Atom shutter using bender piezoactuator Open Access

Atomic beams are used as atom sources in a wide variety of experiments in the field of atomic physics, including applications such as Doppler-free spectroscopy and laser cooling and trapping for the study of ultra-cold atoms. Occasionally, the rapid manipulation of the atomic flux and a blockade of the unwanted effects produced by hot ovens at the right time can become critical issues that affect the success of these experiments. To manipulate the atomic flux, mechanical flags that are driven by a stepper motor,¹ a servo-motor,² a solenoid coil, or a pneumatic device³ are commonly used because they offer a reliable performance. However, these driving methods have several weaknesses in which they produce unwanted magnetic fields and their heavy moving parts limit the speed at which the shutter can operate without associated mechanical vibrations.

With the intention of developing ytterbium (Yb) optical lattice clocks (OLCs) with an uncertainty of 10⁻¹⁸,^4–7 which must be operated using an atom shutter to be capable of nonmagnetic performance and provide good ultra-high vacuum (UHV) compatibility, low mechanical vibration, long lifetimes, and rapid response times, we invented a flag-type atom shutter that is driven using a bender piezoactuator (PZT),⁸ that in principle produces a negligible magnetic field and is satisfactory in terms of the other properties mentioned above.

An atom shutter that was driven by the PZT was reported previously during the development of the Primary Atomic Reference Clock in Space (PARCS),^9,10 which used a stacked PZT with a displacement of several tens of $μ m$ and a special flexure-based architecture that allowed it to obtain flag displacement with a clear 15 mm aperture. In comparison to the previous stack PZT-driven atom shutter, we propose a bender PZT that offers a much greater displacement of a few mm, which then allows us to apply a simpler mechanism based on a rotating lever to obtain a similarly clear aperture. In our atom shutter design, all moving parts were installed in a compact manner in a UHV chamber, including the PZT, which reduced the mass of the moving parts while providing an increased flag speed and a reduced mechanical vibration. However, this all-in-a-vacuum installation method could prove problematic in terms of compatibility with the UHV conditions and robustness for a long-term operation. We performed tests to examine these issues and obtained satisfactory results. In this article, the test results are presented along with the design details shown in Section II; measurements of the mechanical properties of the shutter are described in Section III; the testing of UHV compatibility, lifetime, and applicability to a magneto-optical trap experiment is discussed in Section IV; and conclusions are drawn in Section V.

A bidirectional bender PZT, which was manufactured by Piezotechnology, Inc. (Model PL140.11), was used⁸ as the actuator for the developed atom shutter. The 45-mm-long bender PZT can swing up to a maximal 2 mm if one end of the bender is clamped, as shown in Figs. 1(a) and 1(b). In the experimental setup, an atomic beam that was emitted from an oven was clipped by an aperture with a 5 mm diameter; hence, the flag head was designed to have a diameter of 10 mm. The PZT was driven using a custom-made PZT driver that produced three outputs: fixed ground (GND), 60 V outputs for the two outer electrode layers, and a variable V_con output for the inner electrode layer (Fig. 1(a)). The V_con output was switched using a transistor-transistor logic (TTL) signal between the two voltage states, which were pre-adjusted using the offset and the variable gain shown in Fig. 1(a).

Two types of lever were considered to amplify the bender PZT’s displacement: an extended-arm lever and a rotating-arm lever, as shown in Figs. 2(a) and 2(b), respectively. The extended-arm lever has the advantage of ease of construction and an absence of frictional movement. However, this lever does require an arm of 15–20 cm in length to give an amplification ratio $α ∼ 5$ that provides sufficient flag head displacement to open up the aperture (Fig. 2(a)). This long arm increases the mass of the device and weakens its rigidity, which results in a mechanically unstable flag movement. We therefore selected the rotating-arm-type lever because the length of its longer arm can be reduced to 4–5 cm if the shorter arm is designed to be smaller than 1 cm, which gives an amplification ratio $α > 5$ (Fig. 2(b)). As a result, the total mass and the moment of inertia (MI) of all moving parts in the device can be reduced greatly.

