Abstract

Adsorption was found to be an acceptable treatment option to remove geosmin (GSM) and 2-methylisoborneol (2-MIB). It is meaningful to investigate the adsorption capacity of granular activated carbon (GAC) for the two algal odorants in water, and the influences of natural organic material (NOM) and particle size. The adsorption process was studied with the four isotherm models (Langmuir, Freundlich, Temkin, and modified Freundlich), four kinetic models (pseudo first-order, pseudo second-order, Elovich, and intra-particle), and thermodynamics. The results showed that the adsorption of both compounds could be best described by the modified Freundlich isotherm and pseudo second-order model, and the obtained thermodynamic parameters (changes in heat of adsorption, entropy, and Gibbs free energy) revealed that the adsorption was endothermic and spontaneous. Downsizing the particle size of GAC was effective for improving the adsorption capacity and rate. The concentrations of the two odorants could be reduced from 500 ng L−1 to less than 10 ng L−1 with the presence of NOM (<20 mg L−1 total organic carbon, TOC).

INTRODUCTION

Geosmin (GSM) and 2-methylisoborneol (2-MIB) are two main nuisance odorants in drinking water and are often co-existing in surface water (Chen et al. 2010; Liu et al. 2016). Although non-toxic they can be detected by consumers at levels as low as 10 ng L−1 (Graham et al. 2000; Matsui et al. 2009a). Conventional water treatment processes, such as coagulation and filtration, could not remove these odorants efficiently (Watson et al. 2007; Zamyadi et al. 2015), whereas advanced oxidation processes and adsorption were found to be effective and acceptable treatment options. Because activated carbon (AC) has advantageous characteristics (extended surface area, high adsorption capacity, microporous structure, and special surface reactivity) (Matsui et al. 2013), it has been applied in the adsorption to remove the odor compounds, and the adsorption treatment has become the simplest and the most widely applied method (Mall et al. 1996). It has been cited as one of the best available treatment options by the USEPA (EPA 2019). The two main types of AC used in water treatment are granular activated carbon (GAC) and powdered activated carbon (PAC). The adsorptive capacity of GAC makes it ideal for removing a variety of contaminants from water. Therefore, locating a GAC treatment unit is a common option in water treatment plants (Nekoo & Fatemi 2013; Kose-Mutlu et al. 2017). However, there is still a lack of necessary information for the application of GAC to remove GSM and 2-MIB.

Adsorption is a complex process of transferring specific contaminants from a fluid phase to a solid phase. In order to successfully simulate and optimize the adsorption of a chosen AC, adsorption equilibrium isotherm, kinetics, and thermodynamics have to be studied. The results of the adsorption equilibrium isotherm will be used to evaluate the affinity or capacity of an adsorbent and select a suitable adsorbent and adsorbent dose (Yang & Al-Duri 2005). The study of adsorption kinetics will be used to obtain valuable information on the operation duration, the reaction pathways, and the mechanism of adsorption reactions (Alzaydien & Manasreh 2009). Combining the information from adsorption isotherm and kinetics studies also could enable the estimation of the economic feasibility of an adsorbent's commercial application for specific contaminants, whereas adsorption thermodynamics will provide the basis for exploring the adsorption mechanism. Also, the particle size of GAC has been considered as an important factor affecting the adsorption kinetics and the contact time required for equilibration (Matsui et al. 2015). A previous paper revealed that adsorption capacity of AC substantially increased when carbon particle diameter was decreased for GSM and 2-MIB (Matsui et al. 2015). But the smaller particle may increase the difficulty of filtration, and it is not yet known what particle size of AC would be most suitable. Therefore, choosing the appropriate size of the GAC is important for water treatment. In water treatment processes, the effect of natural organic material (NOM) should be investigated since the concentration of NOM in natural water usually is much higher than the concentrations of the target compounds, which would lead to a large proportion of the adsorption volume within the AC not being available for the GSM and 2-MIB removals (Zoschke et al. 2011; Matsui et al. 2012).

The objectives of this paper were to investigate the adsorption capacity of GAC for 2-MIB and GSM based on the thermodynamics, adsorption isotherm, and kinetics, and to evaluate the impacts of key factors, including NOM and particle size, on the adsorption. The data acquired from this study should provide a scientific basis for controlling the pollution of 2-MIB and GSM in water by GAC adsorption.

