Influence of Soil Conditioners on Infiltration and Water Movement in Soils1

J. W. KlJNE2

ABSTRACTThe infiltration of water at 2mbar suction into horizontal

columns of two soil materials of different textures was studied.The columns had been treated with two soil conditioners, Kriliumand poly(vinyl alcohol). The soil-water diffusivities werecalculated from the water content distribution data. Treatmentof a clay loam with the soil conditioners caused 1.5-fold to 3-foldincreases in soil-water diffusivity over the whole range ofvolumetric water contents. When expressed as a function of soilwater potential, soil-water diffusivity increased by this treatmentmore than 10-fold at the lower soil-water potentials. The sametreatment had far less effect on the soil-water diffusivity of aloamy sand. For both soils the rate of movement of the wettingfront increased as a result of treatment with the soil conditioners.

The calculated mean diffusivity and rate of movement of thewetting front in these soils was greater for n-heptane than forwater. The intrinsic diffusivity and penetrability differed forthe two permeants. Changes in viscosity, surface tension, andapparent contact angle did not account fully for this dependenceof diffusivity and penetrability on the permeating liquid.

THE FLOW OF water through soils could result from acombination of viscous movement and diffusion. For

ideal porous media, the theory of these two types of movementhas been well advanced. Soil, however, does not act as anideal porous medium for water movement because of theinteraction occurring between the fluid and the medium.

The evidence obtained from studies of clay-water inter-actions (9, 11, 12) implies that adsorption and osmotic forcesmay modify the flow process through soils. Kemper (8) inattempting to evaluate the influence of osmotic forces on flowthrough soil obtained qualitative but not quantitative agree-ment between calculated and measured values of osmoticpressure. At least part of the discrepancy was attributed touncertainty in determining the average effective film thicknessfor water movement.

The influence of porosity on soil-water diffusivity wasstudied by Jackson (6) who observed that diffusivity decreasedwith decreasing porosity. It is clear that the roughness of theparticles and the actual pore geometry are also important inunsaturated water movement, but the influence of these factorsis even more difficult to assess. The contact angle in un-

saturated soil is probably not the same as would be obtainedwith the same liquid on a smooth surface. Greenland (4) has-suggested that the adsorption of uncharged polymer moleculesresults in a lining of the soil pores. This would stabilize theaggregate, but a lining of the: soil pores might also have someeffect on the flow properties through the soil. The presentstudy was initiated to investigate the influence of two differentsoil conditioners on the soil-water diffusivity in two differentlytextured soils.

EXPERIMENTAL PROCEDURELiquid flow into horizontal isoil columns was studied wi.h two

permeating liquids, a O.QIN CaSO4 solution in distilled water andw-heptane. The soil materials used were a clay loam (Urrbraered-brown earth) and a loamy sand (solodized solonetz). Thesewere used untreated or treated with Krilium (sodium salt ofhydrolysed polyacrylonitrite) or with poly (vinyl alcohol) (PVA).The treatment consisted of thoroughly stirring 100 g of air-drysoil into 25 ml of a 0.4% solution of Krilium or PVA (GohsinolGL 05, mol wt 25,000). After mixing with the soil conditioners,the soils were dried and passed through a 1-mm screen. Someof the untreated soils were mixed in the same ratio with water,dried, and sieved. The soils were then packed into clear acrylicplastic cylinders. The pressure of the liquid entering the soilcolumn was controlled by a fritted glass bead plate. The platewas filled with the permeating liquid and the desired suction(2 mbar) (15) was applied prior to placing the plate in contactwith the porous material. The suction at a; = 0, the liquid source,was precisely controlled by the use of a Mariotte bottle. x

The columns were sectioned when the wetting front hadadvanced through the fourth of five thin (0.5-cm) sections locatednear the end of the column. Thus four water content measure-ments were obtained within 2 cm of the wetting front. Thelargest part of the fifth thin section, plus a sample from a sub-sequent 3-cm section, allowed water content measurements ot theoriginal air-dried material.

Calculations of the soil-water diffusivity were made accordingto the method developed by Bruce and Klute (1) using theequation:

D(9) == - L<^ f2ta de 101

xdd HI

where D(B) is the soil water diffusivity at the volumetric watercontent 8, 0,- is the initial water content, x is the distance from thesource, and t, is the total time during which water infiltrated intothe column.