The most significant feature of the atom shutter implemented in this work is its use of a Teflon rotator, which converts the bender PZT’s movement into the rotation of the flag and then holds the flag (Figs. 1(b) and 1(c)); the Teflon rotator material plays several important roles in this device. First, it serves as a good electrical insulator, which prevents the PZT electrodes from short-circuiting when the insulation coating on the PZT wears off. Second, the slippery Teflon surface attenuates both the mechanical vibrations and the shocks produced by the friction between the Teflon rotator and the stainless steel axis pole. Third, the Teflon rotator can extend the lifetime of the bender PZT by providing soft contact conditions.

We designed two types of Teflon rotator and flag assemblies, termed type A and type B here, as shown in Fig. 1(c). The type A Teflon rotator has three slots with depths of 3 mm and widths of 0.5–0.6 mm. The slot width is slightly narrower than the thickness of the PZT (0.6 mm). Therefore, when the PZT is inserted into the slot, the slot contacts the PZT elastically and thus provides adequate pressure to eliminate any backlash at the turning point of the PZT; the two other slots were cut to form side walls such that the center slot acts as a hinge by adaptively bending itself based on the variable angle between the Teflon rotator and the PZT. As a result, the positional accuracy of the flag was ensured and the stress on the PZT was also reduced. The insertion depth (1–1.5 mm) of the PZT into the slot was ultimately optimized through testing. The flag was composed of a thin stainless steel plate with a thickness of 0.15 mm, a circular head with a diameter of 10 mm, and a 35-mm-long arm with several holes punched along its center line to reduce its mass. The flag was fixed to the Teflon rotator via the coaxial holes on their bodies and tightened using a tiny flag-fixed bolt, as shown in Fig. 3(c). The total mass of the rotator and flag assembly is 1.22 g. When compared with

the type A shutter, the type B shutter has a longer 50 mm arm and a different hinge design on the Teflon rotator. The slot into which the PZT is inserted has thicker walls that cannot be bent; instead, a 1-mm-wide flexure hinge was carved between the slot and the hole of the rotating axis, as shown in Fig. 1(c). To reduce the overall mass further, the flag of the type B shutter was made using a thinner stainless steel plate with a thickness of only 0.1 mm, and the arm was also punched with larger rectangular cuts shown in Fig. 1(d). Additionally, the flag has two wings near its root. These wings were folded such that they embraced the Teflon rotator, and thus the two bodies were connected without the use of a flag-fixed bolt, as shown in Fig. 1(d). These design changes allowed the total mass of the type B moving part to be reduced to 0.43 g. However, because the flag arm was too thin, it fluttered transversely (in the x-direction in Fig. 1(d)) when it swung rapidly. To prevent this flutter, the flag arm was bent to have the x-direction curvature as an improvisational step, which then strengthened the arm’s rigidity in that direction.

The shutter speeds of the type A and type B shutters were studied using the mechanics of angular motion. The MI was calculated for all moving parts, including the flag, the Teflon rotator, and the bolt, shown in Figs. 1(b) and 1(c), against the rotating axis using Solidworks^® software and summed to provide the combined moment of inertia (CMI), which was listed in Table I with the other physical quantities of the shutters. While the Teflon rotator was the heaviest part, the CMI was dominated by the MI of the flag because its MI is proportional to the square of the distance from the rotating axis.

The angular acceleration of the flag can be determined by dividing the CMI by the torque on the axis of rotation. The torque is obtained as a cross-product of two vectors: the force vector F of the bender PZT when it is pushing the lever and the length vector L of the lever arm that connects the axis to the point at which the force is applied (Fig. 2(b)). The amplitude of the pushing force F is 0.5 N, as specified by the blocking force term provided by the manufacturer,⁸ and the length L of the shorter lever arm was approximately 5 mm, although it had different values that corresponded to the PZT insertion depth into the slot. Accordingly, the shutter time, which is the time required for the head of the flag to be displaced by 10 mm when accelerated by the PZT, was estimated to be roughly 6–8 ms by assuming that F is always orthogonal to L. These values are given in Table I for both the type A and type B shutters.

