Development of a Solid-Phase Microextraction GC-NPD Procedure for

Sandy Owega, Badi-Uz-Zaman Khan, Ryan D'Souza, Greg J. Evans, Mike Fila, .... Robert M. Healy , Greg J. Evans , Michael Murphy , Berko Sierau , Jovann...
2 downloads 0 Views 165KB Size
Anal. Chem. 1999, 71, 3531-3537

Development of a Solid-Phase Microextraction GC-NPD Procedure for the Determination of Free Volatile Amines in Wastewater and Sewage-Polluted Waters Manuela A Ä balos,† Josep M. Bayona,*,† and Francesc Ventura‡

Environmental Chemistry Department, IIQAB-CSIC, Jordi Girona 18-26, E-08034 Barcelona, Spain, and Aigu¨es de Barcelona (AGBAR), P. Sant Joan 39, E-08009, Barcelona, Spain

An analytical procedure for the determination of free volatile C1-C6 amines in aqueous matrixes has been developed and applied to their determination in wastewater, primary and secondary effluents, and sewagepolluted river samples. The developed analytical procedure involves headspace sampling using solid-phase microextraction with a poly(dimethylpolysiloxane) coating (100 µm) followed by GC-NPD determination and GC/ MS confirmation using a tailor-made PoraPLOT amines capillary GC column for volatile amines. Procedural detection limits were compound dependent but ranged from 3 to 56 µg L-1, being close to or lower than the odor threshold concentration, and the reproducibility was ca. 15% (N ) 5) in real water samples. The developed analytical procedure is solvent free, cost-effective (no cryogenic trap needed), and faster than existing methods because no derivatization step is involved in the determination. Linearity was compound dependent but ranged at least from 50 to 600 µg L-1. Control of odors in wastewater treatment plant effluents and in contaminated surface water has received a great deal of attention because they give rise to social alarm and hygienic problems. Volatile amines, among other volatile compounds, are responsible for causing unpleasant odors.1 Aliphatic and aromatic amines may occur as biodegradation products of organic matter like proteins, amino acids, and other nitrogen-containing organic compounds. In addition, amines are used as raw materials or as intermediates in the manufacture of a wide range of industrial chemicals.2 Volatile amines not only have an unpleasant smell but also possess health hazards. Moreover, they may react with nitrosating agents, leading to the formation of potentially carcinogenic N-nitrosamines.2-4 * Corresponding author. E-mail: [email protected]. Fax: 34-93-2045904. † IIQAB-CSIC. ‡ AGBAR. (1) Hwang, Y.; Matsuo, T.; Hanaki, K.; Suzuki, N. Water. Res. 1995, 29, 711718. (2) Sacher, F.; Lenz, S.; Brauch, H. J. J. Chromatogr., A 1997, 764, 85-93. (3) Schade, G. W.; Crutzen, P. J. J. Atmos. Chem. 1995, 22, 319-346. (4) Terashi, A.; Hanada, Y.; Kido, A.; Shinohara, R. J. Chromatogr. 1990, 503, 369-375. 10.1021/ac990197h CCC: $18.00 Published on Web 07/20/1999

