Clays and Clay Minerals, VoL 46, No. 3, 322 329, 1998.

WATER RETENTION BY COLLOIDAL A L L O P H A N E A N D IMOGOLITE WITH DIFFERENT CHARGES

JUTARO KARUBE AND YUKIHIRO ABE

Faculty of Agriculture, Ibaraki University, Ami-machi, Ibaraki-ken, 300-0332 Japan

Abst rac t - -Water retention curves of colloidal allophane and imogolite with different charges and different pretreatments were measured using a tension plate and a pressure plate apparatus. An irreversible effect of air-drying was found for the water retention of colloidal allophane even for particles of less than 50 nm in Stokes" diameter collected from air-dried soil. This was attributed to the irreversible submicroscopic aggregation of allophane. Fresh allophane colloids with a low absolute net charge retained more water above - 1 0 0 J K g -t than did highly charged ones due to the formation of a microporous structure. Allophane did not swell under ordinary conditions, but the relatively highly charged allophane recovered water retention above - 1 0 0 J kg t during the wetting process. Imogolite retained 1.5 times more water than Na-montmorillonite at about - 6 5 0 J kg t due to micropores formed by intertwining fibrous particles.

Key Words--Allophane, Imogolite, Net Charge, pH, Pressure Potential, Water Retention.

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

The character is t ic phys ica l proper t ies o f volcanic ash soils (Andiso l s ) such as h igh hydrau l ic conduc t iv - ity, air pe rmeab i l i ty and h igh water re ten t ion are at- t r ibutable to the h igh ly porous soil s t ructure p roduced by the proper t ies of the col loidal const i tuents . Allo- phane and imogol i te are the character is t ic c lay min- erals s t rongly re la ted to these propert ies , but the i r wa- ter r e ten t ion cu rves have not been m e a s u r e d because of the diff icul ty in prepar ing e n o u g h pure clay min- erals to measu re these curves .

The wa te r r e ten t ion of a l lophanic soils decreases marked ly on a i r -drying ( M a e d a et al. 1977), the cause of w h i c h is r ega rded to be the i r revers ib le aggrega t ion of a l lophane (Kubo ta 1972). One of our purposes was to examine whe the r the col lo idal par t ic les of a l lophane separa ted f rom air-dried soil re ta in the effects of air- drying.

The surface charges of a l lophane and imogol i te can be con t ro l led by chang ing the pH condi t ions because they are typical va r iab le -charge clays (Wada 1980). This p h e n o m e n o n should be useful for inves t iga t ing the re la t ionship be tween the surface charge and water r e ten t ion of a clay. This is ano ther r eason w hy we used a l lophane and imogol i te to exam i ne the effect of the net charge on wate r re tent ion. The effects of pr ior i ron oxide r emova l on wate r re ten t ion were also examined , and a compar i son of wate r re ten t ion by a l lophane, im- ogol i te and mon tmor i l l on i t e was made.

M A T E R I A L S A N D M E T H O D S

A l l o p h a n e Sample

The sample was p repared f rom fresh soil t aken f rom the a l lophane- r i ch lower part of the K a n u m a t s u c h i wea the red pumice bed at K a n u m a City, Tochigi Pre- fecture, Japan. It was pur i f ied by the fo l lowing pro- cedure.

FRESH DEFERRATED SAMPLE. A gel- l ike subs tance was w a s h e d out of the soil wi th wate r before the soil was t reated wi th H202. The aggregates were squashed , and the coarse part icles r e m o v e d us ing a 0.5- turn sieve. The fine por t ion was t reated wi th H20: , then defer ra ted by t rea tment wi th Na2SzO4-NaHCO3-Na ci trate ( M e h r a and Jackson 1960), fo l lowed by 2% Na2CO3 at 90 ~ for 15 rnin ( Jackson 1956; modi f ied by Wada and Green l and 1970). The par t ic les were d ispersed at pH 9 - 9 . 5 wi th NaOH. Par t ic les wi th a S tokes ' d i ame te r smal ler than 50 rtm were co l lec ted by centr i fugat ion. Abou t 11313 m E of sa turated NaC1 solu t ion was added to each li ter of a l lophane col loid to be coagula ted; then the sed iment was d ia lyzed agains t de ion ized wate r un- til the electr ic conduc t iv i ty of the externa l so lu t ion was less than 1 mS/m. A n e lec t ron n l i c rograph of this sam- ple is s h o w n in F igure l a.

