Storage Media

Jan 28, 2015 - A gastight syringe (GTS) is commonly used as a medium for transfer or storage of gaseous standards (or samples) in the analysis of vola...
1 downloads 5 Views 5MB Size
Download PDF
Article pubs.acs.org/ac

Test on the Reliability of Gastight Syringes as Transfer/Storage Media for Gaseous VOC Analysis: The Extent of VOC Sorption between the Inner Needle and a Glass Wall Surface Yong-Hyun Kim and Ki-Hyun Kim* Department of Civil and Environmental Engineering, Hanyang University, 222 Wangsimni-Ro, Seoul 133-791, Korea S Supporting Information *

ABSTRACT: A gastight syringe (GTS) is commonly used as a medium for transfer or storage of gaseous standards (or samples) in the analysis of volatile organic compounds (VOCs). In this study, the sorptive loss patterns of 21 VOCs were examined, using GTS as the transfer medium. The results of the test were evaluated with respect to a number of key variables including concentration, sampling volume, and physicochemical properties (molecular weight and boiling point). The VOCs with relatively high volatility (Group 1: aldehyde, ketone, ester, alcohol, and aromatic hydrocarbons (n = 12)) showed low sorptive losses with a mean (±SD) of 2.56 ± 2.87%, regardless of differences in the aforementioned key variables (p-value by t-test before and after using GTS = mean 0.15 ± 0.13). Conversely, the sorptive losses of seven semi-VOCs (Group 2: carboxyl and cresol (n = 9)) were significantly high, ranging from 18.0 ± 4.10% (propionic acid) to 65.4 ± 10.9% (nheptanonic acid). In addition, we also measured the sorptive losses on the syringe needle (mean sorptive loss of Group 2 = 5.94 ± 5.63%). A linear regression analysis showed that the sorptive losses for Group 2 increased as molecular weight (or boiling point) increased, exhibiting a highly significant correlation (R2 value (0.804 ± 0.084) and mean p-value (0.002 ± 0.003).

V

system by transferring the target analytes from one bag to another. Accordingly, loss of the four VFAs occurred at a rate of 40% or more for each transfer across a supposedly inert bag material (i.e., 1 L polyvinyl fluoride Tedlar bags). Kim et al.10 reported that the recoveries of VFAs (propionic acid ∼nheptanoic acid) from a 25 mL glass vial decreased significantly with increases in molecular weight (60−90%). As the use of a transfer media, such as a gastight syringe, is necessary when injecting or transferring analytes, it is important to be able to assess biases involved in the various steps of analyte quantitation. In this research, we studied the sorptive loss properties of several VOCs, using gastight syringes as the transfer/storage media. To accurately assess the sorptive loss of VOC occurring due to the use of gastight syringe, we took a number of key variables for this investigation such as type of compounds, concentration levels, and collection volumes. To this end, the analysis involved a total of 21 VOCs with low to strong sorptive reactivities (methyl isobutyl ketone to VFA (and phenol)) (Table S-1, Supporting Information).8,10,11 The sorbent tube (ST) method has been shown to be the optimal sampling device for capturing VOCs with varying reactivity ranges.8 Thus, the relative sorptive loss of each target VOC in a GTS made of borosilicate was evaluated based on the ST method. In addition, we also assessed the sorptive loss of VOC occurring in

olatile organic compounds (VOCs) are more commonly known as hazardous pollutants or malodorous substances.1 Benzene and formaldehyde are considered to be carcinogenic, whereas aldehydes and volatile fatty acids (VFAs) are regulated as offensive odorants.2−4 For these reasons, many governments have established diverse guidelines in an effort to effectively regulate such emissions, in both ambient and source conditions. In order to systematically control the VOCs in the air, it is important to be able to accurately analyze VOCs by reducing the biases that exist in the various quantification methods (e.g., sampling, pretreatments, and calibration). There are many pretreatment options for the collection and analysis of VOC samples, including solvent absorption, bag sampling, and sorbent adsorption.3 To reduce the possible sources of biases, it is important to identify potential biases in both the collection procedures as well as the analytical procedures. In the case of bag sampling, gaseous VOC can be subject to sorptive or leakage loss depending on the properties of the bag material or sampling valve.5,6 If the gaseous VOCs are absorbed on a solvent, large biases can occur during the VOC extraction procedure from absorbed solution.7 In the case of the sorbent adsorption, it is important to consider the efficiency of adsorption and desorption to reduce the potential biases.8 In particular, in the case of semi- or low volatile compounds, considerable sorptive losses can occur in each step of the sampling and pretreatment processes. Ahn et al.9 investigated the loss rates of four VFAs (propionic acid, nbutyric acid, i-valeric acid, and n-valeric acid) in a bag sampling © XXXX American Chemical Society

