effects of calcium carbonate on the swelling ... - Wiley Online Library

EFFECTS O F CALCIUM CARBONATE ON T H E SWELLING BEHAVIOUR O F A SOIL CLAY D. L. RIMMERI

AND

D. J. GREENLAND

(Department of Soil Science, University of Reading)

Summary Measurements of the swelling behaviour of a soil clay, in both Ca- and Nasaturated forms, before and after removal of calcium carbonate have been made. For the Ca-clays, three suctions were used, 10.0mbar, 63.1 bar, and 1.59 kbar. There were no significant differences in the swelling before and after carbonate removal at 1.59 kbar. At 63.1 bar, removal of carbonate led to decreases in swelling for all samples. At 10 mbar suction, removal led to increases in swelling for samples initially with > I per cent carbonate and to decreases for samples with < I per cent. For the Na-clays, only 10mbar suction was used and similar results to those for Ca-clays were obtained, except that the swelling was greater and the differences larger. The effect of varying the electrolyte concentration on the swelling behaviour of the Na-clays was also studied. The swelling data can be partly explained in terms of the suppression of diffuse layer formation when calcium carbonate was present. Surface area determinations before and after decalcification led to the conclusion that the carbonate was also acting as a cement. The decrease in swelling after carbonate removal, for all samples at 63.1 bar suction, and for samples initially low in carbonate at 10 mbar suction, is thought to be due to the mobilization and redeposition of hydroxy-alumina polymers during the treatment.

Introduction CLAYparticles play an important role in the formation of soil aggregates and thereby in soil structure. Aggregation is ultimately determined by the forces acting between the soil particles, and of these the swelling forces between clay particles are of prime importance, especially in soils containing large amounts of clay. The swelling behaviour of clays is largely dependent on the nature of the clay minerals and particularly their surface properties. But before any meaningful description of the swelling behaviour in terms of these surface properties can be made it is necessary to know the effect of the presence of materials likely to restrict swelling. Calcium carbonate is one such material. There are at least two possible ways in which it can directly affect the swelling behaviour. Firstly it can act on an inorganic cement, binding adjacent clay particles together and thus prevent them from swelling. In an extreme form this might be encrusting, in which case a cluster of clay particles is com letely covered by a coating of calcium carbonate. Secondly, because o its slight solubility, it can act as a source of calcium ions, the presence of which will tend to suppress the formation of diffuse double layers on the clay surfaces. The tendency to form diffuse double layers is one of the main driving forces which cause clays to swell (Quirk, 1968).

F

I

Present address : Department of Soil Science, School of Agriculture, University

of Newcastle upon Tyne. Journal of Soil Science, 27.129-39, 1976

130

D. L. RIMMER AND D. J. GREENLAND

Many clays in Britain contain large amounts of calcium carbonate, notably the ‘marls’, the Lower Lias, and many of the boulder clays. These clays are often more stable than others free of calcium carbonate, but the mode of action of the carbonate is still incompletely understood. Dumbleton (1966a,b)and others (Sherwood and Hollis, 1966) have attempted to explain the stability of the Keuper Marl in terms of the cementation of the particles into stable aggregates. They discounted organic matter and carbonates as cementing agents because they observed that aggregation persisted after these were removed by chemical treatment. They also eliminated iron and aluminium oxides for other reasons, and concluded that the cement must be poorly crystallized silica. The properties of the material studied may have resulted from artefacts produced by the chemical pre-treatment, because the method used to remove carbonate (treatment with dilute acid) leads to the mobilization and redeposition of aluminium as polymeric hydroxyalumina. More critical work is needed to determine what effects carbonate has on the behaviour of calcareous clays. T o establish its role it is essential to use techniques which do not lead to the formation of artefacts. Liming of heavy soils is well known to lead to improvements in their physical condition, but again the mechanism is uncertain (Russell, 1973). Deshpande et al. (1964)produced evidence of the direct effects of calcium carbonate in suppressing swelling of clay soils. They found that, after removing carbonate by passing carbon dioxide through a sus ension of the soil, the water content at 12.5 mbar suction after sod?ium saturation was increased substantially compared with untreated samples. The removal technique was assumed to be mild enough to prevent unwanted side-effects, such as the mobilization of aluminium. In this pa er the influence of carbonate on the swelling behaviour of Ca-saturate% and Na-saturated clay material obtained from a profile developed on the calcareous Lower Lias clay is examined. Attention has also been given to the mobilization and rede osition of aluminium during removal of carbonate, which was accomp ished by maintaining the soil in a suspension saturated with carbon dioxide, when the carbonate passes slowly into solution as the bicarbonate.

p.

