractionpatterns from the glycolated halloysites gainingwidespread recognition. In part this may be isthe reversal of the 7.2 A Ê /3.58 AÊ intensityratio dueto the fact that although such observations are comparedto their air-dried counterparts. Peculiar easilymade with modern computerized collection of thoughit appears, a simpleexplanation for this XRPD data,the direct comparison of patterns behaviouris suggested from a considerationof the recordedon older XRPD diffractionequipment layerspacings that result following glycolation requiredmoreeffort and attenti onto detail. comparedto those present prior to it. As discussed Furthermore,it was, and still is, not uncommon to previously,it seems reasonably certain that ethylene recordone pattern over a largeangular range but glycolonly replaces the layers, or partial layers, of restrictthe recording of other patterns to much waterpresent in halloysite (7 A Ê ),presumablymore shorterranges, in orderto save time. Procedures such orless completely. Furthermore, it does not result asthis may have caused many workers to miss the inthe opening up or levering open of those layers, changesthat occur to halloysite (7 A Ê ) following orparts of layers, where water is not present, at ethyleneglycol solvation. Others may have dismissed leastnot to any noticeable extent since the result anydifferences as due to small misalignments of wouldbe the end-mem berhalloys ite-ethylene samplepositioning between one recording and the glycolcomplex. Instead, some form of interstrati- next,perhaps because their consistency from one ficationresults between a one-layerglycol halloy- basalpeak to another may not have been apparent. sitecomplex with a spacingof ~10.9 A Ê and other Acomparisonof the response of halloysite (7 A Ê ) layersor parts of layers with no glycol, and a toethylene glycol, and to heating, and most spacingof 7.2 A Ê .Thekey to the behaviour of the Identification of halloysite 493 relativeintensities upon glycolation is that the 003 Modellingof some pa tternswith a discrete peakfrom a halloysiteglycol complex is almost diffractionmaximum in the region of 10 A Ê also exactlycoincident with the 002 from the 7.2 A Ê clearlyrequires an element of R1 (Reichweite = 1, component,whereas this is not the case for the i.e.nearest neighbour) segregation of layer types to hydrate.Additionally, the 001 peaks are corre- beintroduced in accordance with the work of spondinglyle ssc oincident.Simplyap plying Churchman et al.(1972).Thus, segregation may Mering’s principle(e.g. see Moore & Reynolds, needto be considered in order to model the 1997)to this scenario would predict diminished experimentalpatterns precisely. Nonetheless, both intensityfor the 7.2 A Ê peakand increased intensity randomand partially segregated structures indicate forthe 3.58 A Ê peak.In otherwords the effect is due areversalof the intensity ratio from the hydrated to tothe ethylene glycol halloysite complex being theglycolated state. The other point that the almostexactly one and a halftimes the thickness of modellingbrings home is that the basal peaks of theuncomplexed, i.e. 7.2 A Ê layers.That this change partiallydehydrated halloysite samples are really inthe layer spacings will indeed produce the mixed-layerpeaks, i.e. the ‘ 001’at ~7.2 A Ê and observedeffects can be illustrated by calculated ‘002’at ~3.58 A Ê aremore precisely combinations XRPDpatterns.Figure 5 illustratesa pattern of001/ 001and 002/ 003peaks. That is, of course, calculatedfor a randominterst ratificationof assumingo nlytwo co mponents/spacingsare 7.15 AÊ layerswith hydrated 10.1 A Ê layersin the involvedwhereas it may be necessary to consider proportion9:1 compared to the same structure with three,at least where the hydrate interstratificati onis thehydrated layers replaced by glycolated layers concerned(Costanzo & Giese,1985). witha spacingof 10.9 A Ê .Thepatterns were Thevalue of using ethylene glycol solvation to calculatedusi ngthe pro gramdevelopedby identifyhalloysite lies mainly in its routine use for Planc° on& Drits(1999) with experimental para- otherpurposes. If the changes in the XRPD metersappropriat eforcomparison to the real patternsthat we have described are observed, patterns,and assuming a lognormal crystallite theyprovide an unequivocal identificat ionof the sizedistribution with a meanof 10 layers. Trial and presenceof halloysite. Certainly, they are suffi- errormodelling of several different patterns for cientlydistinct from the response of minerals such differentlayer proportions all showed that the asmixed-layer kaolinite-sm ectiteto glycolation reversalof the intensity ratio was a consistent andheating. It is perhaps conceivabl ethatsome effect,but the details were also found to be mixturesmay be too complex for its application , sensitiveto different particle size distributions. butwe have found it tobea valuabletest in studies

