Excitation and emission spectra of Ba⁺ ions in solid xenon

While there have been many experimental and theoretical studies of the spectroscopy of atoms as impurities in solid noble gas matrices over more than half a century, the studies of the spectra of atomic ions in solid noble gas matrix have been few. To our knowledge, only two previous publications report experimental optical spectra for two ion species: Ca⁺ in solid argon (SAr) and Eu⁺ in SAr, SKr, and SXe, with the best data recorded in SXe.^1,2 Interest in the matrix isolation spectroscopy of Ba and Ba⁺ in solid Xenon (SXe) has increased in recent years accompanying work on the search in the EXO-200 and nEXO collaborations for neutrinoless double beta (0νββ) decay via the decay of ¹³⁶Xe into ¹³⁶Ba.^3–5 Observation of 0νββ decay will demonstrate the Majorana nature of neutrinos as well as a violation of the conservation of Lepton number, both of which are physics beyond the Standard Model.⁶ 

To date, 0νββ decay has not been discovered experimentally despite half-life sensitivity in the 10²⁶ range.^7–11 Reach the next generation goal of the 10²⁸ year half-life sensitivity necessitates kiloton scale isotopic mass, high detection efficiency, and supremely low background. The nEXO experiment envisions a TPC filled with 5000 kg liquid Xe (LXe). Although not in the current baseline plan, effective isolation and identification of the ¹³⁶Ba daughter of ¹³⁶Xe 0νββ decay via a technique named Ba tagging¹² could provide background discrimination against all events not from double beta decay and increase the sensitivity by a factor of 2–3.¹³ With the successful implementation of a Ba tagging scheme, the only source of background would then be the ordinary 2νββ decay, observed in ¹³⁶Xe by EXO-200.¹⁴ 

The goal for the next generation is a half-life sensitivity of 10³⁰ years to cover essentially the entire (0νββ) decay phase space for the normal hierarchy of neutrino masses. This would require a nEXO-like LXe detector with 300 ktonnes of enriched LXe.¹⁵ An alternative suggestion to reach 10²⁹ year half-life sensitivity is a large DUNE-type LAr detector with a modest percentage (e.g., 2%) of enriched xenon.¹⁶ In either case, the dominant background competing with detecting 0νββ decay is solar neutrino scattering, which is an irreducible background except with daughter identification. Thus, efficient Ba tagging could be even more advantageous in this long-term scenario.

We are working towards a Ba tagging scheme that involves freezing the Ba daughter within the surrounding LXe onto a cryoprobe at the decay site. Then, the trapped ion or neutralized atom is identified spectroscopically within the frozen sample in a separate chamber. Although the immediate product of the 0νββ decay will be one optically inaccessible Ba⁺⁺ ion, it will quickly recombine to form one Ba⁺ ion in the LXe with visible wavelength transitions.¹² It is possible that through additional recombination with the electron cloud, as it is drawn away, a Ba⁺ ion or a neutral Ba atom will be present in the LXe detector after decay, perhaps with a probability ratio of 76% to 24%, respectively.¹⁷ Previous work has demonstrated the ability to image single neutral Ba atoms in two different types of sites within a SXe matrix.^4,5 This proves the basic concept of the Ba tagging scheme, in which one can count one Ba atom in the matrix. However, a successful Ba tagging scheme should be able to identify the daughter Ba regardless of how it appears in the Xe; as such, it is also of interest to develop a scheme to image single Ba⁺ ions trapped in SXe. This paper presents the excitation and emission spectra of the Ba⁺ ion for the first time in preparation for a single Ba⁺ ion imaging experiment. It also compares these excitation and emission spectra of Ba⁺ in SXe to earlier results obtained for Ba⁺ in LXe. This comparison adds credence to the original Ba tagging suggestion by Moe¹² that Ba⁺ ions can be observed by laser-induced fluorescence in-situ in an LXe detector.

