ClayMinerals (1969) 8, 59.

T H E S E P A R A T I O N O F C L A Y M I N E R A L

F R A C T I O N S W I T H L I N E A R H E A V Y L I Q U I D

D E N S I T Y G R A D I E N T C O L U M N S

G. H A L M A

Department of Regional Soil Science, Agricultural University, Wage~lingen, Holland

(Received 14 October 1968)

ABSTRACT: Because of increased interest in mineralogical analysis of soils a rapid, generally applicable method to separate clay minerals is needed, and such a method is described here. The technique is a modification of the heavy liquid density gradient centrifugation method which is commonly applied in mineralogical practice. Modifica- tions are:

(1) The use of suitable surface active agents to overcome the flocculation problem. (2) The use of the ultra centrifuge at about 33,000 g to hasten sedimentation. (3) The use of a linear density gradient which supplies in addition a quick identi-

fication of the mineral composition. Preliminary experiments, evaluating different variables (e.g. cation form, ultra-

sonic treatment, influence of surfactant), reveal the scope and limitations of the procedure presented.

Clay minerals of different density can be separated. However, if the clay minerals of a mixture have overlapping densities, or if they contain a series of mixed-layer minerals, only a broad fractionation is possible. This is shown by X-ray diffraction and electron microscope studies of isolated zones (or parts of them) which develop in the centrifuged density gradient columns containing various pre-treated natural clays or clay mixtures.

I N T R O D U C T I O N

Soil scientists are interested in the characterization and genesis of soils, in particular in the physical (mineralogical) and chemical nature and properties of the mineral components. Once the properties of the separate minerals are known, it is possible to predict the properties of their mixtures, and possibly certain aspects of the soil as a whole.

In soils a big role is played by the minerals present in the clay fraction (smaller than 2 ~). Clay minerals affect soil fertility, mobility of ions and humus and permeability of soils. Furthermore, as day minerals are frequently the product of rock weathering, it is of interest to study the course of these transformations.

A considerable difficulty in the study of the clay minerals is the lack of adequate

60 G. H a l m a

data about them. This is due to the lack of an easy method for separating and isolating pure clay minerals so that they may be studied by modern techniques.

This subject has been studied extensively, and in the thirties, some investigators used selective flocculation in an attempt to solve the problem. This was unsuccessful, because the flocculation properties of clay minerals are far from specific.

McNeal & Young (1963) experimented with a combined paper chromatography- curtain electrophoresis technique. From a mixture of bentonite, kaolinite and vermiculite only bentonite could be isolated successfully during 2 to 3 days of operation. Recently, an apparatus manufactured by Beckman, called the 'continuous particle electrophoresis system' was claimed to be able to perform promising separations of clay minerals, this apparatus was not tested in the present investigation.

Peters & Wiithrich (1963) worked with dilute suspensions of clays which passed between the poles of a powerful magnet. The separation however was insufficient, and the method time consuming and requiring expensive apparatus.

Bush et al. (1966) subjected differently swelling clays to a silicone treatment, after which dilute suspensions were centrifuged in an organic liquid immiscible with water. The authors claimed success in separating different swelling clay minerals from each other. Finally mention should be made of the work of Gibbs (1967). This author isolated 90-95% montmorillonite and reasonably pure kaolinite, chlorite and mica from specified particle-size fractions by repeated centrifugation at 1600 g of dilute suspensions of natural clays in 15% ethanol-water mixtures or various thallium formate solutions.

All these methods are either unsatisfactory or rather time consuming because of the necessity of handling the sample repeatedly. The object of this study was to obtain a satisfactory separation in one generally applicable, easy and rapid process, giving, in addition, a basic idea of the semi-quantitative mineral composition of the clay under investigation. It is the author's opinion that methods using the electrophoretic or density difference principle should be able to make possible successful separation of clay mineral mixtures and also of natural clays (assuming these to be physical mixtures of clay minerals). It was decided to develop the density difference method using heavy liquid density gradient columns (H.L.D.G.C.) and centrifugation because of its established usefulness in mineralogical practice. This method has the following advantageous features:

(1) It is sensitive: density differences as small as 10 -7 g/ml can be detected when proper precautions are taken (Oster & Yamamoto, 1963).

