Investigation of the Measurement Uncertainties in the Measurement of BTEX in the Volatile Organic Compound Group

Abstract

In this study, repeatability, intermediate precision, and recovery were considered within the Type A uncertainty budget, while measurement uncertainties due to the sampling system used (instrument), VOC mixture standard, internal standard, micropipette, temperature effect, methanol, and carbon disulfide were considered Type B uncertainties. As a result of the studies on the uncertainty components of the BTEX parameters belonging to the group of volatile organic compounds (VOCs), the highest uncertainty component for benzene was intermediate certainty at 24%. The highest uncertainty component for toluene was sampling at 23%. The highest uncertainty component for ethyl benzene was sampling at 25%. The highest uncertainty component for m,p-xylene and o-xylene was sampling at 25%. As a result, intermediate precision, sampling, and calibration uncertainties were identified as the most significant uncertainty components.

1. Introduction

Volatile organic compounds (VOCs) are aliphatic and aromatic hydrocarbons that are used as raw materials in many industries; evaporate rapidly on direct contact with air at room temperature; cause air pollution in the atmosphere through photochemical reactions, emissions from vehicles, and the evaporation of fuel; and are defined as a precursor greenhouse gas in climate change [1]. VOCs in the indoor environment are commonly caused by painting, wood floors, furniture, appliances, etc. [2]. The main sources of VOCs in ambient air are smokestack emissions from industrial processes, solvent use, petroleum refineries, vehicle exhaust, and fuel storage and filling facilities [3,4,5]. While direct exposure occurs when people breathe in these pollutants, which are mixed into the air due to various activities, indirect exposure occurs when they pass from the air into soil, water, plants, animals, and other living things and enter the food chain [6,7]. Adverse health effects from the accumulation and absorption of chemicals into the body are the most critical consequences of air pollution. Benzene, toluene, ethylbenzene, ethylbenzene, and xylene, known as BTEX, are among the volatile organic compounds (VOCs) that pose the most significant health risk and account for about 60% of non-methane VOCs in urban air [7]. Benzene has been classified as a Group 1 carcinogen by the International Agency for Research on Cancer (IARC) [8].

Volatile organic compounds can be measured using a photoionizing detector (PID), flame ionizing detector (FID), Fourier Transform Spectroscopy (FTIR), proton transfer reaction mass spectrometry (PTR-MS), nondispersive infrared (NDIR), gas chromatography with mass spectrophotometry (GS-MS), sorbent tubes and thermal desorption, chemical ionization mass spectrometry (CIMS), metal oxide semiconductor sensors (MOSs), fiber laser-induced fluorescence (FILIF), infrared spectrophotometry (IR), and global atmosphere watch [9,10,11,12,13,14,15]. In the aromatic hydrocarbon group (which includes benzene, toluene, ethylbenzene, ethylbenzene, and xylene), GS-MS is more sensitive than FID and IR and has lower uncertainty values [9].

These measurements’ uncertainties are applied as per EN ISO/CEI 17025:2017 in all accredited laboratories. Method verification, also called validation, defines the analytical requirement, and confirms that the method under consideration can meet that requirement. It confirms that a method fulfills the requirements for a specific use through testing and obtaining objective evidence. When validating a method, a laboratory must check that the method is working correctly before the experiment. This means that some experiments should be carried out for verification [16].

Sources of uncertainty can fall under four main headings: sampling, calibration, the analytical method used, and environmental factors. In sampling, the sample collection type, method, storage conditions, and contaminations affect uncertainties. Uncertainty, accuracy, and precision are considered when evaluating the calibration curve. Calibration certificates of the standards used in calibration are used in uncertainty calculations. The uncertainty of measurement by gas chromatography is used as a reference method. Instrument sensitivity, the column, detector, eluent, and interferences gain importance here. Regarding environmental factors, ambient temperature, pressure, and humidity change the measurement sensitivity of these devices. In addition, measurement errors caused by the analyzer are considered environmental factors [17]. Standard deviations of experimental errors caused by a person are included in uncertainties. In addition, particular factors (recovery-induced and extraction-induced) are calculated as environmental factors in uncertainties. Ultimately, the uncertainties from all sources are calculated as a composite uncertainty. The composite standard uncertainty is calculated by taking the square root of the sum of the squares of all individual uncertainties. The combined standard uncertainty is multiplied by a coverage factor (k) to obtain the expanded uncertainty. The coverage factor is commonly chosen to provide a 95% confidence interval. Considering all uncertainty components, the measurement uncertainty is given as a single value: the expanded uncertainty (BTEX concentration = X ± U µg/m³).