© 1999 American Chemical Society

Analysis of low-molecular-weight amines has been traditionally difficult due to their particular physicochemical properties, i.e., high volatility and polarity, basic character, and high solubility in water (Table 1). Up to now, existing analytical procedures are only partially successful. In fact, GC determination of these amines in their free forms following the addition of alkali suffers from adsorption and decomposition of the amines in the chromatographic column, giving rise to tailing peaks, ghosting phenomena and low detector sensitivity.4-7 Derivatization overcomes some of these problems by the formation of less polar compounds which can be more easily analyzed by GC2,4,7-9 or LC.1,10 Nevertheless, these procedures include time-consuming derivatization steps, resulting in a dramatic increase of analysis time, which is not suitable for monitoring studies. Several preconcentration systems have been evaluated for volatile amine determinations in seawater or weakly polluted waters, including microdiffusion,11 hydrophobic membranes,12,13 and selective purge and trap.9 However, these analytical techniques are not useful to determine volatile amines from wastewaters because their high suspended solid content and volatile amines occur at high concentrations, making it unnecessary to reach high preconcentration factors. Solid-phase microextraction (SPME) is an attractive alternative to conventional analytical methods for the determination of volatile compounds from aqueous samples because it is a fast, simple, low-cost, solventfree technique. The purpose of this study was to develop a headspace SPME/ GC-NPD methodology for the determination of free volatile amines (C1-C6) in aqueous matrixes. Consequently, a tailor-made capil(5) Hoshika, Y. J. Chromatogr. 1975, 115, 596-601. (6) Pfundstein, B.; Tricker, A. R.; Pressmann, R. J. Chromatogr. 1991, 539, 141-148. (7) Da Costa, K. A.; Vrbanac, J. J.; Zeisel, S. H. Anal. Biochem. 1990, 187, 234-239. (8) Kataoka, H. J. Chromatogr., A 1996, 733, 19-34. (9) Scully, F. E.; Howell, G. D.; Penn, H. H.; Mazina, K.; Johnson, J. D. Environ. Sci. Technol. 1988, 22, 1186-1190. (10) Beale, S. C.; Savage, J. C.; Wiesler, D.; Wietstock, S. M.; Novotny, M. Anal. Chem. 1988, 60, 1765-1769. (11) Abdul-Rashid, M. K.; Riley, J. P.; Fitzsimons, M. F.; Wolff, G. A. Anal. Chim. Acta 1991, 252, 223-226. (12) Gibb, S. W.; Mantoura, R. F.; Liss, P. S. Anal. Chim. Acta 1995, 316, 291304. (13) Yang, X. H.; Lee, C.; Scranton, M. I. Anal. Chem. 1993, 65, 572-576.

Analytical Chemistry, Vol. 71, No. 16, August 15, 1999 3531

Table 1. Physicochemical Properties of Volatile Amines parameter solubility (mg (25 °C) log Kow dipole moment (D) pKa(25 °C) Henry constant (dimensionless) odor threshold concn (OTC) (ppb) L-1)

a

MMA naa

-0.57b 1.31 10.6 1.5 × 10-3 20

DMA na -0.84c 1.03 10.7 1.3 × 10-3 47

TMA na 0.27b 0.61 9.8 2.3 × 10-3 0.2

MEA na -1.04c 1.22 10.8 na 39

DEA na 0.57b 0.92 10.5 na na

TEA 10-4

7.3 × 1.44b na 11.0 5.4 × 10-4 na

ref 23 25 15 15, 24 3, 23 26

Not available. b Measured. c Estimated.

