Introduction

The Moon is an excellent location for the storage of interstellar dust. There have been no geological or biological activities for more than a billion years [1-5], except for some mild gardening processes such as micrometeorite bombardment [6]. The dust deposited on the surface of the Moon contains long-lived radionuclides such as ⁶⁰Fe (t_1/2=2.61±0.04 My [7, 8]), which are mainly produced in massive stars and ejected by supernova explosions [9-11] while cosmic rays produce a small amount [12]. Therefore, ⁶⁰Fe can provide evidence for tracing the passing of ejecta of nearby supernova events that have occurred within the last several million years. However, the ⁶⁰Fe/Fe ratio of lunar samples is approximately 10^-15 [12], which falls below the detection limit of most nuclide analytical methods. Accelerator mass spectrometry is the only method capable of detecting ⁶⁰Fe. This method was employed to determine the abundance of ⁶⁰Fe in deep-sea ferromanganese crusts [13, 14], marine sediments [15], Antarctic snow [16] and lunar soils brought back by the Apollo and Luna programs [12].

Chang’e-5 has completed China’s first sample-gathering lunar mission, acquiring scooped and drilled samples from the northeastern Oceanus Procellarum on the Moon at longitudes and latitudes of 51.916°W and 43.058°N. This latitude is considerably higher than that of earlier sample collection sites of Apollo and Luna, which ranged from -8.973°N to 26.133°N. Hence, the new samples from the Chang’e-5 mission may provide more information (such as lunar petrology and volcanism [17-20], lunar geochemistry [21-23], and lunar soil maturity [24]). Motivated by this goal, we stepped up the development of the AMS facility at the China Institute of Atomic Energy to detect ⁶⁰Fe in lunar samples.

AMS setup

The HI-13 tandem accelerator at CIAE was accepted from the HVEC in 1986 and commenced full operations in early 1988 [25, 26]. AMS measurements based on this accelerator began in 1989 [27]. Nuclides such as ¹⁰Be, ³²Si, ³⁶Cl, ⁴¹Ca, etc. have been measured using this facility [28-31].

A schematic diagram of the AMS setup is illustrated in Fig. 1. The injection system was specifically designed for AMS measurements, featuring an NEC multicathode source of negative ions by Cs sputtering (MC-SNICS), which can accommodate up to forty cathodes. Negative ion beams are first filtered using a 90° electrostatic analyzer and a 112° injection magnet. Retractable Faraday cups are placed after each magnet to measure the beam current. Two offset Faraday cups are installed at the focal plane of the injection magnet. A gaussmeter is mounted inside the injection magnet to ensure reproducibility.

Fig. 1Full-screen View

(Color online) Diagram of the HI-13 tandem accelerator AMS system at the CIAE, featuring a photo of the installed Wien filter, along with a quadrupole doublet placed in front of it. Only major parts are drawn and most of the beam guiding devices are not displayed. There is a retractable target holder at the center of the target chamber, which was installed with the Si₃N₄ degrader

The HI-13 tandem accelerator can reach an approximate 12.5 MV terminal voltage, and is capable of foil and gas stripping. Following the accelerator, an analyzing magnet and a switching magnet are present. The analyzing magnet has a mass energy product of 200 amu·MeV (mE/z²). The terminal component of the AMS beamline is a Q3D magnetic spectrograph that sequentially comprises a target chamber, a quadrupole, three dipoles, and a focal plane. A Si₃N₄ foil was installed in the target chamber as an energy degrader for additional isobar separation. The magnetic spectrograph has exhibited an energy resolution of $2\times {10}^{-4}$ and a dispersion along the focal plane, determined by the least squares fit, of $11.37(\text{cm}/1\%\text{Δ}P/P)$ [32]. A multi-anode gas ionization chamber (four anodes in this work) with an entrance window of 65 mm×40 mm was mounted at the center of the focal plane for particle detection. The $\text{Δ}E-\text{Q}3\text{D}$ detection method is developed with this system for isobar identification. Detailed descriptions of the $\text{Δ}E-\text{Q}3\text{D}$ method have been reported [33, 34].