Theory and experiment have found that the neutral Ba atoms exist in vacancies of different sizes in the solid noble gas matrices. Spectroscopic peaks associated with different sites and different atomic transitions can often be distinguished by a combination of correlations in excitation and emission spectra. For example, peaks with similar excitation spectra may suggest a common atomic excited state and a common matrix site. Theory, where it can be done, provides much help in identifying the sites associated with different spectral peaks. To date, no theoretical papers have been published on atomic ion spectra in solid xenon, but calculations of the molecular interaction potential curves of the lower levels of BaXe⁺ lay a solid foundation for progress in that area.^18,19 Since the equilibrium separation of the ground state Ba⁺–Xe is less than that of the Xe–Xe interaction, it is most likely that Ba⁺ ions would be found in single vacancy (SV) or in interstitial (IO_h) sites in SXe.

For optimal single atom or ion imaging in SXe, the atomic fluorescence quantum efficiency must be high, the background must be minimized to obtain adequate signal-to-noise, and the emission of one atom/ion must last long enough before terminating by photobleaching to get sufficient photons detected. These issues have been addressed in previous work on imaging of single neutral Ba atoms. For Ba atoms in both single vacancy (SV) and hexa-vacancy sites (HV), the quantum efficiency is high.^3,20 Background is mainly from the window surface, perhaps from residual gases deposited on the surface. In both cases, it was found that an order of magnitude reduction in this background could be obtained by “pre-bleaching” the window surface background, i.e., by repeatedly scanning a moderately focused medium power laser across the preselected imaging area before a deposit was made.^4,5,20 For Ba atoms in the SV site in SXe, photobleaching is weak, and a large number of emitted photons can be obtained from a single atom.⁴ On the other hand, for Ba atoms in the HV site, the challenge of stronger photobleaching had to be overcome to achieve single Ba atom imaging.⁵ In preparation for single Ba⁺ ion imaging, excitation and emission spectra of Ba⁺ ions are presented in this paper, the grouping of peaks possibly associated with a particular excited state and matrix site are sorted out, steps for overcoming window surface background are discussed, and the first investigations of photobleaching and temperature dependence of emissions are given.

Ba⁺ in SXe at low temperature (∼10 K) has three prominent emission peaks at 531, 553, and 635 nm.³ In that paper, excitation spectra for these peaks in the range of 460–490 nm were presented. A broad excitation peak at 473 nm was found for 531 and 635 nm emission. This set of data is reproduced on the left in Fig. 1 with a rescaling factor. Additional stepwise laser scans from 440 to 520 nm have revealed two additional excitation peaks of Ba⁺ in SXe at around 446 and 500 nm. Since these three data sets were taken on different Ba⁺ in SXe deposits separated by years, the relative signals were scaled using runs that sampled wavelengths near all three peaks on the same deposit. Uncertainties in relative heights of the excitation peaks are as large as a factor of 2 due to temperature and laser exposure differences between runs, as well as the possibility of subtle differences in deposit conditions.

The emission wavelengths have similar excitation peaks but different relative strengths. The emission at 531 nm can be excited in all three excitation peaks but is strongest with 446 nm excitation. The 553 nm emission peak is only excited significantly in the 446 nm peak. In contrast, the 635 nm emission peak is strongest with 500 nm excitation and to a lesser degree in the 473 nm peak. Naively, the three-peak excitation structure resembles the Jahn–Teller and spin-orbit splitting of a typical ²S–²P transition in a one-electron atom. However, the different heights for different wavelengths merit further experimental and theoretical exploration.

In vacuum, the 6 s ²S_1/2–6p²P_1/2, ²P_3/2 transition of Ba⁺ is separated into two strong lines at 493.4 and 455.4 nm, respectively, by the spin-orbit splitting, as seen on the right in Fig. 1. The emission peaks and their excitation spectra of Ba⁺ in SXe described above could be associated with these electronic transitions. In addition, weaker transitions from the two 6p states to metastable ²D_3/2 and ²D_5/2 states occur in vacuum at 649.7 nm (from 6p ²P_1/2) and 614.2 and 585.4 nm (from ²P_3/2) with a total branching ratio on the order of 1/4. Although these weaker lines may overlap for the two systems of Ba⁺ in SXe corresponding to the two strongest peaks, perhaps they can be resolved by association with the strongest lines through excitation spectra as well as temperature and bleaching dependences.

