Multivariate Optimization and Validation of a Modified QuEChERS Method for Determination of PAHs and PCBs in Grilled Meat by GC-MS

Polycyclic aromatic hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs) are recognized as carcinogens and mutagenic food contaminants that threaten public health. As for food safety aspects, control of these contaminants in processed and fatty food is necessary. In this study, eleven factors were screened by the Plackett–Burman design, and four variables were chosen to optimize with the central composite design (CCD) for the improvement of extraction and cleanup procedures of these food contaminants. The optimized variables include 5 g of sample, 2 mL mixture of 2/2/1 ethyl acetate/acetone/isooctane, 1.6 g of ammonium formate, 0.9 g of sodium chloride, and 0.25 g of sorbent Z-Sep+. A 5 min cleanup vortex time with the spike calibration curve strategy, analyzed by gas chromatography–mass spectrometry (GC-MS), led to the validated limits of quantification (LOQs) for 16 PAHs and 36 PCBs of 0.5–2 and 0.5–1 ng/g, respectively, and recoveries of 72–120%, with an average relative standard deviation (%RSD) of 17, for PAHs, and 80–120%, with an %RSD of 3, for PCBs. The method introduces excellent accuracy, precision, and efficiency, and minimizes matrix effects, and ensures a control procedure, adopted with international standards, for food authorities to determine the contaminants of interest in processed meat, and consequently, prevent food-borne disease to improve public health indices.


Introduction
Persistent organic pollutants (POPs), as compounds of anthropogenic and natural (bacterial and plant) sources, have high lipophilicity, a determinant of bioaccumulation in living organisms' fatty tissues.Two groups of these compounds are polycyclic aromatic hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs), and their persistence in the environment is a critical issue due to their toxic effects [1][2][3][4][5][6][7].Therefore, these compounds are included in the priority pollutant list.
Polycyclic aromatic hydrocarbons (PAHs) are categorized as lipophilic compounds consisting of two or more fused aromatic rings.PAHs containing four rings, such as benzo(a)anthracene and chrysene, are weak carcinogenic compounds.In addition, heavyweight PAHs with more than five rings, such as dibenz(a,h)anthracene, benzo(a)pyrene, indeno(1,2,3-cd)pyrene, benzo(b,k)fluoranthene, and benzo(g,h,i)perylene, might have genotoxic and carcinogenic properties harmful to humans, and therefore, are considered as potent contaminants, and are potentially a serious public health concern.
for multiple contaminants with thermal stability, less polar characteristics, and higher volatility [23].
The sample pretreatment process includes several cleanup steps to eliminate the unwanted co-extraction of the sample, which can be time-consuming and labor-extensive.Several sample pretreatment methods are used to determine PAHs and PCBs in food, such as micro-matrix solid-phase dispersion (micro-MSPD) [1], solid-phase micro-extraction (SPME) [34], gel permeation chromatography (GPC) [31], supercritical fluid extraction (SFE) [35], and QuEChERS (quick, easy, cheap, effective, rugged, and safe).
The QuEChERS method aims to simplify and streamline the extraction and purification procedures and minimize cost, and it includes miniaturization and automation to make the analysis easier, while the conventional extraction methods require multiple steps and a considerable amount of solvent and time [36].
Nowadays, the main advantages of QuEChERS, such as the effective elimination of the matrix effect as well as the capacity to obtain a high recovery of target analytes, are extensively known in diverse fields, such as food, environmental, and biological analysis [37].
Moreover, the great versatility of the procedure has allowed its application in other types of matrices and analytes with excellent results [38].
Various modified QuEChERS methods were applied to determine contaminants, such as PAHs and PCBs, in different matrices.Some of these matrices include animal-originated foods, like fish and seafood [39], ham [24], poultry meat [40], eggs [41], and milk [42], as well as rice [43] and vegetable oil [44].C18 and primary-secondary amine (PSA) are repeatedly applied as common dSPE sorbents to increase the cleanup performance in specimens with ≥2% fat; new dSPE sorbents are also available, such as hyper-cross-linked polystyrene, metal-organic frameworks [45], diatomaceous earth [46], graphene, molecularly imprinted polymers [47], and multiwalled carbon nanotubes [48].These have also been applied to increase the effectiveness of sample preparation for various contaminants and matrices.Among the mentioned methods, Z-Sep+ and efficient matrix removal-lipid (EMR-Lipid) are two new commercial sorbents used for the cleanup of fat samples.Z-Sep+ is made of two sorbents, octadecyl (C18) and silica coated with zirconium dioxide (ratio: 2/5) [49].Using Z-Sep+ is an approach to combine the chemical properties of zirconia with the high surface area of C18.Zirconium is a Lewis acid that interacts strongly with Lewis bases such as phospholipids [50].
The EMR-Lipid structure is a proprietary secret, and its mechanism has hydrophobic interactions and size exclusion.The producer claims that EMR-Lipid can selectively remove lipids from QuEChERS extracts of fatty foods in all cases [2].During the extraction, a high surface with elevated absorption capacity and low specific weight is applied as a dispersant.
In this study, our purpose was to improve validation parameters by selecting extraction sorbents and solvents and optimizing the content of salts, solvents, sorbents, and time for extraction and cleanup in the QuEChERS method.The used meat matrix was sheep heart, one of the most nutritive, popular, and delicious grilled foods in Iran.
In dSPE methods, optimizing the condition to achieve maximum efficiency is essential.Statistical optimization techniques, such as response surface methodology (RSM), can anticipate and indicate optimal conditions of the favorable process using validated models [51,52].Such a method is more efficient and valuable than a univariate optimization method, called one-variable-at-a-time (OVAT), with fewer experimental runs [53].
The major disadvantage of the OVAT approach is that it cannot investigate the interaction among factors.RSM-based methods can be considered the best solution to overcome this problem and obtain optimal experimental conditions.
In the first step, independent variables, including solvent type and its volume, vortex time, and the amounts of sample, salts (ammonium formate and sodium chloride) [54], and sorbents, (PSA, Z-Sep+, and EMR-Lipid), were investigated using the Plackett-Burman two-level fractional factorial design.Then, the central composite design (CCD) was used to optimize the QuEChERS important variables for simultaneous GC-MS determination of 16 PAHs and 36 PCBs in grilled sheep heart meat.

