The LC-MS assays of paraquat decomposition were conducted by adding H2O2 to unbuffered paraquat solutions pH¼ 5.85 both in the presence and absence of presaturated silica particles Sampl
Trang 1Silica Adsorbents and Peroxide Functionality for
Removing Paraquat from Wastewater Shama F Barna1; Elizabeth A Ott2; Thu H Nguyen3; Mark A Shannon4; and Alexander Scheeline5
Abstract: To treat wastewater containing paraquat, the interaction of the herbicide with unmodified and modified (i.e., chemically treated) proprietary silicas is explored using liquid chromatography—mass spectrometry and ultraviolet-visible spectrometry Strong adsorption and rapid exchange kinetics (i.e., more than 80% within 30 s) of the contaminant onto both unmodified and modified silica surfaces indicate that paraquat adsorption is inherent to the unmodified silica Although activated carbon sequesters paraquat more effectively than silica at long times (i.e., after 2 h), considering both constant mass and constant area approaches in loading the adsorbents, the best performance in the short-term (i.e., at least up to 150 s) interaction was achieved by using unmodified silicas Comparison of data fitting to a competitive binding model with or without consideration of homogeneous solid diffusion indicates that, at least for times relevant to water-treatment applications, the competitive binding model is an appropriate mathematical tool to explain experimental adsorption data Addition of H2O2 is also at-tempted to initiate paraquat decomposition; however, silica is not an effective catalyst for peroxide decomposition paraquat DOI:10.1061/
CE Database subject headings: Water treatment; Adsorption; Wastewater management; Abatement and removal; Kinetics
Author keywords: Water treatment; Paraquat removal; Silica adsorbents; Peroxide functionality
Introduction
Paraquat dichloride, as an acutely toxic herbicide for agricultural
application, has been a major concern in recent years owing
to the associated risk of accidental or intentional contamination
of water with the compound (Clark et al 1966; Simon and
chemical/photocatalytic oxidation, coagulation, and biological
deg-radation, for removal of pesticides and various dyes from water are
reported in the literature (Churchley 1994; Stephenson and Duff
methods in an economical fashion are still continuing topics of
research Photocatalytic treatment of water is not yet an
energy-efficient solution to the problem because photocatalysts that are
stable and capable of providing sustained photocatalysis using low-cost visible lamps or sunlight are as yet unavailable, ultraviolet (UV) light-emitting diodes (LEDs) are still expensive, and chemi-cal treatments using exogenous agents are only effective if required doses of oxidants are low Many of the contaminants (e.g., dyes) are toxic to the organisms used in biological decontamination proc-esses, and therefore such treatments are not always applicable Consequently, for flexibility, ease of operation, and insensitivity to toxic pollutants, adsorption has been a key technology (Ali and
and for facilitating destruction of the contaminant by catalyzed chemical reaction
Although use of activated carbon as an adsorbent material is widespread, commercial production of the adsorbent produces toxic gases and polycyclic aromatic hydrocarbons during carboni-zation of the precursor material and is considered to be an energy-intensive process because of the high temperatures required for thermal activation, regeneration, and reactivation of the material
adsorbents, including different natural minerals (e.g., clay, siliceous materials, or zeolites), agricultural waste (e.g., sawdust, orange peels, or banana peels), and industrial by-products [e.g., bagasse fly ash, iron(III) hydroxide, or red mud], have been investigated for removal of different metal contaminants and dyes from waste-water (Pollard et al 1992; Crini 2006; Gupta and Suhas 2009) Hence, while simultaneously exploring other potential mechanisms
to treat paraquat-containing water (e.g., oxidation of paraquat with
H2O2), research to identify low-cost alternative materials capable
of effectively adsorbing paraquat is needed The idea is to develop
an efficient, rapidly deployable, and economic paraquat-removal system that may offer the simplicity of an adsorption-based mecha-nism, but to minimize the complications of chemical oxidation– based treatment (such as finding a suitable catalyst, generating oxidants, or providing energy-efficient shortwave illumination)
In addition to reports of paraquat uptake onto treated and un-treated activated-carbon products (Dhaouadi and Adhoum 2010),
1 Graduate Research Assistant, Dept of Mechanical Science and
Engineering, Univ of Illinois at Urbana-Champaign, 1206 W Green St.,
Urbana, IL 61801 E-mail: barna1@illinois.edu
2 Chemist I, Abbott Laboratories, 100 Abbott Park Rd., Abbott Park,
IL 60064; formerly, Undergraduate Research Assistant, Dept of Chemistry,
Univ of Illinois, 600 S Mathews Ave 61, Urbana, IL 61801 E-mail:
eaott2@gmail.com
3 Undergraduate Student, Hanoi Univ of Science —VNU, 19 Le Thanh
Tong St., Hanoi 10000, Vietnam; formerly, Undergraduate Research
Assistant, Dept of Chemistry, Univ of Illinois, 600 S Mathews Ave 61,
Urbana, IL 61801 E-mail: hoaithu.hus@gmail.com
4 Deceased; formerly, Professor, Dept of Mechanical Science and
Engineering, Univ of Illinois at Urbana-Champaign, 1206 W Green St.,
Urbana, IL 61801.
5 Professor, Dept of Chemistry, Univ of Illinois, 600 S Mathews Ave.
61, Urbana, IL 61801 (corresponding author) E-mail: scheelin@illinois
.edu
Note This manuscript was submitted on April 10, 2012; approved on
January 30, 2013; published online on February 1, 2013 Discussion period
open until December 1, 2013; separate discussions must be submitted for
individual papers This paper is part of the Journal of Environmental
En-gineering, Vol 139, No 7, July 1, 2013 © ASCE, ISSN 0733-9372/2013/
7-975-985/$25.00.