An integrated diffusivity value was obtained from:

D = 3(0. - 0,-)5/3 4/•»,f (6 - 0;)2'3 D(B) de [2]

KIJNE: INFLUENCE OF SOIL CONDITIONERS ON INFILTRATION

where D is the weighted-mean diffusivity (3, 6), and 0,is the watercontent at saturation.

Philip (16) proposed the term intrinsic diffusivity to denote adiffusivity which is a function of the geometry of the mediumonly and independent of the properties of the permeant. Thisintrinsic diffusivity D is related to the soil-water diffusivity by:

r , i[T cos H J

D [31

where it is the viscosity, y is the surface tension, and H the angleof contact ^between the fluid and the solid. The intrinsicdiffusivity, 'S), was calculated from the weighted-mean diffusivityvalue D.

It is implied in the mathematical development of [1] that aplane of constant water content advances proportionally to thesquare root of the infiltration time. This leads to the assumptionthat the wetting front maintains a constant 8(6). [Swartzendruber(17), however, considers the Boltzmann transformation a con-sequence of the boundary conditions imposed on the problem.]Thus, a plot of the distance of the wetting front vs. x should yielda straight line through the origin. If X is the slope of this line, anintrinsic penetrability (7) can be calculated from:

[4]

where P is the intrinsic penetrability, which again is dependentonly on the geometry of the medium. Intrinsic penetrabilityvalues were calculated for the different soil materials.

RESULTSThe water-content distribution curves are shown in Fig. 1

for the Urrbrae soil samples and in Fig. 2 for the solodizedsolonetz. In order to compare the distribution curves, theexamples shown in Fig. 1 and 2 are for runs of equal lengths.The diffusivity values, however, are calculated from a largernumber of runs of different duration.

The shape of the water content distribution curves isconsiderably different for the treated and untreated samples.The water contents at x = O are increased by treatment,particularly for. the Urrbrae soil. The infiltration rates alsovary to a large extent between treated and untreated samples.Representative values of the infiltration times are given inTable 1. For each run, the diffusivity was calculated andplotted as a function of volumetric water content. Theresulting average relations are plotted in Fig. 3 and 4.

The diffusivity ranges of the individual samples at eachwater content are shown by vertical lines in Fig. 3. The

Table 1—Time required for penetration of the permeating liquidinto a column of soil 20 cm long

Sample Permeant Minutes

Untreated

Krilium treatedPVA treated

UrrbraeH2OQ.QIN CaSOtn-heptaneQ.01N CaSO40.01JV CaSO4

1,3701,041

121920058

— UNTREATED— KRILIUM TREATED—. PVA TREATED

DISTANCE FROM SOURCE (cm)15 17.5 20 22.5

Fig. 1—Water content distribution curves for Urrbrae soilsamples.

0.5i—HLJ 0.4

O

Q:UJ

So.2

OQ.1

— UNTREATED— KRILIUM TREATED_._. PVA TREATED

DISTANCE FROM SOURCE (cm)2£ 7.5 10 125 15 17.5 20 22.5

Fig. 2—Water content distribution curves for solodized solonetzsoil samples.

UntreatedKrilium treatedPVA treated

Solodized Solonetz0.017V CaSOjn-heptaneO.OlATCaSO.0.01JV CaSO4

10r64

•— O\ C £-

E^ 1°~ 0.6^ 0.4i=> 0.2CO

t .06Q .04

£ -02

< .015.006^.004S.002

.001.0006.0004.0002.0001

19343

171104

HEPTANERTR:K.TR.

UNTR.

O 10 20 0/o 30 40 50VOL. WATER CONTENT

Fig. 3—Soil-water diffusivity vs. volumetric water content forUrrbrae soil samples.

10 SOIL SCI. SOC. AMEH. PEOC., VOL 31, 1967

10060

. 40i 20

1064

i 1g f f iS0-2

< 0.1$.06J .04S.02

.01.006.004.002-D01

HEPTANE

1JNTR.

K.TR.

PTR.