In most studies on VOC measurements, only standard deviation values are given as uncertainty components. Although the number of studies detailing all sources of uncertainty is small, there are studies where uncertainty studies with passive samplers have been conducted. For example, Plaisance et al. determined the uncertainty components in a benzene analysis via the passive sampler method as the measured mass, desorption efficiency, uptake rate, and sampling time, which were 9.6, 2.4, 79.4, and 0.002, respectively [18].

The measurement method with the lowest VOC uncertainty value is GS-MS. Although it is difficult to find studies detailing all components, like our research, in a study by Yang et al. [19], the external standard uncertainty percentages were 1.16, 0.91, 0.99, 1.06, 0.81, and 1.01 for benzene, toluene, ethylbenzene, m-xylene, p-xylene, and o-xylene, respectively, for GC-MS. In the study by Plaisance et al. [18], the external standard uncertainty percentage was 1.20 for benzene under GC-MS. The external standard uncertainty percentages were 1.79, 0.72, 1.56, 1.45, and 0.94 for benzene, toluene, ethylbenzene, m-p, xylene, and o-xylene, respectively, under GC-MS [20]. The repeatable uncertainty percentages for GC-MS were for benzene, toluene, ethylbenzene, m-p, xylene, and o-xylene, 11, 17, 11, 10, and 11 [21].

Using gas chromatography and their weighted percentages, this study aimed to determine the measurement-induced uncertainty components in the outdoor measurement of BTEX, one of the VOC groups. In this study, essential uncertainty components were chosen, and composite uncertainty values were calculated based on the analyses. The Ishikawa diagram was used to calculate the uncertainty budget in the BTEX analysis. This diagram shows all the factors that influence the outcome of the analytical process for uncertainty quantification (Figure 1). Each factor represents the primary factor and its offshoots. The diagram includes the influence of parameters such as the pipettes’ accuracy, the purity of the chemicals, the instrument’s calibration, and the recovery (R). The pressure is neglected below 10% due to very low uncertainties [22]. Therefore, the pressure-induced uncertainty component is not considered in this study. This is then used to construct the uncertainty budget for the proposed method. The Ishikawa diagram has been used to calculate the uncertainty (u) of the result and represents an understanding of how an analytical method works.

2. Material and Methods

The device used for sampling is an Italian origin Eco and Tecora brand and DDS System (Dynamic Dilution Sampler for VOC Emission) model sampling device (Toscana, Italy). This device, which performs sampling by the TSE CEN/TS 13649 method [23], has a measurement range of 0.1–1.5 L/min and a resolution of 0.1 L/min. The activated carbon tubes belong to the British-origin CPL Puragen Company (Wigan, UK). The activated carbon tubes are 0.1 L/min and 1.5 L/min and comply with TS CEN/TS 1349:2015 standard [12]. Desorption for sample preparation was performed according to clause 7.2.1 of TS CEN/TS 13649:2015. Honeywell brand carbon disulfide (CS2) was used as an extraction solvent. Caliskan Ultrasonic Cleaner brand Lab. ult 4032 model was used as an ultrasonic bath. 800D model centrifuge was used (ERTIP, İstanbul, Turkey). J. T. Baker brand methanol solvent, LabKings Brand VOC Mix (2000 ppm), and Internal (2500 ppm) are used in gas chromatography/mass spectroscopy (GC-MS) calibration processes in the laboratory. The above solvents used as samples were spiked onto activated carbon and analyzed. Shimadzu GC-MS–QP2010 of Japanese origin (Kyoto, Japan) was used as gas chromatography in the measurements. In gas chromatography, RESTEK brand Rtx model 60 m, 0.25 mm, 1.40 µm, CatLog: 10969, and serial number: 1498291 analytical column was used.

As a result of the calibration of the Eco and Tecora sampling device, the uncertainties at a 95% confidence interval taken from the calibration certificate are given in

Table 1. Combined uncertainty from the sampling device.

Components of Uncertainty	Reference Standard Flow (q)(L/min)	Value (X)	Distribution Type	Factor	u(x)1
Sampling Device Flowmeter (Main Line)	1.090	0.124	Normal	2	0.062
Sampling Device Flowmeter (Dilution Line)	1.095	0.124	Normal	2	0.062
Sampling Device Temperature Meter (Sample) t = (30.09 °C)	-	0.42	Normal	2	0.210
Sampling Device Temperature Meter (Dilution) t = (30.09 °C)	-	0.42	Normal	2	0.210
Sampling Device Temperature Meter (Probe) t = (30.09 °C)	-	0.42	Normal	2	0.210
From the Sampling Device Combined Uncertainty	Ub(1)				0.0813

. Since the calibration certificate gives the number k as 2 and/or the confidence interval as 95%, the distribution type is considered to be the normal distribution. It is given as X ± C and calculated as u (x) = C/2. Although the flow rate of the sampling device is between 0.1 and 1.5 L/min, the certificate values for uncertainty are 1.090 and 1.095 L/min.

In the uncertainty calculations from the VOC Mix standard, lots of the same brands were used for the calibration curve and experimental studies. According to the findings obtained from the standards used, uncertainty components were calculated separately for each component and given in

Table 2. Combined uncertainty from the VOC mix standard.

Components of Uncertainty	VOC Mix (For 2000 mg.L−1)	Value (X)	Distribution Type	Factor	u(x)1	u(x)b
Benzene (Analysis)	1997	110	Normal	2	55.00	0.028
Benzene (Calibration)	1997	29.23	Normal	2	14.615
Toluene (Analysis)	2000	110	Normal	2	55.00	0.028
Toluene (Calibration)	2000	29.34	Normal	2	14.670
Ethylbenzene (Analysis)	1998	110	Normal	2	55.00	0.028
Ethylbenzene (Calibration)	1998	29.31	Normal	2	14.655
m,p-Xylene (Analysis)	1999	110	Normal	2	55.00	0.027
m,p-Xylene (Calibration)	1999	29.32	Normal	2	14.660
o-Xylene (Analysis)	1998	110	Normal	2	55.00	0.027
o-Xylene (Calibration)	1998	35.41	Normal	2	17.705

. Since the calibration certificate gives the number k as 2 and/or the confidence interval as 95%, the distribution type is considered to be the normal distribution. It is given as X ± C and calculated as u (x) = C/2.

The uncertainty values calculated according to the parameter (Chlorobenzene D5, 1,4-Dichlorobenzene D4, Fluorobenzene) in the certificate of 2500 mg/L 3-component internal standard used in calibration and analysis processes in the laboratory are shown in

Table 3. Compound uncertainty from the internal standard.

Components of Internal Uncertainty	Internal (For 2500 mg/L)	Value (X)	Distribution Type	Factor	u(x)1	u(x)3
Chlorobenzene D5	2507.4	24.9	Normal	2	12.45
1,4-Dichlorobenzene D4	2506.2	26.4	Normal	2	13.20	0.0088
Fluorobenzene	2509.5	24.7	Normal	2	12.35

. Since the number k in the calibration certificate is given as 2 and/or the confidence interval is 95%, the distribution type is considered to be the normal distribution and given as X ± C and calculated as u (x) = C/2.

The study used Ertick brand micropipettes; the reproducibility uncertainty of the pipettes must be less than 0.30% for 50 microns and less than 0.15% for 1000 microns. As a result of the calibration of the micropipettes used in the laboratory, the uncertainty components were calculated according to the values taken from the calibration certificates and the temperature value 20 °C ± 5 for the pretreatment ambient temperature published by the Ministry of Environment, Urbanization, and Climate Change, which should be ±5 °C in the laboratory temperature change in the studies carried out in the laboratory and given in

Table 4. Combined uncertainty from micropipette and laboratory temperature.

Components of Internal Uncertainty	Value (X)	Distribution Type	Factor	u(x)1	u(x)4
Micropipette (1 mL)	0.99	Normal	2	0.495
Temperature Effect Uncertainty $(1\mathrm{m}\mathrm{L}\times 5\times {10}^{-4})$	0.002	Rectangle	1.73	0.001	0.002
Micro pipette (50 µL)	0.19	Normal	2	0.095
Temperature Effect Uncertainty $(0.05\mathrm{m}\mathrm{L}\times 5\times {10}^{-4})$	0.0001	Rectangle	1.73	0.000

. Since the calibration certificate gives the number k as 2 and/or the confidence interval as 95%, the distribution type is considered to be the normal distribution. It is given as X ± C and calculated as u (x) = C/2. If a certificate or other specification gives limits without a confidence level, the distribution type is considered rectangular and calculated as u (x) = C/√3.

For benzene, the recoveries were carried out by spiking activated carbon at 20 ppm. According to Annex B of the TSE CEN/TS 13649 standard, 80% was accepted as the recovery criterion, and the Eurachem/CITAC Guide of CG-4 was taken as the basis for calculations [24]. Recovery studies were carried out in 10 replicates with laboratory personnel.

3. Results and Discussion

The measurement uncertainty components were identified as sampling, VOC mixed standard, internal standard, micropipette use, temperature uncertainty, chemical uncertainty, calibration uncertainty, recovery uncertainty, repeatability uncertainty, and intermediate precision uncertainty and presented in the Ishikawa diagram. The Ishikawa diagram shows all the factors that influence the result of the analytical process (Figure 1).

In general, sources such as sampling heterogeneity, temperature, pressure, effects of gas composition, sample handling, and protection affect the sampling and the uncertainty of the sampler. Table 1 shows the uncertainties of the Eco-Tecora sampler according to the technical information of the main line, dilution line, temperature meter in both lines, and temperature meter in the sample probe.

In the uncertainty calculations of the VOC Mix standard, different batches of the same brand were used for the calibration curve and experimental studies. According to the results obtained from the standards used, the uncertainty components were calculated separately for each component and are presented in Table 2. Since the calibration certificate gives the number k as 2 and/or the confidence interval as 95%, the distribution type is normal and is given as X ± C and calculated as u (x) = C/2.

The uncertainty values calculated according to the parameter (chlorobenzene D5, 1,4-dichlorobenzene D4, fluorobenzene) in the certificate of the 2500 mg/L 3-component internal standard used in the calibration and analysis procedures in the laboratory are shown in Table 3.

As a result of the calibration of the micropipettes used in the laboratory, since the laboratory temperature change in the studies performed in the laboratory should be ±5 °C with the values taken from the calibration certificates, the uncertainty components were calculated using the temperature value of 20 °C ± 5 for the pretreatment ambient temperature and are given in Table 4.

The uncertainties arising from the atomic weights and purity of the methanol and carbon disulfide chemicals used are given in

Table 5. Combined uncertainty from atomic weight and purity of chemicals.

Chemicals	Uncertainty	Purity (%)	Value (X)	Distribution Type	Factor	u(x)1	u(x)5
Methanol (CH3OH)	2507.4	-	0.10	Rectangle	1.73	0.000691
Methanol (CH3OH)	2506.2	99.8		Rectangle		0.060	0.00067
Carbondisulfide (CS2)	2509.5	-	0.05	Rectangle	1.73	0.009817
Carbondisulfide (CS2)		99.9		Rectangle		0.02887

.

The calibration curve for BTEX was extracted at 6 points. Concentrations of 1 ppm, 2 ppm, 5 ppm, 10 ppm, 20 ppm, and 50 ppm were chosen for calibration. The calibration values are given in

Table 6. Calibration values and uncertainty for BTEX (for concentrations 1, 2, 5, 10, 20, and ppm).

BTEX	Sample (n)	Calibration Equation (Y)	Coefficient of Determination (R2)	p Value	Residual (Error) Standard Deviation (So)	Uncertainty Value U(Co)
Benzene	12	$Y=22.702\times X+5.011$	0.99938	<0.05	23.5044	0.795
Toluene	12	$Y=22.889\times X+1.949$	0.99967	<0.05	21.748	0.730
Ethylbenzene	12	$Y=22.327\times X+1.315$	0.99967	<0.05	26.347	0.740
m,p-Xylene	12	$Y=44.905\times X+1.766$	0.99968	<0.05	42.113	0.720
o-Xylene	12	$Y=22.868\times X+0.905$	0.99969	<0.05	23.183	0.778

. Calibration plots with two replicates were prepared for calibration-induced uncertainties, and the appropriate equation system and uncertainty coefficients are shown in Figure 2. The uncertainty associated with the linear calibration curve for BTEX is given as follows:

$$U_{cal}={\displaystyle \frac{S_o}{b}}\times \sqrt{{\displaystyle \frac{1}{p}}+{\displaystyle \frac{1}{n\times }}{\displaystyle \frac{(Y_j-Y_{av})}{b^2\times \sum _{j=1}^n(C_j-C_{av})}}}$$

where S_o is the residual standard deviation of the linear regression model; b is the slope of the curve; p is the number of measurements to determine C_j; n is the number of solutions used to plot the calibration Y_i is the measured area of the aromatic hydrocarbon; C_j is the aromatic hydrocarbon concentration (mg/mL); j is the index of the number of measurements to obtain the calibration curve; av is the average value index.

Uncertainty arising from recovery was realized through the following steps:

$$R={\displaystyle \frac{MeasurementValue}{ReferenceValue\left(20\mathrm{ppm}\right)}}\times 100$$

Combined standard deviation value

$$D=\sqrt{{\displaystyle \frac{{SD}_{x,n-1,A}^2+{SD}_{x,n-1,B}^2}{n_A+n_B-2}}}$$

where

SD_{x,n-1,A, B}: standard deviation values for each analyst (A or B) separately according to n_A,B sample numbers

Uncertainty value from recovery

$$U_R={\displaystyle \frac{SD}{\sqrt{n_A+n_B}}}$$

Average recovery in n sample runs for each analyst (A and B)

$$R_{ort}={\displaystyle \frac{R_1+R_2+\dots R_n}{n}}$$

Overall average recovery value for analysts A and B

$$R_{ort}={\displaystyle \frac{R_{ort,A}+R_{ort,B}}{2}}$$

Uncertainty value for each analyst separately

$${U\left(Rec\right)}_A={\displaystyle \frac{{SD}_{(Rort,A)}}{\sqrt{n}}}\hspace{1em}{U\left(Rec\right)}_B={\displaystyle \frac{{SD}_{\left(Rort,A\right)}}{\sqrt{n}}}$$

Recovery processes for benzene were carried out by spiking activated carbon at 20 ppm. According to the TS CEN/TS 13649 standard [23], 80% was accepted as the recovery criterion, and the Eurachem/CITAC [24] standard was taken as the basis for the calculations. Recovery studies were carried out in 10 replicates with two laboratory personnel. The recovery uncertainty for BTEX is given in

Table 7. Recovery uncertainty for BTEX (for reference value 20 ppm).

BTEX	Std. Dev. (SD) (%)	Rort,A,B	U(Rec) (%)	p Value	Uncertainty Value (UR)
Benzene	12.978 8.497	0.956 0.928	4.104 2.687	<0.05 <0.05	0.0245
Toluene	13.139 22.926	0.986 1.107	4.155 7.250	<0.05 <0.05	0.0418
Ethylbenzene	14.655 3.340	1.094 1.032	4.634 1.056	<0.05 <0.05	0.0238
m,p-Xylene	15.475 3.774	1.073 0.998	4.894 1.193	<0.05 <0.05	0.0252
o-Xylene	17.334 4.205	1.060 0.970	5.481 1.330	<0.05 <0.05	0.0282

.

Reproducibility uncertainties, especially for BTEX, which are low aromatic compounds, lead to a decrease in the stability of the containers in which they are stored and, consequently, to a reduction in the concentration level. This can ultimately result in increased uncertainty [25]. The reproducibility results for BTEX in the laboratory, performed by Person A and Person B at a concentration of 10 ppm using the same instrument and a short time interval, are given in

Table 8. Reproducibility uncertainty for BTEX.

BTEX	Std. Dev. (S)	Relative Standard Deviation (%RSD)	p Value	Uncertainty Value (UR)
Benzene	0.181 0.425	2.077 5.080	>0.05 <0.05	3.8807
Toluene	0.200 0.371	2.347 4.491	<0.05 <0.05	3.5828
Ethylbenzene	0.152 0.327	1.701 3.715	<0.05 <0.05	2.8894
m,p-Xylene	0.162 0.323	1.804 3.651	<0.05 <0.05	2.8797
o-Xylene	0.153 0.303	1.704 3.391	<0.05 <0.05	2.6834

.

As can be seen in Table 8. all BTEX values for both analyses are t_critical> reference value, so there is no difference between the measured values of both analyses and the reference value. The repeatability values are adequate for both analysts. When the pairwise comparisons of the analysts were made by ANOVA, there was no difference between the variances of the analysts since F_critical > F_calculus and p-value > 0.05 for all BTEX values. Pairwise repeatability values are appropriate for all BTEXs.

The intermediate precision results for BTEX analyses in the laboratory for concentrations of 5 ppm, performed successively by persons A and B on different days, are given in

Table 9. Intermediate precision uncertainty for BTEX.

BTEX	Std. Dev. (S)	Relative Standard Deviation (%RSD)	p Value	Uncertainty Value (UR)
Benzene	0.237 0.460	5.586 10.635	<0.05 <0.05	8.4940
Toluene	0.221 0.420	5.218 9.799	<0.05 <0.05	7.8499
Ethylbenzene	0.284 0.396	6.176 8.761	<0.05 <0.05	7.5796
m,p-xylene	0.290 0.403	6.265 8.872	<0.05 <0.05	7.6797
o-Xylene	0.287 0.395	6.207 8.650	<0.05 <0.05	7.5281

.

The combined standard uncertainty was given by the following expression.

$$\begin{gathered} U_c=\sqrt{{\left(U_{samp.dev.}\right)}^2+{\left(U_{Mix.stand..}\right)}^2+{\left(U_{int.std.}\right)}^2+{\left(U_{Mic.pip.\&temp..}\right)}^2+} \\ +{\left(U_{Matm.wei.\&Prufiy..}\right)}^2+{\left(U_{cal\dots }\right)}^2+{\left(U_{Recovery..}\right)}^2 \\ +{\left(U_{MReporducibility..}\right)}^2+{\left(U_{Int.Precision}\right)}^2 \end{gathered}$$

Uncertainty of measurement is usually reported as expanded uncertainty, U_i = k.u (k: coverage factor) and “u” is the combined standard uncertainty (in

Table 10. Extended uncertainty values for BTEX.

BTEX	Relative Compound Uncertainty (Ub)	k Factors (for 95%)	Extended Uncertainty (Ug)
Benzene	0.15231	2	0.3046
Toluene	0.14779	2	0.2955
Ethylbenzene	0.14156	2	0.2831
m,p-xylene	0.14117	2	0.2823
o-Xylene	0.14368	2	0.2873

). Usually, k is taken as 2, which gives a confidence interval of about 95%. Similarly, the expanded uncertainties for BTEX are;

In reporting the uncertainty, the experimental result “Y” is written as follows, together with the expanded uncertainty U calculated using the scope factor k = 2;

$$\mathrm{E}\mathrm{x}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{i}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{R}\mathrm{e}\mathrm{s}\mathrm{u}\mathrm{l}\mathrm{t}\left(\mathrm{Y}\right)\pm \mathrm{E}\mathrm{x}\mathrm{t}\mathrm{e}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{d} \mathrm{U}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{a}\mathrm{i}\mathrm{n}\mathrm{t}\mathrm{y} \left(\mathrm{U}\right)\times \mathrm{Y}$$

As a result of the experiments performed, the reporting of the measurement uncertainty of the BTEX VOC groups in the flue gas is given in

Table 11. Uncertainty reporting for BTEX.

Reporting	Benzene	Toluene	Ethylbenzene	m, p-Xylene	o-Xylene
Relative Compound Uncertainty	0.15231	0.14779	0.14156	0.14117	0.14368
Measurement Result (ppm)	10.000	10.000	10.000	10.000	10.000
Standard Compound Uncertainty	1.5231	1.4779	1.4156	1.4117	1.4387
Extended Uncertainty (k = 2)	3.0462	2.9558	2.8312	2.8234	2.8736
Relative Uncertainty (%)	30.462	29.558	28.312	28.234	28.736
Reporting k = 2 at 95% confidence interval	10 ± 3.046	10 ± 2.955	10 ± 2.831	10 ± 2.823	10 ± 2.873
Uncertainty calculation formula	Y ± 0.304Y	Y ± 0.295Y	Y ± 0.283Y	Y ± 0.282Y	Y ± 0.287Y

.

A summary representation of all uncertainty components for BTEX is given in Figure 3. When all uncertainty components are analyzed, it is seen that especially sampling uncertainty is the highest for all VOC groups, followed by intermediate precision and calibration uncertainties, respectively. Uncertainties of types of repeatability, VOC Mix Standard, and recovery can affect uncertainties moderately. Uncertainties Micropipette and Temperature Effect and Methanol and Carbon Disulfide, below 1% of the total uncertainty, are negligible.