Received: December 18, 2014 Accepted: January 28, 2015

A

DOI: 10.1021/ac504713y Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 1. Flowchart of the experimental procedures to assess relative sorptive loss of gaseous VOCs with different experiment modes (sampling approaches).

S-1, Supporting Information): (1) two aldehydes, i-valeraldehyde (IA) and n-valeraldehyde (VA); (2) two ketones, methyl ethyl ketone (MEK) and methyl isobutyl ketone (MIBK); (3) one ester, n-butyl acetate (BuAc); (4) one alcohol, i-butyl alcohol (i-BuAl); (5) six aromatics, benzene (B), toluene (T), p-xylene (p-X), m-xylene (m-X), o-xylene (o-X), and styrene (S); (6) seven carboxyls, propionic acid (PPA), i-butyric acid (IBA), n-butyric acid (BTA), i-valeric acid (IVA), n-valeric acid (VLA), n-hexanoic acid (HXA), and n-heptanoic acid (HPA); (7) two phenols, o-cresol (o-C) and m-cresol (m-C). The standards for all of the target VOCs were prepared in both liquid and gas phases. The liquid-phase working standards (LWS), which contained all 21 of the target VOCs, were prepared

a stainless steel needle portion of GTS (along with bag sampler).12 If one considers the fact that the analytical uncertainty induced by the use of the gastight syringe has not commonly been accounted for in the quantitative analysis of VOCs, the results of this study will provide valuable insights into balancing all types of sorptive loss encountered in all different stages of VOC quantitation.

■

MATERIALS AND METHODS Preparation of Both the Liquid and Gaseous Working Standards. To assess the sorptive losses of VOCs in a GTS (and bag sampler) as the transfer/storage media, a total of 21 VOCs were selected as the target analytes in this study (Table B

DOI: 10.1021/ac504713y Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry Table 1. Basic Experimental Scheme to Evaluate the Relative Sorptive Loss (RSL, %) of VOCs

using a dilution of the reagent grade chemicals (RGC, SigmaAldrich, USA) in methanol (≥99.8%, Burdick & Jackson, USA). The calibration data were obtained through analysis of the LWS using the ST approach. The gaseous working standards (GWS), although less reliable for semivolatile species, were prepared via vaporization of the L-WS in a polyester aluminum (PEA) sampler bag. The relative sorptive losses of the target VOCs were assessed in terms of transfer/storage capabilities using G-WS samples. The detailed description of standard preparation is provided in the Supporting Information (Table

S-2). In addition, the details of our instrumental system are also provided in the Supporting Information (Table S-3). Experimental Approaches. In this study, the sorptive losses of the VOCs from a gastight syringe were assessed according to various characteristics (i.e., syringe size, needle installation, VOC concentration, sampling volume). In addition, the sorptive loss in a 1 L PEA bag was also evaluated. As shown in Figure 1, a total of four types of experimental procedures (different schemes for sample transfer) were designed and employed to test the sorptive loss in the GTS (or bag sampler). For all four types of experiments, the G-WS C

