Evaluation of an Integrated Adsorption Method for the Isolation and

22

Downloaded by GEORGETOWN UNIV on August 26, 2015 | http://pubs.acs.org Publication Date: December 15, 1986 | doi: 10.1021/ba-1987-0214.ch022

Evaluation of an Integrated Adsorption Method for the Isolation and Concentration of Trace Organic Substances from Water M. F. Giabbai , E. S. K. Chian , J. H . Reuter , H. P. Ringhand , and F. C. Kopfler 1,2

1

3

4

4

School of Civil Engineering and School of Geophysical Sciences, Georgia Institute of Technology, Atlanta, GA 30332 Health Effects Research Laboratory, U.S. Environmental Protection Agency, Cincinnati, O H 45268 1

3

4

A scheme for the isolation and concentration of dissolved trace organic substances from water for toxicological and chemical characterization was evaluated. The principle behind this scheme consists of the separation of organic solutes into fractions by adsorption onto different adsorbents (i.e., XAD-8 resin, AG MP-50 cation-exchange resin, and Carbopack Β graphitized car bon black) under varying pH conditions. Test solutions contain ing 22 model organic substances along with inorganic salts were used to monitor process performance. High-resolution gas chromatography and high-performance liquid chromatography were employed for the quantitation of each model compound. The isolation-fractionation scheme proved to be effective for 16 out of 22 model compounds; average recoveries varied between 30% and 90%.

T H E H I G H C O M P L E X I T Y A N D D I L U T E D F O R M i n w h i c h organic c o m pounds occur i n natural and drinking waters require that isolation, concentration, and fractionation procedures be employed to achieve a suitable sample for chemical and toxicological characterization. T h e use of these methods in analytical schemes has thus far allowed the iden tification of several hundred trace organic substances i n drinking water 2

Current address: EnvironScience Laboratories, Inc., Atlanta International Industrial

Park, Atlanta, GA 30316

0065-2393/87/0214/0467$06.00/0 © 1987 American Chemical Society

In Organic Pollutants in Water; Suffet, I., et al.; Advances in Chemistry; American Chemical Society: Washington, DC, 1986.


468

O R G A N I C P O L L U T A N T S IN W A T E R

(I). Compounds thus identified may then be subjected to various in vivo and in vitro test systems to assess biological activity. Because of the large number of compounds and the possibility of additive or synergistic effects, it is impossible to test all combinations. Alternatively, a primary biological screen may be used to identify those natural and drinking water concentrates or fractions that contain biologically active components. C h e m i c a l identification of the bioactive substances is then attempted (2). Currently, the alternative approach is preferred for the risk assessment of the nonvolatile fraction of organic substances present in water because of analytical limitations. In either approach, the selection of isolation (e.g., solvent extraction, adsorption on carbon and synthetic resins) and concentration (e.g., lyophilization, v a c u u m distillation, reverse osmosis, ultrafiltration) methods is of paramount importance in properly assessing the potential toxicity of waterborae organics. A comprehensive literature review on the development and application of these and other methods to biological testing has recendy been published b y Jolley (3). Several attempts to improve the percent recovery of organic contituents f r o m water have been pursued b y sequentially arranging different methods in a multistep scheme. Baird et al. (4) experienced 80-903) removal of organic carbon b y using a series of ion-exchange and macroreticular resins to prepare organic residues from large volumes of water for chemical and toxicologic testing. Amberlite X A D - 4 / 8 columns in series with activated carbon columns were used b y V a n Rossum and W e b b (5) to process 1000 L of tap water. Organic pollutants were subsequently identified b y gas ehromatographie-mass spectrometric ( G C - M S ) analysis of the solvent-eluted fractions. Recendy, Leenheer (6) proposed a comprehensive analytical scheme whereby the dissolved organic carbon ( D O C ) in natural waters may be separated into operationally defined fractions on the basis of their adsorption onto different substrates (macroreticular adsorbents and ion-exchange resins) under varying p H conditions. Recovery of input and size of the individual fractions has been evaluated in terms of total organic carbon ( T O C ) analysis. If specific classes of compounds or specific components are to be biologically tested, suitable concentration methods can be designed. O n the other hand, to estimate the overall hazard associated with the organic constituents of a water sample, a viable alternative consists in the use of different methods in a sequential scheme. In both cases, several critical areas of concern must be considered before applying such methods to real water samples: (1) The aqueous organic concentrate prepared b y the selected concentration scheme has to be representative of the original water sample with regard to the relative abundance of the individual components. (2) The transformation of organic constituents between preparation of concentrates and biological testing and/or chemical analysis must be avoided. (3) The effect of In Organic Pollutants in Water; Suffet, I., et al.; Advances in Chemistry; American Chemical Society: Washington, DC, 1986.


