Environ. Sci. Technol. 2006, 40, 6038-6043
PAHs and Organic Matter Partitioning and Mass Transfer from Coal Tar Particles to Water KARIM BENHABIB, MARIE-ODILE SIMONNOT,* AND MICHEL SARDIN LSGC- Laboratory of Chemical Engineering Science, CNRS-INPL, 1, rue Grandville BP 20451, 54001 Nancy Cedex, France
The coal tar found in contaminated soils of former manufactured gas plants and coking plants acts as a longterm source of PAHs. Organic carbon and PAH transfer from coal tar particles to water was investigated with closedlooped laboratory column experiments run at various particle sizes and temperatures. Two models were derived. The first one represented the extraction process at equilibrium and was based on a linear partitioning of TOC and PAHs between coal tar and water. The partition coefficient was derived as well as the mass of extractable organic matter in the particles. The second model dealt with mass transfer. Particle diffusion was the limiting step; organic matter diffusivity in the coal tar was then computed in the different conditions. A good consistency was obtained between experimental and computed results. Hence, the modeling of PAH migration in contaminated soils at the field scale requires taking into account coal tar as the source-term for PAH release
Introduction Sites of former manufactured gas plants (MGPs) and coking plants are contaminated by coal tars derived from coal processing (1). Coal tars are multicomponent nonaqueous phase liquids (NAPLs) containing hundreds of aromatic compounds, including polycyclic aromatic hydrocarbons (PAHs), heterocyclic PAHs, phenolic compounds, BTEX, etc. (2). Most of these organics are known to be toxic, and in some cases, carcinogenic or mutagenic (3). Due to their low aqueous solubility and their resistance to degradation, coal tars are persistent in soils and behave as long-term contamination sources (4), threatening sensitive targets such as groundwater and living organisms. The understanding of PAH release from coal tars into water is essential for risk assessment and for the development of remediation processes (1, 5-7). PAH release into water is controlled by phase equilibria and mass transfer. Thermodynamic theory of phase equilibria requires the equality of fugacities for components across phases. For NAPL/aqueous phase systems, the equilibrium concentration of an aromatic compound in the aqueous phase is most often predicted by Raoult’s law (1, 8-10). Assuming ideality, Raoult’s law relates the aqueous concentration of a compound to its mole fraction in the organic phase and the aqueous solubility of the pure liquid or sub cooled liquid. But its applicability may be questionable because of (i) the uncertainty in coal tar composition due to * Corresponding author phone : +33 (0)383 175 260; fax: +33(0)383 322 975; e-mail:
[email protected]. 6038
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 19, 2006
the difficulty of analytical characterization, (ii) the uncertainties in thermodynamic data, -fugacity ratio- (9), and (iii) the assumption of ideality of the organic multicomponent mixture (11). Most contributions consider that PAHs are transferred from an organic liquid phase into water, and few of them take into account the presence of a solid phase. But in some cases, the dissolution of the most soluble compounds may enrich the NAPL in the less soluble ones resulting in their solidification. Peters et al. (12) examined the thermodynamics of PAH-NAPL solidification and observed an ideal behavior for some components (acenaphtene, naphthalene) and a nonideal one for others (phenanthrene and 2-methylnaphtalene). Mass transfer from coal tars under nonequilibrium conditions has also been widely investigated: the limiting step was the transfer through a film forming at the NAPL/ water interface (10, 11, 13-17). In brief, in the cited contributions, except (12), coal tar was viewed as a liquid phase. Also, to our knowledge, there are no published contributions on the release of PAHs from solid coal tar particles into water. As solid coal tars may be found at high rate in contaminated sites, it is important to understand PAH release from this source. Here, we propose an original approach to examine this process, including an experimental study and model development. Water extraction experiments were run with coal tar particles collected from a former coking plant located in the Lorraine district (North East of France). Experiments were undertaken to measure the transfer to water of PAHs and dissolved organic mater (DOM), knowing that PAHs are usually associated to DOM (18). Two models were developed to interpret the experimental data as well as to (i) infer equilibrium partition coefficients and (ii) determine the rate-limiting step of PAH transfer into water.