(2) A quick identification ('finger print') of the mineralogical composition of the clay is obtained.

(3) Samples of different densities are separated and can easily be isolated. (4) It is simple, inexpensive and applicable in every standardly equipped

laboratory. (5) It is independent of particle size. (6) One standard operationaI procedure can be applied for any sample. However, the following three modifications were made to ordinarily practised

Separation of clay mhwral fractions 61 H.L.D.G.C.-centrifugations: firstly the clay suspensions were stabilized with sur- factants to minimize flocculation and aggregation. Secondly ultra-centrifugation was used with speeds up to 18,000 rev/min ( ~ 33,000 g) and thirdly the density gradient columns used were strictly linear.

T H E F L O C C U L A T I O N / A G G R E G A T I O N P R O B L E M

In aqueous media, particles smaller than about 20 ~ tend to agglomerate owing to surface forces. In oil media this tendency is even more pronounced owing to the low dielectric constant of oils (in surface chemistry every non-water-miscible organic liquid is called an 'oil'). This agglomeration (Van Olphen, 1963 distinguishes between flocculation and aggregation--see page 93 et seq.) creates great problems, although some authors do not report any difficulties. Thus, Bush et al. (1966) eliminated agglomeration by violently stirring their suspensions. However, Loughnan (1957) reported that separations were incomplete due to agglomeration. Muller & Burton (1965) discuss the 'use of specific surfactants to inhibit flocculation' in oil media, but no detailed suggestions are given.

There are many references (e.g. in Van Olphen, 1963) indicating the general type of surfactant to be used to stabilize (disperse) a negatively charged alumino-silicate colloid of hydrophilic nature such as clay. In aqueous media an intermediate hydrophilic lipophilic anionic (that is negatively charged) surfactant should be used, whereas in oily media a hydrophilic cationic (positively charged) surfactant is recommended. Thus, the well known quaternaries are excellent 'oleophylizers'. They contain a hydrophylic cationic quarternary ammonium group and one or more aliphatic 'tails' of capryl (C8-) to stearyl (C18-) length.

With these general directions one must still search for the right surfactant for any given medium-colloid complex. This cannot be chosen from theoretical con- siderations aIone and it is not surprising that no details are availabIe on the dis- persion of clayey matter in the oils used in this investigation. Consequently it was necessary to search for an appropriate suffactant by tedious trial-and-error with the available surfactants and various combinations of them, on the clay suspensions in the oils used.

The tests were made by microscopic observation and according to the volume of sediment or flotate accumulating in a fixed time. Occasionally ultrasonic treat- ment at 20 KHz was applied. The intention was to tear apart the particles building up an agglomerate, envelop their surfaces completely with adsorbed surfactant molecules so that they would be shielded and could withstand reagglomeration. Unfortunately the frequency of 20 KHz ruptured the individual clay particles. This resulted in a greater tendency to flocculate, which on standing overcompensated for the better stabilization which resulted from the shielding effect of the surfactant.

The experiments performed with some 50 surfactants of various types indi- cated that lauryl-trimethyl-ammonium bromide (C12H,,~N(CH3)~+Br -) and lauryI pyridinium bromide (C12H25CsH,N+Br -) gave the best deflocculation performance, although all tested surfactants of the alkyl-ammonium- and alkyl-pyridinium type

62 G. Halma

(C8-C18) were satisfactory. Flocculation appeared to be eliminated, i.e. the loose cardhouse structure of the clay particles was broken up, as could be observed macroscopically. The addition of a trace (say 0"1%) of the surfactant to a creamy, high viscosity suspension, turned it to a low viscosity one. However, after standing about half an hour tiny mini-flocs could be observed using shearing illumination. These disappeared on agitation, but reappeared on standing. This could be due to either small flocs building up when undisturbed, or remaining aggregates becoming visible with time. High surfactant concentrations could not eliminate these tiny flocs.

Nevertheless most of the experiments were done using surfactants with this per- formance, and C.T.A.B. (cetyl-trimethyl ammonium bromide, C16HazN(CH~)~+Br -) was chosen because it was available in a pure grade (it is used as a bactericide). Moreover it was thought that the flocs might break up under the large g-force applied during ultra centrifugation; and this indeed, was observed. The gradient columns containing clay-zones showed flocs after standing for a few hours, but never, however, immediately after spinning.