Trang 2adsorption of the particular contaminant onto different natural
minerals and treated/regenerated minerals (spent and treated
diatomaceous earth, activated clay mineral, kaolinite, illite, and
montmorillonite) has already been explored to determine the
envi-ronmental fate of the compound or estimate the potential of the
adsorbents for treating water containing paraquat (Draoui et al
natural minerals are promising as low-cost adsorbents for
water-treatment applications, the primary constituents that govern the
ad-sorption process are still undetermined, and therefore the adsorbing
characteristics reported in the literature are case specific and may
vary depending on the constituents and origins of the materials
With oxygen (46.1% by weight) and silicon (28.2% by weight)
being the first and second most abundant elements on earth,
respec-tively (Lide 1996), silica is a widely available and inexpensive
material to manufacture A thermodynamic study of paraquat
ad-sorption onto different minerals concluded that the cationic
ex-change capacity of silica can play a major role in the interaction
of the compound with cationic organic compounds (Draoui et al
1999) Though a recent investigation reports uptake of paraquat
onto surfaces of silica modified with titania (Brigante and Schulz
2011), a detailed examination of paraquat uptake onto unmodified
silica with insights into the overall uptake kinetics and efficacy
of the material as a water-treatment agent does not appear in the
literature By utilizing ultraviolet-visible absorption
spectropho-tometry (henceforth, UV-Vis) and liquid chromatography—mass
spectrometry (LC-MS) as analytical tools, the individual
interac-tions of five unmodified and 11 modified proprietary silica
materi-als with paraquat are experimentally investigated The idea is to
compare the adsorption performance of the unmodified and
modi-fied materials and thus identify the intrinsic adsorption
character-istics of unmodified silica The nature of the paraquat—silica
interaction is characterized by modeling the experimental data with
well-known mathematical adsorption equations The UV-Vis assay
was repeated for commercially available activated carbon to have a
basis of comparison for paraquat-removal efficiency of the silica
products Knowledge obtained from the particle system can suggest
the required particulate loading and configuration (e.g., serial,
par-allel, tubular, or spiral) of silica-coated membranes to treat water
containing paraquat Additionally, the interaction between H2O2
and dicationic paraquat in aqueous solution was also explored to
estimate the effectiveness of peroxide in removing the contaminant
Materials and Methods
Reagents
Paraquat dichloride as purchased from Sigma Aldrich Because
paraquat salts are highly hygroscopic, the powder was dried at
100°C for 5 h immediately before use to ensure that it reached
con-stant weight before being used as a standard After drying the
pow-der was cooled and stored in a desiccator To prepare stock paraquat
solutions, the dried paraquat salt was weight using an adequately
sensitive Mettler model XP26 microbalance, and then sample
para-quat solutions of various concentrations were prepared by diluting
aliquots of the stock solutions
All silica samples used are proprietary to PPG Industries; to
pro-tect their identities, samples are referred to using a sample number
throughout the article Samples 1–4 and Sample 6 are
unmodi-fied silica, whereas the rest of the silica samples are modiunmodi-fied
(i.e., chemically treated) by PPG Industries Among the unmodified
silica samples, Sample 1 and Sample 3 are commercially known as
Hi-Sil 135 and Lo-Vel 6000 and have particle size of 10 and 16μm, respectively (PPG Industries 2012) Activated carbon (Darco G60) was purchased from Fisher Scientific As per the information pro-vided by the supplier, the product is steam activated, contains 100% (weight) carbon, and particle size distribution is 100–325 mesh (44–149 μm) All the samples except silica samples 13–16 were tested as received Samples 13–16 were dried at 100°C before use,
as recommended by the supplier
For the H2O2 oxidation experiments, commercially available standard 0.1-M ceric sulfate [CeðSO4Þ2] solution was used to standardize H2O2 The concentration of the reagent-grade H2O2 used was found to be 30.5% w=v
Analytical Methods
Spectrophotometric Assay All spectrophotometric measurements were made on a Hewlett-Packard (now Agilent) 8452A diode-array spectrophotometer Two different approaches are available for spectrometric deter-mination of paraquat residues in water (Yuen et al 1967;Ashley
1 Measuring UV absorption of paraquat ions in aqueous solution (because there is a greater tendency of formulation additives
to interfere at the lower wavelengths, this method can only be considered reliable for very pure aqueous solutions of para-quat); and
2 Generating paraquat radicals by alkaline dithionite reduction, followed by determination of paraquat residues from the peak
in the visible absorption spectrum (this method is considered convenient for determination of paraquat in formulated products)
Because the calibration for the method provided a reproducible detection limit, UV-Vis absorptiometry was chosen to be the appro-priate method for this research Further, any degradation might in-terfere chemically or spectroscopically with the assay Because the paraquat solutions prepared in the laboratory are free of additives, the possibility of interference by any undesired compound other than degradation products is moot
Calibration of the spectrophotometer for UV absorption of para-quat at 258 nm was performed at pH¼ 8.06 (a 0.025-M borate buffer) A calibration curve for H2O2 at the peak wavelength of paraquat absorption (258 nm) was also produced to facilitate under-standing of the chemistry in H2O2-added paraquat solutions The second calibration was performed at pH¼ 5.85 (in acetate buffer)