Despite the high sensitivity of the described AMS system, the contaminants in the beam of rare ions of interest cannot be completely removed by the high-resolution magnetic systems of the tandem accelerator. This limitation can lead to detector saturation in certain cases, such as in ⁶⁰Fe measurements. To address this issue, a Wien filter was installed after the switching magnet to reduce the interfering beams entering the final detector along with the rare ions of interest. The Wien filter is discussed in detail in the subsequent section.

2.1

Beam purification with Wien filter

A Wien filter utilizes orthogonal electric and magnetic fields to selectively influence the ions within a beam. Only ions with a specific velocity pass unaffectedly, whereas ions with different velocities are deflected by the electromagnetic force and subsequently blocked. A Wien filter manufactured by Danfysik was installed to purify the beam before it entered the Q3D magnetic spectrometer. The maximum voltage of the Wien filter is ±60 kV and the maximum magnetic field is 0.3T. The Wien filter parameters are listed in Table 1. A quadrupole doublet was positioned in front of the Wien filter to focus the beam at the entrance of the Q3D magnetic spectrograph. A slit was added 2 m from the Wien filter exit to block the deflected and defocused interfering beams. A collimator with a diameter of 5 mm can also be used for this purpose.

Table 1Full-screen View

Parameters of the Wien filter

Parameter	Value
Max. magnetic field (T)	0.3
Effective magnetic length (mm)	1032
Polo gap (mm)	140
Max. electrical field (kV cm-1)	24
Effective electrical length (mm)	1080
Max. electrodes voltage (kV)	60
Electrode gap (mm)	50
Electrode width (mm)	80

Show more

A test experiment with an ⁵⁸Fe beam was performed to evaluate the performance of the Wien filter. During this experiment, the analyzing and switching magnets were optimized for the transmission of ⁵⁸Fe. A multi-anode gas ionization chamber was mounted on the focal plane of the Q3D for detection. The energy-loss spectra for the first anode $E_1$ versus the total energy $E_{\text{total}}$ are illustrated in Fig. 2 When the Wien filter was not activated, there was a significant presence of interfering beams, as illustrated in Fig. 2a Most contaminants were suppressed by the Wien filter when the parameters were optimized for ⁵⁸Fe as illustrated in Fig. 2b Small numbers of ions with the same m/q and E/q as the ions of interest also enter the detector. However, their intensities were significantly reduced, and they were sufficiently separated in the energy spectra. In this experiment, the magnetic field of the Q3D was optimized for the tail of the ⁵⁸Fe because of the high counting rate of the ⁵⁸Fe beam.

Fig. 2Full-screen View

(Color online) Two-dimensional spectra of $E_1$ versus $E_{\text{total}}$. $E_1$ is the energy loss of the first anode and $E_{\text{total}}$ is the total energy of the ions in the detector. (a) Spectrum measured without the Wien filter. (b) Spectrum measured with the Wien filter of ±50 kV voltage and corresponding magnetic field. The areas for ⁵⁸Fe and ⁵⁸Ni are marked by red circles

2.1.1

Isobar suppression

The abundance of ⁶⁰Fe in lunar samples is extremely low, with ⁶⁰Fe/Fe estimated at the level of 10^-15. Separating ⁶⁰Fe from ⁶⁰Ni using most electromagnetic devices is challenging due to the isobar ⁶⁰Ni having nearly the same mass. ⁶⁰Fe and ⁶⁰Ni have to be separated at high energies by the dE/dx method; however, excessive ⁶⁰Ni mixed in the ⁶⁰Fe beam would saturate the data acquisition system. Hence, ⁶⁰Ni must be reduced during each step of the AMS measurement.

First, to reduce the influence of the ⁶⁰Ni, the samples used for ⁶⁰Fe measurements were chemically treated to reduce Ni using a solvent extraction method and an anion-exchange step. Second, copper powder with a relatively high purity of 99.999% was mixed with the ⁶⁰Fe samples (in the form of Fe₂O₃ powder) in an approximately 1:1 weight ratio to increase the beam current at the ion source of the accelerator. Third, the ion extracted for ⁶⁰Fe measurements was ⁶⁰FeO^-, which produced a stronger beam current and higher ⁶⁰fe/⁶⁰Ni ratio than ⁶⁰Fe^- [35]. The holders of the samples in the ion source are composed of high-purity copper.

