Asian Journal of atmospheric environment
Asian Journal of atmospheric environment Asian Journal of atmospheric environment
Asian Journal of atmospheric environment
  Aims and Scope Type of Manuscripts Best Practices Contact Information  
  Editor-in-Chief Associate Editors Editorial Advisory Board  

Current Issue

Asian Journal of Atmospheric Environment - Vol. 11 , No. 1

[ Research Article ]
Asian Journal of Atmospheric Environment - Vol. 11, No. 1
Abbreviation: Asian J. Atmos. Environ
ISSN: 1976-6912 (Print) 2287-1160 (Online)
Print publication date 31 Mar 2017
Received 03 Jul 2016 Revised 07 Dec 2016 Accepted 07 Dec 2016

Quantitative Approaches for the Determination of Volatile Organic Compounds (VOC) and Its Performance Assessment in Terms of Solvent Types and the Related Matrix Effects
Md. Ahsan Ullah ; Ki-Hyun Kim* ; Jan E. Szulejko ; Dal Woong Choi1)
Department of Civil & Environmental Engineering, Hanyang University, Seoul 04763, Republic of Korea
1)Department of Public Health Science, Korea University, Seoul 02841, Republic of Korea

Correspondence to : * Tel: +82-2-2220-2325, E-mail: or

Funding Information ▼


For the quantitative analysis of volatile organic compounds (VOC), the use of a proper solvent is crucial to reduce the chance of biased results or effect of interference either in direct analysis by a gas chromatograph (GC) or with thermal desorption analysis due to matrix effects, e.g., the existence of a broad solvent peak tailing that overlaps early eluters. In this work, the relative performance of different solvents has been evaluated using standards containing 19 VOCs in three different solvents (methanol, pentane, and hexane). Comparison of the response factor of the detected VOCs confirms their means for methanol and hexane higher than that of pentane by 84% and 27%, respectively. In light of the solvent vapor pressure at the initial GC column temperature (35°C), the enhanced sensitivity in methanol suggests the potential role of solvent vapor expansion in the hot injector (split ON) which leads to solvent trapping on the column. In contrast, if the recurrent relationships between homologues were evaluated using an effective carbon number (ECN) additivity approach, the comparability assessed in terms of percent difference improved on the order of methanol (26.5%), hexane (6.73%), and pentane (5.24%). As such, the relative performance of GC can be affected considerably in the direct injection-based analysis of VOC due to the selection of solvent.

Keywords: Volatile organic compound (VOC), Solvent effect, Sandwich injection (SI) technique, Response factor, Effective carbon number (ECN)

1. Introduction

In light of the effect on both human health and global environment, volatile organic compounds (VOC) have become a major issue and concern for many environmentalists. VOCs are well known for their contribution toward global-scale environmental changes such as global warming, stratospheric ozone depletion, and photochemical ozone formation (Kavouras et al., 2013; Sahu, 2012; Evtyugina et al., 2009). It is also perceived that VOCs and their degradation products can be responsible for the high prevalence of various respiratory disorders and cancers (Sahu and Saxena, 2015; Yassaa et al., 2011; Wang and Zhao, 2008; Boeglin et al., 2006). Moreover, many members of VOCs can act as the source of odor nuisance on a regional scale (Lal et al., 2012; Sahu and Lal, 2006).

To gain a better knowledge on the occurrence, behavior, and fate of environmental VOCs, acquisition of accurate concentration data is of primary importance in their research. In view of the physicochemical properties of these target compounds, analytical methods were developed for separation by gas chromatography (GC) and subsequent detection, e.g., via mass spectrometry (MS), flame ionization detection (FID), electron capture detection (ECD), and others (Kim et al., 2011). For all methods, calibration is the key to quantify VOCs contained in an unknown sample. If the calibration is made incorrectly, it can lead to over- or under-estimation of a VOC in a sample. The general procedure of quantitation includes: 1) preparation of accurate standard mixtures; 2) the calibration, i.e., conducting a given analytical process for standard mixtures and; 3) establishing the relationship between final instrument response and analyte content (either in concentration or absolute amount) in a sample (Namieśnik et al., 2000).

To conduct gas chromatographic analysis of VOC in a traditional way, each sample is introduced via direct injection and loaded onto the chromatographic column for the separation of compounds by manual (or automatic) injection (Harper, 2000). In contrast, if airborne VOCs have been collected by trapping on a solid sorbent, it is necessary to volatilize analytes contained in samples or standards with the aid of a thermal desorption system (Ras et al., 2009). To minimize the quantitiation bias due to phase differences between real samples and laboratory standards, the use of gaseous samples and gaseous standards is the most preferable option. However, the use of liquid-phase standards are often inevitable for the analysis of gas samples due to the complexities (and high cost) in the preparation of gas standards and/or to the reactivity of the standard compounds. Hence, if the analyte is in either a liquid or gaseous matrix, their mixing ratio is also an important criterion for GC-based quantitation. The solvent to analyte ratio is commonly set at 1000 : 1 or greater in environmental analysis. Hence, it is important to choose a solvent that exhibits the least interference in the detection of the analytes. As is the case of direct injection, the solvent used for the liquid standards in the thermal desorption (TD) analysis should also be capable of effectively stripping the analytes off the sorbent. Moreover, it is imperative that such stripping should be made with high degree of efficiency and reproducibility (Harper, 2000).

Most commonly used solvents available in laboratory are acetone, acetonitrile, benzene, cyclohexane, dichloromethane, dimethyl ether, ethylacetate, tetrahydrofuran, toluene, methanol, hexane, and pentane (Ahn et al., 2011; Pang et al., 2011). Among those, the latter three (shortnames: MeOH, Hx, and Pn) have been the most common choice for VOC analysis. Although carbon disulfide (CS2) is commonly used in solvent extraction of VOCs off sorbents (highest Hansen solubility parameters among the common solvents), CS2 was not included in our study in light of its toxicity (Beauchamp et al., 1983). In general, their properties seem neutral in that their presence minimally alters or interferes with chromatographic responses. Nonetheless, the results of an in-depth literature search demonstrated that information is insufficient to describe the details of solvent effects or to recommend an optimum solvent for the GC-based analysis of VOCs.

