The quest for AMS of 182Hf – why poor gas gives pure beams

The long-lived radioisotope 182Hf (T1/2 = 8.9 Ma) is of high astrophysical interest as its potential abundance in environmental archives would provide insight into recent r-process nucleosynthesis in the vicinity of our solar system. Despite substantial efforts, it could not be measured at natural abundances with conventional AMS so far due to strong isobaric interference from stable 182W. Equally important is an increase in ion source efficiency for the anions of interest.
The new Ion Laser InterAction Mass Spectrometry (ILIAMS) technique at VERA tackles the problem of elemental selectivity in AMS with a novel approach. It achieves near-complete suppression of isobar contaminants via selective laser photodetachment of decelerated anion beams in a gas-filled radio-frequency quadrupole (RFQ) ion cooler. The technique exploits differences in electron affinities (EA) within elemental or molecular isobaric systems neutralizing anions with EAs smaller than the photon energy. Alternatively, these differences in EA can also facilitate anion separation via chemical reactions with the buffer gas.
We present first results with this approach on AMS-detection of 182Hf. With He +O2 mixtures as buffer gas in the RFQ, suppression of 182WF5− vs 180HfF 5− by >105 has been demonstrated. Mass analysis of the ejected anion beam identified the formation of oxyfluorides as an important reaction channel. The overall Hf-detection efficiency at VERA presently is 1.4% and the W-corrected blank value is 182Hf/180Hf = (3.4 ± 2.1)×10−14. In addition, a survey of different sample materials for highest negative ion yields of HfF 5− with Cs-sputtering has been conducted.


Introduction
Accelerator mass spectrometry (AMS) commonly is the most sensitive technique for the detection of long-lived isotopes, reaching down to the attogram/gram abundance range. Over recent years, it has subsequently been employed in the search for signatures from recent nucleosynthesis in the vicinity of the solar system and corresponding signals of 60 Fe and 244 Pu have been identified in terrestrial and lunar archives [1][2][3][4][5]. The observed scarcity of 244 Pu has fueled the discussion about the astrophysical site of the rapid neutron-capture process (r-process) in favor of neutron star mergers [3]. Currently, this is the only experimentally confirmed r-process production site following gravitational wave detection GW170817 [6] and spectroscopic observation of the associated kilonova SSS17a (e.g. [7]). However, unexplained discrepancies with astronomical observations remain and suggest further sites and possibly multiple r-process components (see [8] for a recent review).
An important anchor point in this puzzle is 182 Hf with a half life of T 1/2 = (8.90 ± 0.09) Ma [9]. As is the case for 60 Fe and 244 Pu, there is no natural production of 182 Hf on Earth and all of the primordial contribution has long since decayed away. Thus, any 182 Hf detected must stem from ⇤ e-mail: martin.martschini@univie.ac.at recent nucleosynthesis. Since 182 Hf is in the middle-mass region of the r-process nuclides, it could potentially be produced in di↵erent scenarios to those for 244 Pu. Based on various yield-and elemental-ratio-calculations for possible 182 Hf production scenarios [10][11][12][13], the estimated 182 Hf/Hf signal intensity is at most a few times 10 13 , but could even be one to three orders of magnitude lower depending on the Hf incorporation e ciencies and stable Hf content in the archives.
The challenge in AMS-detection of 182 Hf at these low abundances is interference from the ubiquitous stable isobar 182 W. This quest started at the 3-MV-Vienna Environmental Research Accelerator (VERA) in 2002 [14]. It soon turned out that extraction of HfF 5 from a HfF 4 sample matrix provided a W-suppression > 10 3 , but no further separation could be achieved with subsequent AMS-filters or single ion detectors [15]. The resulting detection limit was roughly 182 Hf/Hf = 1⇥10 11 , but very susceptible to the W-content of the sputter matrix and the ion source materials [15,16]. The high mass of the isobars and their relatively low Z/Z complicate isobar separation by conventional means like di↵erences in energy loss characteristics or gas-filled-magnets [17]. Despite some progress with innovative detector concepts [18], no better detection limits have yet been achieved at larger tandem facilities [19]. A novel approach at VERA making use of di↵er- Figure 1. Ion intensities of 180 HfF 5 and 182 WF 5 transmitted through the ILIAMS cooler as a function of bu↵er gas pressure for various gases. The respective ion species were measured as 180 Hf 3+ and 182 W 3+ in a Faraday cup or an ionization chamber detector on the high-energy side of VERA. Plotted intensities are corrected for the e↵ective charge state yield of 20 % (including accelerator transmission). Lines between symbols are to guide the eye. ences in electron a nities (EAs) between the isobars may, however, allow us to reach the required sensitivity.

