Isolation of Organic Compounds Present in Water ... - ACS Publications

---

23

Downloaded by FUDAN UNIV on January 24, 2017 | http://pubs.acs.org Publication Date: December 15, 1986 | doi: 10.1021/ba-1987-0214.ch023

Isolation of Organic Compounds Present in Water at Low Concentrations Using Supercritical Fluid Carbon Dioxide Daniel J. Ehntholt, Christopher Eppig, and Kathleen E. Thrun Arthur D. Little, Inc., Cambridge, MA 02140 The use of supercritical fluid carbon dioxide to extract low levels of organic substances from water was investigated for 23 different compounds. In general, compounds that were volatile and/or not highly soluble in water were readily extracted under the conditions used. Compounds of higher water solubility did not show evidence of extraction. In addition, those materials that tended to precipitate or form more soluble species under acidic conditions were not extracted. THE

U S E O F B I O L O G I C A L T E S T S is one approach to understanding and

evaluating the possible toxicological effects of the consumption of organic substances found in drinking waters. Many of these tests using experimental animals or organisms require concentration levels of the organic compounds that are significantly higher than those normally found i n drinking waters. Although hundreds of organic compounds have been identified and quantified in samples of natural waters, much of the organic matter present cannot readily be characterized b y using currently available analytical protocols. Without such prior qualitative and quantitative identification of the substances, they cannot b e purchased or synthesized for use i n the preparation of the concentrated solutions required for health-effects testing. Direct concentration of the organic materials from aqueous samples offers an attractive alternative that circumvents the analytical problems associated with the identification and quantification of w i d e varieties of species present at trace levels. A number of techniques have been studied for their use i n effecting such concentrations. These have 0065-2393/87/0214/0483$06.00/0 © 1987 American Chemical Society

Suffet and Malaiyandi; Organic Pollutants in Water Advances in Chemistry; American Chemical Society: Washington, DC, 1986.

Downloaded by FUDAN UNIV on January 24, 2017 | http://pubs.acs.org Publication Date: December 15, 1986 | doi: 10.1021/ba-1987-0214.ch023

484

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

included the use of reverse osmosis, solid sorbents, and l i q u i d - l i q u i d extractions (I, 2). Serious problems may, however, be encountered in the use of concentrated solutions prepared b y these methods because of inadvertent contamination of the sample. F o r example, membrane techniques may introduce impurities f r o m the membrane and may not selectively isolate organic substances f r o m inorganic species. Collection of organics on sorbents followed b y recovery with organic solvents also poses a number of problems. Concerns have been expressed over the large blank contributions of resins, the possible interactions of the organic substances concentrated with the solvents (or impurities present in them) used for desorption, and the presence of traces of solvent in the prepared sample. The l i q u i d - l i q u i d extraction techniques using organic solvents yield concentrations of organic substances in media that may be undesirable for animal feeding studies. F o r example, immediate concern can be expressed about the use of benzene or the halogenated one- and two-carbon compounds, which are k n o w n or suspect carcinogens, if long-term biological tests are to be performed. This study has examined the possible use of supercritical fluid carbon dioxide for the concentration and/or isolation of specified organic compounds present in waters at trace levels. This type of direct extraction using a nontoxic, nonhazardous solvent such as carbon dioxide represents a new concept for extracting trace levels of organic compounds f r o m water. Solubility phenomena in supercritical fluids were reported b y Hanney and Hogarth (3) as early as 1879. They found that inorganic salts such as cobalt chloride and potassium iodide could be dissolved in supercritical ethanol and ether. Furthermore, they found that the solubility level increased as the pressure increased. In the early 1900s, Bucher (4) studied the solubilities of a number of organic materials in supercritical carbon dioxide. His results showed that the concentration of organic species such as naphthalene, phenanthrene, phenols, and other aromatics dissolved in supercritical carbon dioxide was many times that which w o u l d be expected f r o m the normal increase in vapor pressure due to external pressure. Other supercritical fluid solubility studies of the early 1900s were directed to similar considerations of solution thermodynamics, multiphase equilibria, etc. Booth and B i d w e l l (5) presented an excellent review of the developments during this period. D u r i n g the 1940s, a large amount of solubility data was obtained b y Francis (6, 7), w h o carried out measurements on hundreds of binary and ternary systems with liquid carbon dioxide just below its critical point. Francis (6, 7) found that liquid carbon dioxide is also an excellent solvent for organic materials and that many of the compounds studied were completely miscible. In 1955, T o d d and E l g i n (8) reported on phase equilibrium studies with supercritical ethylene and a number of

Suffet and Malaiyandi; Organic Pollutants in Water Advances in Chemistry; American Chemical Society: Washington, DC, 1986.