Materials and methods The soil, taken from a profile at the Drayton Experimental Husbandry Farm near Stratford-on-Avon, belongs to the Denchworth series. It is a surface-water gley developed on the Lower Lias clay, and is appreciably calcareous below a de th of 60 cm. The clay fraction which was fairly uniform throughout t e profile, consisted of an interstratified com lex of mica and montmorillonite, along with kaolinite. Quartz and ca cite were also detected in the 1-2 pm clay. X-ray and DTA analyses suggested that about 20 per cent of the clay fraction was kaolinite. The potassium contents and internal surface area determined by cetyl pyridinium bromide adsorption (Greenland and Quirk, I 964) indicated that the mica-montmorillonite complex was composed of approximately equal amounts of each component. Infra-red spectroscopic analysis

E

P

SWELLING BEHAVIOUR OF A SOIL CLAY

131

confirmed that the calcium carbonate present in the separated clay sam les was primarily calcite. d e amount of carbonate in the separated clay samples (Table I) was determined by the method of Burford and Bremner (1972). This method, which is sufficiently sensitive to establish that residual carbonate expressed as CaCO, did not exceed 0.05 per cent by weight of sample, was also used to check that carbonate removal by the CO, treatment described below had been complete. Total carbon contents were obtained by combustion analysis using an induction furnace (Carr, 1973) and the organic-C component was calculated by subtracting the amount of carbonate-C from the total. Nitrogen surface areas were calculated by applying BET theory to the nitrogen adsorption isotherms, which were measured on a vacuum microbalance (Mott et al., 1974). The carbonate surface areas were obtained by the isotopic exchange method of Talibudeen and Arambarri (1964). To obtain the clay fraction (< 2 pm e.s.d.) the soil was dispersed in 7 per cent (w/v) sus ension de-ionized water by ultrasonification of for 3 min using a probe-type ultrasonic generator (100 W, 20 k z), and the clay was separated from coarser material by repeated sedimentation of a < I per cent suspension. To remove carbonate from samples with less than I per cent present, carbon dioxide was passed from a cylinder through 5 per cent (w/v) suspensions of the clay fraction for periods of 3-24 h. The suspensions were then centrifuged at 2000 g and the supernatant solution, which contained the soluble bicarbonate produced by the action of the dissolved CO,, was decanted. The conductivity of the solution was measured before discardin . The clay was resuspended in distilled water and CO, passed for a urther period. The dissolution was deemed to be com lete when the conductivity did not differ for three successive readings. %his rocess took from 6 to 18 days, depending on the amount of carbonate. f n later experiments A1 concentration and pH of the supernatant solution were measured in addition to conductivity. This revealed that pH was a better indicator of the completion of the decalcification, and subse uently the reaction was terminated when two successive pH values be ow 5.0 were recorded. Talibudeen and Arambarri 1964) also used pH as an indicator of completion of decalcification. The a uminium in solution was measured using the Eriochrome Cyanine R technique (McLean, 1965). For samples with greater than I per cent carbonate the above decalcification process took an inconveniently long time. Removal of carbonate from such samples was hastened by adding the amount of IM HCl, calculated to react with 90 to 95 per cent of the carbonate, slowly to a well-stirred 5 per cent suspension of the clay fraction. The pH remained above 5-0 during this treatment. The remaining small amounts of carbonate were then removed by the CO, bubbling procedure as above. The Na-clays were prepared by the exchange method of Aylmore and Quirk (1962). To facilitate handling during the swelling measurements all samples N