Setup and Electron Affinities
The Ion Laser InterAction Mass Spectrometry (ILIAMS) setup is the central part of a new injector at the VERA facility dedicated to element selective filtering by laser photodetachment; a recent description of the layout is given in [20]. It consists of a cesium sputter ion source followed by a 90 bending magnet and the ILIAMS radiofrequency quadrupole ion cooler, a 1 m long linear ion guide based on a 2D Paul trap, where electrostatically-slowed anions gently collide with He bu↵er gas and reach almost thermal energies resulting in transit times of several ms [21]. Inside the ion cooler, the ion beam is co-linearly overlapped with an intense cw-or quasi-cw laser beam. All anions with EAs smaller than the photon energy are thereby e ciently neutralized while anions with EAs larger than the photon energy remain una↵ected. This novel technique works extraordinarily well and provides isobar suppression factors of 10 11 and 10 12 for 36 S vs 36 Cl and 26 MgO vs 26 AlO , respectively [20,22].
For Hf, optical filtering has not been implemented so far since no experimental data for the EAs of HfF 5 and WF 5 exist. Theoretical calculations suggest that EAs are suitable and lie in the UV, with calculated vertical detachment energies of 3.9 eV for WF 5 and 8.8 eV for HfF 5 [23]. The only experimental anchor point so far is the finding that the detachment cross section of WF 5 is a factor of 100 higher than that of HfF 5 with a 266 nm laser (photon energy 4.66 eV) [24].

Gas reaction studies
Di↵erences in electronic structure of anions can also be exploited in anion-gas reactions at eV energies and O 2 was found particularly well suited for HfF 5 -WF 5 separation, providing an isobar suppression of at least 10 3 in an earlier study [25]. While the exact reaction channel remained elusive, Zhao et al. [25] argued that HfF 5 , being a superhalogen anion, is extraordinarily stable and, thus, highly immune to electron charge transfer reactions, collisional detachment and association or transformation of molecules at eV energies.
Similar experiments have now been conducted at IL-IAMS using gas mixtures of He+O 2 and He+CCl 2 F 2 . The respective bu↵er gases were volumetrically premixed in a 12 liter gas bottle. For each bu↵er gas, the system was first tuned for optimum 180 HfF 5 transmission at 0.20 mbar pressure at the bu↵er gas inlet and then the pressure was varied while measuring the respective ion species as 180 Hf 3+ and 182 W 3+ in a Faraday cup or an ionization chamber detector on the highenergy side of VERA (Fig. 1). Sputter targets contained mixed powders of HfF 4 +PbF 2 +Ag (1:1:1 by weight) and HfF 4 +W+PbF 2 +Ag (1:1:1:1 by weight), and the ion source produced typically 50 -300 nA of 180 HfF 5 (from either material) and 100 -500 pA of 182 WF 5 , respectively.
While 180 HfF 5 transmissions are almost identical for pure He and He+O 2 (30:1), the amount of 182 WF 5 transmitted through He+O 2 (30:1) is a factor of 10 6 lower compared to pure He. Accounting for the total ILIAMS transmission of 35% for 180 HfF 5 (cf. 3.2), this translates into an isobar suppression of 3⇥10 5 . Raising the O 2 content of the bu↵er gas to He+O 2 (10:1) provides stronger destruction of 182 WF 5 at lower bu↵er gas pressures but at higher pressures the intensity of 182 WF 5 reaches a plateau at al- most the same level as with He+O 2 (30:1). This yet unexplained plateau e↵ect has been observed in e.g. [26] as well and we attribute it to reverse reactions in the gas. With laser photodetachment, no such e↵ects have been observed so far. The transmission of 180 HfF 5 with He+O 2 (10:1) is significantly reduced to around half of that with pure He. Since these losses are independent of the bu↵er gas pressure and, thus, the ion transit time through the gas, they are most likely associated with collisional detachment on O 2 leaking into the electrostatic de-and acceleration areas between the cooler apertures and the first aperture lenses, where ions have hundreds of eV energy. The same is true for He+CCl 2 F 2 (30:1) with even slightly higher loss of 180 HfF 5 . The amount of transmitted 182 WF 5 is reduced by 10 3 , but also independent of bu↵er gas pressure, and hence not in anion-gas reactions at eV energies.
In order to pin down reaction channels and identify negatively-charged reaction products, the anion beam extracted from the ion cooler was mass-analyzed in the range of 250 -320 amu using VERA's injection magnet (Fig. 2).
With pure He, only the respective ion species 180 HfF 5 and 182 WF 5 are detected. Indeed, for an injection mass of 275 ( 180 HfF 5 ), the same is true for the other gases as well with no associated or transformed molecules being observed at any other mass. Significant oxyfluoride formation occurs when injecting 182 WF 5 through an O 2 +He mixture, with an intense peak at mass 293 corresponding to 182 WF 5 O being visible. With He+O 2 (100:1), the intensity of this peak explains at least half of the deficit in 182 WF 5 compared to pure He. This is in agreement with earlier studies [20]. A second peak at mass 290 is attributed to the formation of 182 WF 4 O 2 , albeit with lower intensity. Further peaks are observed at masses 271 and 295, however, their origin remains unsettled so far. The latter might indicate the formation of 182 WF 3 O 2 , but a very weak signal contradicting this explanation also seems to be present in the Hf-scans with He+O 2 . Higher oxygen contents in the bu↵er gas (He+O 2 10:1) produce the same peaks, although their intensity is significantly lower, thus other mechanisms like resonant charge transfer or collisional detachment, either directly with the analyte species or with formed molecules, are likely to reduce the anion intensity.
He+CCl 2 F 2 (30:1) does not yield any transmitted Wanions above the sensitivity limit of the Faraday cup (10 pA), thus, no significant association of molecules is observed, which is in agreement with above conclusions of pure collisional detachment. A mixture He+O 2 +CCl 2 F 2 (150:1:1) produced the same oxyfluoride peaks as He+O 2 , but at much lower intensity. In all cases no formation of WF 6 or WF 7 was observed. Whether the fluorine atoms are too strongly bound in CCl 2 F 2 to be available for reactions and other gases with weaker-bound fluorine potentially allow this reaction path remains to be investigated.

