Environ. Sci. Technol. 1999, 33, 1269-1273
An Optimized Surfactant Formulation for the Remediation of Diesel Oil Polluted Sandy Aquifers MEHDI BETTAHAR,† G E R H A R D S C H A¨ F E R , * , † A N D M A R C B A V I EÄ R E ‡ Institut de Me´canique des Fluides, Institut Franco-Allemand de Recherche sur l’Environnement, UMR 7507 ULP CNRS, 23 rue du Loess, BP 20, F-67037 Strasbourg, France, and Institut Franc¸ ais du Pe´trole, 1 et 4 avenue de Bois-Pre´au, F-92506, Rueil-Malmaison, France
Reduction of hydraulic conductivity during the treatment of hydrocarbon polluted soils by using surfactants may be an inconvenience for a good recovery of the pollutant. The purpose of the present research work is to determine the main factor responsible for this reduction and by the same way to give an application solution to this problem. Experimental studies on sand filled laboratory columns have shown that the reduction of hydraulic conductivity can be due neither to the precipitation of the anionic surfactants in the presence of calcium ions nor to the clay minerals present in small proportion in the porous medium. We point out that this phenomenon is essentially due to the agglomeration of micelles and show that addition of 4 wt % of butyl alcohol under specific conditions contributes to improving both the injectability of the surfactant solution and the efficiency of the treatment. The optimized surfactant formulation, containing a mixture of two commercial surfactants and this cosolvent, palliates the reduction of hydraulic conductivity and decreases the interfacial tension by a factor of 100. As a consequence, after three pore volumes of injected surfactant solution, 85% of the trapped oil is removed compared to 11.4% obtained without cosolvent.
Introduction In the case of groundwater contamination by hydrocarbons spilling, one of the most important problems is how to recover the long-term pollution source, residual hydrocarbons trapped in a porous medium (1-6). Since passive remediation techniques, such as pump and treat of the contaminated groundwater and excavation of the polluted soil followed by on-site treatment are often costly, time intensive, and even insufficient, there is increasing interest in confining the pollutant or applying in situ treatment to the contamination source. While the hydraulic treatment consists in pumping the mobile part of light nonaqueous phase liquids (LNAPL), the remaining residual organic phase is usually displaced by injection of steam, alcohols, polymers, or surfactants (7-9). Surfactant-aided recovery techniques seem to be very efficient and promising in the case of nonvolatile hydrocarbons (1012). * Corresponding author phone: (+33) 388-106797; fax: (+33) 388106795; e-mail:
[email protected]. † Institut Franco-Allemand de Recherche sur l’Environnement. ‡ Institut Franc ¸ ais du Pe´trole. 10.1021/es980598z CCC: $18.00 Published on Web 03/06/1999
1999 American Chemical Society
There are two general mechanisms by which surfactants can enhance the removal of nonaqueous phase liquid (NAPL) sources from the unsaturated and saturated zone. From an engineering standpoint, the easiest to apply is solubilization of the organic phase, resulting from micelle formation (1315). The second mechanism, mobilization of NAPLs, depends on the efficiency of surfactants to lower the interfacial tension between NAPL and injection solution (16). Mobilization has a greater potential than solubilization to increase the rate of remediation but can be riskier because of the movement of free liquid phase. In 1994, large scale experiments were conducted on an experimentally controlled site named SCERES (Site Contro ˆ le´ Expe´rimental de Recherche pour la Re´habilitation des Eaux et des Sols, 25 × 12 × 3 m) where the in situ remediation technique consisted of surfactant flushing from the soil surface (10, 17, 18). This process was carried out after a maximum amount of oil had been recovered by hydraulic pumping in the recovery well. The surfactant infiltration allowed the displacement of the main part of the residual oil trapped in the vertical zone of the impregnated domain. However, during the infiltration, a reduction of the hydraulic conductivity, leading to a quasi-plugging of the porous medium, was observed. This phenomenon has also been observed by others (19, 20) and explained as being due to precipitation of the anionic surfactant with calcium ions and/ or retention of surfactant on clay minerals (21). The focus of the present work is to explain this reduction of hydraulic conductivity of the porous matrix during surfactant flushing and to improve the surfactant solution using column experiments. Several tests in laboratory columns have been performed in order to study both the influence of the hydraulic gradient and the role of clay minerals on the hydraulic conductivity as well as the effect of the structure of the surfactant solution used. Furthermore, results of percolation and decontamination experiments using an improved surfactant solution (containing a cosolvent) are given.
