Phys. Chem Earth (B),
Pergamon
Vol. 24, No. 6, pp. 501-510, 1999 0 1999 Elsevier Science Ltd All rights reserved 1464-1909/99/$ - see front matter
PII: S1464-1909(99)00037-4
Partitioning and Interfacial Tracers to Characterize Non-Aqueous Phase Liquids (NAPLs) in Natural Aquifer Material B. Setarge12, J. Danze?,
R. Klein’ and P. Grathwohl’
‘Applied Geology, Geological Institute, University of Tuebingen, Sigwartstr.10, D-72076 Tuebingen, Germany 2Present Address: P.O. Box 28502, Addis Ababa, Ethiopia 3Corresponding Author: E-mail: danzer@uni-tuebingende Received 19 July 1998; accepted 23 December 1998 development of effective groundwater remediation techniques has become an important field of hydrogeology. One of the most severe environmental problems in groundwater contamination are non-aqueous phase liquids (NAPLs) within the subsurface. Due to the low water solubility and low mobility NAPLs are usually a long term pollution source in groundwater. The characterization of NAPLs (volume, saturation and geometry) is essential for the design of appropriate groundwater protection systems (e.g. funnel-and-gate, reactive walls, natural attenuation) and remediation techniques (e.g. surfactant flushing). A major weakness of current characterization methods, such as core sampling and well sampling, is that they provide data only at discrete points. Therefore, the probability of detecting zones of local NAPL saturation in the subsurface is quite small. Considering the structural and hydraulic heterogeneity the subsurface accurate and comprehensive mapping of the NAPL distribution is usually cost prohibitive using discrete point data (Nelson and Brusseau. 1996). Partitioning tracers are solutes which partition between the NAPL and water. The partition coefficients of partitioning tracers can be determined in the laboratory. The NAPL volume in the pores (saturation) in principle can be determined from the differences of arrival times of the partitioning and conservative tracers. Interfacial tracers adsorb at the interface between NAPL and water. The interfacial area in principle can be determined from the adsorbed mass and the known area per molecule of the interfacial tracer. The particular advantage of these reactive tracer test methods is that they provide integral informations of NAPL saturation in a subsurface volume that is much larger than that available from either core samples or geophysical logs (Mayer and Miller, 1992 cited by Jin et al., 1995). Different alcohols were tested as partitioning tracer by Jin et al. (I 995), Hayden and Linnemeyer (1997) and Meyers et al. (1997) and showed promising results. Sulphur hexafluoride SF6 which is reported to be used as
Abstract. Hydrophobic organic contaminants released from non-aqueous phase liquids (NAPL) impose a serious risk on groundwater quality NAPL can exist as coherent “pools” or residual phase (“blobs” or “ganglia”) within the aquifer. Depending on the NAPL volume the time scale of contaminant emission may be centuries or longer. The concentration in groundwater depends on the NAPL volume and the interfacial area between NAPL and water. The knowledge of these parameters is essential for site characterization, risk assessment and planning of in-situ aquifer remediation techniques, This paper describes the characterization for NAPLs using partitioning (volume) and interfacial tracers. Laboratory studies (undisturbed soil cores, batch experiments) were conducted using coal tar contaminated material from a former manufactured gas plant site. Different alcohols (substituted pentanols) have been tested as partitioning tracers. An anionic surfactant (ethoxylated alkyl sulfate) was used as interfacial tracer. The alcohols have low but significant octanol/water partition coefficients K,, between 24 and 240 resulting in NAPLiwater partition coefficients between 1 and 22. The surfactant accumulates significantly at the NAPLiwater interface. While these reactive tracers were not retarded by the natural aquifer material, they were retarded in the presence of NAPL with respect to the transport of a conservative tracer. A surfactant flushing was applied to the contaminated column in order to enhance solubilization of tar oil constituents. The NAPL saturation calculated from the retardation factor and the equilibrium NAPUwater partition coefficients decreased from 30 % to 17 % in the pre- and post-surfactant flushing system, respectively. 0 1999 Elsevier Science Ltd. All rights reserved 1
Introduction
Due to the enormous toxic wastes deposited in former industrial areas in many countries of the world, the Correspondence
to: .I. Danzer
501
502
B. Setarge et al.: Non-Aqueous Phase Liquids in Natural Aquifer Material
partitioning tracer (Wilson and Mackay, 1993) was also tested in the laboratory and was found to be very sensitive to entrapped air within the column systems (data not shown). The partition coefficients may be predicted e.g. by using the UNIFAC model (Wang et al., 1998). Saripalli et al. (1997a and 1997b), Annable et al. (1997), Kim et al. (1997) and Cary (I 994) described the theory, application and importance of interfacial tracers (IFT) to determine the interface and geometry of subsurface NAPLs. The interfacial tracer technique will be usefnl in assessing changes in NAPL morphology, e.g. pre- and posttreatment of a contamination zone (Saripalli et al., 1997c, Brusseau et al., 1997). The ability to quantify the NAPL/water contact area, even only for relative measurements, offers the contaminant hydrologist an additional tool for site characterization (Annable et al., 1997). Octanoliwater and NAPWwater partition coefficients were determined in batch experiments. Column experiments were conducted to compare the breakthrough curves of the reactive tracers with the conservative tracer in different column systems (non-contaminated, contaminated - pre- and post-surfactant flushing) The specific objectives of the experiments were: (I) determination of coal tar saturation in undisturbed soil cores from a former manufactured gas plant site using different alcohols as partitioning tracers; (2) determination of the specific tar/water interfacial areas using an interfacial tracer; (3) surfactant flushing of the contaminated column for enhanced solubilization of coal tar constituents, and (4) quantification of the efficiency of the surfactant flushing using post-treatment partitioning and interfacial tracers.
