betaFeedback

WorksPhysics and Astronomy

Studying and applying magnetic dressing with a Bell and Bloom magnetometer

Abstract

The magnetic dressing phenomenon occurs when spins precessing in a static field (holding field) are subject to an additional, strong, alternating field. It is usually studied when such extra field is homogeneous and oscillates in one direction. We study the dynamics of spins under dressing condition in two unusual configurations. In the first instance, an inhomogeneous dressing field produces space dependent dressing phenomenon, which helps to operate the magnetometer in strongly inhomogeneous static field. In the second instance, beside the usual configuration with static and the strong orthogonal oscillating magnetic fields, we add a secondary oscillating field, which is perpendicular to both. The system shows novel and interesting features that are accurately explained and modelled theoretically. Possible applications of these novel features are briefly discussed.

Introduction

The physics of a precessing magnetization in a static magnetic field is well-known since the studies performed by Joseph Larmor at the end of the 19th century. The phenomenology of precessing systems is enriched when additional time-dependent fields are introduced. In many applications like magnetic resonance (MR) experiments an oscillating transverse field (which is usually much weaker than the static one) is applied resonantly to the Larmor one ΩL/2π\Omega_{\mathrm{L}}/2\pi. In this case the resonant condition helps in describing the system dynamics in the rotating wave approximation: the oscillating field is seen as a superposition of two counter-rotating components, one of which appears to be static in the reference frame rotating at ΩL\Omega_{\mathrm{L}} around the holding field: it causes a slow precession in that rotating frame.

The presence of a transverse (off-resonant) alternating field arouse interest also in the complementary regime, when its strength exceeds the static one. This condition is technically unfeasible in conventional NMR at Tesla level, while it is more easily accessible in atomic physics experiments, where the MR is commonly studied in much weaker fields, e.g. at microtesla level.

The seminal work studying precessing spins subjected to a strong, off-resonant, transverse field (dressing field) dates back to the late Sixties, when S. Haroche and co-workers [1] introduced the concept of magnetic dressing. That contribution started a vivid activity around the subject [2, 3, 4, 5]. Recently, the magnetic dressing was studied also with cold atoms and condensates [6, 7]. The magnetic dressing of atoms offers a powerful tool in quantum control experiments [6, 8] and high resolution magnetometry [9].

Our group devoted research activity to characterize the magnetic dressing phenomenon occurring when an atomic sample is subject to an anharmonic dressing field [10]. More recently, the potential of applying a non-uniform dressing field to counteract the effects of static field inhomogeneities was demonstrated [11]. This has important implications in NMR imaging (MRI) in the ultra low field regime [12].

As a general feature, the mentioned works studied the magnetic dressing in configurations where the precession around the holding field is modified by one time dependent field oriented along one direction in the plane perpendicular to the holding one (the precession plane).

In a more recent work [13], we studied theoretically and experimentally new features that emerge when the design of the dressing field uses both the dimensions of the precession plane. We addressed the case when such time-dependent two-dimensional field is periodic in time. The periodicity is obtained by superposing two perpendicular field components, which oscillate harmonically with various amplitudes and with frequencies that are equal or in (small) integer ratio pp.

The system is studied by means of a model that produces analytical results thanks to a perturbative approach. This approach requires that one of the time-dependent components is much larger than both the static field and the other oscillating term. Despite its weakness, the latter (denominated tuning field) plays an important role.

This paper is organized as follows: in § 2, we describe the model and derive the expression for the effective precession frequency in terms of the dressing parameters (amplitudes, frequencies and relative phase of the dressing and tuning fields); in § 3, we describe a practical application of magnetic dressing, which makes a Bell and Bloom atomic magnetometer suited to operate also in the presence of a strong field gradient. In § 4, we report experimental results obtained in the tuning-dressing configuration, and consider possible applications where our findings may be of interests. An appendix deepens some technical aspects of the calculations on which the model is based.

