Environ. Sci. Technol. 2005, 39, 2239-2246
Nonderivatization Analytical Method of Fatty Acids and cis-Pinonic Acid and Its Application in Ambient PM2.5 Aerosols in the Greater Vancouver Area in Canada YU CHENG AND SHAO-MENG LI* Meteorological Service of Canada, Environment Canada, 4905 Dufferin Street, Toronto, Ontario M3H 5T4, Canada
A nonderivatization analytical method has been developed to analyze C6-C20 fatty acids and cis-pinonic acid on a GC/ FID and a GC/MSD using a polar DB-FFAP capillary column. On the GC/FID, the response was highly linear over concentration ranges >2 orders of magnitude (R2 ) 1.00). Using a mixed solvent of dichloromethane (DCM): methanol (3:1, v/v) and an extraction temperature of 40 °C, the method recoveries of the acids from spiked filters were 81-115% based on deuterated surrogates, and the relative standard deviations were C14 were 93-100% in the particle phase. Hence, the results of the acids (>C14) are considered reliable while the quantitation of the shorter chain semi-volatile fatty acids (C6-C12) is subject to the uncertainties in the high-volume sampling. The sampled filters were stored at -20 °C with 3 mL of methanol to prevent growth of microbes (13, 38) until analysis in the laboratory. Field blanks were handled the same way as real samples with exposure to air for only several minutes. Sample Extraction, Separation and Concentration. The work concentrated on an optimized method for extraction of aerosol samples from appropriate filter media, separation, and follow-up analysis. Organic extractions from both spiked filters and filter samples from field studies were performed on an accelerated solvent extractor (ASE 200, Dionex Co.). Stainless steel extraction cells of 33 mL volume (Dionex Co.) were used. The cells have cellulose filters (Dionex Co.) at the outlet of each cell to contain potential debris from the samples. The extract was transferred to a collection vial (60 mL vials, Dionex Co.) after extraction. The ASE extraction pressure was set at 1500 psi, static time 5 min, flush volume 60%, purge time 200 s, air pressure 100 psi, and nitrogen pressure 150 psi. Filter samples were extracted at 40 °C with a mixture of DCM: methanol (3:1, v/v) and preheat times were 5 min. The extracts were blown down to 0.5 mL using a TurboVap II concentrator at 30 °C (39). One set of six duplicate spiked filters for the acids at 60 µg for each target compound was prepared to determine the recoveries of the acids during extraction using the ASE. Two sets of six duplicate filters for the acids spiked at 24 µg of each target compound were used to test the absolute and relative recoveries using the following validation method. The validation method was an optimized method for analyzing the acids from aerosol samples. It used the following settings: extraction on the ASE at 40 °C; blowdown on the TurboVap at 30 °C; acid fraction was eluted by a solvent mixture of ethyl acetate and methanol (3:1, v/v) and was finally blown down to a volume of 100 µL-2 mL depending on the target acid concentrations; other instrument settings remained the same as above. The analyses were carried out on a GC/FID and/or a GC/MSD. The absolute and relative recoveries were calculated in the following ways. In method A, the absolute recovery (Ra) was calculated as the ratio of the amount of materials determined from the analysis to the amount spiked onto the filters. In method B, two deuterated standards, added to the spiked filters as surrogates before extraction, were used to calculate the relative recovery (Rr), which is the ratio of the absolute recovery of a given species to the absolute recovery of the surrogates. Method B was applied to the aerosol samples in this study. The relative recovery Rr was then combined with the measured concentrations of the species and the surrogates to derive the final concentration of the species. Details of sample separation were presented elsewhere (39). Briefly, the extracts were separated into four fractions 2240