In this work, we attempted to describe relative response characteristics of VOCs in association with three different solvents (methanol, pentane and hexane). For this comparative analysis, a total of 19 VOCs (Table 1) were selected and investigated for the analysis of solvent effect which included acetaldehyde (AA), propionaldehyde (PA), butyraldehyde (BA), isovaleraldehyde (IA), valeraldehyde (VA), benzene (B), toluene (T), p-xylene (p-X), m-xylene (m-X), o-xylene (o-X), styrene (S), methylethylketone (MEK), methyl isobutyl ketone (MIBK), butyl acetate (BuAc), isobutyl alcohol (i-BuAl), propionic acid (PPA), n-butyric acid (BTA), isovaleric acid (IVA), and n-valeric acid (VLA). All of these compounds (except benzene) are classified as offensive odorants in S Korea (KMOE, 2010). In order to explore the feasibility of each solvent as the medium for VOC analysis, a comparative analysis was carried out based on direct injection of liquid standards of VOCs into a gas chromatograph (GC) equipped with a splitter and a flame ionization detector (FID). To evaluate relative response characteristics of different VOCs between different solvents, the calibration for each compound was done in triplicate. The reliability of each solvent in VOC analysis was then examined by evaluating the relative performance (e.g., sensitivity) and the basic quality assurance (QA) parameters (e.g., method detection limit (MDL) and relative standard error (RSE)).

Table 1. 
Basic information on 19 target volatile organic compounds (VOCs) and three solvent used in this study.
Order Group Full name Short
Formula CASa
RTb RTb RTb Conc.c
(ng μL-1)
(Wax column)
RTI Reference
MeOH Pn Hx
i Solvents pentane Pn C5H12 109-66-0 9.24-14.10 625937 500
ii methanol MeOH CH4O 67-56-1 4.52-6.27 791800 888 Chida et al. (2004)
iii hexane Hx C6H14 110-54-3 4.53-6.25 651526 600
1 Aldehydes acetaldehyde AA C2H4O 75-07-0 4.71 - e - 776160 <500
2 propionaldehyde PA C3H6O 123-38-6 5.58 6.37 6.34 785700 571
3 butyraldehyde BA C4H8O 123-72-8 6.77 8.29 8.28 792000 832 Qian and Reineccius (2003)
4 isovaleraldehyde IA C5H10O 590-86-3 - 9.7 9.76 773090 912
5 valeraldehyde VA C5H10O 110-62-3 - 12.43 12.48 785215 935
6 Aromatics benzene B C6H6 71-43-2 - 10.68 10.78 872118 938 Rahman and Kim (2013)
7 toluene T C7H8 108-88-3 14.90 14.55 14.54 865650 1038 Goodner (2008)
8 p-xylene p-X C8H10 106-42-3 17.82 17.39 17.33 852390 1149 Yanagimoto et al. (2004)
9 m-xylene m-X C8H10 108-38-3 17.95 17.54 17.48 851400 1150 Le Guen et al. (2000)
10 o-xylene o-X C8H10 95-47-6 18.91 18.63 18.56 835170 1182 Binder et al. (1990)
11 styrene S C8H8 100-42-5 20.32 20.19 20.12 899910 1273 Sanz et al. (2001)
12 Ketones & an alcohol methyl ethyl ketone MEK C4H8O 78-93-3 - 9.22 9.19 796950 923 Héberger and Görgényi (1999)
13 methyl isobutyl ketone MIBK C6H12O 108-10-1 14.42 13.43 13.46 797990 974 Héberger and Görgényi (1999)
14 butyl acetate BuAc C6H12O2 123-86-4 16.36 15.65 15.61 875600 1105 Culleré et al. (2004)
15 isobutyl alcohol i-BuAl C4H10O 78-83-1 16.86 16.29 16.20 793980 1103 Goodner (2008)
16 Fatty acids propionic acid PPA C3H6O2 79-09-4 24.43 24.44 24.38 980100 1525 Goodner (2008)
17 n-butyric acid BTA C4H8O2 107-92-6 25.53 25.52 25.48 949905 1628 Goodner (2008)
18 isovaleric acid IVA C5H10O2 503-74-2 26.00 25.98 25.95 915750 1691 Culleré et al. (2004)
19 n-valeric acid VLA C5H10O2 109-52-4 26.81 26.79 26.75 920700 1698 Goodner (2008)
aChemical Abstracts Service (CAS) number, bRetention time (RT) determined in this study, cRaw chemical concentration, dKovats retention time index (RTI), eNot detected

2. Materials and Methods
2. 1 Preparation of Working Standards

In this study, the relative performance properties of the solvents were investigated for the quantitation of 19 selected target VOCs with four functional groups: i) five aldehydes (AA, PA, BA, IA, and VA), ii) five aromatic hydrocarbons (B, T, p-X, m-X, o-X, and S), iii) five fatty acid (PPA, BTA, IVA, and VLA), and iv) three ketones and an alcohol (MEK, MIBK, BuAc, and i-BuAl). For this study, relative performance of the three most widely used solvents in GC, e.g., methanol, pentane, and hexane were investigated. Basic information on the 19 target VOCs and three solvents selected in this study is summarized briefly in Table 1.

Primary grade chemicals were purchased from Sigma-Aldrich, USA at purities of, 97% (PA, IA, VA, and o-X), 99.5% (B, T, MIBK, and BuAc), and 99% for the rest of VOCs. For the three solvents, methanol (purity 100%), pentane (purity 100%) and hexane (purity 99.5% (95% n-hexane, 4.5% other C6H14 isomers, and with trace benzene and cyclopentane)) were purchased from J.T. Baker, USA. As shown in Table 2, four primary standards (PS) representing each of the four chemical groups were prepared. Their liquid phase working standards (L-WS) were then prepared by two step volumetric dilution of the four PS mixtures in each solvent (MeOH, Pn, and Hx). For the preparation of the PS of each chemical group, approximately 70 μL of each raw VOC (except AA (200 μL)) were mixed with each solvent to make a final volume of 1.5 mL. Hence, a mean concentration of each VOC was maintained at 40 μg μL-1 (Table 2). The concentration of each VOC in the L-WS was adjusted to a similar level, except for AA (Table 2). The concentration of AA was set 2.5 times higher than the others due to its reduced FID sensitivity (e.g., Pal and Kim (2008)). In the second step, this PS was diluted further with each solvent to make the final L-WS for the five point calibrations with the concentration of each VOC at approximately 10, 20, 40, 70, and 140 ng μL-1 (except AA). In case of pentane (as solvent), some special precautions were necessary because of its high volatility (and low boiling point); i) maintaining laboratory temperature as low as possible and ii) using a small micropipette (10-100 μL capacity) to take a large volume (>100 μL) solute. After making the final L-WS with each solvent, five point calibration experiments were carried out by introducing (injecting) 1 μL volumes of each (final) L-WS into the GC-FID system.