f

\

9

8

132


were formed into pellets in a press. The clay material after air-drying was ground to pass a 70 mesh sieve and the required weight (usually I g) was placed in an evacuated desiccator over a saturated solution of K,SO, at 22 "C for 24 hours, which maintained a relative humidity of approximately 96 per cent. The powder was subjected for 2 min to a 3000 kg load applied by a hydraulic ram in a pellet press of 1.27 cm diameter. In order to achieve reproducible water contents it was necessary to oven-dry (at 70 "C) the pelleted samples before making the measurements, which ensured that the samples were on the 'final reversible hysteresis loop' described by Croney and Coleman (1954). Swelling was measured by determining the water contents at varibus suctions. Since the specific gravity of water is not significantly different from 1.0, 'swelling' is the difference between the weight of water adsorbed and the weight of water required to fill the initial pores. The initial porosity of the pressed tablets used was 14-4f1-I cm3/1o0 g after oven-drying, irrespective of depth from which the sample originated, and sample treatment. T o obtain water contents at 10 mbar suction the predried samples were allowed to wet in the atmosphere for a week before placing in a desiccator at 96 per cent relative humidity for I to 2 days. They were transferred to a Haines apparatus, initially at a suction of 55 mbar, which was reduced over a period of a few days to 10 mbar. This gradual wetting was carried out to avoid any cracking and disintegration of the pellets. The suction was maintained at 10 mbar for 2 to 3 weeks to allow samples to come to equilibrium. They were then removed, weighed wet and again after drying at 70 "C. Water contents at 63.1 bar and 1-59 kbar suction were obtained by placing the dry samples held at relative humidities of 96 per cent and 32 per cent respectively by appropriate saturated salt solutions at 22 "C. The samples were removed after periods of 2 to 3 days, weighed and returned to the desiccators. This was continued until there were no further weight increases. The weights of the wet samples and after drying at 70 "C were obtained. A single determination of the moisture content was made for each sample at each suction. Results and discussion Swelling behaviour at 10 mbar suction Total water uptake at 10 mbar suction for the untreated clay samples ranged from 59.0 to 75.2 g/Ioo g of clay, calculated on a carbonate free basis (Table I). If the initial porosity of the pellets, 14 cm3/100 g is deducted, the 'swelling' is obtained, as 45 to 61 cm3/100 g. These values may be compared with values for the swelling of Ca-montmorillonite, 78 cm3/100 g, and Ca-illite 30 cm3/100 g, at 10 mbar suction given by Aylmore and Quirk (1959, 1960). The external (N,) surface areas of the clays studied by Aylmore and Quirk were of the order of IOO to 150 m2/g, and those in the present study ranged from 94 to 122 m2/g, and so the agreement with the present results is satisfactory. Down the profile, there was an increase in swelling to a depth of about 105 cm and then a decrease (Table I). The changes may be associated

1.80

2.90

0.50

0.73 0.82 0.88 0.92 0.69

135-50 15-65 165-80 180-95 I 95-2 I 0

6.50 7'70 5.80 5'55 9'75

tr

0'10

105-20 I 20-3 5

2.17 1.21 1.21

0.05

tr tr tr tr

%

2.93 2'74

%

Carbonate

Organic-C

0.96 0.56 0.42

15-30 3-45 45-60 6-75 75-90 90-105

0-15

Sample depth (cm)

27'3

-

36.3

I21 -

1 I3 -

-

Specific Surface Area (m*/g)

Carbonate

64.4 59'7 61.4 67'5 64.1 752 68.7 63.7 60.4 672 59'0 60.6 63.6 62.5

U

10.0

54'0 53.6 57-5 59'7 55'5 68.4 63.7 66.4 72.3 70'3 69.3 69.1 71.8 67.3

T

Suction:

+4.8

+8.2

+2.7 +II'9 +3'I +10*3 +8.5

-5.0

-10.4 -6.1 -3-9 -7,8 -8.6 -6.8

T-U

mbar

23.1

-

21.9

22.6 -

24.2 -

-

23.6

-

21.7

-

21'1

21.8 -

22.9 -

22.9

-

22'0

-

21'1 -

22.8 24.7

T

U

63.1 bar

5.3

-

5.1

6.3 -

5'7 -

6.7

-

-

6.3

5 '5

-

-

U

5 '4

-

4.8

5.6 -

6.4 -

6-2

-

5.3

-

5'4 -

2y

>

%

5

?i

td

0

T ! 2 -

1.59 kbar

zZ

2

TABLE I Properties of the separated clay fractions; U and T are the water content (g/Ioog clay, less carbonate) for untreated M and treated samples respectively at the various suctions r

