2. Slow beam production

Positrons are produced by a $^{22}\textrm {Na}$ source supplied from iThemba labs with the activity of 1.87 GBq in 2013. These fast positrons are slowed down using a neon moderator (Mills & Gullikson Reference Mills and Gullikson1986) grown onto a cone structure located directly in front of the source. The moderator arrangement was previously used at RIKEN (e.g. Oshima et al. Reference Oshima, Kojima, Dumitriu, Mohri, Oyama, Kambara, Kanai, Nakai, Wada and Yamazaki2000); however, the source was replaced upon its arrival at CERN.

Figure 1 shows a scale diagram of a slice through the slow beam production apparatus. Inside the vacuum system, biological shielding close to the source is provided by lead sealed inside aluminium pots. The narrow aperture (15 mm diameter and 120 mm long) also serves to provide a pumping restriction between the moderator chamber and the downstream vacuum. There is a retractable beam stopper mounted on a linear manipulator which is operated by an external pneumatic actuator to stop the beam. This removes any direct line of sight to the source during interventions. A large external shielding made of 316 stainless steel is mounted inside a ‘top hat’ style chamberFootnote ¹ outside of the vacuum which is also visible in figure 1 to the rear of the source. The vacuum chamber is surrounded by six water-cooled magnetic field coils (not shown in figure 1) which can produce fields of up to 500 G (at 60 A) around the source. Finally the system is surrounded by lead bricks to bring the surface dose rate to 1–3$~\mathrm {\mu }\textrm {Svh}^{-1}$ at full source activity (1.87 GBq).

Figure 1. Scale drawing of the apparatus used for slow beam production. The cone structure which freezes a solid Ne layer is located directly in front of the indicated position of the source.

The moderation efficiency ($\epsilon _m$) of the neon ice is

(2.1)\begin{equation} \epsilon_m = \frac{I^+}{A}, \end{equation}

where $I^+$ is the number of positrons in the slow beam per second and $A$ is the source activity in Bq. With a fresh uncontaminated moderator and a 1.87 GBq source the slow beam intensity is approximately $5 \times 10^6$ per second. This is measured using a small CsI detector from which the count rate is corrected for attenuation and solid angle (position indicated in figure 1). We have measured a moderation efficiency of $\epsilon _m = (0.25 \pm 0.1)$ %. A new moderator is grown once per day due to a rapid decay rate measured to be ($16.6 \pm 3.4$) %h$^{-1}$, this rate is due to contamination from buffer gases used in the trap discussed below.

3. Positron trapping

The slow beam produced by the rare gas moderator is magnetically guided from the source into the trap magnet (centre of figure 2). The trap is housed within the cold bore of a superconducting magnet which is energised to 6000 G, this value was found to produce the most stable operating conditions for the trap. The liquid helium reservoir (LHe) and the cold bore are in close contact, hence, it is not possible to run with a room temperature bore. As the bore will heat the reservoir this leads to an extremely high LHe consumption rate ($\sim$300 l per 24 h at room temperature). The temperature is lowered to around 100 K using LHe cooling loops at either end of the vacuum tube. The trap is in contact with the cold bore, although no special attention was given during design to ensure good thermal conduction, nor is the temperature monitored.

Figure 2. Scale drawing of the positron trap system.

The trap has a conventional gas cell structure consisting of a gate electrode (G1), a long narrow electrode (G2) with a high pressure of nitrogen buffer gas (see figure 3) and finally electrodes with a larger inner diameter and lower pressure (G3 and G4). Pumping holes were drilled into the ends of G2 and G3 to allow gas to escape, lowering the pressure more quickly in the subsequent sections. The vacuum tube is split into two parts by a divider after G4, this means that most of the trap gases are pumped away on the upstream side of the apparatus. Beyond G4 on the other side of the pumping divider, there is a multiringed electrode (MRE) trap consisting of 14 (M1–M14) electrodes of the same diameter (42 mm) but varying lengths (see figure 3, the narrowest electrodes are 10 mm wide). These create the final trapping potential well where positrons are stored before extraction to the Cusp trap. A typical trapping potential, and the locations of the electrodes which create it, is also shown in figure 3. The gas cell is constructed from aluminium, the MRE from oxygen-free copper coated with a gold layer. Each electrode has an RC filter mounted inside the vacuum system to reduce electronic noise (by ${\sim }100$ dB in the range 1–30 MHz) picked up from the local environment. Positrons were extracted from the trap using a switching box that sends a fast pulse (fall time ${<}20$ ns) to the confining electrode (M14) to open the well.

Figure 3. Electrical potential generated by the electrodes (left-hand $y$-axis) shown as a black line. Simulation of pressure distribution of nitrogen buffer gas (right-hand $y$-axis) shown as a red line made using Molflow+ (Kersevan & Pons Reference Kersevan and Pons2009). Also shown, a schematic diagram of the positron trap electrodes to indicate position.