2
Theoretical
background
2.1 Solute transport Hydrodynamic and physico-chemical processes were considered as the predominant factors during the tracer transport in the columns, i.e. biodegradation, volatilization etc. were neglected in the evaluation of the breakthrough curves. If a tracer is sorbed to the aquifer material, the column effluent concentration shows a time lag in breakthrough relative to the conservative tracer. The time lag between the arrival of the conservative tracer and the reactive one is referred to as retardation. A linear isotherm (partitioning) model (Jin et al., 1995) was considered to be appropriate to describe and evaluate the transport of the partitioning tracers .in the laboratory columns. The measurement of the NAPWwater interfacial areas using interfacial tracers requires the following four principal assumptions (Saripalli et al., 1997b): (1) the tracer adsorbs at the fluid- fluid interface, but does not partition into the bulk NAPL, (2) the tracer adsorbs in a monolayer at the NAPLiwater interface, (3) each tracer molecule occupies a known molecular area, and (4) the
introduction of a surface active tracer does not alter the configuration or the total volume of the NAPL in the porous medium. The interactions between the reactive tracers and NAPL in the subsurface are schematically shown in Fig. 1.
Fig.
1: interactions of partitioning and lnterfactal tracers at the
NAPL/water
mterface during the transport m a contaminated
aquifer.
2.2 Data Analysis The retardation factor of either partitioning or interfacial tracers was determined by comparing the breakthrough curves of the reactive tracers relative to the conservative The retardation of a reactive tracer is directly one. proportional to the NAPL/water partition coefficients, the NAPL saturation and the specific NAPL/water interfacial area, respectively. The retardation factor for continuous feed tracer inputs was calculated from the first moments ratio of the reactive tracer and the conservative tracer, respectively. p=i
1-g 0i
/I1
dt
(1)
where ,u, C, C, and dt denote the respective first moment, the column outflow concentration [M L”], the column inflow concentration [M L”] and the time interval [T], respectively. By applying the same equation to both the reactive and conservative tracer, the retardation factor Rd [-] is calculated according to: & =& .&
(2)
where p, and A denote the first moment of the reactive tracer (partitioning and interfacial) and the conservative tracer, respectively. The retardation factor for pulse type input tests is given by the ratio of the first moments (areas below the breakthrough curves) according to (Valocchi 1985 as cited by Saripalli et al. 1997b):
B. Setarge et al.: Non-Aqueous Phase Liquids in Natural Aquifer Material
surfactant [M mol.‘I. The concentration at the interface is calculated according to:
4t$t *, Rd = (I-L
$$dt
(3)
2
c,
where the nominator and denominator of the first term denote the first moment of the reactive and conservative tracer, respectively, and t, denotes the finite pulse duration [T] The general equation of partitioning tracer retardation based on an equilibrium mass balance can be written as: R
p””
=
I+
503
f$ +K,, & n
where R,,,,, ph, Kdr n, K,,, and S,, denote the retardation factor [-I, the bulk density [M L”], the solid/water distribution coefficient [L3 Me’], the porosity [-I, the NAPLiwater partition coefficient [-I and the NAPL saturation of the aquifer material [-I, respectively. If the tracers are not adsorbed onto the aquifer material the solid/water distribution coefficient Kd and thus the second term equals zero and the retardation factor depends only on &, and S,,.