Experimental Section Set up. Surfactant flow experiments were performed on columns C1 (length L ) 150 cm, inside diameter D ) 10 cm) and C2 (L ) 40 cm, D ) 5 cm), made of Plexiglas, filled with quartz sand and equipped with piezometric tubes to follow the evolution of hydraulic conductivity K in the different parts of the column. The characteristics of the experimental set up are described in Figure 1. The chosen flow mode is downward. During the experiments, the difference between the water levels in the upstream and downstream tanks was maintained. Flow experiments were carried out at different water head differences (∆h) in order to study the influence of the hydraulic gradient (∆h/L) on the flow behavior of the solution. Applying Darcy’s law, the hydraulic conductivity of column zone (i) can be calculated as follows (see Figure 1)
Ki )
QLi S(hi-1 - hi)
where Q is the measured flow rate, S is the cross section of the column, hi-1 respectively hi is the measured piezometric head in section (i-1) respectively section (i), and Li is the length of the column zone considered. Using a hydraulic gradient of 0.2 corresponds in column C1 to a stationary water flow rate of about 75 mL/min. VOL. 33, NO. 8, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1269
FIGURE 2. Evolution of the normalized hydraulic conductivity of the porous medium: location of the plugging (K represents the hydraulic conductivity using water, V is the discharged surfactant volume, and Vp is the pore volume of the porous medium).
FIGURE 1. Scheme of the experimental set up.
TABLE 1. Composition of the H2F Sand %
SiO2
Al2O3
Fe2O3
TiO2
MgO
CaO
Na2O
K2O
97.50
1.28
0.07
0.04
0.01
0.01
0.07
0.90
TABLE 2. Physical Properties of the H2F Sand hydraulic conductivity porosity diameter characteristics: d60/d10 d50 longitudinal dispersivity
8 × 10-4 m/s 0.4
FIGURE 3. Effect of clay minerals on the surfactant flow rate.
1.46 0.40 mm 0.9 mm
Surfactant. The studied formulation “Resol 30” is a mixture of two commercial surfactants: an anionic surfactant and a nonionic one, which aims particularly at improving the water solubility of the ionic surfactant. The composition of Resol 30 is the following: 0.5 wt % (active material) of sodium dioctyl sulfosuccinate, 0.5 wt % (active material) of ethoxylated (20 OE) sorbitan trioleate, and 0.15 wt % of NaCl. The ionic surfactant (sodium dioctyl sulfosuccinate) is analyzed by the traditional two-phase hyamine titration method (22). The nonionic surfactant (ethoxylated (20 OE) sorbitan trioleate) is titrated by colorimetry: after extraction with CH2Cl2, in presence of ammonium cobaltothiocyanate, the analysis is performed using a spectrophotometer with a wavelength of 620 nm. The water used for solution preparations is tap-water. Selection was also based on the following properties of the surfactants: solubility in oil and water, detergency, oil dispersion, emulsification, foaming, biodegradability and wetting (23).