7.2 ¯ 3.58 ¯

Air-dried Glycolated

4 10 20 30 °2 q Co-Ka

FIG.5. Computer-simulated XRPDpatterns of air-dried and glycolated halloysite with 10% of layers containing water (air-dried) or ethylene glycol (glycolated). 494 S. Hillier and P.C. Ryan ofreasonably complex mixtures from tropical soils thispoint. Finally, comparison of the intensity data containingboth 10 A Ê and 7 AÊ halloysite(Kautz & forthe halloysites with those for the kaolinite Ryan,2001). For samples in which the kaolin KGa-2,as given in Table 1, illustrates the very presentis pure halloysite, the change in the peak commonobservation that the intensity of diffrac- heightinte nsityrat iobetwe enair- driedan d tionfr omhalloysitei norientedspecimens glycolatedstates should be obvious. The simplicity measuredas peak height is invariably very much ofthe method, however, precludes a preciseand/ or weakercompared to kaolinite. Given the similarity universalanalysis of its sensitivity because of a ofcomposition and structure, these differences in varietyof factors that may affect the measured intensitycan be confidently assigned to both lower peakheight ratios. Firstly, the percentage change ‘crystallinity’and less 001 preferred orientation of inthe intensity ratio of different halloysite samples halloysite.In a mixturein which kaolinite and variesbetween 34 and 69% (Table 1). The lowest halloysiteha vee xactlythe same degree of valuewas obtained for Hal-27. This sample has the ‘crystallinity’and of preferred orientation with broadestpeak widths, suggesting that factors such respectto the 00l series of planes then the as‘ crystallinity’and/ orthe initial proportion of relationshipbetween the proportion of halloysite hydrated vs.dehydratedlayers may have important andthe pe rcentagechange in the mea sured effectson the simple peak height ratio. Indeed, the intensityratio should be linear. However, because amountof residual water layers in halloysite (7 A Ê ) halloysiteis generally more poorly ‘ crystalline’ isunlikely to be the same in different specimens, thankaolinite and if, as is likely, halloysite mixed althoughmany will probably lie within a restricted withk aolinitead optsa differentdeg reeof range.Secondly, all five halloysites examined are preferredorientation ,thenthe percentage change onlyassumed to be pure with respect to the kind of inthe intens ityratio betwee nair-driedand kaolinpresent, i.e. halloysite, so that it is difficult glycolatedpatterns will also depend on these toexclude the possibility that some of thevariation differences.Such factors will make the relationship isdue to the presence of other kaolin minerals in ofthe percent change in the intensity ratio vs. somesample s.Examin ationofdata for t he halloysiteconcentrati onnon-linear, as found for percentagechange in the basal peak height ratio themixtures of Hal-9 with the kaolinite KGa-2 frommany more samples will be needed to address (Fig.6). The importance of these effects is that