MATERIALS AND Methods

Chemicals

GSM, 2-MIB, and the internal standard 2,3,6-trichloroanisole (2,3,6-TCA) were purchased from Dr. Ehrenstorfer Gmbh (Augsburg, Germany). The stock solutions of 2-MIB, GSM, and 2,3,6-TCA (100 mg mL−1) were prepared in methanol (HPLC grade) obtained from Sigma-Aldrich® (Saint Quentin Fallaviers, France). All the standard solutions were stored at −20 °C in darkness. All the other solutions were prepared with deionized water obtained from a Milli-Q ultrapure water purification system (Millipore, USA). In order to get the humic acid stock solution, three grams of humic acid (Aladdin®, Shanghai, China) was dissolved in 0.1 M NaOH, and was mixed with a magnetic stirrer for 24 h. The suspension was filtered by a 0.45 μm membrane filter. The pH of the filtered solution was adjusted to 7.5 by adding 0.1 M NaOH. The solution was diluted to 1,000 mL in a measuring flask and was stored in the dark. Total organic carbon (TOC) was measured by a TOC analyzer (Shimadzu®, TOC-VCPN, Japan), and the TOC of the stock humic acid solution was found to be 1,120 mg L−1. This humic acid solution at experiment concentration (5, 10, or 20 mg L−1) was obtained by diluting the stock solution before use.

Coconut shell-based GAC was purchased from Aladdin (Shanghai, China). GAC was ground and sieved to specific sizes, and GAC samples with different size were obtained, including <61, 61–74, 74–106, 106–150, 150–250, and 250–850 μm. All the GAC samples were ultrasonic rinsed with ultrapure water for 30 min, dried in an oven at 110 °C for 12 h, then cooled down to room temperature, and finally were stored in a desiccator prior to use.

Adsorption experiments

For the kinetic experiments, 40 mg GAC was introduced into 2 L solutions in a 3 L volumetric flask with GSM and 2-MIB. The initial concentration of each compound was 500 ng L−1. Then the flask was placed on an orbital shaker with a rotation speed of 100 rpm at 25 °C. The samples were withdrawn from the flask at predetermined intervals. The concentrations of GSM and 2-MIB were determined using solid-phase microextraction coupled with gas chromatography–mass spectrometry (GC-MS).

For the adsorption isotherm, a 500 mL solution containing GSM and 2-MIB, each at concentration of 500 ng L−1, was added into a 600 mL volumetric flask. A specified amount (0.2, 0.3, 0.5, 0.8, 1, 2, 5, and 10 mg L−1) of GAC with particle size less than 61 μm was added, and then the flask was placed on an orbital shaker at 100 rpm for 24 h at control temperature. The flask was sealed with a glass stopper to avoid volatilization, and a blank (without GAC addition) was included with each experiment to evaluate the losses other than that through adsorption. The sample was filtered with a 0.45 μm membrane to remove the solid adsorbents, and the filter had been cleaned in advance by ultrapure water before use. The temperatures in the experiments were set at 293, 303, 313, and 323 K, respectively.

Analytical method

The concentrations of GSM and 2-MIB were analyzed using 65 μm PDMS/DVB solid-phase microextraction (SPME) (Supelco, USA) coupled with GC-MS (Agilent 7890A/5975C, USA) equipped with an HP-5MS column (30 m × 0.25 mm × 0.25 μm, Agilent, USA). High purity helium (99.999%, Rising Corp., China) was used as the carrier gas and kept at 1 mL min−1 constant flow. The injection port was operated at splitless mode with temperature controlled at 250 °C, and the oven temperature program was as follows: initially stayed at 40 °C for 3 min, heated at a rate of 15 °C min−1 to 280 °C, and finally kept at 280 °C for 1 min. The MS ion source temperature was set at 230 °C. For the selected ion monitoring mode, m/z of 167, 195 and 210 for 2,3,6-TCA; m/z of 95 and 108 for 2-MIB; and m/z of 112 and 125 for GSM were monitored, respectively (McCallum et al. 1998).

Data analysis

To get an adsorption isotherm plot, a set of adsorption tests with varied GAC dose D (mg L−1) under a fixed initial solute concentration C0 (mg L−1) should be conducted until the amount adsorbed per unit mass of carbon Qt (ng mg−1) reaches equilibrium; the maximum amount adsorbed per unit mass of carbon Qe (ng mg−1) is measured, and the concentration of adsorbate in solution correspondingly reduces from Ct (ng L−1) to Ce (ng L−1). The relationship can be expressed as follows:  
formula
(1)
 
formula
(2)
where V (L) is the solution volume and m (mg) is the mass of adsorbent. Generally, the adsorbent that has a higher Qe value at a specified equilibrium concentration will be preferred for a given application (Ho et al. 2002).
The adsorption isotherm models and kinetic models were established using Origin 8.0. The models were evaluated by the coefficient of determination (R2) and chi-square (χ2). The chi-square is given as follows:  
formula
(3)
where Qe,ex is the measured Qe, and Qe,cal is the Qe calculated with the models. All data were determined using the Origin 8.0.