FRESH NON-DEFERRATED SAMPLE. This sample was pre- pared in the same way as the f resh defer ra ted sample , except tha t there was no i ron oxide r emova l or 2% Na2CO 3 t rea tment .

AIR-DRIED SAMPLE. The soil was first air-dried, the res t of the p rocedure be ing the same as for the f resh sam- ples. The air-dried sample used cons is ted of a l lophane col loid wi th a S tokes ' d i amete r smal ler than 50 nm, as in the f resh sample. Defer ra ted (Figure l b ) and non- defer ra ted samples were prepared.

Imogol i t e Sample

This sample was p repared f rom f resh soil ob ta ined f rom the imogol i t e - r i ch upper par t of the wea the red Kanuma t such i pumice bed. Af te r the gel- l ike sub- s tance had been col lec ted by wash ing the soil wi th water, it was subjec ted to H202 t rea tment , i ron oxide r emova l and 2% NazCO 3 t rea tment . Pur i f icat ion was pe r fo rmed as fol lows: The pH of the di lute suspens ion

Copyright �9 1998, The Clay Minerals Society 322

Vol. 46, No. 3, 1998 Water retention by allophane and imogolite 323

was adjusted to 8-9 with NaOH, and the suspension was subjected to ultrasonic waves for about 10 rain. After being allowed to settle for several hours, the dis- persed substance (allophane) was removed. This pro- cedure was repeated about 20 times to remove the al- lophane from the sample. The imogolite next was dis- persed by adjusting the pH to 4-5, and colloidal par- ticles with a Stokes' diameter smaller than 50 nm were collected by centrifugation. The pH of the sample was adjusted to 6-7, and about 100 mL of saturated NaC1 solution was added to each liter of the sample, then the sediment dialyzed. Although the particles prepared by this procedure still contained about 5-10% allo- phane (Figure lc; Karube et al. 1992), we designated it the imogolite sample.

Allophane-Imogolite Sample

This sample was prepared from the same soil as the allophane sample. After air-drying and H202 treatment, without deferration treatment, colloidal particles less than 50 nm in size were separated by dispersing at about pH 4 with HCI. Because both allophane and imogolite disperse at that pH, they existed together in the sample. The other procedures were the same as those used for the allophane sample. The relative con- tents of allophane and imogolite were estimated to be about 7:3 from an electron micrograph (Figure ld).

Montmorillonite Sample

Upton montmorillonite with a Stokes' diameter smaller than 2 p.m, 90 cmolc kg -~ of cation exchange capacity (CEC), saturated with sodium (Low 1980), was used.

Experiment

The water retention curves of allophane and imo- golite were measured over the range of - 1 . 4 to - 8 0 0 J kg-L A tension plate apparatus with a membrane filter and sample holder, 2 cm inner diameter and 3 cm high, were used for the range of - 1 . 4 to - 5 0 J kg -~, and a pressure plate apparatus with the same fil- ter and sample holder were used for - 5 0 to - 8 0 0 J kg-L Both apparatuses were used for the drying and wetting process.

For the drying process, samples were left more than 2 d for equilibration before measurement after the pH had been adjusted with HC1 or NaOH. The initial con- centration of the allophane sample after preparation was about i0 g kg -1, and that of imogolite about 1 g kg-k Three to 7 mL of each sample was pipetted into the sample holder, and suction for the tension plate, pressure for pressure plate were applied. The water content was measured by the oven-dry weighing meth- od when the sample reached equilibrium with the pres- sure potential after 1-2 d for suction, or after 2 -4 d for high pressure.

