Journal ofsoil Science, 1983,3423-32

The specific surface of soils determined by water

sorption

A. C . D. NEWMAN
Rothamsted Experimental Station, Harpenden, Herts ALS 2JQ

Summary
The hypothesis that the specific surface of soil can be measured by water sorption
is tested with data for 62 subsoils of widely differing origins. Ethylene glycol and water
sorption at p / p 0 = 0 . 4 7 are found to be very closely related measurements and both
are highly correlated with CEC. Both methods give a satisfactory measure of total
specific surface for soils classed as smectitic and having a large CEC. However, the
application of the multilayer theory to the sorption of water on external surfaces of
clayey soils with small CEC suggests that both sorbates overestimate the specific
surface ofsuch soils. A better estimate ofthe errors would be obtained from isotherm
measurements with water, which is more suitable for this purpose than ethylene
glycol.

Introduction
The total specific surface of soil has considerable importance, being closely related
to shrink-swell behaviour, to liquid limit, to soil water contents at specified suctions,
and cation exchange capacity (Mortland, 1954; Farrar & Coleman, 1967; Banin &
Amiel, 1970; Greene-Kelly, 1974; Ross, 1978; de Kimpe e t a / . , 1979). It is usually
estimated from the amount of a sorbate needed to form a monolayer over the surface.
However, soils are microporous (i.e. contain pores smaller than 2 nm: Gregg & Sing,
1967) and their surfaces heterogenous, so that widely differing results can be obtained
by varying the technique, and it is necessary to specify the sorbate used when quoting
a specific surface. For example, in the correlation studies mentioned above, the
specific surfaces were estimated by retention of ethylene glycol or ethylene glycol
monoethyf ether, methods reviewed by Eltantawy & Arnold (1 973, 1974). These two
polar molecules are considered to gain access to the whole soil surface including the
micropores, and it is claimed that a monolayer capacity can be identified by using
carefully defined conditions.
Water also has access to the whole soil surface so that, in principle at least, it
should be possible to estimate the total specific surface of soil by water sorption (van
Olphen, 1975). This has certain advantages: it is present in the natural soil, so is
most likely to represent field behaviour, and its relatively high vapour pressure at
ambient temperature is easy to control. It is, however, difficult to identify the stage
when a monolayer is complete. In surface chemistry, the equation of Brunauer,
Emmett and Teller (‘BET’ equation) is often used to measure monolayer capacity
(Gregg & Sing, 1967) but its application to smectite and vermiculite minerals does
0022-4588/83/0300-0023 $02.00 0 1983 Blackwell Scientific Publications
23
24 A . C . D. N E W M A N
not correctly calculate monolayer coverage of the total surface (Mooney et al., 1952;
Quirk, 1955).
However, another way of judging when a monolayer is completed is suggested
by X-ray diffraction measurements of interlamellar spacings. Ca-smectites and
Ca-vermiculites have an interlamellar spacing of about 1.5 nm that is stable over a
wide range of relative water vapour pressures (0.3 <p/po<0.8) (MacEwan & Wilson,
1980); this spacing indicates that two sheets of water are present in the interlamellar
space, equivalent to a monolayer on each surface. Further, the weight ofwater sorbed
on Ca-Wyoming montmorillonite at p/p0=0.47 is 0.221 gg-I anhydrous clay
(Mooney et al., 1952) and assuming that one molecule of water covers 0.106 nm2
(Gregg & Sing, 1967), this is equivalent to a specific surface of 798 m2/g, similar to
the specific surface of 780-800 m2/g calculated from the unit cell dimensions and
the particle size.
These data suggest that for the swelling clays that are common in soils, an effec-
tive monolayer is completed when the soil is in equilibrium with p/p0=0.47. This
paper, the first of several on the interaction of water with the solid components of
soil, examines this hypothesis and evaluates the accuracy of specific surface measure-
ments made in this way.

Soils
Samples of 58 British soils, all from sub-surface horizons between 15 and 105 cm
depth, were selected from the collection of the Soil Survey of England and Wales
to represent a wide range of soil series and parent materials. All had organic carbon
contents less than 2% and were taken from B or BC horizons with generally well-
defined blocky or prismatic structures containing between 40% and 60% < 2pm clay
fraction. Many came from surface-water gley soils and others from alluvial gley soils,
less-gleyed pelosols and brown soils (Avery, 1980). Four vertisols from Africa and
Australia were also included. Clay mineralogy ranged from mainly micaceous
(Tedburn series, clay content 56%, cation exchange capacity (CEC)=92 peq g-1 to
mainly smectitic (Swanmore series, clay content 57%, CEC= 3 1 1 peq g-I, and the
vertisols, clay contents 63-64%, CEC range 38 1-567 peq g-I) with smaller variations
in kaolinite, chlorite and free sesquioxides (Avery & Bullock, 1977). Clays inter-
mediate in composition between micas and smectites were usually mixed layer
minerals.