RESULT AND DISCUSSION

Adsorption kinetics

Determining the adsorption rate and identifying the potential rate-controlling step are primary for practical operation. Thus, different kinds of kinetic models, namely pseudo first-order model, pseudo second-order model, intra-particle model, and Elovich model, were used to describe the adsorption kinetic behavior.

Figure 1 shows that the adsorbed amount of GSM and 2-MIB increased dramatically in the first 10 min, and then increased slowly from 10 min to 20 min. The change from 20 min to 30 min and even to 2 h was slight. The intra-particle kinetic model holds that the adsorption rate is controlled by diffusion, and if the line passes through the origin, the process rate is controlled only by internal diffusion (Nassar 1999). In this study, the intra-particle kinetic model fitted the measured adsorption data poorly for 2-MIB and GSM with the low R2 (0.880 and 0.814) and the high χ2 (0.767 and 13.240), which indicated that internal diffusion is not the rate-controlling step in this adsorption process.

Figure 1

Kinetic models for the adsorptions of GSM (a) and 2-MIB (b) onto granular activated carbon (C0 = 500 ng L−1, m = 20 mg L−1 for GAC, T = 25 ± 1 °C, GAC particle size <61 μm).

Figure 1

Kinetic models for the adsorptions of GSM (a) and 2-MIB (b) onto granular activated carbon (C0 = 500 ng L−1, m = 20 mg L−1 for GAC, T = 25 ± 1 °C, GAC particle size <61 μm).

The pseudo first-order model is the empirical kinetic equation for one-site occupancy adsorption which simulates a rapid adsorption due to the absence of sorbate–sorbate interaction. The pseudo second-order model involves all the potential adsorption procedures, such as external film diffusion, surface adsorption, and intra-particle diffusion (Ho & McKay 1999). As shown in Table 1, the pseudo first-order and second-order kinetic models well fitted the measured adsorption data with the high R2 (0.988 and 1.000 for GSM, 0.997 and 0.999 for 2-MIB). Qe calculated with the pseudo second-order model (27.38 ng mg−1 and 29.81 ng mg−1 for 2-MIB and GSM respectively) were close to the measured data (30.25 ng mg−1 and 30.15 ng mg−1 for 2-MIB and GSM respectively), which indicated the sorbate–sorbate interaction was weak in the adsorption process and adsorption kinetics was mainly governed by the one-site occupancy surface adsorption. GSM had a higher adsorption rate than 2-MIB (27.7 min−1 vs 24.9 min−1), and GAC possessed a larger adsorption capacity for GSM, which is probably ascribed to the molecular size since the GSM molecule is bigger than the 2-MIB one.

Table 1

Kinetic parameters for adsorption of GSM and 2-MIB on GAC

Kinetic modelEquationParameterGSM2-MIB
Pseudo first-order  Qe 27.9 24.0 
K1 0.252 0.236 
R2 0.988 0.997 
χ2 1.82 0.314 
Pseudo second-order  Qe 29.8 27.4 
K2 0.019 0.012 
R2 1.00 0.999 
χ2 0.002 
Intra-particle  a 5.34 3.67 
KI 5.09 4.59 
R2 0.814 0.880 
χ2 0.767 13.2 
Elovich  b 17.3 10.9 
Ke 3.32 4.287 
R2 0.994 0.996 
χ2 0.743 0.356 
Kinetic modelEquationParameterGSM2-MIB
Pseudo first-order  Qe 27.9 24.0 
K1 0.252 0.236 
R2 0.988 0.997 
χ2 1.82 0.314 
Pseudo second-order  Qe 29.8 27.4 
K2 0.019 0.012 
R2 1.00 0.999 
χ2 0.002 
Intra-particle  a 5.34 3.67 
KI 5.09 4.59 
R2 0.814 0.880 
χ2 0.767 13.2 
Elovich  b 17.3 10.9 
Ke 3.32 4.287 
R2 0.994 0.996 
χ2 0.743 0.356 

K1, first-order adsorption rate constant; K2, second-order adsorption rate constant; t, contact time; a and KI, intra-particle diffusion rate constants; b and Ke, Elovich adsorption constants.