22.

GIABBAI E T A L .

Evaluation of an Integrated Adsorption Method

469

humic material, w h i c h constitutes the bulk of the organic fraction of natural and drinking waters, on the recovery of trace solutes has to b e taken into account. (4) The introduction of artifacts and constituents' alteration b y the concentration methods must be kept to a m i n i m u m . (5) The co-recovery of toxic inorganic constituents must be evaluated. (6) T h e potential effect of chlorine residual on the material used in the concentration scheme (e.g., resin, membranes) must b e assessed. Moreover, the development and application of several different concentration schemes require that a strict comparison be based on the recovery of selected model organic substances representative of a w i d e range of chemical classes, functional group contents, and molecular weights. These considerations, as w e l l as the necessity for a comprehensive approach toward the isolation, concentration, and fractionation of trace components of D O C i n water, have led to the evaluation of a fractionation scheme wherein selected organic compounds having different functionalities a n d sorption parameters were separated a n d concentrated. Test solutions containing 22 m o d e l compounds at parts-perbillion (micrograms-per-liter) concentration levels were chosen as the basis for process evaluation. The criteria used i n selecting the m o d e l compounds were to provide (1) a variety of functional groups; (2) a range of physical properties such as volatility, solubility, polarity, and molecular weight; (3) k n o w n water pollutants; a n d (4) halogenated derivatives. T h e majority of the compounds were taken f r o m a list of consensus voluntary reference compounds (7). The inclusion of humic acid and inorganic salts in the test solutions was an attempt to simulate d r i n k i n g water. T h e proposed fractionation scheme was initially evaluated o n a laboratory scale and was subsequently adapted for processing several hundred liters of aqueous test solutions.

Experimental Resin and Carbon Adsorbents. Amberlite XAD-8 was obtained from Rohm and Haas as an industrial-grade preparation in 20-50-mesh size beads. The cation-exchange AG MP-50, 20-50 mesh size, was supplied by BioRad Laboratories. Leenheer's (6) procedures for cleanup, preparation, and storage of the resins were followed. In addition, XAD-8 was Soxhlet extracted with methylene chloride immediately after the acetone and hexane extractions. Glass columns (200 X 13 mm i.d.) with Teflon stopcocks were packed with ~15-mL bed volumes of resin. The graphitized carbon black (GCB) Carbopack B, 100-120 mesh particle size, was purchased from Supelco. Acetone, methylene chloride, and organic-free water (OFW) were used to wash the carbon prior to column packing. Because of the fragile nature of this material, care was taken to avoid excessive mechanical stress during its handling. Two hundred milligrams of GCB was packed in a glass column (200 X 5 mm i.d.), as recommended by Bacaloni et al. (8). Reagents. The organic model compounds were purchased from Aldrich Chemical Company, Alfa Products, Fluka Chemical Company, and Analabs; In Organic Pollutants in Water; Suffet, I., et al.; Advances in Chemistry; American Chemical Society: Washington, DC, 1986.