23.

E H N T H O L T ET AL.

Isolation Using Supercritical Fluid CO2

485

Downloaded by FUDAN UNIV on January 24, 2017 | http://pubs.acs.org Publication Date: December 15, 1986 | doi: 10.1021/ba-1987-0214.ch023

low-vapor-pressure organic materials such as fatty acids and high molecular weight alcohols. They found, as d i d previous investigators, that the solubility levels of the organic species were orders of magnitude higher than those predicted b y vapor pressure considerations. Their findings led them to write "The magnitude . . . of solubility . . . is sufficient to consider the gas as an extracting medium, that is, fluid liquid or fluid solid extraction, analogous to l i q u i d - l i q u i d extraction and leaching . . .thus, compression of a gas over mixture of compounds could selectively dissolve one compound, permitting it to be removed from the mixture." This study was the first published reference to potential extraction process applications of solubility in supercritical fluids. A few years later, E l g i n and Weinstock (9) reported that a number of organic-water mixtures could be separated into organic-rich and water-rich phases b y using supercritical ethylene, and they presented process concepts for separating such mixtures. Since T o d d and Elgin's paper (8) in 1955, descriptions of a number of process applications of supercritical fluid solubility have appeared in the literature. M u c h of the effort reported has been directed to the extraction of edible (JO) and essential (II) oils and other food and beverage products such as spices (12), coffee (13), and hops (14) using either supercritical or near-critical liquid carbon dioxide. The attributes of carbon dioxide, such as its l o w cost and absence of safety hazards and toxicity problems, were ideally suited for food applications. In the m i d - and late-1960s, some developmental activity was directed toward supercritical fluid chromatography (15) and to extractions of fuels, namely, supercritical fluid extraction of coal (16), petroleum (17), and lignite (18). Starting in about 1975, developmental efforts at a number of industrial and academic laboratories increased markedly in both the U n i t e d States and E u r o p e . T h e resurgent research activity was motivated b y a number of factors: 1. Increased scrutiny of certain industrial solvents because of associated health and safety problems. 2. Increasing costs of traditional but energy-intensive separation processes such as distillation and evaporation. 3. Increasingly stringent pollution control legislation that increased costs of traditional extraction processes. 4. Identification of certain key areas in w h i c h supercritical fluid processing could be technically, as w e l l as economically, superior to traditional separation processes. Supercritical fluid separation processes operate at pressures ranging f r o m 1000 to 4000 lb/in. , pressures that might be considered high, especially in the foods and essential oils industries. However, because of the factors just listed, supercritical fluid extraction has become eco2

Suffet and Malaiyandi; Organic Pollutants in Water Advances in Chemistry; American Chemical Society: Washington, DC, 1986.

Downloaded by FUDAN UNIV on January 24, 2017 | http://pubs.acs.org Publication Date: December 15, 1986 | doi: 10.1021/ba-1987-0214.ch023

486

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

nomically attractive irrespective of the pressure requirements. Efforts to date have resulted i n the development of several supercritical fluid processes in use throughout the w o r l d . Several large pilot plants for coal (19), coal ash (20), and asphalt (21) separation are in operation i n the United States, U n i t e d K i n g d o m , and Russia. T w o commercial plants came on stream i n 1979 for the extraction of beverage products: one i n Germany for coffee decaffeination (22), and the other in Australia for hops extraction (23). Theoretical and practical efforts leading to several of these developments were summarized at a symposium devoted to "Extraction with Supercritical Gases" (22). These efforts and others suggested there might be process advantages for using supercritical fluid carbon dioxide to extract l o w concentrations of organics f r o m water. T h e U . S . Environmental Protection Agency ( U S E P A ) therefore sponsored the work described herein, which was designed to evaluate such a process; the work was initiated i n late 1980 and completed i n 1983. Twenty-three organic substances were selected as representative of classes of compounds usually encountered in drinking waters. In addition, calcium and sodium salts were added to the spiked aqueous samples (as well as method blanks) to simulate i n organic salt concentrations found i n Cincinnati drinking water. L e a d nitrate was also studied to determine whether supercritical fluid carbon dioxide might extract and/or concentrate metal salts. The ultimate goal of this program was the study of extraction feasibility for large volumes (—500 L ) of drinking waters. However, to facilitate sample preparation steps and subsequent analyses, initial supercritical fluid carbon dioxide extraction studies were conducted b y using small groups of similar compounds. These studies used 400-mL sample volumes. Subsequent 10-L extractions were run with aqueous samples containing all 23 organic substances as well as calcium and sodium salts present at the levels of interest i n this effort.