Although the aforementioned methods significantly reduced the nickel content, the remaining ⁶⁰Ni could not be separated by electromagnetic devices and remained beyond the capacity of the data acquisition system. Thus, aided by an energy degrader foil, the $\text{Δ}E-\text{Q3D}$ [33, 34] method was employed to separate ⁶⁰Ni from ⁶⁰Fe before entering the final detector, because the energies of ⁶⁰Fe and ⁶⁰Ni will be different after passing the foil. In this method, a highly homogeneous Si₃N₄ foil with a thickness of 1 μm was installed in the target chamber as an energy degrader. When ⁶⁰Fe and ⁶⁰Ni with energies of 130 MeV pass through the degrader, the energy difference is approximately 1 MeV, and the energy straggling is approximately 200 keV at FWHM. This difference is sufficient for the Q3D to separate ⁶⁰Ni from ⁶⁰Fe. Although several scattered ⁶⁰Ni ions entered the detector, the remaining intensities were low, ⁶⁰Fe and ⁶⁰Ni can be distinguished using a multi-anode gas ionization chamber.

2.1.2

Experimental procedure

The HI-13 tandem accelerator operated at a terminal voltage of 11 MV for the ⁶⁰Fe AMS measurements. Considering the stripping efficiency and beam energy, carbon foils with a thickness of 3μg/cm² and charge state of ${11}^{+}$ were selected. At this terminal voltage, the stripping efficiency is around $7\%$, and the beam energy is approximately 130 MeV.

As the ⁶⁰Fe ion flux cannot be measured using Faraday cups, it is necessary to simulate ⁶⁰Fe beam transport with another nearby nuclide. ⁵⁹Co was selected as a pilot beam instead of ⁶⁰Ni to avoid heavy contamination in subsequent measurements. The beam-guiding devices were optimized to maximize the overall transmission efficiency. The transmission efficiency is calculated as follows: $\eta =\frac{I_{\text{Q3D}}}{q\cdot I_{\text{InjSys}}}\text{,}$ (1) where $I_{\text{InjSys}}$ and $I_{\text{Q3D}}$ are the beam currents measured with Faraday cups before the accelerator and $\text{Δ}E-\text{Q3D}$ system, respectively, $q$ is the charge of the ions. The efficiency was approximately 2% in this experiment.

Initially, all the accelerator magnet parameters were optimized for the transmission of ⁵⁹CoO^- to ⁵⁹Co¹¹⁺. To calibrate the Wien filter, the ⁵⁹Co beam was first measured using a Faraday cup in the target chamber without the Wien filter. Subsequently, the Wien filter parameters were optimized for ⁵⁹Co to reproduce the beam current. In ⁶⁰Fe measurements, the parameters of the Wien filter were optimized based on ⁵⁹Co. The slit after the Wien filter and collimator in the target chamber was used to block deflected contamination ions. To optimize the ⁶⁰Fe beam, the parameters of the major magnets, including the injection magnet, the analyzing magnet, switching magnet, and the Wien filter, were adjusted based on calculations. These parameters were fine-tuned based on the counting rate of the ⁶⁰Fe.

After the Si₃N₄ degrader, ⁶⁰Fe and ⁶⁰Ni were separated using Q3D. As the counting rate of ⁶⁰Fe was extremely low, initially, the magnets of the Q3D were optimized on ⁶⁰Ni to calibrate the parameters. The Q3D magnet parameters were then scaled to detect ⁶⁰Fe from the set previously tuned using ⁶⁰Ni. The optimized experimental parameters are listed in Table 2.