Recently a surfactant was obtained called DIAM-11C (N-oleyl-l,3-propylene diamine). This surfactant gave very good dispersion of the clays. Sedimentation-times were much larger (2 months) as compared with C.T.A.B. (a week).

Even better results were obtained with equal amounts of DIAM-11C and calcium dodecyl benzene sulphonate (C.D.B.S.). It was thought that the latter, being anionic and lipophilic in character, would be absorbed on the positively charged edges of the clay platelets, thus shielding this part of the crystal, whereas DIAM-11C, possessing two powerful adsorptive electro-positive head amino groups (like a t aw) could build a ring with negative lattice oxygen atoms and acquire a firm grip on the lattice. The cis-formed oleyl 'tails' of this surfactant could shield remaining 'inactive' parts of the flat clay surface. Experiments to let the role of C.D.B.S. be taken by DIAM-11C itself or by the ordinary quaternaries using phosphate as a cement were unsuccessful.

P R O C E D U R E S A N D M A T E R I A L S

Preliminary investigations were made using kaolinite from Zettlitz (near/y pure kaolinite), bentonite from clayspur (nearly pure montmorillonite) and i11ite from Winsum (illitic material of marine origin containing considerable amounts of kaolinite and montmorillonite).

These clays were pretreated according to a procedure outlined by Jackson (1956) and modified by Favejee (1966), to remove organic matter, calcium carbonate and ferruginous material. The clay fractions (-(2 /~ e.s.d.) were isolated by gravity settling in water under slightly alkaline conditions. Part of these H+-saturated clays were transformed into other ion forms by treating them with 1 N solutions of the chlorides.

The clays were 'oleophilized' by washing aqueous suspensions three times with absolute ethanol, and then twice with bromoform. This gave severely flocculated

Separation of clay mineral fractions 63

slurries of creamy consistency. Linear density gradient columns were prepared by a new method (Halma, i969). Heavy liquids used were bromoform, decatine, tetra-chloromethane (purified grades from Brocades, The Netherlands) and di-n butyl phtalate (technical grade from Fluka, Switzerland). Surfactants were dissolved in the heavy liquids prior to the preparation of the gradients (concentrations were 2% w/v). The 'oleophifized' clay suspension containing 2% w/v surfactant was occasionally subjected to ultrasonic treatment for 10 seconds using Branson's sonifier S 75, prior to loading.

The loading was performed either by placing a small layer of the sample on top of the ready-made density gradient using a spatula with a bent tip (top loading), or the clay sample was added to the heavy liquids prior to the preparation of the gradient (bulk loading). Bttlk loading allows about 200 mg of dry clay to be pro- eessed in 30 ml of gradient, whereas in top loading only about 40 mg can be treated without upsetting the gradient.

The gradients were collected in 30 ml celIulose nitrate tubes 2"5 cm wide and 7"5 cm long. These tubes fit into closeable aluminium containers three of which can be fitted into the massive stainless steel rotor (SW 25) of a 'Spinco' ultra- centrifuge. Spinning was performed under vacuum at a speed of 18,000 rev/min (=33,000 g at average tube height). For safety reasons, this speed was not exceeded, although speeds of up to 50,000 rev/min (165,000 g) are applied in biochemical work with sugar gradients. After 1, 2, 4, 8, I6 and 32 or 1, 10, 24 hr of spinning the contents of the tubes were inspected and/or photographed using spot-fight illumination from above and a dark background. The separated zones or samples from them were taken, using a screw operated syringe fitted with a needle the tip of which was bent at 90 ~ Thus a specified layer could be selectively withdrawn. Experiments were made with freezing the column and then cutting it into pieces. The freezing however disturbed the gradient too much. The samples collected were washed twice with acetone or absolute ethanol (in the case of C.T.A.B. and DIAM-11C + C.D.B.S. respectively) to free them from surfactant, and smear oriented slides were prepared. These slides were used for X-ray investigations using a Philips diffractometer PW 1050 provided with a cobalt tube. Relative intensities of the kaolinite, illite and montmorillonite peaks at 20 = 14.35, 10.26 and 7"00 respectively, were compared. Occasionally samples were washed twice with water and transferred to electron microscope grids to obtain E.M.-photographs.

