water research 43 (2009) 3787–3796
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Removal of emerging contaminants of concern by alternative adsorbents Alfred Rossnera, Shane A. Snyderb, Detlef R.U. Knappec,* a
Environmental Science Center EULA-Chile, University of Concepcio´n, P.O. Box 160-C, Concepcio´n, Chile Department of Research and Development, Southern Nevada Water Authority, 1350 Richard Bunker Avenue, Henderson, NV 89015, USA c Department of Civil, Construction, and Environmental Engineering, Campus Box 7908, North Carolina State University, Raleigh, NC 27695-7908, USA b
The effective removal of emerging contaminants of concern (ECCs) such as endocrine-
Received 3 March 2009
disrupting chemicals, pharmaceutically active compounds, personal care products, and
Received in revised form
flame retardants is a desirable water treatment goal. In this study, one activated carbon,
30 May 2009
one carbonaceous resin, and two high-silica zeolites were studied to evaluate their effec-
Accepted 3 June 2009
tiveness for the removal of an ECC mixture from lake water. Adsorption isotherm exper-
Published online 11 June 2009
iments were performed with a mixture of 28 ECCs at environmentally relevant concentrations (w200–900 ng/L). Among the tested adsorbents, activated carbon was the
most effective, and activated carbon doses typically used for taste and odor control in
drinking water (<10 mg/L) were sufficient to achieve a 2-log removal for most of the tested
ECCs. The carbonaceous resin was less effective than the activated carbon because this
Zeolites Emerging contaminants
adsorbent had a smaller volume of pores in the size range required for the adsorption of ˚ ). For the removal of ECC mixture constituents, zeolites were less many ECCs (w6–9 A
Drinking water treatment
effective than the carbonaceous adsorbents. Because zeolites contain pores of uniform size and shape, a few of the tested ECCs with matching pore size/shape requirements were well removed, but the adsorptive removal of others was negligible, even at zeolite doses of 100 mg/L. The results of this study demonstrate that effective adsorbents for the removal of a broad spectrum of ECCs from water should exhibit heterogeneity in pore size and ˚ size range. shape and a large pore volume in the 6–9 A ª 2009 Elsevier Ltd. All rights reserved.
The removal of emerging contaminants of concern (ECCs) such as endocrine-disrupting chemicals (EDCs), pharmaceutically active compounds, personal care products, and flame retardants is an important consideration in the production of safe drinking water and the environmentally responsible release of wastewater. Prescription and non-prescription drugs are present at detectable levels in many US surface
waters (Kolpin et al., 2002), and conventional water treatment processes (coagulation/flocculation/sedimentation, filtration, and chlorination at levels required to achive 0.5-log Giardia inactivation) do not provide an effective barrier against many ECCs (e.g., Adams et al., 2002; Stackelberg et al., 2004). Therefore, treatment technologies that achieve effective ECC removal from water need to be identified. Treatment options that are typically considered for the removal of ECCs from drinking water include adsorption and
* Corresponding author. E-mail address: [email protected]
(D.R.U. Knappe). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.06.009
water research 43 (2009) 3787–3796
(advanced) oxidation processes. While conventional and advanced oxidation processes can be effective for the removal of ECCs (e.g. Adams et al., 2002; Westerhoff et al., 2005; Dodd et al., 2006; Rosenfeldt et al., 2007; Ikehata et al., 2008), these processes lead to the formation of oxidation intermediates that are mostly unknown at this point. In addition, unwanted oxidation byproducts, such as halogenated organic compounds or bromate, form in oxidation processes involving chlorine and ozone. Adsorption processes do not add undesirable byproducts to drinking water, but high adsorbent usage rates can be expected if activated carbon is employed to adsorb polar organic contaminants (e.g., Quinlivan et al., 2005). While several studies have evaluated the adsorption of individual ECCs on activated carbons in ultrapure water and in competition with natural organic matter (NOM) (Ternes et al., 2002; Adams et al., 2002; Yoon et al., 2003, 2005; Choi et al., 2006; Yu et al., 2008), only a few have evaluated the removal of a mixture of contaminants by activated carbon adsorption (Westerhoff et al., 2005; Snyder et al., 2007). No studies have investigated the effects of adsorbent pore size on ECC removal, and no information is available on the effectiveness of alternative adsorbents such as high-silica zeolites and carbonaceous resins for the removal of ECCs from drinking water. The objectives of this study, therefore, were to (1) compare the removal of individual constituents in an ECC mixture in lake water by four adsorbents (two high-silica zeolites, a carbonaceous resin and an activated carbon) with different pore sizes or pore size distributions and (2) to relate the effectiveness of individual adsorbents to their physical characteristics.