Sample preparation chemistry
Samples under investigation in this study were from commercial high-purity HfF 4 material (Alfa Aesar, purity >99.9%) or reference materials from earlier campaigns [15,16]. In parallel, a sample preparation procedure has been developed to allow 182 Hf analysis from neutronirradiated W. Basic steps are: 1.) Dissolving W in 2 ml HNO 3 (70%)/ 2.5 ml HF (48%; drop-by-drop; cooling); 2.) Addition of 1 mg natural Hf as solution; 3.) BaHfF 6 -precipitation by 0. 5   The ion exchange described in [28,29] has been further optimized for higher Hf-yield using inductively coupled plasma mass spectrometry (ICP-MS). A su cient decontamination factor of >10 8 could be reached with chemical yields of 91-100%, thus making the production of a 1 -2 mg HfF 4 -target with a W-content of <10 µg/g starting from a 1 g W matrix feasible. The use of Suprapur R chemicals has proven necessary as otherwise remaining W traces clearly originate from chemical products and consumables. Variants of this procedure will be applied for environmental samples in future studies. All HfF 4 materials for chemistry tests have been mixed (1:3 by weight) with PbF 2 (Alfa Aesar Puratronic R , >99.997% purity) and pressed in Cu cathodes; future samples will be additionally mixed with Ag for more stable sputter rates. Single BaHfF 6 precipitation yields ⇠10 4 as decontamination factor, which is not su cient in most cases. Furthermore, BaHfF 6 , a sputter material suggested by [19], was found to quickly poison the ionizer and the current output from the source dropped by at least a factor of 100 within a few minutes of sputtering.