Model

The evolution of atomic magnetization in an external field is commonly described using the Bloch equations. In this work, the considered atomic sample is a buffered Cs cell at the core of a Bell & Bloom magnetometer [14]. The presence of a weak pump radiation synchronously modulated at the effective Larmor frequency compensates the weak relaxation phenomena and enables the use of a model that determines the steady-state solution on the basis of the simpler Larmor equation.

Consider the Larmor equation M˙=γB×M\dot{\mathbf{M}}=\gamma\mathbf{B}\times\mathbf{M} for the atomic magnetization M\mathbf{M} in presence of a magnetic field of the form

B=B1cos⁡(Ωt)x^+B2cos⁡(pΩt+ϕ)y^+B3z^(1)\mathbf{B}=B_1\cos(\Omega t)\hat{\mathbf{x}}+B_2\cos(p\Omega t+\phi)\hat{\mathbf{y}}+B_3\hat{\mathbf{z}} \tag*{(1)}

with pp an integer. We refer to B3B_3 as static field, to B1B_1 as dressing field and to B2B_2 as tuning field, respectively. Using the adimensional time τ=Ωt\tau=\Omega t one obtains explicitly (ωi=γBi\omega_i=\gamma B_i)

dMdτ=[ω1Ωcos⁡(τ)A1+ω2Ωcos⁡(pτ+ϕ)A2+ω3ΩA3]M,(2)\frac{d\mathbf{M}}{d\tau}=\left[\frac{\omega_1}{\Omega}\cos(\tau)A_1+\frac{\omega_2}{\Omega}\cos(p\tau+\phi)A_2+\frac{\omega_3}{\Omega}A_3\right]\mathbf{M}, \tag*{(2)}

where the three-dimensional matrices AiA_i are antisymmetric with only two elements different from zero i.e. (A1)2,3=−(A1)3,2=−1(A_1)_{2,3}=-(A_1)_{3,2}=-1. A2A_2 and A3A_3 are obtained from cyclic permutations of the indices: see eqs. A.1 in the appendix.

Assuming that ξ=ω1/Ω≫ω2/Ω,ω3/Ω\xi=\omega_1/\Omega\gg\omega_2/\Omega,\omega_3/\Omega the perturbation theory let factorize the time evolution operator U(τ)U(\tau) (M(τ)≡U(τ)M(0)\mathbf{M}(\tau)\equiv U(\tau)\mathbf{M}(0)) in the interaction representation as

U(τ)=exp⁡[ξsin⁡τA1]UI(τ)=(1000cos⁡φ−sin⁡φ0sin⁡φcos⁡φ)UI(τ),(3)U(\tau)=\exp[\xi\sin\tau A_1]U_I(\tau)= \begin{pmatrix} 1&0&0\\ 0&\cos\varphi&-\sin\varphi\\ 0&\sin\varphi&\cos\varphi \end{pmatrix}U_I(\tau), \tag*{(3)}

where φ=ξsin⁡τ\varphi=\xi\sin\tau, obtaining a dynamical equation for UI(τ)U_I(\tau): dUI/dτdU_I/d\tau is given by

[(ω3Ωsin⁡φ+ω2Ωcos⁡φcos⁡(pτ+ϕ))A2+(ω3Ωcos⁡φ−ω2Ωsin⁡φcos⁡(pτ+ϕ))A3]UI≡ϵA(τ)UI,\left[\left(\frac{\omega_3}{\Omega}\sin\varphi+\frac{\omega_2}{\Omega}\cos\varphi\cos(p\tau+\phi)\right)A_2+\left(\frac{\omega_3}{\Omega}\cos\varphi-\frac{\omega_2}{\Omega}\sin\varphi\cos(p\tau+\phi)\right)A_3\right]U_I\equiv\epsilon A(\tau)U_I,

where the parameter ϵ\epsilon is used to identify the various orders of perturbation theory and to label the corresponding terms.