The separation efficiency was correlated with: (1) spinning time, (2) adsorbed ion H+(AI~+), Na +, K +, Mg 2+, Ca 2+ and Ba ~+, (3) ultrasonic treatment at 20 KHz, (4) loading conditions, (5) type and concentration of suffactant, and (6) nature of heavy liquid.

R E S U L T S AND D I S C U S S I O N

Some results of these preliminary experiments are shown in Plates l(a)-(h), and 2(a)-(h) and in Fig. 1, which give representative examples of a large number of pictures obtained. Plates l(a)-(d) show the influence of spinning time. Each figure shows three tubes supported by a holder and illuminated from above. The tubes

64 G. Halma

are filled with a linear density gradient built up from di-n-butyl-phtalate bromoform mixtures containing C.T.A.B. to yield to a density range f rom 1.825 to 2"791 g/ml . The tubes were bulk loaded (initially homogeneously filled) with H+-clay, the left hand one with a mixture of kaolinite (K) and bentonite (B), the middle one with illite (I), and the right hand one with a mixture of these (~ = K + B + I ) . Plates l(a)-(c) were taken after 1, 4 and 32 hr of spinning respectively. Plate l(d) shows a phenomenon often observed after leaving the centrifuged contents to stand un- disturbed for a few days (3 days in this case); macro flocculation has become obvious and peculiar equidistant bands are developed. This phenomenon may be attributed to stabilized turbulences.

The contents of the tubes shown in Plates l(a)-(c) are split up into zones indi- cating that there is some fracfionation. K + B separates into a montmorillonitic layer at 5"2 cm and a kaolinitic zone at 2 cm from the bot tom (the meniscus and bottom are often lit because of their lens action owing to the rounded shape, and sometimes because of contaminants). In all cases montmorillonite separated within 1 hr of spinning, whereas illite required 4--6 hr, and kaolinite was best developed only after 10-30 hr of centrifugation. Thus the mixture ~ requires at least 10 hr to be fractionated. One then observes rather sharp kaolinite and montmorillonite bands, the latter often being obscured by the broad, diffuse central illitic zone, containing varying proportions of kaolinite and montmoriUonite. Illite contains separable kaolinite and montmorillonite as is shown by the middle tube zoning. On close inspection the central illitic zone itself is faintly zoned with diffuse optically denser and less dense regions. Unfortunately this is difficult to show photographically.

Thus K + B separate quickly and well and ~ separates satisfactorily. Only the central illitic zone is contaminated and poorly resolved probably containing a continuous series of physical mixtures or interstratified illites. In retrospect this illite appears to have been a poor choice but the experiments were continued with this sample because no purer samples were available. Remarkably, it was found that in some cases the best separation occurred within 1 hr of centrifugation.

The influence of the exchangeable ion is seen on the mixture (~) in Plates l(e) and (1) and illite (I) in Plates l(g) and (h). The conditions were the same as in the former experiment except that the spinning time was 10 hr. No detectable influence of the exchangeable ion was observed, except for a better resolution of montmoril- lonite with the Ca 2+ and Mg 2+ clays. Diffractograms were made of the isolated zones from tubes EH + and I H +. They are shown in Figs l(a) (~) and l(b) (I)

FIG. 1. Diffractograms of clay fractions isolated from zones as indicated. (a) H + mixture of clays (Plate l(e) left hand tube). (b) H + illite (Plate l(h) left hand tube). (c) Top series. Ca 2+ illitic Swiss clay (Plate 2(a) central tube). (c) Bottom pair. Shows the difference between the K + and Ca 2+ forms of these clays when

untreated (cf. Plate 2(a) left and middle tubes). (a t) H + form of a mixture of kaolinite and bentonite after the mixture had and had

not been ultrasonically pretreated (Plate 2(b) central and left hand tube respectively).