TABLE 3. Physical Properties of the Diesel Oil density at 18 °C dynamic viscosity at 20 °C interfacial tension (oil/water)
0.84 µ ) 2.2 × 10-3 kg/(m s) 30 mN/m
The volume of surfactant solution used in the flow experiments varied considerably depending on the applied hydraulic gradient: from 50 up to 300 pore volumes. The infiltration of surfactant solution at the entrance section of the column was stopped when the flow rate was lower than 5% of the initial water flux. Sand. The porous medium is a uniform and slightly clayey quartz sand H2F (Table 1), which has a specific surface area of 0.5 m2/g and a cationic exchange capacity of 0.34 meq/ 100 g. Tables 1 and 2 summarize the chemical and physical properties of this medium. In order to minimize trapped air in the sand columns, the sand was introduced in the set up initially filled with water. The bulk density of the porous medium was about 1.59 g/cm3. Afterwards the sand columns were flushed downward using deaerated water. Contaminant. The chosen pollutant was a commercial diesel oil. This product contains a wide range of hydrocarbons (67.9% aliphatics and 32.1% aromatics) and, because it is used so widely, is representative of many cases of real-life pollution. Its properties are given in Table 3. A dye (organolred) was added in order to allow visual detection of the pollutant during the experiments. Soil samples are analyzed by infra-red spectrometry after extraction with carbon tetrachloride (CCl4) for oil quantification. 1270
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 8, 1999
Results and Discussion Study of the Surfactant Flow Behavior. The plugging phenomenon (dramatic decrease of the hydraulic conductivity Ki) occurs in the upper zone of the column, while the remainder is spared (Figure 2). If the plugging were due to the precipitation of the anionic surfactants by calcium ions coming from the clay minerals (20), we would have observed a decrease of Ki in all column zones. Referring to the results (Figure 2), it is possible that filtration of surfactants by fine particles takes place in the upstream part of the column and causes the plugging after injection of seven pore volumes (V ) discharged volume; Vp ) pore volume of the porous medium). The role of fine particles is shown in Figure 3. The results of experiments using a porous matrix devoid of clay minerals were obtained by displacement of clay particles present in
FIGURE 4. Effect of the hydraulic gradient on the surfactant flow rate.
FIGURE 7. Schematic representation of monomers-micellescrystals equilibrium (26).
FIGURE 5. Evolution of the surfactant solution flow rate vs time.
FIGURE 8. Comparison of the flow behavior of Resol 30 and Resol+.
FIGURE 6. Effect of the standing time of Resol 30 on the flow rate of the surfactant solution vs time. the medium (24). The plugging phenomenon is persistent in spite of the elimination of clay minerals. This proves that fine particles are not responsible for the flow rate decrease of the surfactant solution in the porous medium. Evidently, we have to consider the properties of the surfactant solution in this process. Furthermore, we notice that the first decrease in the flow ratio Q/Q0 (1.0 to 0.8; Q0 is the initial flow rate), observed in Figures 3-6, is due to the difference of viscosity between the water initially used for the saturation of the porous medium and the surfactant solution. The effect of the applied hydraulic gradient on the surfactant infiltration was also studied. The results obtained are illustrated in Figure 4. The plugging took place whatever value of hydraulic gradient was applied in the experiments. The volume of infiltrated surfactant solution varies in the same way as the hydraulic gradient. The results of all experiments show a significant decrease in the flow rate after 10 h, and flow essentially ceases at nearly 20 h (Figure 5). The fact that time plays a role in the decrease of the infiltration flow rate induced us to take into account the properties of the surfactant mixture. This point of view was understood by studying the effect of varying the ripening time of the solution before its injection. Applying the same hydraulic gradient (∆h/L ) 0.2), three
FIGURE 9. Recovery efficiency of the two surfactant formulations (Resol 30 and Resol+). experiments were realized as follows: a first test on surfactant injection was conducted with a freshly prepared solution, a second with a solution allowed to stand for 24 h, and a third experiment with a solution 52 h old. The results shown in Figure 6 underscore the characteristic variations of the surfactant solution behavior versus time, which means that the solution should not be allowed to stand for a long time before injection. Interpretations. We have shown that the surfactant solution possesses some kind of instability. The experiment carried out with a high hydraulic gradient allows one to inject a considerable volume of surfactant but does not solve the problem of plugging, which is observed after 20 h of infiltration. Furthermore, a surfactant solution which stands for a long time before its utilization quickly plugs the porous medium. The main factor responsible for the plugging phenomenon is due only to the properties of the surfactant solution used and is independent of interactions between the aqueous VOL. 33, NO. 8, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1271
