Reading the spectrum the practical way
The broad O-H stretch of a carboxylic acid is one of the most distinctive features in any IR you'll see, but it's also one of the most frequently misread. It shows up as a huge, ugly drag from roughly 2500 to 3300 cm¹, and because it overlaps the C-H stretching region, you'll often confuse it with an alcohol O-H if you're not paying attention. An alcohol O-H is sharper and sits higher, usually peaking around 3300. The acid version is broad, shallow, and centered lower, dragging all the way down past 3000 until it merges with the C-H signals. You should also be looking for the C=O stretch, which sits right around 1710 cm¹ for a typical aliphatic acid. That's the other anchor point. Between those two, you have enough to make a solid call. When you're actually interpreting one, the C=O band is where people waste the most time. It's strong and sharp, usually between 1700 and 1725 cm¹ for a standard dimerized acid. Here's something that comes up often enough that I wish more people knew: the position shifts depending on whether the acid is hydrogen-bonded as a dimer or existing as a monomer. A dilute solution in carbon disulfide can push the C=O band up to around 1760 cm¹ because the dimers fall apart. If you're running a neat sample or a KBr pellet, you're seeing the dimer. Know which one you're looking at before you assign a structure. I ran into this exact issue last year on a synthesis project. A colleague reported a clean carbonyl at 1758 cm¹ and concluded the product was a methyl ester. I ran a thin film on the NaCl plates next to it and the C=O dropped to 1712 cm¹ with that massive O-H drag underneath. The compound was a carboxylic acid the whole time, and the first reading had been taken too dilute for accurate dimer assignment. We saved a full HPLC purification by checking the concentration effect first.
The O-H bend is another feature worth noting. It appears around 1400 to 1440 cm¹, and the C-O stretch lands in the 1210 to 1320 cm¹ range. These aren't as diagnostic on their own, but when you're confirming a hit after seeing the O-H and C=O, they add confidence. The combination band around 2500 to 2900 cm¹ is also worth keeping in mind. It's the reason the O-H stretch looks so broad and ragged instead of smooth.
What trips people up
One common mistake is confusing the carboxylic acid O-H with water contamination. Water also gives a broad O-H stretch in the same general region, but it tends to be sharper and more symmetric, and it doesn't come paired with a C=O at 1710. If you see a mysterious broad band and a sharp C=O but no C-O stretch in the fingerprint region, check your solvent or your KBr for moisture before declaring an acid. Another trap is anhydrides. They show two C=O bands, usually around 1820 and 1760 cm¹, and sometimes people assume those two peaks mean two different functional groups. They don't. That doublet is the symmetric and asymmetric stretch of the anhydride group itself. No broad O-H comes with it, which is the main way you tell the difference from an acid. Conjugation also changes the game. If your acid is attached to an aromatic ring or an alkene, the C=O stretch shifts down to around 1685 to 1700 cm¹. The O-H band stays where it is. So if you see a carbonyl lower than 1710 without any other obvious structural changes, conjugation is the first thing to consider before reaching for a different functional group entirely.
Get the Full Details
When IR isn't enough
Let me be clear about the limitations here. IR spectroscopy for carboxylic acids is reliable when the sample is pure and the instrument is calibrated. It breaks down in a few scenarios. If you're working with a polymer that has multiple overlapping carbonyls, assigning a single C=O band to a carboxylic acid end group is unreliable. The signal is too weak and too buried. Quantitative work is another weak point. You can detect an acid group, but getting accurate concentration data from an IR spectrum is frustrating. The Beer-Lambert law doesn't behave cleanly in transmission mode for solid samples, and path length variations in pellets or films introduce enough error that NMR or titration is a better choice if you need numbers. Phenols and enols can also produce broad O-H stretches that overlap the acid region. Without the C=O band nearby, you might mistake a phenol for something else, or worse, miss an acid that has an unusually weak C=O due to strong intramolecular hydrogen bonding. In those cases, running a ¹H NMR is the move. The carboxylic acid proton shows up as a broad singlet way downfield, usually between 10 and 13 ppm, and it's much easier to distinguish from phenolic protons that way.
The bottom line is that IR works well for a quick functional group check. It tells you whether a carboxylic acid is present or absent with reasonable confidence. It's not the tool you reach for when the sample is dirty, the structure is ambiguous, or you need precise quantification. Calibrate your instruments regularly, run standards alongside your unknowns, and don't trust a single band to carry the whole assignment.
