CURRENT

---

CURRENT RESEARCH Lindane Adsorption by Lake Sediments Erik G . Lotse,’ Donald A. Graetz, Gordon Chesters, Gerhard B. Lee, and Leo W. Newland Soils Department, University of Wisconsin, Madison, Wis.

Lindane (y-benzene hexachloride) adsorption was determined on eight intact lake sediments using a radiochemical technique. Lindane adsorption was shown to be affected by sediment suspension concentration, organic matter content, lindane concentration, clay content, and lindane-to-sediment ratio and t o fit Freundlich’s equation. As the lindane molecule is neutral and not subject t o ionization, van der Waals forces and hydrogen bonding were considered the most feasible adsorption mechanisms.

T

he accumulation of insecticide residues in lakes is a potential health hazard to fish, wildlife: and humans owing to concentration of insecticides in their food chains (Hickey, Keith, et d., 1966). A n important mechanism of residue accumulation in lakes is insecticide adsorption by lake sediments. Adsorption of insecticides by soils and lake sediments, as evidenced by leaching and toxicity to insects and plants, has been shown to vary with type of insecticide, pH, temperature, and clay and organic matter content (Bailey and White, 1964; Bowman, Schechter, et nl., 1965; Chulski, 1948; Harris, 1964; 1966; Swanson, Thorp, et a / . , 1954). Lichtenstein, Schulz, et d., Although there are several examples of adsorption of insecticides on various adsorbents there are, t o the knowledge of the authors, no reports on quantitative determinations of adsorption of chlorinated hydrocarbons by lake sediments from water solutions. In most earlier investigations of insecticide adsorption, the insecticides have been added as granules or as a powder to the top of a soil column or have been added in a solvent (acetone or hexane) and mixed with the soil prior to leaching with water. Because of their low solubility, a portion of the insecticides may remain in crystalline form even after prolonged leaching with water. Thus, a true measure of the extent of insecticide adsorption cannot be obtained using such techniques. I n the present investigation Present address, Department of Plant and Soil Sciences, The University of Maine, Orono, Me. 04473

the adsorption of a chlorinated hydrocarbon insecticide, y-benzene hexachloride, was investigated using a radiochemical technique making it possible to use water as a solvent and to measure very low insecticide concentrations added in true aqueous solution. Expesiiiiental

Materials. Samples of lake bottom sediments were obtained from six Wisconsin lakes in Vilas and Oneida counties. This area is part of the Northern Highlands Lake District (Martin, 1916) and is covered with acid Woodfordian drift; most lakes are in kettles or morainal depressions. The soils of this region are mainly sandy, shallow, and podzolized. About 85 of the area is covered with second growth forest; agriculture is largely restricted to loamy or silty soils with potatoes as the principal crop. The insecticide chosen for adsorption studies was lindane, the purified gamma isomer of benzene hexachloride reported t o have a solubility in water of 6.6 p.p.m. at room temperature (Robeck, Dostal, et al., 1965). Uniformly labeled 14C-lindane samples were purchased from the Nuclear Chicago Corp., and purified nonlabeled lindane was obtained through the courtesy of Hooker Chemical Corp., Niagara Falls, N. Y . Methods. All analyses were conducted on freeze-dried lake sediment samples. Carbon, hydrogen, ash, and organic matter contents of the sediments were determined on a Sargent programmed microcombustion apparatus in a pure O2 atmosphere. Total N contents of the sediments were determined by a modified Kjeldahl procedure (Peterson and Chesters, 1964). The p H of the sediment samples was determined by the glass electrode method on a thin paste equilibrated for 30 minutes. Size fractionation of sediments was performed on samples treated with H 2 0 2and dispersed by ultrasound using a Branson Sonifier Model LS-75 as described by Fanning (1966). The dispersed samples, treated with a citrate-dithionite buffer to remove free iron oxides were separated into the following size ranges: sand (2000 to 50 microns), coarse silt (50 t o 20 microns), medium and fine silt (20 t o 2 microns), coarse clay ( 2 to 0.2 microns) and fine clay (< 0.2 micron). Volume 2, Number 5, May 1968 353

Table I. Particle Size Distribution of Mineral Portion of Sediments, Organic Matter Content, and pH