The first half of the wetting process was the same as that of drying process. After the sample was equil- ibrated with a certain pressure potential, the potential was raised. The water content was measured when the sample reached a new equilibrium after additional 2 - 4 d. Some samples for the wetting process were placed into screw-capped glass tubes with water and sonicat- ed for different periods. The dispersed samples were measured in the latter half of the wetting process.

Measurements were performed in triplicate. The co- efficients of variation for the mean were estimated to be less than 0.02 for allophane and montmorillonite, and less than 0.03 for imogolite.

The charge characteristics of allophane and imo- golite are based on the experimental data of Karube et al. (1990) determined for different pHs following the procedure of Wada and Okamura (1977) using 20 mol m 3 NaC1 as the saturating salt, and 1 • l03 mol m -3 KNO3 as the displacing agent.

RESULTS AND DISCUSSION

Change in the Water Retention Curve Due to Air- drying

Water retention curves of the air-dried and fresh al- lophane, both non-deferrated, are shown in Figure 2a. The water content of the air-dried allophane at each pressure potential was lower than that of the fresh one by 23-29% and did not recover even when the sample was treated with ultrasonic waves.

In this experiment, colloidal allophane with Stokes' diameters smaller than 50 nm constituted both the fresh and air-dried samples, and no difference was found between these samples by transmission electron microscopy (Figure la, lb). However, when the col- loids were dried to a thickness of 1-2 mm, the air- dried allophane did not become transparent, whereas the fresh allophane did. According to the law of light scattering phenomena, turbidity increases with colloid- al particle size (Hiemenz 1986), which indicates that air-dried allophane forms somewhat larger aggregates. Although the Stokes' diameter of the aggregates was not larger than 10 times that of the unit particles, ag- gregation must be the cause of the irreversible change in the water retention curves.

This irreversibility could not have been caused by free aluminum or iron hydrous oxide because at each pressure potential the water content of the deferrated, air-dried allophane was also lower than that of the fresh sample by 20-23% (Figure 2b). Nor would sur- face charges have been the cause, because there was no difference in the electric mobilities of the air-dried and fresh allophane. Therefore, the irreversible changes in the water retention curves of allophane due to air-drying must have been caused by the irreversible aggregation of allophane, as has been reported by Ku- bota (1972). It is reasonable that in nature an allo-

324 Karube and Abe Clays and Clay Minerals

Figure 1. Transmission electron micrographs, a) Fresh (before observation) deferrated allophane, b) Air-dried deferrated allophane, c) Fresh deferrated imogolite, d) Air-dried non-deferrated allophane-imogolite.

Vol. 46, No. 3, 1998 Water retention by allophane and imogolite 325

Figure 1. Continued,

326 Karube and Abe Clays and Clay Minerals

A

e ~

( h

n

-10

-10

-100

-1000

2 i

Water content (kg kg-')

4 6 8 10 12 i i i i i

o/0 ~ 1 7 6

o / , /

o e

// o e

�9 Fresh p4..I 6.5

o Ak'dried pll 6.4

a)

14

0 -1

A

-10

, i

Q.

f ~ -100 O.

Water content (kg kg-')

2 4 6 8 10 12 14 i i 1 i i

o / e

o/O/./" o/./

do/ o/g / . /

r //

/7 o e

/ �9 Fresh 5 3 pH o Airdried pH 5.3

b) -1000

Figure 2. Water retention curves of allophane with a Stokes' diameter <50 nm. a) Non-deferrated allophane, b) Deferrated allophane. (The unit of pressure potential "J k g ~'' is convert- ible to "kPa" providing the density of water is 1 Mg m 3.)

phan ic soil, once air-dried, wou ld show a m a r k e d ir- revers ib le decrease in water re ten t ion because larger aggrega tes would have b e e n fo rmed i r revers ib ly by a i r -drying (Kubo ta 1976; M a e d a et al. 1977).

