Factors affecting the extraction and analysis of polynuclear aromatic hydrocarbons in water

tinter Re~c~trch V,~I. I1~. pp. 2~1" to 212. P,:rgamon Press 19"6. Printed in Great Britain

FACTORS AFFECTING THE EXTRACTION AND ANALYSIS OF POLYNUCLEAR AROMATIC HYDROCARBONS IN WATER M. A. ACHESON.R. M. HARRISON*.R. PERRY'I"and R. A. WELLINGS Public Health Engineering. Imperial College. London S.W.7. U.K. (Received 10 January 1975) Abstract--Factors expected to affect the el~ciency of extraction of Polynuclear Aromatic Hydrocarbons (P.A.H.) from environmental water samples have been systematically investigated. Such factors include the initial concentration of P.A.H.. the presence of suspended solids and prolonged storage of the sample prior to analysis. Extraction efficiencies between 30 and 85Q~ have been found and these data make possible more accurate measurements of levels of P.A.H. in water. Analyses of environmental water samples have been performed using both Thin layer chromatography /TLC) and Gas-liquid chromatography (GLC). Accuracy of analysis was assessed by addition of P.A.H. to environmental water samples before extraction. Thin layer chromatography was found to be the better procedure for the compounds analysed, and avoided the need for purification of the solvent extract prior to analysis.

I. I N T R O D U C T I O N

2. D I S C U S S I O N A N D R E S U L T S

In any analysis of trace organic materials in water, prior extraction and concentration of the materials is normally essential. Such processes inevitably involve partial loss of the organic material, and hence affect the final analytical result. Consequently, as part of a broader study of the levels of polynuclear aromatic hydrocarbons (P.A.H.) in water, the efficiency of isolation of these compounds prior to analysis has been systematically investigated. Many studies of the levels of P.A.H. in raw, potable and waste waters have been reported (Borneff and Kunte, 1964; 1965; Samoilovich and Redkin, 1968), and although several authors have acknowledged the imperfect efficiency of the necessary processes of extraction, concentration and separation, no satisfactory study of the techniques employed has been reported. As a result, published data are unlikely to be as precise or reliable as is desirable in view of the concern over possible health hazards associated with these compounds (World Health Organization, 1964). In the work described in this paper, all stages of extraction and analysis have been examined and overall efficiencies estimated. Care has been taken to ensure that initial concentrations of P.A.H. in water accurately reflect those found in the aqueous environment, as variations in process efficiency with initial levels were anticipated. The effect of suspended solids, often present in environmental water samples, has been investigated and the analysis of both river water and motorway run-off water is reported.

Initially, three extraction procedures were considered. The first involved conventional extraction in a separating funnel, previously reported by Wedgwood and Cooper (1953; 1954; 1956) for the extraction of P.A.H. from water. Secondly, a continuous solvent extraction procedure was used, which although not reported in the literature for the extraction of P.A.H., was expected to give a high efficiency of extraction. The third method involved the use of an Ultra-Turrax (an ultrasonic mixer/homogeniser) as described by Borneff and Kunte (1964). In preliminary experiments, use of a separating funnel was found to give a low extraction efficiency compared to the other methods, and continuous extraction proved to be of comparable efficiency to the Ultra-Turrax, but was far more time consuming. Consequently the Ultra-Turrax method was chosen for further intensive investigation. Dichloromethane was the preferred extraction solvent because of its good solvent properties for P.A.H., its low b.p. and consequent ease of removal by distillation, and the infrequency of persistent emulsion formation when compared to some other solvents. Reproducible techniques with regard to the volumes of water and dichloromethane used, and the position and speed of the Ultra-Turrax were adopted. 2.1 Extraction procedures

2.1.1 Extraction from pure water. Highly purified water was dosed with P.A.H. and extracted with distilled dichloromethane using the Ultra-Turrax. Subsequent careful evaporation of the solvent, after separ* Present address: Department of Environmental ation of the dichloromethane layer, and addition of Sciences, University of Lancaster, Bailrigg, Lancaster, U.K. octacosane as an internal standard was followed by T To whom correspondence should be addressed. analysis by gas chromatography. A blank extraction 207

P.AH.