Chemicals-Reagents
PAHs, PCBs, and their isotope-labeled internal standard were prepared by Dr. Ehrenstorfer GmbH (Augsburg, Germany).The purity of standards ranged from 95 to 99.8%.Solvents (acetonitrile, acetone, and isooctane) of HPLC grade, sodium chloride (NaCl), and anhydrous magnesium sulfate (MgSO 4 ) of analytical grade were obtained from Merck (Darmstadt, Germany).Also, primary secondary amine (PSA) was obtained from Agilent (Santa Clara, CA, USA), ammonium formate (NH 4 O 2 CH) was supplied by Fluka (Dresden, Germany), Z-Sep+ was supplied by Supelco (Bellefonte, PA, USA), and EMR-Lipid was purchased from Agilent (USA).The intermediate, stock, and working standard mixed solutions of PAHs and PCBs at 100 ng/mL, and internal standard solution at 10 µg/mL, were obtained in acetonitrile and isooctane, respectively, and kept at −18 • C with protection from light.

Apparatus and Instruments
The GC-MS chromatographic analysis of PAHs and PCBs was conducted using an Agilent network GC system 6890N, 5975B Inert MSD (Wilmington, NC, USA) equipped with the 7683B autosampler system Agilent Technologies, (Wilmington, NC, USA).An HP-5MS column (30 m × 0.25 mm id.× 0.25 µm) was used for analysis.Helium was applied as the carrier at 1.0 mL/min.For analyzing PAHs and PCBs in GC-MS, the following GC conditions were used.The oven was set to 75 • C for 3 min, and then the temperature was elevated to 120 • C at 25 • C/min.After that, the oven was heated to 280 • C at 5 • C/min, and remained at this temperature for 21 min.The process was performed at 3 µL injection volume in splitless mode.The electron ionization energy, ion source temperature, and transfer line temperature were 70 eV, 230 • C, and 280 • C, respectively.The total analysis duration was 61.80 min.The screening assessment was conducted based on the selected ion monitoring mode, and a minimum of three characteristic ions were selected for each compound.A centrifuge Eppendorf (Hamburg, Germany) was used at 5816 rcf, and the mixing steps were carried out using a Multi Reax shaker Heidolph, (Schwabach, Germany).Ultra-pure water was obtained from a Millipore (Direct-Q3UV) water purification system (Darmstadt, Germany).