C h a n g e in the Water Re ten t ion Curve by Defer ra t ion Trea tmen t

The water re ten t ion cu rves of the non-defe r ra ted f resh a l lophane at pH 6.5 ( - 4 . 5 cmol c kg 1 of net

Water content (kg kg -~)

0 2 4 6 8 10 12 14 - 1 i i :

P / ' .~

_ z.j -10

m

c ! !

-100 /~ o pH 3.9 1.201 O , . V = �9 pH 5.3 (- 5)

/

/ �9 I)1t 7.5 (-38)

. I)11 9.5 ( -82)

411

-1000

Figure 3. Water retention curves of deferrated fresh allo- phane with different charges. Values in parentheses indicate net charge in cmol c kg ~.

charge) (Figure 2a) and of the defer ra ted one at pH 5.3 ( - 5 cmolc kg -1 of net charge) (Figure 2b) were ve ry similar, t hough the defer ra ted a l lophane had s l ight ly h igher wate r r e ten t ion at about - 1 0 to - 1 0 0 J kg -1. This re la t ion did not change w h e n the pH val- ues were adjus ted to be equal. Therefore , the effect of defer ra t ion t r ea tment on the wate r re ten t ion cu rve of a l lophane was small .

Water Re ten t ion Curves o f A l lophane wi th Di f fe ren t Ne t Charges

F igure 3 shows water re ten t ion cu rves of defer ra ted a l lophane wi th di f ferent net charges con t ro l led by pH. The effects of the net charges appear on the cu rves above the - 1 0 0 J kg -I pressure potent ia l . The h igher the absolute va lue of the net charge, the lower the water re tent ion, and vice versa.

In suspens ion , a l lophane wi th a low absolu te net charge coagulates , p roduc ing low bu lk densi ty, result- ing in h igher wate r re ten t ion than that for a l lophane wi th a h igh absolute net charge. These features are cons i s ten t wi th those of a l lophanic soils which , in na- ture, have a low absolu te net charge, low bu lk dens i ty and h igh water re tent ion. Moreover , because a l lophane forms a p r ima ry f loccule (domain) even w h e n appar- ent ly wel l d i spersed (Karube et al. 1996), the wate r re ten t ion of d ispersed a l lophane also is ascr ibable to the mic roscop ic pores of these domains .

The a m o u n t of water re ta ined in the mic ropores is re la ted to the s t rength of the gel s t ructure agains t pres- sure. Because the gel s t ructure due to f loccula t ion is

Vol. 46, No. 3, 1998 Water retention by allophane and imogolite 327

0

-~ -10

== 4)

-100 a .

-lOOO

Water content (kg kg-')

4 6 8 10 12 , i 2~_10T 10 ' '

�9 . . . . . ) - - - - ~ 5 ~ e

Y ./

/ /

/ a)

14 0

G O~ 1 O.

-10

-100

Water content (kg kg-')

-10 6 8 10 12, 14 2 4

/

-1000

Figure 4. Wetting process for deferrated fresh allophane, a) At pH 5.3 (net charge - 5 cmol~ kg-l). Numerals along the broken line indicate the sonicated time (min). b) At pH 9.5 (net charge -82 cmolc kg-~).

less dense and relat ively weak, the water retention o f flocculated al lophane decreases with pressure potential decrease at a higher rate than for a dispersed one. This difference ceased to be observed below - 1 0 0 J kg ] when the difference in gel strength became negligible. Al lophane has a unique mechan i sm of high water re- tention that differs f rom swelling.

-10

-100

G =

O.

-1000

Water content (kg kg')

2 4 6 8 10 12 14 16 i i

7 " / d / o i~l 3.7 (,,48)

= ~ �9 pH 6.1(- 21 /=/~ �9 pH 7.7 (-24)

Figure 5. Water retention curves of deferrated fresh imo- golite with different charges.