DOI: 10.1021/ac504713y Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

■

Article

RESULTS AND DISCUSSION Recovery of VOCs from the G-WS against the L-WS. To assess the relative sorptive loss (RSL) of the VOCs in the GTS, the G-WSs that were prepared in the PEA bag were analyzed after being transferred to the ST via four pathways (Experiment modes 1−4) (Table 1). For the VOCs with strong sorptive reactivity (like VFAs), it was virtually impossible to prepare gaseous standards in a stable form with any type of container (e.g., cylinder) due to their strong reactivity with the inner wall material.3,10,11 For this reason, the G-WS for all of the 21 target VOCs was prepared in the laboratory by vaporization of the L-WS. Despite this precaution, there was a significant loss of certain target compounds (e.g., Group 2) after the conversion. Although the aim of this study was to analyze the relative loss of gas phase VOC due to the various transfer steps used for each of the experimental modes, it is also important to determine the extent of conversion between liquid and gas phases. To accurately evaluate the sorptive loss patterns of VOCs in each step of the experiment, the efficiency of transition between the two standard phases was assessed by comparing the response factor (RF) values derived from a calibration analysis of the G-WS against those of the L-WS. The actual concentrations of the G-WS were then estimated using the relative recoveries against those of the more reliable L-WS. All of the target VOCs (n = 21) had fairly good linearity, exceeding 0.99 in both the L-WS and G-WS analyses. For the calibration of the G-WS, the analysis involved loading three different volumes ((1) 10 mL, (2) 50 mL, and (3) 100 mL) of the G-WS with four different concentration levels (ideal concentrations: (1) 52.5 ppb, (2) 105 ppb, (3) 262 ppb, and (4) 525 ppb) in the ST. As a result, a total of seven calibration RF values for each target VOC were obtained using the fixed standard volume (FSV) and fixed standard concentration (FSC) approaches.13 The RF values (ng−1) of each VOC obtained using the direct ST method (Exp 1) were similar, regardless of concentration and sampling volume (mean relative standard error (RSE) = 1.63 ± 0.90% (n = 21)) (Table S-5, Supporting Information). The relative recovery (or relative loss) of each VOC in the vaporized standard gas was calculated by comparing the RF values between the G-WS (Exp 1: direct ST) and the L-WS analyses (Table S-5, Supporting Information). The actual concentration of each G-WS was predicted using the results of the relative recovery. The VOCs in Group 1 had a relative recovery mean of 67.1 ± 5.08% (n = 12), while the recorded mean recovery for Group 2 was 48.2 ± 10.1% (n = 9). In line with the general expectation, the transition rate (L-WS to GWS) for Group 2 was much lower than that of Group 1 due to the enhanced reactivity with all surface materials (quartz trap, PEA bag, Teflon tubing, and other surfaces) during transfer after vaporization. In addition, loss may also have occurred due to diffusion caused by the high temperature setting (230 °C) needed for the quartz trap to induce vaporization of the L-WS. For the Group 2 compounds, significantly reduced recovery rates (below 50%) were recorded because of their high sorptive reactivity (i.e., high boiling point > 150 °C). Conversely, the enhanced recovery of Group 1, which had low boiling points generally below 100 °C, can be attributed to the reduced diffusion loss due to the vaporization set up of the experiment (230 °C). The actual concentration of each G-WS was calculated using the relative recovery of each target VOC: (1)