Materials and methods
Adsorption isotherm experiments were performed with Lake Mead water (LMW; Boulder City, NV) that was filtered through a 0.45-mm nylon membrane filter (Magna-R, MSI, Westboro, MA). The dissolved organic carbon (DOC) concentration of filtered LMW was 2.5 mg/L, the pH was 8.1, and the total alkalinity was approximately 140 mg/L as CaCO3. Four adsorbents were tested in this research: one coconutshell based GAC (CC-602 – redesignated as AquaCarb 1230C, Westates Carbon, Siemens, Roseville, MN), one carbonaceous resin (Ambersorb 563, Rohm and Haas, Philadelphia, PA), one mordenite zeolite (HSZ-690HOA, Tosoh USA, Grove City, OH), and one Y zeolite (HSZ-390HUA, Tosoh USA, Grove City, OH). To enhance adsorption rates in isotherm experiments, all pelletized zeolites, the GAC and the Ambersorb 563 resin were pulverized with a mortar and pestle until >95% by mass passed a 74-mm sieve (200 U.S. mesh). Upon sieving, the portion remaining on the sieve was recombined with the portion that passed through the sieve to prevent bias as a result of any physical and/or chemical differences between the two fractions. Pulverized adsorbents were dried at 105 C for one day and stored in a desiccator. Adsorption isotherm experiments were conducted with a mixture of 28 ECCs. Table 1 lists the tested adsorbates and selected physicochemical properties. ECCs were purchased in
neat form and dissolved in methanol to yield a stock solution with ECC concentrations of approximately 50 mg/L. All compounds were obtained from Sigma–Aldrich (St. Louis, MO) except atrazine and N,N-diethyl-m-toluamide (DEET), which were obtained from Accustandard (New Haven, CT), fluoxetine and iopromide, which were obtained from the United States Pharmacopeia (Rockville, MD), and hydrocodone, which was obtained from Cerilliant (Round Rock, TX).
BET surface areas of the four adsorbents were determined from N2 adsorption isotherm data collected at 77 K (Autosorb-1-MP, Quantachrome Corporation, Boynton Beach, FL). Prior to analysis, adsorbent samples were outgassed for 20 h at 423 K. Micropore volume, mesopore volume and pore size distribution of carbonaceous adsorbents were computed from N2 adsorption isotherm data using the Density Functional Theory (DFT) with the N2_carb1.gai kernel (PC software version 1.51, Quantachrome, Boynton Beach, FL). In addition, the mesopore volume was computed using the Barrett, Joyner, and Halenda (BJH) method because this method captures the entire mesopore range (20– ˚ ) while the DFT method was only able to calculate a mes500 A ˚. A opore volume for pores with widths in the range of 20–360 A more detailed description of the procedures used for adsorbent characterization is given elsewhere (Knappe et al., 2007).
In adsorption isotherm experiments, adsorbent doses between 0.3 and 10 mg/L were used for carbonaceous adsorbents while adsorbent doses in the range of 10–300 mg/L were used for highsilica zeolites. To yield an initial ECC concentration of approximately 500 ng/L per compound, 5 or 10 mL of the stock solution containing the ECC mixture shown in Table 1 was added with a constant rate syringe (CR-700-200, Hamilton Co., Reno, NV) to 500- or 1000-mL brown borosilicate glass bottles, respectively. Sodium azide (100 mg/L) was added to all bottles to inhibit aerobic biological activity during equilibration. For zeolites, pre-weighed aliquots were added directly to the bottles while carbonaceous adsorbents were added using slurries of known adsorbent concentration. Adsorbents were added after the addition of ECCs to simulate conditions that would be encountered when powdered adsorbents are added in drinking water treatment; i.e., NOM and all contaminants competed simultaneously for adsorption sites. A mixing time of 3 weeks in a rotary tumbler was used to reach adsorption equilibrium. Apart from bottles containing adsorbents, three types of adsorbent-free blanks were prepared in duplicate: (1) Type 1 blanks were prepared to determine background levels of ECCs and the effects of sample handling on ECC concentrations in LMW used in isotherm experiments. These blanks were prepared by filtering LMW through a 0.45-mm membrane filter and amending the filtrate with 100 mg/L sodium azide. This water was placed into isotherm bottles without spiking chemicals and sent immediately after preparation by overnight carrier for analysis. (2) Type 2 blanks were prepared to verify spiked ECC concentrations. These blanks were prepared by spiking the ECC
water research 43 (2009) 3787–3796
Table 1 – Adsorbate characteristics. Compound name
˚ )c Diameter (A
Acetaminophen Androstenedione Atrazine Caffeine Carbamazepine DEET Diazepam Diclofenac sodium Dilantin Estradiol Estriol Estrone Ethynylestradiol Fluoxetine Gemfibrozil Hydrocodone Ibuprofen Iopromide Meprobamate Naproxen Oxybenzone Pentoxifylline Progesterone Sulfamethoxazole Tri(2-chloroethyl) phosphate (TCEP) Testosterone Triclosan Trimethoprim