The mechanical characteristics, resonant frequency, shutter time, and stability of the type B shutter were measured experimentally in an atmospheric environment. These measurements were performed for the type B shutter only because the type A shutter was already installed in the experimental setup. However, the mechanical characteristics of the type A shutter are believed to be comparable to those of the type B because they use the same mechanism to actuate the flags and have similar CMI values for their moving parts, as shown in Table I.

The mechanical characteristics of the type B shutter were measured using a laser and a photodetector. As shown in Fig. 3(a), a laser beam was expanded until its spot size was wider than the diameter of the aperture to allow it to fill the entire aperture area uniformly, and it was focused on a photodetector such that the flag motion could be monitored when the flag passed across the aperture. The dependencies of the amplitude and the phase delay of the flag on the PZT-driving frequency were measured to determine the resonance frequency of the atom shutter. The PZT-driving power was given a sine-wave form using a function generator and a power amplifier. The flag displacement was set such that it was always smaller than the aperture diameter, even under resonance conditions. The black empty squares shown in Fig. 3(b) are the flag amplitudes, and the red filled circles denote the associated phase delays. In the figure, the amplitude and the phase delay are shown to increase slowly with increasing driving frequency up to 35 Hz. These values then increase more steeply at frequencies over 35 Hz and finally show the resonant behavior in the 45–55 Hz range. At a frequency of 65 Hz, the flag motion was almost frozen, which meant that the phase delay could not be obtained correctly. Consequently, these measurements indicated that the proposed atom shutter can be used reliably at operating frequencies of less than 35 Hz.

Measurements of the response times of the type B shutter were performed using a custom-made PZT driver with two output voltage states that were switched using an external TTL signal, as indicated in Fig. 1(a). The 10-mm-width flag head was set such that it was centered with the 5-mm-diameter aperture in the open state and was set to swing upwards by 10 mm to close the aperture, as shown in the inset images in Fig. 3(a). Because the flag head diameter is greater than that of the aperture, there are dark times, denoted by T₁ for the closing motion and T₃ for the opening motion, as shown in Fig. 3(c), and thus the photodetector cannot sense the flag motion during these times. After the dark times elapse, the photodetector can sense the flag’s motion as it sweeps across the aperture based on the time variations in the laser power, which are called the aperture sweep times, and are denoted by T₂ for the closing motion and T₄ for the opening motion in Fig. 3(c).

As shown in Fig. 3(c), T₁ for the closing motion was 2.9 ms and T₃ for the opening motion was 4.0 ms. This difference in the dark times originates from an asymmetric offset position of the flag. The difference between the dark times induced a difference between the aperture sweep times, where T₂ = 10.1 ms for the closing motion and T₄ = 9.3 ms for the opening motion, because the flag accelerated for a longer time before it started to sweep the aperture with the faster velocity. Therefore, the sums of the dark times and the aperture sweep times, i.e., T₁ + T₂ = 13.0 ms and T₃ + T₄ = 13.3 ms, have similar values for the opening and closing motions. Accordingly, the medium value (13.2 ms) of the two can be used as the experimental shutter time of the atom shutter, which is twice as large as the theoretical shutter time (7.5 ms) given in Table I. This discrepancy can be explained based on the RC time constant of the PZT, which has a capacitance of 4 $μ F$⁠, the time-varying angles of the force vectors, and the mechanical flexibility of the Teflon slot, which serves as a cushion against abrupt motion.