was prepared identically in 1 L PEA bags. In Experiment 1, the G-WS was transferred directly to the ST (direct ST experiment). In Exp 2, the G-WS was initially collected without a needle using two different volumes of GTS (100 mL in Experiment 2A and 10 mL in Experiment 2B) and were then transferred to the ST for quantification. To assess the sorptive loss that occurred on the inner wall of the stainless steel needle, the G-WS was collected using a 100 mL gastight syringe equipped with a needle and then injected into the ST (Experiment 3:100 mL syringe (plus a needle) to the ST). In the last stage, the sorptive loss was analyzed after a single transfer of the G-WS to a new 1 L PEA bag, without the use of a syringe (Experiment 4: bag to the ST).12 To transfer the G-WS that was contained in the 1 L PEA bag or the GTS to the ST, the following steps were taken. The inlet and outlet of the ST were connected to a GTS (or 1 L PEA bag), filled with the G-WS, and the vacuum pump was interfaced with a mass flow controller (Sibata ΣMP-30, Japan). The sampling flow rate was fixed at 100 mL min−1. The standard sample loading volumes for all of the experiments were adjusted to one of three values, 10, 50, and 100 mL, using a 100 mL GTS (Table 1). However, in Experiment 2B, these values were reduced to 2, 5, and 10 mL. The calibration and quality assurance (QA) data were obtained from an analysis of the L-WS. In order to analyze the L-WS, the inlet of the ST was connected via Teflon tubing to a 1 L PEA bag filled with ultrapure nitrogen (>99.999%). The outlet of the ST was then connected to a Sibata vacuum pump (ΣMP-30, Japan). Then, 1 μL of the Final L-WS was injected onto the ST using a 5 μL liquid syringe via a temporary injection port connecting the inlet of the ST with the PEA bag, while supplying back-up gas (from the PEA bag) to the ST (flow rate = 100 mL min−1 for 5 min). The basic experimental information is presented in Table 1. The representative samples of the extracted ion chromatograms obtained from each experimental mode are shown in Figure S-1 (Supporting Information). Quality Assurance/Quality Control (QA/QC) of VOC Analysis with ST/TD-GC−MS System. In this study, the basic quality assurance of ST method was assessed primarily in terms of relative standard errors (RSE (%)) and method detection limits (MDL). The RSE values of ST method were calculated using triplicate analyses of liquid working standard prepared for the third point calibration (mean ± SD concentration: (1) Group 1 = 16.1 ± 0.82 ng μL−1 and (2) Group 2 = 26.6 ± 1.31 ng μL−1) through direct injection approach (analytical volume = 1 μL) (refer to the Experimental Approaches section). According to this analysis, all target VOCs exhibited fairly good reproducibilities with RSE values of 5% (mean ± SD RSE: (1) Group 1 = 0.92 ± 0.32% and (2) Group 2 = 3.05 ± 1.20%). The MDL values were determined by seven repeat analyses of the diluted liquid working standard (mean ± SD concentration: (1) Group 1 = 0.64 ± 0.03 ng μL−1 and (2) Group 2 = 1.06 ± 0.05 ng μL−1) which was made by 5-fold dilution of the first-point calibration standard (Table S-4, Supporting Information). The MDL values of 21 target VOCs were significantly low (range of mean MDL values = 0.023 ± 0.001 ng (T) to 0.541 ± 0.084 ng (PPA)). In addition, regardless of experiment types, their MDL values fell in a narrow range of 0.040 ± 0.017 ng (direct injection of the LWS) to 0.235 ± 0.242 ng (Exp 2B) (Table S-4, Supporting Information). D

DOI: 10.1021/ac504713y Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Table 2. Comparison of the Relative Sorptive Lossesa (RSL, %) of Gaseous VOC Standards According to Experiment Mode (Sampling Transfer Approach (Sampling Material)) exp modes calibration code:

Exp 2A: 100 mL syringe (no needle) to ST FSC1

FSC2

FSC3

FSC4

FSV1

Group 1 (G1): aldehyde, ketone, ester, alcohol, and aromatic IA −4.91 0.02 7.40 4.78 2.82 VA −0.11 −0.58 5.30 4.88 3.29 MEK 1.51 −5.90 4.10 4.36 6.86 MIBK 1.34 5.58 −1.52 2.19 0.55 BuAc 9.33 11.8 0.10 −1.22 2.28 i-BuAl 1.21 4.80 6.14 −0.05 4.05 B 4.61 4.55 −0.32 −0.16 −1.23 T 2.09 4.40 0.11 3.92 1.05 p-X 4.29 5.99 −2.00 1.04 6.21 m-X 3.19 5.18 −1.91 2.13 6.75 o-X 4.33 4.54 0.09 0.64 5.24 S 3.40 6.14 −0.48 1.52 6.63 Mean 2.52 3.88 1.42 2.00 3.71 SD 3.37 4.37 3.35 2.08 2.71 Group 2 (G2): carboxyl and phenol PPA 23.1 20.4 15.0 14.3 24.5 IBA 31.0 12.4 11.7 8.6 32.8 BTA 40.7 24.8 16.9 16.9 49.7 IVA 36.2 20.2 12.7 11.6 39.4 VLA 42.1 28.1 20.1 21.0 43.0 HXA 57.6 39.3 38.6 40.7 54.8 HPA 68.8 62.8 58.8 64.9 62.4 o-C 39.1 30.8 22.5 18.6 60.9 m-C 53.2 41.2 38.3 31.5 70.0 mean 43.6 31.1 26.1 25.4 48.6 SD 14.1 15.0 15.9 17.9 14.9 a

