Quantitative determination of mirex and its degradation products by

1980, 52, 2272-2275. Quantitative Determination of Mirex and Its Degradation. Products by High-ResolutionCapillary Gas. Chromatography/Mass Spectromet...
1 downloads 0 Views 526KB Size
2272

Anal. Chem. 1980, 52, 2272-2275

Quantitative Determination of Mirex and Its Degradation Products by High-Resolution Capillary Gas Chromatography/Mass Spectrometry Francis I. Onuska, Michael E. Comba, and John A. Coburn' National Water Research Institute, Canada Centre for Inland Waters and Inland Waters Directorate, 867 Lakeshore Road, Burlington, Ontario, Canada L7R 4A6

A substantial pari of Lake Ontario has been contaminated with mirex (dodecachlorooctahydro-l,3,4-metheno-2H-cyclobuta[ c,d]pentaiene). Also, some specific wildlife exposures to mirex have been confirmed. The qualitative and quantitative determinations of mirex and a number of related metabolites were evaluated by mass spectrometry utilizing their electron impact (EI) ionization mass spectra. The selected ion monitoring data were compared with gas chromatographicresuns by use of an electron capture detector (ECD). Levels in the low picogram range were detected by using single ion monitoring at m / z 272 (SI-SIM), while the distinction between individual mirex metabolites was made by comparing the ratios of four preselected ions by use of multiple Ion-selected-ion monitoring (MI-SIM).

T h e importance of quantitative analytical measurements in trace organic analysis cannot be overemphasized. Control measures and enforcement of water quality criteria are established on the basis of these data along with the determination of the fate and effects of various contaminants. Hence, t h e development of suitable analytical methodologies to provide for these needs must incorporate the qualities of greater sensitivity, selectivity, and reliability. One of the more promising methods of achieving these goals has been t h e use of capillary column gas chromatography combined with mass spectrometry by utilization of selected-ion monitoring (SIM) features on the more modern instruments (1-4). T h e major benefits derived from quantitative mass spectrometry include improved selectivity and sensitivity and less sample cleanup. Advancements in the design and stability of mass spectrometers have also significantly reduced instrumental error to such a degree t h a t sample handling and operator skills are now the primary source of errors ( 5 ) . Presently, only mirex can be separated from the polychlorinated biphenyl (PCB) matrix on certain conventional packed columns. Glass capillary wall-coated open tubular (WCOT) columns can provide additional separation for some of its related degradation products (8), but chemical pretreatment of the sample is required in order t o obtain an overall quantitative measurement. In addition, the reliability of mirex levels in Lake Ontario fish was questioned by Laseter e t al. (6, 7) on the basis of quantitative mass spectrometry results. I t is the purpose of this paper t o provide a rapid, quantitative procedure for the determination of mirex and the major environmental degradation products of mirex and to establish whether quantitative mass spectrometry does yield significantly different values from the gas chromatographic method employing electron-capture detection.

EXPERIMENTAL SECTION Samples. Ten lake trout samples from eastern Lake Ontario, five lamprey samples from central Lake Ontario, and two Niagara