Modified QuEChERS Sample Preparation
The grilled samples were homogenized before the extraction and cleanup process.The extraction steps were performed as follows.Five grams of the specimen was weighed in a 50 mL Falcon tube, and the blank sample was spiked with a combination of PAH and PCB standards at 100 µg/kg.Then, 20 µL at 10 µg/g of isotopically labeled mix 111, 118, and 188 for PCBs and perylene, chrysene, and pyrene for PAHs was added as the internal standard.Next, 2 mL of mixed solvents, including ethyl acetate + acetone + isooctane (2:2:1) (v/v/v), and 500 µL of deionized water was transferred to the tube, and the mixture was shaken strongly for 60 s.Then, 0.9 g of NaCl and 1.6 g of ammonium formate were added to the mixture, which was mixed in a Multi Reax vortex for 15 min, and centrifugation was conducted at 5816 rcf for 10 min at room temperature.Afterward, the supernatant without a fat layer was added to a 15 mL Falcon tube containing 0.250 g of Z-Sep+ and 0.1 g of PSA, followed by shaking for 5 min and centrifuging at 5816 rcf for 10 min at room temperature.After cleanup, the whole upper phase was added to a vial for GC injection.In comparison with the main QuEChERS method, using the solvent mixture, adding ammonium formate, and applying Z-Sep+ are considered modifications of the pretreatment method.

Optimization of the Modified QuEChERS (Sample Preparation) Technique Using Experimental Design
At first, by performing the Plackett-Burman design (PBD), factors with highly significant effects were identified.Table 1 indicates the primary factors and their levels and notations.The sum of the 52 peak areas in various positions as the most intense peaks relative to the maximum peak in TIC was selected as a response.Table 2 shows the PBD parameter from 11 variables with 24 experimental runs using Design-Expert software (Design-Expert, Stat-Ease, Minneapolis, MN, USA).
Sample weight (A), type of solvent (B), solvent volume (C), water volume (D), NH 4 O 2 CH.weight (E), PSA weight (F), type of sorbent (G), amount of sorbent (H), extraction vortex time (J), cleanup vortex time (K), and NaCl weight (L) were considered essential variables.After the screening design, the changed QuEChERS method was optimized by considering the four most significant variables.Therefore, CCD was conducted with four quantitative variables established as 0.5-2.5 g of NH 4 O 2 CH (C), 0.1-0.3g of sorbent weight (D), 0.1-1.5 g of NaCl weight (E), and 60-600 s for the cleanup vortex time (H).The design matrix of CCD optimization is shown in Table 3.A total of 36 experimental procedures were performed randomly using a quintet at the central point.

Method Validation
The proposed strategy (modified QuEChERS in combination with GC-MS) was validated based on the validation criteria for quantitative approaches for pesticide assessment in food and feed in the SANCO document 2021/11312 [55].The linearity, limit of quantification, accuracy, and precision factors were evaluated based on this instruction.

Optimization Process Screening Design, Plackett-Burman Design, and Experimental Design Approach
For achieving efficient specimen pretreatment, and thus increasing recoveries for all analyses, primary investigations were determinative in the choice of extraction factors.We used PBD to determine the significance of 11 variables influencing the changed QuEChERS approach for the 16 PAHs and 36 PCBs. Figure 1 shows the GC chromatogram of the targeted PAHs and PCBs.Experimental results acquired by PBD are illustrated in Table 2 (Response).

Method Validation
The proposed strategy (modified QuEChERS in combination with GC-MS) was validated based on the validation criteria for quantitative approaches for pesticide assessment in food and feed in the SANCO document 2021/11312 [55].The linearity, limit of quantification, accuracy, and precision factors were evaluated based on this instruction.

