Rb BEC Experiment

MRF-dressed double-well potential

RF-dressed adiabatic potentials allows the creation of smooth trap geometry without optical speckles and allows to trap atoms with extremely low heating rate. By applying multiple independently-generated dressing frequencies, we have extended the scope of the RF dressing method, and demonstrated the application of three frequencies to create a precisely configurable multiple-RF (MRF) double well [1]. 

Our multiple-radiofrequency approach allows for precise control over the double well potential, including the depth of individual wells and the height of the barrier, and enables reliable transfer of atoms between the available trapping geometries. Tight confinement along the static magnetic field direction, compared to the other two directions, allows the control of dimensionality with which low-dimensional physics and dimensional crossover can be studied[2]. Utilising difference in gF factors of different atomic species (or internal states), species-selective trapping and manipulation is possible without any change in apparatus (see below).

[1] Ultracold atoms in multiple-radiofrequency dressed adiabatic potentials. T.L. Harte, E. Bentine, K. Luksch, A.J. Barker, D. Trypogeorgos, B. Yuen and C.J. Foot. Phys. Rev. A 97, 013616 (2017)

[2] Coherent splitting of two-dimensional Bose gases in magnetic potentials. A.J. Barker, S. Sunami, D. Garrick, A. Beregi, K. Luksch, E. Bentine and C.J. Foot. arXiv: 2006.06661



Non-equilibrium dynamics in 2D quantum systems

Non-equilibrium physics is an area of great theoretical and experimental interest. Equilibration of classical systems has been described with great success, but there remain open questions about how quantum system reach equilibrium: what role do local dynamics play in thermalisation, and over what timescale does thermalisation occur? Thermalisation close to a critical point is also interesting since the corresponding equilibrium state is described by universal scaling laws, but it is not known to what extent universality remains valid out of equilibrium. Quantum systems confined to two dimensions (2D) are especially interesting: fluctuations play a large role and prevent true long-range order, but a phase transition to a superfluid at non-zero temperature does exist, the Berezinskii-Kosterlitz-Thouless (BKT) transition. Part of our research is to investigate how the non-equilibrium dynamics evolve in the presence of this critical point [1].

In our apparatus we can prepare a 2D superfluid in an out-of-equilibrium state via a sudden quench. This is achieved by rapidly splitting the 2D cloud into a double well potential, producing two initially identical clouds which then evolve independently. By later recombining the clouds in time-of-flight we can observe the phase fluctuations and their evolution over time via matter wave interference (MWI) [2].

[1] Dynamic Kosterlitz-Thourless transition in two-dimensional Bose mixtures of ultracold atoms. L. Mathey, K.J. Günter, J. Dalibard and A. Polkovnikov. Phys. Rev. A 95, 053630 (2017)

[1] DPhil thesis, K. Luksch (2019)



An experimental scheme to observe non-equilibrium dynamics of 2D gases via matter wave interference.



Machine learning optimisation of BEC experiments

To observe the BEC phase transition in dilute gas experiments, extremely low temperatures of tens of nanokelvin are typically required. The techniques used to reach these ultracold temperatures usually include a combination of optical cooling and forced evaporative cooling. Implementing these cooling processes requires the precise sequencing of time-varying magnetic and optical fields using a control computer. The parameter space that describes a typical experimental sequence is large and locating the optimal experimental settings using exhaustive, brute-force searches is unfeasible.

We have applied machine learning methods to the production of quantum gases in our apparatus and shown they can outperform human optimisation at producing large atom numbers at cold temperatures, even when starting from random initial parameters [1]. Not only these techniques help us in the lab, these can also be applied to optimise other aspects of the experimental sequence such as reducing the sequence time, optimising matter-wave interference visibility and preparing cold gases with specified parameter such as atom number.

[1] Applying machine learning optimization methods to the production of a quantum gas, A.J. Barker, H. Style, K. Luksch, S. Sunami, D. Garrick, F. Hill, C.J. Foot and E. Bentine, Mach. Learn.: Sci. Technol. 1 015007 (2020)


Process of optimising production of a quantum gas.



Species-Selective Manipulation of a Multi-Component Quantum Gas

We have realised a scheme in which species-selective control is achieved using magnetic potentials dressed with multiple radiofrequencies (RFs), exploiting the different Lande gF -factors of the constituent atomic species [1,2,3]. The resulting potentials have low heating rates, and are extremely smooth and free from defects. Species with Lande g-factors that differ in sign or magnitude can be manipulated independently. We use a mixture of either two distinct atoms (Rb-87 and Rb-85) or different hyperfine states of the same atom (Rb-87, F = 1 & F = 2).

[1] Species-selective confinement of atoms dressed with multiple radiofrequencies. E. Bentine, T. L. Harte, K. Luksch, A. J. Barker, J. Mur-Petit, B. Yuen, C. J. Foot. J. Phys. B: At. Mol. Opt. Phys. 50 094002 (2017)

[2] Inelastic collisions in radiofrequency-dressed mixtures of ultracold atoms. E. Bentine, A.J. Barker, K. Luksch, S. Sunami, T.L. Harte, B. Yuen, C.J. Foot, D.J. Owens and J.M. Hutson, arxiv: 1912.02737

[3] Realising a species-selective double well with multiple-radiofrequency-dressed potentials. A.J. Barker, S. Sunami, D. Garrick, A. Beregi, K. Luksch, E. Bentine, C.J. Foot. arxiv:2003.05925







Non-destructive measurement and spin squeezing

In MRF-dressed double-well potential, dynamics of magnetic spins have different frequency components depending on its trapped location ("left" or "right" well). Utilising this, Voigt detection method yields double-well population information encoded in frequency domain. Theoretical and experimental works are underway to achieve conditional spin squeezing with such detection method.




Cold Atomic Source

There is a currently a rapid drive towards transferring cutting-edge quantum research into real-life applications, and we are developing practical cold atom sources for these devices. Ultracold atoms have a wide range of applications - including improved atomic clocks, gyroscopes, and sensing of gravitational and magnetic fields - because they are extremely sensitive to external fields and forces.

The majority of ultracold atom experiments use a 2-stage arrangement: atoms are first cooled from room temperature to ~1 millikelvin in a vacuum chamber, then are transported to a neighbouring chamber where they are further cooled and probed. We are engineering the first stage of this process to produce more atoms while reducing its size, using our patent-pending pyramidal arrangement of mirrors (see below). This compact and robust cold atom source can then be integrated into a wide range of practical quantum devices, and the increased atom number will enhance signal-to-noise and repetition rate.

cold atom source compressed


2019 08 29  21 27 59811