Hongjuan Bai*,
Nelly Cochet,
Audrey Drelich,
André Pauss and
Edvina Lamy
Sorbonne universités, Université de technologie de Compiègne, ESCOM, EA 4297 TIMR, Centre de recherche Royallieu - CS 60 319, 60 203 Compiègne cedex, France. E-mail: hongjuan.bai@utc.fr; Tel: +33 344237933
First published on 25th January 2016
The transport of Escherichia coli (1.1 μm) and Klebsiella sp. (1.5 μm) were performed in three porous media with different grain and pore size distributions under saturated flow conditions to explore the coupled effect of porous size distribution and bacteria cell properties on microbial transport. A two-region mobile–immobile model that account for non-uniform transport in porous media was used to quantify the uniformity of bacteria flow pathways. Bacteria flow pathways were more non-uniform compared to those of water tracer for each porous medium. While the non-uniformity of bacteria flow pathways increased with the increasing of the physical heterogeneity of the porous media for Klebsiella sp., no clear tendency was obtained for E. coli. Different behaviors in term of E. coli and Klebsiella sp. cells retention were observed: similar retention rates were obtained in all porous media for the motile E. coli, whereas the non-motile Klebsiella sp. retention decreased in the medium that exhibited larger pores and a wide range of the pore size distribution. These results indicated that bacteria transport and retention were simultaneously dependent to both pore size distribution and bacteria cell properties.
It has widely been reported in the existing literature that bacteria transport is highly influenced by grain or pore size of the porous media. Thus many previous studies have been focused on bacteria transport in homogeneous porous media,14–16 and several publications are also available on the bacteria transport in heterogeneous porous media with different pore size geometry.17,18 Escherichia coli is the most commonly used bacteria for evaluating bacteria transport in porous media. This bacterium has been used to investigate the factors that control microbial transport in porous media. These factors include bacteria concentration,19 medium characteristics such as grain size,1 the presence of surface coatings,20 matrix structure,21 hydrodynamics properties such as pore water velocity19,22–25 and water content,26 and chemical factors such as pH and ionic strength.27–31 To understand the role of the porous grain size on bacteria transport, porous media with different grain sizes have been employed in literature studies. A porous medium constituted by grains with different sizes implies different pore sizes accessible for bacteria transport. Recent publications have demonstrated that not only pores sizes but also their distribution can strongly affect the transport and retention of colloidal particles in porous media under various conditions.32–35 While the pore size effect has been extensively studied, one drawback associated with the current body of literature is the limited number of studies examining the pore size distribution of the porous media and its effect on bacteria transport and retention.17 Thus, preferential transport of bacteria through macropores has been observed in heterogeneous porous media with simple geometry, constituted by the macropore insertion into homogenous matrix sand.25,36 Other research work has been carried out in real soils with a complicated geometry of macropores.17,37 However the difficulties associated to the control of the hydrodynamic conditions as well the difficulty to obtain an accurate description of macropores geometry makes it hard to reach conclusive results concerning bacteria transport in these complicated real porous systems. The transport of model colloid particles (latex microspheres) under laboratory conditions in porous media composed by mixing sands with different grain size have been studied by Leij and Bradford.38 These authors concluded that the relatively small sample size and the complex flow pattern in the composite medium made difficult to reach definitive conclusions regarding transport parameters for colloid transport. Besides, bacteria transport studies in aggregate media with micro- and macroporosity are very limited in the current literature. In such complex systems, solute migration is mainly controlled by inter-aggregate pores (macropores/mobile phase) in which dispersion and advection occurs and solute diffusion take place from inter-aggregate openings to intra-aggregate pores (micropores).39–41 However the existing studies in aggregate media are only limited to solute transport processes. To the best of our knowledge, there is no study on the bacteria transport and retention in aggregate porous media. Such reports underlined the need for more studies evaluating the effect of pore size distribution of porous media on bacteria transport and retention.
The factors affecting bacteria transport including cell characteristics like cell types and motility,42–44 hydrophobicity,45 cell size and shape,46 population growth44 have also been extensively studied. However, their role on bacteria transport has been mainly investigated in homogenous sandy media. And the role on bacteria transport through heterogeneous porous media has received considerably less attention.
The aim of this study was to investigate the coupled effect of bacteria cell characteristics and physical characteristic of the porous media on the microbial transport. Miscible transport experiments were performed in three porous media with different grain and pore size distributions under saturated steady state flow conditions. Two different representative cell types, Escherichia coli and Klebsiella sp. were used as biotic colloids for transport experiments. Breakthrough curves of bacteria were measured and numerically simulated using a two-region mobile–immobile model,47 which account for non-uniform transport in heterogeneous porous media. Mass balance calculations and the final retained bacteria in the column after transport experiments, deduced from experimental observations, as well fitted model transport parameters were to compare the transport of two bacteria.