so that the spectrometric results for H2O2-added paraquat solu-tions can be compared with the LC-MS results for unbuffered (pH¼ 5.85) H2O2-added paraquat solutions The pH of the
solu-tions was measured using a pH meter, and absorbance measure-ments of the solutions were made by the Hewlett-Packard 8542A diode-array spectrophotometer Linear regression analysis was used to construct Beer’s Law working curves Working curve linear regression equations, including 95% confidence intervals for the regression coefficients, are as follows:
• For paraquat in borate buffer at pH 8.06 (ppm = parts per million): Að258 nmÞ ¼ ð0.0716 0.0015ÞCparaquatðppmÞ þ ð0.03 0.02Þ), R2¼ 0.9956; and
• For H2O2Að258 nmÞ ¼ ð0.0181 0.0001ÞCparaquatðppmÞ þ ð0.001 0.003Þ), R2¼ 0.9997
Liquid Chromatography–Mass Spectrometry Analysis
Chromatographic separation was performed using a Waters 2795 separation module (reverse-phase high-pressure LC) equipped with
a quaternary solvent delivery system, autosampler, and column heater A 2.1- × 50-mm Eclipse XDB C18 column was used
Trang 3Before testing the samples, the column was first decontaminated
from previously used solutions by running a 100-ppm unbuffered
aqueous solution of paraquat through the system
Gradient elution was used for separation of the aliquots by
vary-ing the polarity of the mobile phase Solvent A was a mixture of
95% water and 5% acetonitrile, and Solvent B was a mixture of
95% acetonitrile and 5% water The elution solvent used was
ini-tially 100% Solvent A Over 9 min, the solvent mixture was
gradu-ally changed to 100% Solvent B From 9–10 minutes, the solvent
mixture was rapidly returned to 100% Solvent A
Mass spectrometry was carried out using a Waters Quattro
Ultima mass spectrometer equipped with electrospray ionization
and operated using MassLynx 4.1 MS software (Waters 2012)
Ionization takes place in the source at atmospheric pressure The
ions are sampled through a series of orifices into the first
quadru-pole, where they are filtered according to their mass-to-charge ratio
(m=z) Data acquisition was performed between m=z ¼ 100
and m=z ¼ 500
The BET-N2–specific surface area measurement was performed
using a Quantachrome Nova 2200e instrument The specific
surface areas for N2adsorption at 77.3 K (−195.7°C) were
calcu-lated by the instrument using the BET model in the range of
0.1 < P=Po < 0.2
Experimentation
All experiments were conducted at room temperature (25°C) For
spectrometric analysis of paraquat-adsorption kinetics, each of
the silica samples was mixed separately with buffered proxy
(par-aquat dichloride) solution The initial concentration of par(par-aquat
used in each trial was 25 ppm Each silica sample was added to
the 30-mL paraquat solution to give a final silica concentration
of 1.67 g=L
The system was kept in suspension using a magnetic stirrer For
both UV-Vis and LC-MS (discussed in the following paragraph)
analyses, the solution aliquots were collected using 0.45-μm
Millipore filters to filter out the silica particles The effect (e.g., loss
of paraquat from sample solution) of filters on the measurements
was investigated and found to be negligible For UV-Vis
measure-ments, sample aliquots were initially collected at regular intervals
of 5 s for the first 30 s, and then at intervals of 30 s for the next
120 s To determine the saturation adsorption capacity of silica
par-ticulates, three additional data were also collected for each of the
samples over 48 h The idea was to determine when the paraquat—
silica interaction reaches equilibrium and thus to identify the total
useful adsorption capacity using the equilibrium adsorption data
The first few drops of the filtrate were used to decontaminate the
cuvette from previous solutions, and the rest was used for
absorb-ance measurement with the spectrophotometer
The LC-MS analysis of the samples was conducted to
investi-gate whether the paraquat uptake onto the silica particles is
accom-panied by any chemical dissociation in the presence or absence of
H2O2 To reduce noise in the mass spectra from buffer components,
100-ppm paraquat solutions were prepared using only deionized
water For the LC-MS analysis, three different silica concentrations
(1.3, 1.67, and2 g=L) were added separately to 5 mL of the
para-quat dichloride solutions, and the system was kept in suspension
using a magnetic stirrer The idea behind varying the silica
concen-tration was to investigate whether any reaction of paraquat with
silica can be initiated
The LC-MS assays of paraquat decomposition were conducted
by adding H2O2 to unbuffered paraquat solutions (pH¼ 5.85)
both in the presence and absence of presaturated silica particles
(Samples 1–12) Paraquat solutions were separately mixed with the silica samples for 2 min before adding H2O2to the solutions The aim of this method was to saturate the active adsorption sites
on the surface of the adsorbents and thus to distinguish between the contributions of adsorption and H2O2-assisted decomposition in removing paraquat
To allow adsorption data normalization, BET-N2 analysis was conducted to determine the specific surface areas of the silicas Because there is restriction on which data can be shared on the modified materials that are proprietary to PPG Industries, all the unmodified silicas (Samples 1–4 and Sample 6), two modi-fied silicas (Sample 5 and Sample 10), and the activated-carbon sample were analyzed The instrument allows analyzing two sam-ples at a time Before conducting the analysis, each of the two empty sample cells was weighed Then, samples were loaded into the cells and degassed at 300°C for 3 h to remove volatile species from the materials The sample cell containing the sample was weighed again after degassing The idea was to determine the exact quantity of each sample used because the information is needed for specific surface area calculation by the instrument Results of the specific surface area measurements are presented