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 7, 2005
of different polarity using a glass chromatographic column packed with activated silica gel. The four fractions that contained saturated hydrocarbons (fraction 1), PAHs (fraction 2), ketones and esters (fraction 3), and acids including fatty acids and cis-pinonic acid (fraction 4) were eluted using hexane, a 6:4 (v/v) hexane/DCM mixture, a 5:1 (v/v) hexane/ ethyl acetate mixture, and a 3:1 (v/v) ethyl acetate/methanol mixture, successively. The final volumes of acid fractions were 100 and 400 µL for PM2.5 Interim RM and Air Particulate I (NIST intercomparison samples), respectively; 400 µL for forest samples from Vancouver; and 2 mL for tunnel samples from Vancouver. All the concentrated fractions were stored in a refrigerator at 8 °C. Acid Analysis. The acid analysis was performed on an Agilent 6890 GC/FID and a HP 5890 series II GC/HP 5972 MSD (Hewlett-Packard), respectively, without derivatization. A 30 m polar capillary column, DB-FFAP, 0.25 mm × 0.25 µm film thickness (J&W Scientific), was used for the separation in the GCs. The acid fraction of each sample was injected onto the column on the GC/FID. A 2-µL volume was injected in the splitless mode with a 4-min purge. The initial oven temperature was 70 °C and was held for 1 min. The oven temperature was programmed to increase to 205 °C at 6 °C/ min (hold for 3 min) and then to 240 °C at 7 °C /min and was held for 60 min. The inlet temperature was kept at 240 °C and carrier gas (He) flow was set at 2.5 mL/min. The detector was kept at 250 °C with the hydrogen flow maintained at 40 mL/min and airflow at 450 mL/min. For the GC/MSD analysis, electron impact was used with the mass range from 50 to 550. The flow of the carrier gas (He) was set at 1.5 mL/min to keep acceptable vacuum condition, while other operating conditions were the same as those on the GC/FID. Identification of unknown compounds and confirmation of the target compounds were carried out on the GC/MSD, but the quantification was carried on the GC/ FID because it has a better linearity of response. In this study, 19 authentic compounds from Sigma-Aldrich were used as external standards. They were hexanoic acid, heptanoic acid, octanoic acid, nonanoic acid, decanoic acid, undecanoic acid, dodecanoic acid (lauric acid), tridecanoic acid, tetradecanoic acid (myristic acid), pentadecanoic acid, hexadecanoic acid (palmitic acid), heptadecanoic acid, octadecanoic acid (stearic acid), eicosanoic acid, trans-9octadecenoic acid (elaidic acid), cis-9-octadecenoic acid (oleic acid), and cis-pinonic acid. Two deuterated acids, deuterated octanoic acid (C8-d15) and deuterated octadecanoic acid (C18-d35), were used as the internal standards to improve the accuracy of the analytical procedure from extraction to analysis. Calibration Linearity. The calibration curves of the acids on the GC/FID and the GC/MSD were obtained using mixtures of standards at 9 concentration levels ranging from 0.6 to 127 ng/µL. This range covers the sample concentrations seen in the field samples (see below). All compounds were separated completely, but oleic acid and elaidic acid coelute on both GC/MSD and GC/FID because the structures and characteristics of these two compounds are very similar. The correlation coefficients (R2) of the response versus concentration for the C6-C20 acids including cis-pinonic acid were consistently 1.00 on the GC/FID and from 0.87 to 1.00 on the GC/MSD. Blanks and Quality Control. Three types of blanks, including solvent blanks, filter blanks, and field blanks were prepared and analyzed using the validation method. No target compounds were found in the solvent blank using this nonderivatization method, but small amounts of the acids were found in the filter blanks. The acids in the filter blanks were C10) were not lost (Rbe ∼100%), but the volatile acids (C9 acids) (Table 1). Comparing Rbe and RsRbs, and since Rs must be C12 acids and 67% for cis-pinonic acid (RSDs 0.6-10.2%; Table 1). Since Rbe was 75% for C6, 90% for C9, and ∼100% for >C10 acids, one can deduce the values for Re to be 44% for C6 to 73% for >C12 acids. The extraction step obviously caused a large loss of the acids, particularly for the lighter acids. Thus, it appears that the acid recoveries for the entire method mainly depended on the extraction procedure. The large difference in the Re for the different acids suggests that there must be losses for the lighter acids during extraction. This most likely VOL. 39, NO. 7, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2241