O 10 20 0/o 30 40 50VOL. WATER CONTENT

Fig. 4—Soil-water diffusivity vs. volumetric water content forsolodized solonetz soil samples.

variation is the greatest at water contents near saturation andat the lowest water contents. This variation results from thedifficulty involved in making accurate measuremtnts of dx/ddat these water content ranges, as was noted previously (7).The vaiiation between samples was about the same for bothUrrbrae and solonized solonetz samples, but for clarity thevertical lines were omitted in Fig. 4. There was no significantdifference in the diffusivity-water content relationship foruntreated samples and those which were water treated. Whenheptane was used as the permeant there was no significantdifference between untreated or Krilium- and PVA-treatedsamples.

In Table 2 are listed the weighted-mean diffusivity values(3, 5, 7) and the corresponding intrinsic values (16). Thecontact angles needed for the calculation of intrinsic values in

Table 2—Values of weighted-mean diffusivity D (cm2/min) andthe corresponding intrinsic value for Urrbrae and solodized

solonetz soil samples (cm)

Sample

UntreatedKrilium treatedPVA treatedWith heptane

Weighted-meandiffusivity

D

cm2/min

Urrbrae

0.260.380.474.81

Intrinsicweighted-mean

diffusivity2D

cm

3.5 X 10-=3.7 X 10-65.1 X 10-«

17 X 10-8

UntreatedKrilium treatedPVA treatedWith heptane

Solodized Solonetz

1.901.171.325.02

21 X10-613 X 10-«13 X 10-«

-17 X 10-«

KRILIUMUNTR.-Q.01 N CaSO,

10 15 20 26

INFILTRATION TIMtV (min.1/2)

30

Fig. 5—The position of the wetting front as a function of thesquare root of the infiltration time.

equations [3] and [4] were obtained from capillary riseexperiments (10). Letey and co-workers (10) used ethylalcohol as the standard for capillary rise of a liquid with zerocontact angle for soil. However, comparison of the observedcapillary rise of ethyl alcohol and heptane into the samplesindicated that alcohol still had .a finite contact angle with soil.Therefore, heptane was chosen as the standard.

From the capillary rise of the two liquids the cosine of thecontact angle was calculated. These values and the corre-sponding angles are listed in Table 3. The differencesbetween the cosine of apparent contact angles in the untreatedsamples may appear small, but it should be noted that thesedifferences correspond to a difference of at least 5 cm incapillary rise. Each value of the cosine in Table 3 is anaverage of two or three observations. The maximum differencein capillary rise between replicates was 1 cm.

Figure 5 shows the position of the wetting front as afunction of the square root of the infiltration time for anumber of runs with Urrbrae samples. The lower curve foruntreated soil in Fig. 5 was obtained from the infiltration ofdistilled water into the column. With this sample a smalldeviation from linearity was observed (2) which probablyresulted from slight swelling of the sample. From the slope,X, of curves like those in Fig. i), intrinsic penetrabilities werecalculated. These data are given in Table 4. The rate ofmovement of the wetting front with heptane as the permeantwas not significantly affected by treatment and therefore onlyone value was listed in Table 4.

Table 3—Contact angles for water infiltration from capillary riseexperiments

Sample •

UntreatedKrilium treatedPVA treated

Contact angle

Urrbrae

80" 1376° 4377° 53

Cosine

0.170.230.21

UntreatedKrilium treatedPVA treated

Solodized Solonetz

78° 1079° 376° 7

0.2050.190.24

KIJNE: INFLUENCE OF SOIL CONDITIONERS ON INFILTRATION 11

— URBRRAE--SOL. SOLO METZ

-12 -10 -8 -6 -4 -2 (TSOIL-WATER POTENTIAL (atm.)

Fig. 6—Soil-water diffusivity as a functon of soil-water potential.

DISCUSSIONSoil-Water Diffusivity

Treatment of Urrbrae soil samples with either Krilium orPVA markedly increased the soil-water diffusivity of this soilover the whole range of volumetric water contents (Fig. 3).Both these materials are known to have a stabilizing influenceon the soil structure. Hence, it was expected that the waterconducting qualities would be improved by treatment. Thiswas confirmed both by the shape of the water contentdistribution curves and the infiltration rates.