The zeta potential of each porous medium, measured by a Zetasizer (3000 HAS, Malvern Instruments Ltd, UK), reached −39.6 ± 1.8 mV for Fontainebleau sand, –20.5 ± 1.8 mV for Compiègne sand, and −12.5 ± 1.8 mV for the gravel.
The background electrolyte solution for the bacterial characterization and transport experiments consisted of 0.1 mmol L−1 NaCl solution (pH = 5.89). To characterize the hydrodynamic properties of the porous media, 0.01 mol L−1 KBr solution (for both Fontainebleau and Compiègne sand) and 0.05 mol L−1 NaCl solution (for the gravel) were used as a conservative tracer.
The bacterial cells were harvested from the nutrient broth in their early stationary phase (6 h for E. coli and 7 h for Klebsiella sp.) by centrifugation (Eppendorf, Centrifuge 5810R) (4000 rpm, 10 min, 4 °C). Then they were washed twice with a 0.1 mmol L−1 NaCl (Fisher Scientific) solution (pH = 5.89) and re-suspended in an identical NaCl solution. The same 0.1 mmol L−1 NaCl solution was also used as the background electrolyte solution for the transport experiments.
Each bacteria suspension with a known concentration was prepared with distilled water, adjusted with 0.1 mmol L−1 NaCl solution in this study. This step allows providing a good estimation of the total bacteria mass balance, partitioned between the effluent and soil particles. The actual bacterial concentrations in the influent solution were determined using the method of bacteria enumeration on the nutrient agar plates after incubation at 37 °C overnight52 to monitor for exudates formation and possible cell aggregation. The optical density of the bacterial suspension was measured before and after the experiments. No changes in the optical density was observed, which indicated that the bacterial suspension remained stable over the duration of each transport experiment.53
The zeta potential which governs colloid stability56 was measured by dynamic light scattering (Zetasizer 3000 HAS, Malvern Instruments Ltd, UK) for both bacteria at ionic strength of 0.1 mmol L−1 NaCl. The measurements were conducted in triplicates for each cell suspension. The zeta potential values of the cells and porous media permitted the determination of DLVO interaction parameters and interaction energy profiles, which were calculated using the approach presented by Redman et al.57 The Hamaker constant was set to 6.5 × 10−21 J for bacteria.57
The hydrophobicity adhesion to hydrocarbon (MATH) approach was used to determine the hydrophobicity of both bacterial strains.58,59 The test was performed under the following conditions: the bacteria were harvested at early stationary phase by centrifugation and the bacteria were washed twice with phosphate buffer (pH = 7.2) and the total cell number was determined by counting on agar plates. Then 3 mL of bacteria suspension was mixed with 0.3 mL of hexadecane (Fisher Scientific) and the mixture was vortexed during 2 minutes. After the phases were clearly separated, counting was performed on DEV agar plates containing the sample from the aqueous phase. The fraction partitioned to the hydrocarbon phase was calculated from the difference between the total cell number and the remaining cell number of the aqueous phase. The analysis was performed in triplicate for each sample.
Prior to each experiment, the column was flushed upward under saturated conditions with about 3 pore volumes of the background electrolyte solution at a steady Darcy velocity of 0.42 ± 0.01 cm min−1 using a peristaltic pump (ISMATEC, IDEX corporation). Then the flow was reversed and the column was rinsed with about 10 pore volumes before starting the transport experiments. The solution chemistry conditions were verified by determining both conductivity and pH of the effluent solutions.
A short pulse of tracer solution (20 mL) was injected into each column experiment, followed by 0.1 mmol L−1 NaCl washing solution to background levels. The effluent conductivity was continuously measured to follow the tracer breakthrough using a conductivity meter (SevenMulti, METTLER TOLEDO) and then converted to tracer concentration. The pore water volume for each experiment, measured be weighting the column before and after transport experiment, remained the same. This indicated no change in the total porosity of the porous media, indicating no porous media property alteration.
For bacteria transport experiments, a 20 mL pulse of bacterial suspension (≈108 CFU mL−1) was injected into each column experiment followed by the background electrolyte solution at the same flow rate as for the tracer experiments. The optical density at 600 nm was continuously measured at the column outlet by using a spectrophotometer (Perkin Elmer, Lambda 25). The absorbance bacteria breakthrough was then converted to concentration in order to monitor bacteria breakthrough curves (BTC). The total number of retained bacteria was determined for all columns after transport experiments. In this case, the saturated porous medium was carefully excavated in 9 layers and placed into 9 vials containing excess sterile 0.1 mmol L−1 of NaCl solution. Then the vials were slowly shaken for 15 minutes to liberate any reversibly retained bacteria. Finally, the bacterial concentrations in the excess solution were determined by plate counting. Water and porous media filled vials were placed in an oven (110 °C) overnight to volatilize the remaining solution from porous media. The volume of water and the mass of the dry porous media in each vial was determined from mass balance by measuring the weight of empty vials, water and porous media filled vials, and porous media filled vials. The overall bacteria mass recovery (Mtotal) was subsequently determined as the sum of the amount of bacteria recovered in the effluent (Meff) and the amount of bacteria retained in the porous medium (Mretained). All experiments were conducted in triplicate. The experimental set-up used for transport experiments is shown in Fig. 1 and the overview of experimental conditions is shown in Table 1.
