An orbiting atom interferometer spent 280 days comparing how two rubidium isotopes fall

Both isotope clouds sit in the same instrument. The interferometer on the station's High Microgravity Level Research Rack holds clouds of billions of 85Rb and 87Rb atoms, probed with counterpropagating Raman lasers reflected off a piezo tilt mirror ^[9]. The team excited and detected the fluorescence of the two isotopes sequentially, at slightly different times, which gave two sets of symmetric interference images ^[10]. Taking the differential phase of those sets, after suppressing the station's residual acceleration and the effects of vibration, leaves the difference in acceleration between the isotopes along the falling direction ^[11].

Statistics accumulated slowly. More than 9,700 fringe pairs across 280 days of operation ^[12] averages about 35 pairs a day ^[16]. The gain from orbit is per shot, not per day: Earth's gravity holds the interference time to a few seconds and microgravity stretches it to minutes ^[5], and the resolution of such a test scales as the inverse square of that time ^[6]. Going from about 2 seconds to about 120 seconds is a factor of 60 in time and 3,600 in resolution ^[19].

The new quantum uncertainty is roughly 280,000 times coarser than the best ground-based macroscopic figure and about 10 million times coarser than the best space-based one ^[17], ^[4]. Any model that predicts a composition-dependent violation large enough for those experiments to see has already been tested against them. What the station measurement adds is a constraint in the quantum regime ^[20], where the macroscopic tests say nothing directly, and a violation there would have to be enormous to show up at 2.8 x 10^-8 ^[1].

The laboratory result is a separate confirmation. A group working in Israel, Germany and the UK measured an object's quantum phase in freefall and found the equivalence principle holding there too ^[2]. Physics World reports the two experiments as independent tests, and the 2.8 x 10^-8 figure belongs to the space-station measurement alone ^[1].

Physics World gives the uncertainty as 2.8 x 10^-8 and the test result as around -2.7 x 10^-7 ^[1], which puts the central value about ten times the quoted uncertainty ^[18]. Read literally, those two numbers describe a nonzero violation at high significance; read as an inconsistency in exponents, they describe a null. The error budget in the Science Advances paper ^[8] is where that gets settled.

Zhan said of the work, "We believe that our technique will have a significant impact" ^[13], and put the hardware first: it demonstrates "that an integrated interferometer meeting the requirements of in-orbit operation is possible" ^[14]. The precision the field is aiming at, 10^-17 or better, needs still longer interference times, and that means a permanent microgravity environment ^[7].