X-ray diffraction analyses were made on a General Electric XRD-3 diffractometer using C u K a radiation. Paralleloriented diffraction slides of the mineral fractions were prepared as described by Jackson (1956). Sample treatments included Mg saturation and glycerol solvation and K saturation and heat treatments at 25", 110", 300°, and 550" C. Stock solutions of lindane in hexane, 1250 p.p.m. (mg. of lindane per liter of solution) for the nonlabeled lindane, and 4.5 p.p.m. with I*C activity of 1.10 X 106 d.p.m. per ml. (d.p.m. = disintegrations per minute) for the labeled lindane were prepared. Dilute labeled lindane solutions were prepared by mixing appropriate quantities of the above two stock solutions, diluting with water, stirring until all the hexane was evaporated, and making to volume with water to achieve lindane concentrations of 0.2 t o 5.0 p.p.m. with 14Cactivity of 2000 d.p.m. in water. To effect flocculation of the fine clays, the solutions were made 0.01M with respect to CaCI2 after the addition of CaCI2 was shown to have no effect on lindane adsorption. Rate of lindane adsorption in aqueous systems was determined by equilibrating the adsorbent with 14C-labeled lindane at 23' i 1 C. on an end-over-end shaker for periods ranging from 5 minutes t o 32 hours. Lindane remaining in solution was determined as described below. Extent of lindane adsorption was determined by equilibrating intact sediments in duplicate with I4C-labeled adsorbate at 23" C. on an end-over-end shaker for 24 hours. The lake sediment with adsorbed lindane was separated on a n angle head centrifuge at 3000 r.p.m. for 10 minutes, and the lindane remaining in solution was determined by liquid scintillation spectrometry. Standard solutions, identical t o those used in adsorption experiments, were analyzed and used as the basis for calculating the percentage of insecticide adsorbed. The ratio of nonlabeled lindane to labeled lindane was assumed to be the same in the equilibrium solution and on the surface of the adsorbent. Lichtenstein, Schulz, et al. (1966) showed that lindane is stable in water and that the recovery of lindane is high when the amount of lake sediment per milliliter solution is low. It was, therefore, assumed that under the experimental conditions of low suspension concentrations (0.2 to 50 mg. of lake sediment per ml. of solution) and short reaction time (24 hours), lindane degradation was negligible and decrease in solution I4C activity was caused only by adsorption. Liquid scintillation counting on the aqueous lindane solutions was accomplished by lindane extraction with Skellysolve B (primarily hexane). The lindane concentration in the extract was determined by measuring I4Cactivity in an aliquot (5 ml.) on a Packard Model 3365 liquid scintillation spectrometer using a scintillation cocktail containing 0.5 % PPO (2,5-diphenyloxazole) as primary scintillator and 0.03 % dimethyl POPOP [ 1,4-bis-2(4-methyl-5-phenyloxazolyl)benzene] as secondary scintillator in toluene. Corrections for quenching by the hexane were made from a quench curve prepared by adding different amounts of hexane t o scintillation cocktails containing a known activity of '4C-toluene and plotting efficiency (c.p.m/d.p.m.) against counts produced by passing y-rays of O

354 Environmental Science and Technology

Sediment Sample 1 2 3 4 5 6 7

8

Particle Size Distribution, % Oven-Dried Sediment Total Sand, Silt, clay, >5Op 50 to2w < 2 ~

2.2 2.3 14.4 87.5 25.5 1.6 6.9 70.5

21.9 14.2 27.5 5.1 25.5 16.1 23.6 8.4

36.4 21 . o 26.8 4.0 16.7 18.1 20.2

8.0

Organic Matter 39.3 62.4 31.1

3.0 32.5 64.2 49.4 13.1

PH 5.1 5.7 5.3 5.4 5.4 5.5 5.4 5.5

a Packard 226Raexternal standard through the sample. The counting efficiency of unknown samples was determined directly from the quench curve. Multiple correlation and regression analyses were carried out on an IBM 1620 computer. Resirlts and Discussion