The matrix A(τ)A(\tau) is periodic A(τ+2π)=A(τ)A(\tau+2\pi)=A(\tau) so we can use the Floquet theorem and write

UI(τ)=eΛ(τ)eτF(4)U_I(\tau)=e^{\Lambda(\tau)}e^{\tau F} \tag*{(4)}

with Λ(0)=0\Lambda(0)=0 and Λ(τ+2π)=Λ(τ)\Lambda(\tau+2\pi)=\Lambda(\tau), and the Floquet matrix FF is time-independent. Next, we expand UIU_I following the Floquet-Magnus [15] expansion

Λ=ϵΛ1+ϵ2Λ2+⋯ ,F=ϵF1+ϵ2F2+⋯(5)\Lambda=\epsilon\Lambda_1+\epsilon^2\Lambda_2+\cdots,\qquad F=\epsilon F_1+\epsilon^2F_2+\cdots \tag*{(5)}

where the first terms are obtained from the relations

F1=12π∫02πA(τ) dτ,Λ1(τ)=∫0τA(τ′) dτ′−τF1.(6)F_1=\frac{1}{2\pi}\int_0^{2\pi}A(\tau)\,d\tau,\qquad\Lambda_1(\tau)=\int_0^\tau A(\tau')\,d\tau'-\tau F_1. \tag*{(6)}

Using the relation involving the Bessel functions JnJ_n

eizsin⁡θ=∑n=−∞+∞Jn(z)einθ(7)e^{iz\sin\theta}=\sum_{n=-\infty}^{+\infty}J_n(z)e^{in\theta} \tag*{(7)}

one finds

∫0τcos⁡(φ(τ′)) dτ′=J0(ξ)τ+f1(τ)(8)\int_0^\tau\cos(\varphi(\tau'))\,d\tau'=J_0(\xi)\tau+f_1(\tau) \tag*{(8)}
∫0τsin⁡(φ(τ′)) dτ′=f2(τ)(9)\int_0^\tau\sin(\varphi(\tau'))\,d\tau'=f_2(\tau) \tag*{(9)}
∫0τcos⁡(φ(τ′))cos⁡(pτ′+ϕ) dτ′=1+(−1)p2τJp(ξ)cos⁡ϕ+f3(τ)(10)\int_0^\tau\cos(\varphi(\tau'))\cos(p\tau'+\phi)\,d\tau'=\frac{1+(-1)^p}{2}\tau J_p(\xi)\cos\phi+f_3(\tau) \tag*{(10)}
∫0τsin⁡(φ(τ′))cos⁡(pτ′+ϕ) dτ′=−1+(−1)p2τJp(ξ)sin⁡ϕ+f4(τ).(11)\int_0^\tau\sin(\varphi(\tau'))\cos(p\tau'+\phi)\,d\tau'=\frac{-1+(-1)^p}{2}\tau J_p(\xi)\sin\phi+f_4(\tau). \tag*{(11)}

The functions fi(τ)f_i(\tau) are reported in the appendix (see eqs. A.4 - A.7).

Using the eq. (8) one finds

F1={ω3ΩJ0(ξ)A3+ω2ΩJp(ξ)cos⁡ϕA2p even[ω3ΩJ0(ξ)+ω2ΩJp(ξ)sin⁡ϕ]A3p odd(12)F_1= \begin{cases} \frac{\omega_3}{\Omega}J_0(\xi)A_3+\frac{\omega_2}{\Omega}J_p(\xi)\cos\phi A_2 & p\ \text{even}\\ \left[\frac{\omega_3}{\Omega}J_0(\xi)+\frac{\omega_2}{\Omega}J_p(\xi)\sin\phi\right]A_3 & p\ \text{odd} \end{cases} \tag*{(12)}

and

Λ1=(ω3Ωf2(τ)+ω2Ωf3(τ))A2+(ω2Ωf1(τ)−ω2Ωf4(τ))A3(13)\Lambda_{1}=\left(\frac{\omega_{3}}{\Omega}f_{2}(\tau)+\frac{\omega_{2}}{\Omega}f_{3}(\tau)\right)A_{2}+\left(\frac{\omega_{2}}{\Omega}f_{1}(\tau)-\frac{\omega_{2}}{\Omega}f_{4}(\tau)\right)A_{3} \tag*{(13)}