Irreversibil i ty of the Water Retent ion Curves of Al lophane

Water retention curves in the drying and wett ing processes for fresh deferrated allophane at pH 5.3 and 9.5 are shown in Figure 4. Al lophane at pH 5.3 ( - 5 cmolc kg 1 (Figure 4a) did not recover the water re- tention f rom - 5 J kg -1 of pressure potential. In con- trast, al lophane at pH 9.5 ( - 8 2 cmol c kg -1) recovered f rom - 5 0 J kg -1 showing a different curve, but did not recover f rom - 1 0 0 J kg -1. Al lophane swells only when it has a high charge density and is wel l dis- persed. Because the allophane that swells has low wa- ter retention compared to that o f coagulated al lophane at high pressure potential ( low suction), swel l ing did not produce a very big change in water retention.

Al lophane at pH 5.3, when first treated with - 5 0 J kg 1 of pressure potential, then sonicated through the glass tube, recovered its water retention to some extent at - 1 . 4 J kg -~ (Figure 4a). Ultrasonic treatment ap- parently had the reverse effect with t ime because, when the allophane was wel l dispersed, its water re- tention decreased. Its final water retention, however, was close to that of the wel l -dispersed al lophane at pH 9.5 (Figure 4b).

Differences in Water Retent ion be tween Allophane, Imogol i te and Montmori l loni te

The water retention curves of deferrated imogol i te are shown in Figure 5. The water content o f imogol i te is more than twice that o f al lophane (Figure 3) at - 10 J kg ~, and about 1.4-fold at - 8 0 0 J kg -1. The relation be tween the net charge and water retention was similar to that of allophane, except for imogol i te at pH 7.7, for which the water retention differed be low about - 2 5 0 J kg -1. Under flocculation, imogol i te formed a relat ively dense structure at low pressure potentials. Because this mechanism is different f rom that for al-

328 Karube and Abe Clays and Clay Minerals

. 1 0

-~ -10

Q ,

Q.

-1000

Water content (kg kg-')

2 4 6 8 10 12 14 16 i i i i u i

~e e �9

o ~ o /

/ -

l U /

d ; / o / �9 ,Nophane pH 6.5

/ J o Imogolite I)1t 7.7

~7 Y " MontmorBonite

Figure 6. Water retention curves of deferrated fresh allo- phane, imogolite and Na-montmorillonite.

lophane , the f loccula t ion of imogol i te is ascr ibable to a thick, bund l ed a r r a ngem en t of its f ibrous particles.

The wate r r e ten t ion cu rves of a l lophane, imogol i te and Na-mon tmor i l l on i t e are s h o w n in F igure 6. The wate r re ten t ion of imogol i te was 1.5 t imes as h igh as that of mon tmor i l l on i t e at about - 6 5 0 J kg ~. Mont - mor i l lon i te re ta ined more wate r than a l lophane above - 4 0 0 J kg-] , but the re la t ion was reversed be low that pressure potent ial .

In F igure 7, a l l o p h a n e - i m o g o l i t e was mixed wi th mon tmor i l l on i t e in the mass rat io o f 1:1. This mix ture had a h ighe r wa te r re ten t ion than bo th a l l o p h a n e - i m - ogol i te and mon tmor i l l on i t e in the r ange of - 1 0 0 to - 8 0 0 J kg 1. This m e a n s that the n u m b e r of 3--0.4 jxm mic roscop ic pores increased in the mix tu re due to the associa t ion of the a l l o p h a n e - i m o g o l i t e wi th the mont - mori l loni te .

Re la t ion b e t w e e n the Water Re ten t ion Curve and Specific Sur face Area

Low (1980), who found a re la t ionship be t w een the wate r r e ten t ion cu rves and specific surface areas of clays, p roposed the fo l lowing empir ica l fo rmula based on 34 Na-smect i tes ;

In ('rr + 1) = effmJmw)+ [3 [1]

where rr is the swel l ing pressure (atm; re la ted to the pressure potent ia l by the equa t ion 1 a tm = - 1 0 1 . 3 J kg ~); m~ and mw are the masses of the solids and wa- ter; a , [3 cons tants ; and o~ is the pa rame te r of the spe- cific surface area S, wi th the re la t ionship a = kS, in

0

-10

._t3

Q. -IO0

-1

a.