The results are organized t o provide a description of the lake sediments followed by a discussion of the adsorption properties of the sediments for lindane, a commonly used insect icide. The mechanical composition, organic matter content, and p H of the sediments are shown in Table I. A wide range of sediment composition was obtained between the samples; organic matter contents ranged from 3 to 6 4 z , sand from 1 t o 8 7 z , silt from 5 to 27%, and clay from 4 to 36% (Table I). However, the p H range of the samples, 5.1 to 5.7, was small, and none of the samples contained free carbonates. X-ray diffraction analyses of the clay and silt separates showed that the crystalline mineral composition of all sediments was qualitatively very similar. Kaolinite, illite, montmorillonite, vermiculite, interstratified minerals, feldspars, and quartz were found in most of the separates. Because the x-ray diffraction intensity was low for all samples, the samples contained relatively large amounts of x-ray amorphous material in addition t o the crystalline minerals identified. Lindane adsorption investigations on lake sediments were initiated to determine the effect of the ratio of lindane to sediment in the system, the concentration of the lindane solution, the concentration of the sediment suspension, and sediment characteristics on the extent of adsorption. The rate of lindane adsorption was determined on sediment No. 6 for an initial lindane concentration of 1.86 p.p.m. The lindane was equilibrated with the sediment for periods ranging from 5 minutes to 32 hours (Figure 1). The data show that reaction was complete essentially after 1 hour, but continued slight adsorption was evident, and a 24-hour equilibration period was chosen for all adsorption experiments. The effect of the amount of lindane added per milligram of sediment was investigated for sediments 3, 4, 6, and 8, which provided sediments of a wide range of organic matter and clay contents (Table I). The amount of lindane added per milligram of sediment was varied by varying the sediment concentra-

pl

,E 0 3 r 98 0

b

“0

7

Reaction

time,

hours

Figure 1. Rate of adsorption of lindane b> sediment 6

tion-Le., bk- varqing the amount of lake sediment added to 40 ml. of lindane solution with an initial concentration of 1.86 p.p.m. Figure 2A shows that the lindane-to-sediment ratio had a pronounced effect on the extent of lindane adsorption, also the extent of adsorption was related closely to the organic matter content of sediments. Maximum adsorption for a lindane concentration of 1.86 p.p.m. was approached at a lindane-to-sediment ratio of approximately 5 pg. per mg. and was 0.67 pg. per mg. for sediment 6 (64% organic matter), 0.51 p g . per mg. for sediment 3 (31 organic matter), 0.26 p g . per mg. for sediment 8 (13 organic matter), and 0.16 pg. per mg. for sediment 4 (3% organic matter). Each of the sediments showed a similar adsorption pattern with a steep increase in adsorption at low lindane-to-sediment ratios and a slow rise to the adsorption maximum at the higher lindane-tosediment ratios. Similarly, the percentage of lindane adsorbed varied with the lindane-to-sediment ratio and with sediment type (Figure 2 B ) . Percentage adsorption was greatest at the lowest lindane-to-sediment ratios, and the sediments with high adsorption capacity adsorbed more of the lindane in solution. Thus at ratios of 0.1 pg. lindane per mg. of sediment, almost 90% of the lindane was adsorbed by sediment 6 and 40% by sediment 4 ; at 2 pug. per mg., 30 and 8 % of the added lindane was adsorbed by sediments 6 and 4, respectively. A second factor of importance in determining the extent of lindane adsorption is the solution concentration of the insecticide. Solutions with initial concentration of 0.1, 0.2, 0.5, 1, 2, 3, 4, and 5 p.p.m. were equilibrated with appropriate amounts of lake sediment to give lindane-to-sediment ratios of 0.1, 0.2. 0.5. 1.0, 2.0, and 5.0 pg. of lindane per mg. of sediment for each initial concentration. The equilibrium concentrations were measured and the adsorbed amount calculated. Figure 2 shows that the amount of lindane adsorbed increases with increasing lindane-to-sediment ratio whether the adsorbent is held constant and the adsorbate is increased or the adsorbate is held constant and the absorbent is decreased providing the concentration of lindane is held constant. Figure 3 shows the effect of concentration of equilibrium solution as well as the effect of lindane-to-sediment ratio on the amount of lindane adsorbed. The adsorption increases with increasing concentration of lindane at constant lindaneto-sediment ratio. Plotting equilibrium lindane concentrations

/

0.6

0 4 t

$O’ X-

d

(A:

C

2

I

Added

lindane,

I

1

I

3

&

5

zg/rrc

Figure 2. Lindane adsorption by sediments as a function of lindaneto-sediment ratio

-04

-

-06

m W

g

-08-

tn

m

uQ/mQ

-10-

50

9”