Exp 2B: 10 mL syringe (no needle) to ST

Exp 3: 100 mL syringe (needle) to ST

bag to ST

FSV2

FSV3

mean

SD

FSCα

FSV1

FSC2

FSV2

FSC3

FSV2

4.20 1.97 5.15 3.82 2.00 4.26 0.31 3.67 0.76 0.69 1.03 2.00 2.49 1.66

5.30 5.42 3.65 1.02 −1.14 0.56 −0.08 3.09 0.55 1.59 0.55 1.06 1.80 2.10

2.80 2.88 2.82 1.85 3.31 3.00 1.10 2.62 2.40 2.52 2.35 2.90 2.55 0.59

4.09 2.52 4.17 2.31 5.20 2.39 2.42 1.59 3.12 2.87 2.24 2.65 2.96 1.02

−5.18 5.21 2.39 8.74 5.45 1.49 −1.81 −0.01 4.57 1.58 1.98 4.16 2.38 3.67

−4.38 5.23 1.81 7.15 5.27 2.77 −1.33 −1.15 3.37 1.31 2.70 3.80 2.21 3.24

4.45 5.67 6.73 0.24 0.60 5.13 1.42 3.54 2.29 3.54 2.26 2.27 3.18 2.03

4.06 3.19 5.35 2.52 0.61 2.64 −0.30 3.18 1.14 2.59 1.10 5.00 2.59 1.73

1.38 2.97 −1.48 −3.08 −1.39 3.03 0.13 −0.28 1.57 2.46 1.44 −0.32 0.54 1.93

3.53 −5.61 −1.47 −1.66 −0.52 7.32 1.04 −0.54 1.18 −0.08 1.17 −0.40 0.33 3.10

20.0 11.8 18.8 14.3 23.1 41.8 64.6 27.2 36.8 28.7 16.7

13.3 8.67 16.7 11.5 20.5 39.9 63.5 17.6 32.4 24.9 17.7

18.6 16.7 26.4 20.8 28.3 44.7 63.7 31.0 43.3 32.6 15.3

4.49 10.5 13.4 12.0 10.1 7.99 3.00 15.2 13.8 10.1 4.2

19.9 35.3 49.9 45.1 49.9 65.2 69.8 54.6 68.6 50.9 16.3

20.0 34.2 48.0 44.5 49.3 64.1 69.5 54.0 67.6 50.1 16.2

14.4 19.3 33.2 25.6 37.6 53.2 64.6 36.4 55.0 37.7 17.0

13.1 16.2 24.0 20.1 28.3 49.5 69.9 34.6 50.6 34.0 19.0

21.2 5.35 14.2 2.68 6.74 27.2 56.0 29.6 49.6 23.6 19.1

25.1 3.21 15.2 7.92 11.5 27.8 56.0 30.4 49.8 25.2 18.2

Relative sorptive loss (RSL, %) = [RF value (Exp 1) − RF value (Exp n)]/RF value (Exp 1) × 100.