River suspended sediment samples collected at Niagara-on-theLake were analyzed for chlorinated hydrocarbons and polychlorinated biphenyls. Extraction. Fish. Whole fish were cut into pieces and ground three times to homogeneity by using a stainless steel Hobart grinder. A (10-g) portion of the homogeneous tissue was accurately weighed into a small glass blender and extracted twice with 120-mLportions of acetonitrile. After each extraction, acetonitrile was decanted and filtered through a 15-g column of solvent-washed Celite. The acetonitrile was transferred to a separatory funnel and distilled water added to adjust the aqueous content to 20% (v/v). This mixture was extracted three times with a total volume of 300 mL (3 X 100 mL) of petroleum ether. The combined petroleum ether extracts were dried through a 10-15-g sodium sulfate column, evaporated to approximately 4 mL on a rotary evaporator (water bath 37 "C), and made up to 25 mL with isooctane in a low actinic volumetric flask. Sediment. Suspended sediments were obtained at Niagaraon-the-Lake. Water volumes of 1500-2000 L were processed to obtain each sample. The suspended sediment was rinsed from the bowl with XAD-2 purified water, and this mixture was pressure filtered through 2-5-pm Teflon filters and dried in a desiccator. The dried material was scraped from the filter and ground with a mortar and pestle and a representative portion (3-5 g) was accurately weighed into a stainless steel beaker. The moisture content of each sample was adjusted to 20% (w/w) with XAD-2 purified water. Samples were extracted twice with 100-mL portions of (1:l) hexane/acetone by use of a Sonicator Model 350 cell disruptor equipped with a 19-mmQ-horn probe (Heat Systems Ultrasonics Inc., Long Island, NY). The solvent was decanted after each extraction, and filtered through a prewashed 15-g Celite column. The combined extracts were transferred to a separatory funnel and partitioned with 150 mL of water. The aqueous phase was transferred to another separatory funnel and extracted with 100 mL of benzene. The organic extracts were combined and dried through a 10-15-g column of sodium sulfate and concentrated to about 3 mL on a rotary evaporator. A change to isooctane was made by two additions of 50 mL of isooctane and evaporation to about 3 mL on the rotary evaporator. The extract was made up to 10.0 mL with isooctane in a low actinic volumetric flask. Cleanup. A measured volume (5.0 mL) of the sample extract was placed on a 30-g Florisil (activated at 130 "C until used) column for cleanup and fractionation. Four fractions each of 200 mL were successively eluted and collected. The compositions of the fractions were petroleum ether (PE),6% diethyl ether in PE, 15% diethyl ether in PE, and 50% diethyl ether in PE. Each fraction was concentrated on a rotary evaporator and adjusted to a final volume of 10.0 mL with isooctane. Mirex and its degradation products were eluted in the petroleum ether fraction along with hexachlorobenzene, heptachlor, aldrin, p,p'-DDE, a portion of p,p'-TDE and PCB. Gas Chromatography Analysis. The electron-capture detector gas chromatographic analysis of the Florisil fractions was performed on 1.8 m X 2.5 mm i.d. borosilicate glass columns packed with 4 % OV-101 and 6% OV-210 on Chromosorb W-HP, 80-100 mesh, 3% OV-101 on Chromosorb W-HP, 8C-100 mesh, and 1.5% OV-17 and 1.95% QF-1 on Gas Chrom Q, 1W120 mesh. These columns were operated isothermally at 190-200 "C with argon-methane (5%) carrier gas flow of 30-50 mL/min.

0003-2700/80/0352-2272$01,00/00 1980 American Chemical Society

ANALYTICAL CHEMISTRY, VOL. 52, NO. 14, DECEMBER 1980

Table I. Relative Abundance of Five Characteristic Ions in Electron Impact Mass Spectra compound mirex 8-monohydromirex 1O-monohydromirex

2,8-dihydromirex

Table 11. Comparison of Mirex Concentrations (pg/pL) by Four Different Techniques

masses (re1 abund) 2 7 2 (loo), 237 ( 6 1 ) , 4 0 2 357 ( 5 ) , 5 4 6 ( 3 ) 2 7 2 (loo), 2 3 8 ( 9 6 ) , 2 0 3 475 (38), 370 (18) 2 3 8 ( 1 0 0 ) . 2 7 2 ( 3 8 .).. 2 0 3 3 6 8 (8);.476 ( 2 ) 2 3 8 (loo), 2 0 3 ( 4 1 ) , 3 7 6 334 (3). 441 ( 2 )