Table 2. 
Preparation of liquid-phase VOC standard for the analysis with GC-FID by direct injection (Same for all three solvent types).
A. Step-1: Primary standard (PS)
Group compounds Aldehydea Aromatica Ketonic & alcohola Fatty acida
Concentration (%) 99 97 99 97 97 99.5 99.5 99 99 97 99 99 99.5 99.5 99 99 99 99 99
Reagent volume (μL) 200 76 76 78 77 69 69 71 71 72 67 75 75 69 76 61 63 65 65
Solvent (μL)b 993 1081 1205 1246
Concentration (ng μL-1) 103488 39809 40128 40201 40308 40117 39820 40346 40300 40088 40196 39848 39900 40278 40228 39857 39896 39683 39897
B. Step-2: First stage working standard (1st L-WS)
PS volume (μL) 50×4=200
Solvent (μL)        3800
Compoundsc AA PA BA IA VA B T p-X m-X o-X S MEK MIBK BuAc i-BuAl PPA BTA IVA VLA
Concentration (ng μL-1) 1294 498 502 503 504 501 498 504 504 501 502 498 499 503 503 498 499 496 499
C. Step-3: Final working standard (final L-WS)
Order WS-1
Concentration (ng μL-1)
1 30 1470 25.9 10.0 10.0 10.1 10.1 10.0 10.0 10.1 10.1 10.0 10.0 10.0 10.0 10.1 10.1 10.0 10.0 9.9 10.0
2 60 1440 51.7 19.9 20.1 20.1 20.2 20.1 19.9 20.2 20.1 20.0 20.1 19.9 19.9 20.1 20.1 19.9 19.9 19.8 19.9
3 120 1380 103 39.8 40.1 40.2 40.3 40.1 39.8 40.3 40.3 40.1 40.2 39.8 39.9 40.3 40.2 39.9 39.9 39.7 39.9
4 210 1290 181 69.7 70.2 70.4 70.5 70.2 69.7 70.6 70.5 70.2 70.3 69.7 69.8 70.5 70.4 69.8 69.8 69.4 69.8
5 420 1080 362 139 140 141 141 140 139 141 141 140 141 139 140 141 141 140 140 139 140
aPrimary standard for each group (1500 μL) of compound has been made
bMethanol, pentane, and hexane have been used as solvent
cMixture of all 19 VOC’s made at a total volume of 4000 μL

2. 2 Instrumental System and Analytical Technique

All calibration experiments for 19 VOCs in each of the three solvents were carried out in this study using a GC-FID system (IGC 7200, DS Science, Korea) by direct injection of L-WS in a 1 : 5 split injection mode. The injector was set at 250°C. A 10 μL SGE syringe was used to introduce liquid standard of VOC into GC. The injected VOCs were separated on CP-Wax 52 CB Varian capillary column (60 m (length)×0.25 mm (internal diameter)×0.25 μm (film thickness) at a carrier gas (N2) flow rate of 1 mL min-1. In addition, the FID flow rates of N2 (makeup gas), H2, and air were set at 30, 30, and 300 mL min-1, respectively. To allow parallel comparison of solvent effects, the same operational settings were maintained for all L-WS calibrations for each of the three solvents. For instance, GC oven temperature was initially set at 35°C for 10 minutes and then programmed to ramp at 10°C min-1 to 200°C with the final hold for 5 minutes. The detector temperature was set at 250°C.

The liquid samples were injected using the “sandwich injection (SI) technique” as defined by Grob (2001). In this method, syringe was initially cleaned several times with the solvent used to prepare the L-WS under investigation. Into the syringe were sequentially drawn a plug of air (1 μL), then the sample (1 μL), and finally another air plug (1 μL). The needle of the syringe (SGE, Australia, 10 μL glass fixed needle, needle length 50 mm) used in this study can hold 0.6 μL sample. Hence, by the using SI technique, a more precise sample volume can be injected into the GC to yield optimal calibration results.

3. Results

In this study, the liquid standards containing all target VOCs were analyzed based on SI technique through GC-FID system. To examine the chromatographic separation patterns between liquid standards with three solvent types, individual VOC standard was initially analyzed using methanol as solvent. The experimental retention time (RT) in conjunction with literature Kovats retention time index (RTI) for all compounds was used for the identification of individual from the mixture (Table 1).

Although retention time of different compounds varied slightly with different solvents, their overall trend of elution order remained constant. In all cases elution of all 19 VOCs were seen on the order of AA, PA, BA, MEK, IA, B, VA, MIBK, T, BuAc, i-BuAl, p-X, m-X, o-X, S, PPA, BTA, IVA, and VLA. This elution order also matches well with their RTI values with a few exceptions. In all analyses, three exceptions to the expected (RTI) elution order were noted for the following pairs of analytes (having similar RTIs): 1) MEK (923) and IA (912), 2) B (938) and VA (935), and 3) BuAc (1105) and i-BuAl (1103). In general, all VOCs eluted according to their RTI values. However, in mixture standard, such order was moderately disrupted in certain occasions. For compounds with similar RTI values, elution order is influenced by solvent/analyte thermochemical factors (Gonzalez and Nardillo, 1999).

Typical chromatograms obtained in this study for 140 ng injection mass are presented in Fig. 1. Out of the 19 VOCs, only AA was missing, when either pentane or hexane was used as solvent. As the instrumental setting for all L-WS with different solvents were identical for parallel comparison, instrumental settings are unlikely to be the key factor in the non-detection of AA if using Pn and Hx solvents. In the case of methanol solvent, four out of 19 VOCs was not detected which include MEK, IA, B, and VA. Probably, the peaks for these missing compounds could have merged with solvent peak. Consequently, their elution from methanol (used as solvent) may not be easy to confirm by FID alone, unless aided by MS. An extra peak was observed on the base of toluene peak which may reflect an esterification process of a fatty acid with methanol or formation of an acetal from aldehyde with methanol (Fujiwara and Fujiwara, 1963). Under appropriate conditions, aldehydes and ketones will react with alcohols to produce hemiacetals and acetals in high yields (Guthrie, 1975). The extent of the propanoic acid/methanol esterification reaction was found to be dependent on reaction time and concentration of acid (Lilja et al., 2002). In the case of toluene, the presence of this extra peak can induce errors in its quantitation. However, the least biased quantitation could be achieved if analysis was done within 5 minutes after the L-WS preparation with the least elapsed time.

Fig. 1. 
Chromatogram of 19 VOCs analyzed by GC-FID: 1) AA, 2) PA, 3) BA, 4) MEK, 5) IA, 6) B, 7) VA, 8) MIBK, 9) T, 10) BuAc, 11) i-BuAl, 12) p-X, 13) m-X, 14) o-X, 15) S, 16) PPA, 17) BTA, 18) IVA, and 19) VLA (All compound injection mass was 140 ng, except AA).