Group 1, 28.5 ± 3.97 ppb to 285 ± 39.7 ppb; (2) Group 2, 32.4 ± 11.7 ppb to 324 ± 117 ppb (Table S-5, Supporting Information). Assessment of the Relative Sorptive Losses for VOCs (Group 1) in Each of Four Experiment Modes. In this study, VOCs with relatively high volatility (aldehydes, ketones, esters, alcohols, and aromatic compounds) were designated as Group 1. The sorptive loss properties of the Group 1 compounds were assessed for the GTS (Experiments 2A/B and Experiment 3) and a 1 L PEA bag (Exp 4). As shown in Table S-6 (Supporting Information), Group 1 had fairly good linearity (R2 > 0.99) and excellent experimental reproducibility, with low RSE values (20%) (Table 2). The RSL values (Group 2 (n = 9)) of Experiment 2A, using 100 mL of GTS without a needle, averaged 32.6 ± 17.7% (range: 24.9 ± 17.7 (FSV3) to 48.6 ± 14.9% (FSV1)). In the case of Experiment 2B, the target semivolatile compounds (Group 2) had higher sorptive losses (mean RSL (%) = 50.5 ± 15.8) than those in Experiment 2A (mean RSL (%) = 32.6 ± 17.7). However, when Experiments 2A and 2B were conducted using the same sampling approach (FSV1), the RSL values of the Group 2 compounds were maintained (mean RSL (n = 9): (1) Experiment 2A = 48.6 ± 14.9 and (2) Experiment 2B = 50.1 ± 16.2%) (Figure 2). In Experiment 3, the mean RSL values of Group 2 were 37.7 ± 17.0% (FSC2) and 34.0 ± 19.0 (FSV2). The RSL values for Experiment 3, which involved the use of a needle, were approximately 5% higher than those for Experiment 2A (Figure 2). In Experiment 4, sorptive loss of the Group 2 compounds also occurred on the PEA bag, with a mean RSL value of 24.4 ± 18.1%. E

DOI: 10.1021/ac504713y Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 2. Comparison of the relative sorptive loss (RSL, %) values of 21 target compounds with different gastight syringe size and needle.

PPA to HPA: 18.6 to 63.7% (Experiment 2A), 19.9 to 69.6% (Experiment 2B), 13.7 to 67.2% (Experiment 3), and 23.2 to 56.0% (Experiment 4). For cresols, m-C (with enhanced BP relative to o-C) had higher RSL values than the o-C (mean RSL values of o-C and m-C (%) = 31.0 and 43.3 (Experiment 2A), 54.3 and 68.1 (Experiment 2B), 35.5 and 52.8 (Experiment 3), and 30.0 and 49.7 (Experiment 4)). Accordingly, the results of a correlation analysis between the RSL values and their MWs (or BPs) showed a strong correlation, with a mean R2 of 0.804 ± 0.084 (mean p-value = 2.06 × 10−3 ± 2.88 × 10−3), with the exception of Experiment 4 (Table S-8B, Supporting Information). In the case of Experiment 4, which used a 1 L PEA bag, poor correlations were recorded between the RSL values and the MWs (or BPs) ((1) mean R2 = 0.410 ± 0.109 and (2) mean p-value = 0.078 ± 0.060) (Table S-8B, Supporting Information). Overall, the sorptive losses of the Group 2 compounds in the GTS may be explained primarily by their physical properties, like MW (and BP). A significant sorptive loss of semivolatile compounds can occur not only in a GTS but also in other storage media. For example, Koziel et al.14 examined the sorptive losses of semivolatile carboxyl and phenol compounds, i.e., acetic acid, PPA, IBA, BTA, IVA, VLA, HXA, p-cresol, and 2′-aminoacetophenone, in a Teflon bag. The mean sorptive loss was 24.6 ± 3.82% (at a storage time of 0.5 h). The sorptive loss of semivolatile compounds in the Teflon bag was higher than that in the GTS in the present study (mean RSL = 32.6 ± 17.7% (Exp 2A), 50.5 ± 15.8% (Exp 2B), and 35.9 ± 17.6% (Exp 3)). However, the sorptive loss in the Teflon bag was similar to that