TABLE 4. Oil Recovery Mass BalancesEffects of the Formulations with and without Alcohol surfactant formulation
interfacial tension aqueous solution/oil (mN/m)
1.5 Vp
Resol 30 (without alcohol) Resol+ (with alcohol)
0.1 0.001
5.5 44
oil recovery (in %) after injection of 2 Vp 3 Vp 4 Vp 8.0 78
11.4 85
14.9 85
6 Vp 21 86
TABLE 5. Effect of the Absence of One of the Surfactants on the Oil Recovery composition (%)
oil recovery (in %) after injection of
test
alcohol
anionic surfactant
nonionic surfactant
test 1 (Resol+) test 2 test 3
4
0.5
0.5
4 4
0 1
1 0
1.5 Vp
2 Vp
3 Vp
4 Vp
6 Vp
44
78
85
85
86
16.4 2.1
20.4 2.1
20.5 2.1
0 0.14
0.4 1.4
TABLE 6. Effect of the Alcohol Quantity on the Oil Recovery composition (%)
oil recovery (in %) after injection of
test
alcohol
anionic surfactant
nonionic surfactant
test 4 (Resol+) test 5 test 6
4
0.5
0.5
2 5
0.5 0.5
0.5 0.5
solution and the porous matrix (sorption, precipitation, etc.). The filtration of dispersed objects contained in the aqueous medium leads us to assume that the surfactant forms aggregates which, after a certain time, have a size sufficient to be retained at the surface of the porous matrix and cause its plugging. In our case, the surfactant solution concentration (10 g L-1) is relatively important, and therefore the presence of calcium ions in the water used for preparation favors the aggregation of lamellae. Such mixed surfactant systems may also form cylindrical aggregates (25). The hypothesis retained is the formation of mixed liquid crystals (Figure 7); according to preliminary light-scattering measurements, the average size of the diffusing particles is about 450 nm which is much greater than the average size of a mixed micelle. Test of the Improved Surfactant Formulation. The best way to avoid the formation of aggregates is probably the addition of another component to the surfactant mixture; the cosolvent chosen here is butyl alcohol. The attractive forces between micelles, induced in general by the divalent ions (especially the calcium ions), are reduced due to the solubilization capacity of the added alcohol. The composition concentration of 1% surfactant solution (0.5 wt % of sodium dioctyl sulfosuccinate, 0.5 wt % of ethoxylated (20 OE) sorbitan trioleate) with 1.5 g L-1 sodium chloride and 4 wt % butyl alcohol, which represents the new surfactant formulation “Resol+”, allows the solution to pass from the opalescent phase to the clear phase. The physical change results from the destruction of mixed liquid crystals. In addition, these concentrations were optimized in batch experiments in order to get the lowest interfacial tension. This parameter corresponds to a TYPE III phase behavior (also named Winsor III) (27) in which a microemulsion is in equilibrium with oil and water. A flow experiment using this new formulation was carried out in column C2. A comparison of the flow behavior between the new and the old formulation is illustrated by Figure 8. The new mixture appears to be injected without any decrease in hydraulic conductivity, except at the beginning. As already mentioned, this is due to the difference in viscosity between the new surfactant solution and water with which the column was saturated. 1272
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 8, 1999
1.5 Vp
2 Vp
3 Vp
4 Vp
6 Vp
44
78
85
85
86
16.9 46
37 74
40 77
42 77
42 77
The pollutant recovery experiments were performed in C1 columns filled with 4% oil-saturated sand using the two surfactant formulations Resol 30 and Resol+. The porous medium was first flooded with water, and then the surfactant solution was injected at a fixed hydraulic gradient (∆h/L ) 0.2). The results show a significant difference between the recovery efficiency of the two solutions (Figure 9). In addition to its being well injected in the porous medium, the new solution is more effective. The interfacial tension, measured by the spinning drop method, was shown to be reduced by a factor of 100, and the mass balance of the oil recovered increases from 8% (for the formulation without alcohol (Resol 30)) to 78% (for the formulation with alcohol (Resol+)) after the injection of two pore volumes of surfactants (Table 4). The influence of the two compounds (nonionic and anionic surfactants) has been studied in additional laboratory experiments. The results of tests 1-3 (Table 5) show clearly that the absence of one of the surfactants affects significantly the recovery of oil initially trapped at residual saturation (4%) in the porous matrix. The quantity of alcohol to be used seems of great importance since its variation can also affect the oil recovery from the columns. Results given in Table 6 demonstrate that less than 4% (test 5) could be insufficient for disturbing liquid crystal formation. Test 6 confirms that more than 4% of alcohol is not useful; on the contrary, it can decrease the oil recovery.
Acknowledgments Support of this work by the ZAFA 2 group of IFARE (Institut Franco-Allemand de Recherche sur l’Environnement) is gratefully acknowledged. Financial assistance for this research was provided by the French government (MENRT, MATE), the Regional Council of Alsace and the French National Program in Hydrology (PNRH/INSU-CNRS).