_
[[Q~$$++w
mt -
Mn
(8)
where t, Q , PV, M,, and C,,, denote the actual time [T], the flow rate [L3 T-l], the pore volume [L3], the mass of NAPL [M] and the reactive tracer concentration [M Le3], respectively. The specific interfacial area u,,,+ [L2 L”] compares the NAPLIwater interfacial area to the total volume of the porous medium according to (Saripalli, 1997b):
The NAPL morphology index I [L2 L”] can be calculated by taking the ratio of A,,, to the volume of NAPL in the pores (reformulated from Saripalli, 1997b):
,=+
(10) n
2.3
NAPL saturation and specific interfacial area
The NAPL saturation S, of the contaminated aquifer material can be determined according to Eq. (4) by rewriting it for S,. The NAPL volume within the pores can be calculated according to:
v, = PV-1-
s
I-
s,
where PV and V,, denote the pore volume determined from the first moment of the conservative tracer [L3] and the volume of NAPL determined from the first moment of the partitioning tracer [L3], respectively. Both parameters were obtained at the same time from the respective breakthrough experiment. The NAPLiwater interfacial area A,,, [L*] was calculated from the number of adsorbed interfacial tracer (surfactant) molecules and the area which is occupied by one molecule according to:
A,,,,= A,,, x Z
(6)
where A,y,,,Iand Z denote the interfacial area covered by one surfactant molecule [L2] and the number of moles of adsorbed molecules [-I, which were calculated according to: C
Z=---N,
MWS”d
(7)
where C,,,, NA and Mw,,, denote the tracer concentration at interface [M Me’], the Avogadro’s number the [6.022 1O23mol.‘1 and the molecular weight of the
3
Materials and Methods
3.1
Aquifer material
Undisturbed non-contaminated (FSI) and contaminated (FSI,O,) soil cores from a former manufactured gas plant site at Kehl, Southwest Germany, were taken from the saturated zone at a depth of 3 m to 4 m (FSI,,,) and 7 m to 8 m (FSI), respectively. The main tar oil contamination in the field was located between 3 m and 5 m below the ground surface, just below the ground water table. The aquifer material consists of River Rhine quatemary alluvium sediments with a grain size distribution of about 43 % sand, 56 % gravel and I % silt and clay. The organic matter content, the CaC03 content and the cation exchange capacity CEC determined for the grain size fraction 0.2 mm - 2.0 mm were 0.02 %, 1.2 % - 5.2% and about I meq/lOOg, respectively.
3.2
Tar oil
Since no tar oil from the field site was available an industrial coal tar oil (Rutgers, Germany) was used in the batch experiments to determine the partition coefficients. The coal tar oil is a highly viscous, black liquid with a density between I. I g cm-3 and I .2 g cm-3. The average molecular weight of the coal tar ranges from 220 g mol.’ to 263 g mol.’ (Loyek, 1998). About 40 % of the tar oil consists of the I6 EPA polycyclic aromatic hydrocarbons
B. ‘$,etarge
504
The (PAHs). compounds.
3.3
rest
are
et al.:
insoluble
Non-Aqueous Phase Liquids in Natural Aquifer Material or not
identified
I95 mg L“ (IPA), 104 mg L” (23DM2B), 200 mg L“ (4M2P), 100 mg L-’ (24DM3P), 700 mg L-’ (Marlinat) and 400 mg L-’ to 500 mg L“ (NaCI), respectively. The solution was shaken over head for about 24 hours, to ensure the complete dissolution of the alcohols. A preliminary test showed that the NaCl concentration used had no salting out effect on the solubility of the alcohols. The interfacial tracer was shaken for 1 hour using a horizontal shaker.
Tracers
The alcohols isopropanol (IPA), 2,3Dimethyl-2-Butanol (23DM2B), 4-Methyl-2-Penthanol (4M2P) and 2,4Dimethyl-3-Penthanol (24DM3P) were used as partitioning tracers. The alcohols were purchased as liquids with a activity of 100 % from MERCK (Germany). They are characterized by a high water solubility, a density lower than water and a low volatility at room temperature. They are biodegradable and partition into NAPL and organic matter. The low volatility of the alcohols may be an advantage when conducting tracer tests in a column system with entrapped air or in field systems close to the capillary fringe. The anionic surfactant Marlinat was used as interfacial tracer. It is also highly soluble in water and has an activity of 70 % in the aqueous solution as obtained from the manufacturer (Htils, Germany). The area per molecule of Marlinat (about 60 A’) was determined from the Gibbs adsorption isotherm using surface tension data (Rosen, 1989, Saripalli et al., 1997b). It was assumed that Marlinat adsorbs in mono-layers and the area occupied by one surfactant molecule is the same at both the NAPL/water and air/water interface. No adsorption of Marlinat onto the aquifer material used could be measured in batch experiments (Danzer, 1998). The physico-chemical properties of the tracers used are summarized in Table 1.
3.4
3.5
Batch tests
The tracer stock solution was diluted into different concentrations. 18 mL of each concentration level were added to 2 mL of tar oil in 20 mL head space vials. All samples were prepared in triplicate. The samples were shaken for 72 hours using an overhead shaker. Then, the samples were centritiged and IO mL of the supematant water solution were transferred to an other 20 mL head space vial and analyzed by gas chromatography (GC) using a head space sampling unit and a flame ionization detector (FID). Vials without tar oil were treated same in order to quantify the reference concentrations. The concentrations of the reference samples were always as expected indicating no tracer loss due to system adsorption or volatilization. The same procedure was applied to determine the 0ctanoVwater partition coefficients.