103-90-2 63-05-8 1912-24-9 58-08-2 298-46-4 134-62-3 439-14-5 15307-79-6 57-41-0 50-28-2 50-27-1 53-16-7 57-63-6 54910-89-3 25812-30-0 125-29-1 15687-27-1 73334-07-3 57-53-4 22204-53-1 131-57-7 6493-05-6 57-83-0 723-46-6 115-96-8
Analgesic Steroide Herbicidee Stimulant Analgesic Insect repellent Anti-anxiety Arthritis Anti-convulsant Steroide Steroide Steroide Birth controle Anti-depressant Anti-cholesterol Pain reliever Pain reliever X-ray contrast media Anti-anxiety Analgesic Sunscreen Blood viscosity control Steroide Antibiotic Fire retardant
151.2 286.4 215.7 194.2 236.3 191.3 284.8 318.1 252.3 272.4 288.4 270.4 296.4 309.3 250.3 299.4 206.3 791.1 218.3 230.3 228.2 278.3 314.5 253.3 285.5
9.4,a 9.5d [0/]f – 1.7,a 1.2d [þ/0]f <0d [þ/0] – 1.0d [þ/0] 3.4,a 3.6d [þ/0] 4.2,a 4.1d [0/] 8.3,a 8.4d [0/] 10.5d [0/] 10.5d [0/] 10.5d [0/] 10.5d [0/] 9.5d [þ/0] 4.5d [0/] 9.1d [þ/0] 4.9,a 4.5d [0/] 6.5d [0/]; 12.6d [/] 12.0d [0/] 4.2, a4.5d [0/] 10.2d [0/] 0.1d [þ/0] – 1.7,g 1.8d [þ/0]; 5.6,g 9.1d [0/] –
10 0.0578a 0.0347a 3.7 0.078 1 0.020 44 0.14 0.003 0.015 0.0051 0.002 0.49 35 0.57 120 280 4.7a 39 0.80 0.78 0.0088a 25 7a
0.33 2.90 2.63 0.13 2.67 1.96 2.96 0.57 2.36 4.13 2.94 3.68 4.52 2.06 1.26 1.18 0.36 2.95 0.7 0.06 3.07 0.32 4.04 0.90 0.48
6.5 8.4 7.4 6.9 7.6 7.5 7.9 7.7 7.6 8.3 8.3 8.2 8.4 8.2 8.1 8.3 7.6 9.5 7.5 7.6 7.4 8.0 8.8 7.4 7.5
58-22-0 3380-34-5 738-70-5
Steroide Antibiotic Antibiotic
288.4 289.6 290.3
– 8.0d [0/] 7.1,a 6.6d [þ/0]
0.0234a 0.011 0.49
3.47 4.76 0.73
8.5 7.4 8.1
a Experimentally determined values for compounds in neutral form as listed in EPI Suite v. 3.20 database (used for compounds that were predominantly present in the neutral form at pH 8 (http://www.epa.gov/oppt/exposure/pubs/episuite.htm)). b At pH 8 as listed in SciFinder Scholar (values were calculated with Advanced Chemistry Development (ACD/Labs) Software v. 8.14) unless denoted by footnote a; Cs denotes aqueous solubility, D denotes octanol/water partition coefficient. c Computed from surface volumes that were determined with CAChe WorkSystem Pro Version 6.1.8; spherical molecules were assumed. d Values estimated with SPARC v. 4.2 (http://ibmlc2.chem.uga.edu/sparc/) – N species were treated as acids and bases. e Known or suspected endocrine disrupting chemical. f [0/], transitions from neutral to anionic form; [þ/0], transitions from cationic to neutral form. g From Dodd et al. (2006).
stock solution into LMW. Duplicate type 2 blanks were prepared with 0.45-mm membrane-filtered LMW, amended with 100 mg/L sodium azide and sent immediately after preparation by overnight carrier for analysis. (3) Type 3 blanks were prepared to test emerging contaminant stability over the 3-week contact time that was used for adsorption isotherm experiments. Preparation of type 3 blanks was identical to type 2 blanks. Duplicate type 3 blanks were placed in a rotary tumbler for 3 weeks prior analysis. Upon equilibration, samples containing adsorbent were filtered through 0.7-mm glass fiber filters. Prior to use, filters were baked for two hours at 400 C and cooled in a desiccator for 1 h.
2.4. Analytical method for determining emerging organic contaminant concentrations ECC analyses were completed at the laboratories of the Southern Nevada Water Authority (SNWA) using the analytical
method of Vanderford et al. (2003). Prior to analysis, analytes were concentrated by solid-phase extraction. Compound separation and quantification was accomplished by liquid chromatography/tandem mass spectrometry (LC/MS/MS), using electrospray ionization in both positive and negative modes and atmospheric pressure chemical ionization in positive mode. Method reporting limits for most compounds were 1.0 ng/L in water.
Results and discussion
Properties of the tested carbonaceous adsorbents and zeolites are summarized in Tables 2 and 3, respectively. The BET surface area and micropore volume of the carbonaceous resin Ambersorb 563 was approximately half that measured for the CC-602 GAC (Table 2). In contrast, the mesopore volume of the carbonaceous resin was larger than that of the activated carbon. Fig. 1 depicts micropore size distributions of the
water research 43 (2009) 3787–3796
Table 2 – Physical characteristics of carbonaceous adsorbents. Data from Knappe et al. (2007). Adsorbent CC-602 Ambersorb 563
BET surface area (m2/g)
Micropore volumea (cm3/g)
DFT mesopore volumeb (cm3/g)
BJH mesopore volumec (cm3/g)
4.37 101 2.01 101
2.94 102 2.80 101
6.01 102 3.18 101
˚. a Micropore volume calculated by density functional theory (DFT) for pores with widths less than 20 A ˚ (upper limit for DFT model). b Mesopore volume calculated by density functional theory (DFT) for pores with widths ranging from 20 to 360 A ˚. c Mesopore volume calculated by Barrett, Joyner, and Halenda (BJH) method for pores with widths ranging from 20 to 500 A
Table 3 – Zeolite characteristics. Framework type
Manufacturer’s ID code
BET surface area (m2/g)
0.65 nm 0.70 nm (12-ring) 0.74 nm 0.74 nm (12-ring)
a Information provided by manufacturer.
