Computed IR and Raman Spectra

Introduction

A calculated vibrational spectrum can help assign an experimental peak to a molecular motion. It begins with a stationary molecular geometry and the curvature of its potential-energy surface. The result is a list of normal-mode wavenumbers and predicted activities, not a direct recording of a laboratory sample. Experimental bands may shift, broaden, split or disappear because real vibrations are anharmonic and because temperature, solvent, crystal packing or conformers change the molecular environment. Comparing calculation and experiment is therefore an evidence problem, not a contest to match every peak digit for digit.

Core explanation

For a nonlinear molecule of N atoms, a stable harmonic calculation gives 3N − 6 vibrational normal modes; a linear molecule has 3N − 5. The Hessian of second energy derivatives, transformed into mass-weighted coordinates, gives mode directions and harmonic angular frequencies. A true local minimum should have no genuine imaginary vibrational frequency, apart from tiny numerical artifacts near zero. A transition-state structure normally has one imaginary mode along its reaction coordinate and should not be used as though it were an equilibrium molecular spectrum. Frequencies are usually reported as wavenumbers in cm⁻¹ after conversion from angular frequency.

IR and Raman intensities follow different derivatives. A fundamental vibrational mode is IR active if the electric dipole moment changes to first order along its normal coordinate; its calculated IR intensity depends on the squared dipole derivative. Raman activity is linked to the change in molecular polarizability along the coordinate. A vibration can be strong in one technique and weak in the other. Symmetry may forbid a transition in an ideal isolated molecule, while environmental perturbations can relax selection rules. Therefore, a mode's computed frequency and its computed intensity must be interpreted separately.

The ordinary Hessian calculation is harmonic: it approximates the potential around the minimum as quadratic. Real bond potentials are anharmonic, so observed fundamentals often lie below the corresponding harmonic values. Approximate electronic methods and finite basis sets add their own force-constant errors. A method-and-basis-specific scale factor is commonly applied to harmonic wavenumbers as an empirical improvement, but a factor fitted for fundamentals is not necessarily the best one for zero-point energy. NIST's computational benchmark database explains the frequency-convention distinction, and primary scale-factor research evaluated separate factors and their errors across many model chemistries. A scale factor improves average agreement; it does not cure every individual mode or a wrong molecular structure.

To construct a plotted spectrum, each computed line is usually broadened with a Gaussian or Lorentzian profile. This synthetic width is chosen for visualization and need not equal a sample's actual lifetime or inhomogeneous broadening. Experimental IR absorption and Raman scattering have different intensity scales and instrumental responses. A calculated Raman activity requires choices about laser frequency, temperature and conversion to a relative observed intensity; simply overlaying raw activity sticks with measured intensities can mislead. A peak assignment is strongest when frequency region, isotope shift, displacement pattern, intensity behavior and related compounds all support it.

Conformers complicate spectra. One optimized conformer may put an O–H group into an intramolecular hydrogen bond while another leaves it exposed. Their O–H stretching frequencies and intensities can differ substantially. At the measurement temperature, the sample may contain both, with populations affected by solvent or matrix. A Boltzmann-weighted composite spectrum may be more appropriate than a single geometry, provided conformer free energies and sampling are credible. Intermolecular hydrogen bonds in liquid or solid samples can broaden and shift bands relative to a gas-phase isolated-molecule prediction.

Normal modes are collective, not always local bonds. In a symmetric molecule, two similar bonds can stretch together or oppositely. Labeling one computed peak “the C=O bond” can be useful shorthand, but visualizing the displacement vectors is necessary to see coupled motion. Isotopic substitution offers an additional diagnostic: changing mass alters frequencies and mode mixing without directly changing the electronic potential in the Born–Oppenheimer approximation. A calculated assignment that predicts the observed isotope response is stronger than one that merely places a line near a measured band.

Step-by-step reasoning

1. Optimize the intended species and verify its charge, spin, conformer and minimum character with a frequency calculation. 2. Inspect mass-weighted normal-mode displacement vectors, not only their numerical wavenumbers. 3. Separate IR intensities from Raman activities and apply the appropriate selection-rule reasoning. 4. If needed, use a scale factor documented for the specific method, basis and target property. 5. Construct a plotted spectrum with stated broadening and compare under matched gas, solvent, phase and temperature conditions. 6. Check alternate conformers, hydrogen bonding, isotope shifts and possible overtone or combination bands before declaring a mismatch.

Visual explanation

Show a three-atom bent molecule with arrows on its atoms for symmetric stretch, antisymmetric stretch and bend. Beside each mode, draw separate IR and Raman intensity bars; their heights need not match. Then show a stick spectrum of harmonic lines, a second set shifted by a stated scale factor and a smooth broadened curve. Mark a broad experimental band separately to remind the reader that peak width and environment are not supplied by the bare harmonic Hessian.