in Table 1 The BET surface area obtained by this analysis for activated carbon (Darco G60) was found to be similar to the BET measurement reported in previously published literature
Modeling of Adsorption Kinetics
Langmuir Adsorption Model According to Langmuir, adsorbate molecules are assumed to bind
to distinct empty sites on the surfaces of the adsorbent through the following reversible process (Zuyi and Taiwei 2000; Chiron
A þ S
ka
⇌
kd
A − S
where A = adsorbate molecule; S = adsorption site on surface of adsorbent; ka and kd = adsorption and desorption rate constants, respectively; and A − S = complex formed owing to adsorption
of solute molecule onto adsorbent surface
The basic idea behind the Langmuir adsorption model is the for-mation of a homogeneous adsorbent surface by a monomolecular layer of adsorbate molecules at isoenergetic, uncooperative sites
As per this adsorption model, the fraction of the adsorbent surface occupied by the solute (θ) can be expressed as a nonlinear function
of solute concentration (C) by the following equation:
1 þ KcC Kc¼ka
Furthermore, by balancing the relative rates of uptake and release, the Langmuir adsorption isotherm rate of change of surface coverage (dθ=dt) can be modeled [Eq (2)] and then can be inte-grated to predict the adsorption as a function of time (Chiron
dθ
Integrating Eq (2) and taking K0¼ C=ðC þ kd=kaÞ and
K0 0¼ kaC þ kd, the fractional surface coverage can be calculated
as a function of time:
Trang 42 =g)
2)
2)
Trang 5θðtÞ ¼ K0ð1 − e−K tÞ ð3Þ Competitive Binding Adsorption Model
Competitive binding assays are based on the idea that two
mole-cules, A and B, compete for the same adsorption site, and
adsorp-tion of one molecule (A) onto an adsorpadsorp-tion site is accompanied by
the desorption of the previously adsorbed molecule (B) from that
particular site (Sklar et al 1985;Bachas and Meyerhoff 1986) The
adsorption process can be represented by the following interaction:
A þ B − S
ka
⇌
kd
A − S þ B
Assuming all the surface sites are initially saturated with
mol-ecule B before adsorption of A occurs, fractional coverage of the
surface by adsorption of molecule A (θA) can be expressed as a
function of the bulk solution concentration (C) by balancing the
adsorption and desorption rates of A at equilibrium:
where θ ¼ ðθAþ θBÞ = fraction of surface sites occupied by
adsorbates A and B at equilibrium; and K ¼ ðka=kdÞ = ratio of
ad-sorption constant to dead-sorption constant of A For this model, the
adsorption kinetics is governed by the following equation:
θAðtÞ ¼1 þ KK1t
where Cs= liquid-phase concentration of paraquat at liquid—silica
interface
Homogeneous Solid Diffusion Model
Governing Equation with Boundary and Initial Conditions
The homogeneous solid diffusion model (HSDM) [Eq (6)]
ac-counts for transport of the adsorbate within a spherical particle
and has been used in the literature in combination with popular
adsorption isotherms (i.e., Langmuir or Freundlich isotherms) to
model adsorption (McKay 1998;Meshko et al 1999,2001;Veliev
∂q
∂r¼r12Ds
∂
∂rr2
cq
where Ds= diffusivity of contaminant and can be assumed as
con-stant (Meshko et al 2001) in equation; r = spatial coordinate along
radial direction inside particle; q = solid-phase concentration of
adsorbate and is a function of position (r) and time (t); and R =
radius of spherical adsorbent
The boundary conditions are
∂q
where V = volume of solution used; and M = quantity of silica used
The axisymmetry stated in Eq (7b) arises from the assumption that
the adsorbent is spherical (Meshko et al 1999) If adsorption at
surface sites and internal mass transfer within the particle are considered, Eq (7c) can be imposed, and external mass transfer resistance in the hydrodynamic film of the particle is considered negligible
The initial condition is
Method of Computation The diffusion equation provided in Eq (7a) was discretized follow-ing the Crank-Nicholson scheme (Veliev et al 2006):
2 rj−1
h2
þ rj
h2
where T = total number of time steps; and J = total number of nodes
in radial direction from center to surface of particle
The set of equations that result from Eq (8) at different nodes (j) were solved numerically in MATLAB (MathWorks 2010) for different time steps (m) by formulating a matrix based on the tri-diagonal matrix algorithm (TDMA) (Conte and deBoor 1972)
Results and Discussion
Adsorption Kinetics of Paraquat
A significant decrease in paraquat concentration was noticed within the first 5 s after adding any of the silicas (Figs.1and2) Sample 3,
an unmodified silica, outperformed any of the samples tested when percentage of paraquat-removal data for the first 30 s were consid-ered (Table1) Compared with the 77% paraquat adsorption onto activated carbon, more than 80% of paraquat was removed within
30 s of first paraquat–silica interaction by unmodified silica Samples 3 (91%) and 6 (90%) and modified Samples 5 (88%) and 7 (82%)
For all tested samples except Samples 1, 9, 11, and 15, paraquat adsorption tends to reach equilibrium within the maximum
0 5 10 15 20 25 30
Time (sec)
Sample-1 Sample-2 Sample-3 Sample-4 Sample-5 Sample-6 Sample-7 Sample-8
Fig 1 Measured paraquat concentrations for silica Samples 1–8 up to
30 s after adding particles [sample aliquots were collected at 5-s inter-vals; solution pH was 8.04 in borate buffer; silica concentration in 25-ppm paraquat solution (initial concentration) was1.67 g=L]
Trang 6observation time frame of 48 h (Fig.3) Paraquat adsorption after
48 h was found to be the greatest onto activated-carbon powders;
however, more than 96% of paraquat was removed from solution
with unmodified silica Samples 3, 4, and 6, and modified silica
Samples 5, 8, 13, and 16 (Table 1)
The BET-N2measurements enabled comparison of the
adsorp-tion data of some selected samples (Samples 1–6, Sample 10,
and activated carbon) for normalized surface area (i.e., paraquat
adsorbed per unit surface area) (Table1) For a unit area of
adsorp-tion, the short-term performance of the samples can be ranked