I34


with the decrease of organic matter content to 105 cm and subsequent increase of carbonate. It has reviously been shown that the presence of organic matter can limit swe ling (Emerson, 1963 ; Theng et al., 1967). The decrease below 10 cm is most probably associated with some restricting influence, an since the decrease in swelling coincides with the region where the calcium carbonate content increases, is presumptive evidence of the involvement of the calcium carbonate.

P

d

Time (days)

FIG.I . The cumulative amounts of A1 released and the pH of the supernatant liquid during CO, bubbling over a number of days for a sample containing 0.10per cent CaCO, initially.

After carbonate removal treatment the swelling at 10 mbar suction was increased for Sam les with > 1.8per cent CaCO, initially, and decreased for those with litt e or no calcium carbonate initially present. To explain the reduced swelling it is necessary to postulate that the carbonate removal treatment, which involves the reduction of the pH of the suspensions to below yo, leads to some mobilization of aluminium from the clay. Fig. I shows the cumulative total of aluminium measured in the supernatant solutions with successive CO, bubbling treatments over a period of days. At the end of the removal process some aluminium must therefore have been present with the exchangeable cations on the clay surface, either as exchangeable ions or as hydroxy polymers. The de osition of such hydroxy-aluminium polymers is known to cause re uced swelling (Kidder and Reed, 1972). On the other hand, the removal of calcium carbonate is ex ected to lead to an increase in swelling. The net effect will de en on the relative magnitude of the two processes. Consequently,r!f samples initially low in carbonate,

P

B

B


'35

reduced swelling was observed because the aluminium effect predominated, but for samples initially high in carbonate increased swelling was observed because the carbonate removal effect predominated. The calcium carbonate will maintain a concentration of calcium ions in solution between 5 x 10-4 and 1 0 - 2 ~ depending on the partial ressure of CO, in contact with it (Marshall, 1964). Aylmore and Quir (1959, 1960, 1962 have shown that the swelling of calcium clays is generally only slight y sensitive to changes in electrolyte concentration; for Camontmorillonite there was an increase in swelling from 80 cm3/100 g in 1 0 - 2 ~ CaCl, to 93 cm3/100 g in distilled water, i.e. a 15 per cent increase. They attributed the increase to diffuse double layer formation on the external surface of the clay domains. The increases found in the present work after removal of carbonate from the samples below 105 cm are of the same order of magnitude, but theseincreases are thought to be the result of two opposing effects. The increases in swelling resulting from carbonate removal alone might well have been greater. It must be concluded then that at least in part the increases in swelling are associated with the lower concentrations of calcium ions in the solution in contact with the treated clays.

K

I

Water uptake at suctions of 63.1 bar and 1-59kbar At a suction of 1-5 kbar no significant differences in water uptake were produced by the carbonate removal, but at 63.1 bar the removal led to a small decrease in water uptake in all samples. As the treatment leads to both carbonate removal and aluminium mobilization and redeposition, it follows that the dominant effect is aluminium redeposition which leads to reduced uptake. The carbonate removal appears to have little effect, which can be explained either by postulating that double layer formation will not be important at the water contents corresponding to those suctions, or in terms of the location of the calcium carbonate. Assuming that at least part of the inhibiting action of the carbonate was due to cementation, changes in the swelling behaviour following its removal would depend on its location. The present results could be ex lained if the carbonate acted as a cementing agent primarily between a jacent domains (the ores between domains are in the range 0-1-0.3 pm). Its removal woul thus be expected to affect the swelling behaviour most markedly at suctions less than 10 bar, as observed here.