470


purities varied from 96% to 99%, as specified in the manufacturers' literature. Inorganic salts, hydrogen peroxide (50% solution), mineral acids, and bases were obtained from Fisher Scientific Company. Organic solvents for purifying and extracting and for analytical operations were distilled-in-glass grade (Burdick-Jackson). Heptafluorobutyric anhydride and trifluoroacetic anhydride were obtained from PCR Research Chemicals; acetyl chloride was supplied by Mallinckrodt; Diazald was supplied by Aldrich; and N,0-bis(trimethylsilyl)trifluoroacetamide was supplied by Pierce Chemical Company. The humic acid used in these experiments was provided by the Health Effects Research Labora tory of the U.S. Environmental Protection Agency, and it had been prepared from a commercial-grade humic acid (Fluka). OFW was prepared by passing tap water through a series of treatments in the following sequence: a Millipore no. 360 activated carbon cartridge (Continental Water Systems Company), a Millipore no. 300 deionizer cartridge, and a glass column (60 cm X 2.5 cm i.d.) packed with 50 g of 16-30-mesh size filtrasorb F-400 virgin activated carbon (Calgon Company). Hydrogen peroxide was added to the stream, which then passed through UV light sources (Modified Model H-50, Ultraviolet Tech nology). This process resulted in water having an average TOC concentration of 27 ± 15 ppb and a hydrogen peroxide residue of 3, CaSC>4, and C a C l 2 ' 2 H 2 0 , which are given in milligrams per liter. Data are taken from references 24 and 25; i = insoluble, s = soluble, δ = slightly soluble, and — = no data.

2,4'-dichlorobiphenyl, and 2,2',5,5'-tetrachlorobiphenyl were spiked by sequen tially exposing these compounds to solvents of increasing polarity (hexane X acetone X OFW). Blowing dry with a gentle N stream and sonication were used to remove the solvent and to aid solubilization, respectively. The humic acid stock solution was added as the last component. 2

Isolation-Fractionation Scheme. Figure 1 illustrates the isolation-fraction ation scheme devised and evaluated in this study. Step 1: The test solution was first acidified to pH 2 and passed through the XAD-8 column by gravity flow at a rate of 15 bed volumes/h. The last portion of the test solution remaining in the column was displaced from the resin by 1 bed volume of 0.01 N HC1 rinse, which was combined with the original test solution. Step 2: The hydrophobic acid fraction was desorbed with 0.25 bed volumes of 0.1 Ν NaOH followed by 1.5 bed volumes of OFW. Step 3: The test solution effluent from the XAD-8 (pH 2) was adjusted to pH 10 with 1 Ν NaOH and recycled through the XAD-8 column at a flow rate of 15 bed volumes/h. Following the sample, 2.5 bed volumes of OFW were used to rinse the XAD-8 column. The rinse was com-


472


Water solution pH 2

XAD-8

(2)

Elution with NaOH

(4)

Elution with HCI . . .. , Extraction with C H C I

(5)

Λ

Hydrophobic Acids Hydrophobic Bases

Λ

2

t (3)

2

•

Hydrophobic Neutrals


Water solution pH 10

(6)

I

Water solution pH 2

(7)

AG MP - 50

Elution with NH OH 4

Hydrophilic Bases

I (8) Water solution pH 7 * (9)

Elution with C H C I 2

2

GCB

CARBOPACK Β

I

•

Effluent

Final Effluent

Figure 1. Flow scheme of isohtion-fractionation. (Reproduced with per mission from reference 11. Copyright 1983 Elsevier.) bined with the test solution effluent. Step 4: The hydrophobic base fraction was eluted with 0.25 bed volumes of 0.01 Ν HCI followed by 1.5 bed volumes of 0.1 Ν HCI. Step 5: Finally, the XAD-8 resin was transferred from the column to a separatory funnel and extracted with three 50-mL aliquots of methylene chloride to desorb the hydrophobic neutral fraction. Step 6: The column effluent (pH 10), with its remaining dissolved hydrophilic substances, was readjusted to pH 2 with concentrated HCI and then passed through the AG MP-50 cationexchange column at aflowrate of —15 bed volumes/h. Step 7: The hydrophilic base fraction was desorbed by elution with approximately 0.8 bed volumes of 1 Ν NH4OH. Step 8: Finally, the test solution effluent was adjusted to pH 7 and processed through the Carbopack Β column at aflowrate that allowed a contact time of approximately 0.5 min. Step 9: The GCB was extracted with methylene chloride. Analytical Procedures.

HYDROPHOBIC NEUTRAL FRACTION.

The hydro

phobic neutral fraction, which was desorbed in methylene chloride, was concen trated to an appropriate volume (1 mL) in a Kuderna-Danish apparatus. Then, under a stream of N 2 and after addition of the internal standard (i.e., hexamethylbenzene), this fraction was analyzed by GC-FID and GC-MS.


