1 Introduction

Several kinds of exotic systems composed of three particles with masses of the same order of magnitude and bound via Coulomb interaction, \( p^{ + } \mu^{ - } p^{ + } \), \( d^{ + } \mu^{ - } d^{ + } \), \( p^{ + } \mu^{ - } d^{ + } \) and the positronium negative ion (e− e+ e−, Ps−), have been studied theoretically for many years [1,2,3]. The stability of Ps− was first discussed by Wheeler in 1946 [1]. He found out that this system has ground-state energy of \( - 6.96\;{\text{eV}} \) and a mean lifetime of the order of 0.1 ns. After this work, a number of theoretical studies on Ps− were performed. For example, the energy and the annihilation rate in vacuum were calculated precisely [1,2,3,4,5,6,7,8]. The resonance states and the photodetachment cross sections of Ps− were also studied [9,10,11,12,13].

The first observation of Ps− was made by Mills in 1981 [14]. Following this observation, a few groups determined its decay rate [15,16,17]. In 2008, an efficient production method for Ps− was developed and new experimental investigations have been accomplished for these ten years [18, 19].

This paper reviews the developments of these experimental investigations.

2 Properties of Ps−

Table 1.1 summarises the properties of Ps−. Those of positronium (e− e+, Ps) are also shown. The Ps− properties have been studied only theoretically except for its decay rate. The decay rate has also been obtained experimentally [15,16,17] and compared theoretical values [8].

Table 1.1 Properties of Ps− and Ps
Full size table

3 First Observation of Ps− and Measurements of Its Decay Rate

The first observation of Ps− was made by Mills [14] in 1981. Slow positrons with energy of 470 eV, produced using a 58Co source and a Cu(111)+S moderator, were transported along a magnetic field to a 3.7 nm thick carbon film. A grid, positively biased at 0.5–4.5 kV, was placed 2.5 mm from the film so that any Ps− ions emitted from the downstream side of the film were accelerated. The Ps− produced self-annihilated, with its short lifetime, within a few mm from the grid. A Ge detector monitored the annihilation and blue-shifted γ-rays from the on-coming Ps−. Small bumps on the tails of the γ-ray peaks due to pair annihilation of positrons indicated the emission of Ps− from the carbon film. The Ps− formation efficiency was as low as 0.028%. In his paper, the production of high-velocity Ps by the photodetachment of accelerated Ps− was discussed.

Mills also measured the decay rate of Ps− by changing the acceleration voltages and the distance between the carbon film and the acceleration grid [15]. The obtained decay rate was 2.09(9) ns−1. Its precision has been improved using a stripping-based detection technique, which was originally developed by Mills et al. [22], to 2.086(6) ns−1 [16]. This technique was later combined with an intense slow positron beam from a research reactor in Munich, to give a more precise decay rate value of 2.0875(50) ns−1 [17]. All these values agree with the theoretical values [7, 8].

4 Efficient Production of Ps−

When low-energy positrons impinge on metal surfaces, they are thermalized in bulk and a significant fraction of them diffuse back to the surface. The positrons may be spontaneously emitted if the positron work function \( \phi_{ + } \) is negative. They may also be emitted as Ps. The energy required to emit Ps is written as \( \phi_{\text{Ps}} = \phi_{ + } + \phi_{ - } - 6.8\;{\text{eV}} \), where \( \phi_{ - } \) is the electron work function. The values of \( \phi_{\text{Ps}} \) for most metals are negative; hence, Ps atoms are spontaneously emitted. The energy required to emit Ps− can be written as \( \phi_{{{\text{Ps}}^{ - } }} = \phi_{ + } + 2\phi_{ - } - 7.13\;{\text{eV}} \). For tungsten, this value is negative and spontaneous emission of Ps− is expected [23].

Figure 1.1a shows an experimental setup for the observation of Ps− emitted from a tungsten surface. Positrons with energy of 0.1 keV were guided along a magnetic field and impinged on the negatively biased target through a grounded grid. Emitted Ps− ions were accelerated and annihilated. The emitted \( \upgamma \)-rays were monitored using a Ge detector. Figure 1.1b shows the obtained \( \upgamma \)-ray energy spectra [18]. After annealing the tungsten, a small bump which indicates Ps− emission from the surface appeared. The Ps− ions emission efficiency for this experiment was 0.005%, which was lower than that obtained using a carbon film. However, after Cs deposition, the efficiency increased up to 1.25%.

Fig. 1.1
figure 1

Adapted from [18]

a Experimental setup for the observation of Ps− emission from a tungsten surface. b Obtained \( \upgamma \)-ray energy spectra.

Full size image

After the Cs coating, the electron and positron work functions can be written as \( \phi_{ - } - D \) and \( \phi_{ + } + D \), respectively, where D is the effect of Cs coating. The value of D is 3 eV for the Cs coating of 0.2–0.3 monolayer on tungsten. Accordingly the value of \( \phi_{{{\text{Ps}}^{ - } }} \) changes to \( \phi_{{{\text{Ps}}^{ - } }} - D \) due to Cs deposition. This means that Ps− emission from Cs coated surfaces is more favourable than from clean surfaces.

The Ps− emission efficiency decreased in only half a day because Cs is chemically reactive. We used Na, which is less chemically reactive than Cs, to obtain durability. While the durability of Ps− emission has been improved to a few days, the emission efficiency was still as high as that of Cs coated surfaces [24].

5 Observation of Ps− Photodetachment

The efficient formation of Ps− has enabled the observation of Ps− photodetachment [25]. Figure 1.3a shows an experimental setup conducted to observe this phenomenon. In the setup, pulsed slow positrons produced using linac in KEK were transported and impinged on Na-coated tungsten through grounded grids. The produced Ps− ions were accelerated and blue-shifted \( \upgamma \)-rays were monitored by two Ge detectors. The Ps− ions were irradiated with pulsed photons from Nd YAG laser synchronised with the linac.

The relative amount of para-Ps to ortho-Ps formed from the Ps− photodetachment was 1:3. Therefore, a fraction of the Ps− ions were converted to ortho-Ps and did not contribute to the Ps− Doppler-shifted peak when the photodetachment occurred. Figure 1.3b shows the obtained \( \upgamma \)-ray energy spectra. It is shown that the peak obtained from the annihilation of accelerated Ps− decreased by the laser irradiation. This indicates that the Ps− photodetachment occurred (Fig. 1.2).

Fig. 1.2
figure 2

Reprinted with permission from [25]. Copyright 2018 by the American Physical Society

a Experimental setup for the observation of Ps− photodetachment. b Obtained spectra using (a). Plots of the square (red) and triangle (blue) are the experimental data for laser on and off, respectively.

Full size image

6 Observation of Ps− Shape Resonance

The process of Ps− photodetachment can also be observed by detecting produced Ps atoms, which have almost the same velocity as photodetached Ps−, with a microchannel plate. This method provides clean data with low background [26].

Using this technique, we observed a peak due to the shape resonance of Ps− in a spectrum of photodetachment (Fig. 1.3) [27]. The resonant energy was 5.437(1) eV, which is consistent with the theoretical predictions [10, 11, 13]. Further experiments, e.g. observation of Ps− Feshbach resonance, may feasible using this method.

Fig. 1.3
figure 3

Adapted from [27]

The yield of Ps atoms produced in the Ps− photodetachment plotted against the photon energy.

Full size image

Moreover, the Ps atoms produced in the photodetachment of accelerated Ps− provided an energy tunable Ps beam [26, 28]. This beam can provide new information on the basic science of Ps and in material science.