I O

-I4 I S -‘0

-08

-06

-04

-02

0

02

04

Of

Log equilibrium concentration, p p m

Figure 3. Lindane adsorption by sediment 3 as a function of lindane equilibrium concentration and lindane-to-sediment ratio

Volume 2, Xumber 5 , May 1968 355

Table 11. Freundlich Constants, k , and 12, for Different Lindane-to-Sediment Ratios Lindane-to-Sediment Ratios, pg./Mg. k n

0 7

1

0.5

t

0.1 0.2 0.5 1. o 2.0 5.0

y'-0 0 6 3 ~ 0 9 3 x 1 - G 013x2iG G 3 4 x 3 + G 0 0 3 4 ~ ~ 1

/

0.295 0.302 0.331 0.331 0.339 0.437

0.92 0.89 0.92 0.86 0.75 0.76

o o

against lindane adsorbed as logarithmic functions, a linear relationship was obtained showing that this type of adsorption fits the Freundlich equation (Figure 3). The relationship can be described by the following equation:

0.4

0.3

0.2

X

/?1

0.1 0 y:G

2 0 3 1 0 0 9 5 r l - 0 018a2+0 0 4 2 ~ ~

R.0

944**

05 04

I 0 3

r

0'1

0

/

,

0,

I

1

1

31

0 2

03

0 4

ODserved

adsorption

I

I

I

05

06

07

ug/mg

Figure 4. Scatter diagram of observed lindane adsorption against adsorption predicted from regression equations A refers to sediment 3 only ** significant at the 1 level of probability

kcn

where x,.'mis the amount of lindane adsorbed per unit mass of sediment, c is the equilibrium concentration of lindane, and k and n are constants. Freundlich constants for different lindane-to-sediment ratios are given in Table 11. The constant, k , is the intercept where log equilibrium concentration is 0, and n is the slope of the line. A 50-fold increase in the lindane-to-sediment ratio effects approximately a 1.5 times increase in adsorption at an equilibrium concentration of 1 p.p.m. lindane. The fact that n decreases with increasing lindane-to-sediment ratio and that the straight lines converge implies that lindane adsorption would be equal for ratios 0.1 and 5 at an equilibriiini concentration of 11.6 p.p.m. if such a concentration could be attained. The lindane adsorption did not fit the Langmuir equation. This may be ascribed to the low solubility of lindane in water. Thus, the initial concentration of lindane may not have been high enough to allow complete monolayer coverage. To determine the effect on lindane adsorption of (1) lindaneto-sediment ratio and (2) lindane concentration, it is necessary to vary the suspension concentration of the sediment. In case 1, lindane concentration is held constant while varying sediment concentration and hence the lindane-to-sediment ratio. In case 2, the lindane-to-sediment ratio is held constant while varying the lindane and suspension concentrations. Thus, it is essential to consider the role of sediment concentration in all these investigations. The factors lindane concentration (xl),sediment concentration (x?), and lindane-to-sediment ratio ( x 3 ) were therefore deemed of greatest importance in predicting lindane adsorption by lake sediments and were used further in multiple regression analyses to determine the relative order of their importance in adsorption predictions. A multiple regression equation was established for the three independent variables xl, x?, and x 3 using only one sediment sample, namely, No. 3. The coefficient of multiple determination, R', for the regression equation showed that 89% of the variation in lindane adsorption could be accounted for from variation in the three independent variables (Table 111). Each of the partial correlation coefficients was significant at least at the level of probability. The scatter diagram (Figure 4 4 , which illustrates the relationship between predicted

5z

356 Environmental Science and Technology

=

Table 111. Correlation and Regression Data Describing the Relation Between Lindane Adsorption and Lindane Concentration, (xl), Sediment Concentration (x2), Ratio of Lindane to Sediment (xa), Organic Matter ( x p ) ,and Clay Contents (x5)for Sediments No. 3,4, 6, and 8 IndeMultiple Standard penCorrela- Partial dent Simple Partial tion Regression Dependent VariCorrela- CorrelaCoeffi- Coeffition tion cient Variable able cients SEDlhlENT

Lindane adsorption

S.Z

x3

Regression Equation : J’

= 0.203

3

0 . 874’ - 0. 825’ 0 . 63Sa

0,543‘ - 0 . 700’ 0 . 550‘

x1

0.944‘

0.598’ -0. 561’ 0.319a

+ 0 . 0 9 5 ~-~ 0.018~2+ 0 . 0 4 2 . ~ ~

ALL SEDIMEXTS Lindane adsorption

x:

0.380’ -0.588’ 0.467b

X’P

x3

Regression Equation: JJ

Lindane adsorption

=

0.094

0.380“ -0.588’ 0.467’ 0 557h 0 605b

sI xz

x: X4 x’5

Regression Equation:

0.490’ -0.478’ 0.275

0.724’

0.389b -0.452’ 0.238

+ 0.117~1- 0.014~2+ 0.032~2 0 . 598’ -0.641’ 0 462‘ 0.606b 0.478‘

+

0.919’

0 . 309’

- 0 . 404h 0 249a 0 370b 0 . 276‘

0.063 0093x1 0.013~2 0.034~3 0.0034~4f 0 . 0 0 5 8 ~ j y a

*

+

+

5z level of probabiljty.

Significant a t the Significant a t the 1

level of probability.

(obtained from regression equation) and observed adsorption values, further demonstrates the usefulness of this equation. To determine if these three independent variables could define adsorption for a variety of sediment samples, data for four sediments were included in a regression analysis. Although the multiple correlation coefficient, R , was highly significant, only 52% of the \ariation was accounted for. A scatter diagram determined for the regression equation involving three independent variables and four sediments showed that the points diverged from the line far more than was encountered for the regression equation derived for a single sediment. This increased scatter was attributed t o differences in sediment characteristics.

By including independent variables relating t o sediment characteristics, predictions of adsorption by several sediments should be improved. Many investigators have shown that clay and organic matter are responsible for much of the insecticide adsorption in soils; therefore organic matter and clay contents were included in the regression analysis. With organic matter content, x4, and clay content, x5,included in the regression equation, a significantly better prediction was achieved. A highly significant multiple correlation coefficient ( R = 0.919, significant at 12 probability level) was obtained showing that 8 4 z of the variation in lindane adsorption was accounted for by variations in the five independent variables for a variety of sediments. The scatter diagram shown in Figure 4B indicates the adequacy of the regression equation in predicting lindane adsorption by the four sediments used in this investigation. The order of importance of the independent variables is shown by the standard partial regression coefficient. Of the five independent variables considered, sediment concentration was most important followed in order by organic matter content, lindane concentration, clay content, and lindane-tosediment ratio. I n equations where sediment characteristics were not considered, the order of decreasing importance was: lindane concentration, sediment concentration, and lindanet o-sediment ratio. As the lindane molecule is neutral and not subject t o ionization, van der Waals forces and hydrogen bonding are the most plausible mechanisms of adsorption. Literature Cited

Bailey, G. W., White, J. L., J . Agr. Food Chem 12, 324-32 (1 964). Bowman, M. C., Schechter, M. S., Carter, R. L., J. Agr. Food Chem. 13, 360-5 (1965). Chulski, K., Mich. Agr. Expt. Sta. Quart. Bull. 31, 170-7 (1 948). Fanning, C. D., Ph.D. thesis, University of Wisconsin, Madison, Wis., 1966. Harris, C. R., Nature 202, 724 (1964). Hickey, J. J., Keith, J. A., Coon, F. B., J . Appl. Ecol. 3, Suppl.: 141-54 (1966). Jackson, M. L., “Soil Chemical Analysis-Advanced Course,” pp. 182-9, published by the author, Department of Soil Science, University of Wisconsin, Madison, Wis., 1956. Lichtenstein, E. P., Schulz, K . R., Skrentny, R. F., Tsukano, Y . ,Arch. Enciron. Health 12, 199-212 (1966). Martin, L., Wisconsin Geol. Natural History Siircey Bid. 36, 388-93 (1916). Peterson, L. A., Chesters, G., Agron. J. 56, 89-90 (1964). Robeck, G. G., Dostal, K. A., Cohen, J. M., Kreissel, J. F., J . Am. Water Works Assoc. 57, 181-200 (1965). Swanson, C. L. W., Thorp, F. C., Friend, R. B., Soil Sci. 78, 379-88 (1954). Receiced for reciew June 15, 1967. Accepted March 13, 1968. This incestigation was supported by a grunt (No. B-008 Wis.) froin the Ofice of Water Resources Research, U.S . Department of the Interior. Volume 2, Number 5. May 1968 357