22.

GIABBAI ET AL.

473


H Y D R O P H O B I C A C I D F R A C T I O N . Known amounts of surrogate compounds (i.e., undecanoic acid and 3-quinolinecarboxylic acid) were added to 1 or 2 mL of the hydrophobic acid fraction, and the water was removed under a stream of N at room temperature. The residue was acidified with approximately 0.3 mL of 6 Ν HCI and brought to dryness under a stream of pure N . Finally, it was redissolved in approximately 1 mL of ethyl ether by carefully stirring with a glass rod to help dissolve any acids. The solution was subsequently methylated with gaseous diazomethane (12). Diazomethane was generated by adding 15 drops of aqueous NaOH solution (35$) to a solution of Diazald in methanol (~1 mg in 10 mL). Diazomethane gas was bubbled under N pressure (flow rate of —40-60 mL/min) into the ethereal solution containing the acids for approxi mately 10-20 s. After addition of hexamethylbenzene, the solution was analyzed by GC. For every batch of hydrophobic acid samples, a standard solution con sisting of trimesic, stearic, quinaldic, and surrogate acids was prepared in OFW (concentration of 50 μg/mL), dried, and methylated according to the method just described. This standard solution served as the basis for the quantitative evaluation of the samples. Humic acid was measured in this fraction by spec trophotometry at 430 nm. Standards (10-40 mg/L) and samples were analyzed at identical pH values. 2

2


2

HYDROPHOBIC BASE FRACTION.

The hydrophobic base fraction was ad

justed to pH 10. Α 50-μί aliquot of the aqueous solution was subjected to HPLC analysis to test for the presence of 5-chlorouracil. The remaining aqueous solu tion was solvent extracted with methylene chloride. The extract was first con centrated in a Kuderna-Danish apparatus, then under a stream of N , and analyzed by GC-FID and GC-MS. 2

HYDROPHILIC BASE FRACTION.

An aliquot (1-2 mL) of the hydrophilic

base fraction was dried under a stream of N , acidified with HCI, and analyzed for glycine after derivatization with isoamyl alcohol, acetyl chloride, and heptafluorobutyric anhydride according to the procedure described by Burleson et al. (13). An aliquot (1-2 mL) of the same hydrophilic base fraction was analyzed for quinaldic acid following the procedure mentioned for the hydrophobic acid fraction. The remaining portion of the hydrophilic base fraction was extracted at pH 10 with 50 mL of methylene chloride. The extract was concentrated to 1 mL and analyzed by GC-FID and GC-MS. 2

C A R B O P A C K Β F R A C T I O N . The Carbopack Β column was eluted with 100 mL of methylene chloride. The effluent was concentrated to 1 mL and directly analyzed by GC-FID and GC-MS. F I N A L E F F L U E N T . The final effluent (see Figure 1) was solvent extracted with methylene chloride and analyzed by GC. Details of the analytical procedures for the determination of the model organic compounds are published elsewhere (II).

Results and Discussion Preliminary experiments resulted in the formation of a precipitate due to the presence of inorganic salts when the p H of the test solution was raised to 10 for the first passage through the X A D - 8 column to isolate the hydrophobic base fraction. Initially, desalting the solution with a cation-exchange resin (i.e., AG-50-X8, N a form) was tried before +