As shown in Figure 3, the RSL values of the Group 2 compounds increased with decreasing sample concentration and volume. In Experiment 2A, the mean RSL value of the Group 2 compounds of FSC1 (43.6 ± 14.1%) was approximately 20% higher than that of the FSC4 (25.4 ± 17.9%) (Table 2). In addition, the FSV1 with the smallest sample volume (10 mL) showed the highest mean RSL of 48.6 ± 14.9% (Group 2 − Experiment 2A). To test the effectiveness of the sorptive loss patterns using the key variables of sample concentration and sampling volume, a linear regression analysis was conducted for the RSL values and those key variables (Table S-8A, Supporting Information). The RSL values of the Group 2 compounds increased as the sample concentration and volume decreased: (1) RSL/actual concentration (%/ppb) = −0.01 ± 0.01 (Experiment 2B) to −0.05 ± 0.02 (Experiment 2A) and (2) RSL/sampling volume (%/mL) = −0.05 ± 0.05 (Experiment 4) to −1.25 ± 0.29 (Experiment 2B). However, the strengths of the correlations were not significant, with mean R2 values of 0.681 ± 0.260. In addition, the p-values for all of the experiment modes were high, with a mean of 0.291 ± 0.219 (range: 0.002 to 0.931, shown in Table S-8A, Supporting Information). As a result, the dependence of sorptive loss on the key variables of sample concentration and sampling volume are not predictive in this respect. The RSL values of Group 2 tended to increase as the molecular weight (MW, g/mol) (or boiling point (BP, °C)) increased for all of the experiment modes (Table 2). For example, the RSL values of the VFAs (n = 7) increased systematically with increasing MW (or BP) (RSL values from F

DOI: 10.1021/ac504713y Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

The highly volatile compounds (Group 1: aldehyde, ketone, ester, alcohol, and aromatic hydrocarbons (n = 12)) showed the most reliable trend of significantly low RSL values (±5%), regardless of sample concentration, sampling volume, syringe size, or needle installation. Conversely, the sorptive loss of the semivolatile compounds (Group 2: carboxyl and phenols (n = 9)) in the GTS varied considerably, ranging from 18.0 ± 4.10% (propionic acid) to 65.4 ± 10.9% (n-heptanonic acid). Although the Group 2 compounds recorded higher RSL values at low sample concentrations and sampling volumes, the association between sample concentration (or sampling volume) and RSL was not significant, with a mean p-value of 0.291 ± 0.219. However, the RSL of the Group 2 compounds showed strong correlations with their physical properties of molecular weight and boiling point (mean p-value = 2.06 × 10−3 ± 2.88 × 10−3). The sorptive loss of the Group 2 compounds was not affected by syringe size; however, the sorptive losses occurring in the stainless steel needle with RSL were approximately 5%. In light of these results, it is suggested that the use of gastight syringe should be avoided to collect and analyze semivolatiles with strong sorptive reactivies. This study confirmed that there is a significant amount of sorptive loss of semivolatile compounds due to contact with the inner surfaces of a GTS. Therefore, it is important to examine and remove the sources of sorptive losses of semivolatile compounds during the sampling and pretreatment stages in order to carry out an accurate quantitative analysis. However, if the semivolatiles are analyzed using the gastight syringe (or glass containers), the sorptive loss results determined by this study should be considered to balance the loss of such targets for their reliable quantitation.

Figure 3. Relative sorptive loss (RSL, %) of Group 2 (carboxyl and phenols) in Exp 2A (100 mL syringe (no needle) to ST.

of the PEA bag in the present study (mean RSL = 24.4 ± 18.1%). Kim and Kim3 tested the sorptive loss of five carboxyl acids (acetic acid, PPA, BTA, IVA, and VLA) in stainless steel and quartz materials by forcing the target compounds to pass through empty stainless steel and quartz tubes. The mean sorptive loss in the quartz tube was 23.1 ± 19.1% (n = 5), whereas that in the stainless steel tube was significantly greater, with a mean of 70.8 ± 30.0% (n = 5). As such, when the semivolatile compounds were sampled and analyzed, both the physical properties and the transfer/storage materials were considered in order to provide an accurate quantitative analysis. In the case of semivolatile compounds like VFAs, an accurate assessment of the sorptive loss by all types of sources (including due to the use of a gastight syringe) is critical to obtain the quantitative results. Because the VFAs are strong odorants with their odor threshold at sub-ppb levels, the trace level quantitation of VFAs is essential step for the odor detection. However, if the VFAs were transferred (or collected) by a gastight syringe, the dominant fraction of such analytes are readily lost by the syringe. For this reason, more cautions should be excercised to use a gastight syringe for sampling and analysis of semivolatile compounds.