carbonaceous adsorbents used in this study. The micropore size distribution of adsorbents is important because the pore volume in a size range similar to the molecular size of the targeted contaminant is one important factor that determines the adsorption capacity of a given adsorbent (e.g., Newcombe et al., 1997; Li et al., 2002; Karanfil and Dastgheib, 2004). It is important to note that the ECCs listed in Table 1 exhibit ˚ when spherical molecules are diameters in the range of 6–10 A assumed. Therefore, the results from the adsorbent characterization suggest that the carbonaceous resin may not be as effective as the coconut-shell based activated carbon for the removal of the ECCs listed in Table 1 because the micropore volume of Ambersorb 563 in the size range of the ECCs is considerably smaller than that of the CC-602 GAC (Fig. 1). Zeolites are crystalline, microporous materials with uniform pore dimensions. The primary zeolite building blocks are TO4 tetrahedra, where T is either a Si(IV) or Al(III) atom located at the center of the tetrahedron. Tetrahedra are
0.08 0.06 0.04 0.02
Pore width (Å) Fig. 1 – Micropore size distribution of carbonaceous adsorbents ( y-axis represents differential pore volume [V] ˚ ). Adsorbate in cm3/g adsorbent per unit pore width [W] in A diameters (see Table 1) are indicated for reference; acetaminophen was the smallest adsorbate in the ECC mixture and iopromide the largest.
Concentration in Type 3 Blank (ng/L)
linked via their oxygen atoms to other tetrahedra to form structural subunits. The linking of recurring structural subunits produces the crystalline framework structure of a zeolite, within which exist voids and channels of discrete and regular size. The mordenite zeolite framework contains parallel 12- and 8-ring channels (i.e. the perimeter of an elliptical channel opening is formed by 12 and 8 T atoms, respectively). Perpendicular to these channels are 8-ring channels that connect one 12-ring channel to an adjacent 8-ring channel. The 12- and 8-ring channels have dimensions ˚ and 2.6 5.7 A ˚ , respectively (McCusker and of 6.5 7.0 A Baerlocher, 2001). Because of the small size of the 8-ring channels, they cannot be accessed by organic contaminants. The Y-zeolite has circular, 12-ring windows with a diameter ˚ (or 7.4 7.4 A ˚ ) and supercages with a diameter of of 7.4 A ˚ (Rouquerol et al., 1999). BET surface areas of the about 13 A zeolites were 505 and 810 m2/g for the mordenite and Y zeolites, respectively (Table 3).
Batch 1 Batch 2 1:1 Line
1000 900 800 700 600 500
400 300 Estradiol
Androstenedione, Progesterone, Testosterone
Concentration in Type 2 Blank (ng/L)
Fig. 2 – Stability of emerging organic contaminants over the three-week equilibration time. Type 2 blanks were analyzed immediately upon preparation, type 3 blanks were mixed for 3 weeks prior to analysis. Source: Knappe et al. (2007) ª2007 AwwaRF. Reprinted with permission.
water research 43 (2009) 3787–3796
Adsorbent-Free Blank 1 mg/L Ambersorb 563
h At en ra (6 zi 4. C ar Ca ne 7% ba ffe (4 ) m in 1.3 az e % ep (35 ) in e .9% D ( ) D EE 38.9 ia ze T (3 %) D pa 6. ic lo m 5% fe (3 na 4. ) D c ( 4% 20 ) ila Es ntin .9% tra (2 ) di 6. ol 0% ( Es 1 ) tri 5.6 Et hy Est ol ( %) ny ro 20 le ne .4% st (4 r 8 ) Fl ad uo iol .2% x ) ( 3 e G e tin 0.5 H mfi e ( %) yd br 75 ro oz . co il 8% ) d (2 Ib on 8.2 up e % ( 2 r ) Io ofe 8.0 M pro n ( %) ep m 28 ro ide .7 % ba m (16 ) a N . O ap te ( 7% xy ro ) 3 be xe 0.1 n % P n Su en zon (31 ) lfa tox e ( .2% m ify >9 ) et ho lline 9.8% xa (4 zo 6. ) le 8% TC (3 ) Tr EP 4.2 i % Tr clo ( im sa 36. ) et n ( 6% ho > pr 99 ) im .8% (3 8. ) 3% )
Fig. 3 – Removal of 25 emerging organic contaminants from Lake Mead water by 1 mg/L Ambersorb 563. Error bars represent standard deviations of duplicates. Removal percentages are shown in the x-axis labels. Source: Knappe et al. (2007) ª2007 AwwaRF. Reprinted with permission.
3.1.1. Removal of an emerging contaminant mixture by alternative adsorbents The adsorption of a mixture of 28 ECCs (Table 1) was compared for two high-silica zeolites (mordenite and Y), a carbonaceous resin (Ambersorb 563), and a coconut-shellbased GAC (CC-602). Adsorption isotherm experiments were conducted by spiking the emerging contaminants at
environmentally relevant concentrations (w200–900 ng/L) into Lake Mead water (LMW).