The Elovich kinetic model is an empirical model, which considers a series of reaction mechanisms, such as the diffusion of the solute in the liquid or interface and the activation and deactivation of the surface. It is very suitable for systems with heterogeneous adsorbing surfaces and significant change in the activation energy during the reaction, such as the process on the interface of the soil and the sediment (Ahmad et al. 2014). The Elovich kinetic model also well fitted the adsorption data for GSM and 2-MIB with R2 higher than 0.994 and χ2 lower than 0.743, indicating that the adsorption is related to activation energy. So, it is necessary to investigate the adsorption thermodynamics.

Adsorption isotherms

Adsorption isotherm tests were conducted for GSM and 2-MIB. Measured and modeled single-solute adsorption isotherm data for GSM and 2-MIB are shown in Figure 2. The used models included Freundlich, Langmuir, Temkin, and modified Freundlich equations. The corresponding isotherm parameters obtained by nonlinear regression analysis are summarized in Table 2. When contact time was 24 h, adsorptions of GSM and 2-MIB were best described by the modified Freundlich isotherm, Freundlich isotherm model and Langmuir isotherm model.

Table 2

Constants of isotherm models at 24 h equilibrium time

Isotherm modelsEquationsParametersGSM2-MIB
500 (ng L−1)500 (ng L−1)
Langmuir  Q0 (ng mg−1833 1111 
KL (L mg−10.120 0.0053 
R2 0.980 0.955 
χ2 0.629 0.199 
Freundlich  KF (ng(1−1/n) L1/n g−1116 11.40 
1/n 0.487 0.574 
R2 0.988 0.957 
χ2 0.087 0.016 
Modified Freundlich   (ng1−1/n mg1/n−1130 60.7 
1/n 0.439 0.737 
R2 0.990 0.959 
χ2 0.004 0.007 
Temkin  KT 0.005 0.182 
BT 620 111 
R2 0.737 0.861 
χ2 1.07 0.0432 
Isotherm modelsEquationsParametersGSM2-MIB
500 (ng L−1)500 (ng L−1)
Langmuir  Q0 (ng mg−1833 1111 
KL (L mg−10.120 0.0053 
R2 0.980 0.955 
χ2 0.629 0.199 
Freundlich  KF (ng(1−1/n) L1/n g−1116 11.40 
1/n 0.487 0.574 
R2 0.988 0.957 
χ2 0.087 0.016 
Modified Freundlich   (ng1−1/n mg1/n−1130 60.7 
1/n 0.439 0.737 
R2 0.990 0.959 
χ2 0.004 0.007 
Temkin  KT 0.005 0.182 
BT 620 111 
R2 0.737 0.861 
χ2 1.07 0.0432 

Q0 and KL, Langmuir constants related to the adsorption capacity and intensity of adsorption, respectively; KF and 1/n, Freundlich constants related to the adsorption capacity and energy of adsorption, respectively; , modified Freundlich constant related to the adsorption capacity; KT and BT, Temkin constants.

Figure 2

Adsorption isotherm data for GSM (a) and 2-MIB (b) on GAC (C0 = 500 ng L−1, 24 h contact time, T = 25 ± 1 °C, GAC particle size <61 μm).

Figure 2

Adsorption isotherm data for GSM (a) and 2-MIB (b) on GAC (C0 = 500 ng L−1, 24 h contact time, T = 25 ± 1 °C, GAC particle size <61 μm).

The Langmuir adsorption model is the theoretical formula of adsorption based on homogeneous surfaces where a strong specific interaction between adsorbate and adsorbent only occurs on monolayer coverage; thus it fails to account for the surface roughness of the adsorbent (Song et al. 2013). The adsorption data for GSM and 2-MIB were fitted with Langmuir with the R2 being 0.980 for GSM and 0.955 for 2-MIB and the χ2 being 0.629 for GSM and 0.159 for 2-MIB, indicating that the adsorption mainly is monolayer coverage. The KL value of the Langmuir adsorption model for GSM was higher than that of 2-MIB, suggesting that the adsorptive capacity of GAC for GSM is better than that for 2-MIB.