Real-world analogy

A musical instrument has characteristic notes, but the sound heard in a room depends on the instrument's surroundings, how it is played and the microphone. A harmonic normal-mode calculation predicts idealized note positions and two different ways they can couple to light. It does not automatically predict the complete recorded sound of a warm liquid or crystal. The analogy also explains why a line near the right frequency can be only part of a convincing assignment.

Real-world example

An analyst investigates whether a sample contains a free alcohol O–H group or a strongly hydrogen-bonded one. An isolated-molecule calculation predicts a relatively high O–H stretching wavenumber for a free conformer and a lower value for a conformer with an internal hydrogen bond. A liquid-sample IR spectrum shows a broad low-frequency band. The analyst considers intermolecular hydrogen bonding and conformer populations rather than forcing the isolated free-O–H peak onto the measurement. A Raman spectrum and isotopic substitution can add independent clues, but sample state remains central.

Why?

Why can a mode be Raman active but IR weak? IR probes how vibration changes the molecular dipole; Raman scattering probes how it changes the polarizability tensor. A symmetric stretch may redistribute electron-cloud deformability strongly while changing the net dipole very little. The two derivatives are different physical responses. This is why a missing IR peak is not proof that the molecular vibration itself does not exist.

Common misconception

“Each computed frequency corresponds to one isolated bond motion.” Normal modes can involve many atoms. “All harmonic frequencies can use one universal correction factor.” Factors depend on method, basis and whether the target is a fundamental or zero-point energy. “A strong computed Raman activity is a strong IR absorption” confuses two derivatives. “Matching a measured peak within 10 cm⁻¹ confirms the structure” ignores accidental coincidences, phase effects and alternative assignments.

Worked example

Suppose a method predicts a hypothetical X–H stretch at 3,600 cm⁻¹ harmonically. A validated factor of 0.960 for that method and basis gives a scaled estimate of 3,456 cm⁻¹. A measured liquid-phase band centered at 3,420 cm⁻¹ differs by 36 cm⁻¹. That residual could reflect hydrogen bonding, anharmonicity beyond the average correction or a different conformer; it is not by itself proof of a wrong functional. If an isotopically substituted sample shifts the band in the predicted direction and a second distinctive mode also matches, the assignment gains support. The scale factor and numbers here are illustrative; a real study should use a factor documented for its model chemistry.

Quick check

1. What molecular derivative controls the leading computed IR activity of a normal mode? Answer: The derivative of the electric dipole moment with respect to its normal coordinate. 2. Does a stable nonlinear five-atom molecule have nine harmonic vibrational modes? Answer: Yes. 3N − 6 = 15 − 6 = 9.

Exam focus

Count modes correctly and identify a stable minimum from the absence of genuine imaginary frequencies. Distinguish IR dipole derivatives from Raman polarizability derivatives. Use a stated scale factor with the correct method and purpose, and calculate a scaled wavenumber. Discuss how phase, hydrogen bonding and conformer populations alter experiment–calculation comparison. Do not infer structure from one line when mode motion and independent evidence are available.

Advanced insight

Fermi resonance can mix a fundamental with an overtone or combination band of similar energy, shifting intensity and frequency beyond a simple scale correction. Anharmonic force-field methods can model some of these effects but depend on a reliable potential and can be costly. In condensed phases, vibrational couplings among molecules and fluctuating solvent environments produce spectra that an isolated harmonic molecule cannot capture. An uncertainty statement should therefore distinguish electronic-force-constant error, anharmonicity and environmental modeling rather than using a single unexplained tolerance.

Summary

Computed IR and Raman spectra derive from normal modes at a specified molecular structure. IR intensity follows dipole change, while Raman activity follows polarizability change. Harmonic frequencies are useful approximations but differ from measured fundamentals through anharmonicity, electronic-method error and environment. Scaling, line broadening and conformer weighting are modeling choices that should be reported. Reliable assignments combine mode visualization with intensity, sample conditions and independent spectral evidence.

Practice questions

1. How many normal vibrational modes does a stable nonlinear six-atom molecule have? Answer: 3(6) − 6 = 12. 2. A harmonic line at 1,750 cm⁻¹ is multiplied by 0.97. What is the scaled position? Answer: 1,697.5 cm⁻¹, usually reported with precision appropriate to the model. 3. Can a mode with almost no dipole derivative still appear in Raman scattering? Answer: Yes, if its polarizability changes sufficiently along the vibration. 4. What does one genuine imaginary frequency indicate at an optimized structure? Answer: It indicates a saddle point rather than a local minimum, assuming the mode is not a numerical artifact. 5. Why might a liquid-phase O–H band be broader and lower than an isolated-molecule harmonic prediction? Answer: Intermolecular hydrogen bonding, conformer mixtures, environmental fluctuations and anharmonicity can shift and broaden it.