How To Read Nmr

Getting Started With NMR Spectra

You look at a spectrum and see peaks. Your first task is figuring out what each peak represents, where it sits on the axis, and what it tells you about the molecule you are analyzing. This process is not intuitive when you first encounter it, but once you develop the habit of reading left to right across the chemical shift range while tracking splitting patterns, it becomes routine. The horizontal axis is chemical shift measured in ppm, typically running from about 12 ppm on the left down to 0 ppm on the right for proton NMR. The vertical axis is signal intensity. Each distinct proton environment in your molecule produces one or more signals depending on how neighboring protons couple to it. Integration values underneath each signal tell you the relative number of protons contributing to that signal. That is the basic framework.

How To Read Nmr as a Practical Workflow

Start by identifying the solvent peaks and deuterated residual solvent signals so you know what to ignore. Acetone-d6 leaves a singlet around 2.05 ppm. DMSO-d6 has a residual proton signal at 2.50 ppm. CDCl3 shows a singlet at 7.26 ppm. These are dead giveaways and they consume about five seconds of your time if you know where to look, which saves you from misassigning them as part of your sample later. Next, scan the spectrum from high ppm to low ppm and categorize each signal by its chemical shift region. Aromatic and vinyl protons sit between 6 and 8.5 ppm. Aldehydes appear around 9 to 10 ppm. Carboxylic acids show up past 10 ppm. Oxygen-bound protons like those on alcohols or amines vary widely depending on concentration and temperature, which is one of the most frustrating things about interpreting them. Aliphatic protons occupy the 0.8 to 4.5 ppm range, and anything below that is usually methyl groups attached to carbon chains or silicon. Once you have rough categories, examine the splitting pattern of each signal. A singlet means the proton has no neighboring protons within three bonds. A doublet means one neighboring proton. A triplet means two equivalent neighboring protons. A quartet means three. These follow the n-plus-one rule, and it works reliably for first-order systems where the chemical shift difference between coupled protons is much larger than their coupling constant. When that condition breaks down, you get second-order effects and the patterns become distorted, which is where things get messy.

I spent an entire afternoon once trying to make sense of a spectrum for a compound with two adjacent methine protons that had nearly identical chemical shifts. The expected doublet-of-doublets collapsed into something that looked like a broad multiplet with no clean splitting. Standard textbook rules did not help at all. The workaround was to run a COSY experiment, which showed the correlation between the two coupled protons clearly, and then use simulation software to fit the pattern. That approach took about twenty minutes compared to the five hours I had wasted trying to interpret it by eye.

Key Parameters to Extract From a Spectrum

Chemical shift is the position of a signal. It reflects the electronic environment around the proton. Electron-withdrawing groups deshield nearby protons and push their signals downfield to higher ppm values. Shielding groups like alkyl chains do the opposite and push signals upfield. Oxygen and nitrogen are particularly strong deshielding agents, which is why protons on carbons adjacent to those atoms typically appear between 3 and 4.5 ppm. Coupling constants measured in hertz tell you about the spatial relationship between protons. Geminal coupling between two protons on the same carbon is usually around 12 to 15 Hz. Vicinal coupling through three bonds depends heavily on the dihedral angle according to the Karplus relationship. A trans arrangement across a double bond gives a coupling constant of 12 to 18 Hz. Cis coupling is smaller, typically 6 to 12 Hz. This information alone can confirm the geometry of a double bond without needing any additional experiments. Integration tells you the relative proton count. Modern spectrometers handle this automatically, but baseline correction is critical for accurate integration. If your baseline is tilted or curved, the integration values will be wrong and your proton ratios will be misleading. I have seen people miss the fact that their compound contained a water peak because the integration of the broad signal around 1.5 ppm was included in the analysis of the actual sample peaks. Running a quick shimp or lock signal check before acquisition helps prevent this.

Common Pitfalls and What They Mean

Exchangeable protons are unreliable indicators. Hydroxyl and amine protons can shift position dramatically depending on solvent, concentration, temperature, and trace amounts of acid or base. A broad singlet at 2 ppm in one sample might move to 4 ppm in another sample of the exact same compound. Do not try to use these for structural assignment unless you have confirmed they belong to your molecule through D2O shake experiments. Add a few drops of deuterium oxide, shake the tube, and watch the exchangeable peaks disappear. Whatever vanishes was an OH or NH proton. Symmetric molecules produce fewer signals than you might expect. A para-disubstituted benzene ring with identical substituents shows only two aromatic signals instead of four. If your molecular formula suggests more protons than your spectrum shows, symmetry is the first thing to consider. The reverse is also true. If you see more signals than expected, check for conformational isomerism or restricted rotation that splits what should be equivalent protons into distinct environments at the timescale of the NMR experiment. Amine and hydroxyl protons sometimes show coupling, which violates the usual expectation that they appear as broad singlets. This happens when the exchange rate is slow, typically in very dry samples or non-polar solvents. If you observe coupling to an exchangeable proton, do not immediately assume your structure is wrong. Run the sample again with a drop of D2O or increase the temperature to speed up exchange and simplify the spectrum.

When NMR Is Not the Right Tool

NMR has real limitations that you need to accept upfront. It requires relatively pure samples. Impurities at the 5 percent level or higher will produce visible signals that complicate interpretation and can hide weak peaks from your actual compound. If your reaction yielded a crude mixture, you need to purify it first through column chromatography or recrystallization before running NMR, and that step alone can take anywhere from thirty minutes to several hours depending on the compound. Quantitative NMR is possible but requires careful setup. You need to set the relaxation delay to at least five times the longest T1 relaxation time in your sample, and you need to use a sufficiently large number of scans to get good signal-to-noise. For routine quantification, HPLC or GC is faster and more reliable. NMR excels at structural elucidation, not throughput analysis. Protons that are magnetically equivalent do not split each other. This is a subtle point that beginners often miss. Two protons on a CH2 group in a symmetric environment may appear as a singlet even though they are coupled to each other because the coupling is not observable between equivalent nuclei. Do not assume a singlet always means no neighbors. Check the symmetry of the molecule first.

Building Proficiency Through Practice

The fastest way to improve is to work through known structures and compare your assignments against the literature values. The SDBS database and ChemSpider have curated spectral data for thousands of compounds. Pick a molecule you already know, look up its predicted spectrum, run the actual spectrum if you have the compound, and compare. The discrepancies you find will teach you more than any single tutorial ever will. Learning to recognize common fragment patterns accelerates interpretation significantly. A triplet and a quartet separated by about 7 Hz is almost always an ethyl group. A sharp singlet integrating to nine protons around 1.3 ppm is a tert-butyl group. A pair of doublets in the aromatic region with a coupling constant near 8 Hz suggests ortho-coupled protons on a benzene ring. These patterns repeat across thousands of spectra and recognizing them saves minutes per assignment that add up over a long project. Modern processing software like MestReNova or TopSpin handles phase correction, baseline adjustment, and peak picking automatically, but you still need to verify the results. Automatic algorithms sometimes misassign baseline artifacts as real peaks or merge closely spaced signals into a single integration bucket. Always visually inspect the processed spectrum before trusting the numbers. This step usually takes about two minutes and prevents a significant number of errors downstream.