Picture this: A Michelson interferometer sits gracefully on a state-of-the-art air-bearing vibration isolation table. The interference fringes are so perfectly still they look like they were painted on the screen. Suddenly, the air conditioning compressor kicks in. Not a single microscopic mechanical vibration reaches the table surface, yet before your very eyes, the fringes drift by half a period. The isolation table hasn't failed you. The culprit? The refractive index of the air just changed.
Air Isn't "Empty"
The refractive index of air (nair) is approximately 1.00027 under standard conditions (20°C, 1013 hPa, 50% relative humidity). While this seems remarkably close to the perfect vacuum of space (n=1), in the hyper-precise world of interferometry, that tiny "1.00027" multiplied by the optical path length translates into a highly measurable phase shift.
Let's do the math: if the round-trip physical path length L is 0.5 meters, the single-trip optical path length (nL) becomes roughly 0.500135 meters, making the round-trip approximately 1.00027 meters. Compared to a pure vacuum, this creates an optical path difference of about 270 μm. If we are using a standard laser with a wavelength (λ) of 632.8 nm, that difference equals a staggering 426λ.
Now, imagine the cold breeze from your AC introduces a mere 0.1°C temperature difference to the room. According to the renowned Edlén equation (Metrologia, 1966), the temperature sensitivity of air's refractive index is dn/dT≈-0.92x10-6/℃. This means a 0.1°C shift changes nair by -0.092x10-6. Over a 1-meter round-trip optical path, this results in a path length change of 0.092 μm, or roughly 0.15λ.
However, this calculation assumes a perfectly "uniform temperature change." In reality, the AC creates localized drafts where temperature gradients vary unpredictably across the optical path. Consequently, the actual equivalent phase shift is far more dramatic than this uniform estimation suggests.
Why the Air Conditioner is the Ultimate "Killer"
The turbulence expelled from an AC vent is far from uniform; temperature variations can swing by 0.5 to 1°C across spatial scales of 10 to 50 cm. Suppose one arm of your interferometer sits directly in this draft while the other remains sheltered. The refractive index difference (Δn) between the two arms hits approximately
0.92×10⁻⁶/°C×0.5°C≈0.46×10⁻⁶.
If there is a path length difference (ΔL) of 100 mm between the unequal arms, this creates a phase shift Δφ=(2π/λ)·ΔL·Δn≈0.46 radians. That’s about 26°, or 0.07λ. Keep in mind, this is just for a single pulse of turbulence. Because an AC unit typically pumps out these cold air pulses at a frequency of about 1 Hz, your interference fringes will rhythmically drift back and forth by 0.07λ per pulse, every second.
Acoustic Coupling: The Invisible Saboteur
It’s not just temperature; sound plays a mischievous role too. Infrasound (<20 Hz), which the human ear cannot detect, as well as audible sound (50 Hz - 1 kHz), can dynamically modulate the refractive index of air. A sound pressure of 1 Pa (roughly 94 dB SPL) causes air density fluctuations (Δρ/ρ) of about 7x10-6, leading to a refractive index fluctuation (Δn) of around 2x10-6. Over a 100 mm arm length, this generates an optical path difference of roughly 0.2 μm, or 0.3λ.
In a typical lab setting—with the combined hum of AC fans, computer cooling systems, and people walking around—background noise easily reaches 60-70 dB SPL. This corresponds to a sound pressure of 0.02 to 0.06 Pa, injecting 0.005λ to 0.02λ of phase noise into your system. For large-scale scientific facilities that demand interferometric precision down to 0.001λ, this acoustic interference becomes a massive, unignorable source of noise.
The Thermal "Creep": You Only Thought It Was Stable
When you turn on an interferometer, its internal electronic components heat up. This warms the surrounding air, establishing localized thermal convection currents. Here is the catch: reaching thermal equilibrium takes significantly longer than achieving electronic stability. While a laser might stabilize its frequency in 15 minutes and the circuits might stabilize their current in just 5, the thermal convection within the optical path requires a painstaking 2 to 3 hours to truly settle (Downs & Raine, Precision Engineering, 1979).
If you start recording data during those first two hours, refractive index drifts alone can introduce a massive phase drift of 0.1λ to 0.3λ. Any data collected during this warm-up window is essentially garbage.
Conclusion
Ultimately, the measurement precision of an interferometer is rarely limited by the stability of the laser's wavelength. Instead, the true bottleneck is the spatial and temporal inhomogeneity of the air's refractive index. A 0.1°C air mass from the AC, low-pressure acoustic waves, and post-startup thermal convection—each acts as an invisible hand pushing your interference fringes off their true position.
An air-bearing vibration isolation table flawlessly isolates the system from ground tremors, but the air above it remains a chaotic optical medium filled with refractive index fluctuations. The moment you turn on the AC and a cold draft sweeps across an interferometer arm, your carefully calculated optical path length becomes complete fiction.