80

Theoretical relationship, halloysite and kaolinite Pure halloysites o

i diffracting with equal intensity (peak height ) t a r

t Measured, pure halloysite samples and prepared h

g mixtures with kaolinite (KGa-2 ) i 60 e h

k Theoretical relation ship, halloysite diffracting with a a

e quarter of the intensity of kaolinite (peak height) p

l a s a b

40 n i

e g n a h

c Prepared mixtures

% 20

0 0 20 40 60 80 100 % Halloysite

FIG.6. Plot of %halloysite vs.%change in the basal peak height ratio. The solid line represents atheoretical relationship if both halloysite and kaolinite diffracted with equal intensity (peak height). The dashed line represents atheoretical relationship for apeak height intensity of halloysite which is aquarter of that of kaolinite, for example due to factors such as alower degree of preferred orientation and poorer ‘crystallinity’of halloysite. Both theoretical lines assume that the percentage change in the peak height ratio for pure halloysite is 60%. Identification of halloysite 495 differencesin ‘ crystallinity’and preferred orienta- very constructive reviews of Joe Dixon and an tionwill undoubtedly vary for samples of different anonymous reviewer are gratefully acknowledged. origins,so that the sensitivity of detecting and/ or quantifyinghalloysite will likewise vary. REFERENCES Measurementsof changes in peak ratios for the preparedmixtures of Hal-9 with KGa-2 indicated Bailey S.W. (1980) Summary of recommendations of that20% halloysite produced a 5%change in the AIPEAnomenclature committee. Clays and Clay Minerals, 28, 73 ­ 78. intensityratio, and 40% halloysite an 11% change Brindley G.W. (1951) X-ray Identification and Crystal (Table1, Fig. 6.) For smaller amounts of Hal-9, Structures of Clay Minerals .Pp. 345, Clay Minerals onlya 2%change was observed. A 2%change is Group, Mineralogical Society, London. probablytoo close to the overall precision of Brindley G.W. &Brown G.(1984) Crystal Structures of ‘routine’intensity measurements to be certain that Clay Minerals and their X-ray Identification . asmallamount of halloysite is present. However, Monograph 5,Mineralogical Society, London, p. employingstrategies to increase the number of 495. countsaccumulated and/ ordirecting special atten- Brindley G.W. &Goodyear J.(1948) The transition of tionto ensure the reproducib ilityof intensity halloysite to metahalloysite in relation to relative measurementsbetwee nrunscould no do ubt humidity. Mineralogical Magazine , 28, 407 ­ 422. providebetter sensitivity if needed. Brindley G.W., Santos P.de Souza &Santos H.de Souza (1963) Mineralogical studies of kaolinite- halloysite clays. I. Identification problems. American CONCLUSIONS Mineralogist , 48, 897 ­ 910. Churchman G.J. (1990) Relevance of different inter- Asuiteof XRPD patternsfor air-dried, ethylene calation tests for distinguishing halloysite from glycol-solvated,andheated samples, such as is kaolinite in soils. Clays and Clay Minerals , 38, normallyrecorded routinely for many studies, is 591 ­ 599. sufficientto determine whether or not halloysite Churchman G.J. &Carr R.M. (1975) The definition and (7 AÊ )ispresent in a sample.An effect of ethylene nomenclature of halloysites. Clays and Clay glycolsolvation on halloysite (7 A Ê )wasoriginally Minerals, 23, 382 ­ 388. discoveredby MacEwan (1946). Our experience Churchman G.J. &Theng B.K.G.(1984) Interaction of withthe rediscovery of this fact is that MacEwan’ s halloysites with amides: mineralogical factors affect- observationsdeserve to be more widely publicized. ing complexformation. Clay Minerals , 19, Itis p roposedto call the phenom enonthe 161 ­ 175. Churchman G.J., Aldridge L.P.& Carr R.M. (1972) ‘MacEwaneffect’ in honour of Douglas Maclean Relationship between the hydrated and dehydrated ClarkMacEwan (1917– 2000) who first observed states of ahalloysite. Clays and Clay Minerals , 20, andrecorded it whilst working at the Macaulay 241 ­ 246. Instituteinthe late 1940s. The presenc eof Churchman G.J., Whitton J.S., Claridge G.G.C.& interlayerwater is central to the definition of Theng B.K.G.(1984) Intercalation method using halloysiteand appears to be a prerequisitefor the formamide for differentiating halloysite from kaoli- intercalationof ethylene glycol. There can, there- nite. Clays and Clay Minerals , 32, 241 ­ 248. fore,be no ambiguity concerning the identification Costanzo P.M.& Giese R.F.Jr (1985) Dehydration of ofhalloysite following the use of the ethylene synthetic hydrated kaolinites: amodel for the Ê glycoltest; such as may remain if identification is dehydration of halloysite (10 A ). Clays and Clay basedon morphology alone. Minerals, 33, 415 ­ 423. Dixon J.B. &Weed S.B.(1989) Minerals in Soil Environments .Soil Science Society of America, ACKNOWLEDGMENTS Madison Wisconsin, 1244 pp. Drever J.I.(1973) The preparation of oriented clay SHgratefully acknowledges the able laboratory mineral specimens for X-ray diffraction analysis by a assistance of Caroline Thomson, some sand and a filter-membrane peeltechnique. American few sticks on the shores of Lake Dunmore, and the Mineralogist , 58, 553 ­ 554. Scottish Executive Environment and Rural Affairs Giese R.(1988) Kaolin minerals: structures and Department (SEERAD)for financial support. PCR stabilities. Pp. 29 ­ 66 in: Hydrous Phyllosilicates acknowledges laboratory assistance from CQKautz (exclusive of ) (S.W. Bailey, editor). Reviews and financial support from Middlebury College. The in Mineralogy, 19.Mineralogical Society of 496 S. Hillier and P.C. Ryan

