Journal of Contaminant Hydrology 250 (2022) 104054
Contents lists available at ScienceDirect 
Journal of Contaminant Hydrology 
journal homepage: www.elsevier.com/locate/jconhyd 
Evaluating low-cost permeable adsorptive barriers for the removal of 
benzene from groundwater: Laboratory experiments and 
numerical modelling 
Franklin Obiri-Nyarko a,*, Jolanta Kwiatkowska-Malina b, Samuel Kwame Kumahor c, 
Grzegorz Malina d 
a Groundwater Division, CSIR-Water Research Institute, P. O. Box M 32, Accra, Ghana 
b Department of Spatial Planning and Environmental Sciences, Faculty of Geodesy and Cartography, Warsaw University of Technology, Pl Politechniki 1, 00-661 
Warsaw, Poland 
c Department of Soil Science, University of Ghana, P.O. Box LG 25, Legon, Ghana 
d Department of Hydrogeology and Engineering, Geology AGH University of Science and Technology, Al. Mickiewicza 30, 30-059 Krakow, Poland   
A R T I C L E  I N F O   A B S T R A C T   
Keywords: Permeable adsorptive barriers (PABs) consisting of individual (compost, zeolite, and brown coal) and composite 
Permeable sorption barrier (brown coal-compost and zeolite-compost) adsorbents were evaluated for their hydraulic performance and 
Compost effectiveness in removing aqueous benzene using batch and column experiments. Different adsorption isotherms 
Brown coal and kinetic models and different formulations of the equilibrium advection-dispersion equation (ADE) were 
Zeolite 
Benzene evaluated for their capabilities to describe the benzene sorption in the media. The batch experiments showed that 
Analytical modelling the adsorption of benzene by the adsorbents was favourable and could be adequately described by the Freundlich 
and Langmuir isotherms and the pseudo-second-order kinetic model. Particle attrition and structural reorgani-
zation occurred in the columns, possibly introducing preferential flow paths and resulting in slight changes in the 
final hydraulic conductivity values (4.3 × 10− 5 cm s− 1–1.7 × 10− 3 cm s− 1) relative to the initial values (4.2 ×
10− 5 cm s− 1–2.14 × 10− 3 cm s− 1). Despite the fact that preferential flow appeared to have an impact on the 
performance of the investigated adsorbents, the brown coal-compost mixture proved to be the most effective 
adsorbent. It significantly delayed benzene breakthrough (R = 29), indicating that it can be applied as a low-cost 
effective adsorbent in PABs for sustainable remediation of benzene-contaminated groundwater. The formulated 
transport models could fairly describe the behaviour of benzene in the investigated media under dynamic flow 
conditions; however, model refinement and additional experimental studies are needed before pilot/full-scale 
applications to improve the fits and verify the benzene removal processes. Our results generally demonstrate 
how such studies can be useful in evaluating potential reactive barrier materials.   
1. Introduction and Chen, 2010; Bahadar et al., 2014). As a result, its removal from 
environmental systems is necessary to avoid these health issues. 
Benzene (C6H6) is a naturally occurring volatile organic compound Among the existing groundwater remediation methods, the perme-
(VOC) and one of the contaminants commonly found in groundwater able reactive barrier (PRB) technology has gained popularity in recent 
worldwide. Its occurrence in groundwater is often due to underground times because it is relatively cheap and can be used to remove a wide 
leaks and surface spills of petroleum products (Liu et al., 2010). High variety of contaminants, including heavy metals (e.g., lead, cadmium, 
amounts of benzene exposure in humans via consumption of ground- and arsenic), chlorinated solvents (e.g., tri- and tetra- chloroethylenes), 
water can cause serious health problems such as cancer, mucosal irri- petroleum hydrocarbons (e.g., benzene, toluene, ethylbenzene, and 
tation, haematological abnormalities, permanent damage to the central xylene) and nutrients (e.g., phosphorus and nitrate) from groundwater 
nervous system, respiratory issues, and liver and kidney disorders (Liang (Obiri-Nyarko et al., 2014). The PRB technology generally involves the 
* Corresponding author. 
E-mail addresses: fobiri-nyarko@csir.org.gh (F. Obiri-Nyarko), jolanta.kwiatkowska@pw.edu.pl (J. Kwiatkowska-Malina), skumahor@ug.edu.gh (S.K. Kumahor), 
gmalina@agh.edu.pl (G. Malina).  
https://doi.org/10.1016/j.jconhyd.2022.104054 
Received 24 January 2022; Received in revised form 19 July 2022; Accepted 23 July 2022   
Available online 28 July 2022
0169-7722/© 2022 Elsevier B.V. All rights reserved.
F. Obiri-Nyarko et al.                                                                                                                                                                        J o  u  r n  a  l o  f  C  o  n  t a  m  i n  a  n  t  H  y  d  r o  l o g  y  250 (2022) 104054
emplacement of a barrier filled with a reactive material(s) across the numerical modelling has been performed to predict the long-term per-
flow trajectory of the contaminant plume. As the contaminated formance of PRBs, aid the evaluation of the materials and the design of 
groundwater flows passively through the barrier under the influence of the PRB as well as elucidate the contaminant removal mechanisms. For 
the natural hydraulic gradient, the contaminants are removed (via example, Rabideau et al. (2005) employed the advective-dispersive- 
mechanisms such as sorption, biodegradation, ion exchange, precipita- reactive equation (ADRE) with a zero concentration gradient imposed 
tion, etc.), allowing treated groundwater to emerge downstream of the on the exit boundary of the PRB to determine the PRB thickness. Obiri- 
barrier (Henderson and Demond, 2007; Obiri-Nyarko et al., 2014). Nyarko et al. (2015) performed geochemical modelling using PHREEQC 
Several forms/designs of the PRB technology have emerged since its to predict the long-term performance of a zeolite-PRB to treat lead- 
inception, and these are based on factors such as the employed removal contaminated groundwater. A numerical model based on Visual MOD-
processes and the nature of the contamination. Some variants include FLOW was also developed to evaluate: (i) flow changes due to the 
permeable sorption barrier (PSB), permeable adsorptive barrier (PAB), installation of diverse PRB systems to propose the optimum configura-
permeable reactive bio-barriers (PRbB), and permeable reactive multi- tion, length and thickness, and (ii) impacts of changes in the reactive 
barrier (PRmB) (Erto et al., 2011; Obiri-Nyarko et al., 2014; Freidman material’s hydraulic conductivity overtime on key design parameters: 
et al., 2017; Zawierucha and Nowik-Zajac, 2019). capture zone, residence time and discharge rate (Grajales-Mesa et al., 
In PRBs, adsorption is one of the mechanisms commonly used to 2020). Yang et al. (2021) also recently developed a mathematical model 
remove organic compounds such as benzene from groundwater. This based on Faraday’s law, which integrates iron surface passivation to 
owes to the fact that it allows the utilization of a wide spectrum of ad- describe the porosity change of a hypothetical Fe0-based PRB. Most of 
sorbents, including natural and synthetic materials, as well as low-cost these models are based on the classical advection-dispersion equation 
and eco-friendly waste materials and by-products (e.g., Simpson and (ADE), which assumes instantaneous sorption and a constant dis-
Bowman, 2009; Faisal et al., 2021). Moreover, adsorption has proven to persivity, irrespective of the travel distance (Mahdipanah et al., 2022). 
be highly efficient for removing contaminants at low concentrations In the present study, low-cost adsorbents including compost, natural 
(Simantiraki and Gidarakos, 2015). Furthermore, the adsorption process zeolite, and brown coal as well as their mixtures were evaluated as PABs 
does not produce detrimental derivatives and also allows the recovery of for the removal of benzene from groundwater. Although some of these 
the used adsorbent (Erto et al., 2010). So far, different adsorbents have sorbents have been studied individually, their mixtures have neither 
been tested with different removal efficiencies and adsorption charac- been assessed in a single PAB nor under dynamic conditions. We hy-
teristics. Faisal et al. (2021) studied benzene adsorption in a cement kiln pothesized that the combination of some of these materials may lead to 
dust (CKD)-PRB and found a non-linear sorptive behaviour of benzene, improved removal efficiency and hydraulic performance of the PAB. 
which was best described by the Langmuir isotherm and the pseudo- Batch tests were initially conducted to analyze the individual materials, 
first-order (PFO) kinetic model. They also attributed the adsorption of followed by column studies to evaluate the reactivity of the materials 
benzene by the CKD to electrostatic interaction between the adsorbent (individual as well as their mixtures) and their hydraulic characteristics. 
and benzene. Suponik (2010) reported that the adsorption of benzene The applicability of two analytical solutions of the ADE to describe 
onto granular activated carbon (GAC) largely obeyed the Freundlich benzene behaviour in the investigated media was also evaluated. 
isotherm. The adsorption of Benzene, Toluene, Ethylbenzene, and 
Xylene (BTEX) in a natural zeolite-PRB was also studied by Vaezihir 2. Materials and methods 
et al. (2020). They noted that the barrier efficiency started to decline 
after 132 h due to the occupation of the adsorption sites by the BTEX 2.1. Reactive materials 
molecules. Vignola et al. (2011) also reported an excellent adsorption 
efficiency (96%) for BTEX in a surfactant modified zeolite (ZSM-5)-PRB. Compost was obtained from food and plant waste while brown coal 
Due to the generally high cost of some of these adsorbents, efforts are was acquired from a lignite opencast depot located at Konin, Poland. 
