Anal. Chem. 1980, 52, 1845-1849
Determination of a-Chloroacetanilides in Water by Gas Chromatography and Infrared Spectrometry Jimmy W. Worley," Melvin L. Rueppel, and Fredrick L. Rupel Research Department, Monsanto Agricultural Products Company, 800 North Lindbergh Boulevard, St. Louis, Missouri 63 166
Extractive workup and gas chromatographic analytical techniques are described for characterization and quantitation of a-chloroacetanilides and related materials in water. Nitrogen-selective detection for quantitation leads to good accuracy and precision at the low paris per billion level. On-the-fly gas chromatrography-Fourier transform infrared spectrometry is shown to be quite useful as a trace identification tool for this high-boiling class of compounds.
Table I. Structure of Acetanilides Included in This Study
compd
X
R
1
c1 c1
i-C,H, CH,OCH, CH,OC,H, CH,OCH, H CH,OCH, CH,OC,H,,
2
3
Increasing environmental concerns and regulatory activity require that definitive methods for various classes of organic compounds in water be developed. Both characterization and quantitation techniques are needed t o meet numerous objectives-including evaluation of discharge reduction projects, statistical demonstration of compliance with present and future effluent permits, and general surveys on a nonroutine basis of river water, ground water, a n d other sources. W e report here general sampling and analytical methodology for members of the herbicidal n-chloroacetanilide class at the low parts per billion level in water. Specific compounds investigated were propachlor ( l ) ,alachlor (2), butachlor (3), and t h e three related acetanilides 4-6 (see Table I). T h e basis of the methodology is extraction of a water sample with methylene chloride, concentration and solvent exchange for toluene/methanol, and gas chromatographic (GC) analysis. Quantitation is done by using a nitrogen-selective detector. Structural confirmation is done by using on-the-fly gas chromatography-infrared (GC-IR) spectrometry.
EXPERIMENTAL SECTION Quantitative G a s Chromatography. Materials. Samples of 1-7 were from Monsanto Agricultural Products Co., St. Louis, MO. All materials were 99+ % by gas chromatography with flame ionization or nitrogen-selective detection except 6. This compound was ca. 95% pure. Toluene, methylene chloride, and methanol were distilled-in-glass, pesticide-grade solvents from Burdick and Jackson. Apparatus. A Hewlett-Packard 5840A gas chromatograph equipped with a nitrogen-phosphorus detector operating in the nitrogen mode was used. The injector temperature was 250 "C, and the detector temperature was 300 "C. On-column injection was done on a ' / 4 in. o.d. X 2 mm i.d. x 6 f t length glass column containing 10% OV-11 on 100-120 mesh Gas Chrom W-HP. Carrier gas was 25 mL of helium/min. After injection, the column was held a t 200 "C for 2 min and then programmed to 250 "C at 6 "/min and held for 10 min. Procedure. A 1000-mL water sample is mixed with 5.0 mL of a methanol solution containing 2000 ppb 7 , as an internal standard, and extracted in a separatory funnel (Teflon stopcock and stopper) with one 20-mL portion and three 7-mL portions of methylene chloride. To minimize transfer losses, each organic extract is drawn off directly into a Kuderna-Danish concentrator (Kontes No. K-570050-0425 and K-26500). A boiling chip, 0003-2700/80/0352-1845$01 OO/O
4 5 6 7
c1 H c1 OH c1
R' H
C,H5 C,H, C,H, C,H, C;H; C,H,
name a propachlor alachlor butachlor
a Propachlor, alachlor, and butachlor are the active ingredients, respectively, of Ramrod, Lasso, and Machete herbicides, marketed by Monsanto Co. For illustrative nomenclature, the chemical name of alachlor is 2-chloro. 2', 6'-diethyl-]\'-( methoxymethy1)acetanilide.