The water content distribution curve is determined by thediffusivity at relatively low water content. This is particularlyapparent from Fig. 7 where the water content distributionpresented results from infiltration of distilled water and theQ.QIN CaSC>4 solution into untreated Urrbrae soil. Thederived diffusivity-water content relationship is given in thesame figure. The infiltration of the CaSOi solution results ina steep slope near the wetting front in the water content

Table 4—Values of X and intrinsic penetrabilities for Urrbraeand solodized solonetz soil samples

Sample for 0.01N forCaS04 n-heptane

Intrinsic penetrability

for0.01Ar forCaSOi n-heptane

UntreatedKrilium treatedPVA treated

Jcm/min

Urrbrae

0.62 1.810.660.78

cm

23 X 10-'21 X 10-'26 X 10-'

33 X 10-'

Solodized Solonetz

UntreatedKrilium treatedPVA treated

1.441.531.56

3.06 48 X 10-' 56 X 10-*52 X 10-'48 X 10-'

5 10DISTANCE

FROM SOURCE (cm)

20 0-1 02

Q

Fig. 7—Water content distribution curve and the correspondingsoil-water diffusivity relationship for untreated Urrbrae samplesinfiltrated with water and 0.01N CaSO4 solution.

distribution curve. Thus, the area under the curve is largerfor CaS04 than for water. This corresponds to lower diffusiv-ity values for the CaS04 run at low water contents. The highdiffusivity values for a CaSCh solution at water contents above22% result in a higher infiltration rate for CaSC>4 solutionsthan for distilled water (Fig. 5). This is in agreement withthe observation of Hanks and Bowers (5) that the infiltrationrates are determined more by the diffusivities at the higherwater contents than by those at lower water contents.

The same principles hold for the infiltration of the solution•into the differently treated soils. For example, the area underthe water content distribution curve for PVA-treated Urrbraeis larger than for Krilium-treated Urrbrae, whereas thediffusivity values for PVA-treated Urrbrae are less than thosefor Krilium-treated Urrbrae below a water content of 30%.The water contents near the liquid source, at a; =0, aredifferent for the different treatments and this influences thearea under the water content distribution curve. However,the diffusivity for untreated Urrbrae is less than for Krilium-treated Urrbrae, even at low water contents, although thearea under the water content distribution curve for Krilium-treated soil is larger. This results from the dominatinginfluence of the low infiltration rate into untreated Urrbrae(Table land Fig. 5).

In the case of solodized solonetz (Fig. 2 and 4) the generalpicture is somewhat different since treatment with Kriliumand PVA reduces the soil-water diffusivity. This is particularlyapparent for the lower soil water contents. There is againconsistency between high values of diffusivity at low watercontents and a relatively small, area under the water contentdistribution curve. The treatment with PVA, over the wholerange of water contents, results in higher diffusivities thantreatment with Krilium. This leads to a higher infiltrationrate for PVA-treated soil as will be discussed later. Ingeneral, the differences in diffusivity caused by treatment aresmaller in the solidized solonetz than in the Urrbrae soil.This was to be expected because of the different texturesof these soils.

12 SOIL SCI. SOC. AMEH. PROC., VOL. 31, 1967

The value of the weighted-mean diffusivity for these twosoils (Table 2) supports the conclusions drawn from the firstfour figures. In the calculation of the weighted-meandiffusivities more weight is placed on the higher diffusivityvalues. These higher D values are closely correlated with theinfiltration rates, but they are not necessarily the mostaccurate, since it is difficult to determine the slope of the soilwater distribution curve in this region. Therefore, theweighted-mean diffusivity values also reflect, in the main, thedifferences in infiltration rate between the untreated andtreated soils.

The intrinsic mean diffusivities listed in Table 2 varyconsiderably. One reason for this variation is that theweighted-mean diffusivities, from which the intrinsic valueswere calculated, reflect the diffusivity at higher watercontents. The infiltration rate for heptane into Urrbrae soil isconsiderably higher than for water, and consequently theintrinsic diffusivity is greater for the heptane runs. Theintrinsic values are similar in the case of the solodized solonetzwhere the weighted-mean diffusivity for heptane is only aboutthree to four times as large as for the CaS04 solution. This

• would indicate that differences in viscosity, surface tension,and contact angle are more likely to account for changes inflow characteristics with the two permea'nts in a sandy soilthan in a clay loam. One uncertainty in the calculation of theintrinsic data is the value of the contact angle. In view of theassumptions required in the method developed by Letey et al.(10), which was used here to measure the contact angles, it isinteresting that the intrinsic values were of the same orderof magnitude.