-1000

Water content (kg kg-') 2 4 6 8 10 12 i i i f i i

l

f ~ j f

e~'. ,' - Montmorilionite ,~m / �9 A I l ~ a n e ~ g o H t e

'~" �9 1:1 mixture

14

Figure 7. Water retention curves of non-deferrated air-dried allophane-imogolite, Na-montmorillonite and a mixture of them.

wh ich k is a constant . This re la t ionship was appl ied to each clay (Figure 8) to obta in the a values.

Us ing the specific surface areas o f a l lophane, imo- goli te and mon tmor i l l on i t e ca lcula ted f rom their struc- tural mode ls and water contents at - 6 5 0 J kg -1, the average th ickness of the wate r fi lm at that pressure potent ia l was es t imated (Table 1). The ins ide surfaces

2.5

�9 AIIophane I)1t 6.5 o Imogoate pH 7.7

--~'~ 2.0 o MontmorUlonite/o//~ ;~ /~ / , a

1.5 / / ~" 0 / t /

1.o S ".// 0 / ' ~ / ////r'l O /

oeld/'/o ~ e O0 i , , i , , 0.1 0.2 0.3 0.4 0.5 0.6 0.7

m,/m,, Figure 8. Relationships between In ('rr + 1) and mJmw for fresh deferrated allophane, imogolite and Na-montmorilloni- te. ~r is the pressure (atm) and m, and mw are the masses of the solids and water.

Vol. 46, No. 3, 1998 Water retention by allophane and imogolite 329

Table 1. Relationships between the specific surface area S, parameter et and the average water film thickness at -650 J kg 1 t 650 '

st (m#m~) m~'/m~w t-6~0�82 m 2 kg -I c~:~ kg kg ) kg kg ~ nm

Allophane 6.2 • 105 5.9 1.67 0.108 2.5 Imogolite 8.6 • 105 6.2 2.50 0.103 2.8 Montmorillonite 8.1 • 105 3.5 1.58 - - 2.0

t S of allophane and imogolite was taken as the outside surface area, and calculated from 6 D2/(p~(D~0 - Di3)) for al- lophane, and 4 D0/(p~(D~0 - D~)) for imogolite, where Do and D~ respectively are the outer and inner diameters, and p~ is the density of solids. Numerically, Do = 4.6 nm and D~ = 2.8 nm were used for allophane, Do = 2.2 nm and D~ = 1.03 nm for imogolite, and p~ = 2.7 Mg m 3 for both.

5g a is the parameter related to the specific surface area (Low 1980).

w mw'/m s is the water content of the pore water inside the unit allophane and imogolite (Wada and Wada 1977).

�82 t 6s0 was calculated from (mw - mw')/(m~Sp~), where p~ is the density of water.

o f the unit part icles of a l lophane and imogoli te were neglec ted because the vo lume o f water in the particle did not change over the range of measuremen t used. The specific surface areas are not comple te ly different for these 3 clay minerals , the ct values being close for a l lophane and imogoli te , ev idence that there is struc- ture formation, even during the dry ing process , that differs f rom that of montmori l loni te . Imogoli te had the h ighes t value o f average or apparent water film thick- ness at - 6 5 0 J kg -1, the lowest comparable pressure potential in this exper iment . This h igh water re tent ion of imogoli te , in terms o f the specific surface area, is ascribable to the micropores fo rmed by its in ter twined fibrous particles.

C O N C L U S I O N S

The al lophane separated f rom air-dried soil had 2 3 - 29% lower water retent ion than the al lophane f rom fresh soil at a pressure potential o f - 1 . 4 to - 8 0 0 J kg-L Because the Stokes ' d iameter in both samples was less than 50 nm, the d i f ference in retent ion is as- c r ibed to s u b m i c r o s c o p i c i r revers ib le agg rega t ion caused by air-drying.

The al lophane prepared f rom fresh soil had different water retent ion curves above - 1 0 0 J kg ~ that de- pended on the pH, that is, the net charge o f allophane. The higher the absolute value of the net charge, the lower the water re tent ion above that pressure potential. This p h e n o m e n o n differs f rom swel l ing and is ascrib- able to the format ion o f a microporous structure.