preextracted with methylene chloride, ii; added to the concentrator, and the volume is concentrated to -3 mL with the aid of a steam bath. The upper flask is rinsed down with 2 mL of toluene, and heating is continued another 5 min to remove the remaining methylene chloride. The upper flask then is rinsed down with 1 mL of methanol, and the combined organics are transferred to a 5-mL centrifuge tube (Fisher No. 5-538-35A) with a clean disposable pipet. The Kuderna-Danish apparatus is washed with 2 mL of toluene, which also is transferred to the centrifuge tube. Under a gentle stream of nitrogen, th.e solution then is concentrated to -0.2 mL. Methanol (0.1 mL) is added, the tube is stoppered, and the contents are mixed by shaking. For analysis, 3.0 p L of this solution is injected into the chromatograph. Responses are compared to those obtained for a calibration solution. The calibration solution may be a concentrated tolueneemethanol (2:l) solution containing the equivalent of 2.5 ppb each of 1-6 and 10.0 ppb of 7. However, more uniformity of calibration solution and sample responses will be obtained if the calibration solution is a concentrated extract of a synthetic water sample. This will minimize the effect of some workup artifacts that were observed in the development of this method-for example, elevated chrornatographic base line and broadened response to 6. Contamination. Great care is taken to avoid contamination and memory effects. The gas chromatograph and all glassware used in the sample workup are dedicated to unique roles. For example, equipment used to prepare calibration solutions is not also used for samples. Equipment used to prepare stock solutions and that used for more dilute calibration solutions is not interchanged. All glassware is cleaned initially and immediately after each use with copious amounts of acetone and water. Clean glassware is stored in a closed cabinet; clean hypodermic syringes for injection are stored in glass test tubes. C 1980 American Chemical Society
1846
ANALYTICAL CHEMISTRY, VOL. 52, NO. 12, OCTOBER 1980
Table 11. Summary of Statistical Validation for Six Acetanilides compo- analytical nent precisiona,b
overall recovery, precisionbjc 70
3
0.018 0.010 0.012
0.091 0.090 0.097
4
0.017
0.086
5
0.013
0.095
6
0.05ge
0.122f
1 2
93 94 111
89 99 99
All components were tested at -0.5, 1.0, 5.0, and 10.0 ppb except as noted. Precision measure is the pooled coefficient of variation, or relative standard deviais equal t o the tion, for all test levels. Pooled CV, or square root of the sum of the individual coefficients of variation squared, each weighted by its degrees of freeAll components were tested at -0.5, 1 . 0 , 2.5, dom. 5.0, and 10.0 ppb, except as noted. Recovery is the numerical average of recoveries a t all test levels. This component was tested at 5 and 1 0 ppb only. f This component was tested a t 3.5 and 1 4 . 3 ppb.
i
L
m,
-
-
Validation. For validation of the gas chromatographic analysis only, four toluene solutions were prepared and analyzed five times each. Each solution contained all six acetanilides 1-6, plus the internal standard 7 at a concentration of 33.3 ppm. The four test levels of 1-6 were -0.67, 1.67, 3.33, and 33.3 ppm, which correspond respectively t o a level in water before extraction and concentration of 0.2, 0.5, 1.0, and 10.0 ppb. Statistical evaluation of the data is described in the Results and Discussion. For validation of the entire procedure, synthetic water solutions were prepared at each of five test levels--0.2, 0.5, 1.0, 2.5, and 10.0 ppb of each acetanilide 1-6 (except as noted in Table 11) and 10.0 ppb of internal standard 7. Five solutions were prepared a t each of the five test levels and worked up and analyzed according to the procedure described above. The following description is illustrative of the technique used to prepare the synthetic samples. Twenty-five milliliters of a methanol solution containing 100 ppb each of 1-6 was pipetted into a separatory funnel containing 500 mL of purified water (charcoal, resin, and filter system from Continental Water Conditioning Corp., El Paso, TX). Then 5.0 mL of a solution containing 2000 ppb of 7 in methanol was added by pipet, followed by an additional 470 mL of water. This resulted in 1000 mL of a solution containing 2.5 ppb each of 1-6 and 10.0 ppb of 7. Gas Chromatography-Infrared Spectroscopy. A Nicolet 7199 Fourier transform infrared spectrometer with KBr-Ge beam splitter, gold-plated light pipe (400 mm x 2.5 mm id.), and liquid nitrogen cooled mercury-cadmium-telluride detector (MCT-A, 5000-850 cm-') was used. For the results in Figures 1-9, data were collected with a scanning velocity of 0.712 cm/s (2048 data points/scan) and were apodized with the Happ-Genzel function, a modified trigonometric function. A Varian 3700 gas chromatograph was used. The column was identical with the one used for the nitrogen-selective work above. The end of the column was not connected to a GC detector but directly to the light pipe transfer line, which came inside the GC oven through an opening in the top of the GC. The transfer line was in. 0.d. glass-lined stainless steel tubing. Connection to the column was by means of a in.-'/,6 in. reducing union (stainless steel; Swagelok), using a Vespel ferrule on the column end. The light pipe was held at 275 'C; the transfer line, wrapped with heating tape, also was held at 275 "C. The GC injector temperature was 280 "C. Carrier flow was 22 mL of nitrogen/min. After injection, the column was programmed from 200 to 270 "C a t 12'/min and held a t that temperature.