Tracer/Bacteria | Replicate | Initial concentrationa | Porosity (%) | Bulk density (g cm−3) | Pulse time (min) | Darcy velocity (cm min−1) | |
---|---|---|---|---|---|---|---|
a Tracer concentration is given in mol L−1 and bacteria concentration in CFU mL−1.b The values given in parentheses are the standard deviations of triplicate columns. | |||||||
Tracer | Fontainebleau sandy column | Exp. 1 | 0.01 | 34.5 | 1.74 | 5.48 | 0.439 |
Exp. 2 | 34.2 | 1.74 | 5.48 | 0.436 | |||
Exp. 3 | 33.8 | 1.75 | 5.43 | 0.442 | |||
Average | 34.2 (0.35)b | 1.74 (0.01) | 5.46 (0.03) | 0.439 (0.003) | |||
Compiègne sandy column | Exp. 1 | 0.01 | 45.2 | 1.46 | 5.63 | 0.428 | |
Exp. 2 | 43.9 | 1.50 | 5.67 | 0.427 | |||
Exp. 3 | 44.0 | 1.50 | 5.67 | 0.424 | |||
Average | 44.4 (0.723) | 1.49 (0.023) | 5.66 (0.023) | 0.426 (0.002) | |||
The gravel column | Exp. 1 | 0.05 | 79.0 | 0.52 | 5.65 | 0.419 | |
Exp. 2 | 78.1 | 0.55 | 5.68 | 0.413 | |||
Exp. 3 | 78.5 | 0.53 | 5.58 | 0.428 | |||
Average | 78.5 (0.451) | 0.53 (0.015) | 5.64 (0.051) | 0.420 (0.008) | |||
E. coli | Fontainebleau sandy column | Exp. 1 | 9.08 × 108 | 35.0 | 1.72 | 5.55 | 0.416 |
Exp. 2 | 5.62 × 108 | 34.3 | 1.74 | 5.62 | 0.423 | ||
Exp. 3 | 6.29 × 108 | 34.4 | 1.74 | 5.55 | 0.426 | ||
Average | 6.99 × 108 (1.84 × 108) | 34.6 (0.379) | 1.73 (0.012) | 5.57 (0.040) | 0.422 (0.005) | ||
Compiègne sandy column | Exp. 1 | 9.20 × 108 | 44.2 | 1.49 | 5.55 | 0.418 | |
Exp. 2 | 7.08 × 108 | 44.7 | 1.48 | 5.50 | 0.429 | ||
Exp. 3 | 8.40 × 108 | 44.8 | 1.47 | 5.47 | 0.431 | ||
Average | 8.23 × 108 (1.07 × 108) | 44.6 (0.321) | 1.48 (0.01) | 5.51 (0.040) | 0.426 (0.007) | ||
The gravel column | Exp. 1 | 5.60 × 108 | 77.6 | 0.56 | 5.68 | 0.421 | |
Exp. 2 | 6.52 × 108 | 77.7 | 0.56 | 5.50 | 0.423 | ||
Exp. 3 | 5.30 × 108 | 77.2 | 0.57 | 5.67 | 0.424 | ||
Average | 5.81 × 108 (0.64 × 108) | 77.5 (0.265) | 0.56 (0.006) | 5.62 (0.101) | 0.423 (0.002) | ||
Klebsiella sp. | Fontainebleau sandy column | Exp. 1 | 5.58 × 108 | 34.7 | 1.73 | 5.68 | 0.415 |
Exp. 2 | 4.84 × 108 | 33.3 | 1.77 | 5.67 | 0.416 | ||
Exp. 3 | 5.98 × 108 | 33.8 | 1.75 | 5.68 | 0.416 | ||
Average | 5.47 × 108 (0.58 × 108) | 33.9 (0.709) | 1.75 (0.02) | 5.68 (0.006) | 0.416 (0.0006) | ||
Compiègne sandy column | Exp. 1 | 6.36 × 108 | 45.2 | 1.46 | 5.63 | 0.425 | |
Exp. 2 | 4.08 × 108 | 43.9 | 1.50 | 5.55 | 0.424 | ||
Exp. 3 | 4.52 × 108 | 44.0 | 1.50 | 5.60 | 0.423 | ||
Average | 4.99 × 108 (1.21 × 108) | 44.4 (0.723) | 1.49 (0.023) | 5.59 (0.040) | 0.424 (0.001) | ||
The gravel column | Exp. 1 | 5.59 × 108 | 79.0 | 0.52 | 5.68 | 0.418 | |
Exp. 2 | 6.00 × 108 | 78.1 | 0.55 | 5.68 | 0.414 | ||
Exp. 3 | 6.70 × 108 | 78.5 | 0.53 | 5.50 | 0.428 | ||
Average | 6.10 × 108 (0.56 × 108) | 78.5 (0.451) | 0.53 (0.015) | 5.62 (0.104) | 0.420 (0.007) |
![]() | (1) |
![]() | (2) |
![]() | (3) |
The governing equations of MIM model used in this study are written as:61,62
![]() | (4) |
![]() | (5) |
![]() | (6) |
Due to the limited pore volume of bacteria injection into the columns we assumed that bacteria retention to the porous media was an irreversible process. Thus the detachment coefficients were neglected in this work.