Experimental Most of the extraction studies carried out under this contract were conducted on 400-mL aqueous samples in a stainless steel extractor (extractor volume was approximately 600 mL) operated at about 2500 lb/in. (i.e., 173 bar) and 45 °C. Supercritical conditions are achieved for carbon dioxide at pressures >1070 lb/in. (i.e., 73.8 bar) and temperatures >31.1 °C. In our tests, approximately 300 standard liters of C O 2 was typically passed through the aqueous solutions into the traps. The trapping system usually consisted of a set of three sequential glass U-tubes maintained at —76 °C by a dry ice-acetone bath. Operation at this temperature precludes clogging by solid C O 2 but may be responsible for the loss of some extracted organic materials, as noted later. A scaled-up series of 10-L extractions were also carried out during this program. The apparatus used was similar to that used in the small-scale work but had an internal volume of 15 L. The traps were stainless steel impingers having a volume capacity of approximately 1 L. 2

2

Suffet and Malaiyandi; Organic Pollutants in Water Advances in Chemistry; American Chemical Society: Washington, DC, 1986.

Downloaded by FUDAN UNIV on January 24, 2017 | http://pubs.acs.org Publication Date: December 15, 1986 | doi: 10.1021/ba-1987-0214.ch023

23.

E H N T H O L T E T AL.

Isolation Using Supercritical Fluid CO2

487

The experimental apparatus used to perform the supercritical fluid carbon dioxide extractions is shown diagrammatically in Figure 1. Carbon dioxide provided from supply cylinder 1 is compressed by diaphragm compressor 2 and heated to the desired extraction temperature in heat exchanger 3. The pressurized, temperature-adjusted, carbon dioxide feed flows through the highpressure fluid inlet line 4 to vessel 6, which contains the aqueous solution to be extracted. The extraction vessel is wrapped with electrical heating tape to regulate the extraction temperature, which is measured with thermocouple 7. The supercritical carbon dioxide extract stream is passed from the extractor vessel outlet through pressure reduction valve 8, where the pressure is reduced to atmospheric pressure and the extracted organic substance is precipitated in collection device 9. The atmospheric pressure carbon dioxide then flows from the collection device through a rotameter 10 and dry gas meter 11, which measure C O 2 flow rate and total volume, respectively, to the vent 12. To enhance the CU2-aqueous phase interfacial area and facilitate contact by dispersion of the C O 2 as fine bubbles, a plug of silanized glass wool was placed in the bottom of the extraction vessel. After charging the vessel with 400 mL of aqueous feedstock solution, the vessel was slowly pressurized to the extraction pressure and simultaneously heated to the desired temperature. Carbon dioxide was then passed through the aqueous phase at a velocity of slightly more than 10 cm/min (about 10 standard liters/min at 1 arm and 70 °F). After a predetermined amount of carbon dioxide (typically 300 standard liters) flowed through the sample, the system was depressurized and the extracted aqueous raffinate (stream) was drained through valve 5 into a collection vessel. The extractor and all lines were of stainless steel construction; traps for the smallscale extractions and containers for feedstock and raffinate were all glass.