2

B

Effect of sodium saturation and electrolyte concentrations on swelling behaviour The sodium saturated clays showed similar swelling behaviour at 10 mbar suction to the calcium clays, although the magnitude was greater Table 2). For the untreated clays the presence of 1-80and 5.55 per cent aCO, led to a drastic reduction in water uptake compared with the sam le with only a trace of carbonate, and carbonate removal led to mar ed increases in uptake for these samples. The increased magnitude of the swelling for the sodium clays, compared with the calcium clays, made it ossible to study the effect of lncreasing the concentration of sodium c loride in the external solution.

L

R

K


136

This was found to cause considerable reduction in swelling. The variation of water content with C-*, where C is the salt concentration, is shown in Fig. 2. The observed dependence on C-* is indicative of diffuse double layer formation.

TABLE 2 Water content at

mbar suction for sodium saturated clay samples at various electrolyte concentrations

10

Water content at Distd. H2O

Samples I 5-30 cm untreated Carbonate-free 105-20 cm untreated Carbonate-free 180-95cm untreated Carbonate-free

I

10

683 614 263 564 262 468

I

20

IO-~M

NaCl

-

224

462 -

mbar (gH,O/Ioog clay) IO-~M

CaCOs

NaCl

NaCl

%

385 342 178 224

248 204 91 91 141 161

trace

IO+M

212

24s

-

I 40

-

5'55 -

I

I

30

10

c-%

Distilled water

FIG.2. The relationship between water content and C-k, where C = electrolyte concentration, for the untreated ( A ) and treated (0) sodium clay from 105-zo c m depth. Corrected values for the untreated samples ( 0 )are also shown.

The concentration of calcium ions released to solution from the calcium carbonate in those samples from which it was not removed will be approximately 10-SM, assuming equilibrium with the normal atmospheric concentration of CO,. Consequently the effective electrolyte concentration in the solution with which the clay is in equilibrium is (CN,+4Cc,). If the effect of calcium is allowed for, it is found (Fig. 2) that the corrected values tend towards those of the treated clays. The error bars show how the corrected values would vary if the concentration of calcium ions varied between 5 x I O ~ Mand 5 x I O - ~ M , which


I37

corres onds to a considerable variation in the partial pressures of CO,. An a ditional factor, which will tend to suppress the swelling of the sodium saturated untreated Sam les compared with those which were carbonate free, is the extent o sodium saturation. Because of the presence of the carbonate, which is slightly soluble, it is not possible to

B

F

External areas

a0

100

internal areas

120 140

240 260 280

300

320

340

m2Is

m2Is

FIG.3 . External (BET) surface areas for untreated ( 0 )and treated (0) clay samples from various depths in the profile.

obtain completely sodium saturated samples. However, as the exchangeable sodium is likely to be greater than 70 per cent, the behaviour will not be appreciably different from that of a sodium saturated sample (Glaeser and Mering, 1954). Thus the reduced swelling of the untreated Na-clays can largely be explained in terms of the suppression of diffuse double layer formation due to the presence of calcium carbonate.

Surface area determinations The measurements of the external (BET) surface area of the untreated Ca-clay samples (Fig. 3) showed that the samples with higher amounts of CaCO, below 105 cm had correspondingly lower external surface areas. The lower surface areas cannot he explained by postulating that the carbonate was in the form of separate particles of low surface area, because the higher values could not be accounted for even if the carbonate had a negligibly small specific surface area. I n fact the carbonate had an a reciable specific surface area, varying between 27 and 1 2 1 m2/g (Ta e I). Consequently it must be concluded that the carbonate

EP

6115.5.5

L

D. L. RINIMER A N D D. J. GREENLAND

138

was acting as a cement, or a coating, which prevented access to parts of the external surface (Fig. 4). This is further suggested by the increases in external surface area following carbonate removal (Fig. 3). For samples whose carbonate content was initially negligible or small the increase was small (average : + 9 2 2 per cent); but for those whose carbonate content was initially high ( > 5 per cent) the increase was larger (average: 18.48 per cent).

+

Key:

Calcium carbonate

1

Aluminium compounds

CO2 bubbling

Clay

lamellae

FIG.4. Schematic representation of the changes arising from the CO, treatment.