MS-SIM (7),

2273

sample

EC-GC

1 2 3 4 5 6 7

316' 138 168 106 79 103 112 77 86 124 111 129 t12.8

m/z 546

MS-SIM m/z 2 7 2

MS-SIM 4 ions

Trout (72), (. 2 4 ,, ). (4),

8

Quantitation was based on the identification of a compound on the three above columns using a 2% retention time window with the external standard method (IO)and the SP-4ooO Spectra Physics computing integrator for the integration of peak areas. Mass Spectrometry. Analyses were performed on a Varian MAT 311-A mass spectrometer interfaced to a Varian 2700 gas chromatograph via an open-split interface. Sample separation was achieved on a 20 m X 0.25 mm i.d. glass capillary WCOT column coated with OV-101. Helium was used as the carrier gas a t the rate of 2.5 mL/min with a sample split ratio of 5 to 1. The injection size was 1 pL. The GC/MS conditions were as follows: ion source EI-EID (EID, 20-eV electron impact detector for the total ion current trace); electron energy 65 eV; emission current 2 mA; accelerating voltage 3 kV; multiplier gain 4.3 X lo5 a t 2 kV; multiplier bandwidth 10 kHz; ion source temperature 230 "C; interface temperature 230 "C; GC oven temperature 180 "C. Quantitative data were acquired by three separate measurements on each sample. Selected ion monitoring conditions of the m / z 272 parent ion at a resolving power of 1800 were optimized by introducing a small amount of mirex via the solid probe and adjusting the focusing potentials. One microliter of the original cleaned-up sample extract was injected on the column a t 180 "C isothermal and recorded on a strip chart recorder (10 mV full scale deflection), The sample and standard peak heights were measured manually by a caliper and transferred to linear graph paper. A second measurement was taken by monitoring the molecular ion of mirex at m/z 546 with the instrument optimized in the same manner as previously stated. A l-mL portion of the previously cleaned-up sample was taken and dried under a stream of nitrogen. The conical tube was rewashed with 1 mL of hexane and again gently evaporated to dryness. The sample was made up to 50 p L in isooctane with a 50-pL syringe. Measurements were made by using the same instrumental conditions as previously described except at a resolving power of 1400. The third set of measurements was acquired by using the Varian 166 Spectro System which has the capacity to perform computer-controlled selected ion monitoring for up to eight preselected masses. Four preselected ions corresponding to mirex and its degradation products ( m / z 203,237,238 and 272), shown in Table I, were chosen for multiple ion detection. The peak centroids and maximum responses were optimized at a resolving power of 800 by introducing a small amount of 8-monohydromirex via the solid probe and monitoring the ion intensities on the oscilloscope. Since variations in the accelerating voltage cause small fluctuations in the focusing and sensitivity toward the higher mass, a response ratio for m / z 272/237 of 1.15 of mirex was found to be optimum for all of the products examined. A dwell time of 200 ms, settling time of 18 ms, and computer sampling frequency of 12 kHz were used for the measurements. These conditions provided a minimum of eight points per peak a t the 100-pg level. The samples were analyzed by temperature programming a t a rate of 4 OC/min a t an initial column temperature of 120 "C. The data acquisition was programmed to start 3 min after injection to allow for passage of the solvent front. One-microliter injections of both previously prepared samples (1 mL aliquot and 50 p L ) were analyzed. The sample data were stored on disk and later displayed by using a Tektronix 4010 graphic display terminal and Statos 41 printer/plotter. Quan-

9 gC 10

mean RSD

330" 83 117 75 121 93 121 101 90 90 78 118 r14.2

305' 110 160 93 122 118 122 93 91 103 77 127 i7.5

301' 136 138 94 105 97 115 113 102 103 78 126 i10.6

Lamprey 1 2 3 4 5 mean

18.7' 17.5 11.3 21.9 16.4 17.2

23.2" 14.3 10.0 25.0 13.7 17.2

19.5' 17.7 10.0 20.0 15.5 16.5

17.4a 13.0 10.2 18.1 13.4 14.4

Sediment 1 2

280' 782

273' 692

a Sample volume 5 0 pL. Duplicate.

258' 837

262' 710

' Sample volume 10 mL.

titative data were obtained by manual measurements of the peak heights.

