Time-Resolved Observation of Competing Attractive and Repulsive Short-Range Correlations in Strongly Interacting Fermi Gases

Featured in Physics
Editors' Suggestion

Time-Resolved Observation of Competing Attractive and Repulsive Short-Range Correlations in Strongly Interacting Fermi Gases

A. Amico, F. Scazza, G. Valtolina, P. E. S. Tavares, W. Ketterle, M. Inguscio, G. Roati, and M. Zaccanti

Phys. Rev. Lett. 121, 253602 – Published 19 December 2018

See Viewpoint: The Quest to Make a Ferromagnet with Cold Atoms

Abstract

We exploit a time-resolved pump-probe spectroscopic technique to study the out-of-equilibrium dynamics of an ultracold two-component Fermi gas, selectively quenched to strong repulsion along the upper branch of a broad Feshbach resonance. For critical interactions, we find the rapid growth of short-range anticorrelations between repulsive fermions to initially overcome concurrent pairing processes. At longer evolution times, these two competing mechanisms appear to macroscopically coexist in a short-range correlated state of fermions and pairs, unforeseen thus far. Our work provides fundamental insights into the fate of a repulsive Fermi gas, and offers new perspectives towards the exploration of complex dynamical regimes of fermionic matter.

Received 26 July 2018

DOI:https://doi.org/10.1103/PhysRevLett.121.253602

Physics Subject Headings (PhySH)

Fermi gases Ferromagnetism Spin dynamics

Nonequilibrium systems Ultracold gases

Pump-probe spectroscopy

Atomic, Molecular & OpticalCondensed Matter, Materials & Applied Physics

Viewpoint

The Quest to Make a Ferromagnet with Cold Atoms

Published 19 December 2018

Experimentalists working with cold atoms are closing in on the famous Stoner model of ferromagnetism.

See more in Physics

Authors & Affiliations

A. Amico^1,2,§, F. Scazza^1,2,§, G. Valtolina^1,2,†, P. E. S. Tavares^1,2,‡, W. Ketterle³, M. Inguscio^1,2, G. Roati^1,2, and M. Zaccanti^1,2,*

¹LENS and Dipartimento di Fisica e Astronomia, Università di Firenze, 50019 Sesto Fiorentino, Italy
²Istituto Nazionale di Ottica del Consiglio Nazionale delle Ricerche (INO-CNR), 50019 Sesto Fiorentino, Italy
³Department of Physics, MIT-Harvard Center for Ultracold Atoms, and Research Laboratory of Electronics, MIT, Cambridge, Massachusetts 02139, USA

^*Corresponding author. zaccanti@lens.unifi.it
^†Present address: JILA, University of Colorado, Boulder, Colorado 80309, USA.
^‡Present address: Instituto de Física de São Carlos, Universidade de São Paulo, Caixa Postal 369, 13560-970 São Carlos, SP, Brazil.
^§A. A. and F. S. contributed equally to this work.

Article Text (Subscription Required)

Click to Expand

Supplemental Material (Subscription Required)

Click to Expand

References (Subscription Required)

Click to Expand

Issue

Vol. 121, Iss. 25 — 21 December 2018

Reuse & Permissions

Access Options

Author publication services for translation and copyediting assistance advertisement

Images

Figure 1
(a) A pump pulse converts a weakly interacting 1–2 ${}^{6}{Li}$ mixture into a 1–3 strongly repulsive one. After a variable evolution time the gas is probed by a second pulse. The spectroscopic signal is acquired within a central region of the cloud, denoted by a white rectangle in the image. (b) Spin-injection spectrum at $κ_{F} a ≃ 2.3$ (red line). Both the atomic repulsive resonance located at $Δ_{+ 0} > 0$ and the pair association spectrum are visible, referenced to the spectrum of a spin-polarized state-2 Fermi gas (black line). The pump pulse remains fully selective up to $κ_{F} a \approx 2.5$ , where the atomic and molecular spectral contributions overlap. (c) Probe spectra at $κ_{F} a ≃ 2.3$ for various evolution times (see legend), fitted to a phenomenological model consisting of the sum of Gaussian and Gumbel functions [25]. The spin-injection and pump-probe spectroscopy pulse sequences are also sketched.
Reuse & Permissions
Figure 2
(a) Interaction shift $h Δ_{+ 0} / ϵ_{F}$ as a function of $κ_{F} a$ , measured in balanced mixtures at $T / T_{F} ≃ 0.12$ (orange circles) and $T / T_{F} ≃ 0.8$ (magenta circles), and in the impurity limit (blue squares) [20]. Vertical error bars combine the standard error of the Gaussian fits to the atomic spectra with the uncertainty in determining $ϵ_{F}$ . Horizontal error bars account for the uncertainty in determining $κ_{F}$ . (b) Examples of atomic probe spectra at $κ_{F} a ≃ 0.9$ for various times after the quench (see legend), together with Gaussian fits. (c) and (d) Evolution of relative fermion density $\bar{n} (t) / {\bar{n}}_{0}$ and shift $Δ_{+} (t) / Δ_{+ 0}$ for different $κ_{F} a$ values (see legend). Colored (white) data points are obtained with a $250 μ s$ -long ( $50 μ s$ -long) probe pulse. Error bars account for the standard error of the Gaussian fit, and in panel (c) also for the uncertainty in the calibration of ${\bar{n}}_{0}$ [25].
Reuse & Permissions
Figure 3
Growth rates of anticorrelations in the upper branch $Γ_{Δ}$ (blue circles) and of pairing in the lower branch $Γ_{pair}$ (red diamonds), extracted through linear fits to the measured short-time dynamics of $Δ_{+} (t) / Δ_{+ 0}$ and of $\bar{n} (t) / {\bar{n}}_{0}$ , respectively. Vertical error bars combine the standard error of the fits with the uncertainty in determining $τ_{F}$ . Horizontal error bars account for the uncertainty in the determination of $κ_{F}$ . Inset: $Γ_{pair}$ is compared with the predicted three-body recombination rate $Γ_{3}$ [25, 41].
Reuse & Permissions
Figure 4
(a) Relative atomic populations ${\bar{n}}_{L T} / {\bar{n}}_{0}$ (red triangles) and interaction shifts $Δ_{+ L T} / Δ_{+ 0}$ (blue circles), obtained by averaging data points at evolution times $10 \leq t \leq 30 ms$ , as a function of $κ_{F} a$ . (b) Evolution of the atomic peak full width at half maximum $w_{+} (t)$ at $κ_{F} a ≃ 0.9$ , normalized to the noninteracting value $w_{NI}$ . Inset: atomic probe spectra at $κ_{F} a ≃ 0.9$ recorded at 0 ms (orange squares) and 10 ms (blue circles) after the quench, compared to that of a noninteracting gas (green crosses). The amplitude of each spectrum is normalized to unity and artificially shifted to zero detuning. (c) Spectral width $w_{L T}$ at long times normalized to the $t = 0$ value $w_{+ 0}$ , as a function of $κ_{F} a$ . Vertical error bars in all panels combine the standard error of Gaussian fits to the pump and probe spectra with the standard deviation resulting from averaging several data points.
Reuse & Permissions

Physical Review Letters