Mainstream smoke is a complex mixture containing > 4800 compounds. Within this mixture there are various carbonyls which are also present on the FDA’s abbreviated and extended harmful and potentially harmful constituents (HPHC) list, such as formaldehyde, acetaldehyde, acrolein, and crotonaldehyde. Advances in tobacco science have produced new products aimed at lowering the potential exposure to HPHCs. Among these innovations are heated tobacco products (HTPs), also called heat-not-burn, which eliminate combustion to produce an aerosol. HTPs typically reduce combustible HPHC yields by < 90 %, which limits the usefulness of traditional mainstream smoke methods such as CORESTA Recommended Method (CRM) 74. Due to the lower aerosol yields, HTP methods require better sensitivity or improved stability to allow for the collection of more matrix.

Here, we report the development of a new method for the trapping and analysis of twelve carbonyl compounds (formaldehyde, acetoin, acetaldehyde, diacetyl, acetone, acrolein, propionaldehyde, furfural, pentanedione, crotonaldehyde, methyl ethyl ketone, and butyraldehyde) using UPLC-MS. Method validation included an assessment of selectivity, precision, accuracy, stability and trapping efficiency. Presented here is an improved collection process that traps carbonyls on a Cambridge filter pad followed by a single impinger containing acetonitrile/isopropyl alcohol at -35 °C, thereby eliminating in situ trapping in DNPH. By collecting at low temperatures, we are able to extend stability of the compounds within this solution for up to two hours, allowing for the collection of more matrix. Additionally, by not collecting in DNPH, we are able to minimize degradation of acid sensitive hydrazones, which allows for longer collection times and results in improved recoveries for acrolein and crotonaldehyde (107 % - 110 % and 101 % - 102 % respectively). Complete resolution among all analytes is achieved over a 12.5-minute injection utilizing UPLC-MS operating in SIR mode. The calibration curves were quadratic with 1/x weighting and had a range of 0.01 µg/mL to 8 µg/mL for each analyte except for formaldehyde (up to 4 µg/mL) and acetone (0.05 µg/mL to 4 µg/mL) with r2 = 0.995.