The MIM model was fitted to the experimental BTCs with the proper initial conditions and boundary for the column transport experiments by the HYDRUS-1D code. The code allowed us to fit simultaneously the parameters of λ, α and θm (via θim) for tracer data, and the parameters λ, α, θm (via θim) and katt for bacteria data using the inverse solution. The dispersivity of the medium λ (cm) which was assumed to be an intrinsic characteristic can be determined as follows:47 λ = (Dmθm)/q. The mobile water fraction (θm/θ) was estimated for each transport experiment to characterize the flow uniformity, were θm = θ − θim (θ is the total water content).
Zeta potential (mV) | Energy barrier (kT) | Secondary minimum depth (kT) | Secondary minimum separation (nm) | |
---|---|---|---|---|
Bacteria | Porous media | |||
−41.1 ± 0.65 (E. coli) | −39.6 ± 1.8 (F) | 1586 | 0.40 | 300 |
−20.5 ± 1.8 (C) | 814 | 0.49 | 287 | |
−12.5 ± 1.8 (G) | 43.5 | 2.8 | 27 | |
−33.2 ± 0.29 (Klebsiella sp.) | −39.6 ± 1.8 (F) | 1778 | 0.58 | 309 |
−20.5 ± 1.8 (C) | 909 | 0.63 | 311 | |
−12.5 ± 1.8 (G) | 40.2 | 4.6 | 28 |
![]() | ||
Fig. 5 Measured (symbols) and fitted (lines) breakthrough curves of (a) tracer, (b) E. coli and (c) Klebsiella sp. through three porous media for triplicate columns. |
The hydrodynamic parameters with their confidence intervals for all porous media were obtained from HYDRUS simulations, using the physical non-equilibrium model (MIM). A good fitting of the modeled BTCs (Fig. 5a, lines) to the experimental BTCs (Fig. 5a, symbols) with high regression coefficient (R2 > 0.98) was obtained for all transport experiments (Table 3). The mean mobile fractions of the total water volume, θm/θ, accessible for convective tracer transport were higher for Fontainebleau sandy columns (96.1%), compared to those obtained for Compiègne sandy (79.4%) and gravel (81.7%) columns (Table 3). The lower θm/θ values obtained for both Compiègne sand and gravel implied that in these porous media, a smaller pore water volume than that in the Fontainebleau sand was required for solute transport. These results are in agreement with those obtained by Lamy et al.65 These authors reported θm/θ values of 71.0% for the same heterogeneous gravel. The MIM-derived dispersivity (λ) for chloride was in the same order of magnitude as the grain diameter of porous media (Table 3). Thus, higher dispersivity values were obtained for Compiègne sandy (0.82 cm) and gravel (1.93 cm) media compared to the Fontainebleau sand (0.14 cm). This was expected because the dispersivity increased with the increasing of the physical heterogeneity of the porous media.66 In accordance to these results, Lamy et al. reported similar dispersivity values of 1.97 cm for the same gravel under saturated conditions.47 The solute exchange rate (α) was much higher for the gravel compared to both sands (Table 3). However, it is difficult to compare solute exchange rate values obtained in different porous media, as this parameter is highly dependent to the geometry of the pores and pore water velocity. High dispersivity values and asymmetrical shape of tracer BTCs with tailing obtained for both Compiègne sand and gravel columns confirmed non-uniform flow in these media. This may be explained by the existence of mobile water regions with high velocity and immobile water regions that do not permit convective flow. Thus, a part of the water tracer could preferentially fill the pore regions with high velocity and move through these regions quickly, while other part of the tracer may diffuse into the immobile water regions. Because of the concentration gradient, the tracer in immobile water regions could slowly diffuse into the mobile water regions causing the “tailing” shown in the breakthrough curves of tracer for both Compiègne sand and gravel media.