-
-
RESULTS AND DISCUSSION S a m p l e W o r k u p and Chromatography. The enhanced sensitivity and selectivity of the nitrogen-selective detector
Time (min) Figure 1. Nrtrogen-selective gas chromatogram of a sample containing 2 0 ppb each of 1-6 and 10 ppb of 7 , worked up as described in the Experimental Section (3000-foM concentration) Injection volume was 3 0Ll ./
for nitrogen-containing compounds is well documented (1-3). Its use for the acetanilides helps t o achieve the desired low parts per billion level and to ensure minimum interference from nonnitrogen materials. A chromatogram representing a sample containing 2.0 ppb each of the six acetanilides of interest and 10.0 ppb of the internal standard is shown in Figure 1. Almost base line separation of all components is achieved. Excellent sensitivity is present. Response to the hydroxy compound 6 deteriorates markedly below 2 ppb, however, presumably due to adsorption in the chromatographic system. Significant loss of 6, the most "polar" acetanilide, also was noted in the development of the sample workup. Initial work with toluene only as the solvent exchanged for methylene chloride led t o low, irreproducible recovery of this material. T h e introduction of methanol as a cosolvent resulted in substantial improvement. T h e methanol probably acts as a "polar carrier" for 6, effectively competing with it for active sites on the glass surfaces. A similar effect of methanol for recovery of low levels of polycyclic aromatic hydrocarbons has been reported ( 4 ) . Methylene chloride was chosen because of its low boiling point, and therefore facile concentration, and its superior extracting properties for all of the acetanilides. Chloroform was comparable except it only gave 66% recovery of 6. Several solid absorbents were evaluated as an alternative t o liquidliquid extraction. Activated carbon was very good except no 6 was recovered. XAD-4 resin and Waters Associates C-18 Sep-Paks were less acceptable. The internal standard 7 is ideal for this analysis. It comes in an unobscured region of the chromatogram and should have extraction, chromatographic, and response properties very similar to those of the materials of interest. Its response also makes it suitable for use with other detectors, such as flame ionization and electron capture detectors, should confirmation of the nitrogen-selective results be desired. Statist,icalValidation. Recovery and linearity of response for each of the seven acetanilides were investigated from the low parts per billion level up to 100-300 ppb level and found to be excellent. The method was rigorously validated in the 0.2-10 ppb range. T h e chromatographic analysis alone was validated by doing five replicate analyses of each of four solutions in this range. The four levels were equivalent to water samples containing -0.2, 1.0, 5.0, and 10.0ppb of each material. Each solution contained all six acetanilides plus the
ANALYTICAL CHEMISTRY, VOL. 52, NO. 12, OCTOBER 1980
1847
I I20 - 1070 cm-l
1350- I330 cm-I
1505- 1485 cm-I
1725- 1705 cm-I
0 c
.-c
1 4
0 W
1705- 1685 cm-I I
90
1
180
I
270
1
I
360
450
I
540
I
630
Time (seconds) Figure 2. Set of five chemigrams for 4.0 f i g each of 1-7. Full scale response in each window is 0.03 integrated iabsorbance units. The first
peak after t h e injection mark is the solvent, methylene chloride. internal standard. T h e entire method-extraction, concentration, and analysis-then was validated by working up and analyzing five aqueous samples a t each of the same test levels plus an additional level of -2.5 ppb. Results of the validation are summarized in Table 11. They indicate that the nitrogen-selective GC analysis is extremely precise and contributes only about 10% of the total variance of the method. T h e overall precision is satisfactory for these low levels, indicating 95% confidence limits ( k 2 relative standard deviations) of A2470 or better for each component. Linear regression correlation coefficients were 0.9995 or better for all components, indicating very good linearity in the 0.2-10 p p b range for both t h e GC response a n d t h e extractive workup. Regression analysis was not done for t h e hydroxy compound 6 since only two levels were examined. Average recoveries also were very good for all components. T h e high recovery of butachlor (3) resulted primarily from sub-parts-per-billion column memory effects, which were inexplicably difficult t o overcome. GC-IR Spectrometry. The recent availability of commercial GC-IR systems represents a valuable new tool for trace analysis which is just beginning t o be utilized ( 5 , 6). Previous work on identification of trace organic materials in water has been limited almost exclusively to gas chromatography-mass spectrometry (GC-MS). One notable exception reported happens to be the use of GC-IR by EPA workers t o confirm the presence of alachlor (2) in the much publicized New Orleans drinking water study (7).
The Nicolet 7199 Fourier transform infrared (FTIR) system allows a GC run to be monitored on-the-fly by using a "chemigram" (8). The chemigram is a time-dependent plot of the infrared absorbance integrated over a chosen frequency range or "window". Up to five windows may be monitored simultaneously. T h e effective result is the equivalent of five absorbance-specific detectors and is analogous to the GC-MS technique of mass fragmentography. After the run, t h e chemigrams are examined for areas of interest, and individual infrared spectra may be retrieved. Application of this technique to the acetanilides provides a useful characterization tool. A set of five chemigrams for a mixture containing 4.0 fig each of 1-7 is shown in Figure 2. Near-optimal absorbance windows for various acetanilides were used. Resolution and relative retention of the seven components in Figure 2 are very comparable to those in Figure 1,indicating no apparent degradation of chromatographic quality in the transfer line or light pipe. T h e structural information that may be obtained from the selected absorbance monitoring approach, even for materials as closely related as 1-7,is very evident in Figure 2. All seven compounds are visible in both of the carbonyl (amide I) windows used. However, compound 5, which has the highest wavelength absorbance maximum (1719 cm-'), appears only weakly in the 1705-1685-cm-' window but most prominently in the 1725-1705-cm-' window. All of t h e compounds are tertiary amides except for 5.
1848
ANALYTICAL CHEMISTRY, VOL. 52, NO. 12, OCTOBER 1980 1010,
1011
931
4003
3600
3200
2003 Wavenum bers
2800
1600
2400
'200
800
Figure 3. Single on-the-fly infrared spectrum of propachlor (1).
2BbO
2400
2000
600
I200
'
3600
3200
Wavenumbers
2400
ZOO0
16CC
8b
1203
Wovenumbers
'
3$00
3200
800
Figure 4. Single on-the-fly infrared spectrum of alachlor (2).
2dCC
Figure 5. Single on-the-fly infrared spectrum of butachlor (3).
93
I
32b
970
2&0
24bO 2WO Wave n urn bers
1600
1200
Sbo
Figure 6. Single on-the-fly infrared spectrum of 4.
-
Secondary amides show a strong amide I1 absorbance in the gas phase a t 1495 cm-' (9). This leads to a striking trace in t h e 1505-1485-cm-' chemigram in Figure 2-only 5 gives a significant peak. T h e 1350-1330-cm-' chemigram shows some structural differences but is overall weak and not as useful. The 112C-l070-~m-~ window is very interesting. This is the C-0 stretch region. Compounds 1 and 5 are the only two of the seven which do not contain an N-alkoxymethyl group and, subsequently, they do not respond in the 1120-1070-cm-' chemigram. For the GC run depicted in Figure 2, five to seven infrared spectra, each a coaddition of three scans, were collected across each peak. While these may be coadded to increase the signal-to-noise ratio, a t t h e 4.0 pg level used here this was not necessary to obtain good results. A single spectrum from near the center of each peak is shown in Figures 3-9. Characteristic of gas-phase spectra, 5 (Figure 7 ) shows a sharp N-H band (3428 cm-') and 6 (Figure 8) shows an 0-H band (3509 cm-l). Useful chemigrams and characteristic infrared spectra could be obtained from mixtures containing 1.0 pg of each of the acetanilides. Infrared spectra showing a t least a prominent carbonyl band could still be retrieved from the stored scans from a GC run using 500 ng of material. Given t h a t 1.0 Fg or more of an acetanilide in a mixture is needed to produce a useful chemigram and resulting good quality infrared spectrum, it is evident that GC-IR is a viable tool for confirming structures of acetanilides in water a t the low parts per billion level. T h e concentrated extract of 0.3 m L from 1000 mL of a water sample t h a t results from the workup used for the GC/nitrogen-selective detector analysis contains 2.5 pg of each acetanilide that was present at the 2.5 ppb level in the extracted water samples. The chemigrams in Figure 2 were produced by using a 10-pL injection. Therefore, further concentration of the 0.3-mL sample concentrate to -25 p L and subsequent injection of -10 pL, will give injection of the necessary 1.0 hg of material.
3200
950%600
2dOC
24b0
2000
1600
I200
800
I200
800
1200
800
Wavenumbers
Figure 7. Single on-the-fly infrared spectrum of 5.
965i I
950
3600
3200
2800
2400
2000
1600
Wovenumbers
Figure 8. Single on-the-fly infrared spectrum of 6. 1010,
970
I 3600
3200
28bO
2400
2000
1600
Woven urn ber s
Figure 9. Single on-the-fly infrared spectrum of 7
Anal. Chem. 1980, 52, 1849-1851
LITERATURE CITED Hall, R. C. CRC Crit. Rev. Anal. Chem. 1978, 7 , 323. Albert, D. K. Anal. Chem. 1978, 5 0 , 1822. Marano, R. S.; Levine, S.P.; Harvey, T. M. Anal. Cbern. 1978, 50, 1948. Ogan, K.; Katz, E.; Slavin, W. J . Chromatogr. Sci. 1978, 76, 517. (5) Wall, D. L.; Mantz. A. W. Appl. Spectrosc. 1977, 37, 552. (6) Erickson, M. D.;Pellizzari, E. D.presented at Fourth Annual Meeting of Federation of Analytical Chemistry and Spectroscopy Societies, Detroit, MI, 1977.
(1) (2) (3) (4)
1849
(7) Keith, L. H.; Garrison, A. W.; Allen, F. R.; Carter, M. H.; Floyd, T. L.;Pope, J. D.;Thruston, A. D.,Jr. I n “Identification and Analysis of Organic Pollutants in Water”; Keith, L. H., Ed.; Ann Arbor Science Publishers, Inc.: Ann Arbor, MI, 1976. (8) Vidrine, D. W.; Mattson, D. R. Appl. Specfrosc. 1978, 32. 502. (9) Weiti, D.“Infrared Vapour Spectra”; Heyden and Son, Ltd.: New York, 1970: p 28.
RECEIVED for review May 12, 1980. Accepted July 21, 1980.
Diffuse Reflectance Infrared Spectrometric Analysis of Ultrathin Film Carbowax 20M on Chromosorb W M. A. Kaiser’ and D. B. Chase Central Research & Development Department, E.
I. du POnt de Nemours and Co., Experimental Station, Wilmington, Delaware 19898
Since high-efficiency chromatographic columns containing packings coated with an Ultrathin film organic phase have become popular, much interest has focused on the nature of the coating. I n this study, diffuse reflectance infrared spectroscopy was used to examine the effect of each step in the preparation of packing material coated with ultrathin film Carbowax 20M. The loss of hydroxyl groups from the silica and changes in the Carbowax 20M upon heating are observed in the infrared spectrum; this suggests the formation of Si-0-C bonds, a concept which is not consistent with popular theory on the nature of these packings.
packings although the results of that work were not definitive. Aue proposed that t h e interaction was physical and t h a t the polymer chains rearrange under high temperature t o form a layer strongly held by van der Waals and hydrogen bond interactions between the diatomaceous earth and the polymer. In this work, a specially designed diffuse reflectance infrared spectrometer has been used to follow t h e coating procedure for those packings. T h e spectra showed no change in t h e Carbowax 20M after coating and extracting. Upon heating t o 275 “C, however, loss of hydroxyl groups from t h e diatomaceous earth and changes in the Carbowax 20M spectrum suggested t h a t Si-0-C bonds are formed under these conditions.
Recently many chromatography suppliers have introduced ultrathin film coated stationary phases for gas chromatography. These materials are inert, highly efficient, stable, and selective and are therefore ideal for environmental, trace, and G C / M S (gas chromatography/mass spectrometry) or GC,I E C D (gas chromatography/electron capture detection) analysis. Efforts are now being made to expand the range of available packings by trying novel support/polymer combinations; however, little is known about the nature of the support/polymer interaction. Similar coatings now are being used to “deactivate” silica in WCOT (wall-coated open tubular) chromatography columns. Ultrathin film coated packings were prepared by Aue ( I ) , who observed that Carbowax 20M coated diatomaceous earth, when exhaustively extracted with methanol or benzene, still retained some of its chromatographic properties. Further studies showed efficient columns cannot be obtained with “dry” diatomaceous earth (Chromosorb W) (2) and t h e polarity of the ultrathin Carbowax 20M coated packing was not as great as t h e conventionally coated packing (3). Studies indicated a loading of less t h a n 0.2%; a n d ESCA (electron spectroscopy for chemical analysis) d a t a showed t h a t only three elements were present on the surface, silicon, oxygen, a n d carbon (11, 32, 5 7 % ) ( 4 ) . These d a t a suggest a high coverage of polymer, with a thickness of approximately 15 A. Conventional infrared and nuclear magnetic resonance spectroscopy were not sensitive enough t o give structural information. On t h e basis of extraction studies d a t a , Karasek and Hill ( 4 ) suggested t h a t Si-0-C bonds cannot be present on these
EXPERIMENTAL SECTION
0003-2700/80/0352-1849$0 1.OO/O
Instrumentation. A Hewlett-Packard (Avondale,PA) Model 5710 gas chromatograph with flame ionization detectors was used for column conditioning and all chromatographic analysis. The infrared diffuse reflectance spectra were obtained by using a Nicolet 7199 Fourier transform infrared spectrometer. A diffuse reflectance accessory similar to that of‘ Griffiths ( 5 ) ,but utilizing a paraboloidal collection mirror, was constructed and interfaced to the interferometer. The sample material was held in a cup which had a diameter of 5 mm (two times the focused beam diameter) and was deep enough to obtain a good value of R,. An InSb detector was used for these studies. All spectra were recorded a t 1 cm-l resolution and zero filled four times in the Fourier transformation. For improvement of the signal-to-noiseratio, 256 scans were coadded. The raw interferograms were transmitted to a PDP-10 computer where the data were processed and the Kubelka-Munk transformation was made. The interferograms were apodized by using an algebraic function, F3, defined by Norton and Beer (6). Sample Preparation. The ultrathin, film-coated Carbowax 20M packing material was prepared by treating 40 g of 80/100 mesh Chromosorb W-NAW (Johns-Manville, Denver, CO) with 6 N HCl. The packing was rinsed with distilled water until neutral and dried at 100 “C overnight. A 10% w / v solution of Carbowax 20M (polyethylene glycol, molecular weight 16 000-20 000) in methanol was contacted with the support for 30 min and then the solution was removed by vacuum filtration. The support was air-dried, packed in a glass bulb, and purged with nitrogen for 2 h in the chromatographic oven at ambient temperature. The oven was programmed to 273 “C at 3 “C/min, held there for 16 h, and then cooled to room temperature. The packing was washed with methylene chloride and recovered by filtration several times 6 1980 American Chemical Society