474


processing through the fractionation scheme. However, several model compounds were lost presumably by adsorption on the nonionic lattice of the resin. Precipitate formation was avoided when the sequence of adsorption onto the X A D - 8 column was reversed: the test solution was first adjusted to p H 2 for the adsorption of the hydrophobic acids and neutrals and then adjusted to p H 10 for the isolation of the hydrophobic bases. Therefore, this approach was ultimately adopted in this study. The final effluent of the test solution was readjusted to p H 2 before passing through the cation-exchange resin A g M P - 5 0 to isolate the hydrophilic bases. Six experiments were conducted under these conditions. The results are expressed as average percent recoveries in Table II. M a l c o l m et al. (14) and Thurman et al. (15) noticed that the adsorption of solutes onto X A D - 8 macroreticular resin c o u l d be predicted b y means of a linear correlation between the logarithm of the capacity factor and the inverse of the logarithm of the water solubility of each c o m p o u n d . Their investigation, however, was limited to approximately 20 selected organic compounds in individual aqueous solutions. By comparing the results shown in Table II and the water solubility properties of each model compound used in this study (see Table I), it appears that the predictive model could serve for a first estimate of the recovery of multisolute solutions at trace levels. H o w ever, l o w recoveries and the erratic behavior of several compounds included in this study suggest that additional factors need to be considered. It appears that 2,4-dichlorophenol, w h i c h was expected primarily in the hydrophobic acid fraction, does not follow the predictive m o d e l (see Table II). Solute-solute interactions may be responsible for this unexpected behavior, and the fact that 2,4-dichlorophenol was partially recovered in the hydrophilic base fraction suggests an adsorptive affinity to the styrene-divinylbenzene lattice of the cation-exchange resin. T h e relatively poor recovery of 1-chlorododecane and 2,2',5,5'tetrachlorobiphenyl in the hydrophobic neutral fraction (see Table II) may be attributed to difficulties encountered in solubilizing them in water and to subsequent losses b y adsorption onto glass walls and Teflon tubing, although precautions against solubilization problems had been taken during the preparation of the test solution (see Experimental section). Because of the l o w concentrations, no attempts were made to verify adsorption losses. Bis(2-ethylhexyl) phthalate was found in several fractions (see Table III), a fact that may indicate nonspecific adsorption onto both macroreticular and ion-exchange resins. The small concentrations of methyl isobutyl ketone ( M I B K ) , 5-chlorouracil, and quinaldic acid recovered in their respective fractions made the quantitative analysis of these compounds unreliable. M I B K was detected


In Organic Pollutants in Water; Suffet, I., et al.; Advances in Chemistry; American Chemical Society: Washington, DC, 1986. /

fl

32.4 41.8

Stearic acid Trimesic acid 2,4-Dichlorophenol Quinaldic acid Isophorone Biphenyl 1-Chlorododecane 2,6-Di-^rf-butyl-4-methylphenol 2,4 -Dichlorobiphenyl 2,2' ,5,5-Tetrachlorobipheny 1 Anthraquinone Phenanthrene Bis(2-ethylhexyl) phthalate Furfural Quinoline 5-Chlorouracil Caffeine Glycine Humic acids Chloroform Methyl isobutyl ketone NQ

22.1 ± 10.6

OB

16.4 ± 5.4

80.8 ± 18.5 82.7 ± 5.8 33.8 ± 6.8 50.2 ± 8.6 74.2 ± 5.3 44.4 ± 22.1 58.0 ± 13.3 77.8 ± 13.3 37.6 ± 7.9

ON

fo

C

NQ 3.7 55.5 ± 19.6

fo

2.3

13.8 ± 11.1 NQ

IB

25.2 ± 6.2

9.2 ± 4.2 38.3

23.6 ± 8.9

EF

c

a

N O T E : O A = hydrophobic acid ( X A D - 8 ) ; O B = hydrophobic base ( X A D - 8 ) ; O N = hydrophobic neutral ( X A D - 8 ) ; I B = hydrophilic base ( A G MP-50); E F = final effluent (solvent extraction); N Q = found but not quantitated. Three values. Two values. Four values. S O U R C E : Reproduced with permission from reference 11. Copyright 1983 Elsevier.

88.1 ± 6.5

1.8 ± 1.5'

fe

OA

Compound

Mean Recovery ± Standard Deviation

Table II. Average Percent Recovery of Model Compounds from Resin Scheme


476

O R G A N I C P O L L U T A N T S IN

WATER

Table III. Percent Recovery of Model Compounds on Carbopack Β without Inorganic Salts and at pH 7


Percent Recovery Compound

Desorbed from GCB

Extracted from Water after GCB

2,4-Dichloro phenol Quinoline Isophorone 1-Chlorododecane 2,4'-Dichlorobiphenyl 2,2' ,5,5'-Tetrachlorobiphenyl Anthraquinone Bis(2-ethylhexyl) phthalate Phenanthrene Caffeine Furfural Methyl isobutyl ketone