Replicate | MB (%) | Retardation factor | λ (cm) | θm/θ (%) | α (min−1) | R2 | |||
---|---|---|---|---|---|---|---|---|---|
Value | S.E. Coeff.b | Value | S.E. Coeff. | Value | S.E. Coeff. | ||||
a The values given in parentheses are standard deviations of triplicate columns.b Refers to the standard error coefficient obtained from HYDRUS-1D simulations. | |||||||||
Fontainebleau sandy column | |||||||||
Exp. 1 | 99.4 | 0.98 | 0.19 | 9.86 × 10−3 | 94.8 | 7.54 × 10−4 | 1.03 × 10−4 | 3.38 × 10−4 | 0.996 |
Exp. 2 | 99.4 | 1.01 | 0.11 | 1.68 × 10−3 | 97.3 | 1.23 × 10−3 | 1.91 × 10−4 | 3.51 × 10−4 | 0.999 |
Exp. 3 | 99.2 | 1.01 | 0.11 | 2.86 × 10−3 | 96.1 | 3.52 × 10−4 | 4.83 × 10−4 | 1.31 × 10−4 | 0.999 |
Average | 99.3 (0.12)a | 1.00 (0.02) | 0.14 (0.05) | 96.1 (1.25) | 2.59 × 10−4 (1.99 × 10−4) | 0.998 (0.002) | |||
![]() |
|||||||||
Compiègne sandy column | |||||||||
Exp. 1 | 97.7 | 1.06 | 0.80 | 9.00 × 10−2 | 77.6 | 8.45 × 10−4 | 5.19 × 10−2 | 1.01 × 10−2 | 0.985 |
Exp. 2 | 99.2 | 1.05 | 0.84 | 6.13 × 10−2 | 81.7 | 5.89 × 10−3 | 1.93 × 10−2 | 2.08 × 10−3 | 0.991 |
Exp. 3 | 98.3 | 1.07 | 0.80 | 4.53 × 10−2 | 78.7 | 5.91 × 10−4 | 3.45 × 10−2 | 3.01 × 10−3 | 0.990 |
Average | 98.4 (0.75) | 1.06 (0.01) | 0.82 (0.02) | 79.4 (0.02) | 3.53 × 10−2 (1.6 × 10−2) | 0.989 (0.003) | |||
![]() |
|||||||||
The gravel column | |||||||||
Exp. 1 | 100 | 0.97 | 1.95 | 1.33 × 10−1 | 81.5 | 2.59 × 10−2 | 3.40 × 10−1 | 6.53 × 10−2 | 0.993 |
Exp. 2 | 99.7 | 0.99 | 1.84 | 1.89 × 10−1 | 79.6 | 2.30 × 10−2 | 4.60 × 10−1 | 1.00 × 100 | 0.990 |
Exp. 3 | 100 | 1.19 | 2.01 | 1.25 × 10−1 | 84.1 | 1.49 × 10−2 | 8.60 × 10−3 | 1.33 × 10−3 | 0.997 |
Average | 99.9 (0.17) | 1.05 (0.12) | 1.93 (0.083) | 81.7 (2.2) | 2.69 × 10−1 (2.33 × 10−1) | 0.993 (0.004) |
Replicate | Meff | Mretained | Mtotal | Retardation factor | λ (cm) | θm/θ (%) | α (min−1) | katt. (min−1) | R2 | ||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Value | S.E. Coeff.b | Value | S.E. Coeff. | Value | S.E. Coeff. | Value | S.E. Coeff. | ||||||
a The values given in parentheses were standard deviations.b Refers to the standard error coefficient obtained from HYDRUS-1D code.c The values are not valid because of the large standard deviations compared with the two other experiments. | |||||||||||||
Fontainebleau sandy column (for E. coli transport) | |||||||||||||
Exp. 1 | 46.6 | 41.1 | 87.7 | 1.13 | 0.52 | 3.32 × 10−1 | 88.6 | 7.44 × 10−2 | 1.85 × 10−2 | 4.71 × 10−2 | 0.105 | 2.29 × 10−3 | 0.989 |
Exp. 2 | 45.2 | 41.3 | 86.5 | 1.17 | 0.60 | 2.69 × 10−2 | 86.3 | 1.23 × 10−2 | 1.32 × 10−2 | 1.61 × 10−3 | 0.118 | 1.23 × 10−2 | 0.989 |
Exp. 3 | 53.9 | 36.2 | 90.1 | 1.10 | 0.37 | 9.64 × 10−2 | 83.3 | 1.51 × 10−2 | 1.46 × 10−2 | 7.75 × 10−3 | 0.094 | 1.10 × 10−3 | 0.991 |
Average | 48.5 (4.7)a | 39.5 (2.9) | 88.1 (1.8) | 1.13 (0.04) | 0.49 (0.12) | 86.1 (2.7) | 2.99 × 10−2 (1.60 × 10−2) | 0.107 (0.011) | 0.990 (0.001) | ||||
![]() |
|||||||||||||
Compiègne sandy column (for E. coli transport) | |||||||||||||
Exp. 1 | 50.6 | 41.6 | 92.2 | 0.66 | 0.15 | 2.90 × 10−1 | 64.2 | 1.20 × 10−2 | 1.21 × 10−1 | 7.24 × 10−2 | 0.192 | 1.27 × 10−2 | 0.972 |
Exp. 2 | 39.2 | 49.9 | 89.1 | 0.76 | 0.30 | 9.37 × 10−2 | 62.7 | 4.14 × 10−2 | 1.46 × 10−1 | 7.26 × 10−2 | 0.212 | 7.90 × 10−3 | 0.994 |