lary column (PoraPLOT amines) for high volatile amines was evaluated for their determination. To the best of our knowledge, only one paper related to the analysis of derivatized aliphatic amines by using SPME has been published.14 In the present work, several extraction variables such as coating type, equilibrium, desorption time and temperature, and sample volume were optimized. Linearity, detection limits, and precision of the whole procedure were determined. Finally, to evaluate the suitability of the developed method for real sample application, different wastewater matrixes were analyzed: (i) sewage-contaminated surface waters and (ii) influents and effluents from different treatment plants. EXPERIMENTAL SECTION Chemicals and Materials. Methylamine (MMA, 99%), dimethylamine (DMA, 99%), trimethylamine (TMA, 98%), ethylamine (MEA, 98%), diethylamine (DEA, 99%), and triethylamine (TEA, 98%) as hydrochlorides were obtained from Aldrich-Chemie (Steinheim, Germany). Propylamine (99%), isopropylamine (99%), cyclopropylamine (98%), n-butylamine (99%), sec-butylamine (99%), and 3-methylbutylamine (99%) were also supplied by AldrichChemie. tert-Butylamine (99.5%) and 2-methylbutylamine (97%) were purchased from Aldrich Chemical (Milwaukee, WI). All standards were used as received. Analytical grade HCl (25%) was from Merck (Darmstadt, Germany), and NaOH was from Carlo Erba (Milan, Italy). The SPME holder and coated fibers (poly(dimethylsiloxane), PDMS, 100 µm; Carbowax-divinylbenzene, CW-DVB, 65 µm; polyacrylate, PA, 85 µm; and Carboxen-PDMS, 75 µm) were obtained from Supelco (Bellefonte, PA). Stock standard solutions of each analyte (4000 µg mL-1) were prepared in Milli-Q water acidified up to pH 1-2 with HCl (25%) in order to have the hydrochloride nonvolatile forms of the amines. Standard solutions to prepare spiked samples were obtained by diluting the stocks with acidified Milli-Q water. All the standard solutions were stored at 4 °C in the darkness. Influents and effluents from five wastewater treatment plants and a sewage-polluted surface water were analyzed using the developed method. Samples were acidified until pH 1-2 and stored in amber glass bottles, avoiding the presence of headspaces at the tops of the bottles. All analyses were performed within 4 days after sampling. SPME Procedure. Two types of experiments were carried out. First, a nonacidified aqueous solution containing 500 µg mL-1 each of MMA, DMA, TMA, DEA, and TEA in their hydrochloride (14) Pan, L.; Chong, J. M.; Pawliszyn, J. J. Chromatogr., A 1997, 773, 249-260.

3532 Analytical Chemistry, Vol. 71, No. 16, August 15, 1999

forms was used for studying a two-phase equilibrium (gas-coating). A volume of 20 µL of this solution was introduced into a 40 mL vial sealed with an PTFE-faced silicone septum. Then 50 µL of NaOH (2 M) was injected trough the septum in order to obtain the volatile amines in a basic medium. Under these conditions, the liquid volume was negligible compared to the gas phase. Therefore, a two-phase equilibrium could be considered. The extraction temperature was held at 27 ( 2 °C using a water bath. Four different coatings were tested, the equilibrium time profile was obtained for each coating, and two different desorption temperatures (220, 250 °C) and three desorption times (1, 3, 5 min) were also evaluated. All experiments were performed in duplicate, and the average values are reported. Headspace SPME analysis of spiked and unspiked water samples was considered at the second stage. An acidified aqueous mixture containing 40 µg mL-1 each of MMA, DMA, TMA, DEA, TEA, propylamine, isopropylamine, n-butylamine, sec-butylamine, tert-butylamine, and 2-methylbutylamine was used as the spike solution. Cyclopropylamine was selected as the surrogate. A 20 mL water sample was placed in a 40 mL vial sealed with an PTFEfaced silicone septum. After addition of 600 µL of NaOH (5 M), in the same way that it has been previously described, headspace SPME was carried out within 30 min under magnetic stirring of the liquid phase. The temperature was also held at 27 ( 2 °C. Chromatographic Determination. Two-phase experiments were performed in a 5000 Mega Series GC-FID apparatus (Fisons, Milan, Italy). A 1 µL direct injection of the aqueous amine solution in the split mode (1:50) was carried out while the injector temperature was held at 220 °C. The remaining injections were performed in the splitless mode using SPME and activating the injector purge valves at 3 min from the injection. A GC Fisons Mega 2 Series apparatus equipped with an NPD 800 instrument operating at 3.2 A and 3.5 V was used for the analysis of volatile amines in aqueous samples. A PoraPLOT amines column, 30 m length × 0.32 mm i.d., of 10 µm film thickness (Chrompack) was used in both cases. Helium was used as the carrier gas at a 1.8 mL min-1 flow rate. The operating flow rate was achieved by slowly pressurizing the column (ca. 0.1 bar s-1). Similarly, the inlet column pressure was slowly depressurized. Hydrogen and air at 4 and 40 mL min-1 were used, respectively, as fuel and comburent gases. Helium at 18 mL min-1 was the makeup gas. The initial column temperature (60 °C) was held for 5 min and then ramped at 10 °C min-1 to 220 °C, after which it was held for 15 min. Desorption time and injector port temperature were set at the optimum values (see Results and Discussion). Quantification of real water samples was performed by GC-NPD using cyclo-

