Journal of Chromatography A, 1325 (2014) 40–48

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Estimation of measurement uncertainty of polychlorinated biphenyls, polycyclic aromatic hydrocarbons and organochlorine pesticides in the atmosphere using gas chromatography–mass spectrometry and gas chromatography–electron capture detector Güler Aslan-Sungur a,∗ , Eftade O. Gaga b , Serpil Yenisoy-Karakas¸ c a

Department of Environmental Engineering, Abant Izzet Baysal University, 14280 Bolu, Turkey Department of Environmental Engineering, Anadolu University, 26470 Eskis¸ehir, Turkey c Department of Chemistry, Abant Izzet Baysal University, 14280 Bolu, Turkey b

a r t i c l e

i n f o

Article history: Received 1 October 2013 Received in revised form 2 December 2013 Accepted 3 December 2013 Available online 11 December 2013 Keywords: Gas chromatography Polychlorinated biphenyls Organochlorine pesticides Polycyclic aromatic hydrocarbons Air samples Uncertainty

a b s t r a c t Estimation of uncertainty of measurement is a crucial issue to achieve accurate measurement results. When the target has adverse environmental and health effects, accuracy of the results become more important. POPs are the pollutants that have toxic effects and unfortunately, there is a lack of information about uncertainty of the method for determining POPs in air samples. In this work, uncertainty calculations were carried out for PCBs, OCPs, and PAHs in air samples analyzed by using GC–MS and GC–ECD. The main dominant sources for combined uncertainty were calibration curve, recovery and repeatability. The relative uncertainties were found to be in the range of 23–52% for PCBs, 24–59% for OCPs and 23–90% for PAHs. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Persistent organic compounds (POPs) are the most important class of organic pollutants that consist of three important groups which are polychlorinated biphenyls (PCBs), organochlorine pesticides (OCPs) and polycyclic aromatic hydrocarbons (PAHs). Because of their semi-volatility, POPs are transported at low concentrations over long distances in the atmosphere. POPs are characterized by low water solubility and high lipid solubility, resulting in their bioaccumulation in fatty tissues of living organisms. Therefore, both humans and environmental organisms are exposed to POPs around the world which has chronic and acute toxic effects. PCBs are persistent organic pollutants that have been widely used in the environment that bioaccumulate and biomagnify in the food chain and resist biodegradation. Health hazard of PCBs were identified in a number of reports and due to their persistent nature in the environment they can easily transported one medium to another such as soil and food [1–5].

∗ Corresponding author. Tel.: +90 5557020970. E-mail addresses: aslan [email protected] (G. Aslan-Sungur), [email protected] (E.O. Gaga), yenisoykarakas [email protected] (S. Yenisoy-Karakas¸). 0021-9673/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chroma.2013.12.005

OCPs are bioaccumulative persistent, and have adverse health effects on human and animal health, causing nervous system damage, diseases of the immune system, reproductive and developmental disorders, and cancers [6–8]. Thus, it is still detected and reported in different compartments of the environment all over the world [9,10]. PAHs are generated during incomplete combustion of organic matter and released into the urban environment mainly through anthropogenic activities such as vehicle emissions, coal and fossil fuel combustion for power generation, petroleum refining, industrial processing, chemical manufacturing, oil spills and coal tars [11–13]. Several PAHs are known carcinogens and many studies have been performed about the health effects of PAHs on human also for environment [14–18]. Many analysis methods were developed to measure persistent organic pollutants in the environment. Analysis of PCBs, PAHs and OCPs in air samples currently performed by gas chromatography (GC) coupled to mass spectrometry (MS) and electron capture detector (ECD). Two main important steps, extraction and cleanup procedure in this study are optimized before quantification by chromatographic techniques [19]. Estimation of measurement uncertainties should be taken into consideration for the determination of POPs due to their adverse

G. Aslan-Sungur et al. / J. Chromatogr. A 1325 (2014) 40–48

effects on humans and environment even at low concentrations. According to the international vocabulary of basic and general terms in metrology [20], the uncertainty of a measurement is defined as “a parameter associated with the result of a measurement that characterizes the dispersion of the values that could reasonably be attributed to the measurand”. The interest of calculating the measurement uncertainties is to show the data quality that is fundamentally important for laboratories, their clients and all institutions using these results for comparative purposes [21,22]. However, there is no published paper about a comprehensive evaluation of uncertainty associated to the determination of POPs in air samples. The aim of this study is to determine the uncertainty measurement calculation of the PCBs, OCPs and PAHs in the atmosphere using bottom-up approach.