The water retent ion curves of al lophane generally were irreversible. But when al lophane had a high ab- solute net charge, water re tent ion recovered to a cer- tain extent at a pressure potent ial above - 1 0 0 J kg -1 during the wett ing process .

Imogol i te had h igher water re tent ion 1.5 t imes that of Na-montmor i l lon i te at about - 6 5 0 J kg l, and more than twice that of al lophane at - 1 0 J kg -1. Water re- tention of al lophane also was higher than that o f mont - mori l loni te be low - 4 0 0 J k g - 1 W h e n montmor i l lon i te was mixed with a l lophane- imogol i te , the water reten- tion of the mixture was h igher than for ei ther o f the clays be tween - 1 0 0 and - 8 0 0 J kg -1 because of their association. The high water retent ion of imogol i te is expla ined not only by its h igh specific surface area but by the micropores fo rmed by its in ter twined fibrous particles.

A C K N O W L E D G M E N T S

We are grateful for the helpful advice of the late E E Low, of Purdue University. This research was supported by a Grant-in-Aid for General Scientific Research from the Min- istry of Education, Science, Sports and Culture, Japan.

R E F E R E N C E S

Hiemenz PC. 1986. Principles of colloid and surface chem- istry. New York: Marcel Dekker. 815 p.

Jackson ML. 1956. Soil chemical analysis--Advanced course. Madison, WI: ML Jackson. p 71-76.

Karube J, Sugimoto H, Nakaishi K. 1990. Charge character- istics and electrophoretic mobility of allophane and imo- golite. Abstract of Annual Meeting Jap Soc Irrig Drain Re- clam Eng; 1990; Nagoya, Japan. p 322-323 (in Japanese).

Karube J, Nakaishi K, Sugimoto H, Fujihira M. 1992. Elec- trophoretic behavior of imogolite under alkaline conditions. Clays Clay Miner 40:625-628.

Karube J, Nakaishi K, Sugimoto H, Fujihira M. 1996. Size and shape of allophane particles in dispersed aqueous sys- tems. Clays Clay Miner 44:485-491.

Kubota T. 1972. Aggregate-formation of allophanic soils: Ef- fect of drying on the dispersion of soils. Soil Sci Plant Nutr 18:79-87.

Kubota T. 1976. Surface chemical properties of volcanic ash soil--Especially on the phenomenon and mechanism of ir- reversible aggregation of the soil by drying. Bull Nat Inst Agric Sci B 28:1-74 (in Japanese with English abstract).

Low PE 1980. The swelling of clay: II. Montmorillonites. Soil Sci Soc Am J 44:667-676.

Maeda T, Takenaka H, Warkentin BE 1977. Physical prop- erties of allophane soils. Adv Agron 29:229-264.

Mehra OP, Jackson ML. 1960. Iron oxide removal from soils and clays by a dithionite-citrate system buffered with so- dium bicarbonate. In Swineford A, editor. Clays Clay Min- er, Proc 7th Nat Con; 1958; Washington, DC. New York: Pergamon Pr. p 317-327.

Wada K. 1980. Mineralogical characteristics of andisols. In: BKG Theng, editor. Soils with variable charge. Lower Hut- te: New Zealand Soc Soil Sci. p 87-107.

Wada K, Okamura Y. 1977. Measurements of exchange ca- pacities and hydrolysis as means of characterizing cation and anion retentions by soils. Proc Int Sem Soil Environ Fertility Manag Intens Agric Tokyo. p 811-815.

Wada K, Greenland DJ. 1970. Selective dissolution and dif- ferential infrared spectroscopy for characterization of "amorphous" constituents in soil clays. Clay Miner 8:241- 254.

Wada S, Wada K. 1977. Density and structure of allophane. Clay Miner 12:289-298.

(Received 8 January 1997; accepted 14 October 1997," Ms. 97-003)