RESULTS AND DISCUSSION The column selected in this study (PoraPLOT amines) was a polymeric porous layer open-tubular column, tailor-made for volatile amine determination.15 Porous polymeric phases coated over open capillary columns give rise to reasonably high efficiency and are useful in those cases where liquid phases do not provide sufficient retention, such as for very volatile amines.16 Methylbutyland n-butylamine were not considered in this study because they exhibited too long retention times in the chromatographic column used, leading to broad peak shapes. All the analytes were baseline resolved except TMA and MEA, which exhibited partial coelution. Therefore, the target analytes in this study were MMA, DMA, TMA, DEA, TEA, and propyl-, and isopropyl-, sec-butyl- and tertbutylamine. Figure 1 is a GC-FID chromatogram of the standard mixture of volatile amines showing a satisfactory retention of the analytes of interest and acceptable peak shapes.

Method Development. (a) Equilibrium Time Profiles. In SPME, equilibration time depends on both compound and coating characteristics. In this study, extraction time profiles were obtained for each of the coatings in two-phase experiments using a standard solution containing MMA, DMA, TMA, DEA, and TEA (Figure 2). Four different coatings were evaluated: a nonpolar PDMS phase, two polar coatings (CW-DVB and PA), and CarboxenPDMS, a phase specifically designed for volatile compounds. Amines are polar basic substances (Table 1) that might be extracted by using polar coatings. Due to their high volatility, other coating characteristics (i.e., adsorption surface and thickness) have to be considered. CW-DVB exhibited the shortest equilibration time (2 min) among the coatings. This fact could be explained by considering that extraction is mainly controlled by adsorption mechanisms on the DVB highly porous surface area. Conversely, surface adsorption combined with diffusion into the inner coating must be considered to explain the lower extraction kinetics in PA and PDMS coatings. When PDMS coating was used, equilibrium was reached for most of the amines within 10 min. However, for the PA coating, equilibrium was still not reached at the maximum tested time (30 min). These results agreed with the fact that PA is a low-density solid polymer leading to lower diffusion coefficients of volatile analytes into the coating compared to those observed for the PDMS liquid phase which give rise to longer extraction times.17 The Carboxen-PDMS coating is characterized by a very small pore size.18 This structure makes the mass transfer kinetics slower than those in the CW-DVB coating. Consequently, a 5 min extraction was necessary to reach the equilibrium for the lowest molecular weight compounds (MMA and DMA) while for DEA and TEA the equilibrium was established within 20 min in Carboxen-PDMS. A singular behavior was noticed for the extraction time profile of TMA in this coating (Figure 2d, insert). The maximum area response was obtained after 5 min; however, a steep decrease to half of the response was observed for this compound at the equilibrium time of DEA and TEA. A displacement of small molecules by the larger ones may take place. (b) Selection of Coating in SPME. A long sampling time (30 min) was used to select the coating. Although extraction conditions were not at equilibrium for some coatings, they were applied as a compromise between extraction time and extraction efficiency. As expected, higher responses were obtained for ethylamines than for methylamines according to their higher lipophilicity and higher vapor pressure (Table 1); however, large differences in the extraction kinetics were found among coatings (Figure 3). Polar phases such as the CW-DVB exhibited an extremely broad interference peak close the TMA retention time that precluded the TMA determination. The PA coating gave the smallest peak areas among the evaluated coatings (3 times lower of those obtained with CW-DVB). However, results shown in Figure 3 for the PA coating correspond to a nonequilibrium situation (see Figure 2). From comparison of the two others coatings, it was concluded that Carboxen-PDMS showed areas 4-10 times higher than those obtained with PDMS, except for the trisubstituted amines, for which responses were similar in both

(15) Mohnke, M.; Schmidt, B.; Schmidt, R.; Buijten, J. C.; Mussche, Ph. J. Chromatogr., A 1994, 667, 334-339. (16) De Zeeuw, J.; de Nijs, R. C. M.; Buijten, J. C.; Peene, J. A.; Mohnke, M. J. High Resolut. Chromatogr. 1988, 11, 162-167.

