Open Access
Add to BibSonomy
Abstract
We report a highly efficient loading of 174Yb+ ions in a surface electrode ion trap by using single pulses from a Q-switched Nd:YAG laser to ablate neutral atoms, combined with a two-photon photo-ionization process. The method is three orders of magnitude faster to load a single ion as compared to traditional resistively heated sources and can load large collections of ions in seconds. The negligible thermal load of this method enables the use of this ablation-based loading scheme in ion traps operating under cryogenic conditions.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Typical ion trapping experiments, whether using macroscopic linear four-rod Paul traps [1] or surface electrode traps [2], utilize a variety of methods to generate atomic ions to be trapped. These methods include Joule-heated thermal sources to generate a sublimated stream of neutral atoms [3,4], as well as magneto-optically trapped (MOT) cold neutral atoms [5,6], which are subsequently photo-ionized and trapped. Other experiments have directly loaded ionized atoms from a laser ablated source into a linear [7–9] or a surface electrode [10] Paul trap. Isotope selectivity with ablation lasers using photo-ionization has also been demonstrated, though only in macroscopic linear traps [11–13]. While all these methods work well in certain use cases, they suffer from large heat loads (in the case of Joule heating), a lack of isotope selectivity (in the case of directly ionized laser ablation), or a lack scalability and finer ion chain manipulation (in the case of macroscopic ion traps). Additionally, in the case of thermal ovens the loading process is slow, being limited by the thermal time constant of the oven heating up, which typically takes on the order of a minute. A MOT-based loading scheme can be fast and thermally compatible with cryogenics, though system complexity is increased significantly due to the laser, electronic, and vacuum hardware requirements of the MOT.
Here we describe a method for efficiently trapping target isotope ions, and rapidly repeating the process to trap many ions in seconds. This rapid loading has obvious advantages when trying to re-load a lost ion, as well as quickly loading a chain of ions for quantum computation applications. Loading schemes based on laser ablation are desirable for their relatively modest experimental requirements while maintaining compatibility with cryogenic experiments and offering the potential for isotope selectivity.
2. Experimental details
A schematic of the experimental setup around the trap is shown in Fig. 1(a). The atomic source is a roughly 1 mm×1 mm×3 mm bar of 99.9% pure metallic Yb (natural isotope abundance), housed in a titanium enclosure located 1.1 cm from the trap, with a 500 µm diameter laser entrance hole for optical access, as shown in Fig. 1(c). The titanium enclosure is electrically grounded in order to provide a return path for any ablation-generated electrically charged particles. The sample is loaded from the backside, and then a titanium mounting rod is screwed in. After installation the Yb sample sits approximately 1.5 mm recessed from the laser entrance hole. This geometry was chosen in order to increase the directionality of any ablated material, and therefore minimize contamination of the trap. Incident on the metallic sample through the laser entrance hole is a focused Q-switched Nd:YAG pulsed laser operating at a wavelength of 1064 nm, with a pulse duration of $\approx 6$ ns at a variable 1-20 Hz repetition rate. The laser is focused to a beam waist (radius) of approximately 180 µm. This beam waist is small enough to allow the ablation laser beam to pass over the trap without damaging it, while also being large enought to address a sufficient surface area of the Yb source to generate a sufficiently large neutral Yb atom flux. Although the laser is capable of generating a pulse energy of several Joules, the maximum pulse energy used here is about 0.3 mJ, corresponding to a peak fluence (energy per unit area) of 0.6 J/cm2. A pulse incident on the atomic sample causes some atoms to be ablated from the surface and propagate out of the hole, towards the trapping region.
Fig. 1. (a) Schematic of the main experimental components. (b) Relevant energy levels of $^{174}$Yb$^+$ showing the cooling (370 nm) and repump (935 nm) lasers, and the branching ratios of the decay processes out of the excited states. (c) Cross-section schematic of the ablation oven. (d) Amount of fluorescence emitted by ablated material under 399 nm laser illumination, as a function of ablation laser fluence. (d, inset) Fluorescence emitted by ablated material as a function of the frequency offset of the 399 nm laser from the $^{174}$Yb isotope’s resonance. The different isotopes are labeled by atomic mass and the nuclear spin for odd isotopes).
The trapping region is the central section of a quadrupole RF surface electrode trap (Sandia National Laboratories HOA-2.0 [14]). A combination of radio frequency (RF) and direct current (DC) voltages is used to generate a quasi-static potential to hold the ions [1]. The RF signal source is a home-made board containing a direct digital synthesis (DDS) circuit, operating at 39.5 MHz. The signal is amplified and incident on a quarter-wave resonator [15] with a quality factor $Q \approx 90$, the output of which has an amplitude of $\sim$250 V on the trap’s RF electrodes.