Experimental Section Coal Tar Composition. Coal tar pieces (2 kg) were sampled at a former coking plant located in the Lorraine District (North East of France). Their composition was investigated by elemental analysis, analysis of the organic extract (obtained by accelerated solvent extraction with dichloromethane) by gas chromatography-mass spectrometry (GC-MS) (19, 20), and Fourier transform infrared microspectroscopy (µFTIR) (21). The results were given in detail in ref 22. Coal tar was composed of 95 wt % organic matter and 5 wt % minerals, mainly calcite coming from the soil in which coal tar was buried. Elements were C (85.64 wt %), H (5.27 wt %), and heteroatoms (1.32 wt % N; 3.46 wt % S; 4.31 wt % O). The organic matter was highly condensed (low H/C ratio). The organic extract contained aromatic hydrocarbons (88.9 wt %), polar compounds (10.9 wt %) and saturated hydrocarbons (1.0 wt %). The aromatic fraction contained a wide diversity of polycyclic aromatic hydrocarbons, including the 16 PAHs of the U.S. Environmental Protection Agency list (except naphthalene) as well as O- and S-heterocycles (e.g., dibenzofuran and dibenzothiophene). Numerous polar compounds were detected, with N and O heterocycles of high molecular weight. Saturated hydrocarbons mainly contained high molecular weight n-alkanes (C21 to n-C35). These results were comparable to those of Zander et al. (2). Particle Preparation. Coal tar pieces were split into cubes (side: 5 cm), solidified in liquid nitrogen, crushed, sieved into 3 granulometric classes: C1 (0.9-1.1 mm); C2 (1.9-2.1 mm); C3 (2.9-3.1 mm), and stored at 4 °C. SEM observation and image analysis showed particle sphericity (Figure 1), the equivalent diameters being centered on 1, 2, and 3 mm with 10.1021/es0600431 CCC: $33.50
2006 American Chemical Society Published on Web 08/25/2006
to the concentration of organic matter in this phase. The model was based on the following assumptions: (i) linear partitioning of the extractable organic matter between the coal tar particles and the aqueous phase; (ii) equilibrium was reached at steady state; (iii) particle size and mass remained constant. Thus at steady state, for extraction n°i, partitioning yielded the following:
m/ei ) KC/ei
FIGURE 1. SEM photograph of a 2 mm diameter coal tar particle a standard deviation of 5%. Sphericity was assigned to interfacial tension between air and coal tar material. Water Extraction Experiments. The water extraction experiments were run in a column in which the fluid was re-circulated at a flow rate high enough to consider that this system was analogous to batch experiments (23). The setup was composed of a magnetically stirred recirculation glass tank; a peristaltic pump (Gilson-Minipuls 3); a jacketed glass column (i.d.: 1.6 cm, height: 7 cm, particle weight: 10 g) (Amersham Biosciences, XK column) filled with coal tar particles; and a thermo stated bath to monitor the temperature of the cooling or heating water. Each experiment was run as follows: the glass tank initially contained 100 mL deionized water set at the chosen temperature. From t ) 0, the column was continuously fed upward and the column effluent returned into the tank. The flow rate was 20 mL min-1, so that the mixing time was 5 min (ratio of the solution volume to the flow rate). The solution was monitored by discrete OD measurements at 254 nm (U-2000 spectrophotometer, Hitachi). The samples (2 mL) were reinjected into the tank immediately after OD measurement. The flow was stopped as a steady OD was reached and the solution was analyzed. Each experiment was successively run three or four times with the same column, respectively, with C1 to C3 at 20 °C. The first extractions were run with C1-C3 at three temperatures: 9, 20, and 30 °C. Analyses. The parameters measured at steady-state were pH (PHM210 MeterLab Tacussel), TOC (TOC meter Appollo 9 000), and the concentrations of the 16 PAHs (HPLC KontronBiosciences) after solvent extraction (Soxhlet). HPLC conditions were as follows: Prosphere column (250 mm × 4.6 mm, packing diameter 5 µm); diode array detector (Bio-Teak instruments) (wavelength: 254 nm); mobile phase: acetonitrile/water at 1.2 mL min-1; elution gradient: 10 min isocratic mode 50/50 (v/v); 30 min gradient mode from 50/ 50 to 90 /10; 15 min isocratic mode 90/10 and 7 min of isocratic stabilization 50/50 (v/v). The concentrations of the 16 PAHs were then added to obtain the total PAH concentration. OD at 254 nm was linearly correlated to TOC concentration and to the total PAH concentration, which justified the choice of OD measurement to monitor the continuous transfer of dissolved organic matter from the particles into the aqueous phase.