being made to identify/develop low-cost substitutes, preferably eco- Zeolite was obtained from a private company in Poland and contained a 
friendly waste materials and by-products. substantial (75%) content of clinoptilolite. Other properties of the 
Many studies have shown that the operational life of the PRB can be studied materials are reported in Table 1. Briefly, the pH of the materials 
truncated, owing to early exhaustion of the capacity or passivation- was determined in water (1 part of the material to 2.5 parts of water) 
induced loss of the reactive medium (Kamolpornwijit et al., 2003; using a multifunctional computer meter (Elmetron Cx-742) while the 
Obiri-Nyarko et al., 2014; Yang et al., 2021). Some researchers have also gravimetric method (Topp, 1993) was applied for the moisture content 
reported changes in the pore geometry (i.e. increase or reduction of (MC) determination based on Eq. (1): 
porosity) due to reorganization of grains by the infiltrating solution, M
dissolution/degradation of media, biofouling or production of second- MC% W= × 100 (1)  M
ary products during PRB operation (e.g., Eykholt et al., 1999; Kamol- DS
pornwijit et al., 2003; Henderson and Demond, 2007; Vignola et al., where: Mw (g): mass of water = Mws (mass of moist material) – MDs 
2011). Obiri-Nyarko et al. (2014) reviewed reactive materials used in (mass of dry solid); MDS (g): mass of dry solid = (mass of moisture can +
PRBs and noted that when mixtures of materials are used, they some- dried material) – (mass of clean moisture can). 
times tend to show antagonistic (inhibitory) instead of synergistic The bulk density (ρb) of the materials was determined using Eq. (2). 
(stimulatory) effects. Given the above issues, it is imperative to thor-
oughly assess proposed reactive materials for their effectiveness and to 
identify possible issues that can affect the performance of the PRB before Table 1 
field-scale application. Physico-chemical properties of the studied materials.  
The use of laboratory-scale experiments (batch and column) to Parameters Materials 
evaluate reactive materials for PRBs is common (e.g., Waybrant et al., Zeolite Compost Brown coal 
2002; Chen et al., 2011; Obiri-Nyarko et al., 2020; Faisal et al., 2021). 
Batch tests are used for screening whereas column experiments allow the pH(H2O) 7.13 8.05 4.90 ρb (g cm− 3) 0.80 0.69 1.13 
applicability of the materials to be assessed under simulated field con- CEC [mmol(+) kg− 1] 435.5 480.0 1215.0 
ditions (Waybrant et al., 2002). For instance, the fate and transport of Grain size (mm) 2.0–2.5 1.0–2.5 0.2–0.5 
contaminants in PRBs as well as the hydraulic performance of reactive foc (%) 0.03 0.20 0.17 
materials can be studied concurrently using column techniques (Bilardi ρb: bulk density; CEC: cation exchange capacity; foc: fraction of organic carbon; 
et al., 2016; Obiri-Nyarko et al., 2020). In some cases, mathematical/ and MC: moisture content. 
2
F. Obiri-Nyarko et al.                                                                                                                                                                        J o  u  r n  a  l o  f  C  o  n  t a  m  i n  a  n  t  H  y  d  r o  l o g  y  250 (2022) 104054
The samples were poured into a cylindrical glass with known volume, Vb respectively, when log qe is plotted against log Ce. 
(cm3), and weight, Cm (g). Mb (g) represents the weight of the cylindrical The Langmuir isotherm is valid for monolayer adsorption, and it 
glass filled with the material (Al-Shammary et al., 2018). assumes that the adsorbent has a finite adsorption capacity. Moreover, 
M C all the sites on the adsorbent are assumed to be energetically and ste-ρ b − mb = (2) V rically independent of the sorbed quantity (Osagie and Owabor, 2015; b Mohammadi et al., 2017). Eq. (6) is the linear form of the Langmuir 
The protocol described by Thorpe (1973) was employed to deter- isotherm model used in this study. 
mine the cation exchange capacity (CEC) of the materials. Two (2) g of ( )
each sample was soaked in 100 mL 0.5 N HCl for H+ to displace the Ce 1 Ce= + (6)  
cations in the materials. Thereafter, the samples were soaked in 100 mL qe qmaxb qmax
of barium acetate [0.5 N Ba(OAc)2] to displace the H+ by saturating the 
exchange sites with barium (Ba2+). The suspension was filtered and the where: b is the Langmuir adsorption constant and qmax is the maximum 
samples were washed with water. The combined Ba(OAc) filtrate adsorption capacity (mg g
− 1). The slope and intercept of the straight line 
2 +
wash filtrates solution was titrated with sodium hydroxide (0.1 N NaOH) Ce/qe versus Ce were used to calculate qmax and b of the Langmuir 
to a phenolphthalein endpoint (pH 8.0). The CEC was calculated from isotherm. Furthermore, the Langmuir adsorption constant b (L mg
− 1) 
≈
the number of moles of the NaOH consumed. The fraction of organic was used to determine the separation factor (RL) (Hall et al., 1966), 
carbon (foc) was calculated from the organic carbon (OC) content of the 
which is expressed as: 
samples, which was determined using the loss-on-ignition (LOI) method. 1RL = (7) This method estimates organic matter (OM) based on a gravimetric 1 + bC0
weight change associated with high-temperature oxidation of OM. In the The value of R indicates the nature of the adsorption process. R < 1 
present study, the samples were initially oven-dried at 105 ◦C, after L L suggests favourable adsorption, while R > 1 indicates unfavourable 
which they were ignited in a muffle furnace at 450 ◦C for 4 h. The per L adsorption. 
cent weight loss represented the OM-LOI (% wt. loss), which was con- To obtain additional information about the adsorption processes, the 
verted to OC using the van Bemmelen factor of 1.724 (Sutherland, 1998; experimental data were further analyzed with kinetic models, including 
Wright et al., 2008). the pseudo-first-order (PFO) (Eq. (8)), pseudo-second-order (PSO) (Eq. 
(9)) (Ho and McKay, 1998), and the intra-particle diffusion (Eq. (10)) 
2.2. Batch adsorption isotherms and kinetics experiments (Weber and Morris, 1963) models. 
k1
Batch adsorption experiments were carried out in duplicate using a log(qe − qt) = logqe − t (8)  2.303
150 rpm orbital shaker. Two (2) g of the materials was put in 250 mL 
amber bottles containing benzene solution with initial concentrations t 1 1
ranging from 2 to 50 mg benzene L− 1. No headspace was left in the = 2 + t (9)  qt k2qe qe
bottles to avoid losses via volatilization. After agitation samples were 
√̅̅
taken and analyzed for benzene. Kinetic experiments were also carried qt = kid t+C (10)  
out in duplicate using 2 g of sorbents and benzene-contaminated water 
(20 mg L− 1). Samples were taken at different time intervals (20, 40, 60, where: q and q are the amount of benzene adsorbed (mg g− 1t e ) at time t 
80, 100, and 120 min) and analyzed for benzene. The amount of benzene and equilibrium, respectively. k1 is the rate constant of the PFO kinetic 
adsorbed per unit weight of the adsorbent at equilibrium qe was calcu- model (min
− 1), which is determined from the slope of the line log (qe-qt) 
lated using Eq. (3), while Eq. (4) was used to determine the amount vs t; k2 is the PSO rate constant (min− 1). 1 1k q2 and q are the intercept and 2 e e 
removed at a particular time. the slope of the straight line, respectively. k2q2 e is the initial or instan-
( )
C − C taneous sorption rate (mg g
− 1 min− 1) (Ho and McKay, 1998); kid is the 
q 0 ee = v (3)  m intra-particle diffusion rate constant (mg g
− 1 min-1/2), which can be 
obtained from the slope of the straight line qt versus t1/2. C is a constant 
( )
C − C related to the thickness of the boundary layer (Itodo et al., 2010). 
q v 0 tt = (4)  m
2.3. Column experiment 
where: qe is the amount of benzene adsorbed (mg g− 1) at equilibrium; C0 
and Ce are the initial and equilibrium benzene concentrations in mg L− 1, The column experiments were implemented to study the transport of 
respectively; m is the mass of the adsorbent (g); v is the volume of the benzene in the media (both individual and mixtures) under dynamic 
solution (L); q is the amount sorbed at a time (t) (mg g− 1t ), and Ct is the conditions. The materials were air-dried and packed into the columns in 
concentration at time (t) (mg L− 1). incremental steps based on the materials’ dry density and volume of the 
The linear forms of the Freundlich and Langmuir adsorption iso- column (Fig. 1). Uncontaminated water was flushed through the col-
therms were applied to the experimental data (Osagie and Owabor, umns to establish a steady-state condition. The porosity of the packed 
2015; Mohammadi et al., 2017). The Freundlich model assumes that the bed was quantified by determining the gravimetric moisture content, 
adsorbent’s surface is heterogeneous and can be used to describe both and the values were subsequently converted to volumetric based on the 
monolayer (chemisorption) and multilayer (physisorption) adsorption bulk density of the packed bed. Hydraulic conductivity (K) was 
processes (Osagie and Owabor, 2015; Mohammadi et al., 2017). The measured using the constant head method (e.g., Head and Keeton, 2008) 
linear form of the Freundlich model is represented by Eq. (5). before and after the experiment. Synthetic water spiked with chloride 
1 and benzene was introduced into the columns (in an upward flow di-
logqe = logKF + logCe (5)  n rection) by step from a Teflon bag (to minimize volatilization) using a 
peristaltic pump at a flow rate of 1.67 cm3 min− 1 (i.e. ca. 0.074 cm 
where: q − 1e is the amount of benzene adsorbed (mg g ) at equilibrium; Ce min− 1 linear velocity). The Cl− was used to characterize physical 
is the equilibrium concentration (mg g− 1); KF and 1/n are the Freundlich transport in the media (i.e., to determine whether the transport is 
isotherm constants related to the adsorption capacity and adsorption Fickian or there are flow anomalies under the used experimental con-
intensity, respectively. KF and 1/n represent the intercept and slope, ditions). Table 2 summarises the properties of the packed columns. 