■

ASSOCIATED CONTENT

S Supporting Information *

Preparation of both the liquid and gaseous working standards, instrumental system, list of 21 target VOCs selected for the evaluation of sorptive loss in a gastight syringe, basic information for the preparation of liquid and gaseous VOC standards for TD-based analysis, operational conditions for the analysis of 21 target VOCs using a TD-GC−MS system, quality assurance/quality control (QA/QC) parameters with different experiment types (Exp 1 to 4), comparison of the recovery of 21 target compounds during direct transfer of gaseous working standard to the ST (Exp 1: direct ST)), comparison of the calibration results with different experiment 70 modes (Exp 1 to 4), comparison of recovery (response factor) between Exp 1 and 2A based on a t-test (heteroscedastic), results of correlation analysis between relative sorptive loss (%) and four key variables, chromatograms of the gaseous VOC standards for different experiments, and mean values of relative sorptive loss (RSL, %) of Group 1 compounds (n = 12) with different sampling approaches. This material is available free of charge via the Internet at http://pubs.acs.org.

■

CONCLUSIONS The GTS is one of the most widely used transfer mediums for the collection and analysis of VOCs in air or gas. Sorptive losses of VOCs are expected when using any storage or transfer medium. Accordingly, the most stable material for GTS is expected to be subject to the fewest biases. In this study, the sorptive loss patterns of VOCs in a GTS were assessed according to syringe size, needle installation, sample concentration, sampling volume, and VOCs with different volatilities.

■

AUTHOR INFORMATION

Corresponding Author

*K.-H. Kim. E-mail: [email protected]; kkim61@nate. com. Phone: 82-2-2220-2325. Fax: 82-2-2220-1945. Notes

The authors declare no competing financial interest. G

DOI: 10.1021/ac504713y Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

■

ACKNOWLEDGMENTS This work was supported by a grant from the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (MEST) (No. 20090093848).

■

REFERENCES

(1) KMOE. Annual Report of Ambient Air Quality in Korea; Korean Ministry of Environment: Sejong-si, Korea, 2008. (2) Cometto-Muniz, J. E.; Cain, W. S.; Abraham, M. H. Exp. Brain Res. 1998, 118, 180−188. (3) Kim, Y. H.; Kim, K. H. Anal. Chem. 2013, 85, 7818−7825. (4) Smith, M. T. Environ. Health Perspect. 1996, 104 (Suppl6), 1219− 1225. (5) Kim, Y.-H.; Kim, K.-H.; Jo, S. H.; Jeon, E.-C.; Sohn, J. R.; Parker, D. B. Anal. Chim. Acta 2012, 712, 162−167. (6) Trabue, S. L.; Anhalt, J. C.; Zahn, J. A. J. Environ. Qual. 2006, 35, 1668−1677. (7) Heglund, D. L.; Tilotta, D. C. Environ. Sci. Technol. 1996, 30, 1212−1219. (8) Kim, Y. H.; Kim, K. H. Anal. Chem. 2012, 84, 4126−4139. (9) Ahn, J. W.; Kim, Y. H.; Kim, K.-H.; Song, H.-N. . Anal. Sci. Technol. 2012, 25, 91−101. (10) Kim, Y. H.; Kim, K. H.; Szulejko, J. E.; Parker, D. Anal. Chem. 2014, 86, 6640−6647. (11) Ravi, V. P.; Jasra, R. V.; Bhat, T. S. G. J. Chem. Technol. Biotechnol. 1998, 71, 173−179. (12) Kim, Y. H.; Kim, K. H. J. Sep Sci. 2012, 35, 2914−2921. (13) Kim, K. H.; Nguyen, H. T. J. Sep. Sci. 2007, 30, 367−374. (14) Koziel, J. A.; Spinhirne, J. P.; Lloyd, J. D.; Parker, D. B.; Wright, D. W.; Kuhrt, F. W. Sensors 2009, 10, 8536−8552.

H

DOI: 10.1021/ac504713y Anal. Chem. XXXX, XXX, XXX−XXX