Model development Partitioning Model. We considered the system as a biphasic system including a solid phase (coal tar particles) and an aqueous phase, the transferred solute being the extractable organic matter initially located inside the particles. The total PAH concentration in the aqueous phase was proportional
(1)
m/ei is the average concentration of extractable organic matter in the coal tar particles at equilibrium at the ith step (kg kgcoal tar-1); K is the partition coefficient (m3 kgcoal tar-1), and C/ei is the DOM concentration in the aqueous phase at equilibrium with the coal tar particles at the ith step (kg m-3). The mass balance for extraction n°i (i g 1) yielded the following: / mei-1 ) m/ei + λC/ei
λ ) V/M
(2)
/ me0 is the average initial concentration of water extractable organic matter in the coal tar particles (kg kgcoal tar-1), M is the mass of coal tar particles (kg), and V is the aqueous phase volume (m3). The partition coefficient was then obtained from eqs 1 and 2:
C/ei K)λ / , ∀i 〉 1 Cei-1 - C/ei
(3)
To be consistent, the model must provide the same K value at each extraction step. This condition being checked, the / model enables us to compute me0 . Linear Driving Force Mass Transfer Model. Lumped kinetics at the first extraction step was described by a firstorder equation, namely the linear driving force model. Mass balance on water extractable organic matter yielded the following:
dm(t) ) kMSp[C(t) - C*(t)] dt
M
(4)
/ m(0) ) me0 and C(0) ) 0 as initial condition. m(t) is the average concentration of extractable organic matter in the particles at t (kg kgcoal tar-1); kM is the overall mass transfer coefficient (m s-1); Sp is the external particle surface (m2); C(t) is the DOM concentration in the aqueous phase at t (kg m-3); C*(t) is the DOM concentration in the aqueous phase that would correspond to the partitioning equilibrium with m(t) (kg m-3). Linear partitioning implied that
C/(t) )
m(t) K
(5)
The global mass balance at t yielded the following: / me0 ) m(t) + λC(t)
(6)
Assuming spherical and monodisperse particles, the characteristic time of lumped kinetics tM (s) was defined as follows:
tM )
FpKdp 6kM
VOL. 40, NO. 19, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
(7) 9
6039
Fp is the particle density (kg m-3) and dp is the particle diameter (m). Combining eq 2 for i ) 1 and eq 5-7, and integrating with C(0) ) 0 as initial condition led to the following:
C(t) ) Ce1(1 - e-t/τ) τ)
(8a)
λ t λ+KM
(8b)
τ is the characteristic time of the overall process (s).
then τ ) tM
(9a)
λ then τ ) tM K
(9b)
If λ〉〉K
FIGURE 2. OD evolution for three successive extractions run at 20 °C with 3 mm diameter coal tar particles
or else
{
The initial and boundary conditions are the following:
If λ〈〈K
0 e r e rp,t〈0,CCT )
Thus the characteristic time of the global process τ equalled the characteristic mass transfer time tM provided that V/M was far higher than K. If V/M was far lower than K, then τ would be lower than tM, meaning that the overall process would be faster than the diffusion process. Eqs 8 and 9 enabled us to determine tM for each experiment. Then tM was correlated to diffusion parameters, considering two contrasted limiting steps: film or particle diffusion. (i) Limiting Step: Film Diffusion. The mass-transfer coefficient at the external water film was calculated by the Kunii-Levenspiel correlation, conventionally used for fixed bed, 3 < Re < 3 000 (24):
Sh )
kfdp ) 2 + 1.8Re1/2Sc1/3 Dm
(10)
Sh is the Sherwood number; Re ) u0 dp/ν is the Reynolds number; Sc ) νw/Dm is the Schmidt number; Dm is the molecular diffusivity of the organic matter in water (m2 s-1); kfis the mass transfer coefficient at the external water film (m s-1); u0 is the average Darcy fluid velocity (m s-1); is the bed voidage (-); and νw is the water kinematic viscosity (m2 s-1) -assumed equal to the solution one. Computing kf enabled us to determine the characteristic time of film diffusion for spherical particles tMfd (eq. (11)) and compare its value to tM.
tMfd )
FpKd2p
(11)
6ShDm
/ me0 Fp
∂CCT )0 ∂r m(t) KC(t) r ) rp,t g 0,CCT ) ) Fp Fp
r ) 0,t g 0,
(13)
According to Crank (25), the analytical solution of eq 12 was
C(t) / Ce1
)1-
∞
6R(R + 1)exp(-DCTq2nt/d2p)
n)1
9 + 9R + q2nR2
∑
(14)
with R ) λ/K and qn the nonzero roots of
tan qn )
3qn 3 + Rq2n
(15)
Fifty values of qn were determined by the Newton method to reach a good fitting between the curve calculated from eq 14 and experimental points. DCT was obtained by least-square regression. For the sake of simplicity, we also used a first-order model. According to Villermaux (26), the first-order model could fit the first Fick equation, by equaling the first-order moment of the transfer functions in the Laplace domain of both models. Then, for spherical monodisperse particles, the characteristic transfer time of particle diffusion would be
tMpd )
d2p 60DCT
(16)
Results and Discussion (ii) Limiting Step: Particle Diffusion. Considering particle diffusion as the limiting step, we used the solution of the diffusion equation (eq 12) for solute desorption from spherical particles in which the concentration is assumed to be uniform and constant into a wellstirred solution initially free from solute.