Optimization Process Screening Design, Plackett-Burman Design, and Experimental Design Approach
For achieving efficient specimen pretreatment, and thus increasing recoveries for all analyses, primary investigations were determinative in the choice of extraction factors.We used PBD to determine the significance of 11 variables influencing the changed QuEChERS approach for the 16 PAHs and 36 PCBs. Figure 1 shows the GC chromatogram of the targeted PAHs and PCBs.Experimental results acquired by PBD are illustrated in Table 2 (Response).According to the ANOVA of the evaluated model, the significant variables were distinguished, which are presented in Table 4.According to the ANOVA of the evaluated model, the significant variables were distinguished, which are presented in Table 4. Table 5 shows fit statistics parameters for PBD with excellent values.Also, half-normal probability plots were applied to find the most significant interactions and impacts.In half-normal probability plots, the negligible impacts are in a straight line, while essential impacts are off the line.The half-normal probability plots for PBD are displayed in Figure 2. Another way to recognize significant main and interaction effects is using the Pareto plot.
It specifies an absolute value for different effects and determines reference lines at a defined confidence limit (based on the t-value and Bonferroni limit).Any effect that passes these defined lines is considered a significant effect (Figure 3).The t-value, sample weight, sorbent weight, NaCl weight, cleanup vortex time, type of solvent, water volume, and ammonium formate weight were regarded as essential factors.
As the type of organic solvent significantly affects the extraction process, this factor was also studied.It is concluded that the highest response for the analyses was acquired with a mixture of acetone + ethyl acetate + isooctane (2:2:1) (v/v).The log Kow for PAHs applied in the current research ranged from 3.3 for naphthalene to 6.63 for benzo[ghi]perylene.Acetonitrile presented log Kow = −0.34,whereas amounts for acetone, ethyl acetate, and isooctane were, respectively, −0.24, 0.73, and 4.1.Our finding suggests that the threecomponent extraction solvent was applied in the extraction method.In half-normal probability plots, the negligible impacts are in a straight line, while essential impacts are off the line.The half-normal probability plots for PBD are displayed in Figure 2. Another way to recognize significant main and interaction effects is using the Pareto plot.It specifies an absolute value for different effects and determines reference lines at a defined confidence limit (based on the t-value and Bonferroni limit).Any effect that passes these defined lines is considered a significant effect (Figure 3).The t-value, sample weight,   As the type of organic solvent significantly affects the extraction process, this factor was also studied.It is concluded that the highest response for the analyses was acquired with a mixture of acetone + ethyl acetate + isooctane (2:2:1) (v/v) .The log Kow for PAHs applied in the current research ranged from 3.3 for naphthalene to 6.63 for benzo[ghi]perylene.Acetonitrile presented log Kow = −0.34,whereas amounts for acetone, Extraction with ethyl acetate isooctane and acetone at the molecular level can result in amended recoveries by letting water-solvable ethyl acetate and acetone recover PAHs trapped in water and making these compounds present for transfer to isooctane.High extraction temperatures can disrupt analyte matrix interactions by decreasing the activation energy needed for analyte desorption processes and decreasing solvent viscosity, simplifying more favorable solvent matrix penetration [56].
It should be pointed out that Z-Sep+ and EMR-Lipid functioned similarly in obtaining chromatograms with no interfering peaks.However, we chose Z-Sep+ because it has less volume than EMR-Lipid in similar weight.
Since the maximum sample weight and the mixture of solvents maximize the defined responses and have the same trend in all responses, and also due to avoiding time consumption and the use of dangerous chemical materials, these factors were not considered for the optimization phase.The reason is that increasing each variable and designing a model for optimization forms extra experimental runs.It means that 5 g of the sample (considering the matrix effect) and a mix of ethyl acetate, acetone, and isooctane have been selected as the optimum conditions.Also, the water volume was excluded as it had a lower impact than other factors, and its t-value was close to the reference value.Finally, NaCl (A), sorbent weight (B), ammonium formate weight (C), and cleanup vortex time (D) were considered in the optimization step (Table 6).

Central Composite Design
After determining the critical factors, i.e., (NaCl (A), sorbent weight (B), ammonium formate (C), and cleanup vortex time (D), they were optimized by CCD at 5-level three blocks with a total of 36 experiments.Table 3 shows the response of the CCD experiment.According to different statistical parameters of the model, such as error and precision, the quadratic model was selected as the best one.Equation (1) shows the coefficients of the factors and their interactions selected by the model.
Table 7 shows that the recommended model is highly important, considering its F-value (F model = 139.79> F Lack of fit = 1.29) and probability value (p = 0.0001 < 0.05).Also, the fit derived from the anticipated model is lower than the calculated F value.
Table 8 shows the statistical parameters for CCD, indicating good model fitting in terms of calibration and prediction.
An excellent value of adjusted R 2 (0.98) indicates that the model is not overfitted.In addition, a good value of predicted R 2 (0.95) shows that the model has a high capability to predict unknown points.Also, the plot of actual versus predicted R 2 (Figure 4) shows the quality of the developed model.The points should be close to the fitted line for a proper fit.
As it is clear from the figure, all points are appropriately fitted in the model line, which confirms the quality of the developed model.Therefore, the model is statistically valid and significant.The coefficients related to independent variables based on the evaluated variables are presented in Equation ( 1).According to the equation, NaCl weight (A), sorbent weight (B), cleanup vortex time (D), and interactions between A and C, B and C, B and D, and C and D affect the extraction effectiveness significantly.The coefficients' negative values belonging to D, AB, AD, A 2 , B 2 , C 2 , and D 2 indicate their negative impact on the recovery of contaminants from the specimens.The contaminant recovery is reduced with an increase in their values.Also, the terms A, B, BC, BD, CD, and AC positively affect the recovery value.
The model accuracy was controlled by comparing the anticipated and detected experimental findings, indicating the linear association between the experimental recovery and the anticipated values.