as Sample1 > Sample 2 > Sample 5 > Sample 6 > Sample 3 >
Sample4 > Sample 10 > activated carbon, and the long-term
interaction can be ranked as Sample1 > Sample 10 > Sample 2 >
Sample5 > Sample 6 > Sample 4 > Sample 3 > activated carbon:
Paraquat adsorbed per unit area was maximum for Sample 1
(42.5 g=m2after 30 s and91.2 μg=m2 after 48 h) and the lowest
for activated carbon (11.6 μg=m2after 30 s and16.9 μg=m2after
48 h) Although activated carbon asymptotically adsorbed all para-quat present, it is not obvious that the long-term behavior is suffi-ciently better than silica to warrant its use, particularly given its poorer short-term performance
An apparently exponential trend in the adsorption time series was evident for Samples 10, 11, and 13 in which removal rate for Sample 13 was relatively rapid (Fig 2) Initially, paraquat-adsorption rate by Sample 10 was the slowest; however, total ad-sorptive paraquat removal by the sample was found to be higher than that by Samples 11, 12, and 15 (Fig.3and Table1) As in-formed by the supplier, Sample 10 is granulated and has low sur-face area by configuration, and therefore the sample offers fewer available adsorption sites but more particles in solution compared with the same quantity of other silica particles with larger surface areas (Table 1)
Considering the maximum adsorption and paraquat-removal rates, Sample 3 (which is an unmodified silica) exhibited the best performance among all the tested silica in that 93% of para-quat was removed from the heterogeneous solution within the first
150 s after adding the silica
Investigation of adsorption behavior at different initial paraquat concentrations indicates that the maximum amount of paraquat that can be adsorbed by a silica sample is dependent on the paraquat concentration of the solutions Fig 4(a) presents a comparison
of the measured decrease in paraquat 1 min after adding silica (Samples 1–8) With increasing initial concentration, the percent-age of paraquat removal was found to decrease, whereas the total amount of contaminant adsorbed by a fixed dose of silica (1.67 g=L) increased Further investigation of the adsorption behavior using Sample 6 at different initial paraquat concentrations reveals that the paraquat uptake by silica, at least within the ob-served time frame (1 min), is dominated by equilibrium phenomena until the surface of the adsorbent is saturated [Fig.4(b)] Approx-imately 90% paraquat-removal efficiency was achieved in 1 min for paraquat solutions of different initial concentrations up to 27 ppm, but beyond initial concentrations of 27 ppm, saturation of the silica surface started to control the paraquat-uptake process Because loading of the silica particles was kept constant for all concentra-tions, increasing the initial solution concentration at this stage ac-tually increased the number of paraquat molecules competing for a single adsorption site Consequently, there is a reduction in instant adsorption (e.g., percentage of paraquat adsorbed); however, the mass transfer rate of the solute is enhanced because of a higher-concentration gradient between the bulk solution and adsorption interface resulting in an increase of total paraquat adsorption onto the adsorbents
Fig 5(a) exhibits the chromatograms for the control sample (100-ppm paraquat solution) For all the initially received silica samples (Samples 1–8) except Sample 5 [treated with polyethylene glycol (PEG)], the chromatograms were similar to Fig.5(a)for all the silica concentrations (1.3, 1.67, and2 g=L) attempted and in-dicate that the samples do not react with paraquat
The chromatogram for Sample 5 [Fig.5(b)] displayed additional elution peaks from retention time t ¼ 2.54–4.08 min The com-bined mass spectrum, taken from t ¼ 3.61–4.08 min, displays mass peaks at m=z values higher than that of the precursor ion, ½PQþ, and thus indicates the inclusion of additional
com-pounds into the solution, most likely from the treated silica surface (Sample 5) The presence of the mass peaks at regular m=z intervals
of 44.0 further confirms that oligomers of PEG are present, because the monomer of PEG (-CH2-O-CH2) has a fragment mass of 44.0 Dalton
0
5
10
15
20
25
30
1 10 100 1,000 10,000 100,000 1,000,000
Time (sec)
"Sample-13 (dried)" "Sample-14 (dried)" "Sample-15( dried)"
"Sample-16 (dried)" Activated Carbon
Fig 3 Adsorption kinetics of paraquat onto tested particles over 48 h,
with time plotted on log scale [pH of solution was 8.06 (using borate
buffer); activated carbon or silica in 25-ppm paraquat solution (initial
concentration) was1.67 g=L]
0
5
10
15
20
25
Time (sec)
Sample-9 Sample-10 Sample-11 Sample-12 "Sample-13 (dried)" "Sample-14 (dried)"
"Sample-15( dried)" "Sample-16 (dried)" Activated Carbon
Fig 2 Measured paraquat concentrations after adding silica Samples
9–16 [samples were collected up to 30 s after adding silica particles;
sample aliquots were collected at 5-s intervals; solution pH was 8.04 in
borate buffer; silica concentration in 25-ppm paraquat solution (initial
concentration) was1.67 g=L]
Trang 7Model Predictions
A comparison between mathematical model predictions and
exper-imental adsorption data for paraquat uptake onto Sample 10 is
demonstrated in Fig 6(a) The plot demonstrates a poor fit of the Langmuir isotherm model to the experimental data [Fig.6(a)] Because the nature of solute adsorption onto particulates is more complicated in a solution system compared with gas adsorption on
a homogeneous solid surface, factors like solute–solvent interac-tion, competitive adsorption between a solute and the solvent, sur-face heterogeneity, and solution conditions can come into play
to describe the paraquat-uptake process onto silica
Fitting with the competitive binding model was then attempted Paraquat-adsorption kinetics were modeled, assuming that all the surface adsorption sites are occupied with water before paraquat adsorption occurs The model was found to be in good agreement with the experimental data for Sample 10 through the first 1,200 s [Fig 6(b)] assuming 72% surface coverage (θ) at equilibrium (= 0.72) and constant K1¼ 0.03 s−1 The consistency of the