RESULTS AND DISCUSSION The relative abundance9 of five characteristic ions of mirex and its degradation products in t h e electron impact mass spectra are given in Table I. Mass spectra are presented elsewhere (8). Each method was examined for sources of error. For the SI-SIM measurement of the m/z 546 molecular ion, five 10-mL fractions were separately prepared containing 10 pg/mL of mirex and were subjected to the sample clean-up procedure. The standards were preconcentrated in the same manner as the samples yielding a final concentration of 200 pg/pL. T h e relative standard deviation (RSD) was 14.2% with a mean recovery level of 98.6%. In addition, one standard was run 10 times and gave an experimental error, RSD = 8.3% leaving a sample manipulation error 14.2 - 8.3 = 5.9%. A calibration curve was established for 15G2000 pg injected and was found t o be linear. T h e respective limit of detection at a signal-to-noise ratio (S/N) of 3.0 was 100 pg injected. For the single ion (SI-SIM) measurement at m/z 2 7 2 , l mL of the previously prepared standards was taken for t h e analyses. One microliter of each of the five untreated solutions was injected (10 pg/pL) and gave a reproducibility of RSD = 7.5%. The calibration curve was linear between 1 and 800 pg of mirex with a limit of detection of 500 fg at a S/N of 3.0. For the multiple ion detection (MID) procedure, l-pL injections (200 pg/pL) of t h e five preconcentrated standards yielded a RSD = 16.5%. In addition, five separate standards were prepared by using the sample clean-up procedure to give concentrations of 200 ng/mL without preconcentration. One microliter of each of these samples was injected a n d gave a RSD of 10.6%. The instrumental error was calculated on the variation in the 272 t o 237 m / z ratio over a n 8-h period a n d represented 4.3%. A calibration curve was prepared between 50 and 2000 pg and was found to be linear with a limit of detection of 50 pg

\

2274

ANALYTICAL CHEMISTRY, VOL. 52, NO. 14, DECEMBER 1980

h'

Table 111. Regression Analysis for Trout Samplesa parameter

EC vs. m / z 546

m B

- 5.915

ax aY

64.291 68.834

0.961

0.90

P

a

method EC vs. m/z 272

EC vs. MID

0.822 12.854 64.291 60.209 0.94

0.851 15.766 64.291 57.897 0.95

I

w

I

Formulas: y = ( a x / a y ) X + B ; y = m x + B .

sample

8-mono- 10-mono- 8,8-dihydro hydro hydro

trout 36 lamprey BD sediment BD sensitivity 120 p g (S/N = 3)b

0.1 BD BD 40 p g

12

BD BD 50 p g

b A

m / z 237

Table IV. Mean Concentration of Mirex Degradation Products (pg/pL) by MS-MIDa 10,lO-

dihydro BD BD BD 130 pg

a BD = below detection limit, sample volume 50 pL. Most intense ion.

0

10

5

15

MINUTES

Flgure 2. Multiple-ion detection traces for injected standard of 100 pg of each mirex degradation product: (a)2,8dihydro, (b) 10,10dhydro, (c) 8-monohydro, (d) 10-monohydro, (e) mirex.

C

w

ln z

I

MINUTES

m / z 238

0 n

X400

m W

a

Flgure 1. Single-ion monitoring trace of m l z 272 for standard; 1 pg injected of each (a) 8-monohydro, (b) 10-monohydro, and (c) mirex, on 20-m OV-101 at 180 O C isothermal. injected (SIN = 3) for mirex on the m/z 272 ion. Quantitative results for each determination are presented in Table 11, with the previously calculated total experimental error expressed as RSD. In order to evaluate the methods, we performed a linear regression analysis for the determination of mirex in the trout samples, t o establish any significant difference between the procedures used. The standard deviations of the respective methods were tested by paired analysis and were within the same population a t a confidence interval of 95%. The results of the linear regression analysis, listed in Table 111, show a high degree of correlation between all methods. No significant difference was found between the r n / z 272 values, m / z 546 values, and the gas chromatographic results. This indicates no substantial interferences for the determination of mirex in trout. These observations also appeared to be true for the analyses of lamprey and sediment samples. However, in the case of the sediments, the sample background level deteriorated the sensitivity by a factor of 3, while the other analyses had detection limits equivalent to the instrumental sensitivity. Figure 1 illustrates a trace obtained for 1 pg injected of each component in a mixed standard of mirex and related compounds. The MID feature allows for the simultaneous determination of mirex and a related number of its degradation products. As previously indicated, the method gave good correlation with the mirex results obtained by the other procedures (see Table IV). A comparison of Figures 1 and 2 shows no significant interferences of the ion ratios for the compounds examined, with only one small peak in the vicinity of mirex (approxi-