4. Discussion
4. 1 Comparison of Calibration Result between Different Solvent

To obtain the calibration data of 19 VOCs, 1 μL of each L-WS at five different concentration levels (≈ 10, 20, 40, 70, and 140 ng μL-1) was analyzed by GC-FID. To check the reproducibility of each calibration, triplicate analysis was carried out with two consecutive blank run between each cycle of calibration. The calibration data for each target VOCs are compiled in terms of the response factor (RF) and linearity (by correlation coefficient (R2) of regression analysis) in Table 3. The results of our VOC calibrations indicate a potent role of solvent selection, while the relative calibration patterns may be influenced by solvent vapor pressure. As shown in Table 3, the RF values of different VOCs varied widely with the solvent used.

Table 3. 
Results of five point calibration of 19 volatile organic compounds prepared in three different solvent types.
Order Methanol (n=3)a Pentane (n=3) Hexane (n=3)
1 AA 4356±2534b 0.83±0.20b - c - - -
2 PA 44295±5380 0.99±0.00 10653±994 0.99±0.00 10485±285 0.94±0.01
3 BA 33612±1860 0.99±0.00 18963±1344 0.99±0.01 20375±779 0.99±0.01
4 IA - - 24306±1494 0.98±0.01 24792±1152 0.99±0.01
5 VA - - 17496±967 0.97±0.01 22812±908 0.99±0.00
6 B - - 50501±3021 0.98±0.01 55038±2290 0.99±0.01
7 T 55423±5933 0.99±0.01 50347±3041 0.97±0.01 56537±2822 0.99±0.01
8 p-X 135773±32870 0.99±0.01 49113±3071 0.97±0.01 55573±3061 0.99±0.01
9 m-X 74090±1106 0.99±0.00 53721±3235 0.96±0.01 60777±3253 0.99±0.01
10 o-X 109230±17212 0.99±0.00 53127±3522 0.97±0.01 61552±3220 0.99±0.01
11 S 103030±16420 0.99±0.00 53315±3228 0.96±0.01 59245±3222 0.99±0.01
12 MEK - - 18146±1242 0.99±0.01 32860±1094 0.99±0.01
13 MIBK 9639±788 0.98±0.01 24526±1502 0.98±0.01 42124±2331 0.99±0.01
14 BuAc 51327±7862 0.99±0.01 20661±1209 0.98±0.01 34431±1825 0.99±0.01
15 i-BuAl 60874±9047 0.99±0.00 23130±1481 0.98±0.01 40431±1628 0.99±0.01
16 PPA 26148±3679 0.99±0.01 14938±1665 0.98±0.01 18001±1644 0.99±0.01
17 BTA 33173±4476 0.99±0.01 20212±2097 0.98±0.01 28891±2121 0.99±0.00
18 IVA 37715±4830 0.99±0.01 24364±2449 0.98±0.01 40638±2629 0.99±0.01
19 VLA 33853±3989 0.99±0.01 22697±2467 0.98±0.01 31507±2330 0.99±0.01
Mean 54169 0.98 30568 0.98 38670 0.99
SD 37248 0.04 15779 0.01 16317 0.01
RSE 17.8 1.12 12.2 0.23 9.95 0.28
N 15 15 18 18 18 18
aTriplicate analysis for all solvent, bMean±SD and cNot detected

All aromatic compounds gave much enhanced RF values for all target solvents, while the largest observed RF value was for p-X in methanol (135,773). According to our analysis, methanol was found to be a good solvent, especially for aromatic compounds in terms of higher sensitivity relative to other compounds. In the case of methanol solvent, aromatics exhibited a 2.85 times higher RF values compared to the two other solvents. Unlike the case of either pentane or hexane, methanol is suspected to form a liquid film at the beginning of the wax stationary phase GC column at 35°C due to a combination of two main factors, the largest heat of vaporization (ΔHvap) and second highest boiling point (BP). Under such circumstances, methanol is likely to experience solvent trapping effect more effectively as noted by the increased retention times of analytes eluting after the methanol solvent peak (Grob Jr, 1982) and larger RF values. These phenomena may contribute to generally enhanced response factors for aromatics and alcohols. For pentane and hexane, the lowest RF value was attained from PA. In contrast, the smallest RF value of methanol was found from AA (4356) which is also 9.4 times (89.4%) lower than the mean RF of all other VOCs for all three solvents. In general, the calibration linearity was highly significant and excellent for most analytes e.g., average r2>0.98 (except AA (0.83) for methanol solvent). The mean correlation coefficients of all targeted VOCs, if evaluated for each individual solvent, methanol, pentane, and hexane were 0.991(excluding AA), 0.979 and 0.989, respectively. In Fig. 2, the RF patterns of target VOCs in pentane and hexane solvents were more or less similar. However, the results of methanol are different from others, especially for the heavier aromatics.

Fig. 2. 
Comparison of response factor between 19 volatile organic compounds using three different solvent by GC-FID.

If we compare the relative sensitivity of each solvent type by the average RF values of the detected VOCs common to all three different solvents, it was found on the order of methanol, hexane and pentane. The RF values of methanol and hexane were larger by 84% and 27%, respectively than that of pentane. Although methanol yielded enhanced RF values, it was more selective to cover a limited number (N=10 of 19 VOCs) of targets relative to other solvents in this study. In addition, its standard deviation of RF (for triplicate analyses) is worse than other solvents. There is a wide variation of RF for methanol among the same group of compounds, while those of pentane and hexane were found nearly constant. However, except aldehydes and PPA, hexane generally yielded higher RF than pentane. Hence, for most selected target compounds, there was little solvent effect between the two on their GC-FID based quantitation, although it was not for methanol.

4. 2 The Importance of the Time Factor (Injection Intervals) on the Relative Response of GC-FID with Three Types of Solvent

To further evaluate relative response properties of GC-FID for the three different solvent, an ancillary experiment has been performed using L-WS of a single concentration level. Out of five concentration level L-WS, only the medium level (e.g., 40 ng μL-1) was selected for this purpose. Samples were injected a few minutes after preparation to reduce potential analyte losses over time which can possibly occur depending on solvent types, e.g, the fatty acids in methanol yielding methyl esters (Lilja et al., 2002). The identical amounts of standard sample (1 μL) prepared in three solvents were injected at varying intervals (single injection for each interval in this order: pentane, methanol, and hexane). As depicted in Fig. 3, all 19 VOCs exhibited moderate changes in relative sensitivity trend with different solvents. However, for the same group of compounds, similar sensitivities were maintained except aromatic compounds with methanol. These findings suggest that the response properties are also affected by time interval of injection (or time elapsed after standard preparation) such as sequentially (Fig. 3A) or extended interval (Fig. 3B). In Fig. 3B, sensitivity variations were monitored at 4-hourly intervals for each solvent. If reproducibility of VOCs quantitation is compared between solvents types, the maximum variation occurred with methanol solvent. From Fig. 3B, it is found that methanol represented larger standard deviation of peak area (for all VOCs) than other solvents, with the highest values for p-X. Pentane and hexane showed similar reproducibility for all group of compounds except fatty acid, while methanol and pentane had a similar reproducibility for fatty acid compound.