America, Washington D.C. Miller W.D. &Keller W.D. (1963) Differentiation Giesking J.E. (1975) Soil Components, Volume 2, between endellite-halloysite and kaolinite by treat- Inorganic Components .Springer Verlag, Berlin, ment with potassium acetate and ethylene glycol. 684 pp. Clays and Clay Minerals , 10, 244 ­ 253. Heddle Prof. (1882) Mineralsnew to Britain. Moore D.M.&Reynolds R.C.Jr. (1997) X-ray Mineralogical Magazine , 5, A-2. Diffraction and the Identification and Analysis of Kautz C.Q.& Ryan P.C.(2001) Formation of multiple Clay Minerals .Oxford University Press, NewYork, halloysitic phases in aneotropical fluvial terrace 378 pp. sequence. Geological Society of America , Abstracts Planc° on A.&Drits V.A.(1999) Programs for the with Programs, 33, A437. calculation of diffraction by oriented powders of Kretzschmar R., Ronarge W.P., Amoozegar A.& two- and three-component mixed-layer clay miner- Vepraskas M.J. (1997) alteration to halloysite als.http://www.univ-orleans.fr/ ESEM/plancon/ and kaolinite in soil-saprolite profiles developed Range K.J., Range A.&Weiss A.(1969) Fire-clay type from schist and granite gneiss. Geoderma, 75, kaolinite or fire clay minerals? Experimental classi- 155 ­ 170. fication of kaolinite-halloysite minerals. Proceedings MacEwan D.M.C. (1946) Halloysite organic complexes. Nature, 157, 157 ­ 160. of the International Clay Conference, Tokyo , 3 ­ 13. MacEwan D.M.C. (1948) Complexes of clays with Singer A.(1993) Weathering patterns in representative organic compounds I. Complex formation between soils of Guanxi Province, south-east China, as and halloysite and certain organic indicated by detailed clay mineralogy. Journal of liquids. Transactions of the Faraday Society , 44, Soil Science , 44, 173 ­ 188. 349 ­ 367. Theng B.K.G,Churchman G.J., Whitton J.S. &Claridge MacEwan D.M.C. (1949) Clay mineral complexes with G.G.C.(1984) Comparison of intercalation methods organic liquids. Clay Minerals Bulletin , 3, 44 ­ 46. for differentiating halloysite from kaolinite. Clays Merriman R.J. &Kemp S.J. (1995) Mineralogical and and Clay Minerals , 32, 249 ­ 258. microtextural analysis of altered tuffs associated Wada K.(1961) Lattice expansion of kaolin minerals by with landslips in Hong Kong .British Geological treatment with potassium acetate. American Survey Technical Report, WG/95/30C. Mineralogist , 46, 78 ­ 91.