involving no added ~a~ used tO determine background organic materials in the water and extraction solvent, which were in general found to be extremely low in concentration. Only two P.A.H., pyrene and benzo(ghi)perylene were used in these studies and were selected as representative of lower and higher tool. wt P.A.H. commonly found in water. Figures 1 and 2 show the variation in extraction efficiency with initial concentrations of these two compounds in water. At higher concentrations etSciencies in the region of 80",, comparable to those reported by Borneff and Kunte (1969), are found. At lower initial concentrations, however, efficiencies may drop below 40°,,

2.1.2 Effect of suspended solids upon extraction efficiency. Raw and waste waters frequently contain substantial levels or suspended solids and adsorption of P.A.H. upon these solids may lead to a change in extraction efficiency. Fullers Earth (a highly adsorbent clay), purified by Soxhlet extraction, was added as a suspended solid, and after suspension in the purified water with the aid of the Ultra-Turrax, P.A.H. were added and the water extracted in the usual mannor. Both the variation in extraction efficiency with the level of suspended solids and the variation with P.A.H. concentration at a fixed level of suspended solids were investigated. The resultant efficiencies ~br extraction of pyrene and benzo(ghi)perylene are shown in Figs. 1-3. The reduced efficiency with increasing levels of suspended solids for both P.A.H. studied appears to indicate adsorption on the solids. with a consequent increased ability to remain in the aqueous suspension. The variations in extraction efficiency with solute concentration at a fixed level of suspended solids are not readily explained; in particular the higher extractions obtained for lower concentrations of pyrene with 100 mg 1- ~ of suspended solids than without added solids would not be predicted. Although Fullers Earth may not be entirely representative of all types of suspended solids encountered in raw and waste waters, its use does serve to illustrate that when P.A.H. are extracted from water

I

:oo

=< _J >.

a ~

~

a

i

50 !

~

4o

ud 20

0

o

o -' = 7

0 02

---1~mqt~suspenoeo -

sotids

--I

NO suspended solids I Prolongec~ m i x i n g , l O O m g t " s u s p e n d e d solids[ , P r o t o n g e c l m i x i n g no suspended solids

0.04

006

008

0 tO

0.12

Oq4

Benzo(ghi) perylene concentration, (FgC ~)

Fig. 2. Variation of extraction efficiency with the initial concentration of benzo(ghiJperyleneshowing the effects of suspended solids and prolonged mixing. samples containing suspended solids, efficiencies of extraction may be radically altered. 2.1.3 Effect oJprolonyed mixitu3. Samples of purified water were dosed with P.A.H. and stirred in the dark for 6 h prior to extraction by the usual procedure. Prolonged mixing and extraction were performed both with and without added solids. Results appear in Figs. 1 and 2. A lowered extraction efficiency was found relative to previous extraction experiments, but the addition of suspended solids did not affect the extraction: to within experimental error extraction efficiencies were the same with or without solids after 6 h mixing. These results may be explained either by adsorption of P.A.H. onto the glass of the mixing (and extraction) vessel, or degradation (chemical or biological) within the vessel. Two conclusions were drawn from these results and applied to the analysis of field samples. Firstly, it is essential to sample water directly into the extraction vessel, and secondly the extraction should be carried out as soon after sample collection as possible.

2./.4 Comparison with continuous solvent extraction. Continuous dichloromethane extraction for 36 and 72 h was compared with the above procedures. The results (Table 1) showed that extractions from pure water were very similar to those achieved using the

I ~oo

.<

t

I

I

Key o pyre'he 0 Banzo{ghi)

BO

Derylene 60

8

g

40

40

1

0

?

l~rolm,l~e.a

mixi~,no su~mded ~elld$

OI

Pyrene goncentratiotl,

0"2

03

F.g( "l

Fig. 1. Variation of extraction efficiency with the initial concentration of pyrene showing the effect of suspended solids and prolonged mixing.

1

~°I 0

I

~0

I

IO0 Susoen¢le~ soliCls,

I

I~0 m(:j l -~

200

Fig. 3. Effect of various levels of suspended solids upon the extraction efficiencyfor pyrene and benzo{ghi)perylene at initial concentrations ot"0-3 and 0-! /~g 1-t respectively.