Fig. 3. 
GC-FID sensitivity variation for 19 VOC’s with different solvent; A (peak areas)=standards in 3 different solvents (Pn, MeOH, Hx) were analyzed sequentially 1 hour apart (GC run time=31 minute, cool down time=29 minute) respectively vs, B (standard deviation)=same as for (A) but repeated 3 times, 4 hours apart.

4. 3 Basic Quality Assurance of Instrumental Setup

As a simple means to assess the relative effect of different solvents, the basic quality assurance (QA) parameters (such as method detection limit (MDL) and precision (e.g., reproducibility expressed in terms of a relative standard error (% RSE) of measurements)) were determined by heptaplicate and triplicate analyses of the smallest detectable quantity of L-WS, respectively. The L-WS concentration for MDL calculations was 5 ng (except MIBK for methanol and PA for pentane, 40 ng μL-1). The results of the QA analysis for three different solvent types are summarized in Table 4. In all cases, RSE tended to be <5%, while MDL was below the few nanogram level with a few exceptions.

Table 4. 
Basic quality assurance found for 19 VOCs using three different solvent in this study.
Order Short name Method detection limit (ng)a Relative standard error (%)
MeOH Pn Hx MeOH Pn Hx
1 AA 0.56 - b - 0.11 - -
2 PA 0.58 11.90 1.50 2.19 1.96 3.68
3 BA 1.80 0.56 1.27 4.96 3.13 2.61
4 IA - 0.50 1.43 - 2.07 2.02
5 VA - 0.74 0.85 - 1.85 1.35
6 B - 0.19 0.40 - 2.17 2.34
7 T 1.16 0.32 0.80 3.84 2.10 2.65
8 p-X 0.55 0.31 0.50 4.14 1.93 2.05
9 m-X 0.57 0.33 0.64 2.77 2.15 2.25
10 o-X 0.43 0.37 0.57 2.62 2.22 1.99
11 S 0.37 0.29 0.49 2.50 2.09 1.97
12 MEK - 1.08 1.35 - 2.87 2.73
13 MIBK 5.25 0.57 0.15 3.28 2.04 2.19
14 BuAc 0.42 0.85 0.57 3.12 2.43 2.40
15 i-BuAl 0.37 0.51 0.31 1.33 2.44 1.84
16 PPA 0.12 0.54 0.50 3.85 4.58 4.99
17 BTA 0.40 0.48 0.47 3.54 4.67 4.97
18 IVA 0.39 0.50 0.59 3.83 3.21 4.73
19 VLA 0.46 0.41 0.67 4.19 2.62 4.73
Mean 0.59c 0.50d 0.73 2.73 2.61 2.92
SD 0.42c 0.22d 0.40 1.41 0.91 1.32
RSE 19.0c 10.7d 13.0 13.3 8.20 10.6
N 14c 17d 18 15 18 18
aInjected sample (7 replicate) concentration for all compounds (except MIBK for MeOH & PA for Pn, 40 ng μL-1) was 5 ng μL-1
bNot detected, c,dexcluding MIBK and PA, respectivly to avoid biased calculation

Detection limits (DL) are key criteria to assess the sensitivity performance of any analytical system. For the DL calculation, the concept of MDL was taken in this study. The MDL value for each compound was derived by multiplying the SD of seven replicate analyses with 3.14 (Student’s t-value at the 99.9% confidence interval) and then divided by the relevant RF value (US-EPA, 1986). The average MDLs (ng) of pentane, methanol, and hexane were computed 0.50±0.22, 0.59±0.42, and 0.73±0.40, respectively. As PA (in pentane solvent) and MIBK (in methanol solvent) co-eluted with solvent, they generally had higher MDL values (PA (11.90 ng) and MIBK (5.25 ng)). Hence, these MDL values were not included in the computation of the mean MDL for each solvent. These results suggest that pentane is a better solvent choice for lower MDL values. If we consider reproducibility of analysis, pentane gave the lowest RSE values for all our target VOCs. The average VOC RSE values for Hx, MeOH, and Pn solvents were 2.92±1.32, 2.73±1.41, and 2.61±0.91%, respectively. The MDL values found in this study were significantly larger (10-30 times higher) than Agilent’s GC-FID claimed DL values for propane on a C mass basis. Average MDL values found in this study are highly comparable with the GC-MS/olfactometry results of Zhang et al. (2010). They determined PPA, BTA, IVA, VLA, and 11 other VOCs using the thermal desorption method. Jia et al. (2006) used TD-GC-MS method for airborne VOCs, while Cavalcante et al. (2010) used HS-GC-PID-FID method for the VOCs in environmental aqueous matrices. Their MDL values for aromatics were about 5 to 10 times lower than our study. If we compare average precision level (2.61%-2.91%) of all solvent types determined in this analysis, our values are highly comparable with those reported previously by Jia et al. (2006) in a range of 1.9-4.9%.

4. 4 Comparability of Calibration Results between Experiment and Prediction

In practice, gaseous samples of VOCs are collected using adsorptive cartridges (or bag sampler) and subjected to laboratory GC analysis. To quantify pollutants in a real sample, it is often desirable to carry out calibration using an internal standard. Realistically, because of the presence of complex mixtures of organic compounds in environmental samples, it is nearly impossible or impractical to prepare and spike all authentic analytical standards into the matrix (either gas or liquid) of a sample. Many researchers have been trying to resolve this issue through the generation of predictive equations to compensate the unavailability of authentic standards such as the use of effective carbon number (ECN) concept (Faiola et al., 2012; Kállai et al., 2001; Jones, 1998). According to Morris and Chapman (1961), the basic concept of ECN was described and stated as, “The molar response of a compound in the hydrogen flame detector can be estimated by summing the atomic contributions and multiplying by a proportionality constant characteristic of burner configuration and operating conditions employed.” Based on such theory, the response characteristics of VOCs (e.g., RF values of a given VOC) can be predicted as a function of ECN (Szulejko et al., 2013).

In our study, as a means to test compatibility of each solvent in GC-FID analysis, we examined our calibration data for three different solvents (methanol, pentane, and hexane) in relation to ECN theory. We converted all experimental RF values of each solvent to molar RF values referenced to benzene with the following equation.

RF (mole)= (RF (mass)*MW of each VOC)/MW of Benzene

Then, based on the ECN concept, we constructed an additive scheme (Eqn. 1) using selected molecular descriptors.