To measure background levels of emerging contaminants in LMW and the effects of sample handling on background contaminant levels, ‘‘type 1’’ blanks were prepared by filtering
1200 Adsorbent-Free Blank 1 mg/L CC-602
h At en ra (5 zi n 8.0 C ar Ca e ( %) ba ffe 53 i . m az ne 0% ep (71 ) .3 in D e ( %) D EE 69. ia ze T ( 9% D pa 43. ) ic lo m ( 5% ) fe n 52 D ac .0% ila (2 ) 5 n Es tin .5% tra (4 ) di 0.4 o Es l (7 %) tri 8. Et o 9 hy Es l (5 %) ny tro 8. ne 6% le st ) ( Fl rad 89. uo io 2% x l G eti (67 ) e n .5 H mfi e (9 %) yd br ro oz 8.4 % co il ) d (3 Ib on 1.0 up e % ro (76 ) fe Io . M pro n ( 6% ep m 23 ) .3 ro id e % ba m (17 ) at N .1 a e % O xy pro (38 ) be xe .3 Pe nz n (4 %) Su n on 6 lfa tox e ( .6% m ify >9 llin 9. ) et ho e 8% xa ( zo 69. ) le 9% TC (2 ) E 7.1 T Tr ricl P (4 %) im os 9 et an .8% ho pr (99 ) im .5% (8 0. ) 6% )
Fig. 4 – Removal of 25 emerging organic contaminants from Lake Mead water by 1 mg/L CC-602. Error bars represent standard deviations of duplicates. Removal percentages are shown in the x-axis labels. Source: Knappe et al. (2007) ª2007 AwwaRF. Reprinted with permission.
water research 43 (2009) 3787–3796
100 % Removal this study % Removal Westerhoff et al. 2005
90 80 70 60 50 40 30 20 10
Fig. 5 – Comparison of percentage removal of the different emerging compounds studied in this work with results of Westerhoff et al. (2005). Removal percentages correspond to 1 mg/L dose of activated carbon CC-602 and three weeks contact time in this study and 5 mg/L dose of activated carbon WPM and 4 h contact time of Westerhoff et al. (2005).
LMW through a 0.45-mm membrane filter and amending the filtrate with 100 mg/L sodium azide. Of the ECCs shown in Table 1, type 1 blanks showed that the following compounds were present above the method reporting limit: oxybenzone (58 ng/L), caffeine (37 ng/L), triclosan (16 ng/L), meprobamate (13 ng/L), DEET (9.0 ng/L), sulfamethoxazole (8.2 ng/L), estriol (6.8 ng/L), carbamazepine (3.1 ng/L), dilantin (2.3 ng/L), atrazine (1.2 ng/L), and ibuprofen (1.2 ng/L). For all but three compounds, very similar levels were obtained by Vanderford et al. (2003) in their analysis of LMW samples. Oxybenzone (3.1 ng/L), caffeine (3.0 ng/L), and triclosan (<1 ng/L) were present at lower concentrations in the samples analyzed by Vanderford et al. (2003); the higher levels of these compounds in type 1 blanks suggest that they were introduced during sample handling. Because all samples in this study were handled in an identical manner and because spiked concentrations of ECCs were 1–2 orders of magnitude higher than the levels observed in type 1 blanks, the possible introduction of some compounds during sample handling did not affect the results and associated data interpretation. To test the stability of the contaminants throughout the 3-week equilibration time, results for ‘‘type 2’’ and ‘‘type 3’’ blanks were compared. Fig. 2 illustrates that most compounds were stable throughout the 3-week period used to achieve adsorption equilibrium, and good agreement was obtained between compound concentrations in type 2 and type 3 blanks. However, three compounds were not stable: androstenedione, progesterone, and testosterone, which were spiked at levels of w550–600 ng/L according to type 2 blanks, were below the method reporting limit in type 3 blanks (Fig. 2).
As a result, no adsorption data are presented for these three steroid hormones. In addition, poor agreement between type 2 and type 3 blanks was obtained for estrone, estriol, and estradiol in the second batch of blanks that were prepared, suggesting that initial concentration values for these three compounds are more uncertain than those for the remaining compounds. In all cases, the average concentration obtained in type 3 blanks was used to calculate solid-phase ECC concentrations on the adsorbents.
Fig. 3 summarizes concentrations for each of the 25 stable emerging organic contaminants in type 3 blanks and in samples that were equilibrated with 1 mg/L Ambersorb 563 in LMW. The results in Fig. 3 show that removal percentages between 20 and 50% were achieved for 19 of the 25 compounds at this adsorbent dose. Iopromide and estradiol were the two most poorly removed compounds, but the result for estradiol may have been influenced by uncertainty in the initial concentration (see large error bars in Fig. 3). Removal percentages in excess of 50% were observed for triclosan, oxybenzone, fluoxetine, and acetaminophen. Acetaminophen was the smallest of the studied ECCs ˚ ) and thus may have benefited from (molecular diameter 6.5 A the presence of a large volume of small pores in Ambersorb 563. Triclosan and oxybenzone were among the more hydrophobic compounds in the contaminant mixture (log D values >3, Table ˚ ). It is less 1), and their diameters were also relatively small (7.4 A clear why fluoxetine adsorbed so well on the carbonaceous resin given its greater hydrophilicity (log D w 2) and larger size ˚ ). (dia. w 8.2 A
water research 43 (2009) 3787–3796
1000 Adsorbent-Free Blank 100 mg/L HSZ-690HOA (MOR) 100 mg/L HSZ-390HUA (Y)
800 700 600 500 400 300 200 100
di o Es l tri o Et hy Est l ny ron le st e r Fl adi uo ol xe G em tine f H yd ibro z ro co il do n Ib up e ro fe Io n M pro ep m ro ide ba m at N e a O pro xe xy be n n P Su ent zon e lfa ox m ifyl et lin ho e xa zo le TC E Tr P Tr iclo im sa et n ho pr im
EE D D
in am et Ac
Fig. 6 – Removal of 25 emerging organic contaminants from Lake Mead water by 100 mg/L mordenite or Y zeolites. Source: Knappe et al. (2007) ª2007 AwwaRF. Reprinted with permission.
