Magneto-optical trapping and sub-Doppler cooling have been essential to most experiments with quantum degenerate gases, optical lattices, atomic fountains and many other applications. A broad set of new applications await ultracold molecules 1 , and the extension of laser cooling to molecules has begun 2-6 . A magneto-optical trap (MOT) has been demonstrated for a single molecular species, SrF 7-9 , but the sub-Doppler temperatures required for many applications have not yet been reached. Here we demonstrate a MOT of a second species, CaF, and we show how to cool these molecules to 50 µK, well below the Doppler limit, using a three-dimensional optical molasses. These ultracold molecules could be loaded into optical tweezers to trap arbitrary arrays 10 for quantum simulation 11 , launched into a molecular fountain 12,13 for testing fundamental physics [14][15][16][17][18] , and used to study collisions and chemistry 19 between atoms and molecules at ultracold temperatures.We first focus on the MOT, which is likely to become a workhorse for cooling molecules just as it is for atoms. Previously, only SrF had been trapped this way. For SrF, two types of MOT have been developed, a d.c. MOT where the lifetime was short and the temperature high 7,8 , and a radiofrequency (rf) MOT where longer lifetimes and lower temperatures were achieved 9,20 . In the rf MOT, optical pumping into dark states is avoided by rapidly reversing the magnetic field and the handedness of the MOT laser. It has been suggested that the detrimental effects of dark states can also be avoided in the d.c. MOT by driving the cooling transition with two oppositely polarized laser components, one red-and one bluedetuned 21 . We use this dual-frequency technique to make a d.c. MOT of CaF and find that it works just as well as the rf MOT. Thus, we demonstrate a MOT of a second molecular species, which is important for applications of ultracold molecules, and also verify the effectiveness of this new scheme. Figure 1 illustrates the experiment, which is described in more detail in Methods. A pulse of CaF molecules produced at time t = 0 is emitted from a cryogenic buffer gas source, then decelerated by frequency-chirped counter-propagating laser light, and finally captured in the MOT between t = 16 and 40 ms. Figure 2a shows the molecules in the MOT, imaged on a charge-coupled device (CCD) camera by collecting their fluorescence. We estimate that there are (1.3 ± 0.3) × 10 4 molecules in this MOT (see Methods), with a peak density of n = (1.6 ± 0.4) × 10 5 cm −3 . These are similar to the best values achieved for SrF 20 . To determine the MOT lifetime, we fit the decay of its fluorescence to a single exponential. Figure 2b shows this lifetime as a function of the scattering rate. The lifetime is typically 100 ms and decreases with higher scattering rate, suggesting loss by optical pumping to a state not addressed by the lasers. We do not see the precipitous drop in lifetime observed at low scattering rate in the d.c. MOT of SrF 9 . To watch the molecules ...

Many modern theories predict that the fundamental constants depend on time, position or the local density of matter. Here we develop a spectroscopic method for pulsed beams of cold molecules, and use it to measure the frequencies of microwave transitions in CH with accuracy down to 3 Hz. By comparing these frequencies with those measured from sources of CH in the Milky Way, we test the hypothesis that fundamental constants may differ between the high- and low-density environments of the Earth and the interstellar medium. For the fine structure constant we find Δα/α=(0.3±1.1) × 10−7, the strongest limit to date on such a variation of α. For the electron-to-proton mass ratio we find Δμ/μ=(−0.7±2.2) × 10−7. We suggest how dedicated astrophysical measurements can improve these constraints further and can also constrain temporal variation of the constants.

We demonstrate coherent microwave control of the rotational, hyperfine, and Zeeman states of ultracold CaF molecules, and the magnetic trapping of these molecules in a single, selectable quantum state. We trap about 5×10^{3} molecules for almost 2 s at a temperature of 70(8)  μK and a density of 1.2×10^{5}  cm^{-3}. We measure the state-specific loss rate due to collisions with background helium.

Using frequency-chirped radiation pressure slowing, we precisely control the velocity of a pulsed CaF molecular beam down to a few m s -1 , compressing its velocity spread by a factor of 10 while retaining high intensity: at a velocity of 15m s -1 the flux, measured 1.3m from the source, is 7×10 5 molecules per cm 2 per shot in a single rovibrational state. The beam is suitable for loading a magneto-optical trap or, when combined with transverse laser cooling, improving the precision of spectroscopic measurements that test fundamental physics. We compare the frequency-chirped slowing method with the more commonly used frequency-broadened slowing method.

We present the properties of a magneto-optical trap (MOT) of CaF molecules. We study the process of loading the MOT from a decelerated buffer-gascooled beam, and how best to slow this molecular beam in order to capture the most molecules. We determine how the number of molecules, the photon scattering rate, the oscillation frequency, damping constant, temperature, cloud size and lifetime depend on the key parameters of the MOT, especially the intensity and detuning of the main cooling laser. We compare our results to analytical and numerical models, to the properties of standard atomic MOTs, and to MOTs of SrF molecules. We load up to 2×10 4 molecules, and measure a maximum scattering rate of 2.5×10 6 s −1 per molecule, a maximum oscillation frequency of 100 Hz, a maximum damping constant of 500 s −1 , and a minimum MOT rms radius of 1.5 mm. A minimum temperature of 730 µK is obtained by ramping down the laser intensity to low values. The lifetime, typically about 100 ms, is consistent with a leak out of the cooling cycle with a branching ratio of about 6 × 10 −6 . The MOT has a capture velocity of about 11 m/s.