[
]
∂CCT ∂2CCT 2 ∂CCT ) DCT + ∂t r ∂r ∂r2
(12)
with CCT is the volume concentration of water extractable organic matter in the coal tar particle at r and t (kg m-3); r is the distance from the particle centre (m) (0 e r e rp); rp is the particle radius (m); DCT is the diffusivity of the water extractable organic matter inside a coal tar particle (m2 s-1). 6040
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 19, 2006
Partitioning Experimental results. An example of OD evolution was plotted for three successive water extractions at 20 °C (Figure 2): (i) steady-state was reached in each case and (ii) the OD value on the plateau decreased as the number of successive extractions increased, exhibiting thus an exhaustion of extractable organic constituents of the coal tar particles. pH, TOC, and PAH concentration were measured at steady state (Table 1). TOC was considered to represent DOM in the aqueous phase. Linear relationships were obtained between (i) TOC and OD and (ii) the total PAH concentration and OD (Figure 3), and thus between TOC and PAH concentration. The differences in the straight line slopes were assigned to coal tar sampling. The wt ratio of the total PAH concentration to TOC was between 3.2 and 3.5 w %, which was very low compared to the ratio of the 16 PAHs in the organic compounds of the coal tar particles. This result
TABLE 1. Steady-State Parameters and Global Transfer Time τ for Three Successive Extractions Run at 20 °C with the Three Granulometric Classes
TABLE 2. Steady-State Parameters, Mass Transfer Times τ and tM and Diffusivity DCT for the First Extractions Run at Three Temperatures with the Three Granulometric Classes
dp (mm)
extraction step
OD
(PAH) (mg L-1)
TOC (mg L-1)
pH
τ (h)
dp T PAH TOC (mm) (°C) OD (mg L-1) (mg L-1)
1 1 1 2 2 2 3 3 3
1 2 3 1 2 3 1 2 3
0.863 0.733 0.598 0.862 0.731 0.607 0.862 0.733 0.649
0.986 0.832 0.680 0.883 0.725 0.586 0.840 0.691 0.618
28.56 24.15 20.53 27.55 23.30 19.70 25.27 21.38 18.15
8.47 8.49 8.51 8.69 8.71 8.74 8.51 8.53 8.55
0.48 0.48 0.48 1.91 1.91 1.91 4.29 4.29 4.29
clearly showed that PAHs were hardly extracted from the coal tar particles by water. Despite this, their concentrations in water were higher than the maximum contaminant levels of the U.S. and European standards. pH was always between 8.4 and 8.7, because of calcite dissolution (the dissolved mass was low enough to prevent a modification of the particle geometry). The steady-state parameters obtained for the first extraction step at various temperatures and particle sizes are listed in Table 2. Extraction drastically increased with temperature. At a given temperature, particle size had no effect on OD at steady state. Analyses revealed a slight variability of TOC and PAH concentration at different particle sizes, which was assigned to sampling discrepancies. Equilibrium Parameters. The organic matter partition coefficient K at 20 °C was computed from eq 3. at each extraction step with the data of Table 1 and λ ) 9.71 10-3 m3 kg-1. A constant K value was obtained whatever the particle diameter (Table 3): K ) 5.36 m3 kg-1 ( 1%. This result attested the consistency of the partitioning model, especially the validity of the linearity assumption. The concentration of water-extractable organic matter in the coal tar particles / was then deduced (eq 2): me0 ) 1.73 × 10-3 kg kgcoal tar-1 (
FIGURE 3. For the three granulometric classes (a) PAH against OD at 254 nm (b) TOC against OD at 254 nm.
1 2 3 1 2 3 1 2 3
30 30 30 20 20 20 9 9 9
1.43 1.43 1.43 0.86 0.86 0.86 0.28 0.29 0.29
1.63 0.98 0.36 1.52 0.88 0.29 1.43 0.84 0.26
1.630 1.520 1.430 0.986 0.883 0.840 0.362 0.299 0.262
pH
τ (h)
tMpd (h)
DCT (m2 s-1)
8.62 8.52 8.52 8.47 8.69 8.51 8.41 8.37 8.47
0.27 1.11 2.47 0.47 1.91 4.28 0.94 3.80 8.53
1.75 7.18 15.97 3.04 12.35 27.67 6.08 24.57 55.18
2.65 2.58 2.61 1.52 1.50 1.51 0.76 0.75 0.75
TABLE 3. Equilibrium Concentration of Dissolved Extractable Organic Matter Ci/ and Partition Coefficient Computed after the ith Extraction Step with the Three Granulometric Classes at 20 °C i 1 1 2 2 3 3 dp Ci K Ci K Ci K (mm) (mg L-1) (m3 kg-1) (mg L-1) (m3 kg-1) (mg L-1) (m3 kg-1) 1 2 3
28.56 24.15 20.530
0.0533 0.0533
27.55 23.30 19.70
0.0532 0.0531
25.27 21.38 18.15
0.0534 0.0546
5%. This very low content of water extractable organic matter (
