NMR, IR, MS, and UV are your four tools. Here is how you actually use them together instead of treating each one as a separate puzzle.
Spectroscopic Identification Of Organic Compounds is mostly about learning which questions each technique can answer and, more importantly, which questions it cannot. I spent years watching students try to solve a structure from a single spectrum, which is impossible unless you are extremely lucky. The real workflow is iterative and slightly boring. You run the mass spectrum first to get the molecular weight and formula. You check the IR for functional groups. You look at the NMR for carbon-hydrogen framework. You use UV only if you have conjugation or aromatic systems that need confirmation. That order matters because each technique narrows the possibilities for the next one. Mass spectrometry gives you the molecular ion peak and the isotope pattern. If you see a bromine signature with two peaks separated by two m/z units and roughly equal intensity, you have one bromine atom. If you see chlorine, the isotope ratio is about 3:1 for M and M+2. I once had a student spend forty-five minutes trying to figure out why the molecular ion was missing entirely. The compound was a t-butyl ester. Those things fragment clean and fast, so the molecular ion disappears into noise. We switched to a softer ionization method, electrospray with low collision energy, and got the molecular weight immediately. Never assume the molecular ion is always there just because the textbook shows it that way. Fourier transform infrared spectroscopy is fast but easy to misread if you are not careful about sample preparation. A KBr pellet that is too thick will saturate the peaks and make you think you have a different functional group than you actually do. I remember running an IR on a carboxylic acid and the O-H stretch looked broad but weak because the sample was essentially invisible in the pellet. I re-prepared it at a much lower concentration and the broad absorption around 2500 to 3300 cm^-1 appeared clearly. That trough is characteristically messy and overlaps the C-H stretches, which is why beginners often miss it. Carbonyl stretches sit between 1650 and 1780 cm^-1 depending on what is attached to the carbonyl carbon. Amides are lower, around 1630 to 1690. Esters are higher, around 1735 to 1750. Anhydrides show two bands because of symmetric and asymmetric coupling. If you see two sharp peaks in the carbonyl region, think anhydride before you think contamination.
Proton NMR is where most of the structural information lives, but interpreting it requires patience. The chemical shift tells you the electronic environment. The integration tells you the ratio of protons. The multiplicity tells you how many neighboring protons are coupling to each set. I once spent an entire afternoon trying to assign a spectrum for a molecule with a para-substituted benzene ring and an ethoxy group. The aromatic signals looked like a single messy blob until I changed the spectral width and zoomed in. They were actually two clean doublets centered around 7.0 and 7.8 ppm. The ethoxy methyl was a triplet at 1.4 ppm and the methylene was a quartet at 4.0 ppm. The compound was ethyl 4-methoxybenzoate. The key was not a fancy technique. It was simply increasing the digital resolution and looking at the raw data instead of the processed peak list. Automated peak picking will group overlapping signals into one entry and that is where you lose information. Carbon-13 NMR is less sensitive than proton NMR but gives you the number of unique carbon environments. Quaternary carbons show up weakly because they lack attached protons for NOE enhancement during broadband decoupling. If a carbon signal is unexpectedly weak, check whether it is a quaternary carbon or whether the relaxation delay was too short. A relaxation delay of less than five times the T1 of the slowest relaxing carbon will cause quantitative errors. I usually set the delay to at least two seconds and run enough scans to get a good signal-to-noise ratio, which for a routine identification is about ten to twenty minutes on a modern instrument. UV-visible spectroscopy is the weakest of the four for general identification. It is useful mainly for conjugated systems, aromatic compounds, and anything with a chromophore. The lambda max and molar absorptivity can confirm conjugation length but they cannot tell you the full structure. I use it selectively when NMR and IR leave ambiguity about whether a double bond is conjugated or isolated. A conjugated diene will absorb at a longer wavelength than an isolated one, and the difference is usually several tens of nanometers.
Here is a realistic workflow that I actually use. First, run the mass spectrum and determine the molecular formula. Use the isotope pattern to check for halogens. Second, run the IR and identify carbonyls, hydroxyls, amines, and nitriles. Third, run the proton NMR and assign all signals. Check for symmetry by comparing integration values. Fourth, run the carbon NMR to count unique carbons and identify quaternary centers. Fifth, if the structure still has ambiguity, use UV or a 2D NMR experiment like COSY or HSQC. That last step is where most people skip ahead too early and then waste hours going in circles. One thing beginners consistently get wrong is assuming that a clean spectrum means a pure compound. Impurities below one percent are invisible in NMR integration but they will show up as small stray peaks that confuse interpretation. I always check the baseline for tiny bumps and the mass spectrum for low-abundance peaks that do not belong to the main compound. A good practice is to run a thin-layer chromatography plate before you invest time in full structural analysis. If the TLC shows two spots, no amount of spectral interpretation will give you a single coherent answer. Another common failure mode is ignoring solvent peaks. Dimethyl sulfoxide shows up as a septet at 2.50 ppm in proton NMR. Water in DMSO appears as a broad singlet around 3.33 ppm. Acetone residue gives a clean singlet at 2.05 ppm. I have lost count of the spectra I initially misassigned because I did not account for residual solvent. Always label the solvent peaks first before you start assigning your compound.
Get the Full Details
The biggest limitation of this whole approach is that spectroscopic methods alone cannot always distinguish between certain stereoisomers or conformers. NMR coupling constants can give you information about dihedral angles through the Karplus relationship, but that requires careful analysis and sometimes additional experiments like NOE difference spectroscopy. Chiral compounds will not separate on a standard NMR instrument without a chiral shift reagent or a chiral solvent. If you need absolute configuration, you are looking at X-ray crystallography or optical rotation measurements, not regular spectroscopy. No amount of NMR peak fitting will replace a single-crystal X-ray diffraction pattern when you need to resolve stereochemistry definitively. Another hard limit is that spectroscopic identification struggles with polymers, complex mixtures like natural extracts, and compounds that are thermally unstable or prone to degradation during analysis. In those cases, you couple chromatography with spectroscopy. Gas chromatography-mass spectrometry handles volatile mixtures. Liquid chromatography-mass spectrometry handles everything else. The spectra you get from those coupled systems are usually simpler because each peak is a separate compound, but the interpretation still requires the same fundamental skills. Software helps but it is not a substitute for understanding what the data means. Automated structure elucidation packages exist and they have improved significantly over the last decade. They can propose candidate structures from NMR and MS data in minutes. But they will also propose incorrect structures when the input data is ambiguous or contaminated. I use them as a starting point, not a final answer. A human check against the raw spectra always catches something the software misses.
If you want to practice, start with known compounds and work backward. Run the spectra yourself if you have access to an instrument. If you do not, use public spectral libraries like the SDBS database from the National Institute of Advanced Industrial Science and Technology in Japan or the NIST mass spectrometry library. Compare the literature spectrum to your own. The differences you find between published data and what you measure will teach you more than any textbook explanation. Calibration drift, different solvent, different concentration, and instrument resolution all create variations that matter in practice even if they don not matter in theory.