3.6
Column tests
The undisturbed soil cores (FSI and FSl,J were obtained in PVC liners of 100 cm length and an internal diameter of IO cm resulting in a column volume of 7.86 L. The liners were sealed by using airtight inlet/outlet fittings and PTFE tape. PTFE tubes of an internal diameter of 1.85mm and a wall thickness of I mm were used to deliver the solution via a peristaltic pump (Ismatec MC-MS/CA4). Stainless steel capillaries of an internal diameter of 2.1 mm and a plastic flask was used as the column outflow system. An on-line
Tracer solution preparation
Glass bottles of different sizes were used for the preparation of the tracer solutions. A standard stock solution consisting of 10 mL Methanol, 2 mL IPA, 1 mL 23DM2B, 2 mL 4M2P and 1 mL 24DM3P was prepared. 2 mL of this stock solution and 0.5 g NaCI were added to one liter of deionized water resulting in tracer influent concentrations of
Table 1: Physico-chemical properties of the partitioning tracers Isopropanol (IPA). 2,3-Dimethyl-2-Butanol and 2,4Dlmethyl-3-Penthanol(24DM3P) and the interfacial tracer (IFT) Marlinat, respectively. Tracer
Formula
MW”
[g mol.‘1 IPA
Cd&O
23DM2B
Pb [g
cm-‘1
c
BP”
MP’
lg L’l
[OCI
cw
(23DM2B),
LogH’
Kw ’
1°Cl
l-l
l-l 0.4 f 0.2
60
0.79
++ 1.1
82’
- 89’
- 3.46 m
C6H140
102.2
0.80
12.5’
99’
- 115’
- 2.87”
4M2P
GHuO
102.2
0.81
17’
132’
-90’
- 2.74m
24DM3P
C7H160
116.2
0.82
0.02’
139’
-79’
1
n. d.
Marlinat
C~~HI$ZH~O)~SOJ~\~~’385
4-Methyl-2-Penthanol
K”,;”
l-l 0.9f0.6
24fI
3.1 *0.7
70 f 5
3.6 k
245fl7 CMC
(4M2P)
22.2 + 1.8
’= 950
me/L
a Molecular weight; b Density; ‘Water solubility; d Boiling point; ‘Melting point; ‘Henry’s law constant, g Octanol/water partition coefficient (measured within this study); ’ NAPWwater partition coeffxient (measured within this study), ‘Miscible with water; JStreitwieser (1994); ‘Hayden and Linnemeyer ( 1997); ’Critrcal micelle concentration; m Verschueren (1983)
505
B. Setarge et al.: Non-Aqueous Phase Liquids in Natural Aquifer Material electrical conductivity (EC) detector (Phytec Processor controlled No. 684590) was connected to the outflow system to monitor the conservative tracer breakthrough curve. The general scheme of the column experiments was as follows: The columns were saturated with de-ionized water at a low flow rate of 1.8mL min“ from the bottom to the top of the column to displace entrapped air from the cohunn. Effluent samples from the contaminated columns were collected before and after each tracer test to determine the PAH emission. The column tests were started when a constant effluent EC signal [us] and water turbidity values below 10 NTU were achieved. The tracer solution was introduced as continuous and/or pulse type input. Flow rates range between 1.8 mL min.’ and 10.4 mL mid’ with a variation of about 2 % during each test. The sampling started simultaneously to the monitoring of the breakthrough curves. The setup for the column experiments is shown in Fig. 2. A simple permeameter setup was used to measure the hydraulic conductivity and monitor the variation of the hydraulic conductivity during the surfactant The hydraulic head was breakthrough experiments. measured in polyethylene tubings at the column inlet and column outlet to obtain the headloss along the column.
3.1
Sample collection and analysis
Head space vials (20 mL) sealed by PTFE lined silicon septa were used to collect the partitioning tracer samples. The vials were filled by the stainless steel capillary of the column outflow system penetrating the septum. A hollow needle in the septum allowed the displaced air to escape. Immediately after each head space vial was tilled half (10 mL) with the column outflow solution the septum was changed by a new one to prevent losses due to volatilization. Then, the samples were analyzed by GCFID. The detector signals were acquired by a personal computer. The responses were integrated using a chromatography software (Waters Maxima) and converted into breakthrough curves. The detection limits of the partitioning tracers were 0.300 mg L“ (IPA), 0.15 mg L-’ (23DM2B), 0.300 mg L-’ (4M2P) and 0.015 mg L-’ (24DM3P), respectively. The measurement of the interfacial tracer and the surfactant concentrations were done indirectly by measuring the total organic carbon (TOC) using a TOC-Analyzer (highTOC, elementar. Hanau, Germany), that combusts the organic carbon at 950°C and detects the CO2 by non-disperse infrared detection (NDIR). The detected responses were converted to TOC and surfactant concentration, respectively.
I
Tetlon Tube
Tracer Solution
Fig. 2: Experimental setup of the column tests (modltied from Danzer, 1998): On-line electrical conductwty detector (EC) to monitor the conservative tracer NaCl Automated eftluent water samplmg of the reactive tracers
4
Results and discussion
4.1
Partitioning coefficients
The partitioning tracers show a range of measured octanol/water partitioning coefficients (K,,,,) between 0.42 (IPA) and 245 (24DM3P). The K,,,, is directly proportional to the hydrophobicity of a compound (Karickhoff, 1984). High K,, values may favor the partitioning of a tracer into soil or aquifer organic matter. Thus, using e. g. 24DM3P as a partitioning tracer in soils of high organic matter content may result in a time consuming test on the field scale. The retardation determined by that tracer test may overestimate the actual NAPL saturation if the partitioning into organic matter is not considered separately. The measured NAPL/water partition coefficients K,,,,,ranged between 0.97 (IPA) and 22 (24DM3P) and correlate reasonably well to the K,,,, values. Thus, the expected retardation of 24DM3P in column experiments should be significantly higher than the retardation of the other tracers. The partition coefftcients vary with the NAPL composition and make the comparison of partition coefftcients of different studies difficult (Hayden and Linnemeyer, 1997). Therefore, it is important to use a site representative NAPL sample to determine the partition coefficients.