ECN=CN+a*CNE H+b*CNE O+c*CNE-CH3+d*CNE-CH2-+e*CNE-O-+f*CNE>C=O+g*CNE-CHO+h*CNE-CO2H+i*CNE B ring+J*CNE>C=C<(Eqn. 1) 

In Eqn. 1, the concept of carbon number equivalent (CNE) is used for each atom or functional group descriptor (D=C, H, O, (-CH3), (-CH2-), (-O- or -OH), (>C=O), (-CHO), (-CO2H), (Bnz ring) and aliphatic >C=C<). In addition, a, b, c, d, e, f, g, h, i, and j are the number of occurrences of each descriptor in a given VOC’s molecular structure. Hence, the best R2 values (Fig. 4) are obtained by plotting the relationship between RF vs. ECN. The output of ECN calculation is presented in Table 5. The molar responses of 19 VOCs for each solvent are compared against carbon number (CN) and effective carbon number (ECN) in Fig. 4A and 4B, respectively. In both cases, the slope values for three solvents were found in the order of methanol>hexane>pentane. However, in terms of correlation coefficient (R2), a reversed order (pentane>hexane>methanol) has been seen.

Fig. 4. 
Comparative analysis of calibration results of 19 VOCs for three solvent with carbon number (A) and with effective carbon number (B).

Table 5. 
Comparison of RF value for different solvent using effective carbon number (ECN) method.
Group Short
CNa ECNc RF from experimentd RF from calculation ΔRF error (%)
MeOHf Pnf Hxf MeOH Pn Hx MeOH Pn Hx MeOH Pn Hx
Aldehyde AA 2 1.46 0.70 0.86 2457 - b - 5183 -567e -5239e 111e - -
PA 3 2.13 1.58 1.78 32935 7921 7796 22280 7949 5750 32 0.4 26
BA 4 2.80 2.46 2.70 31028 17505 18808 39378 16466 16739 27 5.9 11
IA 5 3.44 3.58 3.79 - 26801 27338 55710 27304 29758 - 1.9 9
VA 5 3.47 3.34 3.62 - 19293 25154 56476 24982 27728 - 29.5 10
Aromatic B 6 5.98 6.00 6.00 - 50501 55038 95010 50724 56155 - 0.4 2
T 7 6.63 7.00 7.01 65376 59388 66690 111597 60402 68219 71 1.7 2
p-X 8 7.28 8.00 8.02 184535 66751 75531 128185 70079 80283 31 5.0 6
m-X 8 7.28 8.00 8.02 100698 73014 82605 128185 70079 80283 27 4.0 3
o-X 8 7.28 8.00 8.02 148459 72208 83657 128185 70079 80283 14 2.9 4
S 8 7.64 8.00 7.90 137374 71087 78993 137372 70079 78850 0.0 1.4 0
MEK 4 1.88 2.18 3.84 - 16750 30333 15901 13756 30355 - 17.9 0
MIBK 6 3.19 4.18 5.85 12359 31448 54014 49331 33111 54364 299e 5.3 1
BuAc 6 3.47 3.94 5.68 76327 30725 51202 56476 30788 52333 26 0.2 2
i-BuAl 4 4.04 3.03 4.54 57765 21948 38365 71022 21982 38716 23 0.2 1
Fatty acid PPA 3 1.68 2.08 3.03 24798 14167 17072 10797 12788 20680 56 9.7 21
BTA 4 2.35 2.96 3.95 37417 22798 32588 27895 21304 31669 25 6.6 3
IVA 5 2.99 4.08 5.04 49313 31856 53134 44227 32143 44689 10 0.9 16
VLA 5 3.02 3.84 4.87 44263 29676 41196 44992 29821 42658 2 0.5 4
Mean 4.11 4.37 4.97 67007 36880 46640 64642 36880 46640 26.5 5.24 6.73
SD 2.16 2.40 2.17 53322 22255 23873 43725 22155 23684 19.6 7.46 7.58
RSE 12.05 12.59 10.00 21 14 12 19 14 12 22.1 33.7 26.5
N 19 19 19 15 18 18 19 18 18 13 18 18
aCarbon number, bNot detected
cECN=CN+a*(CNE H)+b*(CNE O)+c*(CNE-CH3)+d*(CNE-CH2-)+e*(CNE-O-)+f*(CNE>C=O)+g*(CNE-CHO)+h*(-CO2H)+i*(CNE Bnz)+j*(CNE >C=C<)
dRF values converted to molar RF in referenced to benzene, RF (mole)=(RF (mass)*MW of each VOC)/MW of Benzene
eOutlier values were excluded for statistical calculation
fCalculated carbon number equivalent (CNE) of each descriptor obtained for three different solvent as follows:

Factor a b c d e f g h i j
MeOH -0.17 0.25 -0.01 0.01 1.50 -1.00 -0.10 -0.80 1.00 0.00
Pn 0.00 0.00 0.00 -0.12 -0.85 -1.70 -1.30 -0.80 0.00 0.00
Hx 0.00 0.00 0.01 -0.08 0.60 -0.10 -1.15 0.10 0.00 -0.10

These findings suggest the possibility that the higher responses associated with a selected solvent should be limited to the analysis of complex mixture of analyte. Although pentane and hexane show acceptable correlation coefficients, methanol represent the least acceptable correlation coefficient (R2: 0.692 (experimental) and 0.775 (calculated)). As methanol has yielded different RF values for similar CN/ECN, its reliability (or predictability) in the analysis of complex VOC mixture seemed to decrease. If we compute the mean of ΔRF error (%) (Table 5), then error percentage is found on the descending order of methanol (26.5), hexane (6.73), and pentane (5.24). Although ΔRF error (%) mean for 19 VOCs was the highest for methanol, it yielded zero error percentage for S (styrene). Another solvent hexane also exhibited zero error for S and MEK. Like other calculations (VOC elution pattern, MDL, and RSE), pentane and hexane also exhibited similar types of ΔRF error (%) in ECN calculation approaches. However, hexane yielded 27% higher response than pentane. Thus, overall finding of our study suggests that hexane is good enough for the analysis of complex mixture of VOCs (e.g., 19 VOCs) by GC-FID. In addition, it is noteworthy that the relative ordering in mean response factor values between solvents (methanol>hexane>pentane) in fact inversely correlates with their vapor pressure (calculated using the Antoine equation) (pentane (99.6 kPa at 35°C)>hexane (31.2 kPa at 35°C)>methanol (28.5 kPa at 35°C) (Linstrom and Mallard, 2013)). As such, it is expected that the physical parameter such as solvent vapor pressure should make certain contributions on the GC performance characteristics of VOCs analysis.

