I ran an IR spectrum ($\ce{KBr}$ pellet) and found a sharp absorbance at $2360 \text{cm}^{-1}$, which was unexpected. As far as I can tell, this indicates either a nitrile or $\ce{CO2}$.
I know that there is no nitrile in my product. Would you expect atmospheric $\ce{CO2}$ to show up to a noticeable degree in a typical IR spectrum?
From what I've read (and personal experience with IRs I've run), a relatively intense $\ce{CO2}$ asymmetric bond stretching signal at approximately $\ce{2350cm^{-1}}$ is an extremely common background artifact in IR spectra (along with the broad, weak signals downfield of $\ce{3000cm^{-1}}$ attributable to $\ce{O-H}$ bonds from ambient moisture). Depending on your software, it's often possible to run a "dry" IR reading with no sample present in order to establish a baseline spectrum of the surrounding air, which can then be checked against by your software (so that overlapping signals from the baseline air spectrum are deleted from the spectrum of your sample). Alternatively, you can take measures to dry the optics (I've once used $\ce{N2}$ for that purpose; alternatively, air directed through $\ce{CaCl2}$ drying tubes could possibly be used, but I can't attest to the efficacy of that). That manual drying procedure takes some effort, and might be futile in any case depending on the conditions of the lab. My one experience with that approach was disappointing.
It's rather common in my experience. If you background substraction doesn't work perfectly, you'll get a noticeable $\ce{CO2}$ signal. It can be either negative or positive due to the incomplete substraction of the background signal.
If you measure the spectrum multiple times, this signal should vary in intensity as the $\ce{CO2}$ concentration varies slightly. Though this is probably exaggerated in the situation where I observed it, as we have an open IR spectrometer and often multiple students exhaling $\ce{CO2}$ around it.