2. Experimental 2.1. Target PAH, PCB and OCP compounds The PCB (with IUPAC nomenclature), OCP and PAH compounds investigated in the atmosphere as follows: PCB-18, PCB-20, PCB-28, PCB-31, PCB-44, PCB-52, PCB-101, PCB-105, PCB-118, PCB-138, PCB-149, PCB-153, PCB-170, PCB-180, and PCB-194; Alpha-Hexachlorocyclohexane Beta-Hexachlorocyclohexane (Beta-HCH), (Alpha-HCH), Gamma-Hexachlorocyclohexane (Gamma-HCH) (Lindane), Delta-Hexachlorocyclohexane (Delta-HCH), Dieldrin, 4,4 -DDT, Heptachlor epoxide, Endosulfan I, 4,4 -DDE, Endrin, Endosulfan II, 4,4 -DDD, Endrin aldehyde, Endosulfan sulfate, Endrin ketone, and Methoxychlor; Acenaphtylene (Acy), Acenapthene (Ace), Fluorene (Flu), Anthracene (Ant), Phenanthrene (Phe), Fluoranthene (Flt), Pyrene (Pyr), Benzo(a)anthracene (BaA), Chrysene (Chr), Benzo(b)fluoranthene (BbF), Benzo(k)fluoranthene (BkF), Benzo(a)pyrene (BaP), Dibenzo[a,h]anthracene (DahA), Indeno[1,2,3-c,d]pyrene (Ind), Benzo[g,h,i]perylene (BghiP).

2.2. Sample preparation In this study, high volume PUF sampler (Andersen PUF sampler, USA) was used with a flow of 0.225 m3 min−1 . After twenty-four hours, the samples were taken and wrapped with aluminum foil and stored in desiccators until the extraction time. Particle phase organics were collected by a 90-mm diameter glass fiber filter (GFF), while gaseous phase organics were captured by a two-piece polyurethane foam (PUF) cartridge. The samples were collected in two phases, during two summer months (June 25–August 23, 2007) and two winter months (December 13, 2007–February 12, 2008). A total of 120 samples were collected and analyzed in both seasons. Twelve field blank samples (6 for GFF, 6 for PUF) were collected to determine any contamination during sampling, sample handling, sample preparation, and analyses. Pre-baked GFF and pre-cleaned PUF cartridges were brought to the sampling location in sealed vessels and then placed into the sampler. They were kept in the sampler for several minutes prior to operation. GFF and PUF samples were soxhlet-extracted with a mixture of dichloromethane (DCM) and petroleum ether (PE) (1:4) with volumes of 150 mL and 650 mL, respectively, for 24 h, at 4 cycles per hour. The volume of the extracts was reduced to 10 mL by rotary evaporator at 500 mbar and 40 ◦ C. They were transferred to 15 mL vials by washing the walls of the balloons with solvent mixture (DCM:PE, 1:4) several times for further volume reduction to 1 mL under a gentle stream of nitrogen. Extracts were transferred to aluminum oxide–florisil columns for cleanup and fractionation of the analytes.

41

The adsorption column was 10 cm long and 0.5 cm width (diameter), and prepared by placing glass wool in the tip of the column to support sorbent material. Aluminum oxide, florisil, and sodium sulfate (1 g each) were put in to the column. The sodium sulfate was used to reduce water and moisture content in the samples on the top of the column. The analytical procedure after the extraction procedure is illustrated in Fig. 1. 2.3. Chemicals and materials All solvents and reagents used in the study were of chromatographic grade. PCB congener standard (100 ␮g mL−1 ), internal standard (PCB-30 and PCB-204, 100 ␮g mL−1 ), and surrogate standard (PCB-14, PCB-65, PCB-166, 100 ␮g mL−1 ) were purchased from Absolute Standards (USA). OCP standard solution (2000 ␮g mL−1 ) was obtained from Dr. Ehrenstorfer (Germany). OCP internal standard (Pentachloronitrobenzene, 100 ␮g mL−1 ) and OCP surrogate solution (2,4,5,6-Tetrachloro-m-xylene and Dibutyl chlorendate, 2000 ␮g mL−1 ) were obtained from Absolute Standards (USA) and Fisher Scientific (UK), respectively. PAH standards (PM 610, 100 ␮g mL−1 ) were purchased from AccuStandards (USA). PAH surrogate solution (Acenaphthene-d10, Crysene-d12, Perylene-d12, Phenanthrene-d10, 1000 ␮g mL−1 ) was from Ultra Scientific (USA). Sodium sulfate, florisil (0.150–0.250 mm) and neutral aluminum oxide (0.063–0.2 mm) were purchased from Merck Company (USA). The granulated sodium sulfate was anhydrous and of 99% purity. Glass microfiber filters (90 mm dia., pore size: 2.7 ␮m) were from Whatman Company (USA). All of the samples were spiked with surrogate standards prior to their extraction. Prior to extraction, 20 ␮L of 32 ␮g mL−1 deuterated PAH mixture, 40 ␮g mL−1 OCP surrogates, and 50 ␮g mL−1 PCB surrogates were added to the samples in order to determine the recovery efficiencies for checking the internal quality control. Calibration standards were prepared from the main stock standards with appropriate dilutions with n-hexane. The calibration curve with intercept, the coefficient of determination (r2 ), residual standard deviation (S) and slope (B1 ) data were reported in Table 1. All of the stock, intermediate and standard solutions were stored in the refrigerator. The standard solutions were prepared for the PCBs, OCPs and PAHs at five points. For PCBs calibration, 5, 10, 25, 50, and 100 ppb standard solutions were used. PCB-30 and PCB-204 were used as internal standards. PCB-30 as an internal standard was used for the calculation of PCB-18, PCB-20, PCB-28, PCB-31, PCB-44, PCB-52, PCB-101, PCB-105 and PCB-118. The rest of the PCBs were calculated by using PCB-204 responses. The calibration points were 0.01, 0.05, 0.1, 0.2, and 0.5 ppm for OCPs. Pentachloronitrobenzene was used as an internal standard for OCPs and 20 ␮L from 1 ppm added into standard solutions and samples. The OCPs were analyzed with GC–ECD. The internal standard calibrations were performed for both PCBs and OCPs, however external calibration was used for PAHs since internal standards were not available in our laboratory for PAHs. GC–MS was calibrated with for standard mixtures of PAHs using the concentrations of 0.1, 0.25, 0.5, 1 and 2 ppm. 2.4. Instrumentation and applied methods A Hewlett Packard (HP) 6890N series gas chromatograph equipped with the electron capture detector (Agilent, USA) was utilized for analysis of the OCPs. A 60 m, 0.25 mm id., 0.25 ␮m film thickness, cross linked 5% phenyl methyl siloxane, HP 1MS, capillary column (Agilent Tech.) and 15 mCi of nickel 63 type ECD detector was used for separation and detection of OCPs. A HP 6890N series gas chromatograph (Agilent, USA) coupled with HP 5973 mass spectrometer (Agilent, USA) was used for the