At last, it may be worth considering the effect of the GC initial temperature on the sensitivity between solvent types. Most importantly, it was technically impractical in our laboratory to lower the GC initialization temperature from 35 to say, 20°C so that hexane’s vapor pressure (17.7 kPa) became low enough for on-column condensation to occur to permit solvent trapping. However, if such a situation was possible, then the RF values for hexane solvent would have increased similar to those of the methanol counterpart. Even at 35°C GC initialization temperature, hexane showed limited solvent trapping to yield a 27% enhanced RF values compared to pentane. Bear in mind that the BP of pentane (36°C at 1 atm) is too low (or its 20°C vapor pressure of 57.7 kPA is too high) to permit solvent trapping. The BP and vapor pressure data were from NIST (Linstrom and Mallard, 2013).

5. Conclusion

In this research, the relative response characteristics of 19 target VOCs were evaluated using three different solvents (methanol, pentane and hexane). Most of the selected VOCs for this comparative purpose are well known regulated offensive odorants by many government authorities. Among the commonly used GC detectors, the most widely used GC-FID system has been employed to assess the solvent effect on VOCs quantification. To compare relative response of target VOCs with different solvents, we evaluated the acquired calibration data for standards in each solvent in two different aspects. Firstly, the difference in observed sensitivities of target VOC between different solvent types was used as the main criterion. Accordingly, comparison of the mean RF values of detected VOCs (common to all three different solvents) showed that methanol was the largest in this respect exhibiting an 84% enhancement relative to the lowest, pentane. Moreover, as a means to assess the reliability of each solvent type, the extent of sensitivity deviation across target homologues has also been estimated in terms of ΔRF for each corresponding solvent between experimental and calculated RF data based on ECN concept. The mean ΔRF values for methanol, hexane, and pentane were calculated as 26.5, 6.73, and 5.24%, respectively. The use of methanol hence appeared to be unreliable in the sense of the weakest predictability between homologues (e.g., ECN concept), despite the fact that it had the highest response of all three solvents.

The intricacy of solvent selection was also recognized, when the sensitivity trends for each solvent were compared across different target compounds. Although methanol showed the best sensitivity for aromatic hydrocarbons, it was not so for the other VOCs in the 19 VOC mixture. It was also found that hexane yielded more or less similar results as pentane, while overall, hexane maintained a 27% greater response advantage than pentane. It may hence suggest that hexane can be a potentially good choice for the analysis of complex mixtures of VOCs (e.g., 19 VOCs) through GC-FID, if the GC initialization temperature was low enough to permit solvent trapping. The overall findings of the solvent effect in VOC analysis, as demonstrated in this study, suggest that the solvent selection should be one of the crucial factors in determining GC performance in the analysis of specific target compounds. Thus, a thorough consideration on such variable is recommended in the experimental design stage so as to optimize the GC performance in VOC analysis.


This study was supported by a grant (14182MFDS977) from the Ministry of Food and Drug Safety, Korea, in 2015. The corresponding author (KHK) also acknowledges support made in part by grants from the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (No. 2016R1E1A1A01940995).

