Cerolite_AM64_615

American Mineralogist, Volume 64, pages 615425, t979

Compositions, structures, and properties of nickel-containing mineralsin the kerolite-pimelite series

G. W. BnrNoLry, Dnvro L. Brsul lNo HslrN-MING WnN2

Mineral Sciences Building, The Pennsyluania State UniuersityUniue rsity Park, P ennsyluania I 6802

Abstract

The term kerolite-pimelite series is used for a series of hydrous magnesium-nickel silicateswith talc-like structure and composition but with additional water and with a highly dis-ordered and non-swelling stacking of the layers. The layer spacing, obtained after correctionfor Lorentz-polarization and structure factor variations with diffraction angle, is near 9.64and is larger than that of talc because of an absence of close-packing of adjacent layers. Theaverage composition of nineteen samples is (Mg,Ni), o.(Al,Fe)o 0r(Sis.esAlo 02Fen.or)O,o(OH)r'0.89HrO. The additional "water" is held in a variety of ways and is released graduallyup to 700"C. Infrared spectroscopic data suggest the presence ofsurface hydroxylsyielding anabsorption near 3700 cm-1, Si-OH groups with absorption near4550 cm-r, and interlayer andsurface adsorbed HrO absorbing at 1600, 3400, and 3600 cm-1. The splitt ing of OH stretchingand librational moces in the infrared spectra clearly shows the effects of increasing Ni-for-Mgsubstitution. As expected, optical absorption spectra provide evidence for Ni,+ in octahedralcoordination only. The thermal transformations of these minerals are examined, and evidenceis obtained for a transitional face-centered cubic phase prior to the development of a high-temperature phase.

Introduction

The term "kerolite-pimelite series" is used in thesense defined by Maksimovic (1966) and Brindleyand Maksimovic (1974) for a series of magnesium-nickel hydrous silicates with essentially talc-likecompositions and highly disordered, non-swellingstacking of the layers. The relation of the magnesiumend-member, kerolite, to talc and stevensite has beendiscussed by Brindley et al. (1977). The present studyextends this work to similar nickel-containing miner-als which are named pimelites when the atomic pro-portion of Ni exceeds that of Mg.

A major problem in this study, as in the previouswork on the lizardite-nepouite series (Brindley andWan, 1975), was to obtain material sufficiently puremineralogically to merit detailed examination. Thesenickel-containing minerals commonly occur as in-timate mixtures and are called garnierites (pecora e/

I Present address: Department of Geological Sciences, HarvardUniversity, 20 Oxford Street, Cambridge, Massachusetts 02138.

2 Present address: Mining Research and Service Organization,ITRI, Taipei 105, Taiwan, Republ ic of China.

al., 1949); they are analogous to the magnesium-containing deweylites (Bish and Brindley, 1978). Be-cause it is virtually impossible to separate the com-ponents, the only available course is to examine largenumbers of samples and to select those mainly of oneor other type. In the study of garnierites by Brindleyand Pham Thi Hang (1973), only six nickeloan kero-lites and pimelites were available, and two of thesecontained appreciable serpentine impurity; in thepresent work eighteen samples have been obtained,so that we are now in a better position to study thisseries of minerals.

Experimental proceduresPreliminary selection of material was made under a

binocular microscope by hand-picking green, clay-like materials (garnierites). X-ray examination thenshowed if the selected material belonged to the liz-ardite-nepouite series (basal spacing about 7.3.A), orthe kerolite-pimelite series (basal spacing about l0A),or a mixture of the two. In practice, the selection ofmonomineralic material is often less certain than thissimple description suggests, because the only distin-guishing features are the 001 and 002 reflections of

0003-004x/79 /0506-06 I 5$02.00 6 t 5

6 1 6 BRINDLEY ET AL' KEROLITE-PIMELITE SER/ES

the 7.3A phase and the 001 and 003 reflections of the10A phase. These reflections are commonly broad,and small proportions, such as 5 percent or evenmore, of one component in the presence of the othercan easily be overlooked. All other diffraction fea-tures tend to be of the two-dimensional type becauseof layer stacking disorder, and they overlap com-pletely for the two series. Nevertheless, with care anda practiced eye, mainly monomineralic samples canbe selected.

Samples containing predominantly a 10A spacingmineral were separated from coarse impurit ies (par-ticle size greater than about l0-20 pm) by ultrasonicdispersal in water and settl ing under gravity. The finefraction, less than a few microns particle size, wasallowed to dry at -40'C in air. Further X-ray exami-nation, usually of thin oriented samples on glassslides, was made to check for contamination by otherlayer sil icates and for any changes of basal spacingswith humidity (0-100 percent RH in controlled atmo-spheres), with ethylene glycol vapor (usually for sev-eral days at -50oC), and after heat treatment (vari-ous temperatures up to 550'C).

Chemical analysis of suitable materials was madeby atomic absorption methods using the l ithium me-taborate fusion technique of Medlin et al. (1969).Theignition loss of materials heated to 1100'C after pre-vious drying at l l0'C was determined, and the ig-nited material was used for the total analysis.

Combined gravimetric and X-ray diffraction datawere obtained by taking two samples of each materialstudied, which were given identical heat treatments,mainly 3-4 hr heatings at a sequence of temperatures50'-l00oc apart. One sample was used solely forgravimetric measurements and the other for X-rayexamination. By this procedure, weight (water) losswas correlated with structural changes.