Methods
Particle size distribution and cation exchange capacities were measured by the Soil
Survey methods (Avery & Bascomb, 1974). Ethylene glycol retention was measured
by the following method, similar to that of Bower & Gschwend (1952). The soils
in weighing bottles were first dried to constant weight over fresh P20,, then a small
amount of redistilled ethylene glycol was added to each and the samples were stored
in vacuum over granular anhydrous CaCI,. The bottles were weighed periodically
and the monolayer coverage was assumed to occur at the point where the weight
loss became much slower (Greene-Kelly, 1974). Water sorption at p/p0=0.47 was
measured by weight after equilibrating dried soil over saturated LiNO,; this relative
vapour pressure is equivalent to a water potential of - 100 MPa or pF 6.
 SOIL SPECIFIC S U R F A C E B Y W A T E R S O R P T I O N 25
Results and Discussion
The relationships between water sorbed at pIpo = 0.47 (W(47)), ethylene glycol reten-
tion (EG) and cation exchange capacity (CEC) (Figs 1-3) show that W(47) and EG
are very closely related measurements, and that both are highly correlated with CEC.
Using a reduced major axis analysis (Kendal & Stuart, 1961) the estimated par-
ameters for functional relationships based on the linear equation y = m x + c are given
in Table 1. The close correlation between W(47) and EG for soils extrapolates to
the data for monomineralic smectites. For example, Ca-Wyoming montmorillonite
contains 0.221 g H,O per g clay at plpo=0.47 (Mooney et af., 1952), the weight of
ethylene glycol retained at monolayer capacity is 0.267 g g-1 (Brindley, 1966), and
their ratio W(47)IEG is 0.827, similar to the slope, 0.825 (Table I), of the line relat-
ing W(47) to EG for soils. Because of this close relationship between W(47) and EG,
the hypothesis that W(47) could be used to estimate total specific surface of soils
merits careful consideration.

Fig. 1. Correlation between water sorbed at R =0.47 and ethylene glycol retention for 58 British
soils and four vertisols.
26 A . C . D. N E W M A N

Fig. 2. Correlation between water sorbed at R=0.47 and cation exchange capacity.

There are at least two possible approaches to understanding the sorption of water
by soils. In the first, the classical multimolecular layer (BET) theory is used to calcu-
late the volume of water in a monomolecular layer over the whole surface (Gregg
& Sing, 1967). Another approach considers the hydration states of the cations that
balance the negative charge on the surfaces. These two ways of regarding water
associated with soil surfaces are apparently not incompatible (for a comparison see
Robinson & Stokes, 1965) but here discussion is confined to multilayer theory.

Application of BET theory

One form of the BET equation is
X/X, =cR/[(1 - R)( 1 + (C - I ) R ) ]
where R is the relative vapour pressure, x is the specific weight of water sorbed, x,
the specific weight of a monomolecular layer of water, and c is a parameter that is
constant for the given set of conditions; it very often describes water sorption on
solids, including soils (Quirk, 1955), over the same range of validity, O<R<0.35,
 SOIL SPECIFIC SURFACE B Y WATER SORPTION 27
0126

040 -

ooo-

r
a
a
ooo-
s

004-

002-

Fig. 3. Correlation between ethylene glycol retention and cation exchange capacity.

found for non-polar sorbates. When applied to non-porous solids, however, a mono-
layer is formed somewhere between R=0.1 and R=0.3, depending on the value of
c, which is related to the energy of interaction of the monolayer with the surface.
So for soil containing mica, chlorite and kaolinite clay minerals and no expanding
layer silicates, a monomolecular layer of water should form in the region of R = 0 . 2
for a value ofc=20. At R =0.47, however, the sorbed layer ofwater would be thicker,
as much as 1.7 times the monolayer thickness according to the work of Hagymassy,
et al., (1969). So this extra thickness should be allowed for when calculating the
specific surface from the weight of water sorbed at R=0.47. The equation for a
specific surface, S(W), calculated from water sorption, W(47), is
S(W) = W(47)" A ./M.n (1)
where N is Avogadro's number, 6.023 x 1 0 2 3 , A is the area covered by one H,O mole-
cule, 0.106 nm2, M is the molecular weight of water and n is the ratio 1.70 of the
28 A . C . D. N E W M A N