mod-eling and experimental results suggests that the adsorption kinetics
of the silica particles are governed by competitive adsorption of water and paraquat However, comparison of the competitive bind-ing model prediction with long-time experimental data over 48 h exhibits that the model deviates from experimental data over time
Fig 5 Chromatograms: (a) aqueous solution of paraquat dichloride;
(b) paraquat solution mixed with Sample 5 (initial concentration of
paraquat dichloride solution was 100 ppm)
0
5
10
15
20
25
30
35
5
(a)
(b)
Initial Concentration (ppm)
Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Sample 6 Sample 7 Sample 8
0
10
20
30
40
50
60
70
80
90
100
Initial Concentration of Paraquat
% Paraquat removal 90% removal
Fig 4 Effect of initial concentration on paraquat adsorption: (a)
com-parison among Samples 1–8; (b) detailed analysis for Sample 6
5 10 15 20 25
0
(a)
(b)
20 40 60 80 100 120 140
Time (sec)
Experimental Competitive Binding Model Langmuir Model
1 10 100
1 10 100 1,000 10,000 100,000 1,000,000
Time (sec)
Experimental Competitve Binding Model Langmuir Model
Fig 6 Comparison between mathematical model predictions and ex-perimentally obtained adsorption data for Sample 10, with time plotted
in logarithmic scale: (a) over 150 s; (b) over 48 h [for all experiments, silica in 25-ppm paraquat solution (initial concentration) was1.67 g=L; exponential Langmuir model fit decays more slowly than experimental data; competitive model fits well with experimental data collected for first 2 h, but actual paraquat concentration is much lower than model predicts after 48 h]
Trang 8The final paraquat concentration was found to be lower than the
extrapolated model fit after 48 h [Fig.6(b)]
The initial understanding was that model predictions deviated
from long-time (i.e., after 2 h and up to 48 h) adsorption data
be-cause the competitive binding model does not consider the presence
of pores on the silica particles Therefore, the molecules adsorbed
onto the surface of the silica particles may diffuse into the
adsorb-ents over time, and thus the particles actually offer available surface
sites for more adsorption than the model estimates at later stages of
interaction However, when the homogeneous solid diffusion
equa-tion was applied in combinaequa-tion with the competitive binding
model, the concentration variation of paraquat inside the pore
in-dicates that the paraquat transport to the pore is largely complete
within the first 40 s (i.e., for Sample 1) of paraquat–silica
interac-tion (Fig.7) Fig.7demonstrates the variation of concentration
gra-dient of paraquat inside the pore for Sample 1 at different time
steps The results elucidate that paraquat diffusion through the
pores is not controlling the long-time adsorption behavior, and the
increased paraquat adsorption is likely attributable to the
availabil-ity of more surface sites at long times
Therefore, to model the adsorption data, paraquat adsorption
onto the silica was then modeled by assuming a two-stage
adsorp-tion process—namely, a fast competitive binding phase in
combi-nation with HSDM—to consider external and internal diffusion,
and a slow phase controlled by intraparticle diffusion The idea
posed in this paper is similar to the double exponential model
pro-posed in the literature to model long-time adsorption of copper and
lead onto activated carbon (Wilczak and Keinath 1993) and to the
observation made on diffusion of protein in membranes (Jacobson
The competitive binding process equation considered in this
case is as follows:
θA¼
kfast
1 þ kfasttþ kslow
1 þ kslowt
where kfirstand kslow= rate constants for first and second
competi-tive binding process; and
where tc= critical time after which second process begins; and B was considered to be constant
Fig.8demonstrates the fitting curve, obtained after incorporat-ing the HSDM into the adsorption model When only one competi-tive process was considered, the fitting was similar to the competitive binding curve shown in Fig.6(b); however, the total surface coverage (θ ¼ 0.33) was found to be lower in this case Because the inclusion of the diffusion equation enabled the model
to consider additional adsorption inside the pores, the total adsorp-tion predicted can be the same, even in the presence of the fewer available surface sites [Fig.8(a)] Among the models applied, the two-stage adsorption model was found to be the only method for which the fitting predictions were in the vicinity of the experimen-tal adsorption data [Fig.8(b)] at long times
Addition ofH2O2to Paraquat Solutions Because adsorption was found to be the only active sequestra-tion or degradasequestra-tion mechanism during paraquat–silica interac-tions, the addition of exogenous oxidant (H2O2) to heterogeneous
3.25
3.50
3.75
4.00
4.25
4.50
r (µm)
t=10s t=20s
t=40s t=150s
Fig 7 Solid-phase concentration variation along radius of Sample 1 at
different time (t) steps (r ¼ 0 and 16 μm denotes center and surface of
silica particles, respectively)
1 10 100
1
(a)
(b)
10 100 1,000 10,000 100,000 1,000,000
Time (sec)
1 10 100 1,000 10,000 100,000 1,000,000
Time (sec)
Experimental One competitive binding model in combination with HSDM
1 10 100
Experimental Two-step adsorption model in combination with HSDM
Fig 8 Comparison between: (a)MATLABcomputation results for one competitive binding process in combination with HSDM and experi-mental data [fitting parameters are k1¼ 0.32ðs−1Þ), θ ¼ 0.33, and
Ds¼ 3 × 10−5m2=s]; (b) two-step adsorption model in combination with HSDM and experimental adsorption data [fitting paramters are
kfast¼ 0.32ðs−1Þ), tc¼ 150 s, B ¼ 1.3 × 10−10ðs−2Þ, θ ¼ 0.33, and
Ds¼ 3 × 10−5m2=s
Trang 9proxy solutions (with silica) was attempted to provide a reactant