Adeq Precision 78.0531
An excellent value of adjusted R 2 (0.98) indicates that the model is not overfitted.In addition, a good value of predicted R 2 (0.95) shows that the model has a high capability to predict unknown points.Also, the plot of actual versus predicted R 2 (Figure 4) shows the quality of the developed model.The points should be close to the fitted line for a proper fit.The model accuracy was controlled by comparing the anticipated and detected experimental findings, indicating the linear association between the experimental recovery and the anticipated values.The graphs show the impacts of two variables in their evaluated ranges, and another variable was fixed to the optimum level.The response surface can better display the trend of the variables and their impact on extraction efficiency.Therefore, the achieved recoveries in the extraction process are influenced by target parameters, and the graphs' nonlinearity confirms their interactions.The optimization function was used to achieve an optimal situation for the modified QuEChERS process.The desirability function is an excellent tool for simultaneously optimizing multiple responses.This function works according to the logic that a response is acceptable when its attributes are in the desired range.This process combines all responses into a single one, followed by its optimization.The value of the desirability function varies from 0 (undesirable value) to 1 (desirable value).The response is ideal if this value is closer to 1. Figure 6 shows the chosen solution according to the maximum obtained desirability (0.82).

Response Surface Graphs by CCD and Optimal Values
Finally, NaCl weight, sorbent, ammonium formate, and cleanup vortex time were obtained at 0.9 g, 0.25 g, 1.6 g, and 5 min, respectively.Sample amount (A), type of solvent (B), solvent volume (C), water volume (D), type of sorbent (G), and extraction vortex time (J) were chosen, amounting to 5 g, a mix of ethyl acetate + acetone + isooctane (2:2:1), 2 mL, 500 µL, Z-Sep+, and 15 min, respectively.We used the GC-MS analysis method to simultaneously measure the 16 PAHs and 36 PCBs in grilled meat samples.The compounds, quantification ions, and their retention times are shown in Table 9.

Method Validation and Performance
Following the specification of the optimum experimental situation, the suggested method was validated according to the SANCO guidelines regarding range and linearity, LOQ, intermediate precision, and accuracy.

Linearity and Limits of Quantification
The target method linearity was studied using spike calibration curves.Seven blank matrix samples were spiked with mixed standard working solutions at seven levels ranging between 0.5 and 40 µg/kg in triplicate (pretreated), and experimented upon using the modified QuEChERS method explained earlier.The coefficient of determination (R 2 ) for the spike calibration curves was achieved at 0.999.The results display appropriate linearity for the contaminant residues in the cited matrices within the considered range.LOQs for the PAHs and PCBs were achieved at 0.5-2 ng/g and 0.5-1 ng/g, respectively, according to the lowest level of the spiked samples.The recoveries and RSDs were accepted with a range between 70 and 120% and below 20%, respectively, for LOQ.Regarding MRLs, established by the European Union for PAHs and PCBs in grilled meat, all the LOQ values in grilled meat materials were less than the MRL values.Hence, the studied method is appropriate for detecting contaminants in grilled meat (Table 9).