,

m/z 203

, I

0

5

l

,b

l

10

1

X400

,

1

15

MINUTES

Flgure 3. Multiple ion detection traces for Lake Ontario trout sample 1: (a) 2,8dihydro, (c) 8-monohydro, (d) 10-monohydro, (e) mirex.

mately 2.8%) having a m / z 272 intensity, although it is well separated by capillary column. No interferences were observed in the trout or lamprey samples examined, i.e., all the ion ratios remained constant (see Figure 3). No interferences were observed for the sediment analyses, although a series of earlier eluting peaks gave signals for the m / z 203 trace. This is further substantiated by the various values obtained for the degradation products in trout samples as listed in Table 11, when expressed as a percentage of mirex (8-monohydro, 28.7%; 10-monohydro, 10.1%;and 2,8-dihydro, 1.6%) which are in good agreement with other reported data (8, 9).

CONCLUSIONS Statistically insignificant differences were found, based on the correlation of results from the linear regression analysis, for the determination of mirex by selected ion monitoring mass spectrometric data and gas chromatography with electroncapture detection results. The single-ion detection procedure monitoring ion m / z 272 is a rapid and sensitive means for

Anal. Chem. 1980, 52, 2275-2277

screening and quantitating mirex levels down to the 500-fg level. The multiple-ion detection technique extends the determination to include related degradation products and provides a high probability for correct identification when one considers four separate ion ratios and retention times fit. T h i s procedure does not suffer from the interferences previously reported by Laseter et al. (6) and can be employed for a broad range of samples.

LITERATURE CITED (1) Millard, 8. J. "Quantitative Mass Spectrometry"; Heyden Publishing Co.: London, 1978; Chapter 6. (2) Ciaeys, M.; Markey, S. P.; Maehaut, W. Bbmed. Mass Spectrom. 1977, 4 (2),122-128.

(3) (4) (5) (8) (7)

(8) (9) (10)

2275

Van Vaeck, L.; Van Cauwenbergha, K. Anal. Len. 1977, 10, 467-482. FuJIl, T. Anal. Chlm. Acta 1977, 92, 117-122. Buckle, W. L.; Elchelberger, J. W.; Anal. Chern. 1979, 57, 567A. Laseter, J. L.; Deleon, J. R.; Remel, P. C. Anal. Chem. 1978, 50, 1189-1 172. Deleon, J. R.; Warren, V.; Laseter, J. L. Quant. Mass Spectrom. Life Sei. 1978, 2 , 483. Norstrom, R. J.; Hallet. D. J.; Onuska, F. I.; Comba, M. E. Environ. Scl. Techno/. 1980, 74, 860-866. Hallett, D. J.; Norstrom, R. J.; Onuska, F. I.; Comba, M. E.; Sampson, R. J . Agric. FocdChem. 1976. 2 4 , 1186-1193. "Analytical Method Manual-1979"; Environment Canada, Inland Waters Directorate-WQB: Ottawa, Aug 1979.

for review September 13, 1979. Accepted August 20, 1980.