Emission spectra for excitation wavelengths between 450 and 460 nm are shown in Figs. 2(a)–(c). The first two spectra in (a) with 450-nm excitation on a fresh in SXe deposit containing Ba⁺ exhibit little difference in the large 531 nm peak, but about 40% loss in the 553 nm peak due to strong photobleaching. Weaker emission lines to the red are only partially resolved with components at 595, 611, 635, 651, and 670 nm. Of these weak peaks, only the 651 nm peak bleaches strongly like the 553 nm peak. In (b), the subsequent spectra taken in low to high excitation wavelength order, from 450–460 nm are shown. Both large peaks and the 595, 611, and 655 nm smaller peaks decrease with increasing excitation wavelength. The 635 and 670 nm components have less variation with laser wavelength. Thereafter, in (c), spectra at interlaced wavelengths were taken in reverse order. The strongest emission peaks at 531 and 553 nm still increase with lower laser wavelength as in (b), but the 553 and 655 nm peaks are significantly reduced due to bleaching.

Excitation spectra consisting of emission peak height vs. laser wavelength for Gaussian fits to the 531, 553, 585, and 611 nm peaks in (b) and (c) are shown in Fig. 2(d). The peaks at 595 and 611 nm have an excitation spectrum similar to that of the 531 nm peak. This similarity suggests that these could be emissions from the same excited 6p state of Ba⁺ ions in a common SXe matrix site. One possibility for the weaker peaks suggested by the transitions of Ba⁺ in vacuum, is emission to the metastable 5d states. By analogy, the existence of two weaker peaks could suggest a ²P_3/2 upper state. The branching ratios for the 531, 595, and 611 nm peaks given in Fig. 2(c) for the spectrum with 451 nm excitation are remarkably similar to the branching ratios from the ²P_3/2 in vacuum given in Fig. 1 (right).

To probe more carefully for peaks that are similar to the 553 nm large peak, a subtraction of spectrum 28 (laser 451 nm) from spectrum 2 (laser 450 nm) gives quite well the difference due to bleaching. As shown in Fig. 2(e), the 651 nm emission peak and another smaller 622 nm peak are well resolved by this method. The similarity of the 553 nm peak and these two peaks in relative wavelengths and peak heights to the 531 nm triad suggests similar atomic transitions but with a red shift due to a different matrix site. Assuming a common ²P_3/2 upper state, the branching ratios for these three peaks are given Fig. 2(e). Again, the similarity to branching ratios in vacuum is notable.

Further information can be obtained from spectra taken during the final heating of a sample that had previously been bleached. As seen on the left in Fig. 3, the 531 nm peak decreases with temperature while the 553 nm peak increases up to 40 K. A plot of the fit amplitude in these spectra vs. temperature for the 531 and 553 nm peaks is given on the right in Fig. 3. For the smaller peaks, the 595 and 611 nm peaks have behavior similar to the 531 nm peak, as in the excitation spectra. Structure around wavelengths of the 622 and 651 nm peaks identified previously increases with T up to around 40 K, similar to the 553 nm peak. These results confirm the groups identified above that may correspond to two matrix sites. There is also a new structure around 670 nm that becomes dominant at high temperature and increases up to 70 K.

The goal of this research is to find optimal conditions for the imaging of single Ba⁺ ions in SXe for our Ba tagging scheme. Important considerations are efficient excitation and emission, photobleaching, the amount of competing background in the emission band of interest, and the possibility of Ba⁺ ions existing in more than one SXe matrix site. Some of this information can be gleaned from the above results with large deposits of Ba⁺ in SXe. From Fig. 1, the largest rate of emission photons per laser photon (e.g., in counts/mWs) occurs for 446 nm excitation and detection of the 531 nm emission. On the left in Fig. 4 separate spectra of Ba and Ba⁺ taken on the same smaller deposit of Ba⁺ ions in SXe and normalized to laser exposure are compared. The peak fluorescence signal of Ba⁺ at 531 nm with 446.5 nm excitation is of the same order of magnitude as that of Ba at 619 with 570 nm excitation. The latter conditions produced clear single Ba peaks, and the fluorescence quantum efficiency was high, around 40%.^4,20 Since the relative population of neutral Ba atoms and Ba⁺ ions in different matrix sites has not yet been determined, this only indicates a favorable order of magnitude for the Ba⁺ excitation and emission for single ion imaging. Unfortunately, one can also see in Fig. 4 (left) that the background coming from the window surface is an order of magnitude higher for Ba⁺ imaging conditions than for Ba imaging. We have found that prebleaching the window by laser exposure, e.g., with 532 nm Nd:YAG laser light, before a deposit is made works similarly well for both Ba⁺ and Ba.