42

G. Aslan-Sungur et al. / J. Chromatogr. A 1325 (2014) 40–48

Table 1 Target PAH, PCB and OCP compounds with their calibration curve values. Compound

PCBs PCB-18 PCB-20 PCB-28 PCB-31 PCB-44 PCB-52 PCB-101 PCB-105 PCB-118 PCB-138 PCB-149 PCB-153 PCB-170 PCB-180 PCB-194 OCPs Beta HCH Heptachloro epoxide Endosulfan I 4,4 -DDE Endrin Endosulfan II 4,4 -DDD Endrin aldehyde Endosulfan sulfate Endrin ketone Methoxychlor Alpha HCH Gamma HCH Delta HCH Dieldrin 4,4 -DDT PAHs Acy Ace Flu Phe Ant Flt Pyr BaA Chr BbF BkF BaP Ind DahA BghiP

Standard stock solution concentration (␮g mL−1 ) 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100

Calibration concentration (␮g mL−1 )

r2

Intercept

B1

S

0.005–0.1 0.005–0.1 0.005–0.1 0.005–0.1 0.005–0.1 0.005–0.1 0.005–0.1 0.005–0.1 0.005–0.1 0.005–0.1 0.005–0.1 0.005–0.1 0.005–0.1 0.005–0.1 0.005–0.1

0.9971 0.9964 0.9983 0.9983 0.9994 0.9995 0.9969 0.9997 0.9986 0.9968 0.9984 0.9967 0.9980 0.9966 0.9964

−0.0071 −0.0114 −0.0156 −0.0177 −0.0018 0.0017 −0.0096 0.0004 −0.0072 −0.0077 −0.0058 −0.0180 −0.0086 0.0183 −0.0508

0.0073 0.0094 0.0203 0.0203 0.0055 0.0061 0.0055 0.0073 0.0080 0.0111 0.0106 0.0125 0.0082 0.0090 0.0080

0.0179 0.0254 0.0381 0.0372 0.0061 0.0061 0.0138 0.0055 0.0136 0.0283 0.0191 0.0325 0.0164 0.0237 0.0216

0.01–0.5 0.01–0.5 0.01–0.5 0.01–0.5 0.01–0.5 0.01–0.5 0.01–0.5 0.01–0.5 0.01–0.5 0.01–0.5 0.01–0.5 0.01–0.5 0.01–0.5 0.01–0.5 0.01–0.5 0.01–0.5

0.9998 0.9997 0.9997 0.9997 0.9996 0.9996 0.9997 0.9997 0.9995 0.9998 0.9995 0.9993 0.9992 0.9946 0.9997 0.9915

0.0016 0.0105 0.0085 0.0055 0.0005 0.0004 0.0004 0.0027 −0.0013 −0.0030 −0.0380 −0.0341 −0.0306 −0.0430 0.0031 −0.0256

0.2392 1.3860 1.1561 1.2533 0.5919 0.9395 1.0028 0.7562 0.2710 0.7256 0.2180 1.6321 1.4201 0.7678 1.3128 0.4789

0.0016 0.0108 0.0088 0.0094 0.0052 0.0081 0.0076 0.0064 0.0043 0.0048 0.0022 0.0194 0.0178 0.0257 0.0109 0.0202

0.1–2 0.1–2 0.1–2 0.1–2 0.1–2 0.1–2 0.1–2 0.1–2 0.1–2 0.1–2 0.1–2 0.1–2 0.1–2 0.1–2 0.1–2

0.9997 1.0000 0.9998 1.0000 0.9995 0.9989 0.9988 0.9938 0.9977 1.0000 0.9912 0.9765 0.9886 0.9828 0.9817

−1.4 × 105 −4.9 × 104 −1.0 × 105 −2.0 × 105 −2.7 × 105 −3.4 × 105 −3.8 × 105 −5.2 × 106 −5.4 × 105 −8.8 × 105 −7.6 × 105 −4.9 × 105 −1.3 × 105 −2.0 × 105 −3.8 × 105

5.0 × 106 3.0 × 106 3.0 × 106 5.0 × 106 5.0 × 106 6.0 × 106 6.0 × 106 4.0 × 106 6.0 × 106 4.0 × 106 5.0 × 106 3.0 × 106 9.0 × 105 1.0 × 106 2.0 × 106