Exp. 3 | 41.8 | 43.1 | 86.9 | 0.71 | 0.72 | 3.45 × 10−1 | 65.6 | 1.66 × 10−2 | 1.18 × 10−1 | 1.28 × 10−2 | 0.224 | 3.31 × 10−2 | 0.984 |
Average | 43.8 (6.0) | 44.9 (4.4) | 89.4 (2.7) | 0.71 (0.05) | 0.39 (0.30) | 64.2 (1.5) | 1.28 × 10−1 (1.54 × 10−2) | 0.211 (0.016) | 0.983 (0.011) | ||||
![]() |
|||||||||||||
The gravel column (for E. coli transport) | |||||||||||||
Exp. 1 | 52.4 | 37.1 | 89.5 | 0.48 | 0.50 | 5.27 × 10−3 | 83.3 | 2.10 × 10−3 | 1.15 × 10−2 | 9.78 × 10−4 | 0.156 | 1.16 × 10−3 | 0.996 |
Exp. 2 | 49.3 | 37.9 | 87.2 | 0.45 | 0.51 | 7.90 × 10−2 | 85.2 | 1.01 × 10−2 | 9.71 × 10−3 | 5.26 × 10−3 | 0.181 | 1.83 × 10−3 | 0.996 |
Exp. 3 | 43.3 | 43.6 | 86.9 | 0.44 | 0.57 | 5.57 × 10−2 | 86.2 | 6.43 × 10−3 | 8.91 × 10−3 | 3.23 × 10−3 | 0.230 | 1.20 × 10−3 | 0.998 |
Average | 48.3 (4.6) | 39.5 (3.5) | 87.9 (1.4) | 0.46 (0.02) | 0.52 (0.04) | 84.9 (1.5) | 1.24 × 10−2 (5.31 × 10−3) | 0.187 (0.041) | 0.997 (0.001) | ||||
![]() |
|||||||||||||
Fontainebleau sandy column (for Klebsiella sp. transport) | |||||||||||||
Exp. 1 | 60.5 | 32.0 | 92.5 | 1.64 | 0.236 | 1.66 × 10−2 | 83.9 | 4.55 × 10−3 | 6.17 × 10−3 | 2.61 × 10−3 | 0.043 | 3.06 × 10−3 | 0.926 |
Exp. 2 | 42.5 | 42.5 | 85.0 | 1.46 | 0.11 | 7.79 × 10−2 | 86.1 | 2.07 × 10−2 | 4.26 × 10−2 | 9.81 × 10−3 | 0.075 | 1.93 × 10−3 | 0.905 |
Exp. 3 | 40.2 | 43.7 | 83.9 | 1.38 | 0.13 | 1.34 × 10−2 | 83.1 | 1.42 × 10−2 | 7.56 × 10−2 | 1.86 × 10−2 | 0.087 | 1.97 × 10−3 | 0.914 |
Average | 47.7 (11.1) | 39.4 (6.4) | 87.1 (4.7) | 1.49 (0.13) | 0.16 (0.068) | 84.4 (1.6) | 4.15 × 10−2 (3.47 × 10−2) | 0.068 (0.023) | 0.915 (0.010) | ||||
![]() |
|||||||||||||
Compiègne sandy column (for Klebsiella sp. transport) | |||||||||||||
Exp. 1c | 70.8 | 25.5 | 96.3 | 0.80 | 0.74 | 1.96 × 10−2 | 79.1 | 1.46 × 10−3 | 4.90 × 10−3 | 3.90 × 10−4 | 0.068 | 3.67 × 10−3 | 0.998 |
Exp. 2 | 44.1 | 40.8 | 84.9 | 0.78 | 0.13 | 3.95 × 10−2 | 76.1 | 4.32 × 10−3 | 1.28 × 10−2 | 2.25 × 10−3 | 0.159 | 9.38 × 10−3 | 0.956 |
Exp. 3 | 50.4 | 41.2 | 91.6 | 0.84 | 0.38 | 3.36 × 10−2 | 74.0 | 3.16 × 10−3 | 1.03 × 10−2 | 1.51 × 10−3 | 0.146 | 1.51 × 10−3 | 0.998 |
Average | 55.1 (13.9) | 35.8 (9.0) | 90.9 (5.7) | 0.81 (0.03) | 0.42 (0.31) | 76.4 (2.5) | 9.00 × 10−3 (4.00 × 10−3) | 0.124 (0.049) | 0.984 (0.024) | ||||
![]() |
|||||||||||||
The gravel column (for Klebsiella sp. transport) | |||||||||||||
Exp. 1 | 93.9 | 4.5 | 98.4 | 0.50 | 0.22 | 2.04 × 10−2 | 71.7 | 1.70 × 10−3 | 3.78 × 10−2 | 1.30 × 10−3 | 0.0076 | 1.04 × 10−3 | 0.990 |
Exp. 2 | 98.0 | 1.2 | 99.2 | 0.46 | 0.11 | 1.53 × 10−2 | 70.1 | 1.21 × 10−3 | 4.24 × 10−2 | 4.30 × 10−3 | 0.0020 | 3.50 × 10−4 | 0.988 |
Exp. 3 | 83.9 | 13.7 | 97.6 | 0.46 | 0.42 | 3.71 × 10−2 | 72.4 | 3.89 × 10−3 | 4.60 × 10−2 | 2.88 × 10−3 | 0.0265 | 1.35 × 10−3 | 0.992 |
Average | 92.0 (7.3) | 6.5 (6.5) | 98.4 (0.8) | 0.47 (0.02) | 0.25 (0.15) | 71.4 (1.2) | 4.21 × 10−2 (4.11 × 10−3) | 0.0121 (0.012) | 0.990 (0.002) |
The same physical non-equilibrium model was used to simulate E. coli (Fig. 5b, lines) and Klebsiella sp. (Fig. 5c, lines) BTCs for all porous media. A good fitting of MIM-model to experimental BTCs (with regression coefficients higher than 0.91) permitted to obtain bacteria transport parameters (Table 4).
Lower θm/θ values were obtained for both E. coli (86.1%) and Klebsiella sp. (84.4%) in Fontainebleau sand (Table 4) than that of the tracer (Table 3), indicating that lower pore water volumes were required for bacteria transport, comparing to those of the water tracer. Similar to water tracer, θm/θ values decreased in the more heterogeneous Compiègne sand compared to the homogenous Fontainebleau sand for both bacteria (from 86.1% in F columns to 64.2% in C columns for E. coli and from 84.4% in F columns to 76.4% in C columns for Klebsiella sp.) (Table 4). Similar tendency was obtained for Klebsiella sp. transport in the gravel (θm/θ values decreased to 71.4% in G columns). However this tendency was not confirmed for E. coli transport in the gravel. Quite the same values (84.9%) as for the most homogenous sand (86.1%) were obtained in this medium for E. coli.
The MIM-derived dispersivity (λ) values of E. coli and Klebsiella sp. through Fontainebleau sandy column (Table 4) were in the same order of magnitude as those of tracer (Table 3). Klebsiella sp. presented a lower dispersivity (0.16 cm) compared to E. coli (0.49 cm). However, the dispersivity values of both bacteria through Compiègne sandy and gravel columns (Table 4) were smaller than those of tracer (Table 3). This indicated that bacteria accessed a more restricted part of the pore network and followed a different flow path compared to tracer, because of the size exclusion effect. Lower dispersivity of bacteria compared to water tracer has also been reported by Pang et al.62
Under certain experimental conditions, bacteria may plug or alter the flow. Bacteria may be retained in the porous media, reducing the pore space available for water flow. In addition, bacteria growth during transport experiments may cause biofilm formation, which in turn may modify the permeability. However, under the current experimental conditions of this work bacteria did not plug or alter the flow. The overall mass balances lower than 100% (Table 4) for each bacterium indicated no bacteria growth during transport experiments.
Similar to tracer results, the increase of the pore size and total porosity of sandy media with similar pore size distributions resulted in a decrease of θm/θ values for Klebsiella sp., causing non-uniform transport in the most heterogeneous sand, in accordance with literature studies (Fig. 6). The increase of the total porosity of porous media resulted in a decrease of θm/θ values for this strain, enhancing preferential transport (low θm/θ) in the heterogeneous gravel (Fig. 6). Klebsiella sp. recovery in the effluent, Meff, increased with increasing preferential transport, θm/θ, and porosity of the porous media. These results indicated that Klebsiella sp. collected at the column outlet travel through larger interconnected pores with high pore velocity and is excluded from a portion of the total porosity related to the smallest pores. Similar to these results, Bradford et al. who reported that grain and pore size distribution were positively correlated, and the presence of larger pores resulted in enhanced colloid transport.76
Preferential E. coli transport increased from homogenous (F) sand to heterogeneous one (C) with higher porosity, but this non-uniform transport did not result in an increase of the E. coli mass recovery in the effluent, contrary to what has been observed for Klebsiella sp. in the same sands (Fig. 6). Indeed, no linear relationship between non-uniform transport, porosity and mass recovery in the effluent was obtained for this strain. Similar E. coli pathways, confirmed by similar θm/θ values, were obtained for both the most homogenous sand and the gravel, even though these two media differed in term of pore sizes as well as pore size distribution and total porosity. In addition, similar E. coli recovery was obtained in the effluent for all porous media, contrary to literature results of Bradford et al.76 and to Klebsiella sp. in this study.
The differences obtained between the two bacteria indicated that under these experimental conditions the pore size distribution is not the predominant factor to explain the differences in transport parameters. It should be noted that E. coli is a motile strain, while Klebsiella sp. is a non-motile one. This characteristic should be involved in bacteria transport behavior in porous media. The results obtained by de Kerchove and Elimelech showed that the ability of the cell to swim is an important factor that enhances the transport. They hypothesized that cell motility allows the upstream swimming of bacteria and subsequent cell deposition on regions which are otherwise inaccessible to non-motile cell deposition.43 The diffusion coefficient for motile bacteria has been reported to be up to three orders of magnitude greater than non-motile bacteria.77 This may explain the lower mass recovery of the motile E. coli in the heterogeneous gravel, compared to the non-motile Klebsiella sp. The small pores of this medium may not be accessible for the non-motile bacteria, but the motile E. coli may have access due to its swimming motility. This may increase the possibility to be trapped in these regions, causing low recovery in the effluent.
Both bacteria exhibited a different behavior in term of retention in the porous materials. Similar mean retention (Mretained) was obtained for E. coli in all porous media (Table 4), indicating no influence of the pore size distribution in the retention under the experimental conditions of this work. Conversely, non-motile Klebsiella sp. retention was highly dependent on the pore size distribution of the porous media (Table 4). The mean retention of Klebsiella sp. decreased with the increasing degree of porous media heterogeneity: higher retention was obtained in the Fontainebleau sand (39.4%) compared to 35.8% in Compiègne sand and 6.5% in gravel, presenting a wide range of the pore size distribution (Fig. 2). Similar conclusion was obtained by Bradford et al. who reported that grain and pore size distribution were positively correlated, and the presence of larger pores resulted in enhanced colloid transport.76 However this tendency was not confirmed for motile E. coli.
Results similar to experimental observations were obtained from numerical simulations. Similar MIM-fitted katt values with the same order of magnitude were obtained for E. coli in all porous media (Table 4). For each porous media, the MIM-fitted katt value of E. coli was larger than that of Klebsiella sp. (Table 3), suggesting that there was greater attachment of E. coli than Klebsiella sp. The motility of cells may impact the retention behavior. Thus, higher collisions of E. coli cells to the grains surface than for Klebsiella sp. may be expected due to its motility. This may effectively enhance E. coli “diffusion”, thereby resulting in a higher attachment of this strain. The DLVO calculations (Table 2) also confirmed these findings. Repulsive electrostatic interactions with the three porous media were greater for Klebsiella sp. compared to E. coli, which should greatly reduce the potential of Klebsiella sp. to be retained. This was also in accordance with the observations of Becker et al. who reported greater attachment rate for motile bacteria in comparison with their non-motile mutants in the coated and clean beads column study.42
Other studies suggested a relationship between flow uniformity and colloid retention in porous media. Thus, Lamy et al.47 found that the increasing of flow uniformity promote colloid retention (latex particles of 1 μm diameter). A more uniform flow means more pores accessible to the flow and thus more surface of contact between the colloids and the matrix. The improvement of such contact means a higher possibility of colloid entrapment. Conversely, non-uniform or preferential flow pathways disfavor colloid retention. In this work lower retention was obtained for the non-motile Klebsiella sp. in porous media exhibiting more preferential flow pathways (i.e. gravel), results which are in accordance to what has been reported in the literature for colloids. However contradictory results were obtained for the motile E. coli. Similar retention values were obtained for E. coli for all porous media, indicating no obvious relationship between flow uniformity and motile bacteria retention.
– Water flow was highly dependent to the physical heterogeneity of the porous media. More non-uniform and dispersive flow patterns occurred in both heterogeneous sand and gravel media compared to those of the homogenous sand. However, similar flow pathways obtained for both heterogeneous sand and gravel, even though these two media differed in term of pore size distribution, showed that it is hazardous to reach conclusive results, regarding water flow based only on the pore size distribution of the porous media. Other factor like the connectivity of the pores should be investigated to provide a better characterization of water flow processes.
– Bacteria flow pathways differed from water flow pathways. Bacteria transport occurred through more preferential flow pathways compared to the water tracer. The preferential Klebsiella sp. transport increased with the increasing of the physical porous media heterogeneity. Higher amount of bacteria mass recovery in the effluent with increasing preferential transport indicated that Klebsiella sp. transport occurred through larger interconnected pores with high pore velocity and was excluded from a portion of the total porosity related to the smallest pores. Preferential transport reduced non-motile Klebsiella sp. retention in the porous medium, by reducing the contact between bacteria and retention sites. But this trend was not confirmed for the motile E. coli. No linear relationship between non-uniform transport, porosity, mass recovery in the effluent, and retention was obtained for this strain. These differences in bacteria behavior should be linked to bacteria characteristics, like motility, which greatly affect transport properties that even big differences in the physical heterogeneity of the porous media may not compensate.
This journal is © The Royal Society of Chemistry 2016 |