To overcome the remaining deficit, we have explored the use of thicker SXe deposits of order 100 μm rather than 0.5 μm. A raw CCD image of a ∼130 μm thick SXe deposit with Ba⁺ deposited near the top of the layer is shown in Fig. 4 (right). The blurry upper line in this image is fluorescence from the back window surface, which is broadened by laser defocus there. The brighter and sharper line below it is fluorescence from the front window surface, where the laser is more focused. The separation corresponds to the physical window thickness of 0.5 mm. The two stripes below are the fluorescence from Ba⁺ near the top of the SXe deposit at 531 and 553 nm. These peaks are well separated from the background on the window surface. A factor of an order of magnitude reduction in background is achieved.

A y-profile of a separate CCD image of a thick SXe-only deposit with no Ba⁺ is shown on the left in Fig. 5. The reduced background in the SXe layer is apparent. A hint of the source of the background is shown on the right. After deposit, the background at the SXe surface was very low. However, the SXe surface background increased roughly linearly with time. A possible explanation is condensation of residual gases in the vacuum chamber on the SXe surface.

Results for the photobleaching of the 531 and 553 nm emissions of Ba⁺ ions in SXe are shown in Fig. 6. On the left, the laser intensity incident on the window is 11 nW/μm², which is similar to that used in successful imaging single Ba atoms in a HV site of SXe.⁵ The 531 nm emission peak has very little decrease during the experiment with a total exposure of 2200 nWs/μm². In comparison, single atom imaging of Ba atoms in HV site with 577 nm emission was achieved despite a 50% loss due to bleaching at an exposure of 30–50 nWs/μm². Thus, in terms of photobleaching, the Ba⁺ emission at 531 nm is significantly more favorable than that of Ba in HV site atoms. Even for the 553 nm emission peak, the emission decreases to 50% at an exposure of 340 nWs/μm² that is still significantly greater than that for the single Ba imaging in Ref. 5. On the right is data taken at 12x higher intensity. At this intensity, both peaks bleach, with the 531 nm peak decreasing to 50% at about 3600 nWs/μm². Thus, although bleaching is not as favorable for imaging a single Ba⁺ ion in SXe as for single Ba atoms in a SV site in SXe,⁴ for which an order of magnitude higher intensity was used, it is significantly more favorable than for Ba atoms in a HV site.⁵ 

It is interesting to compare the excitation and emission spectra of Ba⁺ in SXe at low temperature to previous unpublished spectra of Ba⁺ in liquid xenon (LXe). For the best excitation spectra, the inverted electrode geometry was chosen with Ba⁺ ions created in a small plasma by laser ablation of a BaAl getter in liquid xenon. This produced 3 × larger charge pulse fluctuations (∼80% pulse-to-pulse standard deviation) but more consistent emission signals with 2 × better standard deviation (∼90%).

An excitation spectrum of average fluorescence counts per mW of laser power per average picocoulomb of detected ion charge for 10 ablation laser pulses with the different argon ion laser wavelengths is shown in Fig. 7. Data for an experiment with the colored glass filter OG 530 (∼50% transmisson at 530 nm) are shown in red dots. The OG 530 filter did not have sufficient attenuation at the 514 nm laser wavelength to obtain an emission spectrum. Two additional data sets with an OG 550 filter (∼50% transmisson at 550 nm) and more limited wavelengths but including 514 nm are shown in blue and green dots with ±1 nm displacement to distinguish the data. This spectrum has remarkable similarities to that of Ba⁺ in SXe at 12 K in Fig. 2(d). Two excitation peaks near the Ba⁺ in SXe peaks at 473 and 500 nm are evident. The lowest argon laser wavelength of 454 nm had lower laser power (15 mW) and larger fluorescence uncertainty, so comparison to the Ba⁺ in SXe peak at 446 nm cannot be made. At the time these data were taken, nothing was known about the spectrum of Ba⁺ in LXe or SXe. Part of the motivation for beginning our research on the spectrum of Ba and Ba⁺ in SXe was to verify that the observed fluorescence of Ba⁺ in LXe was indeed due to Ba⁺ ions. Some confirming evidence is now seen. More details and results are given in Refs. 21 and 22.