The freundlich isotherm model is an empirical equation and particularly appropriate for hydrophobic adsorbates (Sotelo et al. 2002) in water. It is often appropriate to use the Freundlich model instead of the Langmuir adsorption model (Jaroniec 1975). The validity of the Freundlich model (R2 being 0.988 and 0.957 for GSM and 2-MIB, respectively, and χ2 being 0.087 and 0.016 for GSM and 2-MIB, respectively) and Langmuir model indicated that the adsorptions were controlled by one-site occupancy surface adsorption with monolayer coverage, and the effects of other factors were negligible. Similar findings on GAC adsorption for other taste and odor compounds were reported by Zhang et al. (2011) and An et al. (2012). The KF parameter can roughly indicate the adsorption capacity and the parameter 1/n gives an indicator of the favorability of adsorption. The value of 1/n usually ranges from 0.1 to 1, and the smaller value of 1/n implies the stronger interactions between the adsorbent and adsorbates (Tan et al. 2008). The values of 1/n for GSM and 2-MIB are comparable, but GSM had a higher KF than 2-MIB, implying that GAC had a slightly higher adsorption capacity for GSM than for 2-MIB.

The modified Freundlich model is a special case of dosage-dependent heterogeneous surface adsorption with non-negligible intermolecular reactions, which is an appropriate explanation of adsorption behavior in diluted solution. The modified Freundlich model well fitted the measured adsorption data with R2 being 0.990 for GSM and 0. 959 for 2-MIB, and χ2 being 0.004 for GSM and 0.007 for 2-MIB, respectively. GSM had a higher than 2-MIB, representing that GAC had a slightly higher adsorption capacity for GSM than for 2-MIB. Smaller 1/n values were obtained for GSM than for 2-MIB, implying that GAC had a higher affinity with GSM than with 2-MIB, and the adsorption for 2-MIB was more readily affected by equilibrium concentration. The result agreed well with that obtained with the Freundlich model.

The Temkin isotherm is another two-parameter isotherm model established on the theoretical hypothesis that the decline of the heat is linear with the amount of adsorption, due to adsorbate with adsorbent interactions (Tan et al. 2007). The bad fitness indicates that the heat of adsorption is independent of the adsorption amount, and the adsorption may not be chemical adsorption.

Adsorption thermodynamics

Gibbs free-energy change (ΔG0) of the adsorption processes can be determined by the classical Van 't Hoff equation:  
formula
(4)
where ΔG0 is the free-energy change (kJ mol−1), R is the universal gas constant (8.314 × 10−3 kJ mol−1 K−1), T is the absolute temperature (K), and Kad is the adsorption equilibrium constant determined above. Also, ΔG0 can be calculated from Equation (5).  
formula
(5)
where ΔS0 is the change in entropy (kJ mol−1 K−1), and ΔH0 is the change in heat of adsorption (kJ mol−1) at a constant temperature. Combining Equations (4) and (5) leads to Equation (6):  
formula
(6)
In Equation (6), ΔH0 can be determined from the slope of the linear Van 't Hoff plot, i.e. versus (1/T), using Equation (7):  
formula
(7)

This ΔH0 corresponds to the isosteric heat of adsorption (ΔHst,0) with zero surface coverage (i.e. Qe = 0). Kad at Qe = 0 was obtained from the intercept of the ln(Qe/Ce) versus Qe plot at different temperatures (293, 303, 313, and 323 K). Figure 3 shows the Van 't Hoff plot, from which the values of ΔH0, ΔS0, and ΔG0 were determined (Table 3).

Table 3

Thermodynamic parameters for the adsorption of GSM and 2-MIB onto GACs

Organic mattersΔH0 (kJ mol−1)ΔS0 (kJ mol−1 K−1)ΔG0 (kJ mol−1)
293 K303 K313 K323 K
GSM 10.2 0.00671 −9.44 −10.1 −10.8 −11.4 
2-MIB 19.0 0.00794 −4.21 −5.00 −5.79 −6.58 
Organic mattersΔH0 (kJ mol−1)ΔS0 (kJ mol−1 K−1)ΔG0 (kJ mol−1)
293 K303 K313 K323 K
GSM 10.2 0.00671 −9.44 −10.1 −10.8 −11.4 
2-MIB 19.0 0.00794 −4.21 −5.00 −5.79 −6.58 
Figure 3

Van 't Hoff plot for the determination of ΔH0, ΔS0, ΔG0 (C0 = 500 ng L−1 for 2-MIB and GSM, respectively, m = 20 mg L−1 for GAC).

Figure 3

Van 't Hoff plot for the determination of ΔH0, ΔS0, ΔG0 (C0 = 500 ng L−1 for 2-MIB and GSM, respectively, m = 20 mg L−1 for GAC).