Extraction and analysis of hydrocarbons in water Table 1. Comparison of extraction efficiencies l%t using continuous extraction and the Ultra-Turrax Suspended solids (ppm) 0 100 100

Ultra-Turrax P?r" [ll~hilP+ 84 68 68

Continuous extraction Timeth) Pyr* B!$hilP+

S2 53 53

36 36 72

:')3 27 3.1.

55 16 15

* Initial concentration 0.3 gg l- t. ~"Initial concentration 0. l ,ug 1- t. Ultra-Turrax, but in the presence of suspended solids the efficiency of extraction was reduced substantially. Again, no clear explanation of the effect of suspended solids can be put forward.

2.2 Analysis 2.2.1 Purification of the solvent extract. Solvent extracts of environmental water samples, unlike extracts of the highly purified water used in the above studies, contain substantial quantities of organic compounds other than P.A.H. These impurities were found not to affect adversely the analysis by TLC and fluorescence (vida infra), but made GLC analysis of P.A.H. extracts quite impossible. Hence, separation of extraneous organic compounds by the procedure of Hoffmann and Wynder (1960) was employed, and was found to produce a solution sufficiently purified to be readily amenable to GLC analysis. The efficiency of this procedure, which involves consecutive solvent-solvent partitions of P.A.H. from aqueous methanol into cyclohexane and thence into nitromethane, was examined by the use of standard mixtures of P.A.H. and was found to lie between < 10 and 839/0, depending upon the compound and the initial concentration (Table 2). At the higher initial concentration studied, in three determinations the GLC analysis failed to reveal perylene in significant quantities, although an additional GLC peak, eluted between octacosane and benzo(k)fluoranthene was apparent. This phenomenon is not readily explained. as both in the examination of this procedure at a lower initial perylene concentration and in the anal)'sis of field samples using this purification procedure, perylene was detected in quantities comparable to those found by TLC examination of an unpurified extract.

209

The procedure of Hoffmann and Wynder {1960) was not highly efficient and appeared to have two major drawbacks. The partition coefficient between cyclohexane and methanol for P.A.H. was such that four time-consuming extractions were needed to allow a high recovery. Further. the high temperature necessary for evaporation of the nitromethane encouraged loss of P.A.H. both by volatilisation and thermal degradation. Haenni. Howard and Joe 119621 have described the use of dimethylsulphoxide as a solvent suitable for extraction of P.A.H. from aliphatic solvents. It was incorporated in the procedure of Hoffmann and Wynder 119601 by use in place of nitromethane, and the extracted P.A.H. were isolated by subsequent dilution with water and back-extraction into cyclohexane, which was then evaporated. This procedure had a much improved efficiency of 84-110°,o (Table 2). but was only developed in the final stages of the work described. 2.22. TLC and GLC analysis of extracted P.A.H. Two techniques of analysis were employed. Firstly, a TLC procedure reported by Borneff and Kunte 11969) and recommended by the World Health Organization 11970: 1971) was employed after slight modification of the solvent system used for development of the plate. This method has the advantage that the P.A.H. are determined by fluorescence of the spots on the plate, without prior removal and associated losses (Kunte, 19671.The drawback of the technique is the imprecision of visual comparison of the size and intensity of the spots on the plate with those of standards. After development of the plate it was sprayed' with tetrachlorophthalic anhydride (Dubois, Corkery and Monkman, 19601 prior to fluorescence measurement, and this was found to improve greatly the sensitivity for benzo(a)anthracene, coronene, benzo(e)pyrene and chrysene. The compounds chosen for study and the detection limits for this method appear in Table 3. Although Sawicki et al. (1967) have discussed the relative merits of various methods for the determination of benzo(a)pyrene in airborne particulates, it appeared valuable to assess the sensitivity and scope of two methods, each with the low detection limit necessary for analysis of P.A.H, in water. For a range of P.A.H. found in environmental samples. The second method selected was G.L.C. (Bhatia, 1971;

Table 2. Efficiencies of solvent partition purification procedures Recovcr.~ 1",,)

Compound Fluoranthcne Pyrene Benzola~nthraeen¢ + chrytene Benzolk~uorandmn¢ Bcnzola)pyrtn¢ + Benzole)pyrcn¢

Peryh'ne I n d~,"not I ,'2.,3-cd Ipy r ene Itenzolghilperylcne

Coronene

Initial weight tug)

Nitromcthan¢ procedure

DMSO modification

Initial weight (.ugi

0.52 1.21 1.54 0.60 1.01 0.95 0.38 0.79 0.50

48 42 50 .54 36 24 49 45 44

92 94 90 9[ 95 84 97 91 93

10.4 _.'4.2 30.8 12.0 20.2 19.0 7.6 15.8 I0.0

Nitrom¢than¢ procedure 83 38 64 59 33 < I0 41 39 82

DMSO modification 96 96 92 94 99 90 100 90 110

2[L'+

\1-\.