Using the eq. (12) (see appendix for details) we can calculate the dressed Larmor frequency measured in the experiment for the component of the magnetization parallel to the dressing field:

ΩL={ω32J02(ξ)+ω22Jp2(ξ)cos⁡2ϕp even∣ω3J0(ξ)+ω2Jp(ξ)sin⁡ϕ∣p odd(14)\Omega_{L}= \begin{cases} \sqrt{\omega_{3}^{2}J_{0}^{2}(\xi)+\omega_{2}^{2}J_{p}^{2}(\xi)}\cos^{2}\phi& p\ \text{even}\\ \lvert\omega_{3}J_{0}(\xi)+\omega_{2}J_{p}(\xi)\sin\phi\rvert& p\ \text{odd} \end{cases} \tag*{(14)}

The eq. (14) is at the focus of this paper. It includes also the well known and simpler case of a single dressing field: for zero values of the tuning field (i.e. ω2=0\omega_{2}=0) it gives ΩL=∣J0(ξ)ω3∣\Omega_{L}=\lvert J_{0}(\xi)\omega_{3}\rvert. The latter is the instance discussed in § 3, where only the dressing field is applied, with the peculiarity of a position-dependent parameter: ξ=ξ(x)\xi=\xi(x). The more general case, with the presence of both the dressing and the tuning fields, is discussed in § 4.

Atomic resonance enhancement with inhomogenous dressing

An interesting magnetometric application of the dressing phenomenon is based on applying a position-dependent dressing (and/or tuning) field to compensate atomic MR broadening caused by inhomogeneities of the static field. Unless counteracted, the broadening induced by strongly inhomogeneous static field would severely deteriorate the performance of the magnetometer.

Details of the magnetometric experimental apparatus can be found elsewhere [16]. Beside the field control system described in ref. [16], here several additional coils enable the application of the time dependent (dressing and tuning) fields. In particular, the stronger dressing field is applied with a solenoidal coil surrounding the sensor and the weaker tuning field is applied by means of an Helmholtz pair. Alternatively, when an inhomogenous dressing field is required, it is produced by a small source (a coil wound on a hollow-cylinder ferrite, with the laser beams passing across the hole) placed in the proximity of the atomic cell.

As discussed in Refs.[11, 12], the application of inhomogenous dressing field paves the way to detect NMR imaging signals in loco, that is with an atomic sensor operating in the same place where the NMR sample is. In fact, the frequency encoding technique used in MRI requires the application of static gradients which would destroy the atomic resonance.

The basic principle of operation can be summarized as follows: let the NMR sample and the sensor atom be in a static field (B3(x))(B_{3}(x)) with an intensity dependent on the xx co-ordinate, thus ω3=ω3(x)=γ(B3(0)+Gx)\omega_{3}=\omega_{3}(x)=\gamma(B_{3}(0)+Gx), where G=∂B3/∂xG=\partial B_{3}/\partial x is the gradient applied to the frequency encoding purpose, which in our case is up to about 150 nT/cm. Over the centimetric-size of the atomic sample, such a gradient would broaden the atomic MR from its original few-Hz width up to kHz level, smashing it completely and preventing any magnetometric measurement.

The narrow width of the atomic magnetic resonance can be restored by applying tailored tuning and/or dressing fields. We will consider here the case of single field dressing, where the gradient GG makes ΩL\Omega_{L} position-dependent according to

ΩL(x)=∣J0(ξ(x))ω3(x)∣,(15)\Omega_{L}(x)=\lvert J_{0}(\xi(x))\omega_{3}(x)\rvert, \tag*{(15)}

where we have considered a position dependent dressing field, making the dressing parameter ξ\xi dependent on xx, as well.