115.2 97.5 16.3 51.2 48.6 54.1 92.1 51.1 114.0 92.1 NF 6.7

NF NF 92.4 NF 0.9 3.7 NF 64.3 NF NF 26.0 65.5

N O T E : N F indicates not found. S O U R C E : Reproduced with permission from reference 11. Copyright 1983 Elsevier.

primarily in the hydrophobic neutral fraction, whereas 5-chlorouracil and quinaldic acid were found at very l o w concentrations in the hydrophilic base fraction (see Table II). The behavior of quinaldic acid confirms the findings of Leenheer and H u f f m a n (16), who used test solutions spiked at the milligrams-per-liter level through a similar frac tionation scheme. Quinaldic acid was recovered in the hydrophilic base fraction after the test solution had gone through the X A D - 8 column under acidic and alkaline conditions. This result supported the sugges tion of the amphoteric behavior of this compound. C h l o r o f o r m could not be detected in the hydrophobic neutral fraction probably because it was lost b y volatilization. That several model organic compounds were only partially or incompletely retained b y the resins prompted us to investigate the use of Carbopack Β as an alternative or complementary adsorbent. Test solutions without humic acids were used to verify the sorptive-desorptive behavior of several model compounds under the experimental condi tions proposed by Bacaloni et al. (8), except that the compounds were desorbed with methylene chloride. The results of duplicate experiments are given in Table III. Isophorone and M I B K were not effectively retained b y Carbopack B, whereas bis(2-ethylhexyl) phthalate was almost equally distributed between the aqueous phase and the carbon. The relatively poor recovery of 1-chlorododecane, 2,4'-dichlorobiphenyl, and 2,2',5,5'-tetrachlorobiphenyl may be ascribed to sorptive losses onto reservoir glass w a l l , whereas furfural may be inefficiently



22.

GIABBAI E T A L .


477

desorbed. Phenanthrene, quinoline, anthraquinone, and, in particular, caffeine and 2,4-dichlorophenol were quantitatively recovered. In consideration of the results presented in Tables II and III, it was decided to integrate the resins and carbon columns in a single scheme as shown in Figure 1. The test solution was adjusted to p H 7 i m mediately following the A G M P - 5 0 column and processed through the Carbopack Β column. The results f r o m one experiment conducted under these conditions practically confirmed the overall process perfor mance anticipated f r o m the individual experiments (see Table I V ) . Trimesic acid, stearic acid, and humic acids were found in the hydrophobic acid fraction; quinoline was primarily quantitated in the hydrophobic base fraction, whereas a small amount of it was also detected in the hydrophobic neutral fraction. Isophorone, biphenyl, 1-chlorododecane, 2,6-di-ferf-butyl-4-methylphenol, 2,4'-dichlorobiphenyl, 2,2'5,5'-tetrachlorobiphenyl, anthraquinone, and phenanthrene were re-

Table IV. Average Percent Recovery of Model Compounds from Integrated Adsorption Scheme Percent Recovery Compound

OA

Stearic acid Trimesic acid 2,4-Dichlorophenol Quinaldic acid Isophorone Biphenyl 1 -Chlor ododecane 2,6-Di-terf-buryl-4methylphenol 2,4'-Dichlorobiphenyl 2,2\5,5'-Tetrachlorobiphenyl Anthraquinone Phenanthrene Bis(2-ethylhexyl) phthalate Glucose (NA) Furfural Quinoline 5-Chlorouracil Caffeine Glycine Humic acids Chloroform Methyl isobutyl ketone

31.3 39.7

OB

ON

75.6 66.8 40.1

IB

GCB

11.6 NQ

23.1

49.3 70.1 55.6 62.3 60.1 39.1 31.4

NF 3.5 11.3

83.1

12.3 NF NQ 2.3 44.5

31.3

NF NF

N O T E : O A = hydrophobic acid (XAD-8); OB = hydrophobic base (XAD-8); O N = hydrophobic neutral (XAD-8); IB = hydrophilic base; G C B = Carbopack B; N Q = found but not quantitated; Ν A = not analyzed; N F = not found.