3
F. Obiri-Nyarko et al.                                                                                                                                                                        J o  u  r n  a  l o  f  C  o  n  t a  m  i n  a  n  t  H  y  d  r o  l o g  y  250 (2022) 104054
effects of pore-scale fluctuations (Köhne et al., 2006). The ADE can, 
however, be modified to account for other biotic and abiotic processes 
such as retardation due to reversible sorption, biodegradation, and/or 
nonequilibrium processes. In the present study, one-dimensional (1-D) 
analytical solutions of the ADE (Eqs. (11) and (12)) were formulated to 
describe, respectively, chloride and benzene transport in the studied 
media (Domenico and Schwartz, 1998; Thornton et al., 2000). 
⎡ ⎤ ⎡ ⎤
( )
C 1 z − v
erfc ⎣ p
t ⎦ 1 v z z + v t= ( ) + exp
p erfc⎣ p ⎦α ( ) (11)  C 2 10 2 αv t 2 2 v
1
p
p 2 αvpt
2
⎡ ⎤
{ [ ( ( ))1
( ) ( ( ))1
]} 4μα 2
C 1 z 4μα 2 ⎢Rz − vpt 1 + ⎥v
= exp α 1 − 1 + × erfc
⎢ p ⎥
C 2 2 v ⎢⎣ ( )1 ⎥0 p 2 αvptR
2 ⎦
(12)  
where: C and C0 are the effluent and input concentrations of the solute 
− 1
Fig. 1. Setup of the column experiment.  (mg L ); erfc is the complementary error function; vp is the pore water 
velocity (cm min− 1); R is the retardation factor [− ] which is a fully 
reversible process between the solution and the solid phases; z is the 
Table 2 column length or the distance (cm) from the source of contamination 
Properties of the packed columns.  after time, t (min); μ (min− 1) is a sink that is used to represent irre-
Material Mixing vexperiment (cm ρb (g n (− ) versible sorption due to physical/sorption-related nonequilibrium pro-
ratio min− 1) cm− 3) cesses or biodegradation based on first-order decay (Thornton et al., 
Brown coal – 0.19 1.28 0.39 2000; Baek et al., 2003; Köhne et al., 2006); α is the dispersivity (cm). 
Zeolite – 0.20 1.28 0.38 The chloride simulation was performed assuming no retardation (R 
Compost – 0.21 1.24 0.36 = 1) and decay (μ = 0) of the chloride, and the approach involved 
Zeolite-compost 1:1 0.19 1.26 0.40 reproducing one known parameter (i.e., pore-water velocity) and the 
Brown coal- 
compost 5:1 0.21 1.27 0.35 other unknown (i.e., dispersivity). These parameters were then fixed in 
Eq. (12) to simulate the transport of benzene. The μ and R were adjusted 
ρb: bulk density of column; vexperiment: experimentally determined pore water until best fits were obtained in the case of benzene. The initial concen-
velocity; and n: total porosity. tration was set to zero for the entire sample, and the boundary condi-
tions were C = 0 at an infinite distance from the inlet and C = C0 at the 
Effluent samples were taken at specific intervals (i.e. 1, 3, 5, 8, 10, 15, inlet for all the solute transport processes in this study. Additional pa-
20, 24, and 30 pore volumes (PVs) for benzene and from 0.1 to 2 PVs for rameters describing chloride and benzene transport, namely: effective 
chloride) using a disposable glass syringe fixed to the columns to pre- porosity, ne, hydrodynamic dispersion coefficient, D (cm2 L min− 1), 
vent/minimize benzene loss via volatilization. Observed concentrations, Péclet number, Pe, and experimental partition coefficient (Ked), were 
C, were normalized to the initial concentration, C0, to enable the com- calculated using Eqs. (13) to (16), respectively. The theoretical distri-
parison of the breakthrough curves (BTCs). Time was also normalized to bution coefficient (Ktd), was also determined using Eq. (17) for com-
the mean residence time (MRT) to obtain the pore volumes flushed. parison with the Ked. This relationship normally holds when the foc is 
>0.1% (Appelo and Appelo and Postma, 1993; Thornton et al., 2000; 
2.4. Analytical methods Zheng et al., 2002). 
Q
The liquid-liquid extraction (LLE) method (Glaze and Lin, 1983) was ne = (13)  A × v
used in this study. Five (5) mL of the effluent sample was transferred into p
a vial capped with Teflon-lined septa after which 1 mL of n-pentane was D = α × v (14)  
added and agitated for 5 min to reach liquid-liquid equilibrium. The L L p
pentane extract was then transferred into 2.5 mL glass vials capped with vp × z
Teflon-lined septa and analyzed for benzene using gas chromatography Pe = (15)  DL
(Shimadzu GC 17A, PAF/A/5/Sb). Chloride was determined titrimetri-
cally using AgNO3 as the standard and K2CrO4 as the indicator e (R − 1) • nK e= (16)  
(Richards, 1968). For reproducibility of the results, all the experiments d ρb
were performed in duplicate and average values were used. Addition-
ally, blank experiments were carried out to assess losses via Ktd = K
t
oc × foc (17)  
volatilization. 
where: Q: discharge (cm3 min− 1); A: the cross-sectional area of the 
column (cm2); DL: the hydrodynamic dispersion coefficient (cm2 min− 1) 2.5. Modelling approach assuming negligible diffusion; α: the dispersivity (cm); ρb: the bulk 
density (g cm− 3); Pe: Péclet number [− ]. Pe indicates the relative 
The ADE is an equilibrium model that considers advection and importance of mechanical dispersion and molecular diffusion in the 
dispersion as the fundamental processes governing the fate and trans- transport of solute in the media, where Pe < 0.4 indicates that the solute 
port of solute. Dispersion is assumed to behave macroscopically as a transport is controlled mainly by molecular diffusion, Pe between 0.4 
Fickian process with dispersivity remaining constant in space and time. and 6 indicates the combined effects of molecular diffusion and me-
Consequently, the basic form of the ADE is unable to reproduce non- chanical dispersion on the solute transport, and Pe > 6 indicates the 
Fickian or anomalous processes due to its inability to account for the 
4
/
/ /
F. Obiri-Nyarko et al.                                                                                                                                                                        J o  u  r n  a  l o  f  C  o  n  t a  m  i n  a  n  t  H  y  d  r o  l o g  y  250 (2022) 104054
dominance of mechanical dispersion over molecular diffusion (Fetter, 3. Results and discussion 
2001); Ke d and Ktd, respectively, represent the experimental and theo-
retical partition coefficient (cm3 g− 1); foc represents the fraction of OC in 3.1. Batch experiment 
the reactive material; Kt oc is the theoretical solute distribution coefficient 
to solid OC. The value of Kt oc used in this calculation was taken from 3.1.1. Adsorption isotherms and kinetics models 
Weiner (2008). The accuracy of the simulations was evaluated based on The results of fitting the two isotherms to the experimental data are 
the correlation coefficient (R2) and root mean square error (RMSE) (Eq. illustrated in Fig. 2a & b, while the estimated parameter values are 
(18)). The smaller (closer to 0) the RMSE, the better the model predic- summarized in Table 3. A high correlation coefficient (R2 ≥ 0.96) was 
tion. Higher values of RMSE indicate a large over-or under-estimation of obtained for the two isotherms, indicating that both models can be 
the experimental data by the model. utilized to characterize benzene adsorption onto the examined sorbents. 