An appropriate inhomogeneity of ξ\xi makes possible to achieve the condition ∂ΩL/∂x=0\partial\Omega_{L}/\partial x=0, under which the atomic resonance width is restored. It is worth noting that the nuclear precession is unaffected by the dressing field. The much smaller gyromagnetic factor, makes their dressing parameter vanishing and the effects of B1B_{1} negligible: the MRI frequency encoding based on GG is preserved.

In this configuration, B1B_1 is inhomogeneous, and, in a dipole approximation, at a distance xx from the cell center, the dressing field BxB_x is

Bx(x,t)=μ02πm(t)(x0+x)3=B1(x)cos⁡(Ωt),(16)B_x(x,t)=\frac{\mu_0}{2\pi}\frac{m(t)}{(x_0+x)^3}=B_1(x)\cos(\Omega t), \tag*{(16)}

where μ0\mu_0 is the vacuum permittivity, m(t)=m0cos⁡(Ωt)m(t)=m_0\cos(\Omega t) is the oscillating dipole momentum, and x0x_0 is the distance of the dipole from the cell center. Taking now into account the dependence on xx of both the static and the dressing fields, accordingly with the eq.11, the dressed angular frequency results

ΩL(x)=γCs2π(B0+Gx)J0(γCsB1(x)Ω),(17)\Omega_L(x)=\frac{\gamma_{\mathrm{Cs}}}{2\pi}(B_0+Gx)J_0\left(\frac{\gamma_{\mathrm{Cs}}B_1(x)}{\Omega}\right), \tag*{(17)}

or, in a first-order Taylor approximation,

ΩL(x)=ΩL(0)+ΩL′(0)x+O(x2)≈γCs2π(B0J0(α)+[3B0αJ1(α)x0+GJ0(α)]x),\Omega_L(x)=\Omega_L(0)+\Omega'_L(0)x+O(x^2)\approx\frac{\gamma_{\mathrm{Cs}}}{2\pi}\left(B_0J_0(\alpha)+\left[\frac{3B_0\alpha J_1(\alpha)}{x_0}+GJ_0(\alpha)\right]x\right),

where α=(μ0/2π)(γCsm0)/(Ωx03)\alpha=(\mu_0/2\pi)(\gamma_{\mathrm{Cs}}m_0)/(\Omega x_0^3). In conclusion, the condition for compensating the effect of the gradient GG is found to be:

−3B0x0αJ1(α)J0(α)=G.(18)-3\frac{B_0}{x_0}\frac{\alpha J_1(\alpha)}{J_0(\alpha)}=G. \tag*{(18)}

This shows that with an appropriate choice of α\alpha, the width of the atomic magnetic resonance is substantially restored, as to recover the magnetometer performance to a level making it suited to detect weak MRI signals.

The Figure 1 shows MRI profiles obtained with this methodology. The frequency encoding is here obtained with a field gradient G≈100 nT/cmG\approx100\,\mathrm{nT/cm}. The presence of GG broadens the atomic MR from its original 25 Hz up to about 800 Hz. Its width is then restored down to 35 Hz by inhomogeneous dressing.

Image not extracted

Figure 1. 1-D imaging of a structured (four disks) water sample contained in the plastic cartridge shown in the picture (b). The plots (a) represent MRI profiles, as obtained when the sample occupies four different positions xsx_s. Both the structure details and the positions (actual xsx_s values are reported the right-side y-axes) are determined with submillimetric precision.

Tuning-dressing experiment

The Eq. (14) expresses the dependence of dressed angular frequency ΩL\Omega_{\mathrm{L}} on the experimental parameters (strengths, relative phase and frequencies of the tuning and dressing fields). The predicted behaviour is verified using the mentioned Bell and Bloom magnetometer. Measurements are made with p=1,2,3p=1,2,3, and ΩL\Omega_{\mathrm{L}} is recorded as a function of ω1\omega_1 and ϕ\phi.

Figure 2 shows comparisons of measured and calculated values of ΩL\Omega_{\mathrm{L}} as a function of ϕ\phi for given ξ\xi and as a function of ξ\xi for given ϕ\phi, respectively. The data shown in Figure 2(b) are measured with the ϕ\phi values that maximize the tuning effect. i.e. is ϕ=π/2\phi=\pi/2 for the odd pp values and ϕ=0\phi=0 for the even one.

