70 point Acetic acid curve by GC/FID demonstrates heterogeneity of variance in analytical data

Published: 18-07-2017| Version 2 | DOI: 10.17632/dwf4ddww3w.2
Keith Chamberlain


These data are from a calibration of Acetic acid on a GC/FID instrument with a headspace autosampler with 70 calibrated points (10 points were targeted for each level, but replicates were weighed individually resulting in 70 unique amount points measured in mg). One major outlier was added after the fact to illustrate how outliers can pull the calibration model. These data illustrate problems with analytical data from chromatography experiments that largely go unnoticed because there's not enough data collected to see them, namely non-normal error, heteroscedasticity, and outliers. There are 8 columns in the dataset, called "Level", "Amount", "Response", "ISArea", "DMIAmt", "STDAmt", "ISAmt", "ISAmount". "Level" contains integer values corresponding to the specific vial prepared, numbered 1:70, not in that order. "Amount" contains the amount of Acetic acid, in mg, in each vial run on the instrument, which was injected into the vial from a stock standard solution with a weighed concentration of Acetic acid and DMI. These data are sorted by "Amount". "Response" is the gas chromatograph's (GC's) resulting area counts for the Acetic acid peak on each run. "ISArea" is the GC's resulting area counts for the Ethylbenzene peak on each run. The "Amt" columns are the balance weights for the internal standard solution, the reagent (DMI), and the stock standard solution added to each vial, measured in mg. These are not the Amounts of the individual standards in a vial. "ISAmount" is the Amount of Ethylbenzene in the vial in mg.


Steps to reproduce

17Jul17 - Clarified method description to use one column, as ran (the 624 column in the description in version 1 is unnecessary and doesn't retain Acetic acid. The current description is easier). Note that this document is not an endorsement of the method, as I encountered a severe loss in sensitivity after 6 months. Try ultra-inert column/inlet supplies. A operationally qualified Tekmar 7000 headspace autosampler was connected to a qualified Agilent 5890 series II Chromatograph via a high temperature rated 1mL heated inert sample loop and heated transfer line. 20 mL headspace vials with silicone/PTFE septa were equilibrated in the oven of the 7000 for 5 min at 140 degC, then incubation with shaking was performed for 10 min. Sample loop temperature=230 degC, line temperature = 240 degC. Vial pressurization = 13 psi, Injection time = 1 min. The 5890 had inlet and detector (FID) temperatures of 250 degC. Inlet and detector was connected via a Phenomenex Zebron ZB-WaxPlus 30m x 0.32 mm x 0.25 µm column (P/N 7HM-G013-11). The flow was set at ~5.0 mL/min, tested with a calibrated flow meter. A (4:1) split was used with a Restek 1mm internal diameter borosilicate glass inlet liner (P/N 23333). The oven program was as follows: 35 °C for 10 min, then increasing to 220 °C at 10 °C/min. Hold at 220 °C for 5 min. Standard Prep: The Acetic acid standard solution was prepared by adding 50 µL Acetic acid (Sigma-Aldrich P/N 27225), via a 50 µL gas tight syringe, to a headspace vial containing 19950 µL 1,3-Dimethyl-2-imidazolidinone (DMI; TCI P/N D1477), which was added to a headspace vial using a 20mL class A pipette then extracting 50 µL with a 50 µL gas tight syringe. Weights for the DMI and the addition of Acetic acid were recorded. The Acetic acid standard addition was made through the septum that was crimped on after addition of the DMI. After the day’s preps were completed, the Acetic acid stock standard (AA SSTD) solution’s cap was replaced and the solution was stowed in a -20 °C freezer until next use. Internal Standard Preparation: The internal standard solution was prepared by adding 100 µL of Ethylbenzene (EtBzn; Fluka P/N 03080) to a headspace vial containing 9900 µL of DMI added as above but using a 10 mL class A pipette, and a 100 µL syringe. Weights were recorded. The internal stock standard solution (ISTD) was stowed as the AA SSTD at the end of each day. Levels for the curve were prepped with 4, 10, 20, 50, 100, 150, 250 uL of stock solution and weighed, and enough DMI to get to 1mL, and weighed, in a headspace vial, prepped through the septum with a gas tight syringe. Then 10 uL of internal standard solution was added. The levels were placed on the instrument for analysis within 24 hours (to prevent leaching of standard through the hole in the septa). Piercing septa for prep was found to be superior to the open cap method (more accurate sub 10 uL transfers) as long as the vials were ran in 24 hours.