What problem is this page solving?
A high-NA objective, SLM, vacuum viewport, or imperfect relay does not just make light dimmer; it changes the phase across the pupil. Zernike modes are the clean vocabulary for naming those phase errors. Once you know the mode, you can predict the PSF damage, estimate Strehl, and decide whether to shim, realign, pre-compensate on an SLM, or stop worrying.
Start with one Zernike mode. Learn what defocus, astigmatism, coma, and spherical aberration look like.
Combine modes, compute RMS wavefront error, and watch the Strehl ratio and PSF respond.
Use SLM and Fourier-optics panels to see how phase masks become tweezers, OAM beams, or distorted spots.
Use Gerchberg-Saxton for phase retrieval, then use the mode atlas as a reference sheet.
Aberrations broaden the focus, lower collection efficiency, and make single-atom images harder to classify.
A distorted phase front changes waist, depth, spacing, and uniformity across large tweezer arrays.
The same math used to diagnose bad optics also creates LG beams, OAM modes, and programmable spot arrays.
Start with one aberration at a time.
Move through radial order n and azimuthal order m. The goal is pattern recognition: when a real image looks comet-tailed, stretched, or ringy, you should know which knob to suspect.
Mathematical convention used here (OSA/ANSI)
Orthonormal over the unit disk ρ ≤ 1 in polar coordinates (ρ, θ):
Orthonormality: ∫|Z_n^m|² ρ dρ dθ / π = 1. Any wavefront W(ρ,θ) = Σ c_n^m Z_n^m; RMS WFE = √(Σ c²).
Mode selector
Z_n^m (OSA, units λ)
Radial cross-section at θ=0
Now combine modes like a real optical system.
Real objectives rarely have a single pure aberration. Add terms, read RMS wavefront error, and use the PSF/Strehl output as the practical verdict.
Wavefront from Zernike Expansion
Any wavefront over a circular aperture: W(ρ,θ) = Σ c_n^m · Z_n^m(ρ,θ), coefficients in waves (λ). RMS WFE = √(Σ c²). Strehl ratio (Maréchal approximation):
S ≈ exp[−(2π · σ_W)²]Strehl > 0.8 → diffraction-limited (Rayleigh criterion). The PSF is computed via 2D FFT of the pupil function: P(ρ,θ) = A(ρ) · exp[i·2π·W(ρ,θ)].
Zernike expansion coefficients
Wavefront W(ρ,θ) (λ)
PSF = |FT{P}|²
Turn pupil phase into focal-plane light.
This is the bridge from aberration language to AMO hardware: SLM phase masks, vortex beams, diffraction orders, and tweezer-array Fourier optics.
SLM Phase → Far-Field Intensity
A spatial light modulator (SLM) imprints a programmable phase φ(x,y) on an input Gaussian beam. In the focal plane of a lens, the field is the Fourier transform of the modulated input:
E_ff(u,v) = FT{ E_in(x,y) · exp[iφ(x,y)] } → I_ff = |E_ff|²Helical phase φ = l·θ converts a Gaussian to a Laguerre–Gaussian LG₀ˡ donut (for l ≠ 0), carrying orbital angular momentum l·ℏ per photon.
SLM Parameters
Input intensity |E_in|²
SLM phase φ (mod 2π)
Far-field intensity |FT{E}|²
Laguerre–Gaussian LG₀ˡ beams
Carry OAM l·ℏ per photon. At the beam waist (p=0): |LG₀ˡ| ∝ (r√2/w)^|l| · exp(−r²/w²) Intensity ring at r = w√(|l|/2). Larger ring for larger |l|.
1. Input: Gaussian beam
2. Phase: $\varphi = l\cdot\theta$ (helical/vortex phase)
3. Far field $\approx$ LG$_0^l$ (donut for $l\neq0$)
Blazed grating $l\cdot\theta + k\cdot x$: shifts donut off-axis, separates OAM orders spatially
| l | Name | Far-field shape |
|---|---|---|
| 0 | Gaussian (LG₀⁰) | Central Airy peak |
| ±1 | First-order vortex | Single donut ring |
| ±2 | Second-order vortex | Larger donut ring |
| ±3 | Third-order vortex | Even larger ring |
| 0 + defocus | Defocused Gaussian | Broadened central peak |
| 0 + astigmatism | Astigmatic Gaussian | Elongated / elliptical |
| 0 + grating | Blazed diffraction | Off-axis Gaussian spot |
Ask the inverse question: what phase makes this light?
Gerchberg-Saxton is the workhorse idea behind many SLM tweezer-array holograms: alternate between planes, keep the desired amplitudes, update the phase.
Gerchberg-Saxton (GS) Algorithm — iterative phase retrieval
Given a desired far-field intensity |T(u,v)|² (e.g. a reconfigurable tweezer array), GS finds the SLM phase pattern φ(x,y) that produces it. The algorithm alternates between the SLM plane and the target plane, enforcing known amplitude constraints in each:
GS Parameters
Target intensity
SLM phase φ (mod 2π)
Reconstructed far-field
Convergence: normalised RMSE vs iteration
Algorithm details
Step 1 (SLM plane): Combine input Gaussian amplitude A(x,y) with
current phase estimate φₖ(x,y).
Step 2 (FFT → target plane): Propagate the field to the focal plane
using a 2D discrete Fourier transform.
Step 3 (target constraint): Replace the amplitude with √T(u,v) while
keeping the retrieved phase arg(FT{E}).
Step 4 (IFFT → SLM plane): Propagate back; extract new phase φₖ₊₁.
Typically converges in 20–50 iterations. Efficiency (fraction of power in target spots)
reaches 70–90% for typical arrays; adding a blazed grating shifts the 1st order off-axis
to separate it from undiffracted light.
Use this as the cheat sheet.
Once you understand the workflow, the atlas is a fast visual reference for identifying modes by eye and remembering the OSA/ANSI indexing structure.
Zernike Pyramid, all modes at a glance
Every Zernike polynomial Z_n^m arranged in the standard pyramid (OSA/ANSI order). Columns index the azimuthal frequency m, rows the radial order n. Cosine modes (m > 0) on the right, sine modes (m < 0) on the left. Blue = negative wavefront, white = zero, red = positive (RdBu colormap).