6.4 × 105 4.0 × 104 5.3 × 105 9.6 × 105 1.8 × 105 2.8 × 105 2.1 × 105 6.8 × 105 2.7 × 105 1.2 × 106 7.6 × 105 4.4 × 105 9.3 × 104 3.2 × 105 5.3 × 105

Table 2 Operating GC–MS and GC–ECD conditions for PAHs, PCBs and OCPs. Operating parameters

PCBs

OCPs

PAHs

GC column

30 m × 250 ␮m × 0.25 ␮m 5% Phenyl methyl siloxane, HP 5 MS, capillary column Ultra purified helium 99.999%, 1 ml min−1 Splitless 280 ◦ C 70 ◦ C (2 min), 25 ◦ C min−1 to 150 ◦ C (1 min), 3 ◦ C min−1 to 200 ◦ C (1 min), 8 ◦ C min−1 to 280 ◦ C (5 min) 1 ␮l MS

60 m × 250 ␮m × 0.25 ␮m 5% Phenyl methyl siloxane, HP 1 MS, capillary column Ultra purified helium 99.999%, 1 ml min−1 Splitless 250 ◦ C 50 ◦ C (1 min), 25 ◦ C min−1 to 170 ◦ C (5.8 min), 5 ◦ C min−1 to 300 ◦ C (2 min)

30 m × 250 ␮m × 0.25 ␮m 5% Phenyl methyl siloxane, HP 5 MS, capillary column Ultra purified helium 99.999%, 1 ml min−1 Splitless 280 ◦ C 70 ◦ C (4 min), 7 ◦ C min−1 to 250 ◦ C (5 min), 5 ◦ C min−1 to 300 ◦ C (8 min)

1 ␮l ECD

1 ␮l MS

Mobile phase Type of injection Temperature of the injection port Temperature programs

Injection volume Detector

G. Aslan-Sungur et al. / J. Chromatogr. A 1325 (2014) 40–48

43

Fig. 1. Scheme of analytical procedure.

analysis. A 30 m, 0.25 mm id., 0.25 mm film thickness, cross linked 5% phenyl methyl siloxane, HP 5MS, capillary column (Agilent Tech.) was used for the separation of PAHs and PCBs throughout the study. Agilent 7683B series automatic injector (Agilent, USA) was used in both instruments. The operating conditions for the instruments for PCBs, OCPs, and PAHs were given in Table 2. The MS was operated in electron impact mode (70 eV) and PCBs and PAHs were identified on the basis of their retention time, target and qualifier ions. The quadrupole temperature was set to 150 ◦ C. The source of mass instrument was operated at 230 ◦ C. Nitrogen gas was used as a make-up gas in the GC–ECD instrument with 99.999% purity at a rate of 30 mL min−1 . 2.5. Blank results Filter blanks and PUF blanks were subjected to the same analytical procedure applied to the samples. Field and laboratory blank samples were routinely analyzed in order to evaluate analytical bias and precision. Blank levels of individual compounds were normally

very low and, in most cases, not detectable. The amounts in the field blank samples were less than 0.4% of the sample amounts for PCBs and for PAHs, and less than 0.03% of the sample amount for OCPs. Therefore, no blank correction was applied to the results. 3. Results and discussions 3.1. Estimation of uncertainty 3.1.1. Specification of the measurand The PCB, OCP and PAH concentrations in the air samples, expressed in ng m−3 , ␮g m−3 and ␮g m−3 , are obtained from the following model equation, respectively: Concentration (ng m−3 ) =

C (ng mL−1 ) × Vsample (mL) Vair (m3 )

(1)

where C is the concentration of the target compound obtained from the calibration (ng mL−1 and ␮g mL−1 ), VSample is the final

44

G. Aslan-Sungur et al. / J. Chromatogr. A 1325 (2014) 40–48

Vsample

Recovery temperature calibration

Concentration

Repeatability

Certified conc.of elements Certified Conc.(IS)

Dilution

Flow rate

temperature

calibration

calibration

Timer Calibration curve calibration

Concentration obtained from calibration curve

Vair

Fig. 2. Cause and effect diagram for the determination of POPs in air samples.

calibration; c0 , analyte concentration in air sample; c¯ , mean value of the different calibration standards (n number of measurements); i: index for the number of measurements to obtain the calibration curve, ci : individual calibration standard value obtained from the calibration equation. In this study, five calibration standards were prepared for PCBs, OCPs and PAHs and three replicate measurements were performed for each solution. In this case, p = 3 and n = 3 × 5 = 15. Thus, for PCB18 the calculation is as follows:

volume of the sample (mL) and VAir is the volume of the air sample (m3 ). 3.1.2. Identification of the uncertainty sources According to the model equation (Eq. (1)), the main parameters that affect the measurand concentrations are obtained from calibration curve (C) and preparation of the standard (std) and internal standard solutions (instd), volume of sample and air (VSample and VAir ), recovery (R) and repeatability (Rep). Thus, the uncertainty sources that contribute to the POPs concentration in air are as follows: (1) (2) (3) (4) (5) (6) (7)



0.0179 u(c0 ) = 0.0073

Calibration curves (C). Stock solutions of standards (std). Stock solution of internal standards (instd). Volume of sample (VSample ). Volume of air (VAir ). Recovery (R). Repeatability (Rep).