I I I i I I i I I , 2U=15 1~ 7 315 Ill 7 3 ~15 1117 3 , 5 111 7 3115 Tll 7 3 t5 I I I 11 7 3

/!y// 3~

t 16 - 3"0

~ ' 6 - 6 "2 0 - 1 . 5

2U= 15 7 3 I 1 7 3 15 7 3 I 1 7 3 15 11 7 3 15 11 7 3 15 7 3

/

L l t O- 0-8

J 5 ' ' ' ' 1'1 ' , . . . . I , 1 , , ~ I ; , I 5 ,1 I , , 111 , , 21)=1 11 7 3 15 7 3 15 11 7 3 15 7 3 1 11 3 ! 7 3 15 7 3

(c) [d) H . cm K Clay Ca Clay 4.~ Ultrasonically treated Unfreot:d ~ j~j ~ j ~ Untreated

1-0-1,7 3,8 - 4.n~j

kd . . j f I I I I , I I l I I I I I I | I J t l I I

2V=15 1 1 7 315 1 1 7 315 7 315 11 7 315 7 315 II 7 315 11 7 3

FIG. 1

6 6 G. Halma

respec t ive ly . T h e per fec t s e p a r a t i o n of k a o l i n i t e and m o n t m o r i l l o n i t e in the e x t r e m e

zones of the m i x t u r e is eas i ly seen (zones 0 -2-0 and 4-9-5-2) bu t the i n t e r m e d i a t e

zones a re c o n t a m i n a t e d to a ce r t a in extent . A r e a s o n a b l y g o o d f r a c t i o n a t i o n is

i nd ica t ed by the d i s t r i bu t i on o f the th ree c lay minera l s . I n the tubes c o n t a i n i n g

i l l i te (Fig. l (b)) the h igher zones c o n t a i n m o r e m o n t m o r i l l o n i t e and less kao l in i t e ,

a n d in the l o w e r zones this is reversed . A t the ve ry b o t t o m , h o w e v e r , p u r e i l l i te is

found .

L E G E N D S TO P L A T E S

PLAIE 1 (a) Tubes containing linear density gradients (1-825 to 2-791) showing clay fraction

zoning of kaolonite + bentonite (left), illite (middle) and a mixture of all three (right) after centrifugation for 1 hr at 18,000 rev/min. Surfactant is 2% C.T.A.B.

(b) Same as in (a) except centrifugation time was 4 hr. (c) Same as in (a) except centrifugation time was 32 he. (d) Same as in (c) except the tubes had been allowed to stand for 3 days. (e) Tubes containing the same density gradients as above, loaded with a mixture of kao-

linite, bentonite and illite, showing the influence of exchangeable cation. Left: H--clay; middle: K ~-clay; right: Ca2 +-clay. Spinning was for 10hr at 18,000 rev/min.

( f ) Same as in (e) except exchangeable cations were left: Na +, middle: Mg2+ and right Ba 2 ~.

(g) Same as in (e) except the clay was only illite. (h) Same as in ( f ) except the clay was only illite.

PLATE 2 (a) Tubes containing linear density gradients (1.825 to 2.791) showing zoning of an

illitic Swiss clay saturated with K + (left) and Ca 2+ (middle) after centrifugation for 10 hr at 18,000 rev/min. Surfactant: 2% of C.T.A.B.

(b) Tubes containing linear density gradients (1-592 to 2"805) showing the separation of an artificial mixture of kaolinite and bentonite (in H +-form) after centrifugation for 10 hr at 18,000 rev/min, pretreated as follows: left: top loaded and ultra- sonically treated; middle: top loaded and and not ultrasonically treated; right: suspension loaded and not ultrasonically treated. Surfactant: 1% D]AM-11C + 1 ,% C.D.B.S.

(c) Electron micrograph of H*-bentonite ultrasonically treated ; magnification x 4305. (d) Electron micrograph of H~-bentonite, not ultrasonically treated; magnification

x 4305. (e) Electron mierograph of clay from the top layer of the kaolinite + bentonite mixture

of (b) central tube (about 4 cm from the bottom). Magnification x 4305. ( f ) Ditto, except bottom layer of (b) right hand tube .(at about 1-5 cm from the

bottom); magnification x 4305. (g) Tubes containing linear density gradients (1.847 to 2.819) containing H~-illite

spun for 1"5 hr at 18,000 rev/min showing the influence of surfactant and of ultrasonic treatment: left: 1% DIAM-I1C + 1% C.D.B.S., not ultrasonically treated. Middle: ditto but ultrasonically treated. Right: 2 ~ C.T.A.B. and not ultrasonically treated.