1. Ahn, J.-W., Pandey, S.K., Kim, K.-H., (2011), Comparison of GC-MS calibration properties of volatile organic compounds and relative quantification without calibration standards, Journal of Chromatographic Science, 49, p19-28.
2. Beauchamp, R.O., Bus, J.S., Popp, J.A., Boreiko, C.J., Goldberg, L., McKenna, M.J., (1983), A Critical Review of the Literature on Carbon Disulfide Toxicity, Critical Reviews in Toxicology, 11(3), p169-278.
3. Binder, R.G., Flath, R.A., Mon, T.R., (1989), Volatile components of bittermelon, Journal of Agricultural and Food Chemistry, 37, p418-420.
4. Binder, R.G., Turner, C.E., Flath, R.A., (1990), Volatile components of purple starthistle, Journal of Agricultural and Food Chemistry, 38, p1053-1055.
5. Boeglin, M.L., Wessels, D., Henshel, D., (2006), An investigation of the relationship between air emissions of volatile organic compounds and the incidence of cancer in Indiana counties, Environmental Research, 100, p242-254.
6. Cavalcante, R.M., de Andrade, M.V.F., Marins, R.V., Oliveira, L.D.M., (2010), Development of a headspacegas chromatography (HS-GC-PID-FID) method for the determination of VOCs in environmental aqueous matrices: Optimization, verification and elimination of matrix effect and VOC distribution on the Fortaleza coast, Brazil, Microchemical Journal, 96, p337-343.
7. Chida, M., Sone, Y., Tamura, H., (2004), Aroma characteristics of stored tobacco cut leaves analyzed by a high vacuum distillation and canister system, Journal of Agricultural and Food Chemistry, 52, p7918-7924.
8. Cremades, L.V., (2004), Simulation of the dispersion of VOC from an MSW landfill via a lagrangian particle model: Efficiency of a solid barrier, Environmental Engineering Science, 21, p291-302.
9. Culleré, L., Escudero, A., Cacho, J., Ferreira, V., (2004), Gas chromatography-olfactometry and chemical quantitative study of the aroma of six premium quality Spanish aged red wines, Journal of Agricultural and Food Chemistry, 52, p1653-1660.
10. Evtyugina, M.G., Nunes, T., Alves, C., Marques, M.C., (2009), Photochemical pollution in a rural mountainous area in the northeast of Portugal, Atmospheric Research, 92, p151-158.
11. Faiola, C., Erickson, M., Fricaud, V., Jobson, B., Van-Reken, T., (2012), Quantification of biogenic volatile organic compounds with a flame ionization detector using the effective carbon number concept, Atmospheric Measurement Techniques, 5, p2415-2447.
12. Gonzalez, F.R., Nardillo, A.M., (1999), Retention index in temperature-programmed gas chromatography, Journal of Chromatography A, 842, p29-49.
13. Goodner, K.L., (2008), Practical retention index models of OV-101, DB-1, DB-5, and DB-Wax for flavor and fragrance compounds, LWT-Food Science and Technology, 41, p951-958.
14. Grob, K., (2001), Split and splitless injection for quantitative gas chromatography, 4th ed., WILEY-VCH Verlag GmbH, Weinheim (Germany), p37.
15. Guthrie, J.P., (1975), Carbonyl addition reactions: Factors affecting the hydrate-hemiacetal and hemiacetal-acetal equilibrium constants, Canadian Journal of Chemistry, 53, p898-906.
16. Harper, M., (2000), Sorbent trapping of volatile organic compounds from air, Journal of Chromatography A, 885, p129-151.
17. Héberger, K., Görgényi, M., (1999), Principal component analysis of Kováts indices for carbonyl compounds in capillary gas chromatography, Journal of Chromatography A, 845, p21-31.
18. Hyötyläinen, T., Grob, K., Riekkola, M.L., (1997), Reversed phase HPLC coupled on-line to GC by the vaporizer/precolumn solvent split/gas discharge interface; analysis of phthalates in water, HRC-Journal of High Resolution Chromatography, 20, p410-416.
19. Jia, C., Batterman, S., Chernyak, S., (2006), Development and comparison of methods using MS scan and selective ion monitoring modes for a wide range of airborne VOCs, Journal of Environmental Monitoring, 8, p1029-1042.
20. Jones, F.W., (1998), Estimation of flame-ionization detector relative response factors for oligomers of alkyl and aryl ether polyethoxylates using the effective carbon number concept, Journal of Chromatographic Science, 36, p223-226.
21. Kállai, M., Veres, Z., Balla, J., (2001), Response of flame ionization detectors to different homologous series, Chromatographia, 54, p511-517.
22. Kavouras, I.G., DuBois, D.W., Etyemezian, V., Nikolich, G., (2013), Spatiotemporal variability of ground-level ozone and influence of smoke in Treasure Valley, Idaho, Atmospheric Research, 124, p44-52.
23. Kim, K.H., Susaya, J., Sohn, J.R., Brown, R.J.C., (2011), A study of relative performance between direct injection and thermal desorption for several volatile organic acids and reference VOCs by gas chromatographymass spectrometry, Fresenius Environmental Bulletin, 20, p2875-2882.
24. KMOE, (2010), Korean ministry of environment, annual report of ambient air quality in Korea.
25. Lal, S., Sahu, L.K., Venkataramani, S., Mallik, C., (2012), Light non-methane hydrocarbons at two sites in the Indo-Gangetic Plain, Journal of Environmental Monitoring, 14(4), p1158-1165.
26. Le Guen, S., Prost, C., Demaimay, M., (2000), Characterization of odorant compounds of mussels Mytilus edulis) according to their origin using gas chromatographyolfactometry and gas chromatography-mass spectrometry, Journal of Chromatography A, 896, p361-371.
27. Lilja, J., Aumo, J., Salmi, T., Murzin, D.Y., Mäki-Arvela, P., Sundell, M., Ekman, K., Peltonen, R., Vainio, H., (2002), Kinetics of esterification of propanoic acid with methanol over a fibrous polymer-supported sulphonic acid catalyst, Applied Catalysis A: General, 228, p253-267.
28. Linstrom, P.J., Mallard, W.G., (2013), NIST Chemistry WebBook, NIST Standard Reference Database Number 69, National Institute of Standards and Technology, Gaithersburg MD, 20899, USA, (available at:
29. Morris, R.A., Chapman, R.L., (1961), Flame lonization hydrocarbon analyzer, Journal of the Air Pollution Control Association, 11, p467-489.
30. Namieśnik, J., Zygmunt, B., Jastrzȩbska, A., (2000), Application of solid-phase microextraction for determination of organic vapours in gaseous matrices, Journal of Chromatography A, 885, p405-418.
31. Pal, R., Kim, K.-H., (2008), Gas chromatographic approach for the determination of carbonyl compounds in ambient air, Microchemical Journal, 90, p147-158.
32. Pang, X., Lewis, A.C., Hamilton, J.F., (2011), Determination of airborne carbonyls via pentafluorophenylhydrazine derivatisation by GC-MS and its comparison with HPLC method, Talanta, 85, p406-414.
33. Qian, M., Reineccius, G., (2003), Potent aroma compounds in Parmigiano Reggiano cheese studied using a dynamic headspace (purge-trap) method, Flavour and Fragrance Journal, 18, p252-259.
34. Rahman, M.M., Kim, K.-H., (2013), Parallel analysis of volatile fatty acids, indole, skatole, phenol, and trimethylamine from waste-related source environments, Journal of Chromatography A, 1314, p241-248.
35. Ras, M.R., Borrull, F., Marcé, R.M., (2009), Sampling and preconcentration techniques for determination of volatile organic compounds in air samples, TrAC Trends in Analytical Chemistry, 28(3), p347-361.
36. Sahu, L.K., (2012), Volatile organic compounds and their measurements in the troposphere, Current Science, 102(12), p1645-1649.
37. Sahu, L.K., Lal, S., (2006), Distributions of C2-C5 NMHCs and related trace gases at a tropical urban site in India, Atmospheric Environment, 40(5), p880-891.
38. Sahu, L.K., Saxena, P., (2015), High time and mass resolved PTR-TOF-MS measurements of VOCs at an urban site of India during winter: Role of anthropogenic, biomass burning, biogenic and photochemical sources, Atmospheric Research, 164-165, p84-94.
39. Sanz, C., Ansorena, D., Bello, J., Cid, C., (2001), Optimizing headspace temperature and time sampling for identification of volatile compounds in ground roasted arabica coffee, Journal of Agricultural and Food Chemistry, 49, p1364-1369.
40. Szulejko, J.E., Kim, Y.-H., Kim, K.-H., (2013), Method to predict gas chromatographic response factors for trace level analysis of volatile organic compounds based on Effective Carbon Number concept, Journal of Separation Science, 36, p3356-3365.
41. US-EPA, (1986), Definition and procedure for the determination of the method detection limit - revision 1.11, code of federal regulations, title 40, part 136, appendix b, 1984, Fed. Regist, 51, p23703.
42. Wang, P., Zhao, W., (2008), Assessment of ambient volatile organic compounds (VOCs) near major roads in urban Nanjing, Atmospheric Research, 89, p289-297.
43. White, L.D., Taylor, D.G., Mauer, P.A., Kupel, R.E., (1970), A convenient optimized method for the analysis of selected solvent vapors in the industrial atmosphere, American Industrial Hygiene Association Journal, 31, p225-232.
44. Yanagimoto, K., Ochi, H., Lee, K.-G., Shibamoto, T., (2004), Antioxidative activities of fractions obtained from brewed coffee, Journal of Agricultural and Food Chemistry, 52, p592-596.
45. Yassaa, N., Ciccioli, P., Brancaleoni, E., Frattoni, M., Meklati, B.Y., (2011), Ambient measurements of selected VOCs in populated and remote sites of the Sahara desert, Atmospheric Research, 100, p141-146.
46. Zhang, S., Cai, L., Koziel, J.A., Hoff, S.J., Schmidt, D.R., Clanton, C.J., Jacobson, L.D., Parker, D.B., Heber, A.J., (2010), Field air sampling and simultaneous chemical and sensory analysis of livestock odorants with sorbent tubes and GC-MS/olfactometry, Sensors and Actuators B: Chemical, 146, p427-432.