Table 1
Estimates of the parameters of equation y= mx+c for the re-
lationships between watersorbed at p/p0=0.47 (W(47),g g-’),
ethylene glycol retention (EG, g g-l) and cation exchange
capacity (CEC, Feq g-1)

rn E

y= W(47) 0.8249 1- 0.0199 0.0065f0.001I

x = EG Variance accounted for=96.5%

y=W(47) (1671-6)x 10-6 (5.1+1.7)xlO-3

,Y=’CEC Variance accounted for = 9 1.%o

)I= EG (204 f 7) x (2.2+ 1 . 9 ) 10-3

~
x=CEC Variance accounted for=93.2°/o

multilayer thickness at R =0.47 to the monolayer thickness. This calculation con-

trasts with that given earlier for the expanding layer silicate, Ca-montmorillonite,
which showed that a monolayer is formed at R=0.47.
Essentially, these calculations indicate that on an external surface where multi-
layers of water can develop unhindered, the water film is thicker than in the inter-
lamellar space of expanding layer silicates. The clay mineralogy of clayey soils in
Britain tends to be dominated by micaceous and smectitic minerals so that clays with
the largest K 2 0 content have the smallest CEC and little smectite, and vice versa
(Fig. 1 1 in Avery & Bullock, 1977). In consequence, soils with a large CEC will have
a considerable proportion of interlamellar surface on which the build-up of multi-
layers is restricted, whereas soils with a small CEC should be able to develop a
normal multilayer.
This was tested in the following way. As ethylene glycol retention is widely used
to measure soil surface, it was supposed for this discussion that this gives a ‘correct’
measure of total surface, S(EG). This initial assumption contains the further implied
assumptions that under the precise experimental conditions used a monolayer of eth-
ylene glycol is formed on soil and clay surfaces, and that the specific surface calcu-
lated from the mass of ethylene glycol retained is based on a predetermined surface
coverage by the ethylene glycol molecule. The average number of monolayers of
water, n, can then be calculated from the expression
n = W(47)N.A .IM.S(EG) (21
derived from Equation 1. For this calculation, it was assumed that I mg glycol covers
3 m2 and that 1 mg H,O covers 3.50 m2, both values being derived from the sorption
of water and glycol on montmorillonite of known specific surface (Brindley, 1966).
,Arguments presented above suggest that n should depend inversely upon CEC,
from n= I for soils with a large CEC and smectite content, but tending towards the
limit of n = 1.7 for soils with no micropores smaller than 2 nm and with very small
CEC, 1.7 being the statistical number of monolayers formed on non-porous solids
at R=0.47 (Hagymassy et al., 1969). For soils with a large CEC, n is close to unity
as predicted and tends to increase as CEC decreases, but there is no evidence that
 SOIL SPECIFIC SURFACE B Y WATER SORPTION 29

5
"i
1.3
. .
.
1.2 -
.- .*
* +
1.1 - s-
.-
..
w 0

.. -
e.

.
0

I ,- % ,em 0
1.0'

09 -
CEC
p q 0"
*x) 200 -300 - 4 0 0
. . 500 600

Fig. 4. Variation of multilayer thickness of water, n, with cation exchange capacity.

the relationship is extrapolating towards n = 1.7 as CEC + 0 (Fig. 4). The maximum
value for n was 1.34 for only two soils with CEC values of 170 and 2 10 peq g-I. Other
soils with small CEC values gave smaller multilayer thicknesses: the mean n for soils
with CEC < 200 peq g-' (19 soils) was 1.13; for soils with CEC < 150 peq g-I (eight
soils) mean n was 1.12.
It may be questioned whether it is appropriate to apply to soils isotherm data
obtained for sorbents with very small exchange capacities like quartz, rutile, anatase,
etc. However, the compilation by Hagymassy et a/. (1969) also included sorbents
like zirconium silicates well known for their exchange properties, and in addition,
work at present in progress here on the sorption of water by well-characterized illites
shows that the proposed standard isotherm data apply equally well to layer alumino-
silicates. It seems quite reasonable therefore to expect a multilayer with n = 1.7-1.8
to be formed on soil surfaces at the relative vapour pressure given by saturated
LiNO,.
The average value of n calculated from Equation 2 is effectively controlled by
the ratio of water sorption to ethylene glycol sorption. Small values could arise
because water sorption is less than predicted, or because ethylene glycol retention
is larger. Smaller water sorption might be due to the presence of expanding layer
silicates with internal surface on which multilayer formation is restricted. This is
unlikely; the amount needed is too large to be compatible with small exchange capa-
cities (mean n was 1.13 for soils with CEC < 200 peq g-I). The more likely reason
for small n values is that ethylene glycol sorption is also larger than a monolayer
for soils with small CEC.