Coconut-shell based activated carbon
Fig. 4 summarizes concentrations for each of the 25 stable emerging organic contaminants in type 3 blanks and in samples that were equilibrated with 1 mg/L CC-602 in LMW. The results in Fig. 4 show that removal percentages exceeded 50% for 15 compounds at this adsorbent dose. For nine compounds, removal percentages between 20 and 50% were achieved, and for one compound, iopromide, removal was
CC-602 - Co = 426 ng/L CC-602 - Co = 100 µg/L Ambersorb 563 - Co = 426 ng/L Ambersorb 563 - Co = 100 µg/L 1
Adsorbent Dose (mg/L) Fig. 7 – Effect of initial SMX concentration and presence of trace level emerging contaminants in Lake Mead water on SMX removal by activated carbon CC-602 and carbonaceous resin Ambersorb 563. Source: Knappe et al. (2007) ª2007 AwwaRF. Reprinted with permission.
<20%. These results illustrate that the CC-602 GAC was more effective for the removal of emerging organic contaminants than Ambersorb 563, most likely because the GAC exhibited ˚ range. Fig. 4 a larger volume of pores with widths in the 6–9 A also illustrates that the estradiol removal percentage was 79%, which is higher than the 40–50% removal from Colorado River water that was measured for a 1 mg/L carbon dose by Yoon et al. (2005). However, a contact time of 4 h was used in their study, which is insufficient for adsorption equilibrium to be established. Fig. 5 compares removal percentages of the 25 stable ECCs in LMW obtained with a 1 mg/L CC-602 activated carbon dose and a three-week contact time (this study) with the average removal percentages obtained in four natural waters for the same 25 ECCs with a 5 mg/L WPM (coal-based) activated carbon dose and 4-h contact time (Westerhoff et al., 2005). Despite the differences in experimental conditions, Fig. 5 shows that results are very comparable between the two data sets. Linear regression analysis yielded a coefficient of determination (R2) of 0.9, demonstrating the good agreement between the data sets. The similarity between the results confirms the relative adsorbability of the tested ECCs on activated carbon; it is unclear, however, to what extent contact time, differences in background water matrices, and differences in activated carbon properties can explain why a carbon dose of 1 mg/L produced results that were similar to those obtained with a carbon dose of 5 mg/L in the prior study (Westerhoff et al., 2005).
water research 43 (2009) 3787–3796
Additional adsorption isotherm experiments were conducted with the carbonaceous adsorbents to determine the effects of initial contaminant concentration and of other contaminants in the mixture on ECC removal percentages. For this purpose, sulfamethoxazole (SMX) adsorption isotherm data for Ambersorb 563 and CC-602 were collected in LMW at an initial concentration of 100 mg/L and compared to SMX adsorption isotherm data obtained with an initial SMX concentration of 426 ng/L in the presence of 24 other emerging organic contaminants that were stable over the 3-week equilibration time. As shown in Fig. 7, removal percentages of SMX at a given adsorbent dose were similar for the two initial concentrations. The similarity in removal percentages at initial SMX concentrations that differed by 2.5 orders of magnitude was somewhat surprising for an anionic contaminant (pKa of the sulfonamide group of SMX is 5.6) because it was anticipated that electrostatic interactions between the adsorbent surface and SMX anions would depend on the SMX surface concentration (Mu¨ller et al., 1980, 1985). However, the results obtained here are similar to previous observations for many neutral organic contaminants (e.g., Knappe et al., 1998; Graham et al., 2000; Westerhoff et al., 2005); i.e., in the presence of competing background organic matter, the removal percentage of a trace organic contaminant at a given adsorbent concentration is independent of the initial trace organic contaminant concentration, or, in other words, isotherms describing the adsorption of trace organic contaminants in the presence of competing background organic matter are linear when experiments are conducted with a fixed adsorbent dose and varying initial adsorbate concentrations. To obtain 90% SMX removal in LMW, the results in Fig. 7 suggest that w7 mg/L of CC-602 or w11.5 mg/L Ambersorb 563 would be required if adsorption equilibrium can be obtained. From a practical standpoint, these results show that adsorption studies that are conducted in the mg/L concentration range, where chromatographic analyses can be conducted with less sophisticated equipment and with greater accuracy, yield percent ECC removal data at a given adsorbent dose that are comparable to those obtained from experiments conducted in the ng/L range.
Adsorbent Dose (mg/L)
Iopromide Meprobamate Dilantin Sulfamethoxazole
effective if the removal of a specific organic contaminant is targeted (e.g. MTBE, Rossner and Knappe, 2008), but they are not effective when the removal of a broad mixture of contaminants is targeted.
Hydrocodone Ibuprofen Carbamazepine Ethynylestradiol
Fluoxetine Triclosan Oxybenzone
Adsorbent Dose (mg/L) Fig. 8 – Percent removal of ECCs from Lake Mead water as a function of (a) activated carbon and (b) carbonaceous resin dose. Data from Knappe et al. (2007).