The amount of water retained in both soils at each equili-brium pressure was slightly greater after treatment with PVA(maximum difference in water content about 0.5% nearsaturation) and again greater (a difference of up to 2%) aftertreatment with Krilium. Soil-water diffusivity is expressedfor these soils in Fig. 6 as a function of water potential. Therelationship is nearly identical for the treated and untreatedsolodized solonetz samples, but the Urrbrae samples follow adifferent pattern after treatment. The small difference nearsaturation between the two untreated soils results from theirdifferent textures. Curves relating thermal diffusivity to soilwater potential also fall close together irrespective of theirtextures, except near saturation (14).

Infiltration Rates

Swartzendruber (17) has pointed out that the Boltzmanntransformation, X = xt1/2, appears as a consequence of theboundary condition in which the initial water content aheadof the wetting front is constant. Since this boundary conditionwas satisfied, it is not necessary to justify the Boltzmanntransformation as a separate assumption. As is apparentfrom Fig. 5, plots of the distance of the wetting front vs. i1/2

did indeed yield straight lines through the origin; The slopes,X, of these plots, Table 4, are an indication of the infiltrationrates. They are in good agreement with the diffusivity valuesat the higher water contents, illustrating that the infiltrationrates are determined mainly by the diffusivity values at thesehigher water contents (5). .

With both soils the infiltration rate decreased in the orderPVA treated, Krilium treated, untreated; the difference beingparticularly significant with the Urrbrae soil. The .rate ofmovement of the wetting front in both soils was greater forheptane than for the CaS04 solution. Michaels and Lin (13)have noted that the permeability of kaolinite increased with adecrease in polarity of the permeating liquid. Here the sameeffect was observed but the increase in intrinsic penetrabilitywith heptane is apparently greater than would be expectedfrom the differences in viscosity, surface tension, and contactangle as they were incorporated in the calculation of theintrinsic values. As was the case with the intrinsic diffusivi-ties, the intrinsic penetrability is more nearly independent ofthe permeant for the loamy sand than for the clay loam.

The intrinsic quantities of the PVA-treated Urrbrae arelarger than those of untreated or Krilium-treated Urrbrae. Itappears from a comparison of these values with those obtainedwhen heptane was the permeant that water (or the CaSOisolution) interacts least with the PVA-treated soil. Greenland(4) obtained evidence that stabilization of soil aggregates bythe introduction of PVA was due to the lining of soil pores,similar to spreading of a "coat of paint" over adjacent surfaces.From the present study it is not possible to distinguish betweenthe effects on the flow characteristics due to stabilization ofsoil pores and those due to the lining'of the pores. On theone hand, the possibility of a lining is evident from thedecreased value of the contact angle of the PVA-treatedUrrbrae sample as compared with the untreated sample(Table 3). The low diffusivity values at the lower watercontents, and the shape of the water content distributioncurves, on the other hand, indicate that the soil pores arestabilized by PVA treatment. PVA treatment appears to bemore effective in stabilizing the soil particles or the poresbetween them than treatment with Krilium.

Solodized solonetz samples were rendered slightly morehydrophobia by treatment with Krilium, as is indicated by thehigher contact angle for Krilium treated soil in Table 3.Hydrophobia behavior has also been reported by Letey et al.(10) for Krilium-treated silt loam. The adsorption of poly-electrolyte soil conditioners, e.g., Krilium, onto soil waspostulated by Greenland (4) to differ from that of PVA, theattachment now being only by a limited number of segmentsof the molecule. From the two soil conditioners studied,PVA, due to its mode of attachment, is the most effective ininfluencing the water conducting properties of these soils.

It can be concluded that the interaction between permeantand medium cannot be entirely accounted for by differencesin those physical quantities which were used in the calculationof the intrinsic diffusivity and penetrability. It is, thereforenecessary to incorporate other physical properties in order toobtain meaningful intrinsic quantities. A measure of thestability of the pore spaces should also be taken into accountin these experiments.

RUBIN: POST-INFILTRATION REDISTRIBUTION OF SOIL MOISTURE 13