CapiIfary condensation
Around interparticulate contact points in soil aggregates small capillaries exist that
till with water or ethylene glycol when a meniscus of curvature determined by the
Kelvin equation forms. Such capillary condensation modifies the analysis of isotherm
data and might be a factor in the retention of ethylene glycol and sorption of water
at R=0.47.
30 A. C. D. NEWMAN

Although experimentally closely controlled, the retention ofethylene glycol is not

defined in terms of relative vapour pressure, so little can be deduced about whether
glycol is indeed held in such capillaries.
At R=0.47, the Kelvin radius for water held in planar pores is about 0.7 nm at
20°C. From the physical dimensions of aluminosilicate sheets, pores of a size that
would allow menisci of water to form could exist where single ahminosilicate layers
overlap and this might occur in aggregates of true smectite. With mixed layer mica-
smectite structures and illitic clays the particles are thicker and such small pores must
be rare: Aylmore & Quirk (1967) showed that very little porosity in Willalooka illite,
a very fine-grained illite, was less than 1 nm in size. The contribution of capillary
condensation of water at R =0.47 is therefore considered negligible for illitic soils.

Apparent charge densities

The permanent negative charges on the surfaces of layer silicate mineral arise when
structural cations are replaced by cations of lower valency, e.g. A1 for Si in tetrahedral
sheets, Mg for A1 in octahedral sheets, and so on. Only limited ranges of substitution
are common and so the range of net negative charge per structural unit or charge
density is restricted. For a typical fine-grained smectite, Redhill montmorillonite, the
charge density is 1.3 peq m-2, for Black Jack Mine beidellite it is 1.6 peq m-* and
for muscovite, with the highest structural charge of the common layer silicates, it
is 3.43 peq m-2; a clay mica, Beavers Bend illite, has a charge density of 3.23 peq
m-,.
The linear relations between W(47), EG and CEC (Figs 2 and 3) enable average
apparent charge densities am to be calculated from the parameter m in Table 1.
Assuming as before that 1 mg ethylene glycol covers 3 m2 and 1 mg water covers
3.5 m2,
am(H,O)= l/(m.3500)= 1.71 peq m-2
am(EG)= l/(m.3000)= 1.63 peq rn-,
Both charge densities lie within the range possible for soil minerals, but for 19 soils
with CEC<200 peqg-1, am(H,O)=1.51 peq m-2, which is smaller than the overall
mean. As mentioned earlier, this group ofsoils is predominantly micaceous in miner-
alogy and would be expected to have a surface charge density in the region of
3 peq m-2. It therefore seems very probable that in soils ofthis group, specific surfaces
estimated from water sorption at R = 0.47 by assuming monolayer coverage are over-
estimated.
If, on the other hand, a multilayer thickness of n= 1.7 is used to calculate charge
density,
am(H,0)=2.56 peq m-2,
with a range I .94-3.08 peq m-2. This is a much more reasonable range of charge
densities for this set of soils.

Measurement of soil speciJic surface

For the reasons given above, neither water sorption at R=0.47 nor ethylene glycol
retention measures soil surface accurately in all instances. Both appear to measure
surface of soils containing expanding layer silicates correctly but overestimate the
surface of clayey soils with small exchange capacity. The degree of overestimate
cannot be determined from the present set of data because both measurements are
 SOIL SPECIFIC S U R F A C E BY W A T E R SORPTION 31
‘single point methods’; a more general procedure for estimating soil surfaces should
be based on water sorption over a range of relative vapour pressures.
As a method for estimating total soil surface approximately, water sorption at
R =0.47 offers advantages over ethylene glycol retention. The procedure is simple
and as no predrying is required, field-moist soil can be used: soil is stored in a vacuum
desiccator over saturated LiNO, until it reaches constant weight. The soil is then
dried in an oven at 105°C and the water content used to calculate the specific
surface, assuming that 1 mg H,O is equivalent to 3.5 m2.

Conclusions
1. Water sorption and ethylene glycol retention by subsoils are closely related
measurements, apparently because both solvate surface cations.
2. Specific surfaces calculated by assuming a monolayer of water forms at
p/p,=0.47 are little larger (1.07 times) than those calculated from ethylene glycol
retention.
3. Water sorption at p/p0=0.47 is a convenient method that offers advantages
over ethylene glycol retention for measuring total soil specific surface, but like eth-
ylene glycol retention tends to overestimate the surface of clayey soils with small
CEC.