478


covered exclusively i n the hydrophobic neutral fraction. Bis-(2-ethylhexyl) phthalate was primarily quantitated i n the hydrophobic neutral and Carbopack Β fractions and partially found i n the final effluent. As expected, caffeine was primarily recovered i n the Carbopack Β fraction, although a small amount was also detected i n the hydro phobic neutral and hydrophilic base fractions. Glycine, quinaldic acid, and 5-chlorouracil were detected i n the hydrophilic base fraction. Furfural was quantitated only i n the final effluent fraction. This result confirmed that neither the resins (i.e., X A D - 8 , A G MP-50) nor the carbon (i.e., Carbopack B) was able to effectively isolate it f r o m the water solutions. T h e results of integrated experiments and those ob tained f r o m the separate resin and carbon experiments (see Tables II and III) allow us to conclude that the proposed isolation-fractionation scheme (see Figure 1) is effective for recovering 16 out of 22 model compounds under this study at recoveries ranging f r o m 30% to 90%. One of the major concerns over the use of synthetic resins for the isolation of trace organic compounds is the potential contamination of the isolated samples, w h i c h is a major limitation particularly when attempting to collect organic concentrates for biological testing. During these experiments, the G C - F I D trace of the fractions generated f r o m the resin scheme revealed the presence of organics other than those of the selected model compounds. T h e hydrophobic neutral fraction was the relatively more contaminated. The bulk of the impurities appeared to be i n small quantities, except for t w o or three major ones, whose amount was comparable to that of the recovered model compounds. Attempts to confirm the origin of the contaminants were pursued b y G C - M S analysis of each isolated fraction and of the methylene chloride extract of a similar aliquot of the test solution. A list of the tentatively identified contaminants is shown i n the box. The detection of several compounds of high volatility in both resin fraction and solvent extract can be ascribed to contributions f r o m the lab environment (e.g., refrigerator for stock solution storage) and/or to the humic acid solids used to prepare the test solution. Chlorocyclohexene is an impurity commonly found i n the best grade methylene chloride commercially available, whereas phthalates are widespread contaminants because of their large use as plasticizers.

Conclusions A n isolation-fractionation scheme for the separation of trace organic solutes f r o m natural and drinking waters has been developed. This process involves the separation of a number of organic solutes into several fractions on the basis of their sorptive characteristics onto different adsorbents under varying p H conditions. The specific adsor-


22.

GIABBAI E T AL.



Tentatively Identified Artifact Contaminants from Resin Fractionation Scheme

ON Chlorocyclohexene Phenol 5-Amino-2,4-(l//,3H)-pyrimidinedione N,4-Dimethylbenzenesulfonamide Phthalate Phthalate Bromoform Xylene Ethylbenzene Chlorobenzene 4- Methyl-3-pen ten-2-one Dibromochloromethane EF l-(4-Hydroxyphenyl)ethanone Dichlorocyclohexane Chlorocyclohexanol Phenol Chlorocyclohexene Tetrachloroethane Bromoform Ethylbenzene Chlorobenzene 3-Methylenepentanone Dibromochloromethane Solvent-Extracted Test Solutions, pH 2 and 10 Chlorobenzene 1- Cyclohexene 2- Cyclohexen-1 -one Trichloropropene Chlorocyclohexanol Phthalate Phenol Toluene Trichloroethane Ethylbenzene Xylene Bromoform N O T E : O N = hydrophobic neutral and E F = final effluent.


479


480


bents evaluated include Amberlite X A D - 8 and A G M P - 5 0 resins and Carbopack Β G C B . O f the 22 m o d e l compounds used in this study, 16 have been separated at recoveries varying f r o m 30% to 90%. The recovery of the model compounds on X A D - 8 resin appears to be controlled b y their water solubility properties, except for the highvolatile compounds (i.e., chloroform), which may be lost b y volatiliza tion during sample handling. Highly polar solutes, present as cations in acidic water solutions, can be effectively recovered on A G M P - 5 0 . Meanwhile, nonionic solutes, w h i c h have water solubilities not suitable for adsorption on X A D - 8 and a strong affinity for the graphite structure of Carbopack B, are effectively recovered on Carbopack B. Still, the poor recovery of several compounds cannot be fully explained, and further experiments are required to elucidate the fate of these com pounds on the scheme. When the samples and resins were handled properly, contaminants introduced throughout the isolation-fractionation scheme were found to be m i n i m a l . Therefore, it is felt that the proposed process can be properly scaled up to handle large quantities of water for the prepara tion of concentrates for biological and chemical characterization. However, because several classes of organic compounds cannot be recovered effectively, the investigation of other supplemental isola tion-concentration methods is warranted. F o r example, the highly volatile purgeable organic compounds (i.e., chloroform, M I B K ) may first be analytically identified and quantitated and then spiked at a level that w o u l d be expected in the concentrate for the toxicologic study. Other methods, such as reverse osmosis or freeze-drying processes, can be used as an integral part of the proposed isolation-fractionation scheme to concentrate the highly polar, water-soluble compounds (e.g., glucose, furfural).