The timing jitter (i.e., the short-term stability) and the drift (i.e., the long-term stability) of the shutter time are also important factors for many experiments. To evaluate these values, the time intervals from the times when the TTL triggered the PZT driver to the times when the photodetector signal reached half of its maximum were measured using a time interval counter. The measurements were performed every 1 s for 42 h using two time interval counters, where one counter measured the opening motion of the flag, which swung at 1 Hz, while the other monitored the closing motion of the flag (Fig. 3(d)). The measured time interval for each motion was analyzed in terms of the Allan time deviation,¹¹ which describes the short-term and long-term stabilities together. As shown in Fig. 3(e), the short-term stabilities for both the opening and closing motions of the flag were as low as 0.03 ms at 1 s, which represents the standard deviation of the time interval jitter for each cycle. The long-term stability levels were also as low as 0.02 ms after an average time of ∼5.5 h (20 000 cycles). Additionally, the timing jitter and drift of the PZT driver output versus the TTL signal were measured to determine the origin of the drift in the shutter time (indicated by the black empty squares in Fig. 3(e)). As shown in the figure, these figures were negligibly small when compared with the corresponding figures for the flag motion, which indicates that the shutter time stability behavior was mainly caused by engineering factors. Consequently, the time response properties obtained for the atom shutter are satisfactory for OLC experiments, which tolerate stabilities of several ms, and for many other experiments in atomic physics that require tighter stability levels.

The UHV compatibility and lifetime of the implemented atom shutters were tested. We also provide miscellaneous comments about the mechanical vibration behavior and the possible contamination on the apparatus from the activated shutter.

The type A shutter was installed in an UHV chamber of the OLC experiment (Fig. 4(a)). The chamber was baked gently (100 °C) for 7 days while being pumped using a turbo-molecular pump. After the completion of the baking process, the pressure in the vacuum chamber reached 1 × 10⁻⁶ Pa, which was estimated based on a reading of the ion current of an ion pump that was installed adjacent to the atom shutter, as shown in Fig. 4(a), and then slowly decreased to 3 × 10⁻⁷ Pa after several weeks of pumping with the ion pump alone at room temperature. We checked any possible increment in the ion current produced by outgassing from the atom shutter when it was activated. However, no such increment occurred, which means that the outgassing is small enough to be detected at this vacuum pressure level. Consequently, the implemented atom shutter was proven to be compatible with UHV conditions. UHV tests on the type B shutter have not been performed to date. However, the test results from the type A shutter will be similar to those from the type B shutter because they are composed of the same materials, use the same operating mechanism, and differ only in terms of their shapes.

The lifetime tests on the atom shutters were performed on both types of shutter. Based on the specifications provided by the manufacturer, the bender PZT lifetime exceeds 4 × 10⁹ cycles under 116 Hz sine-wave excitation and free load conditions.⁸ While the PZT lifetime is long enough for it to be applicable to many experiments, it can also be reduced when the PZT is loaded using the flag. Additionally, the overall atom shutter lifetime can be limited by ageing of the Teflon rotator caused by its frictional and repetitive bending motions. Therefore, the lifetime of the shutter must be confirmed experimentally.

To measure the lifetime of the type A shutter when installed in the vacuum chamber, a pulse counter was used to count the number of triggering TTL signals that were applied to the PZT driver input from the start of the experiment. The accumulated number of cycles during the OLC experiment over eight months reached 2.0 × 10⁶ cycles. Tests using larger cycle numbers were performed on the type B shutter in an atmospheric environment with a room temperature of 21 °C and humidity of 40%. The shutter was operated continuously with a displacement of 10 mm at a frequency of 5 Hz using the custom-made PZT driver. The flag swung for 2.6 × 10⁷ cycles over 60 days without any major problems such as short circuiting of the PZT, release of the moving parts, or deviation from the offset position. It is believed that the type A shutter can also reach at least the same number of cycles without any problems because its hinge is more robust than that of the type B shutter.

Quantitative measurements of the vibration formed by the shutter activation process were not performed at this stage. To provide readers with a rough inference of the vibration status, it is useful to comment on the status of an optical cavity that is locked to a laser frequency in our experimental setup. In our setup, the atom shutter and cavity mirrors, which have a finesse of 220 and a cavity linewidth of 1.75 MHz (full width at half maximum), are attached to the same vacuum chamber (Fig. 4(a)). Therefore, if a small amount of mechanical vibration produced by the activation of the atom shutter transfers to the main chamber, it can then be monitored based on the fluctuations in the error signal of the cavity lock-servo device. To attenuate the vibration that is transferred from the atom shutter, the main chamber was linked to a Zeeman slower using a bellows nipple that was as short as 7 cm; a rigid connection between the two parts was thus largely realized through an optical table. As a result, no fluctuations were found in the error signal during the activation of the atom shutter at up to 30 Hz. In general, the mechanical vibration of the shutter is proportional to the mass and speed of its moving parts; consequently, the significantly reduced mass (0.5–1 g) of the moving parts in the implemented atom shutter played an important role in reducing the vibration on the experimental setup of the shutter. If necessary, further reduction would also be possible through the use of a thinner flag and by smoothing the PZT driver output electronically.