It is observed from Table 4 that ΔH0 and ΔS0 are positive, and ΔG0 is negative in both adsorption processes. The positive ΔH0 value confirms the endothermic nature of the overall sorption process, and the positive value of ΔS0 suggests increased randomness at the solid/solution interface with some structural changes in GSM and 2-MIB affinity to GAC. The negative ΔG0 indicates the feasibility and spontaneity of the adsorption process.

Table 4

Adsorption kinetic parameters (m = 20 mg L−1 for GAC, T = 25 ± 1 °C)

Organic mattersTOC
mg L−1

ng mg−1
Pseudo first-order
Pseudo second-order

ng mg−1

R2χ2
ng mg−1

g ng−1 min−1
R2χ2
GSM 27.7 27.4 0.310 0.999 0.0633 29.8 0.0194 0.999 0.166 
26.4 26.0 0.34 0.999 0.124 27.3 0.0246 0.999 0.102 
10 27.17 26.6 0.325 0.995 0.364 25.6 0.0295 0.991 0.066 
20 26.04 25.6 0.231 0.997 0.209 26.9 0.0177 0.989 0.758 
2-MIB 25.6 24.9 0.215 0.994 0.436 26.2 0.0162 0.997 0.199 
25.4 24.5 0.209 0.990 0.682 25.9 0.0154 0999 0.0815 
10 25.6 24.5 0.164 0.987 0.865 26.18 0.0109 0.998 0.106 
20 25.3 24.3 0.143 0.986 0.931 25.99 0.00926 0.999 0.0844 
Organic mattersTOC
mg L−1

ng mg−1
Pseudo first-order
Pseudo second-order

ng mg−1

R2χ2
ng mg−1

g ng−1 min−1
R2χ2
GSM 27.7 27.4 0.310 0.999 0.0633 29.8 0.0194 0.999 0.166 
26.4 26.0 0.34 0.999 0.124 27.3 0.0246 0.999 0.102 
10 27.17 26.6 0.325 0.995 0.364 25.6 0.0295 0.991 0.066 
20 26.04 25.6 0.231 0.997 0.209 26.9 0.0177 0.989 0.758 
2-MIB 25.6 24.9 0.215 0.994 0.436 26.2 0.0162 0.997 0.199 
25.4 24.5 0.209 0.990 0.682 25.9 0.0154 0999 0.0815 
10 25.6 24.5 0.164 0.987 0.865 26.18 0.0109 0.998 0.106 
20 25.3 24.3 0.143 0.986 0.931 25.99 0.00926 0.999 0.0844 

Effect of GAC particle size

As shown in Figure 4, the smaller the particle is, the shorter the equilibration time needed, but the odorants could be removed almost completely if the contact time is sufficiently long (such as 24 h) regardless of the particle size of the GAC. Reduction of particle size from 250–850 μm to 150–250 μm or from 150–250 μm to 74–106 μm was effective in shortening the equilibrium times to 20 min, meaning that the removal rates were improved. But the removal rates have no obvious difference in 6 h when particle size was reduced from 61–74 μm to <61 μm.

Figure 4

Effect of GAC particle size on adsorption for GSM (a) and 2-MIB (b) (C0 = 500 ng L−1 for 2-MIB (a) and GSM (b), respectively, m = 20 mg L−1 for GAC, T = 25 ± 1 °C).

Figure 4

Effect of GAC particle size on adsorption for GSM (a) and 2-MIB (b) (C0 = 500 ng L−1 for 2-MIB (a) and GSM (b), respectively, m = 20 mg L−1 for GAC, T = 25 ± 1 °C).

A conventional drinking water treatment plant consists of coagulation, sedimentation, filtration, and disinfection units (Angreni 2013). AC can be added in the source water, or the process waters after the coagulation or before the filtration. Adding AC to the source water, the contact time between AC and adsorbates would be sufficient. Thus, the various sizes of AC in this study could all be applied to this unit. After coagulation, the coagulation and sedimentation time was longer than 2 h (China 2006); thus the AC with the particle size smaller than 106 μm might be suitable. However, the flocculents added in the water tend to adsorb on the AC and cause an encapsulation effect (Cook et al. 2001). And the encapsulation greatly depends on the particle size of the AC (Matsui et al. 2009b). Thus the particle size should be chosen sophisticatedly. Before the filtration, the contact time of filtration is less than 30 min (China 2006), while the smaller size AC easily penetrates the filter layer and increases effluent turbidity (Matsui et al. 2009b). Therefore, choosing a suitable AC size is important. In general, the optimal particle size of GAC needs to be combined with specific water treatment processes.

