The $\ensuremath{\pi}$-phase shift between
∣F=1,mF=0⟩ and
∣F=2,mF=0⟩ states for an adiabatic rotation of a magnetic field observed in Ramsey atom interferometry arises from the negative sign of the transition amplitude between the
∣1,0⟩ and
∣2,0⟩ states when the wave functions are rotated to the opposite direction. Using a two-photon Raman atom interferometer with a cold ensemble of sodium atoms, the phase shift for a half-rotation of the magnetic field was confirmed to be $\ensuremath{\pi}\text{ }\text{rad}$ with an uncertainty of 4%, and the phase shift for a noncyclic adiabatic rotation of the magnetic field was investigated.
The Berry phase acquired by an electromagnetic field undergoing an adiabatic and cyclic evolution in phase space is a purely quantum-mechanical effect of the field. However, this phase is usually accompanied by a dynamical contribution and cannot be manifested in any light-beam interference experiment because it is independent of the field state. We here show that such a phase can be produced using an atom coupled to a quantized field and driven by a slowly changing classical field, and it is manifested in atomic Ramsey interference oscillations. We also show how this effect may be applied to one-step implementation of multiqubit geometric phase gates, which is impossible by previous geometric methods. The effects of dissipation and fluctuations in the parameters of the pump field on the Berry phase and visibility of the Ramsey interference fringes are analyzed.
In a recent paper we have studied the peculiar features of the Berry and Aharonov-Anandan geometric phases for isolated spins $S\ensuremath{\geqslant}1.$ We have assumed that they are submitted to a dipole and quadrupole coupling to external
E and
B fields with the mild restriction $\mathbf{E}\ifmmode\cdot\else\textperiodcentered\fi{}\mathbf{B}=0$. This implies discrete symmetries leading to remarkable simplifications of the geometry and algebra involved. The aim of the present work is to describe realistic proposals, within the realm of atomic physics, for the verification of some of our most significant theoretical predictions. There are several challenges to be overcome. For alkali-metal atoms, most commonly used in atomic interferometers, the only practical way to generate quadrupole coupling, with a strength comparable to the dipole one, is the ac Stark effect induced by a nearly resonant light beam. One then has to face the instability of the``dressed'' atom hyperfine (hf) level, a candidate for our isolated spin. One deleterious effect is the apparition of an imaginary part in the quadrupole to dipole coupling strength ratio, $\ensuremath{\lambda}$. Fortunately we have found a simple way to get rid of $\mathfrak{I}(\ensuremath{\lambda})$ by an appropriate detuning. We are left with an unstable isolated spin. This implies an upper bound to the quantum cycle duration
Tc. In the case of the Berry's phase,
Tc has a lower bound due to the necessity of keeping the nonadiabatic corrections below a predefined level. We have found a compromise in the case of the
F=2,m=0 87Rb ground-state hf level. This is our candidate for the measurement of the somewhat ``exotic'' Berry's phase acquired by the
S=2,m=0 state at the end of a quantum cycle involving a rotation of $\ensuremath{\pi}$ of the
E field---in practice, the linear polarization of the dressing beam ---about the
B field direction. We have found a way to implement, in a Ramsey-type interferometric measurement, the procedures aimed at controlling the nonadiabatic corrections, as described in detail in our previous theoretical paper. A numerical simulation of our experimental proposal shows that a
0.1 accurate determination of Berry's phase, free of nonadiabatic corrections, can be achieved. Measurements could also be considered for cold
52Cr chromium atoms with
S=3, where values of $\ensuremath{\lambda}\ensuremath{\simeq}1$ can be obtained with an instability smaller than in the
87Rb case, due to a more favorable spectroscopic structure. The
F=1,m=1 hf level of the
87Rb ground state offers the opportunity to extend the measurement of Aharonov-Anandan's phases beyond the case
S=21. We construct, using ``light shift,'' the Hamiltonian ${H}_{\ensuremath{\parallel}}(t)$ generating a closed circuit in the density-matrix space which satisfies at any time the ``parallel transport'' condition, thus making the quantum cycle free from the adiabaticity condition. We also consider the case of half-integer spins (e.g.,
201Hg,
135Ba, and
137Ba), with their own specific features. We show how the difference of Berry's phases for states
S=23 and
S=21, with
m=21, can be exploited to achieve a holonomic maximum entanglement of three qubits.
The full effect of adiabatically varying Hamiltonians on systems prepared in eigenstates of the stationary Hamiltonian often includes geometric Berry's phases. For degenerate systems these effects may result in transitions between distinct degenerate eigenstates that can be described in terms of a non-Abelian gauge potential. I demonstrate the explicit dynamic origin of such transitions between degenerate angular-momentum sublevels in an atom that is subject to collinear electric and magnetic fields that rotate adiabatically. The origin of the effects becomes clear when eigenstates of the total Hamiltonian, which includes the rotating fields, are calculated. The total Hamiltonian removes the degeneracy, and the eigenstates are in some cases linear combinations of the angular-momentum sublevels. Thus a system prepared in a specific sublevel may not remain in that sublevel, and the transition probability is exactly that given by an analysis of the geometric phase.
Berry's phase for a whole turn in a conical rotation of the magnetic field with a semiangle \ensuremath{\theta} has been clearly manifested free from the dynamical phase shift using the magnetic-field-insensitive two-photon transitions between sodium-ground hyperfine states having different signs of the
g factors. The solid angles for states with a positive
g factor and with a negative
g factor are verified to be $2\ensuremath{\pi}(1\ensuremath{-}\mathrm{cos}\ensuremath{\theta})$ and $\ensuremath{-}2\ensuremath{\pi}(1+\mathrm{cos}\ensuremath{\theta})$, respectively, for a right-handed rotation of a magnetic field and a semiangle of $0\ensuremath{\leqslant}\ensuremath{\theta}\ensuremath{\leqslant}\ensuremath{\pi}/2$.