B. Setarge d al.: Non-Aqueous Phase Liquids in Natural Aquifer Material
506 4.2
Tracer transport
The hydrodynamic parameters such as porosity, pore volume, dispersion coefficient, dispersivity and average linear velocity were determined from the breakthrough The results of the curves of the conservative tracer. different experiments are summarized in Table 2. The porosity (n) and hydraulic conductivity (KY) of the noncontaminated column FSI was found to be 35 % and 3.8 10e4m se’, respectively. In contrast, the corresponding values of the contaminated column FSI,Or were 15 % and 3 Joe6m se’. The smaller values may be due to the tar oil present in the pore space. The difference of dispersion coefficients determined by pulse type and continuous feed inputs, respectively, was found to be small (Table 2). 0
4.3
The breakthrough curves of the partitioning tracers and the interfacial tracer are shown in Fig. 3 and Fig. 4, respectively. The breakthrough curve of the conservative tracer NaCl is included in both figures for comparison. All reactive tracers show the same transport behavior as the conservative tracer, i.e. no retardation due adsorption onto the aquifer material or partitioning into soil organic matter could be observed.
3:
Breakthrough
partitioning column
tracers
through
FSI. The column
transported
hke
adsorption
onto
of
conservative
the undisturbed, properties
the conservative the
the
aquifer
are given
tracer, material
tracer
N&l
non-contaminated in Table
2
All tracers
I.e. no retardation or
partitioning
and
occurs
into
the
soil
soil
core were
due to organic
matter
I 1.1
0.8
FSI,,,: pre-surfactant flushing
0.7-
The breakthrough curves of the partitioning tracers and the conservative tracer through the column FSl,Or before surfactant flushing are shown in Fig. 5. The partitioning tracers are significantly retarded compared to the conservative tracer (NaCI). The retardation factors range between 1.5 (IPA) and 4.6 (24DM3P). The retardation of the different partitioning tracers seems to be significant after one pore volume exchanged, i.e. the retardation was mainly due to the tailing part of the breakthrough curves. The percentage of the sorbed tracer mass with respect to the total mass of tracer which had entered the column varies from 4 % (IPA) to I4 % (24DM3P). The calculated tar oil saturation S, of FSl,a, ranges between 17 % (24DM3P) and 34 % (4M2P. 23DM2B). The partitioning tracers IPA, 23DM2B and 4M2P yield almost the same S, values. The average tar oil volume of about 47 1 mL was calculated from the saturation S, according to Eq. (5), i.e. about half of the pore volume is filled by tar oil. It should be noted that the duration of the column experiments was the same for all tracers. The effect of flow rate variation on tracer partitioning was investigated using FSI,O, and the results are summarized in Table 3. A four fold decrease (9.4 m de’to 2.4 m d’) of the linear velocity, i.e. an increase of the mean residence time (NAPL/water contact time) of the partitioning tracers from 3 to 5 hours,
curves
4
3
vo/umes
Pore
FSI (without tar oil) Fig.
4.4
2
1
??
0.6 0.5 0.4 0.3 0.2
”
0
2
4
6
8
10
12
tracer
Marlinat
Pore Volumes Fig.
4:
Breakthrough
conservative Table
2.
The
conservatwe absence
tracer
curves
of the
NaCl
through
mterfacral
tracer
tracer
due
interfaclal
FSI. The column showed
to mteractlons
properties
no retardatron wth
the
aquifer
and
arc given
the in
compared
to the
materral
tn the
of NAPL.
resulted in a small but significant increase of the apparent tar oil saturation (6 %). Similar observations were reported by Hayden and Linnemeyer (1997). Jin et al. (1995) mentioned that the analysis of partitioning tracer breakthrough curves becomes much more complicated and subject to large errors if the flow velocity is too high to allow local equilibrium partitioning of the tracer between NAPL and water. However, the retardation factors R,, calculated from the equilibrium partition coefficients determined in this study
B. Setarge et al.: Non-Aqueous Phase Liquids in Natural Aquifer Material for different tar oil saturations S, according to Eq. (4) show reasonably good agreement with the retardation factors obtained from the breakthrough curve of the respective tracer (Table 5). The three partitioning tracers (IPA, 23DM2B and 4M2P) show almost the same values of R,, and Rdand a tar oil saturation of 30 %. This may indicate that the coal tar composition of the field site is similar the industrial coal tar (Rutgers) which was used for the equilibrium batch experiments. Contrary, the measured Rd of 24DM3P is only half the value of the expected equilibrium transport retardation factor R,,. This may indicate a rate limited partitioning of that tracer. The equilibrium retardation R,, versus NAPL saturation of the different partitioning tracers is shown in Fig. 7. The graphs show almost an exponential increase of the retardation factor with increasing NAPL saturation S,. Theoretically, the partitioning tracers would have an infinite retardation factor for complete NAPL saturation (S,, = 100 %) with an already tremendous increase for S, values above 50%. At S,, values above 50 % a strong tailing of the partitioning tracers breakthrough curves may be expected due to non-equilibrium conditions. The breakthrough curves of the interfacial tracer, the anionic surfactant Marlinat, and the conservative tracer through the column FSI,,, are shown in Fig. 6. Significant retardation of Marhnat was observed right from the beginning of the breakthrough curve compared to that of the conservative tracer NaCI. A retardation factor of 2 was determined resulting in an interfacial concentration of 4.8 mmol (kgNAPL)-‘, a specific interfacial area a,,,. of 2,320 cm-’ and an index of morphology I of 36,914 cm-‘. The results are included in Table 4. The interfacial area compared to the total volume of the column is quantified by a,,,, while the interfacial area to volume ratio of the tar oil present in the porous medium is quantified by the index of morphology I. The high values of a,,. and I may indicate a highly dispers tar oil distribution in the column FSI,,, and/or a coating of the grains by tar oil, respectively.