Effect of NOM

NOM is a complex mixture of organic compounds, primarily composed of fulvic acids, humic acids, hydrophilic acids, and carbohydrates. Humic acid is the main organic substance in soil and water at forestry protected watersheds, which can account for 50%–80% of the total organic matter (Davies & Ghabbour 2004). Several previous studies have shown that the presence of NOM could cause a significant reduction of the target organic pollutants adsorbed by the AC (Moore et al. 2010; Zhang et al. 2011; Zoschke et al. 2011). In this study, the effect of humic acid in the background matrix on the adsorption process was considered.

Figure 5 shows that the removal ratios for both GSM and 2-MIB with the initial concentration of 500 ng L−1 could reach above 98% after 2 h in these solutions with different concentrations of TOC, and the residue concentrations could be decreased to 10 ng L−1 for 2-MIB and 4 ng L−1 for GSM.

Figure 5

Effect of initial TOC of the aqueous solution on the removal of GSM and 2-MIB by GAC (C0 = 500 ng L−1 for 2-MIB (a) and GSM (b), respectively, m = 20 mg L−1 for GAC, T = 25 ± 1 °C).

Figure 5

Effect of initial TOC of the aqueous solution on the removal of GSM and 2-MIB by GAC (C0 = 500 ng L−1 for 2-MIB (a) and GSM (b), respectively, m = 20 mg L−1 for GAC, T = 25 ± 1 °C).

The adsorption kinetics for GSM and 2-MIB still exhibited pseudo first-order or pseudo second-order kinetic behaviors (R2 = 0.98–0.99), as shown in Table 4. The values of K1 and K2 for GSM adsorption were comparable for TOC concentrations of 0, 5, and 10 mg L−1 in water, while they showed a slight decrease at 20 mg L−1 TOC. As for 2-MIB, the values of K1 and K2 adsorption showed a slight decrease with TOC increasing from 0 to 20 mg L−1. These results indicated that the inhibitory action of NOM was not significant with the TOC increasing in the present experiments, which was in accordance with the findings of other studies (Hepplewhite et al. 2004; Zoschke et al. 2011). It was generally considered that the larger NOM molecules could block pores of AC, or the smaller NOM molecules directly competed for adsorption sites with those low molecular weight compounds such as 2-MIB (Kilduff et al. 1998). The competitive mechanism was important for the effect of NOM on microcontaminant adsorption (Newcombe et al. 2002). However, in the present study, the insignificant impact of NOM suggested that GSM and 2-MIB are more competitive than NOM for adsorption.

CONCLUSIONS

The adsorption kinetics and isotherm of GSM and 2-MIB could be best described by the pseudo second-order model and modified Freundlich isotherm. Other kinetic models also provide valuable pieces of information on the reaction pathways and the mechanism of adsorption reactions. The calculated thermodynamic parameters (ΔH0, ΔS0, and ΔG0) revealed that the adsorption of the two compounds was endothermic, there was increased randomness at the solid/solution interface and the adsorption process was spontaneous.

The effect of particle size of GAC for removal of GSM and 2-MIB was studied and the results showed that downsizing the particles of GAC was effective in shortening the necessary contact times both for GSM and 2-MIB. When the particle size of GAC (20 mg L−1) was less than 61 μm, the equilibrium could be reached within 2 h, and the concentrations of the two odorants would be reduced from 500 ng L−1 to 10 ng L−1 for 2-MIB and to 4 ng L−1 for GSM with the presence of NOM (<20 mg L−1 TOC).

ACKNOWLEDGEMENTS

This work was supported by the Special Fund for Agro-scientific Research in the Public Interest of China (No. 201503108) and Science & Technology Project of Hunan Province (No. 2017WK2091).