T h e carbonate removal treatment leads to aluminium mobilization and it seems reasonable therefore to suggest that the increases in surface area observed for samples with initial1 very small amounts of carbonate are partially due to dissolution of a uminium from oxides or hydroxides which were in positions where they were blocking access to external surfaces or cementing clay particles together (Fig. 4).

r

Acknowledgements T h e authors would like to acknowledge financial assistance from the Agricultural Research Council, and assistance in the experimental work from Mr. K. Huxford, Miss S. Morgan, and Mrs. N. Hasirci. T h e mineralogical aspects of the work were carried out by Dr. A. A. Jones. Thanks are also due to Dr. C. J. B. Mott, Mr. D. Payne and Prof. E. W. Russell for useful discussions. REFERENCES L. A. G., and QUIRK, J. P. 1959. Swelling of clay-water systems. Nature AYLMORE, Lond. 183, 1752-3. --1960. Swelling and shrinkage of clay-water systems. Trans. 7th int. Congr. Soil Sci. 2, 378-87. --1962. The structural status of clay systems. Clays and Clay Minerals, 9, 104-30. --1967. T h e micropore size distribution of clay mineral systems. J. Soil Sci. 18, 1-17. BURFORD, J. R., and BREMNER, J. M. 1972. Gas chromatographic determination of carbon dioxide evolved from soils in closed systems. Soil Biol. Riochem. 4, 1917.


I39

CARR, C. E. 1973. Gravimetric determination of soil carbon using the LECO induction furnace. J. Sci. Fd. Agric. 24, 1091-5. CRONEY, D., and COLEMAN, J. D. 1954. Soil structure in relation to soil suction (pF). J. Soil Sci. 5, 75-84. DESNPANDE, T. L., GREENLAND, D. J., and QUIRK,J. P. 1964. Role of iron oxides in the bonding of soil particles. Nature, Lond. 201, 107-8. DUMBLETON, M. J., and WEST, G. 1 9 6 6 ~ .Studies of the Keuper Marl. Min. of Transport R. R. L. Reports 39 and 40.

--1966b.

Some factors affecting the relation between the clay minerals in soils and their plasticity. Clay Minerals, 6, 179-93. EMERSON, W. W. 1963. The effect of polymers on the swelling of montmorillonite. J. Soil Sci. 14, 52-63. GLAESER, R., and MERING,J. 1954. Hydration isotherms of mixed Na-Ca montmorillonite. Clay Minerals Bull. 2, 188-93. GREENLAND, D. J., and QUIRK,J. P. 1964. Determination of the total specific surface areas of soils by adsorption of cetyl pyridinium bromide. J. Soil Sci. 15, 178-91. KIDDER, G., and REED,L. W. 1972. Swelling characteristics of hydroxy-aluminium interlavered clavs. Clays and Clay Minerals, 20. 12-20. MARSH ALL;^. E. 1964. Physical Chemistry and Mineraiogy of Soils, Vol. I , pp. 92-4, Wilev. New York. McLEAN,‘E.0. 1965. Aluminium. In Methods of Soil Analysis, 978, Ed. Black, C. A., Am. SOC.Agron., Madison, Wisc. Mom, C. J. B., RIMMER, D. L., and GREENLAND, D. J. 1974. The effects of calcium carbonate on the surface area and pore size characteristics of some selected soil clays. In Pore Structure and Properties of Materials, B 139, Ed. Modry, S., Vol. 111, Academic, Prague. QUIRK,J. P. 1968. Particle interaction and soil swelling. Israel J. Chem. x, 213-34. RUSSELL,E. W. 1973. Soil Conditions and Plant Growth, pp. 511-12, Longman, London. P. T., and HOLLIS, B. G. 1966. Studies of the Keuper Marl, Min. of SHERWOOD, Transport, R.R.L. Report 41. TALIBUDEEN, O., and ARAMBARRI, P. 1964. The influence of the amount and origin of calcium carbonates on the isotopically exchangeable phosphate in calcareous soils. J. agric. Sci., Camb. 62,9 3 7 . ’I”G, B. K. G., GREENLAND, D. J., and QUIRK,J. P. 1967. Swelling in water of complexes of montmorillonite with polyvinyl alcohol. Aust. J. Soil Res. 5,69-76.

-

(Received 22 August 1975)