507
1.2 1.1 1.0 0.9 0.8 u”
0.7
3
0.6 0.5
1-e
23DM28
/
10
12
1
0.4 0.3 0.2 0.1 0.0
2
4
6
8
14
P‘ore Volumes Fig. 5: Breakthrough curves of the conservatwe tracer NaCl and the partlttonmg tracers through FS/,O, (pre-surfactant flushing) showing a significant retardation of the partitionmg tracers relative to NaCl The column properks are given in Table 2
0.04 0
2
4
6
8
10
12
14
16
fore Volumes 4.5
Surfactant flushing
The column FSI,,, was flushed with a surfactant solution consisting of 70 % non-ionic and 30 % anionic surfactant. Tar oil constituents such as PAHs were removed due to an increase in their apparent water solubility in the presence of surfactant micelles (solubilization). The effluent concentration of the 16 EPA-PAHs before the reactive tracer and surfactant flushing test ranged between 90 ug L” and 140 ug L”. It increased during the partitioning and interfacial tracer test to concentrations between 150 ug L*’ and 400 ug L-‘. The PAHs outflow concentrations during
Fig. 6: Pre- and post-surfactant flushing breakthrough mterfacial tracer Marlinat through the column FS/,o, propertles are given in Table 2.
curves of the The column
the surfactant flushing were increased by - orders of magnitude up to concentrations of about 11,000 ug L-‘. An increase of the headloss by a factor of 4 to 5 was observed during the surfactant flushing. This may be due to the precipitation of the anionic surfactant or the formation of liquid crystals. The post-surfactant flushing PAHs emission was higher compared to the pre-surfactant flushing emission.
508
B.
Table
2:
Hydrodynamic
parameters
IQ/,,,
a Flow
‘Values
Table
3:
Effect
used.
[m d-‘1
[%I
0 648
0.256
35
average
of flow
0 098
velocity;
from
rate variations
[m d-‘1
0 432
94
0 108
3.8.
saturation
24
determined
4: Results of the partitioning
by parbtioning
[m2 h-‘1
[ml
1-l
0.0167
0.065
(0.021)’
(0.08)’
0.042
043’
fLongitudinal
dispersion
15
23
coeff~clent,
s Dispersivity,
h Peclet
rate used and the tar oil present in the column.
tracers (pre-surfactant
flushing),
DL
d
Pd
sy
[m2 h?]
[m]
[-I
WI
0.079
0.20
5.0
23
d
0.265 dispersion
tracers breakthrough
043
coefficient,
’ Disperswity:
curves through
MhC
M”dd
I-l
WI
WI
IPA
1.5
32
1746
23DM2B
2.5
34
4M2P
3.1
34
24DM3P
46
I7
lFTh
2.0
RdY
Peh
nc
s,*
Tracer
at
WI
d Longitudinal
and interfacial
volume;
Due to the very low flow
of IO cm
DLf
I.1
“Pore
curves.’
IS.2 ‘Porosity;
2.356
IO”
conductwity;
I52
average velocity;
CL1
IO4
3.
d Hydraulic
v.z”
[L h-l]
PV’
[m s.‘]
pulse type breakthrough
on the NAPL
of 100 cm and an inner diameter
6”
I5
‘Porosity;
obtained
had a length
n‘
[L K’]
in brackets
a Flow rate; b Lmear
Both column
vxb
Q"
Table
of the columns
0.108
rate; b Linear
number,
Non-Aqueous Phase Liquids in Natural Aquifer Material
ef al.:
Q”
Column
FSl
Setarge
2.3 f Peclet number,
FSI,ar (pre-surfactant
s NAPL
29 saturation.
flushing) c
Mud Min.’
C/l
[msl
WI
[mmol
74
4
938
130
I4
I795
342
I9
o,,’
I8
[cm-‘]
[cm-‘]
2320
36914
(kg NAPL)-‘1
‘Retardation
factor obtained
concentration
of the interfacial
Table
5:
NAPL
saturations
Comparison
from
the breakthrough
tracer, ‘Specific
of the calculated
Tracer
&w ”
23DM2B
32 40
bNAPL
saturation;
‘Tracer
area; s Morphology
retardation
factors
index;
4.8
mass injected h Interfacial
(Req) and the measured
into the column;
dAdsorbed
tracer mass; “Adsorbed
tracer.
retardation
factors
(RJ of partitioning
tracers for different
flushmg).