REFERENCES

REFERENCES
Alzaydien
A. S.
,
Manasreh
W.
2009
Equilibrium, kinetic and thermodynamic studies on the adsorption of phenol onto activated phosphate rock
.
International Journal of Physical Sciences
4
(
4
),
172
181
.
An
N.
,
Xie
H. H.
,
Gao
N. Y.
,
Deng
Y.
,
Chu
W. H.
,
Jiang
J.
2012
Adsorption of two taste and odor compounds IPMP and IBMP by granular activated carbon in water
.
Clean – Soil Air Water
40
(
12
),
1349
1356
.
Angreni
E.
2009
Review on optimization of conventional drinking water treatment plant
.
World Applied Sciences Journal
7
(
9
),
1144
1151
.
Davies
G.
,
Ghabbour
E.
2004
Humic Substances: Nature's Most Versatile Materials
.
Taylor & Francis
,
New York, USA
.
EPA
2019
Drinking Water Treatability Database
. .
Graham
M. R.
,
Summers
R. S.
,
Simpson
M. R.
,
MacLeod
B. W.
2000
Modeling equilibrium adsorption of 2-methylisoborneol and geosmin in natural waters
.
Water Research
34
(
8
),
2291
2300
.
Hepplewhite
C.
,
Newcombe
G.
,
Knappe
D. R.
2004
NOM and MIB, who wins in the competition for activated carbon adsorption sites?
Water Science and Technology
49
(
9
),
257
265
.
Ho
Y. S.
,
McKay
G.
1999
Pseudo-second order model for sorption processes
.
Process Biochemistry
34
(
5
),
451
465
.
Ho
L.
,
Newcombe
G.
,
Croue
J. P.
2002
Influence of the character of NOM on the ozonation of MIB and geosmin
.
Water Research
36
(
3
),
511
518
.
Kose-Mutlu
B.
,
Ersahin
M. E.
,
Ozgun
H.
,
Kaya
R.
,
Kinaci
C.
,
Koyuncu
I.
2017
Influence of powdered and granular activated carbon system as a pre-treatment alternative for membrane filtration of produced water
.
Journal of Chemical Technology & Biotechnology
92
(
2
),
283
291
.
Mall
I. D.
,
Upadhyay
S. N.
,
Sharma
Y. C.
1996
A review on economical treatment of wastewaters and effluents by adsorption
.
International Journal of Environmental Studies
51
(
2
),
77
124
.
Matsui
Y.
,
Ando
N.
,
Sasaki
H.
,
Matsushita
T.
,
Ohno
K.
2009a
Branched pore kinetic model analysis of geosmin adsorption on super-powdered activated carbon
.
Water Research
43
(
12
),
3095
3103
.
Matsui
Y.
,
Hasegawa
H.
,
Ohno
K.
,
Matsushita
T.
,
Mima
S.
,
Kawase
Y.
,
Aizawa
T.
2009b
Effects of super-powdered activated carbon pretreatment on coagulation and trans-membrane pressure buildup during microfiltration
.
Water Research
43
(
20
),
5160
5170
.
National Standard of the People's Republic of China
2006
Code for Design of Outdoor Water Supply Engineering
.
GB50013-2006
.
Nekoo
S. H.
,
Fatemi
S.
2013
Experimental study and adsorption modeling of COD reduction by activated carbon for wastewater treatment of oil refinery
.
Iranian Journal of Chemistry & Chemical Engineering
32
(
3
),
81
89
.
Song
X.
,
Zhang
Y.
,
Yan
C.
,
Jiang
W.
,
Chang
C.
2013
The Langmuir monolayer adsorption model of organic matter into effective pores in activated carbon
.
Journal of Colloid and Interface Science
389
(
1
),
213
219
.
Sotelo
J. L.
,
Ovejero
G.
,
Delgado
J. A.
,
Martínez
I.
2002
Comparison of adsorption equilibrium and kinetics of four chlorinated organics from water onto GAC
.
Water Research
36
(
3
),
599
608
.
Tan
I. A. W.
,
Hameed
B. H.
,
Ahmad
A. L.
2007
Equilibrium and kinetic studies on basic dye adsorption by oil palm fibre activated carbon
.
Chemical Engineering Journal
127
(
1
),
111
119
.
Watson
S. B.
,
Chariton
M.
,
Rao
Y. R.
,
Howell
T.
,
Ridal
J.
,
Brownlee
B.
,
Marvin
C.
,
Millard
S.
2007
Off flavours in large waterbodies: physics, chemistry and biology in synchrony
.
Water Science and Technology
55
(
5
),
1
8
.
Yang
X. Y.
,
Al-Duri
B.
2005
Kinetic modeling of liquid-phase adsorption of reactive dyes on activated carbon
.
Journal of Colloid and Interface Science
287
(
1
),
25
34
.
Zamyadi
A.
,
Henderson
R.
,
Stuetz
R.
,
Hofmann
R.
,
Ho
L.
,
Newcombe
G.
2015
Fate of geosmin and 2-methylisoborneol in full-scale water treatment plants
.
Water Research
83
,
171
183
.
Zhang
K. J.
,
Gao
N. Y.
,
Deng
Y.
,
Shui
M. H.
,
Tang
Y. L.
2011
Granular activated carbon (GAC) adsorption of two algal odorants, dimethyl trisulfide and β-cyclocitral
.
Desalination
266
(
1
),
231
237
.