TABLE 1. Recoveries and RSDs of Target Acids in Spiked Filters and Concentrations in PM2.5 Samples from NIST Using the Nonderivatization Methoda spiked filters ASE at 40 °C
compd name
abs recov (ReRbe)
hexanoic acid heptanoic acid octanoic acid nonanoic acid decanoic acid undecanoic acid dodecanoic acid tridecanoic acid tetradecanoic acid pentadecanoic acid hexadecanoic acid heptadecanoic acid octadecanoic acid eicosanoic acid oleic acid + elaidic acid cis-pinonic acid
33.0 35.4 39.5 60.9 65.5 70.6 72.6 73.4 72.9 72.5 72.4 71.9 71.1 69.8 69.4 67.2
sepn
RSD
abs recov (RsRbs)
7.9 6.7 5.3 10.2 5.2 3.8 2.7 1.8 1.1 0.8 0.6 0.9 2.0 1.2 1.9 2.8
83.8 92.0 93.1 95.8 98.8 96.4 92.6 98.6 97.8 95.4 94.1 93.9 92.5 91.7 92.3 95.1
NIST samples
method A
RSD
abs recov ( R)
5.7 3.5 9.1 9.0 8.8 7.7 7.1 7.1 6.7 6.1 5.2 4.4 4.2 5.5 5.6 5.8
32.2 39.4 43.9 45.5 50.7 54.5 56.1 62.7 64.5 66.7 66.8 66.9 68.5 65.3 65.7 63.0
method B PM2.5b
PM2.5c
RSD
rel recov (Rr)
RSD
concn
RSD
concn
RSD
6.0 4.3 8.5 11.8 0.7 0.8 2.9 3.6 1.7 0.5 2.4 1.9 1.7 8.0 3.2 0.8
94.2 104.8 115.1 81.1 89.0 95.1 102.5 104.1 105.4 104.3 104.7 101.9 101.7 90.6 93.0 95.3
11.8 8.8 7.0 5.9 3.8 4.0 4.3 4.8 4.8 4.9 5.1 5.2 5.3 5.3 9.1 2.3
9990 2880 4490 5720 3410 2370 4400 8090 11000 4710 170400 15000 87100 21600 21900 4040
8.7 3.1 14.8 21.6 24.8 13.9 19.1 7.3 17.9 23.9 17.4 18.0 17.5 14.6 9.1 13.8
121000 52400 12000 20900 12100 5990 19500 8650 27500 12000 91900 11700 52100 16300 na na
12.9 13.3 5.2 10.8 5.6 6.1 1.2 2.3 0.9 0.6 1.8 2.9 1.4 22.2 na na
a abs recov, absolute recovery %. rel recov, relative recovery %. sepn, separation. RSD, relative standard deviation %. R, R , R , R , and R e be s bs as defined in text. Method A, the absolute recovery as the ratio of measured concentration to spiked concentration (see text). Method B, the relative b recovery (Rr) with the surrogate standards added before extraction (see text). concn, concentration in ng/g. na, not available. Air Particulate I of NIST sample. c PM 2.5 Interim RM of NIST sample
occurred when the extracts were transferred from the extraction cell to the collection vials in the ASE. The recoveries ReRbe of the fatty acids extracted at 80 °C were 24% for C6 to 77% for >C12 acids; the RSDs increased up to 22%. While for the >C12 acids the ReRbe values were essentially the same at both extraction temperatures, the ReRbe values for the shorter chain acids were reduced. This indicates that the higher temperature increased the volatilities of the acids from C6 to C9 and caused additional loss during the transfer to the collection vials from the extraction cells in the ASE. Method Intercomparison using NIST PM2.5 Matrix. Our lab and three other labs (lab 4, lab 7, and lab 23) conducted hexadecanoic acid (palmitic acid) analysis in the interlaboratory comparison program organized by NIST (35). Two ambient PM2.5 matrixes from NIST, Air Particulate I (API) and PM 2.5 Interim RM (PIR), were analyzed in our lab using the nonderivatization method as described above. Three duplicates of the samples, about 150 mg for the API and 50 mg for the PIR, were quantified on the GC/FID and confirmed on the GC/MSD. The final volumes before GC analysis were concentrated to 400 µL for the API and 100 µL for the PIR to be within the working range of the calibration curves. The concentrations of the target acids were from 2880 to 170 400 ng/g for the API with four-fifths of RSDs