R, (S.=
IPA
288 I008
curve;
mterfacral
equilibrium
S. of FSls,(pre-surfactant
904 2450
’
R,
10%)
(S,=
*
Rl/’
Rqb
20 %)
S””
(S, = 30 %)
c-1
[-I
[-I
[-I
r-1
0.9
I.1
1.2
I .4
I.5
32
WI
3.2
1.4
I8
2.4
2.5
34
4M2P
3.6’
1.4
I.9
2.5
3.1
34
24DM3P
22.2
3.5
6.6
IO5
46
I7
a NAPWwater obtained
partition
coefficient
from the breakthrough
determmed
curve; d NAPL
in equihbrium saturation;
batch
‘Hayden
systems,
b equilibrmm
and Lmnemeyer
(I 997)
retardation
factor
calculated
from
K,,;
’ Retardation
factor
FSI,,,: post-surfactant flushing
4.6
20
Table
6:
satoratIon
Results
of the partitioning .S, decreased
and mterfaclal
compared
tracers
breakthrough
to the pre-surfactant
-
24DM3P 23DM2B
16
4M2p .. . ,pA
_._._..
14 12 z oc 10 86m 42 0
0
10
20
30
40
5
5, 1w Fig.
7:
The
values
Equihbrium
equiiibrmm
retardation
of R,,
were
NAPWwater
factors
R, versus NAPL
calculated
according
to equation
partition
coefficient
determined
saturation
.S,
2-4
the
from
in the
batch
experiments.
Another reason could be that the tar oil had formed ganglia with lower surface to volume ratios during the surfactant flushing. This would result similar to the preferential flow paths in a very limited tar/water contact area which is not detected by the interfacial tracer. A third reason could be that the interfacial tracer Marlinat adsorbs only to the more liquid (less viscous) components of the tar, but not to the residual highly viscous components. The less viscous, soluble components were removed during the surfactant flushing and the interfacial tracer does not interact with the remaining residual semisolid phase.
curves
flushing
.._~__
18
The breakthrough curves of the partitioning tracers and the conservative tracer through the column FSl,Or after surfactant flushing are shown in Fig. 8. The partitioning tracers were still considerably retarded relative to the conservative tracer. The retardation factors of the partitioning tracers range between 1.4 (IPA) and 3.3 (24DM3P) allowing a calculation of about 17 % tar oil saturation. An average tar oil volume of about 225 mL was calculated from S,, according to Eq. (5). The results are summarized in Table 6. The lower values of the tar oil saturation in the column FSI,,, after surfactant flushing may indicate that a significant amount of tar oil was removed by the surfactant flushing. However, the decrease of tar oil saturation could also be due to following reasons: (1) the formation of preferential flow paths during the surfactant flushing resulting in a smaller contact area or a shorter contact time between tar oil and water. However, this should be indicated by an unusual breakthrough curve of the conservative tracer, which was not observed. (2) The partition coefficients may have changed due to changes of the post-surfactant flushing tar oil composition (Meyers et al., 1997). However, the relative differences in retardation factors of the partitioning tracers are the same pre- and post-surfactant flushing indicating that the partition coefficients did not change due to a surfactant induced composition change of the tar oil. The breakthrough curve of the interfacial tracer through KSI,,, was included in Fig. 6 and the results are included in Table 6. No retardation of the interfacial tracer was observed after the surfactant flushing, i.e. it was transported similar to the conservative tracer NaCI. This is not quite clear since the tar oil saturation should be at least about 17 % as determined by the partitioning tracers. The phenomenon may be explained by the formation of preferential flow paths as discussed above.
NAPL
509
Setarge et al.: Non-Aqueous Phase Liquids in Natural Aquifer Material
B.
tracer
through test (Table
FSl,,, (post-surfactant 4)
The
mterfaclal
flushing).
The retardation
tracer
not retarded
was
Rd and the
factor after
the surfactant
flushing.
Tr&X!r
&in post-surfactant
IPA
Sb
Sb
post-surfactant
pre-surfactant
flushmg
flushing
flushmg
I-1
WI
WI
14
28.6
32
Min kl
MO/ [mz4
2794
Mud.W”” WI
74
26
49
32
23DM2B
IS
14.3
34
1500
4M2P
I.8
16.7
34
287
145
50
24DM3P
3.3
9.5
17
1446
128
88
35000
0
0
I
IFT ’
’ Retardatmn tracer
factor
obtamed
from
the breakthrough
curve,
b NAPL
saturatmn.
‘Tracer mass InJected
I
mto the column:
d Adsorbed
tracer
mass.
’ lnterfaclal
B. Setarge et al.: Non-Aqueous Phase Liquids in Natural Aquifer Material
0.011
0
1
10
20
30
40
50
Pore Volumes Fig. 8 Breakthrough curves of the partitioning tracers through FSI,O, (post-surfactant flushing) showing a smaller retardation compared to the pre-surfactant flushing breakthrough curves.