Image not extracted

Figure 2. (a) ΩL\Omega_{\mathrm{L}} as a function of the relative phase ϕ\phi between the fields B1B_1 and B2B_2. In black dots the experimental measurements, in blue the theoretical predictions. Upper, middle and bottom panels refer respectively to the cases with p=1,2,3p=1,2,3, in which ξ\xi has been fixed to 1.38,3.83,1.541.38, 3.83, 1.54. (b) ΩL\Omega_{\mathrm{L}} as a function of the Bessel functions argument ξ\xi. In black dots the experimental measurements, in blue the theoretical predictions. Upper, middle and bottom panels refer respectively to the cases with p=1,2,3p=1,2,3. In the cases with an odd pp the phase between B1B_1 and B2B_2 is fixed to ϕ=π/2\phi=\pi/2, when pp is even the phase ϕ\phi is set to zero. In each plot, ξ\xi changes varying the value of B1B_1, which ranges from zero to ≈15 μT\approx15\ \mu\mathrm{T}.

There is an excellent accordance between the experimental data and the theoretical prediction. In particular, both the dependence of the relative phase ϕ\phi and on the pertinent JpJ_p (with p=1,2,3p=1,2,3) Bessel functions are perfectly verified by the experiment. Minor deviations can be attributed to experimental imperfections, mainly to not-exact perpendicularity of the applied fields B1B_1 and B2B_2.

The developed model and its excellent correspondence with the experimental results demonstrates the possibility to enhance the control of spins evolution by means of the described tuning-dressing field arrangement.

This enhancement opens up to a variety of new experimental configurations in which the new set of parameters (B2B_2, pp and ϕ\phi) add to B1B_1 and Ω\Omega to make new handles available to finely control the atomic magnetization dynamics. Compared to the known cases of harmonic [1] or anharmonic [10] dressing field oscillating along one direction, noticeably here the dressed frequency ΩL\Omega_{\mathrm{L}} (eq. (14)) may exceed ω3\omega_3.

The tuning-dressing scheme makes possible to chose a parameter set to achieve an arbitrary dΩL/dξd\Omega_{\mathrm{L}}/d\xi in conjunction with an arbitrary (to some extent) value of ΩL\Omega_{\mathrm{L}}. For instance, it is possible to produce a condition of critical dressing (equalization of precession frequencies of different species) [17, 9] with no first-order dependence on ξ\xi, so to attenuate the detrimental effects caused by B1B_1 inhomogeneities, which constitute a severe limiting problem in high-resolution experiments [9].

Similarly, it is possible to fulfill the condition of a large dΩL/dξd\Omega_{\mathrm{L}}/d\xi avoiding the constraint of a strong ΩL\Omega_{\mathrm{L}} reduction. The latter may help in applications like that described in § 3, where ΩL\Omega_{\mathrm{L}} is made deliberately position-dependent by means of a spatially inhomogeneous ξ\xi. In addition, for that kind of application, it is worth noting that the presented scheme makes it possible to render ΩL\Omega_{\mathrm{L}} space dependent by means of an inhomogeneity of the field B2B_2, which is of easier implementation and control, being B2≪B1B_2 \ll B_1.

Other applications where the tuning-dressing scheme may offer important potentials is suggested by the dependence on ϕ\phi. As recently reported [18], an emerging application of highly sensitive magnetometers concerns the detection of targets made of weakly conductive materials. In that case, the typical setup is based on a radio-frequency magnetometer, where the target modifies the amplitude or the phase of a (resonant) radio-frequency field driving the magnetometer. Alternative setups could be developed, where the target modifies the field B2B_2, whose frequency is not required to match the atomic resonance. In this case, provided that Jp(ξ)J_p(\xi) is large (e.g. ξ≈1.84\xi\approx1.84 in the case of p=1p=1), the system would have a large response to any variation of either the amplitude (if ϕ=±π/2\phi=\pm\pi/2) or the phase (if ϕ=0,π\phi=0,\pi) of B2B_2 caused by eddy currents induced in the target.