Table 2Full-screen View

Optimized accelerator parameters for ⁶⁰Fe measurements

Parameter	Value	Parameter	Value
Isotope	60Fe	Material of cathode	Fe2O3 + Cu (1∶1 by wt.)
Negative ion	60FeO-	Terminal voltage	11 MV
Accelerator stripper	Carbon foil (3 μg/cm2)	Charge state/stripping yield	11/~7%
Wien filter voltage	±50 kV	Wien filter magnet	B=0.092 T
Degrader	Si3N4 (1 μm)	Detector medium gas	Isobutane (35 mbar)

Show more

Results

In these measurements, an ⁶⁰Fe blank sample was initially measured. Subsequently, a sample with an ⁶⁰Fe abundance of $1.125\times {10}^{-10}$ was used for testing. The results are presented in Fig. 3. $E_1$ and $E_3$ are the energy losses at the first and the third anode, respectively. ⁶⁰Fe is clearly distinguished from ⁶⁰Ni, as illustrated in Fig. 3b. In the spectra, a substantial amount of ⁶⁰Ni and a few other contaminants are present; however, they are far away from the region of ⁶⁰Fe and do not affect the identification. The sensitivity of AMS measurements, r, was calculated as $r=\frac{N_{\text{blank}}/Q^{\text{'}}}{N_{{}^{60}\text{Fe}}/Q}\cdot r_{\text{sample}}\text{,}$ (2) where $N_{\text{blank}}$ and $N_{{}^{60}\text{Fe}}$ are the event counts in the area for the ⁶⁰Fe of the blank sample and ⁶⁰Fe sample measurements, respectively. $Q^{\text{'}}$ and $Q$ are the numbers of ⁵⁸Fe¹⁶O^- collected at the injection system for the blank sample and the ⁶⁰Fe sample measurements, respectively; $r_{\text{sample}}$ is the abundance of the ⁶⁰Fe sample. In the measurements of the ⁶⁰Fe blank sample, data were accumulated for 5.8 h and the average current of ⁵⁸Fe¹⁶O^- was 40 enA. No ⁶⁰Fe events were detected in this area. 152 ⁶⁰Fe ion counts accumulated over 6 h, and the average current of ⁵⁸Fe¹⁶O^- was 2.2 enA. Consequently, the sensitivity of ⁶⁰Fe measurements in the test experiment was estimated to be better than $4.3\times {10}^{-14}$. The sensitivity could, in principle, be expected to be approximately $2.5\times {10}^{-15}$ when the data are accumulated for 100 h. This, with the use of a pristine set of ion source components in the ionizer region, is feasible for accumulation over multiple cathodes for future lunar sample measurements because the main contaminants are sufficiently separated, as illustrated in Fig. 3b.

Fig. 3Full-screen View

(Color online) Two-dimensional spectra of $E_1$ versus. (a) Spectrum measured with a blank sample. (b) Spectrum measured with the ⁶⁰Fe sample. It is noted that the pressure of the detector during the blank sample measurements was different from that of the ⁶⁰Fe sample measurements. Therefore, the energy ranges of Fig. 3a and b are not identical. The areas for ⁶⁰Fe and its isobar ⁶⁰Ni are marked by red circles

Conclusion and outlook

The AMS facility at the HI-13 tandem accelerator has been developed for several decades and includes an NEC multicathode source of negative ions by Cs sputtering, the $\text{Δ}E-\text{Q3D}$ isotope separation system, and a multi-anode gas ionization chamber. The sensitivity of AMS mainly depends on its ability to suppress contamination. Given that the previous system alone could not achieve the required sensitivity for detecting ⁶⁰Fe, a Wien filter was installed after the accelerator magnetic system to purify the beam and improve its sensitivity. The new setup was tested for ⁶⁰Fe measurements using a sample with ⁶⁰Fe/Fe at the level of $1.125\times {10}^{-10}$. The results demonstrated the following: Nearly all the contaminants in the beam of ⁶⁰Fe were effectively separated. The sensitivity of ⁶⁰Fe measurements with 5.8 h blank sample measurements at the AMS facility was evaluated to be better than $4.3\times {10}^{-14}$. For the lunar sample measurements, the duration would be approximately a few hundred hours. Thus, the sensitivity could in principle be expected to reach the level of ${10}^{-15}$. Furthermore, the ion source and transmission efficiencies of the tandem accelerator did not achieve the best performance during the test experiment. Therefore, the sensitivity could be enhanced through further improvements in the future.