5
Conclusions
The partitioning tracers tested seem to be suitable for the quantification of the tar oil saturation in undisturbed soil columns and the quantification of the effkiency of a surfactant flushing. The interfacial tracer gave a first indication of the tar/water interface and thus changes in geometry of the tar present in the porous medium pre- and post-surfactant flushing.
Acknowledgements. This study was funded by the Landesanstalt fUr Umweltschutz (LfU) of Baden-Wttrttemberg, Germany. The authors thank Bernice Nisch, Renate Riehle, Anne Hartmann-Renz and Renate Seehg for their technical assistance in the hydrogeochemistry laboratory
References Annable, MD, Jawtz, J W., Rao, P.S.C., Dai, D.P., Kim, H.K., and Wood, A.L , Field evaluation of interfacial and partrtioning tracers for characterization of effective NAPL-water contact areas, accepted for publication in Groundwater, 1997. Brusseau, M.L., Popowcova, J., and Silva, J.AK., Characterizing gaswater interfacial and bulk-water partitioning for gas phase transport of organic contaminants in unsaturated porous media, Enwon. .!&I Technol.. 3/(6), 1645-1649, 1997 Gary, J W., Estimating the surface area of fluid phase interfaces in porous media, J. Cam. Hydrol.. 15. 243-248, 1994
Danzcr, I., Transport of surfactant and coupled transport of PAHs and surfactants in natural aquifer material, Ph.D. Thesis, University of Tuebingen, in preparation, 1998. Hayden, N.J and Linnemeyer, H.C.. Investigation of partitioning tracer for determining coal tar saturatlon in soils, Presented at the 213th ACS National Meeting, American Chemical Society, San Francisco, California, April 13-17,37(l), 1997. Jm, M., Delshad, M., Dwarakanath, V., McKinney, D.C., Pope, GA., Sepehmoori, K., Tilburg, C.E., and Jackson, R.E., Partitioning tracer tests for detection, estimation, and remediation performance assessment of subsurface nonaqueous phase liquids, @‘a&r. Resow Res., 3/(5). 1201-1211, 1995. Karickhoff, S.W.: Organic pollutant sorption in aquatic systems, J Hydraui. Eng, IO(61, 707-735, 1984. Kim, H., Rae, P.S.C., and Annable, M.D., Determination of effectwe auwater interfaclal area in parttally saturated porous media using surfactant adsorption, Warer. Resow Res., 33(/2), 2705-271 I, 1997. Loyek. D. and Grathwohl, P., Ermitthmg und Reduzierung der Schadstoffemwon bei teer- und teeralkontamimerten Boden. I Schadstoff-Freisetzung im Kontaminationsherd. AbschluOberlcht zum PWAB-Projekt PD 94 159, Projekt Wasser-Abfall-Boden (PWAB), Kemforschungszentrum Karlsruhe, 1998. Meyers, S.L , Wright, C.L , Lee. C.M., Falta. R.W., and Coates, J.T., Effect of co-solvent flooding on partitioning tracer behavior. Presented at the 213th ACS National Meeting, Amerxao Chemical Society, San Francisco, California, April 13-17, 1997. Nelson, N.T. and Brusseau, M L., Field study of the pa&toning tracer method for detection of dense nonaqueous phase liquid in a trichloroethene-contaminated aquifer, Envrron. Ser. Technol.. 30(g). 2859-2863, 1996. Rosen, M.J., Surfactants and interfacial phenomena, 2nd Edition, John Wiley & Sons, New York, 1989. Sanpalli, K.P., Annable, M.D., and Rao, P.S.C., Estimation of nonaqueous phase liquid (NAF’L) - water interfacial areas in porous media following mobilization by chemical flooding, accepted for publication m Emron. Sci. Technol, l997a Saripalll, K.P.. Kim H., Rao, P.S C., and Annable, M.D., Measurement of specific fluid-fluid interfacial areas of immtscible fluids in porous media, Envrron. Ser. Technol, 3/(3), 932-936, 1997b. Sarlpalli, K.P., Rao, P.S.C., and Annable, M.D., Determination ofspecific NAPL-water interfacial areas of residual NAPLs m porous media, accepted for publication in J. Cont. Hydrol., 1997~. Strethvieser, A. et al., Organ&he Chemie, zweite Auflage: Verlaggesellschal? mbH, Weinheim, New York, Basel, Cambridge, Tokyo. 1309 p., 1994. Verschueren. K Handbook of enwronmental data on orgamc chemicals.second editlon, Van Nostrand Reinhold Company Inc, New York: 13lOp., 1983. Wang, P., Dwarakanath, V , Rouse, B.A., Pope, GA., and Sepehmoori, K., Partition coefficients for alcohol tracer between non-aqueous phase liquids and water from UNIFAC - solubihty method, Advances in Water Resources, 21(2/. 71 -I 8 I, 1998. Wilson, RD. and Mackay, D.M., Direct detection of residual nonaqueous phase liquid m saturated zone using SF6 as partitioning tracer, Envrron. Scr Technol., 29(j), 1255-1258, 1993.