In separate experiments, Ba⁺ ions created by laser ablation of Ba metal in gaseous xenon above the liquid xenon were drawn into and through the LXe to a collection plate in the same manner as for mobility measurements.^21–23 Spectra for all the argon laser lines are shown in Fig. 8. The emission spectra of Ba⁺ in LXe have fluorescence peaks that are in the vicinity of the sharper peaks at 531 and 553 nm observed in SXe, but the fluorescence width is on the order of 100 nm. A tail to the red is reminiscent of the 635 nm peak of Ba⁺ in SXe. It is unknown whether this broadening is mainly an effect of higher temperature or is a function of the constantly changing environment around the ion in the liquid. The emission spectra of Ba⁺ in SXe up to 70 K do not show this amount of broadening.

These spectra represent those of the largest fluorescence signals in 4–14 ablation shots, normalized to charge and laser power. Due to the ∼200% pulse-to-pulse average standard deviation in emission amplitude, the relative heights of these spectra are not useful for determining an excitation spectrum. It was also observed in this configuration that the larger fluorescence signals were less frequent as the laser beam was moved away from the LXe surface. Thus, the spectra in Fig. 8 were taken with the laser ∼1 mm below the surface. The reason for this emission decrease with depth is not known. It is possible that a small amount of impurities, e.g., condensed water molecules, accumulates on the liquid xenon surface and could contribute to quenching of the Ba⁺ fluorescence if they are drawn into the LXe by the movement of the ion pulse. Nevertheless, it was also observed from LXe purity measurements by electron creation and transport across the electrodes that the concentration of electronegative impurities in the LXe does not change at the < 1 ppb level even after hundreds of laser ablation shots.²⁴ 

The additional evidence that these emissions are due to laser excitation of Ba⁺ ions in LXe confirms a basic feature of the Ba tagging premise of Moe in 1991¹² that the daughter ¹³⁶Ba⁺ ion of neutrinoless double beta decay could in principle be excited by a laser and its fluorescence detected efficiently in a liquid xenon detector. Unfortunately, the broad emission of Ba⁺ in LXe at higher temperature is not as desirable for discrimination against background as the sharper emissions of Ba⁺ in SXe at low temperature. As mentioned in Refs. 5 and 25, in-situ Ba tagging by the Moe scheme could work at lower temperature in a solid xenon detector. Another intriguing possibility is a liquid argon detector with 2% xenon that would operate closer to 90 K.¹⁶ In an operating LXe or SXe double beta decay detector, the possible issue of fluorescence quenching by impurity species may be negligibly small, as sub-ppt concentrations of electronegative species in the detector are routinely obtained by advanced purification methods to ensure long electron lifetime.

The apparatus used for spectroscopy of Ba⁺ in solid xenon (SXe) is essentially the same as that used for the imaging of neutral Ba atoms in previous work.^4,5 Samples were prepared on a cold sapphire window of 0.75 in diameter and 0.5 mm thick under ∼10⁻⁷ torr vacuum. Deposits were made by sending Xe gas (GXe) toward the sapphire window for 25–60 s via an inlet tube with a calibrated leak valve. The window was held at 50 K to minimize condensation of residual gases in the deposit. After the first ∼10 s of the xenon deposit, a Ba⁺ ion beam produced by a commercial Colutron ion gun was unblocked for a few seconds of nA level current or a specified number of 5 μs pulsed deposits at 1000 Hz. The GXe leak continued ∼10 s after the Ba⁺ deposit, which secured the Ba⁺ ions in the SXe matrix. Typical deposition thicknesses were 0.5 μm. Upon completion of the deposition, the Xe flow was stopped and the sample was cooled to 12 K for these studies. Laser excitation of the Ba⁺ ions in the matrix was done by one of five wavelength-selected blue diode lasers from 442 to 447 nm or by a Hubner C-Wave continuous wave optical parameter oscillator (OPO) with computer-selected wavelengths at 450 nm and above. The laser beams were coupled from the laser table to the ion beam table by a single mode optical fiber. The beam was expanded to a parallel beam by a collimating lens. Laser beam diameters on the sample were adjusted by longitudinal translation of the f = 7.8 cm aspherical laser focusing lens. The emission spectra were separated by an Acton SpectroPro 300i spectrograph and imaged in an Andor iXon3 CCD Camera cooled to –100 °C. Spectra were obtained from the CCD images by integrating a region in the y-direction of roughly constant laser intensity. The adjustable spectrometer input slit provided selection for constant intensity in the x-direction. A typical slit width was 40 μm, which corresponds to 3 pixels on the CCD camera. A short-pass interference filter was generally inserted in the laser beam to block stray long-wavelength components from the OPO or Raman scattering in the optical fiber. Tilted long-pass interference filters were used in the collection optics to block stray light at the laser wavelength and pass the desired Ba⁺ fluorescence peaks. The laser was blocked by an automated mechanical shutter synced to the CCD camera frames to avoid bleaching the sample between exposures.