AMS detection efficiency
As for any low-level AMS measurement, the overall detection e ciency of 182 Hf is a crucial parameter and highest losses generally occur in the production of negative ions in the Cs sputter ion source. Therefore, a study of the HfF 5 ionization yield with di↵erent admixing ratios of PbF 2 powder to the final sputter matrix was conducted.
Batches of sputter targets containing known amounts of typically 1 -4 mg HfF 4 were prepared and sputtered to exhaustion while collecting the 180 HfF 5 current in the Faraday cup right after the first bending magnet (Fig. 3). Of the three mixtures tested, the best HfF 5 yields are achieved with highest PbF 2 admixture, i.e. HfF 4 +Ag+PbF 2 1:1:3 by weight. Within the first two hours of sputtering, 2% of the sample material was turned into HfF 5 , peaking at finally about 6%. On the other hand, increasing the proportion of PbF 2 results in higher WF 5 formation from the target material, as demonstrated in [16]. The study here was conducted with ultrapure commercial HfF 4 and in first tests, no enhanced WF 5 content in the anion beam was observed (cf. 3.3). For real environmental samples with a slightly elevated W-content, it might however be necessary to determine the optimum PbF 2 mixing ratio as a trade-o↵ between high HfF 5 and still strong WF 5 suppression in the ion source. For these reasons, an ionization yield of only 2% achievable with low admixture (1:1) of PbF 2 is assumed for the considerations below.
The transmission of 180 HfF 5 through the ion cooler is 35% at an injected current of 200 nA. In contrast to e.g. Cl, where the cooler transmission is limited by the total charge limit of the RFQ ion guide [20], losses of HfF 5 are mostly caused by the large emittance of the ion beam from the ion source. The intense F current from the sample on the order of 50 µA blows up the size of the entire ion beam between the source and first mass-separation in the magnet. Therefore, even the mass-separated ion beam becomes too large to fit through the 3 mm entrance aperture of the ion cooler. Subsequently, this issue cannot be overcome by attenuation of the stable reference isotope beam but would require a redesign of the cooler injection optics, which is yet beyond our scope. In addition, around ⇠15% of the HfF 5 beam is lost in collisional detachment with O 2 leaking into the injection and extraction area of the cooler.
The tandem accelerator is operated at 2.4 MV with He as stripper gas providing an e↵ective 3+ charge state yield of 20 %. The resulting 8.8 MeV 182 Hf 3+ ion beam is directed without losses into a split anode ionization chamber with a 10⇥10 mm 2 large, 100 nm thick silicon nitride entrance window. Since no W isobar separation is achieved within the ionization chamber, virtually all events fall within the wide ROI of the 3+ charge state, thus, the detector e ciency is close to 100%. Stable 180 Hf 3+ is measured in the Faraday cup following the analyzing magnet.
The performance is summarized in Table 1 and results in an overall Hf detection e ciency of >1.4 %⇠. Assuming that at least 30 events need to be collected in

Results of first AMS measurements
First AMS measurements were conducted on a set of inhouse reference materials also employed in [16] with nominal ratios of 182 Hf/ 180 Hf = 5.59⇥10 10 (now labeled Vienna-Hf-10) and 182 Hf/ 180 Hf = 5.88⇥10 11 (Vienna-Hf-11) and blank material from commercial HfF 4 (Alfa Aesar). Very recent measurements suggest that both reference materials might in fact have (2.0±0.2)-times higher ratios and certainly require thorough remeasurement and cross-calibration in the very near future. For the following evaluation, the above nominal values were used. Hence, all measured Hf-ratios may require scaling by this factor once the proper isotopic ratios of these materials have been determined.  During the measurements, the ILIAMS cooler was operated at 0.30 mbar He+O 2 (30:1) bu↵er gas pressure. Sample spectra from the split anode ionization chamber are shown in Fig. 4. Measured 180 Hf 3+ -currents were typically 30 -40 nA. Since 182 Hf 3+ and 182 W 3+ cannot be distinguished in the ionization chamber, the amount of 182 W 3+ is monitored via separately counting 183 W 3+ and subsequent correction for the 182 W-contribution to the m=182 signal [15]. The 182 W/ 183 W seen by the detector is experimentally determined on samples contaminated with W on purpose at the 1000 µg/g level and was 2.43 ± 0.21 for the first measurement in comparison to the natural ratio of 1.85. Di↵erent masses are injected sequentially by adjusting the field of the ILIAMS bending magnet, the voltage of the insulated chamber of the injection magnet before the accelerator and the accelerator terminal voltage with switching times of ⇠3 s. Each run consists of four sequences for determination of the stable reference isotope currents in 5 s and in between 3 pairs of counting sequences for masses 182 (⇠150 s) and 183 (variable, here ⇠150 s). The system ran automatically and unattended for three days.
Final results normalized to the Vienna-Hf-10 reference material are plotted in Fig. 5. The reproducibility for the Vienna-Hf-11 material is better than 5% and in agreement with its nominal value. After correction of the W-induced background, the average blank value is 182 Hf/ 180 Hf = (3.4 ± 2.1)⇥10 14 . The upper limit translates into a present detection limit of 182 Hf/Hf ⇡ 6⇥10 14 , which