Acknowledgments This research was supported b y the Health Effects Research Laboratory of the U.S. Environmental Protection Agency under Contract N o . 68-03-3000. The excellent technical assistance of Z . Geskin, P. M a y , B. Ghosh, and J. S. K i m is gratefully appreciated.

Literature Cited 1. Lin, D. C . K.; Melton, R. G . ; Kopfler, F. C.; Lucas, S. V. In Advances in the Identification and Analysis of Organic Pollutants in Water; Keith, L. H., Ed.; Ann Arbor Science: Ann Arbor, MI, 1981; Vol. 2, pp 861-906. 2. Tabor, M. W.; Loper, J. C.; Barone, K. In Water Chlorination: Environmen tal Impact and Health Effects; Jolley, R. L.; Brungs, W. Α.; Cumming,


22.

3. 4.


5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

GIABBAI E T A L .


481

R. B.; Jacobs, V. Α., Eds.; Ann Arbor Science: Ann Arbor, MI, 1980; Vol. 3, pp 899-913. Jolley, R. L. Environ. Sci. Technol. 1981, 15(8), 874-880. Baird, R.; Cute, J.; Jacks, C.; Jenkins, R.; Neisess, L.; Scheybeler, B.; Van Sluis, R.; Yanko, W. In Water Chlorination: Environmental Impact and Health Effects; Jolley, R. L.; Brungs, W. Α.; Cumming, R. B.; Jacobs, V. Α., Eds.; Ann Arbor Science: Ann Arbor, MI, 1980; Vol. 3, pp 925-935. Van Rossum, P.; Webb, R. G. J. Chromatogr. 1978, 150, 381-392. Leenheer, J. A. Environ. Sci. Technol. 1981, 15(5), 1578-1587. Keith, L. H. In Advances in the Identification and Analysis of Organic Pollutants in Water; Keith, L. H., Ed.; Ann Arbor Science: Ann Arbor, MI, 1981; Vol. 2, pp 1165-1170. Bacaloni, Α.; Goretti, G.; Lagana, Α.; Petronio, B. M.; Rotatori, M. Anal. Chem. 1980, 52, 2033-2036. Grob, K. J. High Resolut. Chromatogr. Chromatogr. Commun. 1980, 3, 493. Giabbai, M.; Shoults, M.; Bertsch, W. J. High Resolut. Chromatogr. Chromatogr. Commun. 1978, 1, 277. Giabbai, M.; Roland, L.; Ghosal, M.; Reuter, J. H.; Chian, E. S. K. J. Chromatogr. 1983, 279, 373-382. Schlenk, H.; Gilerman, J. L. Anal. Chem. 1960, 32, 1412. Burleson, J. L.; Peyton, G. R.; Glaze, W. H. Environ. Sci. Technol. 1980, 14(11), 1354-1359. Malcolm, R. L.; Thurman, Ε. M.; Aiken, G. R. In Trace Substances in Environmental Health; Hemphill, D. D., Ed.; University of Missouri: Columbia, MO, 1977; Vol. 11, pp 307-314. Thurman, Ε. M.; Malcolm, R. L.; Aiken, G. R. Anal. Chem. 1978, 50(6), 775-797. Leenheer, J. Α.; Huffman, E. W. D. J. Res. U.S. Geol. Surv. 1976, 6, 737-751.

RECEIVED for review August 14, 1985. ACCEPTED A p r i l 7,

1986.


Evaluation of an Integrated Adsorption Method for the Isolation and

Recommend Documents