Fig. 6 summarizes concentrations for each of the 25 stable emerging organic contaminants in type 3 blanks and in samples that were equilibrated with 100 mg/L of mordenite (HSZ-690HOA) or Y (HSZ-390HUA) zeolite. The results in Fig. 6 show that 15 compounds were either completely or partially removed with the mordenite zeolite; in contrast, only three compounds (fluoxetine, oxybenzone, and triclosan) were removed by the Y zeolite. Thus, mordenite was a more effective adsorbent than the Y zeolite for the removal of emerging organic contaminants that were spiked at levels between w200 and 900 ng/L into LMW. It is interesting to note that seemingly small differences in pore size and shape had an important effect on removal efficiencies observed with the two high-silica zeolites. Mordenite has elliptical channels with minor and ˚ , respectively, while Y major axis dimensions of 6.5 and 7 A ˚ zeolites contain channels with circular openings that are 7.4 A ˚ in in diameter that open into supercages that are about 13 A diameter. Compared to the zeolites, carbonaceous adsorbents were more effective for ECC removal. This result was primarily attributed to the fact that activated carbons exhibit a broader micropore size distribution, in which compounds of different shapes and sizes can be effectively accommodated. In contrast, high-silica zeolites, with their uniform pore sizes, appear to be
Effect of initial ECC concentration
Adsorbent dose requirements
Fig. 8 summarizes adsorption isotherm data for contaminants in the emerging contaminant mixture on CC-602 and Ambersorb 563. Isotherm data are presented by plotting the percentage of the contaminant remaining in solution as a function of adsorbent dose, a format that eliminates the initial concentration dependence of the trace organic contaminant isotherm in the presence of NOM (e.g., Knappe et al., 1998). For CC-602, the shaded area bracketed by the isotherms for triclosan and diclofenac includes isotherms for 19 of the 23 remaining compounds (Fig. 8a). Thus, the results in Fig. 8a suggest that CC-602 doses between w1 and 8 mg/L would be required to achieve 99% removal for 21 of the 25 tested emerging contaminants. The remaining four
water research 43 (2009) 3787–3796
compounds, iopromide, sulfamethoxazole, meprobamate, and ibuprofen, were not as efficiently removed. For the most poorly adsorbed compound, iopromide, a CC-602 dose of 10 mg/L would have led to w78% removal from LMW. For Ambersorb 563, the shaded area bracketed by the isotherms for fluoxetine and ethynylestradiol includes isotherms for 10 of the 23 remaining compounds (Fig. 8b). Thus, the results in Fig. 8b suggest that Ambersorb 563 doses between w3 and 10 mg/L would be required to achieve 99% removal for 12 of the 25 tested emerging contaminants. Two compounds, triclosan and oxybenzone, were more adsorbable than fluoxetine and would require Ambersorb 563 doses <3 mg/L to obtain 99% removal from LMW. On the other hand, 8 compounds (iopromide, meprobamate, dilantin, sulfamethoxazole, diclofenac, hydrocodone, ibuprofen, and carbamazepine) were not as adsorbable as ethynylestradiol. At an Ambersorb 563 dose of 10 mg/L, iopromide removal was <10%; at the same Ambersorb 563 dose, removal of the second-most poorly adsorbed compound, meprobamate, was w70%.
Overall, the results obtained in this study suggest that activated carbon adsorption processes can effectively remove antimicrobial compounds, EDCs and other pharmaceuticals from lake water. For the tested coconut-shell-based activated carbon, a dose of 10 mg/L was sufficient to achieve 98% contaminant removal for 24 of the 25 stable compounds in the mixture. If adsorption equilibrium can be reached, an activated carbon dose that might typically be added by a utility for taste and odor control is therefore sufficient to achieve a 2-log removal of many emerging organic contaminants. The greater effectiveness of activated carbon relative to the carbonaceous resin was primarily related to GAC’s larger volume of pores ˚ range. Finally, the results obtained with widths in the 6–9 A with the zeolites showed that this adsorbent class with its uniform pores does not provide an effective broad-spectrum barrier against the wide variety of compound classes that a utility might want to remove from its source water.
Acknowledgements The authors of this study would like to thank the Water Research Foundation (formerly AwwaRF) for funding this project through agreement #2905. In addition, the authors would like thank Janie Zeigler at the Southern Nevada Water Authority for LC-MS/MS analyses.