4. Conclusions

Volatile Organic Compounds (VOCs) are extremely important because they are classified as carcinogens or toxins and include liquid particles in the form of particles with very long chain and aromatic rings and their vapors. For this reason, measuring VOCs is an issue that must be approached with great care. It is believed that the measurement uncertainties will add value to VOC measurements by following the path taken by internationally recognized accredited organizations in the precision of VOC measurements. To this end, this work has attempted to identify the steps that cause uncertainty in VOC measurements and which sources of uncertainty have a more significant impact on the measurement results. For this purpose, a laboratory accredited by TÜRKAK was used and all stages of the study were carried out with equipment, laboratory, methods, and personnel according to accreditation standards.

The study’s sampling, pretreatment, and analysis sections were prepared according to the TSE CEN/TS 13649 standard [12]. The uncertainty budget was evaluated using TSE CEN/TS 13649 and ISO GUM methods. Only verification studies were performed to establish the uncertainty budget.

BTEX (benzene, toluene, ethylbenzene, xylene), standard components in VOC measurements, were used in the uncertainty calculations. In the uncertainty calculations, 10 sources of uncertainty were identified, and an uncertainty budget was constructed for these sources. These sources of uncertainty were identified as sampling equipment uncertainty, VOC mixture standard uncertainty, internal standard uncertainty, micropipette and temperature effect uncertainty, methanol and carbon disulfide atomic weight and purity uncertainty, calibration curve uncertainty, recovery uncertainty, reproducibility uncertainty, intermediate precision uncertainty. The uncertainty budget identified sources of uncertainty, starting with sampling, pretreatment, and analysis. In the laboratory studies, the highest uncertainty value was found for the parameter “benzene” and the lowest for the parameter “m, p-xylene” at k = 2 95% confidence interval.

The reasons for the high uncertainty of benzene: As mentioned, it is likely due to the choice of chromatographic system, analysis of the stability of the sample, analytical sensitivity, deviations due to calibration, and lack of adaptation in the calibration function. Benzene has more isomers than xylene and is less reactive than benzene. Xylene is also less volatile than benzene. For these reasons, there are fewer losses in the analysis of xylene, which reduces uncertainty.

When the uncertainty components are analyzed for all VOC groups as summarised in

Table 12. The Summarizing uncertainty percentages for BTEX [27].

Components of Uncertainty Percentages (%)	Benzene	Toluene	Ethylbenzene	m,p-Xylene	o-Xylene
Intermediate Certainty	24	23	23	24	23
Sampling System (Device)	23	23	25	25	25
Calibration Curve	23	21	23	22	24
Repeatability	11	10	9	9	8
VOC Mix Standard	8	8	9	8	8
Recovery	7	11	7	8	8
The Internal Standard	3	3	3	3	3
Micropipette and Temperature Effect	1	1	1	1	1
Methanol and Carbon Disulfide	0	0	0	0	0

, the highest uncertainty components were found to be 24% intermediate precision, 23% sampling system and 23% calibration for benzene, 23% intermediate precision, 23% sampling system and 21% calibration for toluene, 25% sampling system, 23% intermediate precision and 23% calibration for ethylbenzene, 25% sampling system, 24% intermediate precision and 22% calibration for m-p-xylene, 25% sampling system, 24% calibration and 22% intermediate precision for o-xylene. In the laboratory studies for the BTEX parameters of the VOC groups, it was found that the uncertainty components for all the BTEX parameters consist of the sampling system (instrument) uncertainty component, the intermediate precision uncertainty component, and the calibration curve uncertainty component for all the parameters, and since the other sources of uncertainty are small, their contribution to the measurement uncertainty is insignificant. EPA has reported that uncertainty values below 25% can be neglected [26]. Therefore, the uncertainty components to be considered are intermediate precision, sampling, and calibration.

Different isomer structures of gases are thought to be effective in calibration uncertainties.

Internal standards are essential in uncertainty assessments as they contribute to determining an isolated quantity of gas [28].

Pre-preparation of the samples when brought to the laboratory, incomplete extraction procedures, and interferences from sample transportation and storage lead to increased uncertainties.

Therefore, when determining the measurement uncertainties and essential elements in VOC analysis, it is necessary to focus on the sampling system (instrument), calibration curve, and intermediate precision studies.

As a result, because of detailed component analysis in GS-MS measurement methods, which have the lowest uncertainty values in VOC measurements, it was determined that intermediate precision, sampling, and calibration uncertainties are at non-negligible levels for the compound uncertainties determined.