1 1 25 − 382 + + = 1.601 ␮g L−1 3 15 10.84

(4)

3.1.3.2. Estimation of the uncertainty derived from standard solution preparation, u(std) and u(instd). Standard preparation for the calibration curve is an important uncertainty source. This uncertainty is a combination of uncertainty of glassware used and stock solution. In this case, the standards were prepared using pipette (pip) and volumetric flask (flask). The related uncertainty data given by manufacturers are introduced in Table 3.

A cause and effect diagram is introduced in Fig. 2 which explains the combination of several uncertainties causes these seven main uncertainty sources. According to these main five uncertainty sources, the standard combined uncertainty of the POPs is calculated with the following equation: uCombined = Conc

  u(C) 2 C

+

 u(std) 2 std

+

 u(instd) 2 instd

 +

u(VSample ) Vsample

3.1.3. Calculation of the uncertainty sources In this paper, we only report a unique analyte (PCB-18) to show the uncertainty calculations in order to minimize the length of the paper. The same calculations were performed for the rest of the analytes. 3.1.3.1. Estimation of the uncertainty derived from a linear calibration curves, u(C). This uncertainty is associated with the calibration curve which is calculated with Eq. (3) defined by Eurachem/Citac guide (2012) [23]:



 1 (c0 − c¯ )2 1 S + + u(c0 ) = Sxx = (ci − c¯ )2 (3) B1 p n Sxx i−1 where S, residual standard deviation; B1 , slope; p, number of measurements to determine c0 ; n, number of measurements for the n

2 +

 u(V ) 2 Air VAir

+

 u(Rep) 2 Rep

+

 u(R) 2 R

(2)

Table 3 The manufacturer specification data of analytical volumetric glassware, and POP standards. Items

Purity (%)

Quantity

Manufacturer’s specification

Volumetric flask Pipette Air sampler (PUF) Hexane thermal expansion PCBs OCPs PAHs Internal standard (PCBs) Internal standard (OCPs)

– – – – – 99 – – –

100 mL 1 mL 324 m−3 day−1 – 100 ␮g mL−1 2000 ␮g mL−1 20 ␮g mL−1 100 ␮g mL−1 100 ␮g mL−1

±0.014 ±0.0023 ±%10 1.41 × 10−3 ◦ C−1 ±0.0202 – ±0.05 ±0.0396 ±0.01965

G. Aslan-Sungur et al. / J. Chromatogr. A 1325 (2014) 40–48

The following Eqs. (5)–(8) are used for calculation of the u(std):

u(std) = Cstd



u(pip) Vpip

2 +

 u(stock) 2 Cstock

+

 u(flask) 2 Vflask

(5)

where u(pip) is the uncertainty of the pipette and u(stock) is the uncertainty of the stock solution. Calculation of the u(pip) is the combination of the uncertainty comes from the calibration of the pipette [u(pipcal)], and the uncertainty source from the temperature effect [u(temp)]. Preparing the standard solution by using a flask, u(flask), and the purity of the stock solution, u(purity) or uncertainty of stock solution, u(stocksoln) should be taken into account in the uncertainty calculation of standard solution. In order to prepare 1000 ␮g L−1 standard solution, we are using 100 ␮g mL−1 PCB stock solution with 0.0202 ␮g L−1 uncertainty (k = 2). u(pip) =



u(flask) =

2

(u(pipcal)) + (u(temp))2

(6)



(u(flaskcal))2 + (u(temp))2

(7)

The temperature effect is the result of the variation in temperature in laboratory which is generally accepted as ± 3 ◦ C in Eurachem/Citac guide [23]. 3×V ×Q u(temp) = 1.73

(8)

where u(temp), standard uncertainty of the temperature effect; V, measured volume (e.g. 1 mL for pipette); and Q, the coefficient of volume expansion of the hexane (Qhexane : 1.41 × 10−3 ◦ C−1 ). Rectangular distribution is accepted. According to the equations given below and the necessary data taken from Table 3, the calculations were performed for PCB-18 as follows: u(temp) =

3 × 1 × 1.41 × 10−3 = 0.0024 mL 1.73

u(temp) =

3 × 100 × 1.41 × 10−3 = 0.014 mL 1.73

u(pip) =



for pipette

(9)

for flask

(10)

(0.0023)2 + (0.0024)2 = 0.0034 mL

(11)

(12)

or u(purity) =

1 − 0.99 = 0.0058 for OCP stock solution 1.73

 0.014 2

u(std) = 1000 100 = 0.00415 ␮g L−1

+

 10.1 2 100000

+

 0.0034 2

(16)

Air sampling was carried out with high volume PUF sampler adjusted to 0.225 m3 min−1 flow rate for 24 h. The volume of air collected in a day is equal to 324 m3 . The manufacturer specification uncertainty for the air sampler is ±%10 (Table 3), thus the uncertainty comes from the air volume is as follows: u(VAir ) = 32.4 m3

(17)

3.1.3.4. Estimation of the uncertainty derived from recovery, u(Rrep ). Repeatability of the analysis gives the variation among the analyses which is defined with standard deviation of the successive analysis. The repetition of the analysis for each analyte (three times) was performed in different days. The uncertainty comes from the repeatability is calculated by using the following equation: RSD u(Rep) = √ n