Accuracy and Precision
The repeatability intraday and interday (three various days) precision were evaluated through %RSDs of recovery studies from the mean of five replicates of the spiked blank matrix at 0.5, 1, 2, 4, 7, 15, and 30 µg/kg levels in grilled meat.The recoveries were between 72 and 120% and 80 and 120% for the PAHs and PCBs, respectively.Also, the %RSD values were below 17% for the repeatability of the PAHs and 3% for the PCBs.Therefore, the changed QuEChERS method was sufficiently sensitive, accurate, and precise to determine simultaneously the assessed contaminants in the grilled meat samples.The linear range, LOQ values, coefficient of determination, recovery, and RSD are shown in Table 9.The graphs show the impacts of two variables in their evaluated ranges, and another  Finally, NaCl weight, sorbent, ammonium formate, and cleanup vortex time were obtained at 0.9 g, 0.25 g, 1.6 g, and 5 min, respectively.Sample amount (A), type of solvent (B), solvent volume (C), water volume (D), type of sorbent (G), and extraction vortex time (J) were chosen, amounting to 5 g, a mix of ethyl acetate + acetone + isooctane (2:2:1), 2 mL, 500 μL, Z-Sep+, and 15 min, respectively.We used the GC-MS analysis method to simultaneously measure the 16 PAHs and 36 PCBs in grilled meat samples.The compounds, quantification ions, and their retention times are shown in Table 9.Our proposed methods have some advantages, such as low consumption of the solvent, high recovery, short extraction time, no matrix interference, and good merit figures, compared to the other methods.Also, it should be noted that to eliminate the matrix effect, the calibration curve is drawn based on the spike calibration method, and the range of linearity is from 0.5 to 40 ng/g.Recoveries and RSD data of the analytical method met the criteria as set by European Commission Regulation No. 1881/2006.Also, the LOQ value was investigated at the levels of 0.5-40 ng/g, and after evaluating the accuracy and precision and meeting the criteria, it was determined as the LOQ of the method, and it is not only based on 10 times the S/N ratio.The comparison of this study with other methods was shown in Table 10.

Conclusions
We attempted to further develop a sensitive and effective method for the simultaneous determination and extraction of trace amounts of a mixture of PAHs and PCBs, as common food contaminants, from grilled sheep heart meat.The modified QuEChERS method was successfully used for high-performance extraction of the analytes of interest before determination by GC-MS in grilled meat.The multivariate optimization technique employing

Figure 1 .
Figure 1.GC chromatograms of PAHs (a-p) of interest and PCBs (1-36) of interest that are referred to in Table in Section 3.5.

Figure 2 .
Figure 2. The half-normal plot of factors.Factors that are far from the line are significant in the experiment.

Figure 2 .
Figure 2. The half-normal plot of factors.Factors that are far from the line are significant in the experiment.

Foods 2023 ,
12, x FOR PEER REVIEW 9 of 19 sorbent weight, NaCl weight, cleanup vortex time, type of solvent, water volume, and ammonium formate weight were regarded as essential factors.

Figure 5
Figure 5 Show the desired responses concerning the selected factors and interactions.

Figure 4 .
Figure 4. Predicted versus actual R 2 plot for CCD optimization.As it is clear from the figure, all points are appropriately fitted in the model line, which confirms the quality of the developed model.Therefore, the model is statistically valid and significant.The coefficients related to independent variables based on the evaluated variables are presented in Equation (1).According to the equation, NaCl weight (A), sorbent weight (B), cleanup vortex time (D), and interactions between A and C, B and C, B and D, and C and D affect the extraction effectiveness significantly.The coefficients' negative values belonging to D, AB, AD, A 2 , B 2 , C 2 , and D 2 indicate their negative impact on the recovery of contaminants from the specimens.The contaminant recovery is reduced with an increase in their values.Also, the terms A, B, BC, BD, CD, and AC positively affect the recovery value.The model accuracy was controlled by comparing the anticipated and detected experimental findings, indicating the linear association between the experimental recovery and the anticipated values.

Figure 5
Figure 5 Show the desired responses concerning the selected factors and interactions.

Figure 4 .
Figure 4. Predicted versus actual R 2 plot for CCD optimization.

Figure 6 .
Figure 6.Optimized values of the considered factors in CCD.

Figure 6 .
Figure 6.Optimized values of the considered factors in CCD.
SourceSum of Squares df Mean Square F-Value p-Value Source Figure 1.GC chromatograms of PAHs (a-p) of interest and PCBs (1-36) of interest that are referred to in Table in Section 3.5.

Table 3 .
Experimental design matrix for CCD.

Table 6 .
Factors, notation, type, unit, and their levels in central composite design (CCD).

Table 7 .
Analysis of variance (ANOVA) results for CCD.

Table 8 .
Fit statistics parameters for CCD optimization.

Table 10 .
Comparison of the LOQ, recovery%, %RSD, and extraction method for simultaneous analysis of PAHs and PCBs in meat matrices by GC-MS.