Acknowledgements
I am grateful to C. L. Bascomb, lately of the Soil Survey of England and Wales, for
bringing this problem to my notice and am indebted to the late R. Greene-Kelly and
his co-workers for many of the results on which this paper is based.

REFERENCES
AVERY,B.W. 1980. Soil Class$cation for England and Wales, Technical Monograph No. 14.
Soil Survey, Harpenden.
AVERY,B.W. & BASCOMB, C.L. (eds) 1974. Soil Survey Laboratory Methods. Technical Mono-
graph No. 6. Soil Survey, Harpenden.
AVERY,B.W. & BULLOCK, P. 1977. Mineralogy of clayey soils in relation to soil classification.
Technical Monograph No. 10. Soil Survey, Harpenden.
AYLMORE,L.A.G. & QUIRK, J.P. 1967. The micropore size distributions of clay mineral sys-
tems. Journal of Soil Science 18, 1-1 7.
BANIN,A. & AMIEL, A. 1970. A correlative study of the chemical and physical properties of
a group of natural soils of Israel. Geoderma 3, 185-198.
BOWER,C.A. & GSCHWEND, F.B. 1952. Ethylene glycol retention by soils as a measure of sur-
face area and interlayer swelling. Soil Science Society ofAmerica Proceedings 16,342-345.
BRINDLEY,G.W. 1966. Ethylene glycol and glycerol complexes of smectites and vermiculites.
Clay Minerals 6,231-259.
ELTANTAWY, I.M. & ARNOLD, P.W. 1973. Reappraisal of ethylene glycol mono-ethyl ether
(EGME) method for surface area estimations of clays. Journal of Soil Science 24,232-238.
ELTANTAWY, I.M. &ARNOLD, P.W. 1974. Ethylene glycol sorption by homoionic montmorillo-
nites. Journal of Soil Science 25,99-110.
FARRAR, D.M. & COLEMAN, J.D. 1967. The correlation of surface area with other properties
of nineteen British clay soils. Journal of Soil Science 18, 118-124.
32 A . C . D. N E W M A N

GREENE-KELLY, R. 1974. Shrinkage of clay soils: a statistical correlation with other soil proper-
ties. Geoderma 11,243-257.
GREGG,S.J. & SING,K.S.W. 1967. Adsorption. Surface Area and Porosity. London: Academic
Press.
HAGYMASSY, J., BRUNAUER, S. & MIKHAIL, R.S. 1969. Pore structure analysis by water vapour
adsorption. I. t-Curves for water vapour. Journal of Colloid and Interface Science 29,
485-49 1.
KENDAL, M.G. & STUART,A. 196 1. Advanced Theory of Statistics. Vol. 2. London: Griffin.
DE KIMPE, C.R., LAVERDIERE, M.R. & MARTEL,Y.A. 1979. Surface area and exchange capacity
of clay in relation to the mineralogical composition of gleysolic soils. Canadian Joirrnal
of Soil Science 59, 341-347.
MACEWAN,D.M.C. & WILSON.M.J. 1980. Interlayer and intercalation complexes of clay
minerals. Chapter 3 in Crystal Structures of Clay Minerals and their X-ray IdentiJication
(eds G. W. Brindley and G. Brown), pp. 197-248. London: Mineralogical Society.
MOONEY.R.W., KEENAN, A.C. & WOOD,L.A. 1952. Adsorption of water by montmorillonite.
Journal of the American Chemical Society 14, 1367-1 374.
MORTLAND, M.M. 1954. Specific surface and its relationships to some physical and chemical
properties of soils. Soil Science 78,343-347.
VAN OLPHEN,H. 1975. Water in soils. Chapter 15 in Soil Componenfs, Volume 2 , Inorganic
Components (ed. J. E. Gieseking), pp. 497-527. New York: Springer.
QUIRK,J.P. 1955. Significance of surface areas calculated from water vapour isotherms by the
use of the B.E.T. equation. Soil Science 80,423-430.
ROBINSON, R.A. & STOKES,R.H. 1965. Electrolyte Solutions, pp. 55-58. London: Butterworths.
Ross, G.S. 1978. Relationships of specific surface area and clay content of shrink-swell poten-
tial of soils having different clay mineralogical compositions. Canadian Journal of Soil
Science 58, 159-166.

(Received 23 February 1982)