(18)

where RSD is the relative standard deviation that is obtained by dividing the standard deviation to the mean of the analysis and n is the number of the repetition (in this case, n = 3). In this study, percent recovery of the analyte was determined three times that is reported in Table 4 with their mean of recovery (%), standard deviation (STD) and RSD. According to Table 4, u(Rep) for PCB-18 is calculated as: u(Rep) =

0.054 = 0.031 √ 3

(19)

3.1.3.5. Estimation of the uncertainty derived from recovery, u(R). Recovery experiments were repeated three times for each analyte (Table 4) by applying extraction procedure. Recovery (R) can be obtained from Cobs /CCRM where Cobs : observed concentration after extraction, CCRM : the CRM concentration. In this study, 1 ppm CRMs was prepared and after extraction, cleanup and analysis, the observed mean values reported as mean recovery (%) in Table 4. Uncertainty term for recovery (R) can be calculated by using the following equation: u(R) = R

 u(C

obs )

Cobs

2

+

 u(C

CRM )

2

CCRM

(20)

u(Cobs ) =

0.054 = 0.031 √ 3

(21)

RPCB-18 =

Cobs 0.74 ppm = = 0.74 CCRM 1 ppm

(22)

1 (14)

u(CCRM−PCB−18 ) =

8.308 × 10−3 Uncertainty for 1 ppm = 2 k

= 4.15 × 10−3 ␮g L−1

u(std) = 4.15 ␮g L−1 The same calculation procedure for u(std) was followed to estimate the u(instd) and the same result for PCB-18 was obtained: u(instd) = 4.15 ␮g L−1

u(VSample ) = 0.0034 mL

For PCB-18: (13)

where the k is the coverage factor that is used in the calculation of the expanded uncertainty (k is taken as 2 in the certificate).



3.1.3.3. Estimation of the uncertainty derived from sample volume u(VSample ) and air volume u(VAir ). The sample volume was taken as 1 mL using pipette, which means that the pipette uncertainty [u(pip)] (Eq. (6)) is the source of u(Vsample ). According to Eq. (6), the u(pip) was calculated for PCB-18 as 0.0034 mL, so:



20.2 20.2 u(stocksoln) = = 10.1 ␮g L−1 = 2 k

45

(15)

(23)

where uncertainty for 1 ppm was calculated using the uncertainty values from the certificate results of CRM and 1 pipette and 100 mL capacity volumetric flask (Table 3) as in the case of preparing the standard solution.

46

G. Aslan-Sungur et al. / J. Chromatogr. A 1325 (2014) 40–48

Table 4 Mean recovery results with three replicates (n), standard deviation (STD) and relative standard deviation (RSD) for POPs. Compound

n

Mean recovery (%)

STD

RSD

PCBs PCB-18 PCB-20 PCB-28 PCB-31 PCB-44 PCB-52 PCB-101 PCB-105 PCB-118 PCB-138 PCB-149 PCB-153 PCB-170 PCB-180 PCB-194

3 3 3 3 3 3 3 3 3 3 3 3 3 3 3

74 77 75 70 90 83 105 105 93 103 78 101 102 98 106

4 6.6 5.4 5.4 4 4.9 3.2 3.7 3 2.8 4.6 3.5 3.6 4.7 3.7

0.054 0.0857 0.072 0.077 0.044 0.059 0.031 0.035 0.032 0.027 0.059 0.035 0.035 0.048 0.035

OCPs Beta BHC Heptachloro epoxide Endosulfan I 4,4 -DDE Endrin Endosulfan II 4,4 -DDD Endrin aldehyde Endosulfan sulfate Endrin ketone Methoxychlor Alpha HCH Gamma HCH Delta HCH Dieldrin 4,4 -DDT

3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3

82 74 82 101 91 95 79 90 101 98 92 49 73 80 96 94

8.0 11.1 8.0 10.8 2.6 8.8 15.7 4.2 13.1 6.2 13.2 6.4 13.1 3.1 12.8 5.8

0.098 0.150 0.098 0.107 0.029 0.093 0.197 0.047 0.130 0.063 0.144 0.130 0.180 0.039 0.134 0.062

PAHs Acy Ace Flu Phe Ant Flt Pyr BaA Chr BbF BaP BkF Ind DahA BghiP

3 3 3 3 3 3 3 3 3 3 3 3 3 3 3

66 56 81 86 90 107 99 108 112 118 115 102 109 114 113

14 9.1 8.8 6.4 5.4 3.7 5.8 4.5 7.9 8.5 4.7 4 5.2 7.4 6.3

0.098 0.150 0.098 0.107 0.029 0.093 0.197 0.047 0.130 0.063 0.144 0.130 0.180 0.039 0.134

Table 5 Calculation of relative uncertainty (%) for PCB-18 in air samples. PCB-18

Value (X)

u(X)

u(x)/X

Std solution (␮g L−1 ) Internal standard (␮g L−1 ) Volume of air (m3 ) Volume of sample (mL) Calibration uncertainty (␮g L−1 ) Recovery Repeatability Relative combined uncertainty Measurement result (␮g L−1 ) Standard combined uncertainty Expanded uncertainty (k = 2) Relative uncertainty (%)

1000 1000 324 1 25 0.74 1

4.15 4.15 32.4 3.4 × 10−3 1.60 0.134 0.031

0.00415 0.00415 0.1 0.0034 0.064 0.181 0.031 0.2264

0.01 2.26 × 10−3 4.53 × 10−3 45

achieve this purpose, significance test should be applied on R; if R is significantly different from 1, then the model equation should be corrected by adding R into the model equation. The significance test statistic t was calculated using Eq. (25).



1 − R

t=

(25)

u(R)

This t value was compared with two-tailed critical values (tcritic ) at 95% confidence level (in this case tcritic is 2.92 for 3 replicates). The significance test result for PCB-18 obtained as follows:



1 − 0.74

t=

0.031

= 8.39

(26)

So; t > tcritic which indicates that R value was significantly different from 1 indicates that R value should be included into the model equation. The R values were estimated and tested for the other analytes and the following analytes do not require any correction with R in their model equations (i.e. t < tcritic ): PCB-194, PCB-180, PCB170, PCB-153, PCB-138, PCB-105, PCB-101, 4,4 -DDE, Endosulfan II, 4,4 -DDD, Endosulfan sulfate, Methoxychlor, Gamma HCH, Dieldrin, 4,4 -DDT, Ant, Pyr, BaA, Chr, BbF, and BaP. In general case, applying this R factor correction in the model equation is not the practical way thus Yenisoy-Karakas¸ [24] suggested that the bias comes from this uncorrected term should be considered in calculation of u(R) and the following equation was defined:





u(R ) =

 1 − R 2 k

+ (u(R))2

(27)

u(R ) calculation for PCB-18:



Then, substituting Eqs. (21)–(23) into Eq. (24) for PCB-18:

 u(R) = 0.74

 0.031 2 0.74

 +

4.15 × 10 1

2 −3

= 0.031



u(R ) =

  1.60 2 25

+

 4.15 2 1000

2

+ (0.032)2 = 0.134

(28)

(24)

The u(R) and recovery factor (R) values were calculated for each analyte according to formulas (20)–(24). The R values should be

uCombined = 0.01

 1 − 0.74 2

+

 4.15 2 1000

3.1.4. Estimation of the combined uncertainty The combined uncertainty was calculated by using Eq. (2) for each analyte and PCB-18 is given as an example calculation in Eq. (29): +

 0.0034 2 1

uCombined = 2.26 × 10

tested whether they should be taken into consideration during the calculation of the analyte concentration in the sample or not. To

−3

+

␮g L

 32.4 2 324

+

 0.031 2 1

+

 0.134 2 0.74

(29)

−1

Table 5 summarizes the calculation of relative uncertainty for PCB-18. The relative uncertainty results for all PCBs, OCPs and PAHs are introduced in Table 6.

G. Aslan-Sungur et al. / J. Chromatogr. A 1325 (2014) 40–48

47

Table 6 Results of combined uncertainty (UCombined ) and relative uncertainty (%) for all POP analytes. PCBs

uCombined (×10−3 )

Relative uncertainty (%)

OCPs

uCombined (×10−3 )

Relative uncertainty (%)

PAHs

uCombined

Relative uncertainty (%)

PCB-18 PCB-20 PCB-28 PCB-31 PCB-44 PCB-52 PCB-101 PCB-105 PCB-118 PCB-138 PCB-149 PCB-153 PCB-170 PCB-180 PCB-194

2.19 1.08 1.09 1.29 0.67 0.81 0.67 0.59 0.64 0.66 0.97 0.67 0.63 0.69 0.69

45 43 43 52 27 32 27 24 26 26 39 27 25 27 28

Beta BHC Hep. epoxide Endosulfan I 4,4 -DDE Endrin Endosulfan II 4,4 -DDD End. aldehyde End. sulfate Endrin ketone Methoxychlor Alpha HCH Gamma HCH Delta HCH Dieldrin 4,4 -DDT

8.0 10 8.0 6.8 5.9 6.6 11 6.1 7.5 5.8 8.2 15 11 9.2 7.6 9.2

32 41 32 27 24 27 44 24 30 23 33 59 46 37 31 37

Acy Ace Flu Phe Ant Flt Pyr BaA Chr BbF BkF BaP Ind DahA BghiP

0.039 0.045 0.030 0.030 0.014 0.013 0.012 0.025 0.014 0.154 0.024 0.022 0.017 0.043 0.037

74 90 46 46 26 24 23 35 27 64 35 32 28 56 49

3.2. Relative uncertainties of POPs The uncertainty components effects are shown in graphics for PCBs, OCPs and PAHs in Figs. 3–5, respectively. According to Figs. (3–5), u(std), u(instd) and u(VSample ) have minimum effects on the relative combined uncertainty. The uncertainty associated

with air sampling, u(VAir ) is constant (0.1) for all POP analytes. Calibration curve, u(C), values are varied for PCBs, OCPs and PAHs (Figs. 3–5). The largest u(C) values are obtained for PAHs, but the variations were high for all POPs. The calibration curve is the important step to be careful. The uncertainty related to recovery, u(R), is the most important component in the

Fig. 3. The contribution of each effect to the total uncertainty for PCB samples.

Fig. 4. The contribution of each effect to the total uncertainty for OCP samples.

48

G. Aslan-Sungur et al. / J. Chromatogr. A 1325 (2014) 40–48

Fig. 5. The contribution of each effect to the total uncertainty for PAH samples.

standard combined uncertainty because the extraction is a critical step in this method. Repeatability is another source for the combined standard uncertainty, u(Rep). These two components (u(R) and u(Rep)) have dominant effects on combined uncertainty for the most of PCBs and OCPs (Figs. 3 and 4). However, u(C) showed higher values than u(R) and u(Rep) for the most of PAHs (Fig. 5). 4. Conclusions Measurement uncertainties for the persistent organic compounds (POPs) in air samples were estimated. The combined uncertainty calculations were performed by following bottom up approach. Calibration curve, standard solutions preparation, volume of sample and air, recovery and also repeatability were the main sources of uncertainty. The combined uncertainty results showed that the most important effects on combined uncertainty were due to the uncertainties from recovery, repeatability and calibration curve. The uncertainty calculations for POPs in air samples were reported as clear as possible in this paper in order to make an easy and useful guide to be followed. The important steps for this analysis method were highlighted with the help of the study which was another aim of estimation of uncertainty. Acknowledgements This work was supported by The Scientific and Technological Research Council of Turkey (Grant Number 107Y238) and by Abant Izzet Baysal University Research Fund (Grant Number BAP 2007.03.03.267). We would like to thank Hatice KARADENI˙ Z, Muhammed ÖZ and Akif ARI for their helps in the preparation and in the analysis of the samples.

References [1] B. Ulbrich, R. Stahlmann, Achieves of Toxicology 78 (2004) 252. [2] L. Turrio-Baldassarri, S. Alivernini, S. Carasi, M. Casella, S. Fuselli, N. Lacovella, A. Laura Lamiceli, C. La Rocca, C. Scarcella, C. Laura Battistelli, Chemosphere 76 (2009) 278. [3] A. Fromberg, K. Granby, A. Hojgard, S. Fagt, J.C. Larsen, Food Chemistry 125 (2011) 1179. [4] B.A. Cohn, P.M. Cirillo, R.I. Sholtz, A. Ferrara, J.-S. Park, P.J. Schwingl, Reproductive Toxicology 31 (2011) 290. [5] I. C¸ok, B. Mazmanci, M. Mazmanci, C. Turgut, B. Henkelmann, K.W. Schramm, Environment International 40 (2012) 63. [6] J.A. Harding, C.M. Engbers, M.S. Newman, N.I. Goldstein, S. Zalipsky, Biochimica et Biophysica Acta 1327 (1997) 181. [7] R. Dietz, F.F. Riget, C. Sonne, R. Letcher, E.W. Born, D.C.G. Muir, Science of the Total Environment 331 (2004) 107. [8] T. Harner, K. Pozo, T. Gouin, A. Macdonald, H. Hung, J. Cainey, A. Peters, Environmental Pollution 144 (2006) 445. [9] S. Harrad, H. Mao, Atmospheric Environment 38 (2004) 1437. [10] C. Backe, I.T. Cousins, P. Larsson, Environmental Pollution 128 (2004) 59. [11] A. Masih, A. Taneja, Chemosphere 65 (2006) 449. [12] C. Peng, W. Chen, X. Liao, M. Wang, Z. Ouyang, W. Jiao, Y. Bai, Environmental Pollution 159 (2011) 802. [13] W. Wang, M. Huang, Y. Kang, H. Wang, A.O.W. Leung, K.C. Cheung, M.H. Wong, Science of the Total Environment 409 (2011) 4519. [14] R.M. Dickhut, E.A. Canuel, K.E. Gustafson, K. Liu, K.M. Arzayus, S.E. Walker, G. Edgecombe, M.O. Gaylor, E.H. Macdonald, Environmental Science of Technology 34 (2000) 4635. [15] R.K. Larsen, J.E. Baker, Environmental Science of Technology 37 (2003) 1873. [16] K.C. Cheung, H.M. Leung, K.Y. Kong, M.H. Wong, Chemosphere 66 (2007) 460. [17] Y. Wei, I. Han, M. Hu, M. Shao, J. Zhang, X. Tang, Chemosphere 81 (2010) 1280. [18] J. Øvrevik, V.M. Arlt, E. Øya, E. Nagy, D.H. Phillips, M. Lag, J.A. Holme, Toxicology and Applied Pharmacology 242 (2010) 270. [19] S. Yenisoy-Karakas¸, E.O. Gaga, Talanta 115 (2013) 150. [20] Eurachem Guide, in: V. Barwick, E. Prichard (Eds.), Terminology in Analytical Measurement – Introduction to VIM 3, 1st ed., EURACHEM, 2011, ISBN 978-0948926-29-7. [21] V.R. Meyer, J. Chromatogr. A 1158 (2007) 15. [22] P. Konieczka, J. Namiesnik, J Chromatogr. A 1217 (2010) 882. [23] Eurachem/Citac Guide CG 4, in: S.L.R. Ellison, A. Williams (Eds.), Quantifying Uncertainty in Analytical Measurement (QUAM), 3rd ed., EURACHEM/CITAC, 2012. [24] S. Yenisoy–Karakas¸, Food Chemistry 132 (2012) 1555.