The frictional motion of the proposed atom shutter can yield some unwanted particles and cause outgassing. While the UHV compatibility of the shutter materials was proven under the pressure of 10⁻⁷ Pa in our tests, this can become a critical issue in certain experiments when using high-purity materials or high-quality optics. In experiments of this type, it is suggested that the implemented atom shutter should be positioned such that it is distant from the critical point of the experimental setup.

An ytterbium atomic beam that was directed towards a magneto-optical trap (MOT) was manipulated using the type A shutter. As shown in Fig. 4(a), the shutter was positioned between an atomic beam oven and a Zeeman slower, and its motion was monitored through a viewport. The shutter body was mounted on a linear motion feedthrough to align the centers of the shutter apertures such that they were in line-of-sight with the aperture of the atomic beam oven and the center of the MOT. Yb atoms were effused at a velocity of 300 m/s from the atomic beam oven, decelerated to a velocity of 20–30 m/s by the Zeeman slower and trapped in the MOT in the main chamber. The black line in Fig. 5(a) shows the time-varying number of Yb atoms that were trapped in the MOT when the atomic beam was switched on/off using the TTL signal (red line shown in Fig. 5(a)). As shown in Figs. 5(b) and 5(c), the atom numbers of the MOT begin to vary with a delay time of 20 ms (30 ms) during the opening motion (closing motion) controlled by the TTL signal. Given that it took 10–15 ms for the slowed atoms leaving the exit of the Zeeman slower to reach the MOT region after passing through the 30-cm-long space, the shutter time of the type A shutter was roughly estimated to be 5–15 ms (15–20 ms) for the opening motion (closing motion). The difference between the shutter times stemmed from the position offset of the shutter flag which was adjusted by OFFSET input of the PZT driver shown in Fig. 1(a). Therefore, the averaged shutter time of the type A shutter is 10–18 ms, which is comparable with that of the type B shutter at 13.2 ms. The atom shutter has functioned robustly without any problems like variation of the shutter time and deviation of the offset position for almost 1 year since it was installed in the UHV chamber.

We developed flag-type atom shutters that were driven using a bender PZT that offers negligible magnetic field induction and performed tests of the applicability of these shutters to experiments in atomic physics. The test results are summarized as follows. The UHV compatibility of the shutters was proven down to operating pressures of the low 10⁻⁷ Pa level. The shutter time was as short as 13 ms (10–18 ms) for the type B (type A) shutters. The shutter time stability was 0.03 ms for each cycle and 0.02 ms over a 20 000 cycle average for the type B shutter. However, because the stability of the type A shutter time has not been confirmed experimentally to date, we recommend the use of the type B shutter more strongly for experiments where the shutter time stability is an important issue. The type B (type A) shutter lifetime was experimentally proved to be longer than 2.6 × 10⁷ cycles (2.0 × 10⁶ cycles) in an atmospheric (vacuum) environment. However it is believed that the lifetimes of both types of shutter will be longer than 2.6 × 10⁷ cycles in the vacuum condition. Mechanical vibrations and impulses were blocked with minimal effort by virtue of use of the Teflon rotator and the significantly low mass of the moving parts. In conclusion, the overall performance levels of the proposed atom shutters were satisfactory for use in the required experimental conditions for many atomic physic research areas that use atomic beams, as well as that of an optical lattice clock.

This work was supported by the Korea Research Institute of Standards and Science under the project entitled Research on Time and Space Measurements through Grant No. 16011007.