Spectroscopic analyses were made in the near-in-frared and visible regions using a diffuse reflectanceattachment to a Beckman DK-2A spectrometer. In-frared spectra were obtained with the KBr pellet tech-nique and a Perkin-Elmer spectrophotometer, model62r .

X-ray diffraction data

Table I records d spacings and 00/ and hk indicesfor representative samples. Diffractometer patternshave been published previously by Maksimovic(1966) and by Brindley and Pham Thi Hang (1973).The apparent values of d(001), taken directly fromdiffraction recordings, are usually in the range 9.8-10.2A^, but when corrected for the variation of the

Lorentz-polarization factor and the structure factorover the range of 20 values involved, the value ofd(001) is near 9.6+0.054 and agrees with 3 X d(003)= 9.554. The increase in the basal spacing as com-pared with the value for talc, about 9.35-9.38A, isattributed to the random stacking of the layers whichdoes not permit their partial close-packing as in thestructure of talc (see Rayner and Brown, 1973). The002 diffraction peak is always weak and is resolvedonly partially from the adjacent 02, ll diffractionband. It is seen more clearly from Mg-rich than fromNi - r i ch samples . Ca lcu la ted in tens i ty ra t ios ,/(001):(002):1(003) are respectively 100:7:41 and100:3:15 for the Mg- and Ni-end-members; the calcu-lations assume the z atomic parameters for talc givenby Rayner and Brown (1913). The sharp 06 diffrac-tion band corresponds closely to the 060 position andgives a value of b : 9.15A.

The effect of the Lp and structure factor variationson the apparent basal spacings, d'(001) and2d'(N2),of talc crystallites containing small numbers of layersis shown by the following calculated values:

No. of layers 3 4 6 l0 12 @d'(001), A 10.98 10.30 9.81 9.54 9.45 9.352d'(N2), A 9.91 9.7r 9.51 9.38 9.38 9.35

Evidently the number of layers must exceed about 20before the effect becomes negligible.

The angular breadth B ofthe basal diffraction pro-files at half-maximum intensity, after correction forinstrumental broadening, gives the average thicknesst of the crystallites by application of the Scherrerequation, Bcos d : 0.9)tlt, provided broadeningarises only from small crystal size. On this basis,values of I of the order of 40-604 or 4-7 structurallayers are obtained. The mean diameter D of thecrystallites is estimated similarly from the profile ofthe 02, I I diffraction band using the expression givenby Warren and Bodenstein (1965): Bcos d : l.9lx/D.Values obtained range from about 90-170A.

The question whether kerolites and pimelites areswelling or non-swelling minerals is crucial in decid-ing whether they belong to the smectite group or tothe talc-willemseite group; willemseite is a nickel-richform of talc described by DeWaal (1970). We havealready shown that kerolites are essentially non-swel-ling minerals (Brindley et al., 1977). Similar testshave been made for nickeloan kerolites and pimelites.Samples were exposed for several days or longer tovarious relative humidities, to ethylene glycol vapor,and to various low-temperature heat treatments.Reoresentative results are shown in Fisure l. The

BRINDLEY ET AL,: KEROLITE-PIMELITE S'RI'S 617

hk and

000

Table l. X-ray powder diffraction data for representativekerolites and oimelites

ily in ethylene glycol to give a regularly expandedspacing of about l7A. When exposed for long peri-ods, the basal spacings of kerolites and pimelites insome cases show modifications which suggest a par-tial penetration of glycol into the structures. Samples( l) , (13) and (18) [see Figs. 1(a), (c), (e)] show theseeffects; samples (10) and (17) [see Figs. l(b), (d)] donot show these effects. Evidently when penetrationoccurs it is partial and irregular, as shown by thepoorly defined diffraction effect. The mechanism bywhich ethylene glycol causes expansion of smectites isstil l poorly understood but probably is quite differentfrom the expansion caused by water which is relatedto cation hydration. Ethylene glycol will penetrateinto reduced charge montmorillonites when waterwill not penetrate (Brindley and Ertem, l97l). Possi-bly glycol penetrates because of an interaction withthe Si-O surfaces of the layers. Since the layer stack-ing in kerolites and pimelites is extremely disordered,such that the layer spacing is near 9.64, as comparedwith 9.35A for talc, there may well be regions wherethe normal interlayer bonding is appreciably or evenconsiderably reduced. The calculations of Giese(1975) show that the interlayer bonding in talc andpyrophyllite is partially ionic and is considerably re-duced when the layer spacing is increased; an increaseof spacing from 9.35 to 9.6,{ produces a 40 percentdecrease in the ionic and van der Waals forces ofinterlayer attraction. Regions of weaker bondingcould well be the cause of a partial penetration ofethylene glycol.

1 ( a ) 2 ( b )

d I d I d3 ( b ) 4 ( a ) 5 ( b )

r o L o t

0 0 1 1 0 . 00 0 2 4 . 8 3

o z , L I 4 , 5 4003 3 , 19

1 3 , 2 0 2 . 5 50 4 , 2 2 2 . 2 8

1 5 , 2 4 , 3 L L . 7 20 6 , 3 3 r . s 2 22 6 , 4 0 1 . 3 r 4

t 1 , 3 5 , 4 2 r . 2 6 8

6515

1005 55 0

5020

5

9 . 8

4 . 5 93 . 1 82 . 6 22 . 3 1I . 7 2 5L , 5 2 51 . 3 1 7L . 2 6 5

9 . 6

4 , 5 63 . 1 92 , 6 2

7 , 7 31 .528

L . 2 7 4

100

1007 07 53020

1002020

100

70506 0

30100

3010

100

7 0501 0

20100

305

9 , 6 1 0 0 9 . 64 . 7 3 1 54 , 5 4 6 0 4 . s 63 . 1 s 1 0 3 , r 72 . 5 3 5 5 2 . 6 2

_ _ 2 , 2 71 , 7 2 1 0 r . 7 3r . s 2 4 4 5 t , 5 2 61 . 315 10 1 . 317

- - r , 2 6 1

1. Kerol iEe from Carter 's Hioe, N. Carol ina. Dried at 110"C2. Nickeloan kerol i te, sanple (6) of Table 2; room condit ions3. Nickeloan kerol i te, sanple (8) of Table 2; roon condit ioos4. Nlckeloan kerol i te, 6anp1e (8) of Table 2; af ter 650oC heat iog5. Pinel i te, sample (17) of Table 2; room condit ions

(a) Dlf f ractoneter data; (b) Debye-Scherrer camera data

compositions of the samples, listed in Table 2, arediscussed later. The various treatments used are listedin the caption to Figure l From patterns labelled (b),samples in 100 percent RH, to patterns labelled (f),samples after 550oC for 3 hr, no change in peakposition is observed, but some samples do show asharpening of the peak profile. This result suggeststhat the layer spacings become somewhat more regu-lar with the heat treatment, but there is no obviousimprovement in the layer stacking order. The resultsindicate an absence of swelling in humid atmospheresand show that these minerals are not smectites.

Their behavior with respect to ethylene glycol va-por is less simple but stil l confirms that they aredistinctly different from smectites which expand read-

9 r 2 2 9 L 2 2 9 1 2 2 9 r 2 2 9 L 2

2 9 d e g r e e s

Fig. l Diffractometer patterns of 001 reflections from kerolites and pimelites; Ni-filtered, CuKa radiation. Figs. (a)-(e) correspond tosamples ( l) , (10), (13), (17), (18) ofTable2. Treatments ofsamples are as fol lows: (a) exposed to ethyleneglycol vapor,25'C,4 weeks; (b)and (c) exposed to 100 percent and 50 percent relative humidity; (d), (e), (f) heated at ll0'C, 350"C, 555"C for 3 hr and cooled indesiccator.

II

IIIFIIIl-IIII

( a )

( b )

( c )

( d )

( e )

( f )

4 _

6 @ *i N o - i E. d o E oJ d d oo d 6 a' i + . Ei ' @H Z r .P . i l dA E F Z

O ' dE t s

e i = , X 1^ . 6 =p € o 9 x

N t s i - : ' F. o F t t a

o d 6 6 o i . i !$ t i o 6 F i J oa s o o o . i i u . di r a i 6 t s d . d dr + t i t t o a B a

d H o d d o o oT . N N T N O . ! J@ s o + o @ F a p 5+ N + + + 6

6 . t > O. i - o . . E p - d

o o i o o o ; . do € o > o o € - dz a z o z z E o h oH P H E H H O . i F !

o . d H i n o> o h o > > i a ; dp d p J p p p g - : d

d d e t J g! o E E € E ) 6 ^o 9 o o o @ F 3

i . i . i i . - od i N i d r 6 vA . g A A . da o a > a p o oa c S F r S n o d x

6 . 4 6 @ 6 A O i Pi E o o o oo o d o o x. t s . . d . . i . E i d

6 d d d d d 6 ^ q 4 2i a . i a . i i l 9 E oo o a o oo E o 5 o o i o . . i od u E @ c c d o - lo t s o C o . O J . . i! o ! E 5 E E d t r E .d z d . d d . d d N d

H F d H H O O d O O N. n ! H P t v

^ ! . p . . o . F q ad d o ! o o t r @ E t r o oa I 3 0 J J 0 0 0 > a

o o d o d o 3 0 ! : l II O O O O E d d c C @d E ! . ! t s O ! P d O O, o o d o o t s d E t s d d

6 A A A @ 6 h F < r F +

o i N o $ 4 a t s @ 6i d i i i i d d - -

o p oN N i i s it t t t r l

$ d s d i iN t s T+ o + @ 4 6

+ + + +

N o N o o o6 Q 6 O O Q/ 2 4 Z Z Z

H H H H

o h o! ! u p p p

o ! o € d €i o i o o oi n i . d . d . io d o d d d

. 6 . @ A @

o o

i d i d d 6o . i o . d . d . dt s o ! a @ @

. d O d O O O^ a C o t r C t r4 0 0 0 04 . ! . € E €o z c z c d cd H H H H

! - !h d d d 6 6 6

l k 6 9 d 6 6o o i o H d dG O d O d d 6

i o o o o o o@ Q A N A A A

.;i ^ P> Fo r EE 6 0

. i i p@ v p . d

J O t sO . i o

: i I od H o

. N €

N ! Oo +

h tF h ^ o

o o >! i I Fo E z d

a h 5

o c E >. i o F

d ! i Id o a o> p a . io o p i

i i @ ao d q

@ o . 65 0 d

h o. . i O O

o c 6 d . do . i c o E

i d d E o

E d + H Od t 6 Fo o E

E O O -q . d l oo o ! d N

o . i o EO H E F Oo o ! o Ho o < o v

a a N o $ 6 0 r @ o

N

N

N

s

N

a

d

j

+

6 @ 4t t 6 d i

o N 6

= t N @ o 6e o r o d

o o o d N

N N { @I O N F O

o i d N

= ! i @ d Nv o N o 6

o o d i N

Q N 6t t i @ o

N O O

@ o 9t t o r i

^ i c i ;

@ N h ot o i N 6

O N O N

. d N d 6r o + 6 6

o i l i N

E ] N 6 N $e o s s 6

o o d d N

N 0 @t t 9 0 6

i . i . ;

r r st l @ N d

i d o

N r 6r r 1 ? 1i i N

o @ s| | ? 9 ' 1N H O

N 6 rr t i 6 0

d c ; ;

d s i t st o o 6 0

Q N O N

6 d or r n : 1 " . :

N O O

o r Nt l s 6 6

N O N

@ r €t t 6 0 0

N O 6

6 d or t o o i

o o o

;H UH U

d a i l i< r E z H

BRINDLEY ET AL,: KEROLITE-PIMELITE SERIES

o i i N6 0 Q O

o i o6 0 t o

o o s

o t t o

d;6

+I

Nd

6o

i

o

a

66

Ii

i

c;

i

o

N

o

"i

N

6

I

oo

d

6

d

ots

@ o d o 9 6 i oi o o N N t t o o o 6

@ o o o @ o o o o @$ o

@ $ 6 @ S O $ Na o o N 6 r t N o o o

o o o Q a N o o o @

r 6 0 0 6 r N No o o o o N t i o o 4

N O O O O @ O O A r4 H N

r 9 0 6 r @ @ ro o o d d i t d o o N

N O O O d € O O O @6 i N

N 9 n i N o a Qi o o @ Q @ t N 6 0 i

d o o i o + a o o N6 i N

o s 6 0 0 0 N an o N o $ 0 t o + d a

6 0 0 0 6 0 0 0 0 ah i N

o N i o 6 N O Nn i O N 6 @ t d O i r

o o o o s N o o o N6 i N

s N N 0 i @ 6+ t N s r o i i i d 6

d o o 9 d a o o o @6 i N

4 0 i d t s N S6 t N N @ + A i O d 9

s O O N @ O O O O T4 i i

d @ e @ N O @o o d N 6 o t o N i t s

o o o o @ @ o Q o Nn i d

d a i o 4 r o $$ o i N d o t o i d o

h o o o o @ o o o rn N d

i 6 r d @ @ i6 t 6 i d 6 o i o o 6

6 0 0 6 4 0 0 0 0 an i H

o 6 N 6 + @ 6i l i d $ N o 4 d o 6

+ o o d 6 0 0 0 0 r6 N i

s N o s @ + 6t s r N 4 4 $ O i O i O

4 0 0 0 6 0 0 0 0 a4 N i

i @ N s o i Ns t o s + o o i o i 6

$ Q O N O O O O O t s4 N i

@ Q N O @ O 4N l o 4 0 0 0 N h H O

n o d 6 0 0 0 0 0 N6 d i

@ o s N 6 9 6 6o o o N o o t N o o o

N O O O O O O O O @6 N d

o r o o o @ o i l9 t N @ N @ O H N O 6

$ 4 o o 9 0 0 0 0 F6 N

o n o s 6 0 6@ t o d @ o t o t t N

_ @4 0

o o

J o ' o - o - o o o o o - o

3n . r d o @ . d d d d N N@ t s < h Z z = e z V E

N 6 0 0@ o d o

6 0 0 $

( n i Ao o t a

o o {

@ o d 6

r N d o

o o o $

o o o o

o o o o

t s o oo o t o

o o s

@ o o @

O Q Q 6

o o o o

4 0 N O6 0 0 0

o o o s

N O O A

o o o o

H F

H OF

. i i o

N O $ d

6 0 0 $

s

i

i

o

i

O

o

O

o

Io

o

o

o

o

q

6

dI

o

(-)al

o

F

BRINDLEY ET AL.: KEROLITE-PIMELITE SERIES 619

Chemical analyses snd formulae

Chemical analyses of nineteen kerolites and pime-lites arranged in order of increasing NiO content arepresented in Table 2 together with the sources of thematerials. The structural formulae, calculated on thebasis of a total cation valence of -t22, also are given.Tetrahedral positions are filled with Si ions togetherwith Al and Fe3+ ions, if available, to fil l four posi-tions. Mg and Ni ions and any remaining Al and Fe8+ions are placed in octahedral positions. The smallamounts of Na, K and Ca ions are neglected; theymay arise from traces of impurities and/or occupysurface positions. The numbers of ions in tetrahedraland octahedral positions are usually close to 4.0 and3.0 respectively; in six cases where )Tetr. falls below3.95, )Octa. rises correspondingly above 3.0 to keepthe total valence at -t22.If poorly crystalline serpen-tine minerals are present as impurities, they wouldtend to raise the ratio of R2+ ions to Sia+ ions. Takenas a whole, the results confirm that kerolites andpimelites deviate little from a talc-like compositionexcept for the high value of HrO. The average com-position of the nineteen samples in Table 2 is[(M& Ni)r.,. (Al, Fe)o.or] (Sis.o8Al0.o2F%.or) O,o (OH),.0.89 H,O.

Three procedures have been used to study the na-ture of the additional water: (i) thermogravimetricanalyses to ascertain loss of weight (water) as a func-tion of temperature, (ii) dehydration-rehydration ex-periments to study if water once lost is easily recov-ered, and (iii) spectroscopic study.

Thermogravimetric curves (typical results areshown in Fig. 7) show a distinct difference betweenthe relatively sharp loss of "hydroxyl" water due tothe decomposition of (OH)r, and the gradual loss ofthe additional water from ll0oC to about 730oC.Dehydroxylation occurs between about 730-800"Cfor kerolites, 730-840oC for nickeloan kerolites, and800o-900o for pimelites. For fine-grained, well-crys-tallized talc the corresponding temperature range is840"-910'C. The long temperature range, ll0o-730oC, in which the additional water is lost suggeststhat it is held by a range of bonding forces. Brindleyet al. (1977) showed that water lost up to 300oC isalmost completely regained and that lost at 650'C isonly 70-90 percent regained; the results suggest thatmuch of this water is held by external surfaces aseither OH or molecular HrO.

Spectroscopic study of kerolite and pimelites

Thirteen samples were studied by infrared spec-troscopy, and the O-H vibrations in particular were

examined in detail. Figure 2 presents near-infraredspectra of a kerolite and a nickeloan kerolite. For allsamples this region is dominated by overtones andsummations of hydroxyl vibrations in the mid-rangeinfrared. Molecular water is indicated by absorptionnear 1900 nm, and most of this water is lost by300"C. The band near 2200 nm is present in all kero-lite and pimelite spectra and is possibly an Si-O-Hcombination band (Scholze, l960a,b), arising fromthe Si-O v, and /s modes plus an O-H stretchingvibration; a similar band is present in the spectra offinely ground talc and quartz but not in the spectra ofthe well-crystallized minerals.

Mid-range infrared spectra of kerolites and pime-lites are analogous to that of talc, with broadenedabsorptions and additional adsorptions due to hy-droxyl and adsorbed water. Figure 3 shows typicalspectra of several nickeloan kerolites, pimelites, and anickeloan talc, Ni,.roMg,.rrSi.O,o(OH)r, in the rangeof 300-4000 cm-1. Figures 4 and 5 show more de-tailed spectra in the range 3000-4000 cm-1 and 600-800 cm-1 respectively. Absorptions of particular in-terest are those at 670-700 cm-r assigned to a libra-tional motion of structural hydroxyl, the water bend-ing vibration near 1600 cm-r, and various hydroxylstretching vibrations between 3400 and 3700 cm-1.The broad band at about 3400 cm-r is due to ad-sorbed water, and the various sharper bands between3630 and 3680 cm-' are due to structural hydroxyl.The spectra in Figure 4 show the 3600 and 3700 cm-1absorptions to varying degrees; as Brindley et al.(1977) noted, the adsorption at 3600 cm-' is attrib-uted to O-H stretching associated with H-bonds be-tween interlayer water and surface oxygens, and theabsorption at about 3700 cm-r is attributed to vibra-tions of surface OH groups. Virtually all spectrashow this high-frequency OH vibration, although tovarying degrees.

The behavior of the nickeloan kerolites and pime-lites with increasing temperature is the same as thatof kerolites; the amount of adsorbed water decreasesreversibly, and the absorption at about 3700 cm-ldecreases irreversibly after heating to 600oC. Accom-panying the large reduction in the intensity of the3700 cm-' absorption is a decrease in the shoulderbetween 850 and 900 cm-' adjacent to the 1014 cm-1Si-O vibration. This shoulder may be due to Si-Ostretching of Si-OH groups; such groups are charac-terized by similar shifts in the Si-O absorption bandof silica gels (Moenke, 1974, p. 369).

Infrared spectroscopy also provides evidence ofchanges taking place with increasing nickel sub-

K e r o l i t e

1 4 0 0N i - k e r o l i t e

1 4 0 0

l \ uoo ' " 2220

o()coltoott

620 BRINDLEY ET AL.: KEROLITE-PIMELITE S'RIES

I 6 0 0 2 0 0 0 2400 | 6 0 0 20 00 2400

W a v e l e n g t h , n m

Fig .2 . Near - in f ra redspec t rao fkero l i te (Car te r ' sMine ,N.Caro l ina ; chem.ana l . inBr ind leye ta l . ,1977,p .448; sample48 l .90)andnickeloan kerolite (Sample 6, Table 2).

stitution. Hydroxyl stretching in pure well-crystal-lized talc occurs at about 3680 cm-l, and kerolitescontaining little or no nickel exhibit a broad absorp-tion close to 3680 cm-1 which sharpens appreciablyafter heating to 660oC. As noted by several investiga-tors (Vedder, 1964; Wilkins and lto, 1967; DeWaal,1970), with an increase in nickel content, up to fourO-H stretching absorptions are possible. These cor-respond to the four possible combinations of thenickel and magnesium ions in the three octahedralsites linked to a hydroxyl, i.e., (Mg,Mg,Mg),(Mg,Mg,Ni), (Mg,Ni,Ni), and (Ni,Ni,Ni) (Wilkinsand lto, 1967). The absorption in the 3680 cm-rregion of the kerolite spectra does indeed split withincreasing nickel, but four individual absorptions arenot observed. The spectrum in the region 3000-4000cm-1 of the nickeloan talc mentioned previously iscompared in Figure 4 with those of a series of kero-lites and pimelites with increasing nickel contents. Asthe amount of nickel increases, additional absorp-tions become noticeable at about 3645 cm-'and 3625

cm-1, corresponding to the arrangements (Mg,Ni,Ni)and (Ni,Ni,Ni) respectively, and are very close tocorresponding absorptions in the nickeloan talc. It isstrange that an absorption corresponding to the ar-rangement (Mg,Mg,Ni) does not appear. The in-tensities of the 3680 and 3625 cm-l absorptions areapproximately equal when the composition is be-tween MgrNi, and Mg,.uNi1.u, and the 3625 cm-Labsorption becomes stronger with more nickel. In-spection of Figure 4 reveals that there is no rigorouslysystematic variation in hydroxyl stretching band in-tensities with composition for the kerolites and pime-lites; the majority of the spectra have 3625 cm-rbands with intensities higher than the calculatedvalue. Thus, it appears that the hydroxyl stretchingvibrations in kerolites and pimelites are not particu-larly useful for composition or ordering determina-tions, as they are for well-crystallized talcs (see Ved-der, 1964).

The relative intensities of the absorptions at 670and 705 cm-' also have been used to estimate compo-

BRINDLEY ET AL.: KEROLITE-PIMELITE SERI,ES

sitions (Stubican and Roy, 1961). Equal intensities ofthe two bands indicates a Ni:Mg ratio of about 2:1.Figure 5, displaying the region between 600 and 800cm-r, shows the effects of Ni-for-Mg substitution andreveals a gradual increase in the 705 cm-1 band withincreasing nickel; for qualitative analysis of theNi:Mg ratio in these minerals, this region appearssuitable.

The effects and nature of the Ni-for-Mg sub-stitution also can be examined by optical absorptionspectroscopy which enables us to describe quan-titatively the color of the minerals and the coordina-tion and oxidation state of the nickel. Typical spectraare shown in Figure 6 and the individual absorptionsare listed and assigned in Table 3. All spectra areconsistent with the presence of octahedrally-coordi-nated Ni2+ ions and agree with spectra of other Ni2+-containing minerals in the literature (White et al.,l97l; Faye, 1974; Wood, 1974). Bands at about 380,660, and I 130 nm are spin-allowed transitions, andshoulders at about 425 and 715 nm are assigned tospin-forbidden transitions using the energy-level dia-gram calculated by Berkes (1968). The band at about1400 nm is due to an OH overtone vibration and isnot electronic in nature. The green color of theseminerals is due to the "window" at about 530 nmbetween the 380 and 660 nm absorptions.

The crystal field splitting parameter, Dq, which is ameasure of the strength of interaction between theNi2+ ion and its surroundings, is given by Dq : l/10u1. Dq is directly related to the crystal field stabili-zation energy by a factor 6/5.The Racah B parame-ter, obtained by analytically solving the Tanabe-Su-gano matrices, is

g : (ut .--2vr)(v" - vt\3(5vs - 9v1)

and is related to the "covalency" of the Ni-O bond.The Dq and B parameters are included in Table 3.Values of Dq, which range from about 850 to 900cm-r and average 885 cm-t, are close to the values fornickeloan chrysoti le, 910 cm-r (Faye, 1974),MgO:Ni'+, 860 cm-', and CaNiSirO., 840 cm-l(White et al., l97l). The plot of Dq us. M-O bonddistance constructed by Faye (1974) gives an averageNi-O bond length of 2.06+0.02A, in good agreementwith the average Mg-O bond distance in talc of2.074 (Rayner and Brown, 1973). The Racah B pa-rameter averages 950 cm-1, compared with 947 cm-1for nickeloan chrysotile (Faye, 1974) and 1030, 858,881, and 1039 crn-r for the free Ni2+ ion, MgO:Ni2+,

Fig. 3. Infrared spectra J;;?" rrained, Manchuria),stevensite (NMNH, R4710; Faust and Murata, 1953), kerol i tes (W186 and 48l .90; Brindley et al. , l9 '17 , p. 448). Al l other samples, seeTable 2

CaNiSirOu, and nickeloan clinoenstatite respectively(White et al., l97l). We find no evidence for theexistence in the natural unheated minerals of trivalentnickel or for Ni'+ in other than octahedral coordina-tion.

Thermal transformations

The phases formed by heating kerolites, nickeloankerolites, and pimelites to temperatures of about1000'C were discussed by Pham Thi Hang andBrindley (1973). They observed that kerolites gaveenstatite as the main product, and that nickel-con-taining forms gave first an enstatite product with littleolivine, but subsequently olivine with minor enstatite.

2:i

4

o

co

E

c3

F

BRINDLEY ET AL.,' KEROLITE-PIMELITE S'RI'S

---r---T_---r---t

3 2 0 0 3 8 0 0

wavenumber , cm- l

3 8 O o 3 5 0 0 3 2 O o

Fig. 4. Hydroxyl-stretching region of the infrared spectra of kerolites, pimelites and nickeloan talc. Simple descriptions given mainlyin Table 2. Additionally, sample l* is a kerolite, WI86, described by Brindley et al. (1977, p. zl48); samples 6* and l2* are from the sameveins as samples 6 and 12 respectively and illustrate variations within one vein.

6 s O 7 5 O 6 5 0 ? s O 6 5 0 7 5 o

Wavenumber , cm- l

Fig. 5. Hydroxyl-libration region of the infrared spectra ofkerolites, pimelites, and nickeloan talc. For sample designations,see F ie .4 .

The influence of the initial structure as distinct fromthe total composition was uncertain because the ini-tial samples appeared to have considerable propor-tions of a serpentine component.

We have examined the phases formed by heatingsmall samples in platinum thimbles for periods of 3-4hr at temperatures of710,855,960, 1080 and 1230"C.Figure 7 shows schematically the results obtained forfour samples: (i) kerolite from Wiry Mine, L. Silesia,given by Antoni Wala, Cracow, Poland, which isquite similar to sample (l) in Table 2; (ii) nickeloankerolite, sample (3) of Table 2; (ii i) and (iv) pimelites,samples (17) and (18) of Table 2. Each section ofFigure 7 shows the thermogravimetric weight-losscurve and the amounts of hydroxyl water, marked2(OH), and additional water. The stippled areasshow the variation of the basal reflected intensities asthe minerals are heated; evidently little change in theoriginal structure occurs until dehydroxylation be-gins around 700-800"C. Enstatite, marked E, is thefirst product phase in all cases, together with a poorlydefined phase marked FCC which may be face-cen-tered cubic (see later). In agreement with the earlierwork, enstatite remains as the final phase formed bykerolites, but pimelites develop an olivine-type phasenear l000oC, marked F in Figure 7. Nickeloan kero-

ITEcg

12'

6

N i -a l c

400 700 t000 1300W a v e l e n g t h , n m

Fig. 6. Visible absorption spectra of nickeloan kerolites andnickeloan talc. For sample designations, see Fig. 4.

lites retain both the enstatite (E) and forsterite (F)phases at 1200"C. Cristobalite (marked Cr) also isobserved.

These results agree broadly with the equilibriumdiagram of the MgO-NiO-SiO, system at l400oCgiven by Campbell and Roeder (1968). Our presentwork, however, is concerned more particularly withthe sucession of phases when the minerals are heatedthan with the ultimate equilibrium. Two results arenoteworthy. In the first place, enstatite develops firstfrom the talc-like initial structures irrespective of therelative proportions of Mg and Ni. This result reflectsthe fact that talc transforms topotactically to ensta-tite when dehydroxylation occurs (Nakahira, 1964).The lattice parameters of the enstatite are related verysimply to those of talc, so that this transformationpresumably takes place easily. The forsterite phaserequires considerably more structural reorganization;it does not develop from low-nickel kerolites but is

BRINDLEY ET AL.: KEROLITE-PIMELITE StR/tS

. . ( b )v l ( n m l -

v2 (nn)

!3 (nm)

- lD q ( c m

- )

- lB ( c n

- )

z N10

obtained from high-nickel kerolites and pimelites,The appearance of this phase in nickel-containingsamples is due presumably to the instability of thenickel analog of enstatite.

A second interesting feature is the transitional de-velopment of the FCC phase. Figure 8 reproducesdiffractometer patterns of pimelite sample 18 ofTable 2 after various heat treatments. The broadFCC reflections, respectively (lll), (200) and (220),are sketched by the dashed line. In the earlier study ofthe lizardite-nepouite series, no enstatite phase wasrecorded and an FCC phase was seen immediatelyprior to the formation of olivine. The FCC phase isless obvious in the present study, but there is littledoubt of its formation. The broad reflections indicatevery small crystals. In the temperature range 800o-1000'C, the initial green color of the mineralschanges to black, and the higher the nickel content,the deeper is the black color. Beyond I l00oC, thegreen color is restored due to Ni2+ in the olivine. Theblack color and the FCC phase could well be due tothe formation of a defective NiO. The estimated lat-tice parameter, 4.2-4.34, can be compared with4.178A for NiO and 4.213 for MgO; mixed oxides,(Mg,Ni)O, have intermediate parameters.

AcknowledgmentsWe thank those who have made mineral samples available for

this study, Dr. Z. Maksimovic, Mr. Antoni Wala and Dr. B.Ostrowicki, Dr. J. Ulrych, Mlle S Cail ldre, Dr. J. P. Golightly andthe International Nickel Company of Canada, Mr. J. T. Cumber-lidge and the Hanna Mining Company, and the Smithsonian Insti-tution. The work forms part of a program supported by NSF grantEAR74-02821. One of us (H.M.W.) thanks Mr. T. T. Feng, Direc-tor, Mining Research and Service Organization, Taipei, Taiwan,for leave of absence to participate in this work.

Table 3. Visible absorptions and assignments for nickeloankerolites and nickeloan talc

sanples (aJ

Ni-talc Assignment

0)o

oo

oo

1131 1117 1138

725 703 7r7

660 648 6s8

442 445 451

392

898

890

28.2

385 383 370

884 895 878

94I 939 1025

r 5 . 4 1 s . 9 1 8 . 4

1 1 1 3

654

3or, {")*3rra {r)

1s (o )' I (F )

l r r r {o ) ,1or " {a )

'T - (P)a g

(a) Sanp le nwbers 6 , I and 10 re fe r to those in Tab le 3

(b) Wave lengths are accura te to + 10 nD.

624 BRINDLEY ET AL.: KEROLITE-PIMELITE S'RIES

200 400 600 800 1000 1200 T, oc 20o 400 600 800 1000 1200

Fig. 7. Thermal transformation data for (a) kerolite, WI86, see caption to Fig. 4; (b) nickeloan kerolite, sample (3)' Table 2; (c) and(d) pimelites, samples (17), (18), Table 2. Stippled areas indicate persistence of initial minerals. Phases formed: E : enstatite, F : olivine-

type phase, Cr : cristobalite, FCC : face-centered cubic phase. Each diagram shows thermogravimetric weight-loss curve. Arrows

indicate hydroxyl water loss and molecular water loss.

- 2

oo

x

q)

od

I

'

ooo

d

odo3

0 .85 B^r 'ItI

2 (oH)

II

I

I

II

\;Fcd

Sltl' \ l

o tI

o - -9

( a )

( b )

( c )

(d )

Et ^ f

Il E

E O

Ii4,

Il r

l ooo l

FCC

1fl.I

I

I' , , -__- l - - " '

I_ - ) .