Appendix

The explicit expressions for the matrices in the main text are

A1=(00000−1010),A2=(001000−100),A3=(010−100000)(19)A_1=\begin{pmatrix}0&0&0\\0&0&-1\\0&1&0\end{pmatrix},\quad A_2=\begin{pmatrix}0&0&1\\0&0&0\\-1&0&0\end{pmatrix},\quad A_3=\begin{pmatrix}0&1&0\\-1&0&0\\0&0&0\end{pmatrix} \tag*{(19)}

Defining Lj=iAjL_j=iA_j and taking the base that diagonalizes L3L_3 one obtains the familiar form for the angular momentum operators acting on the ∣L=1,ML⟩\lvert L=1,M_L\rangle states. This demonstrates that the algebra generated by the AiA_i matrices is the same of the quantum angular momentum operators. Moreover, it is straightforward to demonstrate that the quantum mean value ⟨γL⟩\langle\gamma L\rangle satisfies the same classical Larmor equation.

The action of a general antisymmetric matrix W=a1A1+a2A2+a3A3W=a_1A_1+a_2A_2+a_3A_3 on a given vector v\mathbf{v} is the cross-product

Wv=a×va≡(a1,a2,a3).(20)W\mathbf{v}=\mathbf{a}\times\mathbf{v}\qquad\mathbf{a}\equiv(a_1,a_2,a_3). \tag*{(20)}

Using the Cayley-Hamilton theorem, the needed matrix exponentials can be analytically evaluated. Let write a=θ^a^\mathbf{a}=\hat{\theta}\hat{\mathbf{a}} with a^⋅a^=1\hat{\mathbf{a}}\cdot\hat{\mathbf{a}}=1 then

eWv=∑n=0+∞Wnn!v=v+sin⁡θ a^×v+(1−cos⁡θ)a^×(a^×v).(21)e^W\mathbf{v}=\sum_{n=0}^{+\infty}\frac{W^n}{n!}\mathbf{v}=\mathbf{v}+\sin\theta\,\hat{\mathbf{a}}\times\mathbf{v}+(1-\cos\theta)\hat{\mathbf{a}}\times(\hat{\mathbf{a}}\times\mathbf{v}). \tag*{(21)}

The auxiliary fif_i functions introduced in the main text are defined as

f1(τ)=∑n=1∞J2n(ξ)nsin⁡(2nτ)(22)f_1(\tau)=\sum_{n=1}^{\infty}\frac{J_{2n}(\xi)}{n}\sin(2n\tau) \tag*{(22)}
f2(τ)=4∑n=0∞J2n+1(ξ)2n+1sin⁡2((n+1/2)τ)(23)f_2(\tau)=4\sum_{n=0}^{\infty}\frac{J_{2n+1}(\xi)}{2n+1}\sin^2\left((n+1/2)\tau\right) \tag*{(23)}
f3(τ)=R(g(τ))(24)f_3(\tau)=\mathfrak{R}(g(\tau)) \tag*{(24)}
f4(τ)=S(g(τ))(25)f_4(\tau)=\mathfrak{S}(g(\tau)) \tag*{(25)}

where

g(τ)=eiϕ∑n≠−pJn(ξ)i(n+p)(ei(n+p)τ−1)+e−iϕ∑n≠pJn(ξ)i(n−p)(ei(n−p)τ−1)(26)g(\tau) = e^{i\phi} \sum_{n\ne-p} \frac{J_n(\xi)}{i(n+p)}\left(e^{i(n+p)\tau}-1\right) + e^{-i\phi} \sum_{n\ne p} \frac{J_n(\xi)}{i(n-p)}\left(e^{i(n-p)\tau}-1\right) \tag*{(26)}

These functions have a limited and oscillating behaviour and are needed in the evaluation of Λ1(τ)\Lambda_1(\tau). One can see by inspection that eΛ1≈1e^{\Lambda_1}\approx1 is a good approximation.

