BONUS

Fourier Analysis

Decomposing signals into frequencies — the mathematics of spectroscopy

The Hook

Your FT-IR spectrometer doesn't measure absorbance vs frequency directly.

It measures an interferogram — intensity vs mirror position. A wiggly, complicated signal.

Then it computes the Fourier transform. Out comes the spectrum.

How does mathematics convert position information into frequency information?

The Core Idea: Any periodic function can be written as a sum of sines and cosines. Sines and cosines form a complete basis for periodic functions.

1. Building Waveforms from Harmonics

A square wave is a sum of odd harmonics with decreasing amplitudes.

Fourier Series: Square Wave
1

Square wave: \(f(x) = \frac{4}{\pi}\sum_{n=1,3,5,...} \frac{1}{n}\sin(nx)\)

Notice the Gibbs phenomenon — the overshoot near discontinuities!

2. The Uncertainty Principle

A Gaussian transforms to a Gaussian. Narrow in time → wide in frequency.

Time-Frequency Uncertainty
0.5
\[\Delta t \cdot \Delta \omega \geq \frac{1}{2}\]

This is why ultrashort laser pulses have broad spectra!

3. Spectral Lineshapes

Different decay mechanisms produce different lineshapes.

Lineshape Explorer
0.5
LineshapeOriginTime Domain
LorentzianLifetime/naturalExponential decay
GaussianDoppler/inhomogeneousGaussian distribution
VoigtBoth mechanismsConvolution

4. FT-IR: Interferogram to Spectrum

The Fourier transform converts mirror position data to frequency spectrum.

FT-IR Simulation
1000
1500

FT-IR Advantages:

Fellgett: All frequencies measured simultaneously → better S/N

Jacquinot: No slits needed → higher throughput

Connes: Internal laser calibration → precise frequency scale

5. Sampling and Aliasing

Undersampling causes high frequencies to appear as low frequencies.

Nyquist Theorem Demonstration
5
20

Summary

Transform Pairs

Time DomainFrequency Domain
NarrowWide
WideNarrow
ConvolutionMultiplication
Exp. decayLorentzian
GaussianGaussian
DeltaConstant

Chemistry Applications

TechniqueFT converts
FT-IRInterferogram → Spectrum
FT-NMRFID → Spectrum
UltrafastE(t) → E(ω)
QMψ(x) ↔ φ(p)

The Core Message: Fourier analysis decomposes signals into frequencies. Time and frequency are conjugate domains — narrow in one means wide in the other.

Exercises

  1. A laser pulse has Δt = 50 fs. Estimate its spectral width in nm (around 800 nm).
  2. An excited state has lifetime τ = 10 ns. What is the natural linewidth (FWHM)?
  3. What sampling rate is needed to measure up to 4000 cm⁻¹ in FT-IR?
  4. The convolution of two Gaussians with widths σ₁ and σ₂ gives width = ?