(h) Tubes containing kinked density gradients (1.592 to 2"805) (left and central) and from 1.903 to 2.294 (right). They are loaded with illite (left and right) and an artificial mixture of kaolinite, bentonite and illite (middle).

Contents were ultrasonically treated and dispersed with I ~ DIAM-11C + i % C.D.B.S. Left and middle tubes were stirred between 3 and 5 cm from the bottom to give an expanded gradient in that region.

PLATE 1

(Facing page 66)

PLATE 2

Separation of clay mineral fractions 67

Plate 2(a) shows a photograph of a tube containing an illitic clay from Switzer- /and, coded 964 I. Here the K + and Ca 2+ forms are quite different from one another. A diffractogram of the untreated clay (Fig. l(c)) confirms this observation.

�9 Diffractograms of isolated zones of the Ca 2+ form (Fig. l(c) top series) again show fractionation to have occurred. This is not reflected by the appearance of zones in the photograph, probably because they overlap continuously.

Loading conditions and ultrasonic treatment were investigated in the next experiment. K + B was processed in a linear gradient of CCI~ and CHBr3 giving a density range of 1"592 to 2"805. The surfactant added was 1% of DIAM-11C + 1% of C.D.B.S. The results are shown in Plate 2(b). The left hand tube was toploaded and its contents were ultrasonically treated. The middle and right hand tubes were toploaded and bulk-loaded respectively. They were not ultrasonically treated.

Ultrasonic treatment results in broadening of the zones and slightly poorer separation. Seemingly, the disrupted particles are more adsorptive to each other and more resistant to the separating g-force. The effects of U.S. treatment are clearly demonstrated by electron microscope photographs of bentonite treated ultrasonically (Plate 2(c)) and untreated (Plate 2(d)).

Loading methods do not basically influence the picture although an effect related to quantity is observed in Plate 2(b) where the zones are thicker in the right hand tube. Overloading must be avoided in this technique, because it gives rise to broadened and overlapping bands (Britten & Roberts, 1960).

Plate 2(b) demonstrates a perfect separation of a montmorillonite layer at about 4 cm and a kaolinite layer at about 1'5 crn from the bottom of the tube. This is also shown by the diffractograms (Fig. l(d)) which demonstrate the slightly poorer separation of the U.S. treated sample. Plates 2(e) and (I) conclusively show the perfect separation electron microscopically. Plate 2(e) shows only montrnorillonite in the top layer of the central tube of Plate 2(b), whereas Plate 2(I3 shows only kaolinite in the lower zone of Plate 2(b).

Plate 2(g) illustrates the influence of surfactant as well as U.S. treatment. The gradients were prepared from bromoform-decaline mixtures to give a density range of 1-847 to 2"819. They were bulk loaded with illite. The contents of the central tube were ultrasonically treated. The suspension in the right hand tube was dispersed with C.T.A.B. and the others with DIAM-11C+C.D.B.S. Again U.S. treatment resulted in more broadened, overlapping layers. DIAM-11C+C.D.B.S. yielded better dispersion, manifested by the extended, split up and finer textured layer (cf. right and left hand tubes). In addition, on standing for periods up to some weeks, no coarsening (due to flocculation) was observed, as in the case of C.T.A.B.

Plate 2(h) shows an experiment using a density gradient with an extended central region. The left hand tube contains illite, and the central one the mixture K + B + I in a gradient prepared from CCI~ and bromoform (range 1-592 to 2"805) the central region of which was mixed to give a kinked over-all height-density relationship. This was achieved by stirring the central region with a bent spatula in up and down movements between 3 and 5 cm from the bottom of the tube. A certain fraction of the clay can be trapped in this central extended zone, excluding

68 G. Halma

other fractions. The photographs show such a trapped illite zone, the excluded zone above the trapped one being separated from it by a clayless zone.

A still better definition of zones, applicable to clay minerals of continuously varying density, can be obtained by the use of staircase-shaped gradients. Certain fractions can be made to occupy predetermined zones by this technique. The right hand tube of Plate 2(h) contained a gradient ranging from 1-903 to 2-294 and was bulk loaded with illite to show the effect of an extension of the central part of the illitic zone. At least six diffuse sub-zones can be distinguished. The contents of all tubes in Plate 2(h) were uItrasonically treated, spun for 9"5 hr and treated with DIAM-11C + C.D.B.S.

C O N C L U S I O N S

The technique outlined here enables the separation of physical mixtures of clay minerals provided their densities do not overlap one another.* The zones developed in the density gradient allow a quick identification of the mineral composition to be made. Interstratified clay minerals, and minerals with a continuous, overlapping density range do not yield separate zones. They can only be fractionated by taking samples from different positions in the zone. A kinked, or staircase-shaped gradient may create artificial zoning and facilitate isolation of samples.

Ultrasonic treatment is probably able to loosen weakly-bound aggregates of natural clay particles. The frequency however is important and should probably be in the order of 100 KHz rather than the 20 KHz used in these experiments.

In order to obtain reasonable amounts of pure minerals most efficiently, bulk loading of the gradient with the sample is recommended. Overloading should be avoided. For identification purposes, toploading with very small quantities of the sample gives best results.

Spinning time depends on individual circumstances. Generally 10 hr of spinning at 18,000 rev/min is sufficient to obtain optimum separation.

In some instances the type of adsorbed ion exerts a pronounced influence. Ca ~+ and Mg 2+ forms of the tested clays behave slightly differently from the other forms investigated. K + clays also may give differing results. Saturation with a heavy cation (e.g. Th 4+) may aid separation.

The choice of the best surfactant will depend both on the clay mixture to be separated and on the liquids used in the gradient. Its chemical structure is of utmost importance. The best results so far have been obtained with equal amounts of DIAM-11C and C.D.B.S. in a 2% w/v total concentration but this is currently being investigated in more detail. It is possible that phosphorus or silicon contain- ing organic surfactants may show better performance. The choice of heavy liquids is probably of less importance as long as the oils used are chemically compatible. The possibility of making the density gradients with aqueous heavy liquids, such as clerici solution, should also be borne in mind.

* It is to be expected that the method will also be of use in the separation of mineral~ in the 2-50 # range. In this case ordinary low speed centrifugation will be sufficient, but surfactants must still be used.

Separation of clay mineral fractions 69 A C K N O W L E D G M E N T S

The author wishes to thank Dr Van der Plas and Mr Schoorl for their kind advice and Mr Henstra, Mr Van Druuten and Mr Buurman for their help with electron microscopy, photography and design.

He is even more indebted to Dr Van Kammen and Dr Peeters of the virology department for their advice and willingness to supply ultra centrifuge facilities.

R E F E R E N C E S

BRITTEN R.J. & ROBERTS R.B. (1960) Science 131, 32. Bosh D.C., JENKn~S R.E. & MCCALEB S.B. (1966) Clays Clay Miner. 14, 407. FAVEJEE J.CH.L (1966) The Identification of Detrital Feldspars, (L. Van der Plas, editor) Elsevier,

Amsterdam. GIaas R.J. (1967) Clay Mines. 7, 79. HALMA G. (1969) Clay Mines. 8, 47. JACKSON M.L. (1956) Soil Chemical Analysis Advanced Course, University of Wisconsin. LOUGHNAN F.C. (1957) Am. Miner. 49, 393. MCNEAL B.L. & YOUNG J.L. (1963) Nature, Lond. 197, 1132. MULLER L.D. & BURTON C.J. (1965) 8th Commonwealth Mining and Metallurgical Congress,

Australia and New Zealand, p. 1151. OLermN H. VAN (1963) An Introduction to Clay Colloid Chemistry Interscience, New York. OSTER G. & YAMAMOTO M. (1963) Chem. Rev. 63, 257. PETERS TJ. & W~rTHRICH H. (1963) Eclog. geol. Hetv. 56, 113.