I

q"i"llc "

2e, deg,reegFig. 8. Diffractometer patterns of pimelite, sample (18) of Table 2,CvKa radiation. (a) Room temperature; (b)-(f) after 4 hrs heat

treatment at 710o, 855", 960', 1080', 1230'C. Difraction peaks labelled as follows: Q = quartz, E = enstatitp, O : olivine-tyPe phase,

FCC = face-centered cubic ohase.

References

Berkes, J. S. (1968) Energy leuel diagrams fdr transition metal ionsin cubic crystal fields. MRL Monograph No. 2. Materials Re-search Laboratory, The Pennsylvania State University.

Bish, D. L. and G. W. Brindley (1978) Deweylites, mixtures ofpoorly crystalline hydrous serpentine and talc-like minerals.M ineral. Mag., 42, 75-80.

Brindley, G. W., David L. Bish and H.-M. Wan (1977) The natureof kerolite, its relation to talc and stevensite. Mineral. Mag., 41 ,443-452.

- and G. Ertem (1971) Preparation and solvation propertiesof some variable charge montmorillonites. Clays Clay Minerals,t9, 399-404.

- and Z. Maksimovic (1974) The nature and nomenclature ofhydrous nickel-containing silicates. C lay M ine ral s, I 0, 27 | -2'l 7 .

- and Pham Thi Hang (1973) The nature of garnierites-LStructures, chemical compositions, and color characteristics.Clays Clay Minerals, 2l, n-40.

- and H.-M. Wan (1975) Compositions, structures and ther-mal behavior of nickel-containing minerals in the lizardite-ne-pouite series. A m. Mineral., 60, 863-87 1.

Campbell, F. E and P. Roeder (1968) The stability of olivine andpyroxene in the Ni-Mg-Si-O system. Am. Mineral., 53, 257-268.

DeWaal, S. A (1970) Nickel minerals from Barberton, SourhAfrica: III. Willemseite, a nickel-rich talc. Am. Mineral.. 55.3l-42 .

Faye, G. H. (1974) Optical absorption spectrum of Nir+ in garnier-ite: a discussion. Can Mineral., 12, 389-393.

Giese, R. F. (1975) Interlayer bonding in talc and pyrophyl l i te.Clays Clay Minerals, 23, 165-166.

Maksimovic, Z. (1966) B-Kerolite-pimelite series from GolesMountain, Yugoslavia. Proc Int. Clay Conf., Jerusalem 1,9'l-105.

Medlin, J. H., N. H. Suhr, and J. B. Bodkin (1964) Atomicabsorption analysis of silicates employing LiBOrfusion. AtomicAbsorption Newsletter, 8, 25-29.

Moenke, H. H W. (1974) Silica, the three-dimensional silicates,borosilicates and beryllium silicates. In V. C. Farmer, Ed., TheInfrared Spectra of Minerals, p. 365-382. Mineralogical Society,London.

625

Nakahira, M. (1964) Thermal transformations of pyrophyllite andtalc as revealed by X-ray and electron diffraction studies. C/aysClay Minerals, 1 2, 2l-27.

Ostrowicki, B. (1965) Mineraly niklu strefy wietrzenia serpentynitow w Szklarach [Nickel minerals of the weathering zone ofserpentinites at Szklary (Lower Silesia)). Polska Akad. Nauk,Prace Mineral.,l, 7-87 (with English summary, 88-92).

Pecora, W. T., S. W. Hobbs, and K. J. Murata (1949) Variat ions ingarnierite from the nickel deposit near Riddle, Oregon. Econ.Geol.,44, 13-23.

Pham Thi Hang and G. W. Brindley (1973) The nature of garnier-ites-IIL Thermal transformation s. C lavs C la v M ine rals. 2 I . 5 | -

Ruyn"r, J. H. and G. Brown (1973) The crystal structure of talc.Clays Clay Minerals, 21, 103-114.

Scholze, H. (1960a) Zur Frage der Unterscheidung zwischen HrOMolekulen und OH-Gruppen in Gliisern und Mineralen. Natnr-wissenschaften, 47, 226-227 .

- (1960b) Eine weitere Unterscheidungsmdglichkeit ob inGlZisern HrO-Molekiile oder OH-Gruppen vorliegen. Glastech.B e r . . 3 3 . 3 3 .

Slansky, E. (1955) The Ni-hydrosilicates from Kremze in SouthBohemia. (Czech text, English abstract). Unio. Carol., Geol. l,r-28.

Stubican, V. and R. Roy (1961) A new approach to assignment ofinfrared absorption in layer-structure silicates. Z. Kristallogr.,I t s, 240-214.

Vedder, W. (1964) Correlations between infrared spectrum andchemicaf composition of mica. Am. Mineral.,49, 736-768.

Warren, B. E. and P. Bodenstein (1965) The diffraction pattern offine particle carbon blacks. Acta Crystallogr. 18, 282-286.

White, W. B., G. J. McCarthy, and B. E. Scheetz (1971) Opticalspectra of chromium, nickel, and cobalt-containing pyroxenes.A m. Mineral., 56, 72-89.

Wilkins, R. W. T. and J. I to (1967) Infrared spectra of somesynthetic talcs. Am. Mineral., 52, 1649-1661.

Wood, B. J. (1974) Crystal field spectrum of Ni'z+ in olivine. Am.Minerat.. 59. 244-248.

Manuscript receiued, July 3, 1978;accepted for publication, October 9, 1978

BRINDLEY ET AL.: KEROLITE-PIMELITE S'RIES