Results and Discussion Analytical Methods. Twenty-three organic substances were selected for study during this effort. These compounds and the concentration levels at which they were investigated in this program are shown in Table I. All of the aqueous solutions for evaluation by using supercritical fluid C 0 were prepared by spiking a small aliquot of the organic compounds dissolved in acetone into a distilled, deionized, water sample containing 70 ppm of N a H C 0 , 120 ppm of CaS0 , and 47 ppm of CaCl H 0. Special care must always be exercised in the study of parts-perbillion concentrations of organics in water to ensure minimal losses due to sample degradation, adsorption or absorption to process materials, and other similar losses. These issues were addressed by dividing the measurement procedures into two parts: (1) sample preparation and (2) analytical method, or finish. Because well-defined sample preparation steps were not available from the literature for the quantitative determination of parts-per-billion concentration levels of most of the model organic compounds in water, a considerable amount of effort was placed on the development of appropriate procedures for such measurements. In particular, each method was developed with the in ent to have a procedure that could verify the presence of appropriate concen2

3

2

4

2

Suffet and Malaiyandi; Organic Pollutants in Water Advances in Chemistry; American Chemical Society: Washington, DC, 1986.

Suffet and Malaiyandi; Organic Pollutants in Water Advances in Chemistry; American Chemical Society: Washington, DC, 1986.

Figure 1. Schematic representation of supercritical fluid extraction apparatus.

Downloaded by FUDAN UNIV on January 24, 2017 | http://pubs.acs.org Publication Date: December 15, 1986 | doi: 10.1021/ba-1987-0214.ch023

23.

E H N T H O L T E T AL.

489

Isohtion Using Supercritical Fluid C0

2

Table I. Organic Substances Selected for Study

Group 1

Downloaded by FUDAN UNIV on January 24, 2017 | http://pubs.acs.org Publication Date: December 15, 1986 | doi: 10.1021/ba-1987-0214.ch023

2

3

4 5

6 7 8 9 10 0

Compound 1-Chlorododecane 2,2' ,5,5'-Tetrachlorobiphenyl Biphenyl Bis(2-ethylhexyl) phthalate 2,4'-Dichlorobiphenyl Crotonaldehyde Furfural Isophorone Methyl isobutyl ketone Anthraquinone Quinoline Caffeine 2,4-Dichlorophenol 2,6-Di-terf-butyl-4-methylphenol Quinaldic acid Trimesic acid Stearic acid Glucose Glycine Chloroform Phenanthrene 5-Chlorouracil Humic acid

Concentration Level

Sample Preparation and

Analysis

(m/L) 5 5 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 1 50 2000

0

microextraction, G C - F I D

D N P H derivatization, HPLC-UV

microextraction, G C - F I D

microextraction, G C - F I D sorbent extraction, C H N , GC-FID 2

2

evaporation, derivatization, GC-FID purge and trap, G C - E C D H P L C - U V-fluorescence HPLC-UV HPLC-UV 0

Abbreviations are defined in the text.

trations of the organics in the feedstock and also monitor their concentrations after extraction by carbon dioxide, that is, to quantify concentrations of organics in the raffinate (effluent aqueous stream), which might be as low as one-fiftieth of the starting concentration level. For some groups of compounds, this effort involved an extensive study of sample preparation steps. However, the goal of developing accurate and reproducible methods for studying the concentrations of the model compounds in water was met for the compounds in Table I. Table I identifies the sample preparation and analytical technique used to quantify each compound in this study. To accurately determine the levels of compounds present in the aqueous solutions before and after C 0 extraction, a sample prepartion step involving concentration of the compounds was necessary for most samples. In the case of Groups 1, 2, 3, 4, and 8, microextraction techniques with organic solvents were used (24). For Group 5 acids, resin concentration was used. Group 6 was concentrated by evaporation; Group 7 (chloroform), Group 9 (5-chlorouracil), and Group 10 (humic acids) samples were analyzed as received. Derivatization was used to enhance the detection limits for three of the groups. The formation of 2

Suffet and Malaiyandi; Organic Pollutants in Water Advances in Chemistry; American Chemical Society: Washington, DC, 1986.

Downloaded by FUDAN UNIV on January 24, 2017 | http://pubs.acs.org Publication Date: December 15, 1986 | doi: 10.1021/ba-1987-0214.ch023

490

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

2,4-dinitrophenylhydrazone (DNPH) derivatives of the Group 2 aldehydes and ketones permitted their determination by an H P L C method after microextraction. Methyl esters of the Group 5 acids were formed b y using diazomethane and were subsequently detected b y gas chromatography-flame ionization detection ( G C - F I D ) . Glucose and glycine (Group 6) were quantified after treatment with a hydroxylamine hydrochloride solution in pyridine, followed b y N-trimethylsilylimidazole. The trimethylsilyl-glucose-oxime derivative and the glycine-trimethylsilyl derivative thus formed were analyzed b y using G C - F I D methods. Analytical methods for monitoring the compounds were developed or m o d i f i e d to permit the quantification of all 23 compounds of interest. As noted earlier, the compounds were initially studied in small-scale extractions b y groups. This approach assured minimal interferences in the analyses conducted during the initial supercritical fluid carbon dioxide extractions. Table II summarizes the data on the recovery of organics f r o m aqueous samples containing the compounds of interest at concentration levels listed in Table I when the sample preparation techniques and analytical methods described were used. For each experimental run, blank and spiked aqueous samples were carried through the sample prepration and analytical finish steps to ensure accurate and reproducible results. Analyses of sodium, calcium, and lead content were also conducted on selected samples b y using standard atomic abTable II. Summary of Recovery Data (Microextraction-Derivatization Techniques)

Analyte

No. of Experiments

Mean Recovery

Standard Deviation

Coefficient of Variation

2,4'-Dichlorobiphenyl 2,2' ,5,5'-Tetrachlorobiphenyl Bis(2-ethylhexyl) phthalate 1-Chlorododecane Biphenyl Furfural Crotonaldehyde Isophorone Methyl isobutyl ketone Anthraquinone Quinoline Caffeine 2,4-Dichlorophenol 2,6-Di-terr-butyl-4-methylphenol Quinaldic acid Trimesic acid Stearic acid Chloroform Phenanthrene^

6 6 6 3 6 3 3 3 3 3 4 4 7 7 3 3 3 12 6

96 87 85 131 98 61.5 83.7 82.4 63.1 66.7 70.8 86.1 86 76 9.1 95.8 78.1° 80 69.2

18.1 17.9 22.5 7.8 11.8 16.5 12.0 10.7 28.1 19.3 19.9 10.6 19.8 10.5 0.7 3.3 8.8

18.8 20.6 26.5 6 12.0 26.8 14.3 12.9 44.4 66.7 28.1 12.3 23.1 13.9 7.7 3.4 11.2

— 9.7

14.0

e

e

N O T E : All values are percentages. Recovery from X A D - 7 resin. Data are taken from reference 26.

a

b

Suffet and Malaiyandi; Organic Pollutants in Water Advances in Chemistry; American Chemical Society: Washington, DC, 1986.

—

Downloaded by FUDAN UNIV on January 24, 2017 | http://pubs.acs.org Publication Date: December 15, 1986 | doi: 10.1021/ba-1987-0214.ch023

23.

E H N T H O L T ET A L .

Isolation Using Supercritical Fluid CO2

491

sorption spectroscopic methods for inorganic cations present in drinking waters (25). A l l of the techniques developed were amenable to the analysis of the concentrated compounds in collection devices (traps). Because these compounds were expected to be present in neat form in the traps, it was not anticipated that the sample preparation aspect of the analysis w o u l d pose any difficulty (i.e., it was assumed that any adsorption to glass walls, etc., that might occur would affect only a proportionately very small amount of the collected sample). As discussed later, this situation d i d not prove to be the case for some of the compounds studied. In addition, the difficulties inherent in trapping trace quantities of organics in the effluent C 0 stream were not obvious during the early stages of this program. 2

Small-Scale Extractions of Organic Compounds. Table III details the experimental results obtained for the small-scale supercritical fluid carbon dioxide extraction of the organic compounds. The compounds investigated, nominal spiking levels, and number of experiments performed are listed in the first three columns. The mean recoveries for each of the three U-tube traps connected in series are then presented along with the total recoveries obtained f r o m all three traps. The quantity of c o m p o u n d recovered f r o m the raffinate (effluent) solution after C O 2 extraction is contained in the last column. Although four of the five G r o u p 1 compounds spiked into the aqueous samples could be recovered f r o m the traps, only 20$ to 31$ of the total mass of each compound could be accounted for when the amounts in the traps and raffinate were summed. Losses may be due to incomplete trapping because of the l o w , but measurable, vapor pressures of these compounds and the large volume of C O 2 passed through the traps. Results f r o m the two experiments conducted on the aldehydes and ketones are listed in Table III as separate sets of data to illustrate the care that must be exercised in conducting and evaluating these runs. Both C O 2 extractions were performed under similar conditions. H o w ever, in the second run, the U-tube traps were contacted with the 2,4-dinitrophenylhydrazine derivatizing solution for longer periods of time. This modification in the analytical procedure permitted higher total mass accountabilities in the second experiment, ranging f r o m 64.9? for isophorone to 28.7$ for methyl isobutyl ketone. The recoveries f r o m the raffinate for each of these compounds remained relatively constant. This result suggested that the trap recoveries in the first case were artificially l o w . In the case of the Group 3 compounds anthraquinone, caffeine, and quinoline, it seems likely that the l o w trap recoveries and high residual concentrations of quinoline and caffeine in the raffinate were due to the

Suffet and Malaiyandi; Organic Pollutants in Water Advances in Chemistry; American Chemical Society: Washington, DC, 1986.

Suffet and Malaiyandi; Organic Pollutants in Water Advances in Chemistry; American Chemical Society: Washington, DC, 1986. 50 50 50 50 50 50 50 50 50 50 50 1 50 50 50 50

Isophorone Methyl isobutyl ketone Quinoline Caffeine Anthraquinone 2,4-Dichlorophenol 2,6-Di-tert-butyl-4-methylphenol o-Bromophenol Stearic acid Quinaldic acid Trimesic acid Phenanthrene Stearic acid Quinaldic acid Trimesic acid 5-Chlorouracil

Standard deviations are given in parentheses. S C F denotes supercritical fluid.

50

Furfural

0

5 5 50 50 50 50

Concentration Level (μβ/L)

1-Chlorododecane 2,2' ,5,5'-Tetrachlorobiphenyl Biphenyl Bis(2-ethylhexyl) phthalate 2,4'-Dichlorobiphenyl Crotonaldehyde

Compound

1 1 1 1 1 1

2

2

2

f c

2

2

2

1 3 1 3 3 1 1 1 1 1 1 1 1 2 2 2 3(SCF C 0 ) 1 (liquid C 0 ) 3(SCF C 0 ) 1 (liquid C 0 ) 3(SCF C 0 ) 1 (liquid C 0 )

No. of Experiments 0 18.7 9.1 11.3 15.7 0.8 7.0 3.7 8.3 0 39.2 0 15.6 1.7 0 56.0 35.8 15.3 26.6 8.7 24.6 10.8 25 0 0 58 25 0 0 0

Trap 1 0 0 11.0 0 4.6 0 0.8 0 0.7 0 0.7 0 1.5 1.7 0 14.3 9.6 16.8 6.0 10.7 17.0 12.3 2.5 0 0 36 2.5 0 0 0

Trap 2 0 0 3.3 0 0 0.2 0 0.1 0 1.7 0.5 0 0.2 0 0 14.3 0 8.2 0 6.6 0 10.8 20 0 0 3 20 0 0 0

Trap 3

0

0 18.7 (18.0) 23.4 11.3 (3.6) 20.3 (2.6) 1.0 7.8 3.8 10.8 1.7 40.4 0 17.3 3.4 (3.7) 0 84.6 (38.3) 45.4 (15.0) 40.3 32.7 (3.0) 26.0 41.7 (15.6) 33.9 47.5 0 0 97 47.5 0 0 0

Total

Mean Trap Recoveries (%)

Table III. Small-Scale (400 m L of Aqueous Samples) Extractions

Downloaded by FUDAN UNIV on January 24, 2017 | http://pubs.acs.org Publication Date: December 15, 1986 | doi: 10.1021/ba-1987-0214.ch023

20.7 12.0 (20.8) 3.8 15.4 (17.5) 8.5 (10.8) 25.1 31.0 43.4 22.3 17.8 24.5 7.4 11.4 46.1 (14.3) 81.4 (11.4) 21.4 (30.2) 28.0 (14.0) 13.4 0 6.8 31.6 (9.6) 19.5 22 85 0 (91) 0 22 85 91 96

Mean Raffinate Recoveries (%)°

23.

E H N T H O L T ET AL.