Literature Cited (1) Van Dorpe, F. Ph.D. Thesis, Universite´ Louis Pasteur de Strasbourg, 1997; p 256. (2) Ducreux, J.; Bocard, C.; Gatellier, C.; Minssieux, L. Proceedings of the International Conference on Subsurface Contamination by Immiscible Fluids; Calgary, Canada, 1990; pp 337-350.
(3) Razakarisoa, O. Ph.D. Thesis, Universite´ Louis Pasteur de Strasbourg, 1990; p 149. (4) Zilliox, L.; Muntzer, P.; Fried, J. J. In Proceedings of International Symposium on Groundwater Pollution by Oil Hydrocarbons; Prague, Chzechoslovakia, 1978; pp 209-227. (5) Van Der Waarden, M.; Bridie, A. L. A. M.; Groenewoud, W. M. Water Res. (GB) 1971, 5, 213-226. (6) Schwille, F. GWF-Wasser/Abwasser 1971, 112, 307-311, 331378, 465-472. (7) Jabbour, C.; Quintard, M.; Bertin, H.; Robin, M. J. Petroleum Sci. Eng. 1996, 16, 109-130. (8) Mackay, D. M.; Cherry, J. A. Environ. Sci. Technol. 1989, 23, 630-636. (9) Bourdarot, G.; Sardin, M. In Proceedings of the Conference on Solid-Liquid Interactions in Porous Media; Nancy, France, 1984; pp 567-585. (10) Bavie`re, M.; Bocard, C.; Ducreux, J.; Monin, N.; Muntzer, P.; Razakarisoa, O. Paper SPE 37258, Proceedings, International Symposium on Oilfield Chemistry; Houston, TX, 1997; pp 481491. (11) Abdul, A. S.; Ang, C. C. Ground Water 1994, 32, 727-734. (12) Abdul, A. S.; Gibson, T. L.; Rai, D. N. Ground Water 1990, 28, 920-926. (13) Lundsten, G.; Backlund, S.; Kiwilsza, G. Prog. Colloid Polym. Sci. 1994, 97, 194-198. (14) Bavie`re, M.; Rouaud, T. Revue de l’Institut Franc¸ ais du Pe´trole. 1990, 45, 605-620. (15) Kile, D. E.; Chiou, C. T. Environ. Sci. Technol. 1990, 24, 205208.
(16) Borkovec, M. J. Chem. Phys. 1989, 91, 6268-6281. (17) Van Dorpe, F.; Bettahar, M.; Ott, C.; Razakarisoa, O. In Proceedings of the 12th French Congress on Mechanics; Strasbourg, France, 1995; pp 85-88. (18) Bettahar, M. Ph.D. Thesis, Universite´ Louis Pasteur de Strasbourg, 1997; p 163. (19) Celik, M. S.; Manev, E. D.; Somasundaran, P. Proc. Am. Institute Chem. Engineers Symp. 1982, 78, 212, 86-96. (20) Allred, B.; Brown, G. Ground Water Monitoring Revue 1994, 174-184. (21) Zundel, J. P.; Siffert, B. In Proceedings of the Conference on Solid-Liquid Interactions in Porous Media; Nancy, France, 1984; pp 447-462. (22) Brewer, J. L. J. Institute Petroleum 1972, 58, 41-46. (23) Bavie`re, M.; Ducreux, J.; Monin, N. First World Congress on Emulsion-Paris, October 19-22; 1993; paper no. 430333. (24) Behra, P. Ph.D. Thesis, Universite´ Louis Pasteur de Strasbourg, 1987; p 191. (25) Buler, E. Ph.D. Thesis, Universite´ Louis Pasteur de Strasbourg, 1996; p 190. (26) Beauvais, G. Ph.D. Thesis, Universite´ de Paris VI, 1989; p 186. (27) Maunier, C. Ph.D. Thesis, Institut Polytechnique de Lorraine, 1996; p 197.
Received for review June 12, 1998. Revised manuscript received December 17, 1998. Accepted January 18, 1999. ES980598Z
VOL. 33, NO. 8, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1273