Additional unpublished results from spectroscopy of Ba⁺ ions moving through liquid xenon (LXe) are also reported here. The liquid xenon cell, Xe gas system, electrodes and laser ablation apparatus is similar to that used for measurements of mobility of alkaline earth ions and is described in Refs. 21–23. For the excitation spectra reported, a pulse of Ba⁺ ions was created in a small plasma by laser ablation of a BaAl getter in liquid xenon. The electrodes were inverted from that used for ion mobility measurements. The Ba⁺ ions were drawn upwards through the LXe to a collection plate, where the ion current was measured. Half way along the path, in an 8 mm gap between electrodes, the Ba⁺ ions were excited by an unfocused argon ion laser, which has nine discrete wavelengths from 454 to 514 nm that can be selected individually. Fluorescence was imaged on a first aperture by a lens to block scatter from the electrodes, passed through an OG530 or OG550 long-pass optical glass filter to further reject scatter at the laser wavelength, focused again through a second aperture to discriminate against any fluorescence from the filter, and photons were counted in a photomultiplier tube. For recording emission spectra, Ba⁺ created by laser ablation of Ba metal in gaseous xenon above the liquid xenon was drawn into and through the LXe to a collection plate in the same manner as for mobility measurements.^21–23 Two optical improvements were implemented over the excitation spectra experiments. First, a 514 nm Raman edge interference filter was used to more sharply block scatter at the laser wavelength and to pass the Ba⁺ fluorescence. Second, the collected light was passed through a the Acton spectrometer to a Roper Spec-10:400B photon counting CCD camera that produced a spectrum for each pulse.

The basic spectroscopy of Ba⁺ ions in solid xenon has been presented for the first time. Excitation peaks at 446, 473, and 500 nm are found, which might be associated with a 6p triplet with Jahn–Teller and spin orbit splitting. Two large emission peaks at 531 and 553 nm with strongest laser excitation around 446 nm have been investigated, and two smaller peaks each associated with them are found at (595 nm, 611 nm) and (622 nm, 651 nm), respectively. A third strong emission peak at 635 nm is excited primarily in 473 and 500 nm excitation peaks. Conditions necessary for imaging of single Ba⁺ ions in SXe, such as strong emission, modest photobleaching and surface background at these wavelengths are discussed and found to be favorable when compared to neutral Ba if thick SXe deposits can be utilized to mitigate the background. This is an important step toward demonstration of the principles of a scheme for detecton and tagging daughters of ¹³⁶Xe neutrinoless double beta decay that exist in both the 0 and +1 charge state. Finally, agreement with excitation spectra for Ba⁺ in liquid helps confirm the assignment of such observed spectra. This gives credence to the original Ba tagging idea of Moe for reducing non-beta decay background in a liquid xenon time-projection chamber.

The authors thank Eli Bayarsaikhan and Cherie Bornhorst for their assistance with the experiments and John McCaffrey and Alexei Buchachenko for their helpful discussions of the energy levels of Ba⁺ ions in the noble gas matrix. This material is based upon work supported by the National Science Foundation under Grants No. 2011948 and 1649324 and by the Department of Energy under Grant No. DE-FG02-03ER41255. The authors acknowledge many discussions and the encouragement of the nEXO collaboration.

Joe Soderstrom and David Fairbank contributed equally to this work.