(17) Pawliszyn, J. Solid-Phase Microextraction: Theory and Practice; Wiley-VCH: New York, 1997. (18) Shirey, R. E.; Mani, V. Paper presented at the Pittcon Conference, Atlanta, GA, 1997.

Figure 1. GC-NPD chromatograms of the standard mixture of volatile amines: (a) 1 µL aqueous direct injection in the split injection mode (1:50); (b) headspace SPME with PDMS (100 µm) in the splitless injection mode. Compound identification: 1, MMA; 2, DMA; 3, TMA; 4, DEA; 5, TEA.

propylamine as the internal standard. All the determinations were performed in duplicate except the evaluation of precision, which was performed in five replicates. The linearity of the method was investigated over the 50-600 ng mL-1 range expressed as the initial concentration in water. Real samples were diluted if necessary in order to fall within the linear dynamic range. Detection limits were calculated from a procedural blank as the concentration corresponding to 3 times the signal-to-noise ratio. Confirmation of analytes in real samples was performed by GC/ MS in the EI mode using an MD 800 instrument (Fisons).

Analytical Chemistry, Vol. 71, No. 16, August 15, 1999

3533

Figure 2. Extraction time profiles of different coatings obtained from a two-phase experiment performed in duplicate (see Experimental Section). Range and mean values are indicated at each experimental point as a vertical line. Coatings evaluated were (a) PDMS, (b) Carbowax-DVB, (c) polyacrylate, and (d) Carboxen-PDMS. The inserts included in parts a and d correspond to expanded views of the shaded areas. Table 2. Kd Values (mL) vs Initial Concentrations in Two-Phase Experiments Using a PDMS (100 µM) Coating spiked amt in gas phase (µg)

Figure 3. Comparison of the extraction efficiencies of spiked volatile amines according to the coatings in two-phase experiments (see Experimental Section). Extraction time was 30 min, and determinations were performed in duplicate.

cases. The higher peak area for the trialkyl-substituted amines in PDMS are consistent with the higher lipophilicity of these compounds compared to the mono- and disubstituted homologues. Similar results have been reported on the SPME of methyltin19 and alkylmercury20 using the PDMS coating. Although the best response was obtained for the most polar compound (MMA) with the CW-DVB coating, on the whole, (19) Morcillo, Y.; Cai., Y.; Bayona, J. M. J. High Resolut. Chromatogr. 1995, 18, 767-770. (20) Cai. Y.; Bayona, J. M. J. Chromatogr., A 1995, 696, 113-122.

3534 Analytical Chemistry, Vol. 71, No. 16, August 15, 1999

compound

10

1

0.1

MMA DMA TMA DEA TEA

0.052 0.180 0.262 0.819 2.279

0.107 0.356 0.578 1.809 4.978

0.103 0.227 0.933 1.554 5.472

polar coatings were found to be less effective for the extraction, particularly for DEA and TEA. Carboxen-PDMS was the most appropriate coating for the extraction of the five amines on the basis of the area responses. However, due to the high capability of that coating to extract very volatile compounds, some additional peaks appeared on the FID or NPD trace of standard mixtures complicating its application to wastewater analysis. (c) Optimization of Desorption Conditions. Two desorption temperatures, 220 and 250 °C, were evaluated for the PDMS, PA, and CW-DVB coatings. Desorption time was set at 5 min. Results showed that there were no differences in the area responses between the two temperatures for both coatings. The RSDs of the mean values for each temperature were below 7%, and they were similar to the experimental error calculated among replicates performed at the same temperature. In addition, carryover was not detected either at 220 °C or at 250 °C; therefore, the lower