A plot of positron number versus accumulation time is shown in figure 4. Results from the trap described above are shown as solid red points, results using the compact trap described by Imao et al. (Reference Imao, Michishio, Kanai, Kuroda, Enomoto, Higaki, Kira, Mohri, Torii and Nagata2010) are shown as solid blue points. For comparison, results from the compact trap, corrected by the different source activity are shown as blue hollow points. The present results show a trapped lifetime of 40 s and a peak trapping efficiency defined by the ratio of positrons entering the trap to those leaving of $(17.4 \pm 1.8)\,\%$. The lifetime was calculated by fitting an exponential rise to maximum of the form $N_{t}(t)=N_{t}^{\max }(1-\exp (-t/\tau ))$ where $N_{t}$ is the number trapped as a function of fill time $t$, $N_{t}^{\max }$ is the maximum number trapped and $\tau$ is the lifetime. The present trapped number is approximately 25 times higher than achieved with the compact trap. These improvements are a result of improved moderator efficiency (replacing W foil with Ne ice), and the much longer gas cell ensuring each positron interacts with the gas once along its length in a single pass.

Figure 4. Number of positrons versus accumulation time. The present results are shown in red, previous results using the compact trap (Fujii Reference Fujii2012) are shown in blue. The hollow blue points show the compact trap data normalised by the difference in source activity for direct comparison. Error bars shown on red points are smaller than the symbols, these show the standard error.

4. Positron accumulation

During typical operation, the positron trap is filled for 40 s collecting approximately $15 \times 10^6$ particles. Multiple bunches from the positron trap are accumulated in the Cusp trap to increase the number available for mixing. A scale drawing of the Cusp trap as used by Kolbinger et al. (Reference Kolbinger, Amsler, Cuendis, Breuker, Capon, Costantini, Dupré, Fleck and Gligorova2021) is shown in figure 5. Positrons are transferred via a magnetic conduit which is operated in a pulsed mode and has a magnetic field strength of up to 1000 G. Despite the 3 m length and two 90$^\circ$ turns, it is possible to achieve transport efficiencies of ($90\pm 5$) % which is determined by comparing the number leaving the trap to those arriving at the Cusp trap microchannel plate (MCP). As discussed above, the positron trap uses nitrogen buffer gas whereas the Cusp trap must have excellent vacuum for long antiproton lifetimes. To lower the contamination rate, gate valves along the transfer line are opened 5 s before the pulse is extracted from the trap and closed shortly after.

Figure 5. Scale drawing of the Cusp trap.

Positrons enter from the left-hand side of the arrangement and are stored in the upstream section of the MRE. The Cusp MRE shown here consists of 21 electrodes made of oxygen-free copper coated with gold, held in thermal contact with the cold bore using CuBe springs and mechanical wedges which can be tightened after the installation of the trap; the electrodes reach a temperature of approximately 15 K. Heat shields coated with a black nickel layer are located at either end of the trap to stop blackbody radiation from room temperature regions entering the MRE. An MCP and phosphor screen arrangement are located at the upstream entrance of the trap on a linear manipulator which can be used to image and count the number of particles extracted from the trap.

Figure 6 shows the magnetic field produced by the Cusp magnet (blue line upper plot), the potentials used during the catch and store operation, and the position of the MRE electrodes. For each subsequent stack of positrons, the catching open potential (green solid line) is applied. Once the stack is inside the trap, the catching potential is closed (black dashed line). While the next stack is prepared, the particles are moved to the storage potential (red line) and a rotating wall electric field is applied to the segmented electrode to compress the plasma. The timing for this catching operation is controlled by a SRS-DG645 unit (Stanford Research Systems) which allows nanosecond-level control over the triggering of the fast pulser to open the Cusp (fall time ${\sim }20$ ns). The open catching potential is applied for $t_{\rm open}\sim 900$ ns during transfer (this includes the transit time). The number of positrons accumulated versus filling time is shown in figure 7. Each stack takes approximately 50 s to fill the trap and perform transfer. The catching efficiency is typically ($50\pm 5$) % depending on the vacuum conditions within the Cusp trap. The reproducibility in the number of positrons is approximately 5 %.

Figure 6. Magnetic field produced by the coils of the Cusp magnet and the electrical potentials used when catching positrons from the trap. The location of the MRE electrodes are indicated the second and seventh electrodes are segmented.

Figure 7. The number of positrons stored in the Cusp trap as a function of filling time.

The optimum parameters for the rotating wall were found experimentally to be a sweep in the frequency from $f=7.0$ to $7.5$ MHz during the filling time and peak-to-peak voltage of 5 V. Using the MCP and phosphor screen arrangement on the upstream side of the Cusp, a diameter of 1.6 mm is measured when the plasma is extracted. During mixing cycles, approximately $30\times 10^6$ positrons were used, which corresponds to a density of $\sim 2\times 10 ^{14}$ m$^{-3}$, assuming a uniform magnetic field, and hence plasma radius, across the trapping region.