so that silica might be able to catalytically decompose paraquat
Initially, the interaction between the compounds (i.e., H2O2
and paraquat) was investigated spectrometrically by introducing
varying concentrations of H2O2 to 0.48 mM of buffered paraquat
solutions (Fig 9) The idea was to look for unstable complex
formation and better understand the nature and stoichiometry of
interaction of paraquat and H2O2 so that viable techniques to
catalyze paraquat decomposition could later be attempted
As illustrated in Fig 9, the absorbance peak of the solution
shifted to lower wavelengths for higher concentrations of H2O2
in solution, which was initially interpreted as signifying the
pres-ence of light absorbing unknown compounds However, an
alter-native explanation for the peak shift in Fig.9 was that the shift
of the absorption peak was attributable to addition of H2O2 and
an apparent shift independent of any association or reaction
Therefore, the absorption spectra in Fig 9 were replotted,
sub-tracting the putative absorbance of H2O2 (Fig 9) No additional
absorbance peak was identified, and the shift of the absorption
peak corresponding to paraquat became less evident with increas-ing H2O2 concentrations in solutions (Fig 10) The apparent shift observed was actually attributable to the relative position
of the paraquat- and H2O2-absorption peaks, such that increasing peroxide absorption raised the short-wavelength band wing ab-sorption and shifted the wavelength of peak abab-sorption In addi-tion, H2O2 modifies the solution hydrogen bonding matrix, which ultimately shifts the absorption band of water to longer wavelengths (Higashi et al 2008) The results point to the con-clusion that H2O2alone does not have any significant interaction with dicationic paraquat
The LC-MS analysis of H2O2-added paraquat solutions with
or without the silica (Fig 11) was also in agreement with the spectrometric results No decomposition was evident for any aliquots tested To identify whether the absence of any reaction was attributable to the reduction of surface active sites (before adding H2O2) on silica by preadsorbed paraquat, a silica sample (Sample 2) and H2O2 were added simultaneously to paraquat solution However, no change in the chromatogram of Fig 11
was observed, which indicates that the heterogeneous solution remains nonreactive, even in the presence of putatively reactive surface sites
0.0
0.5
1.0
1.5
2.0
Wavelength (nm)
0.48 mM paraquat with 2.4 mM hydrogen peroxide 0.48 mM paraquat with 14.2 mM hydrogen peroxide 0.48 mM paraquat with 37.5 mM hydrogen peroxide 0.48 mM paraquat with 48.9 mM hydrogen peroxide 0.48 mM paraquat with 71.2 mM hydrogen peroxide 0.48 mM paraquat with 93.0 mM hydrogen peroxide 98.3 mM hydrogen peroxide
Fig 9 Absorbance spectrum of H2O2-added paraquat solutions for
varying H2O2 concentrations; ½paraquat ¼ 0.48 mM; labels indicate
concentration of H2O2added (solution pH was 5.87 in acetate buffer)
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Wavelength (nm)
0.48 mM paraquat with 2.4 mM hydrogen peroxide 0.48 mM paraquat with 8.3 mM hydrogen peroxide 0.48 mM paraquat with 14.2 mM hydrogen peroxide 0.48 mM paraquat with 20.1 mM hydrogen peroxide 0.48 mM paraquat with 25.9 mM hydrogen peroxide 0.48 mM paraquat with 31.7 mM hydrogen peroxide 0.48 mM paraquat with 37.5 mM hydrogen peroxide 0.48 mM paraquat with 43.2 mM hydrogen peroxide 0.48 mM paraquat with 48.9 mM hydrogen peroxide 0.48 mM paraquat with 54.5 mM hydrogen peroxide 0.48 mM paraquat with 60.1 mM hydrogen peroxide 0.48 mM paraquat with 65.7 mM hydrogen peroxide 0.48 mM paraquat with 71.2 mM hydrogen peroxide 0.48 mM paraquat with 76.7 mM hydrogen peroxide 0.48 mM paraquat with 82.1 mM hydrogen peroxide 0.48 mM paraquat with 87.6 mM hydrogen peroxide 0.48 mM paraquat with 93.0 mM hydrogen peroxide 0.48 mM paraquat with 98.3 mM hydrogen peroxide
Fig 10 Corrected absorbance spectrum (i.e., corrected absorbance= absorbance of H2O2-added paraquat solution minus computed, scaled
ab-sorbance of H2O2in that solution) paraquat solutions for varying H2O2 concentrations; ½paraquat ¼ 0.48 mM; labels indicate concentration of
H2O2added (solution pH was 5.87 in acetate buffer); only shift is attributable to offset in baseline, which may be attributable to relative position
of paraquat and H2O2 absorption peaks or modification of solution hydrogen bonding matrix by H2O2
Fig 11 Chromatogram of 100-ppm paraquat solutions when mixed with approximately 50 mM H2O2and no silica; similar chromatogram were observed by adding both H2O2and silica to paraquat solutions
Trang 10Silica was found to be a promising adsorbent for
water-treatment applications in which maximum removal of paraquat
within a limited time of interaction is critical Although activated
carbon sequestered paraquat more effectively than silica at long
times (i.e., after 2 h and up to 48 h), considering both constant mass
and constant surface area approach in loading the adsorbents, the
best performance in short-term (i.e., at least up to 150 s) interaction
was achieved by using unmodified silicas
When the constant mass approach was adopted for loading the
adsorbents, all the unmodified silica samples except Sample 1 were
found to be good adsorbents for time periods relevant to water
puri-fication, and percentage of paraquat removal within 30 s was higher
than that achieved by activated-carbon sample for three (Samples 3,
4, and 6) of the five unmodified silicas In fact, the unmodified
silica Samples 3 and 6 outclassed all the tested samples when
short-term adsorption data were compared
After long-term interaction, at least 90% paraquat removal was
achieved for all the unmodified silicas The paraquat adsorption
observed onto the surfaces of different base materials (e.g., silica)
can certainly aid removal of the contaminant when appropriate
re-agents are incorporated into the adsorbent, thus holding the promise
to optimize efficiency of a regenerable membrane-based