Isolation Using Supercritical Fluid CO2

493

l o w p H of the extracting media ( p H 3) and the resulting poor solubility of these nitrogenous compounds in the C O 2 effluent stream. Anthraquinone was recovered in good yields f r o m the same extractions. Table III also presents our data for the extraction of G r o u p 4 phenols f r o m aqueous solutions. The o-bromophenol was added as an internal standard when some initial recovery problems were noted for the 2,6-di-ter£-butyl-4-methylphenol; results for its extraction are also reported here. The three phenols show good recoveries in the traps and overall good mass recoveries. One experiment was conducted under liquid C 0 extraction conditions (temperature = 30 °C and pressure = 1500 lb/in. ) in an attempt to compare the relative efficiencies of the two states of C 0 for phenol extraction. Unfortunately, the phenols showed evidence of substantial breakthrough f r o m the trapping system. The experiment does, however, demonstrate that liquid C O 2 is also a good extractant for phenols present in water at parts-per-billion concentration levels. 2

Downloaded by FUDAN UNIV on January 24, 2017 | http://pubs.acs.org Publication Date: December 15, 1986 | doi: 10.1021/ba-1987-0214.ch023

2

2

G r o u p 5 acids present at the 50-ppb level in aqueous solutions containing the specified inorganic salts were also extracted. The feedstock solution ( p H 7.2) and the raffinate ( p H 4.5) were analyzed b y concentration on X A D - 7 resin, elution b y methanol, methylation b y diazomethane, and, finally, measurement of peak areas b y G C - F I D . Comparison of the peak areas of the raffinate with the peak areas in the original solution indicated a 78$ reduction for stearic acid, 15$ reduction for quinaldic acid, and 9$ reduction for trimesic acid. The three traps, including the glass w o o l plugs and connecting tubes, were also analyzed for acid content. Only stearic acid was detected. The G r o u p 6 compounds glucose and glycine were not tested at the small-scale level because previous work suggested that they w o u l d not be soluble in supercritical fluid C 0 . Additionally, chloroform was not tested at this level because it is undoubtedly extractable but w o u l d pose significant trapping problems because of its relatively high vapor pressure. The Group 8 compound phenanthrene was tested in a 1-ppb (micrograms-per-liter) solution and showed virtually complete recovery, as detailed in Table III. The G r o u p 9 compound 5-chlorouracil, on the other hand, d i d not exhibit any extraction. Recovery of the compound in the raffinate (at the 50-ppb level) was quantitative; this result indicated that little, if any, was extracted. Several extractions were also conducted on 2.0-mg/L humic acid solutions. These solutions were prepared b y dissolving a k n o w n quantity of humic acid (Fluka, further purified b y U S E P A pesonnel) in 0.20 M sodium hydroxide f o l l o w e d b y dilution with water to a 0.02 M sodium hydroxide solution. Subsequent neutralization to p H 7.0 with 0.100 M H C I and dilution with water containing the salts noted earlier gave a 2

Suffet and Malaiyandi; Organic Pollutants in Water Advances in Chemistry; American Chemical Society: Washington, DC, 1986.

494

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

2.0-mg/L humic acid solution for extraction studies. Three supercritical fluid carbon dioxide extractions were carried out o n this solution; no humic acid could b e detected in the traps. Table I V lists extraction conditions tried in these runs. Although the analyses of feedstock solutions showed the presence of 97.7$ to 104.7$ of the expected concentration levels of humic acids, analyses of raffinate solutions showed lower humic acid concentrations. The raffinate obtained after the C 0 extraction indicated that the humic acids were present at 39.4$ to 44.9$ of the feedstock levels (as measured b y highperformance liquid c h r o m a t o g r a p h y - U V spectroscopy ( H P L C - U V ) . This result suggested that the acidic conditions present in the extractor caused some precipitation and resultant loss of material. Cleaning of the extractor after these experiments indicated that a dark organic material had precipitated o n the walls of the extractor. In addition, an experiment was conducted o n the neat humic material (in the absence of water) to verify that the organic substance was not dissolved or carried over into the trapping system. In that case, approximately 500 m g of humic acid material was treated sequentially with 100 standard liters of carbon dioxide (2500 lb/in. at 45-50 °C) and 50 standard liters of carbon dioxide (4000 lb/in. at 38-41 ° C ) . There was no apparent collection of material in the traps (visually or gravimetrically), and there was only a very slight loss (