\(

HES<,'-,, R \ l H,',RRISO'-,,R.

Table 3. Detection limits for anal>sis of P.A.H. D,-'rcctio n hmlt mg~ G LC FI D

Y LC tluore~cence

t ,,mpound Fiuor:mthcne {FD P?rene IP?r~ gcnzoLt kmth ra¢cne IBlalA~ (hr>senc iChrt Benzol k itiuoranthene I BI k IF) ll:nzola4p~renc B~.oP) {cnzo+clpyrene Ble!P Pcr'.lcne IPerj hld~oot 1,2.3-cd Ip? rene tiP) Bcnzolghltperylene I Blghi~PI ( oronene iCorl

5 5{) Or) '~ ) "¢ 5 3o II) 5 3 50

5 3 [/) l I) I j I0 II) 10 10 20

Lane, Moe and Katz, 1973) on balanced dual columns ( 3 ° 0 0 V 7 on Gas Chrom Q) with flame ionisation detection. In our hands this method did not resolve benzo(a)anthracene and chrysene, or the 3 benzofluoranthenes, and gave only partial resolution of benzo(alpyrene and benzo(e}pyrene. The detection limits (Table 3) are in a number of instances better than for the TLC procedure, although the latter method gives a more complete separation of the P.A.H. encountered in environmental samples. 2.2.3 Analysis offield samples. Samples of water originating from two types of source were analysed. Three samples of Thames river water and one sample collected from a sewer draining the M4 motorway were extracted with dichloromethane. Each sample was collected in duplicate, and in each case one pot-

PFP,M'*

and R \

V',iL~',,}'~

tion ~as dosed ~ith P.A.H. prior to extraction. One tenth of each extract ~:~s. after concentration, spotted directly onto a TLC plate and analysed by the abo~e procedure. Seven spots were readily identified in each undosed sample: perylene and the six compounds determined bv Borneff and Kunte (1969L Since no standard of benzolblfluoranthene was available, the remaining six compounds were determined by fluorescence and the results appear in Table 4. It is apparent that assuming a high efficiency of extraction, as would be anticipated from the earlier work. the added amounts of P.A.H. were accurately determined for the compounds measured. Increase of the measured levels in the undosed samples by a factor of approximately I00 75 to account Ibr losses in extraction should give a fair[? ztcct£1zttcestimate of P.AH. levels in the water samples. Gas-liquid chromatographic analysis was performed on the remaining dichioromethane extracts, after purification by the procedure of Hoffmann and Wynder (1960). Results {Table 5) show a less accurate measurement of added P.A.H., and indicated a generally lower precision than TLC analysis, probably as a result of the difficulty in the assessment of GLC peak areas in the presence of a variable background. In a small proportion of cases measurement of peak areas was rendered impossible by a high background, and no estimate of P.A.H. concentration was made. To account for the inefficiency of the extraction and

Table 4. T.L.C. analysis of water samples (#g I-t) (uncorrected for extraction efficiency) Albert Bridge*

Kew Bridge P.A.H.

P.A.H.

measured in Compound Fluoranthene Benzotk)fluoranthene Benzo~ a!pyrene

Perylene Indeno( 1.2.3.cd)pyrene Bcnzotghilperylen¢

undosed

P.A.H.

sample

added

0,11 0.06 O. I 0 0.03 O.0d 0.04

0.05 0,1~5 0,07 0. I 0 0.04 0.08

P.A.H.

Tower Bridge P.A.H.