Several lasers are needed to photo-ionize the neutral atoms present in the ablated plume of atoms, as well as to subsequently cool the trapped ions. Photo-ionization is accomplished by a resonant two-photon process. The first photon comes from a frequency-stabilized 399 nm laser (the neutral fluorescence laser) with up to $\sim$120 µW of optical power used to drive a resonant dipole transition, thus allowing for isotope selectivity. Figure 1(d, inset) shows the measured natural abundance neutral Yb spectrum in the region around the $^{174}$Yb resonance. A second laser at 391 nm drives the atoms from this excited state into the continuum, thereby ionizing the atom. The optimal (and longest) wavelength capable of exciting the atom to the continuum is 394 nm, but 391 nm is easier to combine with the fluorescence laser using a dichroic beam splitter, and it is only marginally ($\approx 2.5\%$) less effective [16]. It also means that it is not necessary to frequency-stabilize this laser. In order to maintain isotope-selectivity, both photo-ionization lasers are oriented near-perpendicular to the mean atomic velocity in order to minimize the effects of Doppler broadening that can cause different isotopes’ lines to overlap. Additionally, this allows for addressing of the entire atomic sample for ionization.
If the sum of an atom’s kinetic energy (from its velocity) and potential energy (from the trapping potential) at the point of photo-ionization is sufficiently low, the ion will be trapped. At this point a 370 nm laser addresses a cycling transition between the ion’s $^2S_{1/2}$ ground states and $^2P_{1/2}$ excited states used to Doppler cool the ion. A repump laser at 935 nm pumps the ion back into the cycling transition’s ground state if it decays to the long-lived $^2D_{3/2}$ state. A schematic showing the relevant energy levels is shown in Fig. 1(b).
A high numerical aperture (NA = 0.6) imaging lens collects 10% of the emitted light, either at 399 nm in the case of fluorescence, or at 370 nm in the case of a trapped ion. The collected light is spatially filtered by an iris to eliminate unwanted background scatter, and then incident on a photo-multiplier tube (PMT). The PMT signal, the RF signals controlling the AOM and trap RF voltages, and the trigger for the ablation laser are connected to an FPGA-based controller which allows for precise temporal control of the hardware, as well as recording time-of-arrival information of the photons incident on the PMT.
The number of fluorescence photons incident on the PMT, as a function of ablation laser fluence, is shown in Fig. 1(d). A threshold for fluorescence is found at a fluence of 0.2 J/cm2. Beyond this threshold the amount of fluorescence signal scales roughly linearly. However, while this data tells us the minimum ablation laser fluence needed to generate ablated atoms, it does not show whether the atoms are of sufficiently low energy to be trapped.
The ablated atoms’ velocity distribution can be obtained by looking at the times-of-arrival of the fluorescence photons. This data was collected in two different ways. The data presented here is collected by sending the PMT counts to the controller FPGA, which tags each photon with its arrival time $t$. There are two corrections that need to be made in order for the data to correctly represent the atomic velocity. First, the finite width of the fluorescence laser causes slower-moving atoms to contribute more fluorescence signal when compared to faster-moving atoms, as they spend a longer amount of time in the beam. Therefore a scaling factor $\propto 1/t$ needs to be applied to correct for this. Second, the FPGA has a finite rate at which it can record the arrival time of photons, which acts as a rate limiter on the measured photon collection rate. At any given time interval, the actual photon collection rate $R_a$, measured photon collection rate $R_m$, and FPGA recording rate $R_r$ are related by $1/R_m = 1/R_a + 1/R_r$. A histogram showing these corrected arrival times can then be plotted to see the photons’ time-of-arrival distribution, as shown in Fig. 2(a). Each curve is a histogram showing the arrival time of photons relative to the ablation pulse (bin width is 1 µs), and aggregated for 310 ablation pulses per fluence. Since the distance from the atomic source to the trapping region is $d= 1.1$ cm, this arrival time distribution can be easily converted to a velocity distribution, and the curves can then be fitted to a one-dimensional Maxwell-Boltzmann (thermal) distribution given by:
where $\bar {v} = \sqrt {2 k_b T/m}$ is the (temperature dependent) standard deviation of the velocity distribution, $k_b$ is the Boltzman constant, $m$ is the mass of the atom, $T$ is the effective temperature of the plume, and $v_s$ is the plume’s stream (center-of-mass) velocity. The data shown in Fig. 2(a) is converted to a velocity distribution and fitted using Eq. (1) to extract $T$ and $v_s$. As shown in Figs. 2(b) and 2(c), the plume’s temperature shows a roughly linear dependence on the fluence, while the stream velocity shows a much weaker dependence on the fluence. Also shown in Fig. 2(a) is the arrival time cut-off (shown as the vertical blue line) corresponding to the velocity below which the atoms have low enough kinetic energy to be captured in our trap. The atoms arriving at the trapping region after this cutoff time have sufficiently low kinetic energy and are available for trapping, and Fig. 2(a) indicates that higher fluence lead to a larger number of such atoms.Fig. 2. (a) Time-resolved fluorescence spectroscopy of the ablated atomic plume. The vertical line is the arrival time cut-off, above which the atoms are moving slowly enough to have a chance of being trapped, given the trap depth in our system. (b) The temperature of the atomic plume as a function of ablation laser fluence. (c) Stream (center-of-mass) velocity of the ablated plume.