water-treatment system with a low carbon footprint
Silica was also found to be a suitable adsorbent over activated
carbon for both short-time and long-time application when size or
area of the membrane system is an important design consideration
Normalized adsorption data exhibit that, for the same surface area
of the adsorbents, all the silicas analyzed (Samples 1–6 and 10)
adsorb more paraquat than the activated carbon during the entire
observation period (i.e., 48 h)
Concentration-dependency data demonstrate that paraquat
up-take is equilibrium limited for the first 1 min after adding fresh
silica The results suggest that when immediate protection against
paraquat contamination is the primary concern, effective cleanup of
the paraquat solution is possible by simply passing the solution
through a series of membranes coated with clean unmodified silica
Agreement between short-time (i.e., adsorption data up to
1,200 s) experimental adsorption data and single-step competitive
binding model predictions also indicates competitive adsorption
of both paraquat and water molecules onto silica particles Among
the adsorption models attempted, predictions using a two-step
com-petitive binding approach with consideration of the homogeneous
solid diffusion model were found to be in the vicinity of the
long-time adsorption data However, neither the modified nor
unmodi-fied silica reacted with dicationic paraquat, either in the presence or
the absence of H2O2 in solution
Future work on this research will be to utilize the adsorption
results obtained for the materials in suspension to investigate
para-quat removal using membranes containing silica so that the best
paraquat-adsorbing silica candidate can be identified, and
mecha-nisms of decomposing or sequestering paraquat can be successfully
implemented
Acknowledgments
This material is based upon work supported by the Engineer
Research and Development Center–Construction Engineering
Regulation Laboratory (ERDC-CERL) under Contract No
W9132T-09-C-0046 Work was further supported and, in part,
collaboratively guided by PPG Industries Sincere
acknowledg-ment is given to Professor Kenneth Suslick, Maryam Sayyah,
Edward Chainani, and Furong Sun from the School of Chemical
Sciences, and Dr Glennys A Mensing from the Department of Mechanical Science and Engineering at the University of Illinois
at Urbana Champaign, for their continued support in providing the logistics and scientific infrastructure required to conduct this research
References Ali, I., and Gupta, V K (2007) “Advances in water treatment by adsorp-tion technology ” Nat Protoc., 1(6), 2661 –2667.
Ashley, M G (1970) “Collaborative study of the determination of para-quat and dipara-quat in formulated products ” Pestic Sci., 1(3), 101 –106.
Bachas, L G., and Meyerhoff, M E (1986) “Homogeneous enzyme-linked competitive binding assay for the rapid determination of folate
in vitamin tablets ” Anal Chem., 58(4), 956 –961.
Brigante, M., and Schulz, P C (2011) “Adsorption of paraquat on mesoporous silica modified with titania: Effects of pH, ionic strength and temperature ” J Colloid Interface Sci., 363(1), 355 –361.
Çeçen, F., and Özgür, A (2011) Activated carbon for water and waste-water treatment: Integration of adsorption and biological treatment, Wiley, Weinheim, Germany.
Chiron, N., Guilet, R., and Deydier, E (2003) “Adsorption of Cu(II) and Pb(II) onto grafted silica: Isotherms and kinetic models ” Water Res.,
Churchley, J H (1994) “Removal of dye wastewater color from sewage effluent: The use of a full-scale ozone plant ” Water Sci Tech., 30(3),
275 –284.
Clark, D G., McElligott, T F., and Hurst, E W (1966) “The toxicity of paraquat ” Br J Ind Med., 23, 126–132.
Conte, S D., and deBoor, C (1972) Elementary numerical analysis, McGraw-Hill, New York.
Crini, G (2006) “Non-conventional low-cost adsorbents for dye removal:
A review ” Bioresour Technol., 97(9), 1061 –1085.
Dhaouadi, A., and Adhoum, N (2010) “Heterogeneous catalytic wet peroxide oxidation of paraquat in the presence of modified activated carbon ” Appl Catal B, 97(1), 227 –235.
Draoui, K., Denoyel, R., Chgoura, M., and Rouquerol, J (1999) “Adsorp-tion of paraquat on minerals: A thermodynamic study ” J Therm Anal.
Dun, J W., Gulart, E., and NG, K Y S (1985) “Fischer tropsch synthesis
on charcoal supported molybodenum: The effect of preparation and potassium promotion on activity and selectivity ” App Catal., 15(2),
Gupta, V K., Carrott, P J M., and Ribeiro Carrott, M M L., and Suhas., Z (2009) “Low-cost adsorbents: Growing approach to wastewater treatment —a review.” Crit Rev Environ Sci Tech., 39(10), 783 –842.
Gupta, V K., and Suhas., Z (2009) “Application of low-cost adsorbents for dye removal —a review.” J Environ Manage., 90(8), 2313 –2342.
Higashi, N., Ikehata, A., Kariyama, N., and Ozaki, Y (2008) “Potential
of far-ultraviolet absorption spectroscopy as a highly sensitive method for aqueous solutions Part II: Monitoring the quality of semiconductor wafer cleaning solutions using attenuated total reflection ” Appl.
Jacobson, K., Ishihara, A., and Inman, R (1987) “Lateral diffusion of proteins in membranes ” Ann Rev Physiol., 49(1), 163 –175.
Kuo, T L., Lin, D L., Liu, R H., Moriya, F., and Hashimoto, Y (2001).
“Spectra interference between diquat and paraquat by second derivative spectrophotometry ” Forensic Sci Int., 121(1 –2), 134–139.
Lide, D R (1996) CRC handbook of chemistry and physics, Special Student Ed., CRC, Boca Raton, Florida.
MassLynx 4.1 [Computer software] Milford, Massachusetts, Waters MATLAB R2010b (2010) [Computer software] Natick, Massachusetts, MathWorks.
McKay, G (1998) “Application of surface diffusion model to the adsorp-tion of dyes on bagasse pith ” Adsorption, 4(3 –4), 361–372.
Meshko, V., Markovska, L., and Mincheva, M (1999) “Two resistance mass transfer model for the adsorption of basic dyes from aqueous sol-ution on natural zeolite ” Bull Chem Technol Maced., 18(2), 161–169.