0.16 0.08 O. 16 0.10 0.08 0.08

0.15 0.03 O.12 0.05 0.08 0.08

%14 Motorv, ay P.A.H.

measured in measured in measured in do~d undosed uodosed sample sample sample 0.27 0,0~ O 26 0. I0 0. I6 0.12

PAH

PA,H,

added

sample

sample

added

measured in dosed sample

01 I 0.13 0 13 O.20 0,08 0 17

0.44 025 O.52 (I.30 0.24 0.25

t.10 0.32 0 65 0.20 0.32 0.12

0. l l 0.13 0.13 0.20 0.08 0.17

1.10 0.50 0.98 0,40 0.48 0, [ 7

P.A.H.

measured in measured in dosed undoscd

P.A.H.

* This determination was not carried out in duplicate. Table 5. G.L.C. analysis of water samples (,ug 1- ~) (uncorrected for extraction efficiency} Albert Bridge*

Kew Bridge P.A.H.

P.A,H.

P.A.H.

M4 Motorway

Tov, er Bridge P.A.H.

P.A.H

measured in dosed sample

0.49 0.43

0.11 0.25

0,64 086

0.32

0.39

0.32

0.84,

0.13

0.68

0,4,9

0.13

1.06

0.13

0.21

033

0.57

O. 12 O. I I 0,03

0.20 0.08 0,17

0.43 0.33 0 lS

0.52 0.14 0.07

0.2 I ~ } 0,201 0.08 0.17

P.A.H. added

0.18 0.26

0.05 0.13

0.20 0.29

0.02 0,05

0. i8 0.23

0. I I O 25

0.2t 0 32

0 14

0.16

--*

0.27

0.33

0. 32

Benzo{j Ifluoranthene

0.24

0.06

0.52

0.15

0.43

Benzo(blfluoranthene Benzo(a)pyrelle B,anzc4e)pyrene Pervlene

0.21

0.11

0.36

--¢

0.04 0. I0 0.04

0, I0 O.Oa 0,08

0.12 0.22 0.08

--¢ 0.07 0.04

Fluorant hene

Pyrene Bcnzolalanthracene Chrysenc Benzoik)fluorant heoe

I ndeno{ 1,2.3-¢d )pyrene

Benzo{ghilpery lene

* This determination was not carried out in duplicate. t Not determined due to high background. ++Peaks not resolved.

P.A.H.

P.A.H. added

measured in measured in measured in dosed undosed undosed sample sample sample

measured in measured in undosed dosed sample sample

measured in undosed sample

Compourld

P.A.H.

P.A.H. added

l,OO~ 0.29 0.21

Extraction and analysis of hydrocarbons in water purification procedures. P.A.H. levels measured by G L C should be increased by a factor of approximately 100/40 to give the levels present in the water samples. For P.A.H. determined by both methods results are in tolerable accord, although clearly much scope remains for improvement in accuracy. The differences between results given by the two methods. however, are not large in view of the results of Bravo et al. (1970) who found discrepancies of approximately tenfold in levels of benzo(a)pyrene and benzo(k)fluoranthene in airborne particulates when determined by two different procedures, although the total quantity present was far greater than that encountered in these small water samples. Of the compounds used for calibration of the G L C procedure, only coronene was not present in the field samples at a measurable level. An increasing level of P.A.H. pollution of the River Thames as it passes through London is indicated. The T L C results (Table 4) show steadily increasing levels of P.A.H. from Kew Bridge (west London; 8 6 k m from the sea) to Albert Bridge (central London; 73 km from the sea) and to Tower Bridge (east London; 65 km from the sea). The levels show a distinct similarity to those measured by Borneff and Kunte (1964) in the River Rhine at Mainz. The motorway run-off sample contains high levels of P.A.H., and this finding is consistent with the observation of Borneff and Kunte (1965) that levels of P.A.H. in sewage increase substantially in wet weather periods, probably as a result of road run-off. 3. EXPERIMENTAL 3.1 Extraction of P.A.H. from pLirified water Water was purified by distillation and passage through 2 I-m columns of activated carbon at a rate of 0-25 1h- ft. After production of 20 I., which was mixed thoroughly, four aliquots, each of 51. were separated, one used as blank and three dosed with P.A.H. for extraction. Polynuclear Aromatic Hydrocarbons were added as a solution in dichloromethane, the total volume added being ca. 0.2 ml, easily miscible with the water (5 l.) upon mixing with the Ultra-Turrax for 2 min. After addition of distilled dichloromethane (300 ml). the liquids were vigorously stirred with an UItra-Turrax T45N high speed blender at 80'~,~full revolutions for 5 rain, placed in the centre of the extraction vessel, the generator ca. 8 mm from the bottom. The vessel was stood overnight in the absence of light, and the dichloromethane layer separated, dried (CaSO.0 and evaporated to near dryness by distillation through a 15cm vigreux column. After transfer to a smaller pear-shaped flask, the solution and washings were evaporated to a small volume and an octacosane solution (0.04 ml; 40.7/ag mladded). After further evaporation to a total volume of ca. 20 td. aliquots were withdrawn and analysed by GLC, 3.2 Extraction in the presence of suspended solids Fullers Earth (B.D.H. Chemicals Ltd.) was purified by Soxhlet extraction with dichloromethane for 7 days, and after addition to purified water (5 I.), it was mixed with the UItra-Turrax for 2 rain, followed by addition of the P.A.H. in dichloromethane solution and further mixing for 2 min. A 20-1. water sample provided three sample solutions and one blank. The solutions were extracted and analysed as above.