A second method to characterize the velocity distribution is by using an acousto-optic modulator (AOM) to shutter the fluorescence laser for a short time window $\Delta t_w$ with a variable time delay $\Delta t_D$ relative to the trigger of the ablation laser pulse. Increasing $\Delta t_D$ ensures that only atoms with a sufficiently low velocity are in the trapping region while the fluorescence laser is on. This method has an advantage that the photons do not need to be time-tagged, and therefore the tagging rate is not an issue. However, because the light is only on for a small fraction of the time, accumulating sufficient statistics is much more time consuming. We primarily used this method to verify that the rescaling of the time-tagged data is justifiable.
3. Performance characterization
The fluorescence data shown in Fig. 2(a) can be used to estimate the number of trappable ions per ablation pulse. This estimate, which includes the overlap between the trapping volume and the atomic flux, the finite trap depth, and the speed of the atoms, is shown in Fig. 3(a). We expect about 4–10 trappable atoms per pulse at the typical fluence used for the ablation beam, and when combined with a photo-ionization laser beam width that allows several photon absorption-emission cycles during an atom’s path through the beam, this means we expect a trapping probability of near unity per pulse.
Fig. 3. (a) Number of trappable atoms per ablation pulse as a function of ablation laser fluence. (b) Probability histogram of the number of trapped ions per ablation attempt for 201 ablation pulses and an ablation laser fluence of 0.5 J/cm2.
The loading efficiency of $^{174}$Yb$^+$ ions into an empty (no ions present) trap was measured by sending an ablation pulse, probing the presence of a trapped ion by monitoring ion fluorescence (at 370 nm), and recording the number of ions each ablation pulse generates. This loading procedure is performed 201 times, and a histogram summarizing the loading efficiency is shown in Fig. 3(b). A single ablation pulse yields at least one trapped ion 85% of the time, with probability distribution being approximately geometric, as expected. The average number of trapping attempts to successfully load one ion was 1.17 pulses which, for an ablation repetition rate of 20 Hz, leads to a mean time-to-trap of 9 ms. This time is dominated by the repetition rate of the ablation laser, as the trigger delay for a single pulse is only 120 µs and represents an improvement over typical thermal sources by more than three orders of magnitude. Lasers with faster repetition rates (upwards of 10 kHz is widely available) could be used to reduce this time even further. This method enables on-demand re-loading of lost ions during experiments. Additionally, the energy deposited in the sample is only 0.25 mJ per ablation pulse, which is manageable even in cryogenic systems operating at 4 K.
Ions already present in the trap are typically not perturbed by the ablated plume of atoms passing through the trapping region. This enables us to load many ions into the trap simply by repeating the loading process many times. The loading efficiency is lower when ions are already present in the trap, though it is still possible to rapidly load many ions. We were able to load a collection of $\gtrsim 50$ ions in the span of 30 s, or at least 1.6 ions/s over a prolonged period, as shown in Fig. 4.
Fig. 4. Fluorescence counts from trapped ions as a function of time spent ablating. At low trapped ion numbers, a single ion generated about 30 detected photon counts per 1 ms of integration time (inset). This number decreases with increased numbers of trapped ions due to the Gaussian profile of the fluorescing laser. This sets a lower bound of 50 ions trapped within 30 s.
It should be noted that, unlike similar experiments, no noticeable reduction in the atomic flux was observed, even after many thousands of ablation pulses. Furthermore, no visible accumulation of material, Yb or other, has been observed to contaminate the trap surface over the course of several thousand loaded ions. The trap’s DC electrodes, which are separated from each other by only a few microns, have also shown no noticeable change in the resistance or capacitance to ground. This indicates that this method is at least moderately safe to use, even with the oven mounted within line-of-sight of the trapping region. Presumably, this is due to the very small amount of material that is ablated per pulse.