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ This suggests the existence of both monolayer and heterogeneous sur-
√ [( ) ( ) ]√∑ 2
√ N C C face conditions on the adsorbents. The RL values obtained from the 
√ j=1 C −0 cal C0RMSE exp (18)  Langmuir isotherm ranged from 0.03 to 0.45 for zeolite, 0.01–0.14 for =
N brown coal, and 0.02–0.37 for compost indicating that the adsorption of 
( ) ( ) benzene onto the studied adsorbents is favourable. Similarly, the values 
where: CC and 
C
C , respectively, are the predicted and measured of 1/n determined from the Freundlich isotherm were < 1, indicating 0 cal 0 exp favourable adsorption (Özer and Pirincci, 2006). Among the studied 
relative concentration at time t; N is the number of observations. materials, brown coal demonstrated the highest capacity to remove 
Additional properties including the removal efficiency and removal benzene. The maximum adsorption capacity (q ) determined with the 
capacity of the adsorbents were determined using the equations below maxLangmuir model was 6.057 mg g− 1 for brown coal, which is generally 
(Obiri-Nyarko et al., 2020): higher when compared to the qmax of the natural adsorbents (e.g., clay 
⎡ ⎤ and sand) but lower when compared to the qmax of synthetic or modi-
⎢⎛ t ⎞ ⎥ fied/treated adsorbents (Table 4). ⎢ ∫total / ⎥
⎢ Q ( )⎢⎝ ⎠ C Qt ⎥⎥ The kinetics of contaminant removal is important in PRB studies as it RE% = ⎢ Cremdt 0 total ⎥× 100 (19)  
⎢ 1000 1000 ⎥ indicates the rate and removal mechanisms, which are crucial for the ⎣ 0 ⎦ design of PRBs. The results of the kinetics studies are shown in Fig. 3a-c 
while the kinetic constants and R2 are reported in Table 5. Although the 
values of R2 ⎡ ⎤ for both PFO and PSO kinetic models are high (>0.94), the 
adsorption capacities calculated by the PSO model are closer to those 
⎢⎛ t ⎞ ⎥⎢ ∫total / ⎥ determined experimentally when compared to those determined with 
⎢
⎢⎝ Q ⎥q = Cremdt⎠ ⎥ (20)  the PFO model. This indicates that the adsorption of benzene onto the TRC ⎢
⎢ 1000 m⎥⎥
⎣ 0 ⎦
Table 3 
where: m (mg) is the mass of the adsorbent in the columns; t rep- Isotherm parameters for adsorption of benzene onto compost, brown coal and total 
resents the total experimental time (min), Q is the flow rate (cm3 min− 1); zeolite.  
Crem = (C0 - Ct) is the reduction of the benzene concentration in the Isotherm Parameter Brown coal Zeolite Compost 
effluent due to sorption (mg dm− 3); C0 and Ct (mg dm− 3) are the initial b (L mg− 1) 3.052 0.612 0.837 
benzene concentration and concentration at time t; RE%, (%) is the rate q (mg g− 1max ) 6.057 3.484 2.823 Langmuir 2 
of benzene removal; q − 1 R 0.983 0.979 0.960 TRC (mg g ) is the total removal capacity of the 
material. RL 0.01–0.14 0.031–0.45 0.02–0.37 K (mg g− 1)(L mg− 1)1/n F 12.368 6.066 7.836 
Freundlich 1/n 0.477 0.367 0.303 
R2 0.997 0.986 0.995 
b: Langmuir constant related to the energy of adsorption or adsorption intensity; 
qmax: maximum adsorption capacity determined from the Langmuir isotherm; 
KF: Freundlich constant related to the saturation capacity of the sorbent; 1/n: 
adsorption intensity; and R2: correlation coefficient. 
Fig. 2. Linear fittings of the (a) Langmuir and (b) Freundlich isotherm models to the experimental data for zeolite, brown coal, and compost. (For interpretation of 
the references to colour in this figure legend, the reader is referred to the web version of this article.) 
5
F. Obiri-Nyarko et al.                                                                                                                                                                        J o  u  r n  a  l o  f  C  o  n  t a  m  i n  a  n  t  H  y  d  r o  l o g  y  250 (2022) 104054
Table 4 intra-particle diffusion rate constant, and an intercept, C, which in-
Comparison of benzene adsorption capacities of adsorbents in this study with dicates the thickness of the boundary layer or surface adsorption. If the 
other.  straight line passes through the origin, it indicates that intra-particle 
Adsorbent q − 1max (mg g ) References diffusion is the only mechanism and the rate-limiting step in the 
PMO 7.04 Moura et al. (2011) adsorption process. It can be seen from Fig. 3c that the linear plots did 
CKD 130.39 Faisal et al. (2021) not pass through the origin, indicating that intra-particle diffusion is not 
PEG-Mt 5.92 Nourmoradi et al. (2012) the rate-limiting step and the only mechanism controlling the adsorption 
SMZ 150.42 Vidal et al. (2012) process. 
F-400 183.29 Wibowo et al. (2007) 
F-400Cox 114.41 Wibowo et al. (2007) 
F-400Tox 240.07 Wibowo et al. (2007) 
CuO-NPs 100.24 Mohammadi et al. (2017) 
Natural clay 1.26 Osagie and Owabor (2015) 
Sandy soil 1.14 Osagie and Owabor (2015) Table 5 
Brown coal 6.06 Present study Kinetic parameters for adsorption of benzene onto brown coal, compost, and 
Zeolite 3.48 Present study 
Compost 2.82 Present study zeolite.  
Model Parameter Brown coal Zeolite Compost 
PMO: periodic mesoporous organosilicas; CKD: cement kiln dust; PEG–Mt: 
montmorillonite (Mt) modified with poly ethylene glycol; SMZ: HDTMA- Pseudo- qe − 1(exp) (mg g ) 2.509 2.212 2.291  
modified Y zeolite; (F-400): granular activated carbon Filtrasorb400; F- 1st-order (PFO) 
− 1
400Cox: chemically treated granular activated carbon Filtrasorb400; F-400Tox: qe(cal) (mg g ) 2.221 1.112 1.039  
− 1
thermally treated granular activated carbon Filtrasorb400; and CuO-NPs: copper k1(min ) 0.020 0.006 0.006  R2 0.946 0.963 0.997 
oxide nanoparticles. Pseudo- k2(g mg− 1 min− 1) 0.015 0.015 0.017  2nd-order (PSO) 
studied materials is a chemical process (chemisorption) involving q (mg g− 1e(exp) ) 2.509 2.212 2.291  
2 − 1 − 1
sharing of electrons between benzene and the surface of the sorbents. k2q (mg g min ) 0.112 0.078 0.095  
q (cal) (mg g− 1) 2.750 2.260 2.352  
Similar results were reported by Simantiraki and Gidarakos (2015). The eR2 1.000 0.994 0.999 
intra-particle diffusion model (Weber and Morris, 1963) was employed Intra-particle − 1 -1/2
to explore the effect of diffusion on the mass transfer of benzene onto the diffusion Kid (mg g min ) 0.083 0.066 0.067  
studied materials. According to Tan and Hameed (2017), solute mass C 1.291 1.032 1.151  2 
transfer occurs in three steps: (i) film diffusion (external diffusion); (ii) R 0.923 0.984 0.980 
pore diffusion (i.e. intra-particle diffusion); and (iii) surface reaction, qe(cal): theoretical adsorption capacity; qe(exp): experimental adsorption capac-
which involves the attachment of the adsorbate to the internal surface of ity; k1: PFO adsorption rate constant; R2: correlation coefficient; k2: PSO rate 
2
the adsorbent. According to Weber and Morris (1963), the plot of qt vs constant; k2q : initial or instantaneous sorption rate; Kid: intra-particle diffusion 
t1/2 will produce a straight line with a slope K , which represents the rate constant; and C: intra-particle diffusion boundary layer thickness. id
Fig. 3. Fittings of the experimental kinetic data with the (a) PFO; (b) PSO; and (c) Intra-particle diffusion models.  
6
F. Obiri-Nyarko et al.                                                                                                                                                                        J o  u  r n  a  l o  f  C  o  n  t a  m  i n  a  n  t  H  y  d  r o  l o g  y  250 (2022) 104054
3.2. Column studies column, the measured BTCs reflect the macroscopic behaviour of the 
column. As can be seen, the Cl− BTCs have different shapes, reflecting 
3.2.1. Chloride breakthrough curves different transport conditions in the media. The values of the inverted 
Physical or transport-related nonequilibrium, also known as anom-
alous (non-Fickian or nonideal) transport (e.g., preferential flow), is a Table 6 
major issue in PRBs as it limits the mass transfer of contaminants to the The hydrodynamic parameters estimated from the chloride curve fitting.  
reactive sites, resulting in lower than expected barrier performance 
(Kamolpornwijit et al., 2003). The chloride experiments were conducted Reactive vp (cm DL (cm
2 α Pe n 2 e R RMSE 
materials min− 1) min− 1) (cm) (− ) (− ) 
to ascertain the presence of physical or transport-related nonequilibrium 
in the columns based on the experimental data and curve fitting with the Brown coal 0.20 0.42 2.10 14.2 0.37 0.985 0.050 
ADE. Several researchers (e.g., Singh and Kanwar, 1991; Köhne et al., Zeolite 0.23 1.15 5.00 6.0 0.32 0.983 0.049 Compost 0.21 0.43 2.10 14.7 0.36 0.980 0.058 
2006) have noted that physical or transport-related nonequilibrium (e. Zeolite- 
g., preferential flow and diffusion into non-advective zones) exists if the 0.19 0.61 3.17 9.3 0.39 0.986 0.048 compost 
ADE is unable to adequately describe the non-reactive tracer transport, Brown coal- 0.22 0.24 1.10 27.5 0.32 0.990 0.042 
or if 50% of the injected chloride (C/C0 = 0.5) is measured in the 
compost 
effluent well before or after 1 PV. DL: hydrodynamic dispersion coefficient; α: dispersivity; vp: pore-water velocity; 
Fig. 4a-e shows the experimental and simulated Cl− BTCs for the Pe: Péclet number; ne: effective porosity; RMSE: root mean square error; and R2: 
different media. Since samples were collected at the outlet of each correlation coefficient. 