211

33 Effect of prolonoed mixing P.A.H. in dichloromethane solution (ca. 0.2ml) were added to purified water as before, with and without prior addition of suspended solids. The solution was kept in the absence of light for 6 h. stirred gently with a magnetic stirrer. then extracted and analysed as above. 3.4 Continuous soh'ent extraction Purified water 120 1.1 was split into four equal portions and one portion placed in a continuous extraction apparatus. Alter addition of P.A.H. in dichloromethane solution (ca. 0.2 mll and suspended solids it was agitated strongly to ensure mixing and then extracted continuously with dichloromethane for 36 or 72 h. Blank extractions were performed, and analyses carried out by G.L.C. 3.5 Efficiency oj" the extract pto'ification procedure (a) A dichloromethane solution lca. 0-2 roll of P.A.H. was dissolved in aqueous methanol and partitioned sequentially according to the procedure of Hoffmann and Wynder (1960). Finally the nitromethane extract was evaporated by distillation through a 15cm vigreux "column. After addition of octacosane solution (0-2 ml: 54-0 t~g ml- 1) as internal standard, and evaporation to ca. 50 ~d, an aliquot was withdrawn and analysed by GLC. (b) The above procedure was followed, the P.A.H. being extracted from aqueous methanol into cyclohexane (30 ml). The cyclohexane was shaken with distilled dimethylsulphoxide (3 x 15 ml) for 3 rain each extraction, the combined extract diluted with water/90 ml) and extracted with cyclohexane (2 x 25 ml) for 5 min shaking each extraction. The cyclohexane was removed b) distillation through a 15 cm vigreux column, and after addition of octacosane solution (0"2 ml; 54.0 #g ml- *). the analysis was performed as above. 3.6 Aimlysis by TLC amt GLC (a) TLC analysis. The procedure of Borneff and Kunte (1969) was followed with slight modifications. The first development with iso-octane was omitted, and the plates were developed for a fixed distance of solvent front advance, rather than a fixed time. After development, the plate was sprayed with a 2~/ow v- t solution of tetrachlorophthalic anhydride in acetone/chlorobenzene (10:1) and dried prior to measurement of fluorescence by irradiation with a u.v. lamp (,~. = 360 nm) and visual comparison with standard chromatograms. (b) GLC analysis. Aliquots of P.A.H. solution were injected onto a Hewlett-Packard 7620 G.C. using balanced dual columns (3.5m x 0'5cm o.d. stainless steel; 3% OV7 on 60/80 mesh AW-DCMS Gas Chrom Q) and nitrogen carrier (35 ml rain- 1) fitted with dual F.I.D. A temperature of 260~C was maintained for 8 min and then increased to 300 ~ at 8~ rain- 1. Octacosane was used as an internal standard. and a linear relationship between the ratio of peak areas (P.A.H.: octacosane) and the ratio of concentrations was found for each compound upon calibration with standard solutions in dichloromethane. 3.7 Collection and analysis of field samples Water samples (9.5 1.) were collected in duplicate in borosilicate glass vessels which were stoppered and returned immediately to the laboratory. One sample was dosed with a solution of P.A.H. in dichloromethane (0-1-0-2ml) and mixed thoroughly. To each. dichloromethane (600 ml) was added and the resultant liquids emulsified with the UltraTurrax at 80°/o full revolutions for 5 rain. After stoppering, the mixture was left overnight in the absence of light. The bulk of the aqueous layer was decanted and the resultant liquids filtered under vacuum through a glass fibre paper which was subsequently washed with dichloromethane. The filtrate and washing were transferred to a separating funnel and the dichloromethane layer separated, dried (CaSO.0