4. Conclusion
We have demonstrated fast ablation loading of trapped ions into a surface electrode trap, with both on-demand loading of a single ion as well as rapid loading of large numbers of trapped ions. The system is isotope-selective by tuning the frequency of the laser driving the first resonant stage of the photo-ionization process. The low energy required per ablation pulse means that this scheme is compatible with a cryogenic system which, when viewed in combination with the loading speed and efficiency in a surface electrode trap, demonstrate a scalable and flexible loading mechanism for quantum computing applications. The only drawback compared to thermal ovens is the initial experimental overhead and the necessary investment of the ablation laser and associated optics, though this investment is still significantly less than MOT-based ion sources. The low required pulse energy indicates that it may be possible to use solutions at yet a lower cost than the Q-switched Nd:YAG laser used here.
Funding
Intelligence Advanced Research Projects Activity (W911NF16-1-0082); National Science Foundation (EFMA-1741651).
Acknowledgments
The authors would like to thank S. Crain and B.-H. Ahn for discussions.
Disclosures
The authors declare no conflicts of interest.
References
1. W. Paul, “Electromagnetic traps for charged and neutral particles,” Rev. Mod. Phys. 62(3), 531–540 (1990). [CrossRef]
2. S. Seidelin, J. Chiaverini, R. Reichle, J. J. Bollinger, D. Leibfried, J. Britton, J. H. Wesenberg, R. B. Blakestad, R. J. Epstein, D. B. Hume, W. M. Itano, J. D. Jost, C. Langer, R. Ozeri, N. Shiga, and D. J. Wineland, “Microfabricated surface-electrode ion trap for scalable quantum information processing,” Phys. Rev. Lett. 96(25), 253003 (2006). [CrossRef]
3. N. Kjaergaard, L. Hornekaer, A. Thommesen, Z. Videsen, and M. Drewsen, “Isotope selective loading of an ion trap using resonance-enhanced two-photon ionization,” Appl. Phys. B: Lasers Opt. 71(2), 207–210 (2000). [CrossRef]
4. S. Gulde, D. Rotter, P. Barton, F. Schmidt-Kaler, R. Blatt, and W. Hogervorst, “Simple and efficient photo-ionization loading of ions for precision ion-trapping experiments,” Appl. Phys. B: Lasers Opt. 73(8), 861–863 (2001). [CrossRef]
5. M. Cetina, A. Grier, J. Campbell, I. Chuang, and V. Vuletić, “Bright source of cold ions for surface-electrode traps,” Phys. Rev. A 76(4), 041401 (2007). [CrossRef]
6. J. M. Sage, A. J. Kerman, and J. Chiaverini, “Loading of a surface-electrode ion trap from a remote, precooled source,” Phys. Rev. A 86(1), 013417 (2012). [CrossRef]
7. R. D. Knight, “Storage of ions from laser-produced plasmas,” Appl. Phys. Lett. 38(4), 221–223 (1981). [CrossRef]
8. S. Olmschenk and P. Becker, “Laser ablation production of ba, ca, dy, er, la, lu, and yb ions,” Appl. Phys. B: Lasers Opt. 123(4), 99 (2017). [CrossRef]
9. K. Zimmermann, M. V. Okhapkin, O. A. Herrera-Sancho, and E. Peik, “Laser ablation loading of a radiofrequency ion trap,” Appl. Phys. B: Lasers Opt. 107(4), 883–889 (2012). [CrossRef]
10. D. R. Leibrandt, R. J. Clark, J. Labaziewicz, P. Antohi, W. Bakr, K. R. Brown, and I. L. Chuang, “Laser ablation loading of a surface-electrode ion trap,” Phys. Rev. A 76(5), 055403 (2007). [CrossRef]
11. R. Hendricks, D. Grant, P. Herskind, A. Dantan, and M. Drewsen, “An all-optical ion-loading technique for scalable microtrap architectures,” Appl. Phys. B: Lasers Opt. 88(4), 507–513 (2007). [CrossRef]
12. K. Sheridan, W. Lange, and M. Keller, “All-optical ion generation for ion trap loading,” Appl. Phys. B: Lasers Opt. 104(4), 755–761 (2011). [CrossRef]
13. H. Shao, M. Wang, M. Zeng, H. Guan, and K. Gao, “Laser ablation and two-step photo-ionization for the generation of 40ca+,” J. Phys. Commun. 2(9), 095019 (2018). [CrossRef]
14. P. Maunz, “High optical access trap 2.0,” Documentation Manual (2016).
15. J. D. Siverns, L. R. Simkins, S. Weidt, and W. K. Hensinger, “On the application of radio frequency voltages to ion traps via helical resonators,” Appl. Phys. B: Lasers Opt. 107(4), 921–934 (2012). [CrossRef]
16. C. Blondel, R. J. Champeau, and C. Delsart, “Photoionisation laser spectroscopy of ytterbium atoms in the presence of a dc electric field,” J. Phys. B: At. Mol. Phys. 18(12), 2403–2422 (1985). [CrossRef]
