Determination of Volatile Compounds in Wine by Gas Chromatography-Flame Ionization Detection: Comparison Between the U.S. Environmental Protection Agency 3σ Approach and Hubaux-Vos Calculation of Detection Limits Using Ordinary and Bivariate Least Squares

Abstract

A capillary GC-flame ionization detection (FID) method to determine volatile compounds (ethyl acetate, 1,1-diethoxyethane, methyl alcohol, 1-propanol, 2-methyl-1-propanol, 2-methyl-1-butanol, 3-methyl-1-butanol, 1-butanol, and 2-butanol) in wine was investigated in terms of calculation of detection limits and calibration method. The main objectives were: (1) calculation of regression coefficient parameters by ordinary least-squares (OLS) and bivariate least-squares (BLS) regression models, taking into account errors in both axes; (2) estimation of linear dynamic range (LDR) according to International Conference on Harmonization recommendations; (3) performance evaluation of a method by using three different internal standards (ISs) such as acetonitrile, acetone, and 1-pentanol; (4) evaluation of LODs according to the U.S. Environmental Protection Agency (EPA) 3sigma approach and the Hubaux-Vos (H-V) method; (5) application of H-V theory to a gas chromatographic analytical method and to a food matrix; and (6) accuracy assessment of the method relative to methyl alcohol content through a Unione Italiana Vini (UIV) interlaboratory proficiency test. Calibration curves calculated via BLS and OLS show similar slopes, while intercepts are closer to zero in the first case, independent of the chosen IS. The studied ISs show a substantially equivalent behavior, even though the IS closer to the analyte retention time seems to be more appropriate in terms of LDR and LOD. Results indicate an underestimation of LODs using the EPA 3sigma approach instead of the more realistic H-V method, both with OLS and BLS regression models. Methanol contents compared with UIV average values indicate recovery between 90 and 110%.

A Fully Automated Online SPE-LC-MS/MS Method for the Determination of 10 Pharmaceuticals in Wastewater Samples

“…The limits of detection (LOD) and quantification (LOQ) were determined from the standard calibration curves using the Hubaux-Vos’s method [ 58 ]. LODs were all below 10 ng L −1 except for aliskiren which was only slightly higher ( Table 3 ).…”

“…For each analyte, linear calibration curves were constructed by correlating the analyte peak area of the quantifier transition (y-axis) to the analyte standard concentration (x-axis). The Hubaux-Vos method [ 58 ] was used to evaluate linearity, lack-of-fit, and residuals of the calibration curves, as well as to calculate LOD and LOQ values. In contrast to the most commonly used ordinary least squares regression methodology, which considers errors only on the y-axis, the Hubaux-Vos method analyzes errors on both axes.…”

A Fast Liquid Chromatography Tandem Mass Spectrometric Analysis of PETN (Pentaerythritol Tetranitrate), RDX (3,5-Trinitro-1,3,5-triazacyclohexane) and HMX (Octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine) in Soil, Utilizing a Simple Ultrasonic-Assisted Extraction with Minimum Solvent

“…Calibration curves of RDX, HMX and PETN were constructed using a minimum of 6 analysis results for each concentration; the dynamic range was determined by a linear regression analysis, using the least-squares method. 30 …”

Gas Chromatographic Quantitative Analysis of Methanol in Wine: Operative Conditions, Optimization and Calibration Model Choice

“…This is more pronounced when limits of quantification are considered. Comparing the calibration curve made up of fifteen data points [16], while the 3σ LOD is substantially unvaried, if six levels are employed, the H-V detection limits are rather different. The calibration curve built up on the 5 -100 mg/L concentration range shows a LOD approximately 2.5 times higher than that of the 0.3 -510.6 mg/L range [16].…”

“…Resolution between methanol and the identified acetal (1,1-diethoxyethane) peak was not less than 0.9. The asymmetry factor, calculated at 10% of the maximum height for methanol, was d In parenthesis, limits of detection obtained with fifteen data points, reported in [16].…”