The initial condition appropriate for the experiment is M(0)∝(1,0,0)\mathbf{M}(0)\propto(1,0,0) and the quantity monitored is Mx(τ)M_x(\tau). So applying eq. (A.3) to evaluate eτF1e^{\tau F_1}, and then eq. (3), one finds

Mx(t)∝cos⁡(ΩLt).(27)M_x(t)\propto\cos(\Omega_L t). \tag*{(27)}

References

  1. [1]Haroche S, Cohen-Tannoudji C, Audoin C and Schermann J P 1970 Phys. Rev. Lett. 24 861–864
  2. [2]Yabuzaki T, Tsukada N and Ogawa T 1972 Journal of the Physical Society of Japan 32 1069–1077
  3. [3]Kunitomo M and Hashi T 1972 Physics Letters A 40 75–76
  4. [4]Ito H, Ito T and Yabuzaki T 1994 Journal of the Physical Society of Japan 63 1337
  5. [5]Holthaus M 2001 Phys. Rev. Applied 64 011601
  6. [6]Gerbier F, Widera A, Fölling S, Mandel O and Bloch I 2006 Phys. Rev. A 73(4) 041602 URL https: //link.aps.org/doi/10.1103/PhysRevA.73.041602DOI
  7. [7]Hofferberth S, Fischer B, Schumm T, Schiedmayer J and Lesanovsky I 2007 Physical Review A - Atomic, Molecular, and Optical Physics 76 013401
  8. [8]Pervishko A A, KIBIS O V, Morina S and Shelykh I A 2015 Phys. Rev. B 92(20) 205403 URL https: //link.aps.org/doi/10.1103/PhysRevB.92.205403DOI
  9. [9]Swank C M, Webb E K, Liu X and Filippone B W 2018 Phys. Rev. A 98(5) 053414 URL https: //link.aps.org/doi/10.1103/PhysRevA.98.053414DOI
  10. [10]Bevilacqua G, Biancalana V, Dancheva Y and Moi L 2012 Phys. Rev. A 85 042510 (Preprint 1112.1309)
  11. [11]Bevilacqua G, Biancalana V, Dancheva Y and Vigilante A 2019 Phys. Rev. Applied 11(2) 024049 URL https://link.aps.org/doi/10.1103/PhysRevApplied.11.024049DOI
  12. [12]Bevilacqua G, Biancalana V, Dancheva Y and Vigilante A 2019 Appl.Phys.Lett. 115 174102 (Preprint https://arxiv.org/abs/1908.01283) URL https://doi.org/10.1063/1.5123653
  13. [13]Bevilacqua G, Biancalana V, Vigilante A, Zanon-Willette T and Arimondo E 2020 Phys. Rev. Lett. 125(9) 093203 URL https://link.aps.org/doi/10.1103/PhysRevLett.125.093203DOI
  14. [14]Bell W E and Bloom A L 1957 Phys. Rev. 107(6) 1559–1565 URL https://link.aps.org/doi/10.1103/PhysRev.107.1559DOI
  15. [15]Blanes S, Casas F, Oteo J and Ros J 2009 Physics Reports 470 151 – 238 ISSN 0370-1573 URL http: //www.sciencedirect.com/science/article/pii/S0370157308004092
  16. [16]Bevilacqua G, Biancalana V, Chessa P and Dancheva Y 2016 Applied Physics B 122 103 ISSN 1432-0649 URL http://dx.doi.org/10.1007/s00340-016-6375-2
  17. [17]Golub R and Lamoreaux S K 1994 Physics Reports 237 1 – 62 ISSN 0370-1573 URL http://www.sciencedirect.com/science/article/pii/0370157394900841
  18. [18]Marmugi L, Deans C and Renzoni F 2019 Applied Physics Letters 115 083503 (Preprint https://doi.org/10.1063/1.5116811) URL https://doi.org/10.1063/1.5116811

Paper details

vlnc:W_msmf5afc
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents