Efficient Loading of an Yb MOT on the ¹S₀ → ¹P₁ Transition

Abstract

We demonstrate efficient loading of an Yb three-dimensional magneto-optical trap (3D MOT) on the ¹S₀ → ¹P₁ transition. The experiment employs a two-dimensional magneto-optical trap (2D MOT) as an efficient cold atom source. Through optimization of the 2D MOT, auxiliary Zeeman slower, and push beam parameters that govern atomic capture, we achieve an atomic loading rate of 5 × ${10}^6$ atoms/s in the 3D MOT, with approximately ∼10⁷ trapped atoms. Our experimental results confirm the successful transfer of a substantial number of atoms from the 2D region to the science chamber via the push beam, providing an experimental foundation for subsequent implementation of narrow-line MOT and two-color MOT.

1. Introduction

Alkaline-earth atoms play a crucial role in quantum technologies, particularly in quantum information processing [1,2], high-precision quantum simulation [3,4,5,6], and quantum interferometry [7,8]. Among them, ytterbium atoms have emerged as an ideal system for studying many-body interactions, quantum degenerate gases, and optical lattice clocks [9,10,11,12], owing to their narrow-line metastable state.

The narrow-linewidth cooling of ytterbium atoms presents significant technical challenges due to weak cooling forces and limited capture velocity. To overcome these challenges, researchers have made extensive efforts. These include crossed-beam slowing techniques that enhance loading rate by combining a Zeeman slower with additional broad-line transition laser beams [13]. Two-color MOT systems utilize both broad-line and narrow-line transitions for rapid loading and efficient cooling [14]. For single-atom manipulation, narrow-linewidth cooling and imaging in optical tweezer arrays has been demonstrated using the narrow-line transition with frequency broadening [15]. Additionally, compact 2D-/3D-MOT setups achieve narrow-linewidth trapping and quantum-degenerate gases without a Zeeman slower [16].

Among the various approaches, the two-dimensional magneto-optical trap (2D MOT) has emerged as a particularly promising solution, owing to its compact structure and efficiency as an atomic source [15,16,17,18,19]. This approach effectively circumvents the technical limitations associated with traditional methods such as the Zeeman slower [20,21], including vacuum contamination and interference from blackbody radiation generated by high-temperature ytterbium ovens. While the 2D MOT offers advantages in compactness and efficiency, it delivers relatively low atom flux to the 3D region. This makes it difficult to directly observe atomic fluorescence and reliably determine whether the push beam has successfully transferred atoms from the 2D to the 3D region. This confirmation is crucial for the successful implementation of narrow-line three-dimensional magneto-optical trap (3D MOT). To address this challenge, the broad-line 3D MOT offers a viable solution, while certain experimental aspects of this approach remain insufficiently explored.

Herein, we employed a 2D MOT as the atomic source to investigate the loading of Yb atoms in a broad-line 3D MOT. The loading rate was systematically characterized as a function of various experimental parameters, including the optical power for the 2D MOT, the power for the Zeeman slower beam, and different push beam configurations. The results provide a robust foundation for the investigation of narrow-linewidth laser cooling based on the ¹S₀ → ³P₁ intercombination transition at 556 nm.

2. Experimental System

The experimental protocol was executed under ultra-high-vacuum (UHV) conditions (∼2.2 × ${10}^{-10}$ Torr) shown in Figure 1a. The system comprises three chambers: an oven chamber, a two-dimensional magneto-optical trap (2D MOT) for transverse cooling, and a science chamber. The science chamber houses a three-dimensional magneto-optical trap (3D MOT) dedicated to the preparation of ultracold atoms. Notably, this setup also contains an auxiliary Zeeman slower, as shown in Figure 1b.

Yb samples were heated in a home-made oven chamber to 380 ± 2 °C to achieve a saturated vapor pressure of 3 × ${10}^{-3}$ Torr for continuous atomic flux generation. The thermal atomic flux was hydrodynamically collimated using an array of precision-engineered microchannel nozzles. This optimized geometry facilitated the delivery of a high-flux low-divergence atomic beam to subsequent 2D laser cooling stages.

In the 2D MOT region, eight sets of permanent magnets were arranged in a symmetric configuration in the orthogonal plane defined by the cross-arm structure. As shown in Figure 1b, the magnet array generated a magnetic field with axial and radial gradients of 74.90 and 6.185 G/cm, respectively. A push beam transported atoms from the 2D MOT region into the science chamber, where they were cooled by a 3D MOT on the ¹S₀ → ¹P₁ transition. The $1/e^2$ diameters of the 2D MOT and push beam were approximately 16.8 and 6.8 mm, respectively. A pair of anti-Helmholtz coils positioned above and below the science chamber provided the magnetic field gradient necessary for the 3D MOT. The axial magnetic field gradient of the 3D MOT was typically maintained at 46 G/cm. The atom number and MOT loading rate were determined by collecting the fluorescence.

The excited states of Yb gave rise to two important transitions (Figure 1c). The broad-line transition (¹S₀ → ¹P₁, 399 nm, linewidth 29 MHz) was used for the blue MOT, with a fiber laser generating the required 399 nm laser for cooling. Additionally, the ¹S₀ → ³P₁ transition, which is a narrow-line transition characterized by its low scattering force and slow acceleration, can also serve as a push beam [15].

The optical path system is schematically presented in Figure 2. There are 2D MOT cooling beams, a Zeeman slower beam, a push beam and blue MOT trapping beams. The laser frequency is locked at a detuning of −109 MHz from the ¹S₀ → ¹P₁resonance transition; the 2D MOT trapping beams are frequency-shifted by the first-order diffraction of an acousto-optic modulator (AOM) driven at +74 MHz, corresponding to a detuning of −35 MHz. The Zeeman slower (ZS) beam shares the same AOM with the 2D MOT trapping beams and uses the zeroth-order diffracted beam, producing a detuning of −109 MHz. The push beam is modulated by an AOM at +109 MHz, matching the resonant frequency. Both the horizontal and vertical trapping beams of the 3D MOT are shifted by AOMs operating at +80 MHz, resulting in a detuning of −19 MHz. An ultra-low-expansion (ULE) cavity was employed to lock the fiber laser [22]. The fineness of this cavity is approximately 1500, and the free spectral range (FSR) is 1.5 GHz; this exhibits a full width at half maximum (FWHM) of the cavity resonance for near ∼1 MHz. Compared with the wavelength meter locking scheme, the laser frequency locked to the cavity is more stable due to smaller drift. This study also employed a 556 nm laser as a push beam to examine its influence on the atom loading rate. Like the blue laser locking scheme, the 556 nm laser system was also frequency-stabilized by another ultrastable optical reference cavity (stable laser systems).

3. Results and Discussion

The real-time fluorescence was detected via an Amplified Detector (APD) (PDA20X2). The loading curves are shown in Figure 3. We analyze the curves by the function of $N\left(t\right)={\displaystyle \frac{R}{\gamma }}(1-e^{-\gamma t})$ [23,24], where R is the loading rate and $\gamma$ is the collisional loss rate. The insert in Figure 3 is the fluorescence image of the 3D MOT with atom number around 10⁶ at a low field of ∼4 G/cm.

We characterized the dependence of the atomic loading rate on several important optical parameters (Figure 4), including the distance from the laser power for the 2D MOT and the Zeeman slower beam.For the 2D MOT, the loading rate increased with the optical power. The limited laser power was insufficient to achieve saturation, although the loading rate was expected to saturate at high power [25]. The solid line represents the fitting curve with a theory. The loading rate L is proportional to the atomic flux Φ, and the atomic flux is proportional to $v_c^4$; it follows that $\mathrm{\Phi }\propto v_c^4$, where $v_c\propto \sqrt{s(s+1)}$, $s=I/I_{sat}$ and $v_c$ represents the capture velocity [17].

Figure 4b shows the power effect of the Zeeman slower beam, which has similar behavior to Figure 4a; we use the same model to fit the data.

For the 399 nm push beam, the loading rates are maximized around 0.01 $I_{sat}$ ($I_{sat}$ = 60 mW/cm²), with higher powers resulting in lower loading rates (Figure 5a). This behavior occurs because larger push beam intensities result in the finite capture efficiency of longitudinal velocities beyond the capture velocity. Consequently, when the push beam intensities exceed a certain threshold, the atomic loading rate decreases.

We had taken the broad-line transition (29 MHz) laser as push beam to research the power dependence of the loading rate. However, owing to the large natural linewidth of the ¹S₀ → ¹P₁ transition, the scattering force due to this transition would push the atoms across the vacuum chamber at a velocity beyond the capture velocity of the narrow-line MOT; this makes it difficult to capture atoms in a narrow-line MOT in further experiments. Hence, we also researched the green laser (556 nm) with a narrow-line transition (182 kHz) as a push beam. In order to find the resonance frequency for the narrow-line transition, spectroscopy was performed, shown in Figure 5b.

Owing to the Doppler effect, the atomic loading rate was offset relative to the resonant transition. The detuning for the maximum atomic loading rate was approximately −0.5 MHz with respect to the resonant frequency. When the detuning exceeded this critical threshold, the system entered the detuning regime ($|\Delta |\gg \Gamma$), the scattering force decreasing because of a decrease in the photon scattering cross-section. This decrease resulted in a reduction in the atomic loading rate. The power effect is not shown here, as it should behave similarly to the blue laser shown in Figure 3. The maximum loading rate was lower than that observed for the blue laser because of the notably smaller scattering force in the former case. The loading curves for the 556 and 399 nm push beams are shown together in Figure 3; it can be found that the atomic loading rate was much higher for the 399 nm beam.

The atomic loading rate increased with the oven temperature (Figure 6). After setting each temperature, we allowed the system to stabilize for 30 min to reach thermal equilibrium before commencing measurements, ultimately recording a maximum loading rate of 9.19 × 10⁵ atoms/s. The solid line in the figure is a result obtained through simple theoretical fitting, where the loading rate is proportional to $P_{\mathrm{Yb}}\left(T\right)\times T^{-5/2}$ [26]. Although higher loading rates could theoretically be obtained through further temperature elevation, practical limitations prevented additional optimization. Specifically, to ensure equipment integrity and prevent potential thermal damage to the vacuum system components, we maintained the oven temperature at levels below the operational safety threshold.

4. Conclusions

This study demonstrates efficient loading of ytterbium atoms in a blue MOT. We obtain the relationship between the laser power for the 2D MOT and Zeeman slower within an adjustable power range and the loading rate of the 3D MOT. Even with limited laser power, where the atomic loading rate does not reach saturation, a sufficient number of atoms could still be obtained, which laid the foundation for subsequent optical tweezer experiments [27]. We also studied the influence of push beams at two different frequencies on the loading rate. The results show that using a 399 nm push beam leads to a higher atomic loading rate. Moreover, we found that a significant blue MOT could be established even at a low gradient of ∼4 G/cm, holding promise for the development of a two-color Yb MOT.

In addition, we also obtain the loading rate of multiple isotopes for ytterbium in the MOT. At an oven temperature of 643.15 K, the parameters of each isotope are summarized in

Table 1. Overview of loading rates of the stable isotopes of ytterbium. The natural abundances are referenced [28]. Δf is the relative frequency shift to ¹⁷⁴Yb for the singlet cooling transition. L shows the relative loading rates obtained in this work.

Isotope	Statistics	Abundance (%)	Δf (MHz)	L (106 Atoms/s)
171Yb (F = 3/2)	Fermionic	/	860	0.886
172Yb	Bosonic	21.9	540	0.648
173Yb (F = 7/2)	Fermionic	/	600	0.003
174Yb	Bosonic	31.8	0	2.171
176Yb	Bosonic	12.7	−500	0.235

, including statistics, abundance, detuning and loading rate. From the data in the table, it can be seen that the loading rate of isotopes is positively correlated with their natural abundance.