Fig. 4. Experimental and modelled BTCs of chloride in (a) brown coal, (b) zeolite, (c) compost, (d) zeolite-compost, and (e) brown coal-compost. (For interpretation 
of the references to colour in this figure legend, the reader is referred to the web version of this article.) 
7
F. Obiri-Nyarko et al.                                                                                                                                                                        J o  u  r n  a  l o  f  C  o  n  t a  m  i n  a  n  t  H  y  d  r o  l o g  y  250 (2022) 104054
parameters (i.e. pore-water velocity and dispersivity) are summarized in measurements were terminated after 2 PVs due to analytical difficulties. 
Table 6. The model results for pore-water velocity compared well with Therefore, these observations reflect the transport conditions in the 
the calculated pore-water velocity obtained from the ratio of flux rate to media at the beginning and not for the entire duration of the experiment. 
water content (Table 2). Effective porosity was slightly lower than the 
total porosity in all the columns except the column with compost, 3.2.2. Benzene breakthrough curves 
indicating negligible immobile water zones in the columns. Dispersivity BTCs showing the behaviour of benzene in the studied media are 
ranged from 2.1 to 5.0 cm, with lower values indicating lower variability displayed in Fig. 5a–e. In relation to the chloride BTCs (Fig. 4a-e), the 
in the microscopic velocity distribution (e.g., Kumahor et al., 2015) and benzene BTCs shifted to the right on the time axis (PV), indicating 
reduced spread in solute breakthrough. The corresponding Péclet benzene retardation by the media. The effluent concentration of ben-
numbers for the columns ranged from 6 to 27.5, indicating that the flow zene from all the columns was also below the inlet concentration (C/C0 
regime within the studied columns was dominated by mechanical < 1) except for zeolite where there was 100% recovery of the applied 
dispersion and diffusion into the non-advective or immobile zones was benzene, i.e. C/C0 = 1. The benzene BTC of the brown coal column 
not significant. As shown in Fig. 4a-e, C/C0 = 0.5 was measured a little additionally showed a plateau between PV 10 and 15 after which ben-
before or after 1 PV. Moreover, the ADE described all the Cl− BTCs zene concentration started to decline again. Volatilization, biodegra-
reasonably well (R2 ≥ 0.98 and RMSE <0.06), suggesting that physical dation, and nonequilibrium processes (e.g., sorption- and transport- 
or transport-related nonequilibrium processes, such as preferential flow related nonequilibrium) have been adduced as possible causes of such 
and diffusive transport into immobile zones at the beginning of the ex- behaviour (Selim et al., 1999; Thornton et al., 2000; Baek et al., 2003; 
periments were marginal. It should be mentioned that the chloride Köhne et al., 2006; Parashar et al., 2007). In the present study, however, 
Fig. 5. Modelled and experimental BTCs of benzene in: (a) zeolite, (b) brown coal, (c) compost, (d) zeolite-compost, and (e) brown coal-compost columns. (For 
interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 
8
F. Obiri-Nyarko et al.                                                                                                                                                                        J o  u  r n  a  l o  f  C  o  n  t a  m  i n  a  n  t  H  y  d  r o  l o g  y  250 (2022) 104054
we attribute the observed C/C0 < 1 as well as the plateau to physical/ Table 7 
sorption nonequilibrium processes with biodegradation playing little or Estimated parameters from the benzene BTCs.  
no role. Volatilization is ruled out since benzene mass loss in the blank Material R Ke d (cm3 Kt d (cm3 μ RE qTRC (mg 
experiments (without sorbents) was insignificant. g− 1) g− 1) (min− 1) (%) g− 1) 
In a related study by Grajales-Mesa and Malina (2019), the authors 8.0 × 8.05 ×
observed that the application of compost resulted in the biodegradation Brow coal 4 0.87 10.20 − 4 51.87 10 10− 2 
of trichloroethylene. In the present study, biodegradation has been 1.95 ×Zeolite 2.8 0.45 1.78 0.0 11.11 − 2 
considered a benzene mass loss mechanism only because of the nature of 10
the benzene BTCs. We initially did not consider that benzene biodeg- 2.0 × 9.62 ×Compost 6 1.45 11.92 10− 4 33.33 10− 2 
radation would advance significantly in the columns with compost due Zeolite- 3.5 × 4.63 ×
to the short duration (70 h) of the experiment. The nonequilibrium 2.9 0.58 6.85 22.08 compost 10− 4 10− 2 
processes are classified into physical/transport- and sorption-related Brown coal- 29 7.11 10.49 0.0 90.12 16.15 ×
nonequilibrium. Sorption-related nonequilibrium can be caused by compost 10
− 2 
chemical nonequilibrium (i.e. rate-limited interactions between the R: retardation factor; Ked: experimental partition coefficient; Ktd: theoretical dis-
sorbate and sorbent) and/or rate-limited diffusive mass transfer (film tribution coefficient; μ: degradation rate; qTRC: total removal capacity; and RE: 
diffusion, retarded intraparticle diffusion, and intra-sorbent diffusion) removal efficiency. 
while physical nonequilibrium may occur when there is preferential 
flow and solute diffusion into non-advective/immobile regions of the Table 7. The Kt d indicates the expected sorption in a model system where 
medium (Brusseau et al., 1991). Köhne et al. (2006) noted that both other influencing processes are limited. The use of Eq. (17) to calculate 
nonequilibrium processes can cause C/C0 < 1. In the present study, the the distribution coefficient is justified since the OC values measured for 
nonequilibrium processes playing a major role are the intra-organic all the studied materials except zeolite were > 0.1% (foc > 0.001). This 
matter diffusion and benzene transport in preferential flow paths. represents a condition where hydrophobic partitioning onto OC is equal 
Diffusive mass transfer into non-advective zones is not favoured since to non-hydrophobic sorption by minerals (Thornton et al., 2000). The 
results from the tracer experiment indicated the dominance of me- calculated Kt d values ranged from 1.78 to 11.92 cm3 g− 1, which are 1.5 to 
chanical dispersion over molecular diffusion. Brusseau et al. (1991) also 11.8 times higher than the Ke d values, implying that the performance of 
noted that the specific sorbate-sorbent interaction is relatively irrelevant the materials was lower than expected. This could be due to benzene 
for hydrophobic organic compounds (HOCs) since their sorption is transport in preferential pathways as explained earlier and/or the high 
generally driven by partitioning between the solution and OM compo- pore-water velocity used, which is inversely related to R (e.g., Pang 
nents of the sorbent. On the contrary, they found intra-sorbent (intra- et al., 2002; Kim et al., 2006a). Pang et al. (2002) explained that at low 
organic matter) mass transfer diffusion as the main cause of the sorption flow rates, there is a longer retention of the contaminants in the col-
nonequilibrium of HOC due to the polymeric nature of OM. Sander and umns, which allows for more complete sorption. The R factors and some 
Pignatello (2009) reported that intra-sorbent diffusion is mostly an experimental conditions reported in other studies are summarized in 
irreversible process, resulting in effluent concentrations generally lower Table 8 for comparison with the values from the present study. As can be 
than the influent solute concentrations. The preferential flow, on the seen, the R obtained for the brown coal-compost mixture is generally 
other hand, causes contaminants to bypass reactive sites, resulting in higher than those for the other reported media. 
higher effluent solute concentration (Kamolpornwijit et al., 2003). It is The values of μ ranged from 0 to 8 × 10− 4 min− 1. For zeolite and the 
possible that in addition to intra-sorbent (intra-organic matter) mass brown coal-compost mixture, μ = 0 was obtained, indicating that ben-
transfer diffusion, preferential flow also occurred in the columns. The zene removal was due to its retardation or reversible sorption. For the 
ascending part of the plateau observed in the brown coal column, for other columns, benzene removal was possibly due to reversible sorption 
example, is reminiscent of preferential flow. It is worth noting that, as and biodegradation, and/or irreversible sorption due to the diffusive 
the experiments proceeded, particles from columns with brown coal and mass transfer of benzene into the matrix of OM. The effect of biodeg-
compost were detected in the effluent, indicating attrition and possible radation could have been quantified by performing sterile (involving the 
rearrangement in those media. This structural rearrangement possibly application of a biocidal compound to annihilate microorganisms) and 
created preferential flow paths in those columns, allowing contaminants non-sterile tests (Kiecak et al., 2020). However, as indicated earlier, we 
to bypass the reactive sites. did not anticipate that biodegradation would occur within the experi-
The capability of the used model to predict benzene transport in the mental period, so such experiments were not performed. Advanced 
studied media and to estimate the benzene attenuation parameters was models together with additional experiments (e.g., two-site and two- 
also explored. The results are shown in Fig. 5a-e. The ADE where only R region models and sterile and non-sterile experiments) to distinguish 
was considered could only reproduce sufficiently the experimental data between these processes will be needed (Pang et al., 2002; Kiecak et al., 
for zeolite and the brown coal-compost mixture, indicating that benzene 2020). Additional parameters determined to evaluate the performance 
removal by these sorbents is a reversible process. In the case of the of the materials such as their removal efficiency and removal capacity 
brown coal-compost mixture, the BTC was incomplete, so it is unclear are also presented in Table 7. The removal efficiencies ranged from 
whether other processes apart from reversible sorption were involved. 11.11 to 90.12% while the removal capacity ranged from 1.95 × 10− 2 to 
For the other systems (brown coal, compost, zeolite-compost mixture), 16.15 × 10− 2 mg g− 1. The highest removal efficiency and adsorption 
significant disparities between the experimental and modelled results capacity were obtained for the brown coal-compost mixture. This im-
were observed. However, when both R and μ were coupled and fitted to plies that a smaller amount of this mixture will be needed to treat a given 
the experimental data, significant improvement in the simulation was volume of benzene-contaminated groundwater as compared to the other 
observed. adsorbents. 
The estimated values of R and μ are summarized in Table 7. Benzene 
was most retarded by the brown coal-compost mixture (R = 29) while 3.3. Hydraulic performance 
zeolite exhibited the least affinity for benzene (R = 2.8). This is possibly 
due to the relatively high foc of the brown coal-compost mixture and the Evaluating the hydraulic performance of potential PRB materials for 
relatively small grain size of the brown coal (Table 1) (Müller, 2010). in situ groundwater remediation is crucial because many of the reported 
The corresponding experimental distribution coefficient, Ke d ranged from PRBs failures were caused by changes (increase/decrease) in porosity 
0.45 to 7.11 cm3 g− 1. The Kt d calculated for each column material based and hydraulic conductivity of the medium (e.g., Henderson and 
on literature data of Koc and OC contents of the media is also shown in Demond, 2007). Table 9 displays the hydraulic conductivity values 
9
F. Obiri-Nyarko et al.                                                                                                                                                                        J o  u  r n  a  l o  f  C  o  n  t a  m  i n  a  n  t  H  y  d  r o  l o g  y  250 (2022) 104054
Table 8 
Comparison of sorption parameters reported in the literature and this study.  
Reactive material K (cm3 g− 1d ) R (− ) Initial concentration (mg dm− 3) Flow velocity (cm s− 1) Reference 
SAM – 1.02 5.20 0.025 [1] 
Triassic sandstone 0.14 1.56 0.39 5.21 × 10− 5 [2] 
SAM + PAC – 1.5–3.2 300 1.69 × 10− 3 [3] 
Aquifer material – 14.3 0.52–1.25 5.2 × 10− 4 [4] 
SAM – 2.46 30 9.72 × 10− 3 [5] 
SMZ – 18.22 9.37 0.027 [6] 
SMZ 13.1 – 1.37 0.042 [7] 
Brow coal 0.87 4 2.40 1.24 × 10− 3 Present study 
Zeolite 0.45 2.8 2.40 1.24 × 10− 3 Present study 
Compost 1.45 6 2.40 1.24 × 10− 3 Present study 
Zeolite-compost 0.58 2.9 2.40 1.24 × 10− 3 Present study 
Brown coal-compost 7.11 28.9 2.40 1.24 × 10− 3 Present study 
SAM: sandy aquifer material; SMZ: surfactant-modified zeolite; Kd: partition coefficient; R: retardation factor; PAC: powdered activated carbon. The amount of PAC 
added ranged from 0 to 2%. [1] Thierrin et al., 1995; [2] Thornton et al., 2000; [3] Kim et al., 2006b; [4] Priddle and Jackson, 1991 [5] Maraqa et al., 1998; [6] 
Simpson and Bowman, 2009; and [7] Altare et al., 2007. 
mechanistic understanding of the dynamics of benzene removal by the 
Table 9 sorbents as well as for the design of an effective PAB system for the 
Initial and final hydraulic conductivities of the studied media.  remediation of benzene-contaminated groundwater. 
Reactive material Initial Final 
Brow coal 4.20 × 10− 5 4.30 × 10− 5 CRediT authorship contribution statement 
Zeolite 2.14 × 10− 3 1.71 × 10− 3 
Compost 1.12 × 10− 3 1.03 × 10− 3 
3 3 Franklin Obiri-Nyarko: Conceptualization, Methodology, Formal Zeolite-compost 1.42 × 10− 1.71 × 10−
Brown coal-compost 5.77 10− 5 6.67 10− 5  analysis, Investigation, Writing - original draft, Writing - review & × ×
editing. Jolanta Kwiatkowska-Malina: Conceptualization, Methodol-
ogy, Validation, Formal analysis, Resources, Writing – review & editing. 
measured before and after solute injection into the columns. The initial Samuel Kwame Kumahor: Software, Formal analysis, Writing - review 
K ranged from 4.20 × 10− 5 to 2.14 × 10− 3 cm s− 1, with the observed & editing. Grzegorz Malina: Conceptualization, Methodology, Valida-
discrepancy due to differences in the grain size of the materials. The tion, Formal analysis, Resources, Writing – review & editing. 
initial K of the mixtures either increased or decreased relative to their 
individual components due to similar reasons as adduced above. Declaration of Competing Interest 
Although the final K remained within the same order of magnitude after 
about 70 h of operation, slight changes (increase/decrease) were The authors declare no conflict of interest in relation to this work. 
observed. This is attributed to either particle attrition due to the 
employed flow rate or reorganization of the pore space by the flowing Acknowledgement 
water as some of the material grains were observed in the effluent from 
the columns containing compost and brown coal. In general, all the The research leading to these results has received funding from the 
studied media can conduct water fairly well. As a general rule, however, European Community’s Seventh Framework Programme (FP7/2007- 
the initial K of the PRB must be greater than that of the aquifer material, 2013 under grant agreement no 265063). 
and changes in K during the operation of the barrier must not be extreme 
as this will result in insufficient contact between the contaminant and References 
the reactive material. 
Al-Shammary, A.A.G., Kouzani, A.Z., Kaynak, A., Khoo, S.Y., Norton, M., Gates, W., 
4. Conclusion 2018. Soil bulk density estimation methods: a review. Pedosphere 28, 581–596. 
https://doi.org/10.1016/S1002-0160(18)60034-7. 
Altare, C.R., Bowman, R.S., Katz, L.E., Kinney, K.A., Sullivan, E.J., 2007. Regeneration 
Laboratory (batch and column) experiments coupled with analytical and long-term stability of surfactant-modified zeolite for removal of volatile organic 
modelling enabled us to evaluate reactive materials of a PAB for sus- compounds from produced water. Microporous Mesoporous Mater. 105, 305–316. 
tainable removal of benzene from contaminated groundwater. The re- https://doi.org/10.1016/j.micromeso.2007.04.001. Appelo, C.A.J., Postma, D., 1993. Geochemistry, Groundwater and Pollution. AA 
sults showed that the brown coal-compost mixture can be used as a low- Balkema, Rotterdam.  
cost effective reactive material in PABs for remediating benzene- Baek, D.S., Kim, S.B., Kim, D.J., 2003. Irreversible sorption of benzene in sandy aquifer 
contaminated groundwater. It proved to be the most suitable option materials. Hydrol. Process. 17, 1239–1251. https://doi.org/10.1002/hyp.1181. 
Bahadar, H., Mostafalou, S., Abdollahi, M., 2014. Current understandings and 
among the studied media as it exhibited relatively high (i) ability to perspectives on non-cancer health effects of benzene: a global concern. Toxicol. 
retard benzene migration (R = 29), (ii) removal efficiency (>90%), and Appl. Pharmacol. 276, 83–94. https://doi.org/10.1016/j.taap.2014.02.012. 
(iii) total removal capacity (0.16 mg g− 1), owing to its relatively high Bilardi, S., Ielo, D., Moraci, N., Calabrò, P.S., 2016. Reactive and hydraulic behavior of 
permeable reactive barriers constituted by Fe0 and granular mixtures of Fe0/ 
OM content and small gain size. Moreover, it can conduct water fairly pumice. Procedia Eng. 158, 446–451. https://doi.org/10.1016/j. 
well (K = 10− 3 m s− 1). Benzene generally exhibited a non-Fickian/ proeng.2016.08.470. 
nonideal behaviour in all the studied media, which is attributed to Brusseau, M.L., Jessup, R.E., Rao, P.S.C., 1991. Nonequilibrium sorption of organic 
sorption- and transport-related nonequilibrium processes (preferential chemicals: elucidation of rate-limiting processes. Environ. Sci. Technol. 25, 134–142. https://doi.org/10.1021/es00013a015. 
flow and intra-organic matter mass transfer diffusion), and possibly Chen, L., Liu, F., Liu, Y., Dong, H., Colberg, P.J., 2011. Benzene and toluene 
biodegradation. The 1-D ADE modified to include retardation with or biodegradation down gradient of a zero-valent iron permeable reactive barrier. 
without degradation/irreversible sorption allowed satisfactory fit of the J. Hazard. Mater. 188, 110–115. https://doi.org/10.1016/j.jhazmat.2011.01.076. Domenico, P.A., Schwartz, F.W., 1998. Physical and Chemical Hydrogeology. Wiley, 
model to the experimental benzene BTCs. Further modelling and New York.  
experimental studies are, however, needed to differentiate between the Erto, A., Andreozzi, R., Lancia, A., Musmarra, D., 2010. Factors affecting the adsorption 
benzene removal processes. Such discrimination is indispensable for a of trichloroethylene onto activated carbons. Appl. Surf. Sci. 256, 5237–5242. https://doi.org/10.1016/j.apsusc.2009.12.110. 
10
F. Obiri-Nyarko et al.                                                                                                                                                                        J o  u  r n  a  l o  f  C  o  n  t a  m  i n  a  n  t  H  y  d  r o  l o g  y  250 (2022) 104054
Erto, A., Lancia, A., Bortone, I., Di Nardo, A., Di Natale, M., Musmarra, D., 2011. Nourmoradi, H., Nikaeen, M., Khiadani, M., 2012. Removal of benzene, toluene, 
A procedure to design a permeable adsorptive barrier (PAB) for contaminated ethylbenzene and xylene (BTEX) from aqueous solutions by montmorillonite 
groundwater remediation. J. Environ. Manag. 92, 23–30. https://doi.org/10.1016/j. modified with nonionic surfactant: equilibrium, kinetic and thermodynamic study. 
jenvman.2010.07.044. Chem. Eng. J. 191, 341–348. https://doi.org/10.1016/j.cej.2012.03.029. 
Eykholt, G.R., Elder, C.R., Benson, C.H., 1999. Effects of aquifer heterogeneity and Obiri-Nyarko, F., Grajales-Mesa, S.J., Malina, G., 2014. An overview of permeable 
reaction mechanism uncertainty on a reactive barrier. J. Hazard. Mater. 68, 78–101. reactive barriers for in situ sustainable groundwater remediation. Chemosphere 111, 
https://doi.org/10.1016/s0304-3894(99)00032-1. 243–259. https://doi.org/10.1016/j.chemosphere.2014.03.112. 
Faisal, A.A., Jasim, H.K., Naji, L.A., Naushad, M., Ahamad, T., 2021. Cement kiln dust- Obiri-Nyarko, F., Kwiatkowska-Malina, J., Malina, G., Kasela, T., 2015. Geochemical 
sand permeable reactive barrier for remediation of groundwater contaminated with modelling for predicting the long-term performance of zeolite-PRB to treat lead 
dissolved benzene. Sep. Sci. Technol. 56, 870–883. https://doi.org/10.1080/ contaminated groundwater. J. Contam. Hydrol. 177, 76–84. https://doi.org/ 
01496395.2020.1746341. 10.1016/j.jconhyd.2015.03.007. 
Fetter, C.W., 2001. Contaminant Hydrogeology, 4th ed. Prentice-Hall, Inc., New Jersey.  Obiri-Nyarko, F., Kwiatkowska-Malina, J., Malina, G., Wołowiec, K., 2020. Assessment of 
Freidman, B.L., Terry, D., Wilkins, D., Spedding, T., Gras, S.L., Snape, I., Mumford, K.A., zeolite and compost-zeolite mixture as permeable reactive materials for the removal 
2017. Permeable bio-reactive barriers to address petroleum hydrocarbon of lead from a model acidic groundwater. J. Contam. Hydrol. 229, 103597 https:// 
contamination at subantarctic Macquarie Island. Chemosphere 174, 408–420. doi.org/10.1016/j.jconhyd.2019.103597. 
https://doi.org/10.1016/j.chemosphere.2017.01.127. Osagie, E., Owabor, C.N., 2015. Adsorption of benzene in batch system in natural clay 
Glaze, W.C., Lin, C.C., 1983. Optimization of Liquid-Liquid Extraction Methods for and sandy soil. Adv. Chem. Eng. Sci. 5, 352–361. https://doi.org/10.4236/ 
Analysis of Organics in Water. US Environmental Protection Agency, Environmental aces.2015.53037. 
Monitoring and Support Laboratory. Özer, A., Pirincci, H.B., 2006. The adsorption of Cd (II) ions on sulphuric acid-treated 
Grajales-Mesa, S.J., Malina, G., 2019. Pilot-scale evaluation of a permeable reactive wheat bran. J. Hazard. Mater. 137, 849–855. https://doi.org/10.1016/j. 
barrier with compost and brown coal to treat groundwater contaminated with jhazmat.2006.03.009. 
trichloroethylene. Water 11, 1922. https://doi.org/10.3390/w11091922. Pang, L., Close, M., Schneider, D., Stanton, G., 2002. Effect of pore-water velocity on 
Grajales-Mesa, S.J., Malina, G., Kret, E., Szklarczyk, T., 2020. Designing a permeable chemical nonequilibrium transport of Cd, Zn, and Pb in alluvial gravel columns. 
reactive barrier to treat TCE contaminated groundwater: numerical modelling/ J. Contam. Hydrol. 57, 241–258. https://doi.org/10.1016/s0169-7722(01)00223-6. 
Diseño de una barrera permeable reactiva para el tratamiento de aguas subterráneas Parashar, R., Govindaraju, R.S., Hantush, M.M., 2007. Temporal moment analysis for 
contaminadas con tricloroetileno: modelo numérico. Tecnol. Cienc. Agua 11, volatile organic compounds in dual-porosity porous media: loss fractions and 
78–106. https://doi.org/10.24850/j-tyca-2020-03-03. effective parameters. J. Environ. Eng. 133, 879–890. https://doi.org/10.1061/ 
Hall, K.R., Eagleton, L.C., Acrvos, A., Vermeulen, T., 1966. Pore- and solid- diffusion (ASCE)0733-9372(2007)133:9(879). 
kinetics in fixed-bed adsorption under constant pattern conditions. Ind. Eng. Chem. Priddle, M.W., Jackson, R.E., 1991. Laboratory column measurement of VOC retardation 
Fundam. 5, 212–223. https://doi.org/10.1021/i160018a011. factors and comparison with field values. Groundwater 29, 260–266. https://doi. 
Head, K.H., Keeton, G.P., 2008. Permeability, shear strength & compressibility tests. In: org/10.1111/j.1745-6584.1991.tb00518.x. 
Manual of Soil Laboratory Testing, 2. Whittles Publishing, UK.  Rabideau, A.J., Suribhatla, R., Craig, J.R., 2005. Analytical models for the design of iron- 
Henderson, A.D., Demond, A.H., 2007. Long-term performance of zero-valent iron based permeable reactive barriers. J. Environ. Eng. 131, 1589–1597. https://doi. 
permeable reactive barriers: a critical review. Environ. Eng. Sci. 24, 401–423. org/10.1061/ASCE0733-93722005131:111589. 
https://doi.org/10.1089/ees.2006.0071. Richards, L.A., 1968. Diagnosis and Improvement of Saline and Alkali Soils. USDA 
Ho, Y.S., McKay, G., 1998. Kinetic models for the sorption of dye from aqueous solution Handbook, 60. U.S. Government Printing Office, Washington, DC.  
by wood. Process. Saf. Environ. Prot. 76, 183–191. https://doi.org/10.1205/ Sander, M., Pignatello, J.J., 2009. Sorption irreversibility of 1,4-dichlorobenzene in two 
095758298529326. natural organic matter–rich geosorbents. Environ. Toxicol. Chem. 28, 447–457. 
Itodo, A.U., Abdulrahman, F.W., Hassan, L.G., Maigandi, S.A., Itodo, H.U., 2010. https://doi.org/10.1897/08-128.1. 
Intraparticle diffusion and intraparticulate diffusivities of herbicide on derived Selim, H.M., Ma, L., Zhu, H., 1999. Predicting solute transport in soils second-order two- 
activated carbon. Researcher 2, 74–86. site models. Soil Sci. Soc. Am. J. 63, 768–777. https://doi.org/10.2136/ 
Kamolpornwijit, W., Liang, L., West, O.R., Moline, G.R., Sullivan, A.B., 2003. Preferential sssaj1999.634768x. 
flow path development and its influence on long-term PRB performance: column Simantiraki, F., Gidarakos, E., 2015. Comparative assessment of compost and zeolite 
study. J. Contam. Hydrol. 66, 161–178. https://doi.org/10.1016/S0169-7722(03) utilisation for the simultaneous removal of BTEX, Cd and Zn from the aqueous phase: 
00031-7. batch and continuous flow study. J. Environ. Manag. 159, 218–226. https://doi.org/ 
Kiecak, A., Breuer, F., Stumpp, C., 2020. Column experiments on sorption coefficients 10.1016/j.jenvman.2015.04.043. 
and biodegradation rates of selected Pharmaceuticals in three aquifer sediments. Simpson, J.A., Bowman, R.S., 2009. Nonequilibrium sorption and transport of volatile 
Water 12, 14. https://doi.org/10.3390/w12010014. petroleum hydrocarbons in surfactant-modified zeolite. J. Contam. Hydrol. 108, 
Kim, S.B., Ha, H.C., Choi, N.C., Kim, D.J., 2006a. Influence of flow rate and organic 1–11. https://doi.org/10.1016/j.jconhyd.2009.05.001. 
carbon content on benzene transport in a sandy soil. Hydrol. Process.: Int. J. 20, Singh, P., Kanwar, R.S., 1991. Preferential solute transport through macropores in large 
4307–4316. https://doi.org/10.1002/hyp.6164. undisturbed saturated soil columns. J. Environ. Qual. 20, 295–300. https://doi.org/ 
Kim, S.B., Kim, D.J., Yun, S.T., 2006b. Attenuation of aqueous benzene in soils under 10.2134/jeq1991.00472425002000010048x. 
saturated flow conditions. Environ. Technol. 27, 33–40. https://doi.org/10.1080/ Suponik, T., 2010. Adsorption and biodegradation in PRB technology. Environ. Prot. 
09593332708618614. Eng. 36, 43–57. http://epe.pwr.wroc.pl/contents.html. 
Köhne, S., Lennartz, B., Köhne, J.M., Šimůnek, J., 2006. Bromide transport at a tile- Sutherland, R.A., 1998. Loss-on-ignition estimates of organic matter and relationships to 
drained field site: experiment, and one-and two-dimensional equilibrium and non- organic carbon in fluvial bed sediments. Hydrobiologia 389, 153–167. https://doi. 
equilibrium numerical modelling. J. Hydrol. 321, 390–408. https://doi.org/ org/10.1023/A:1003570219018. 
10.1016/j.jhydrol.2005.08.010. Tan, K.L., Hameed, B.H., 2017. Insight into the adsorption kinetics models for the 
Kumahor, S.K., de Rooij, G.H., Schlüter, S., Vogel, H.-J., 2015. Water Flow and Solute removal of contaminants from aqueous solutions. J. Taiwan Inst. Chem. Eng. 74, 
Transport in Unsaturated Sand—A Comprehensive Experimental Approach. https:// 25–48. https://doi.org/10.1016/j.jtice.2017.01.024. 
doi.org/10.2136/vzj2014.08.0105. Thierrin, J., Davis, G.B., Barber, C., 1995. A ground-water tracer test with deuterated 
Liang, C., Chen, Y.J., 2010. Evaluation of activated carbon for remediating benzene compounds for monitoring in situ biodegradation and retardation of aromatic 
contamination: adsorption and oxidative regeneration. J. Hazard. Mater. 182, hydrocarbons. Groundwater 33, 469–475. https://doi.org/10.1111/j.1745- 
544–551. https://doi.org/10.1016/j.jhazmat.2010.06.066. 6584.1995.tb00303.x. 
Liu, J.H., Maity, J.P., Jean, J.S., Chen, C.Y., Chen, C.C., Ho, S.Y., 2010. Biodegradation of Thornton, S.F., Bright, M.I., Lerner, D.N., Tellam, J.H., 2000. Attenuation of landfill 
benzene by pure and mixed cultures of Bacillus spp. World J. Microbiol. Biotechnol. leachate by UK Triassic sandstone aquifer materials: 2. sorption and degradation of 
26, 1557–1567. https://doi.org/10.1007/s11274-010-0331-9. organic pollutants in laboratory columns. J. Contam. Hydrol. 43, 355–383. https:// 
Mahdipanah, H., Tashakori, A., Emamgholizadeh, S., Maroufpoor, E., 2022. An doi.org/10.1016/S0169-7722(99)00105-9. 
experimental study on the determination of dispersion coefficient in layered soils. Thorpe, V.A., 1973. Collaborative study of the cation exchange capacity of peat 
Water Supply 22, 2377–2394. https://doi.org/10.2166/ws.2021.359. materials. J. Assoc. Anal. Chem. (AOAC) 56, 154–157. https://doi.org/10.1093/ 
Maraqa, M.A., Zhao, X., Wallace, R.B., Voice, T.C., 1998. Retardation coefficients of jaoac/56.1.154. 
nonionic organic compounds determined by batch and column techniques. Soil Sci. Topp, G.C., 1993. Soil water content. In: Carter, M.R. (Ed.), Soil Sampling and Methods 
Soc. Am. J. 62, 142–152. https://doi.org/10.2136/ of Analysis. Lewis Publishers, London.  
sssaj1998.03615995006200010019x. Vaezihir, A., Bayanlou, M.B., Ahmadnezhad, Z., Barzegari, G., 2020. Remediation of 
Mohammadi, L., Bazrafshan, E., Noroozifar, M., Ansari-Moghaddam, A., Barahuie, F., BTEX plume in a continuous flow model using zeolite-PRB. J. Contam. Hydrol. 230, 
Balarak, D., 2017. Adsorptive removal of benzene and toluene from aqueous 103604 https://doi.org/10.1016/j.jconhyd.2020.103604. 
environments by cupric oxide nanoparticles: kinetics and isotherm studies. Vidal, C.B., Raulino, G.S., Barros, A.L., Lima, A.C., Ribeiro, J.P., Pires, M.J., 
J. Chemother. https://doi.org/10.1155/2017/2069519. Nascimento, R.F., 2012. BTEX removal from aqueous solutions by HDTMA-modified 
Moura, C.P., Vidal, C.B., Barros, A.L., Costa, L.S., Vasconcellos, L.C., Dias, F.S., Y zeolite. J. Environ. Manag. 112, 178–185. https://doi.org/10.1016/j. 
Nascimento, R.F., 2011. Adsorption of BTX (benzene, toluene, o-xylene, and p- jenvman.2012.07.026. 
xylene) from aqueous solutions by modified periodic mesoporous organosilica. Vignola, R., Bagatin, R., Alessandra De Folly, D., Flego, C., Nalli, M., Ghisletti, D., 
J. Colloid Interface Sci. 363 (2), 626–634. https://doi.org/10.1016/j. Sisto, R., 2011. Zeolites in a permeable reactive barrier (PRB): one year of field 
jcis.2011.07.054. experience in a refinery groundwater—part 1: the performances. Chem. Eng. J. 178, 
Müller, B.R., 2010. Effect of particle size and surface area on the adsorption of albumin- 204–209. https://doi.org/10.1016/j.cej.2011.10.052. 
bonded bilirubin on activated carbon. Carbon 48, 3607–3615. https://doi.org/ 
10.1016/j.carbon.2010.06.011. 
11
F. Obiri-Nyarko et al.                                                                                                                                                                        J o  u  r n  a  l o  f  C  o  n  t a  m  i n  a  n  t  H  y  d  r o  l o g  y  250 (2022) 104054
Waybrant, K.R., Ptacek, C.J., Blowes, D.W., 2002. Treatment of mine drainage using Wright, A.L., Yu, W., Reddy, K.R., 2008. Loss-on-ignition method to assess soil organic 
permeable reactive barriers: column experiments. Environ. Sci. Technol. 36, carbon in Calcareous Everglades wetlands. Commun. Soil Sci. Plant Anal. 39, 
1349–1356. https://doi.org/10.1021/es010751g. 3074–3083. https://doi.org/10.1080/00103620802432931. 
Weber, W.J., Morris, J.C., 1963. Kinetics of adsorption on carbon from solution. J. Sanit. Yang, H., Hu, R., Ruppert, H., Noubactep, C., 2021. Modeling porosity loss in Fe0-based 
Eng. Div. 89, 31–60. https://doi.org/10.1061/JSEDAI.0000430. permeable reactive barriers with Faraday’s law. Sci. Rep. 11, 1–13. https://doi.org/ 
Weiner, E.R., 2008. Applications of Environmental Aquatic Chemistry: A Practical Guide, 10.1038/s41598-021-96599-8. 
2nd ed. CRC Press, Taylor and Francis Group, Boca Raton, FL.  Zawierucha, I., Nowik-Zajac, A., 2019. Evaluation of permeable sorption barriers for 
Wibowo, N., Setyadhi, L., Wibowo, D., Setiawan, J., Ismadji, S., 2007. Adsorption of removal of Cd (II) and Zn (II) ions from contaminated groundwater. Water Sci. 
benzene and toluene from aqueous solutions onto activated carbon and its acid and Technol. 80, 448–457. https://doi.org/10.2166/wst.2019.288. 
heat treated forms: influence of surface chemistry on adsorption. J. Hazard. Mater. Zheng, Z., Aagaard, P., Breedveld, G.D., 2002. Sorption and anaerobic biodegradation of 
146, 237–242. https://doi.org/10.1016/j.jhazmat.2006.12.011. soluble aromatic compounds during groundwater transport. 1. laboratory column 
experiments. Environ. Geol. 41, 922–932. https://doi.org/10.1007/s00254-001- 
0470-2. 
12