Determination of Trace Amounts of Alkylbenzenesulfonates by High-Performance Liquid Chromatography with Fluorimetric Detection Atsuo Nakae," Kazuro Tsujl, and Makoto Yamanaka Tochigi Research Laboratories, Kao Soap Co., Ltd., 2606, Akabane, Ichikai-machi, Haga-gun, Tochigi, Japan

Trace amounts of aikylbenzenewlfonates (ABS) In river water were determined by reversed-phase hlgh-performance liquid chromatography (HPLC) without pretreatment. The fluorhetrk detector was operated at 225 nm (excitation) and at 295 nm (emission). Alkyl chaln distributions and partial phenyl isomer compositions of the determined ABS were also Obtained. By addltion of a large amount of sodium dodecyi sulfate to the collected samples and the standard ABS solution, the adsorption loss of ABS was negllgible. Relative standard deviation for the river water, contalning 0.097 Fg/mL of ABS, was 2 % .

Trace amounts of alkylbenzenesulfonates (ABS) in the environment have been determined by colorimetry with methylene blue ( I ) , infrared spectrometry (2, 3 ) , gas chromatography (GC) (4-6), atomic absorption spectrometry (7, 8),a n d high-performance liquid chromatography (HPLC) ( S I 2 ) . Colorimetric methods based on methylene blue have long been used but often give erroneous results due to interference by many organic and inorganic materials. Colorimetric results, therefore, have been presented not in terms of ABS but of methylene blue active substances. Infrared spectrometric methods based on the extraction of amine salts of ABS require several different separation steps and a concentration step. The atomic absorption spectrometric method based on the extraction of an ion-association compound with t h e bis(ethylenediamine)copper(II) or t h e tris(1,lOphenanthroline)copper(II) cations is a highly sensitive method, but ABS is determined indirectly. GC has very high efficiencies for the analysis of individual components of ABS. Moreover, specific and highly sensitive detectors such as flame photometry (13),electron capture ( 5 ) ,and mass spectrometry (6) enable trace amounts of ABS to be determined accurately without interferences. GC, however, requires the coversion of ABS into volatile derivatives such as the methylsulfonates or sulfonyl chlorides before the analysis, and i t is not appli0003-2700/80/0352-2275$01 .OO/O

cable to multisample analysis. HPLC is the most suitable method for the analysis of ABS, because it does not require the conversion of ABS into volatile derivatives. In a previous paper (9),we reported that 1-5 hg of ABS was determined by HPLC employing UV detection a t 225 nm, with silica gel as a stationary phase and n-hexanelethanol (8:2, v/v) containing sulfuric acid as a mobile phase. Since ABS eluted under the above chromatographic conditions giving a single peak,information on the alkyl chain distributions and the isomer compositions was not obtained. We also reported that even- or odd-carbon-numbered ABS were separated by HPLC using porous microspherical poly(styrene-divinylbenzene) gel as a stationary phase and 0.5 M perchloric acid in methanol as a mobile phase (14). B u t this method was not applicable to the determination of trace amounts of ABS in the environment, because commercial ABS gave a complicated chromatogram. In this paper, we describe improved chromatographic conditions for determining ABS in river water with improved results.

EXPERIMENTAL SECTION Apparatus. The liquid chromatograph consisted of a Hitachi 635 pump with a Hitachi 650-10 LC fluorescent spectrophotometric detector (Hitachi Scientific Instruments, Inc., Tokyo, Japan), a NS NHV-5000-6MP injection six-way valve with a 500-fiL loop (Nippon Seimitsu Co., Tokyo, Japan), an ATTO perista minipump (ATTO Co., Tokyo, Japan), and a Haake Model FE circulator (Haake Inc., Karlsruhe, West Germany) for column temperature control. Peak areas and retention times were obtained with a Shimadzu Chromatopac-E1A data processor (Shimadzu Scientific Instruments, Inc., Kyoto, Japan). Reagents. Deionized water was distilled once in an all-glass apparatus. ABS and sodium dodecyl sulfate (SDS) were obtained from our company and were purified to remove unreacted and inorganic substances. All other reagents were of analytical-reagent grade. Standard ABS Solution. ABS solutions (0.01-0.50 pg/mL) containing 0.5% (v/v) formalin, which prevents the biodegradation of ABS, and SDS (100 gg/mL) were prepared. Sample Preparation. A 0.5-mL formalin solution containing 10 mg of SDS was added to 100 mL of collected samples from 0 1980 American Chemical Society