Figure 4. Effect of the aqueous sample volume in the extracted amount of volatile amines. The coating was PDMS (100 µm). The area of TEA corresponds to the right scale, and those of the remaining compounds correspond to the left.

temperature was chosen for further experiments. For the Carboxen-PDMS coating, only the higher temperature was considered because it was only 10 °C above the minimum recommended temperature for this coating. Desorption time for the PDMS and Carboxen-PDMS coatings was optimized in a second stage. Three different desorption times, 1, 3, and 5 min, were evaluated. As happened with the desorption temperature, the responses among the evaluated times were not significantly different (RSD ca. 10%). Low desorption times might cause higher variability in the response if the process is not carefully controlled. On the contrary, an unnecessary increase in desorption time can shorten coating lifetime. A compromise value of 3 min desorption time was selected for the subsequent experiments. (d) Estimation of Kd. Coating/gas distribution constants (Kd) of MMA, DMA, TMA, DEA, and TEA were estimated for the PDMS coating. Kd can be calculated from

Kd ) Cfeq/Cgeq where Cfeq and Cgeq are the analyte concentrations in the coating and in the gas phase at equilibrium, respectively. However, in the present work, Kd was calculated as

Kd ) nfeq/Cgeq where nfeq is the total amount of analyte extracted into the coating at the equilibrium because the fiber volume is difficult to measure and it is constant throughout the experiments. Kd was then expressed in milliliters. Calculation of the amount of each analyte extracted into the coating (in nanograms) was based on a response factor obtained from direct injections of the aqueous mixture of the amines. The concentration of each analyte in the gas phase at equilibrium (in nanograms per milliliter) was calculated as the

Table 3. Linear Dynamic Ranges, Response Factors, and Procedural LODs of Volatile Amine Determinations by GC-NPD Using Headspace SPME with the PDMS Coating compd

linear range (µg L-1)

response factor

LOD (µg L-1)

MMA DMA TMA DEA TEA propylamine isopropylamine sec-butylamine tert-butylamine

32-686 20-608 47-563 52-611 60-714 66-492 105-546 46-542 51-324

0.7 2.4 8.7 8.3 46 2.4 1.7 3.4 1.4

27 21 11 3 14 56 nq 14 24

a

Not quantifiable.

total amount of compound introduced into the vial minus the amount introduced into the coating, the difference then being divided by the vial volume. Theoretically, Kd (s) should be independent of the spiked concentration. In this study, three different initial concentrations were considered. Calculated values of Kd(s) are reported in Table 2. Surprisingly, a concentration dependence was found for Kd(s) decreasing down to half of its value upon increasing the spiked amount. This concentration effect on Kd was previously reported in a study of PDMS coating performance for a broad spectrum of analytes with different polarities showing variations in Kd for hydroxy- and nitrogen-containing compounds.21 The Kd concentration behavior can be ascribed to the low solubility of the highly polar analytes in the coating leading to nonlinear partitioning isotherms at high concentrations in the gas phase. (e) Analysis of Spiked Water Samples. The results obtained during the optimization of two-phase experiments were heeded for the analysis of a wider group of amines from the headspaces (21) Bartelt, R. J. Anal. Chem. 1997, 69, 364-372.

Analytical Chemistry, Vol. 71, No. 16, August 15, 1999

3535

Table 4. Fate of Volatile Amines in Wastewater Treatment Plants amt (mg L-1) treatment plant

equiv pop.

treatment process

1 2 3 4 5

110 000 223 000 80 000 180 000 36 000

primary secondary secondary primary secondary

a

amt (µg L-1)

nitrogen BOD5 influenta effluenta influent effluent influent effluent MMA DMA TMAb DEA TEA MMA DMA TMAb DEA TEA