Adams, C., Wang, Y., Loftin, K., Meyer., M., 2002. Removal of antibiotics from surface and distilled water in conventional water treatment processes. J. Environ. Eng. ASCE 128 (3), 253–260. Choi, K.J., Kim, S.G., Kim, C.W., Park., J.K., 2006. Removal efficiencies of endocrine disrupting chemicals by coagulation/flocculation,
ozonation, powdered/granular activated carbon adsorption, and chlorination. Korean J. Chem. Eng. 23 (3), 399–408. Dodd, M.C., Buffle, M.O., von Gunten., U., 2006. Oxidation of antibacterial molecules by aqueous ozone: moiety-specific reaction kinetics and application to ozone-based wastewater treatment. Environ. Sci. Technol. 40 (6), 1969–1977. 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 Res. 34, 2291–2300. Ikehata, K., El-Din, M.G., Snyder., S.A., 2008. Ozonation and advanced oxidation treatment of emerging organic pollutants in water and wastewater. Ozone Sci. Eng. 30, 21–26. Karanfil, T., Dastgheib, S.A., 2004. Trichloroethylene adsorption by fibrous and granular activated carbons: aqueous phase, gas phase, and water vapor adsorption studies. Environ. Sci. Technol. 38, 5834–5841. Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D., Barber, L.B., Buxton., H.T., 2002. Pharmaceuticals, hormones, and other organic wastewater contaminants in US streams, 1999–2000: a national reconnaissance. Environ. Sci. Technol. 36 (6), 1202–1211. Knappe, D.R.U., Matsui, Y., Snoeyink, V.L., Roche, P., Prados, M.J., Bourbigot., M.M., 1998. Predicting the capacity of powdered activated carbon for trace organic compounds in natural waters. Environ. Sci. Technol. 32, 1694–1698. Knappe, D.R.U., Rossner, A., Snyder, S.A., Strickland., C., 2007. Alternative Adsorbents for the Removal of Polar Organic Contaminants. American Water Works Association Research Foundation, Denver, Colorado. Li, L., Quinlivan, P.A., Knappe., D.R.U., 2002. Effects of activated carbon surface chemistry and pore structure on the adsorption of organic contaminants from aqueous solution. Carbon 40 (12), 2085–2100. McCusker, L.B., Baerlocher, C., 2001. Zeolite structures. In: van Bekkum, H., Flanigen, E.M., Jacobs, P.A., Jansen, J.C. (Eds.), Introduction to Zeolite Science and Practice, Second Completely Revised and Expanded Edition. Elsevier Science, Amsterdam, pp. 37–67. Mu¨ller, G., Radke, C.J., Prausnitz., J.M., 1980. Adsorption of weak organic electrolytes from aqueous solution on activated carbon. Effect of pH. J. Phys. Chem. 84, 369–376. Mu¨ller, G., Radke, C.J., Prausnitz., J.M., 1985. Adsorption of weak organic electrolytes from dilute aqueous solution onto activated carbon. Part 1. Single-solute systems. J. Colloid Interface Sci. 103 (2), 466–483. Newcombe, G., Drikas, M., Hayes., R., 1997. Influence of characterized natural organic material on activated carbon adsorption: II. Effect on pore volume distribution and adsorption of 2-methylisoborneol. Water Res. 31, 1065–1073. Quinlivan, P.A., Li, L., Knappe., D.R.U., 2005. Effects of activated carbon characteristics on the simultaneous adsorption of aqueous organic micropollutants and natural organic matter. Water Res. 39 (8), 1663–1673. Rosenfeldt, E.J., Chen, P.J., Kullman, S., Linden, K.G., 2007. Destruction of estrogenic activity in water using UV advanced oxidation. Sci. Total Environ. 377 (1), 105–113. Rossner, A., Knappe, D.R.U., 2008. MTBE adsorption on alternative adsorbents and packed bed adsorber performance. Water Res. 42, 2287–2299. Rouquerol, F., Rouquerol, J., Sing., K., 1999. Adsorption by Powders and Porous Solids. Academic Press, San Diego, CA. Snyder, S.A., Adham, S., Redding, A.M., Cannon, F.S., DeCarolis, J., Oppenheimer, J., Wert, E.C., Yoon, Y., 2007. Role of membranes and activated carbon in the removal of endocrine disruptors and pharmaceuticals. Desalination 202 (1–3), 156–181. Stackelberg, P.E., Furlong, E.T., Meyer, M.T., Zaugg, S.D., Henderson, A.K., Reissman., D.B., 2004. Persistence of pharmaceutical compounds and other organic wastewater
water research 43 (2009) 3787–3796
contaminants in a conventional drinking-water-treatment plant. Sci. Total Environ. 329 (1–3), 99–113. Ternes, T.A., Meisenheimer, M., McDowell, D., Sacher, F., Brauch, H.J., Haist-Gulde, B., Preuss, G., Wilme, U., ZuleiSeibert., N., 2002. Removal of pharmaceuticals during drinking water treatment. Environ. Sci. Technol. 36, 3855–3863. Vanderford, B.J., Pearson, R.A., Rexing, D.J., Snyder., S.A., 2003. Analysis of endocrine disruptors, pharmaceuticals, and personal care products in water using liquid chromatography/tandem mass spectrometry. Anal. Chem. 75, 6265–6274. Westerhoff, P., Yoon, Y., Snyder, S., Wert., E., 2005. Fate of endocrine-disruptor, pharmaceutical, and personal care
product chemicals during simulated drinking water treatment processes. Environ. Sci. Technol. 39 (17), 6649–6663. Yoon, Y., Westerhoff, P., Snyder, S.A., Esparza., M., 2003. HPLC-fluorescence detection and adsorption of bisphenol A, 17b-estradiol, and 17a-ethynyl estradiol on powdered activated carbon. Water Res. 37, 3530–3537. Yoon, Y., Westerhoff, P., Snyder., S.A., 2005. Adsorption of 3 H-Labeled 17-b estradiol on powdered activated carbon. Water Air Soil Pollut. 166, 343–351. Yu, Z.R., Peldszus, S., Huck, P.M., 2008. Adsorption characteristics of selected pharmaceuticals and an endocrine disrupting compound – naproxen, carbamazepine and nonylphenol – on activated carbon. Water Res. 42, 2873–2882.