212

:,,1 -\ \,Hlq,,.. R \ 1 H.~.RttlSO',. R P~a,, :rod R \. ~A~:'.-.v,-

and e',aporated b~ distillation through a !5 cm vigreux column, When reduced to a small ~olume. the extract aas split for analysis by TLC and GLC. Thin laser chromatography anabsis ~as performed directb on the crvdc extract, ,a,hilst the GLC sample ~as purified b~ the procedure of Hoffmann and W~nder (19601 described abo~e. The GLC sample was then split into two portions: one portion was examined ~ith the addition of octacosane as internal standard and one without, in order to subtract background peaks of similar retention time to octacosane. 4. COXCLtSJO~S As anticipated, the factors influencing extraction efficiency are complex, and it is not possible to predict precisely an efficiency for a given ~tmple. The results presented here, however, do offer a uset\d guide for prediction of probable extraction efficiency and allow statement of P.A.H. concentrations in water wkh more confidence than would otherwise be possible. In our hands, the extract purification procedure of Hoffmann and Wynder (19601 showed serious drawbacks, which modification by use of dimethylsulphoxide overcame. Disadvantages of both G L C and TLC analytical procedures were apparent. The T L C method avoids extract purification and associated losses, but is of poor sensitivity for a number of compounds. Accuracy does. however, appear high for the compounds determined, as shown by analysis of P.A.H. added to field samples, Gas-liquid chromatography has a more uniform sensitivity, but is unable to separate some isomeric compounds, and consequently some carcinogenic P.A.H. are determined onl) in combination with non-carcinogenic isomers. The G L C procedure is more susceptible to interference by background organic materials and appears

to be of lo~er accurac~ than TLC. although it cl~'arl 5 is abic to gi',e results of the correct order of magnitude. REFERENCES Bhatia K, 11971t 4mdyt. Chem. 43. 609-610. Borneff J. & Kunte H, 1/964)Arch, Hy 9. Bakt. 148, 5S5-597. Borneff J. & Kunte H. 119651 .4rch. Hyy. Bakt. 149, 226-243. Borneff J. & Kunte H. (1969~ .4rch. Hyg. Bakt. 153, 220-229. Bravo H.. Nulman R.. Monkman L. & Stanley T. (1970) Proc. Secoml Int. Clean Air Conf, pp. I18-121. Washington D.C. Dubois k., Corker~ A. & Monkman J. L. (1960} Int. J. Air Polha. 2, 236--252. Haenni E. O.. Hoaard J. W. & Joe F. L. (1962) J..4~s. o f agric. Che,l. 45. 67-7t). Hoffmann D. & Wynder E. L. (1960),4nal.vt. Chem. 32. 295-296. Kunte H. (19671 Arch. H)V. Bakt. 151, 193-201. Lane D. A.. Moe H. K. & Katz M. 119731 Analyt. Chem. 45. 1776-1778. Samoilovich L. N. & Redkin Y. R. (1968) Hyg. Sanit. 33, 165-~168. Sawicki E.. Stanley T. W.. Elbert W. C., Meeker J. & McPherson S. 11967) Atmos. Environ. !, 131-145. Wedgwood P. & Cooper R. L. (1953) Analyst, Lond. 78, I70-173. Wedgwood P. & Cooper R. L. {1954) Analyst, Lond. 79, 163 169. Wedgwood P. & Cooper R. L. (19561 Analyst. Lond. 81. 42M4. World Health Organization 11964) Wld Hlth Org. tech. Rel,. Sere. 276. World Health Organization 11970) European Standards lbr Drinkmq H,ater. Second Edition. Geneva. World Health Organization (1971) International Standards jot Drinking Wuter, Third Edition. Geneva.