An evaluation of resolution, accuracy, and precision in FT-IR spectroscopy

Infrared spectroscopy is a foundational technique for the elucidation of chemical structures. The advancements in interferometric spectroscopy, and specifically the development of Fourier transform infrared (FT-IR) spectroscopy, are responsible for the widespread usage of IR spectrometers ranging from teaching labs to pharmaceutical quality control. FT-IR affords an excellent signal-to-noise ratio that permits sensitive sampling with quantitative accuracy and high wavenumber precision based on well documented advantages (Jacquinot, Fellgett, Connes). However, the effect of resolution and instrument-to-instrument variation on wavenumber accuracy is not well understood, with previous work grossly overestimating error. Here, a recommendation of wavenumber accuracy as a function of spectral resolution, accounting for instrument variation among leading manufacturers, is given based on an experimental study of polystyrene and acetaminophen. For peaks that are well resolved and not saturated, the position can be known within 1.1 cm^-1 at a spectral resolution of 4 cm^-1 or higher, and within 2.2 cm^-1 at 8 cm^-1 resolution. Other sources of variation are also discussed (e.g., poorly resolved peaks, peak saturation, water interference, spectral noise) to give general recommendations on when IR peak positions can be considered significantly different. Such guidelines are critical for interpreting subtle positional variations, as are often present in different crystal forms of pharmaceuticals.