- Email: [email protected]
Electron cooling of D− at the ASTRID storage ring
Nuclear Instruments and Methods in Physics Research A 441 (2000) 150}153
Electron cooling of D~ at the ASTRID storage ring J.S. Nielsen*, S.P. M+ller!, L.H. Andersen", P. Balling", M.K. Raarup" !Institute for Storage Ring Facilities (ISA), University of As rhus, Ny Munkegade, Bygn 520, DK-8000 As rhus C, Denmark "Institute for Physics and Astronomy, University of As rhus, DK-8000 As rhus C, Denmark
Abstract A report of recent results on electron cooling of D~ at an energy of 1.6 MeV in the ASTRID storage ring is given. The longitudinal velocity spread has been reduced from &4]10~4 (FWHM) to &7]10~5 (FWHM) at a current of &0.1 lA. A drift in the mean velocity of the cooled beam has been reduced by application of a small RF signal on four sets of plates in the cooler. Initially, the velocity spread is found to decrease with ion current, indicating equilibrium between cooling and intra-beam scattering, whereas at later times (lower current) the velocity spread becomes constant, indicating equilibrium with the electron beam. To diagnose cooling, a simple system allowing to follow the frequency width and position of a Schottky harmonic on a sub-second time-scale, has been developed. The system uses a standard data acquisition card to digitize a down-mixed Schottky-signal and a FFT routine in Labview on a standard PC. The electron-cooled ion-beam is used for high-resolution vacuum ultra-violent spectroscopy of H~ and D~ in the region near the H(n"2) threshold. The velocity spread of the ion beam can be directly extracted from these experiments. ( 2000 Elsevier Science B.V. All rights reserved. PACS: 29.20.Dh; 29.27.-a; 41.75.Cn Keywords: Storage ring; Electron cooling; Negative ion-beam
1. Introduction The ASTRID storage ring [1] has now been in operation for more than 10 years, serving half the time as a 580 MeV Synchrotron Light Source, and half the time as a low-energy heavy-ion storage ring, primarily for atomic and molecular physics studies. The ring has from the start been equipped with an electron cooler [2], but apart from a single
test experiment with electron cooling of D` in the early 1990s [3] the electron cooler has only been used as a target for recombination studies [4]. However, recently an experiment performing VUV spectroscopy on H~ and D~ [5,6] has been limited in resolution by the Doppler broadening due to the velocity spread of the ion beam. An interest in cooling of H~ and/or D~ has therefore emerged. 2. Experimental setup
* Corresponding author. Tel.: #45-89422899; fax: #4586120740. E-mail address: [email protected] (J.S. Nielsen)
ASTRID consist of four pairs of dipole magnets, and four straight sections, each of 8 m length,
0168-9002/00/$ - see front matter ( 2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 9 0 0 2 ( 9 9 ) 0 1 1 2 5 - 0
J.S. Nielsen et al. / Nuclear Instruments and Methods in Physics Research A 441 (2000) 150}153
Fig. 1. Layout of the ASTRID storage ring together with the experimental setup.
151
current about 3.5 mA, and the expansion factor &6. The electron density in the cooler is about 4]106 cm~3. The details of the VUV spectroscopy experiment can be found in Ref. [5,6]. The experiment aims at observing and characterising the 1P0 dipole series of autodetaching resonances in H~ (and D~) located just below the H(n"2) threshold. In order to measure the photodetachment cross section, the ninth harmonic (118 nm) of a Nd:YAG laser is overlapped with the ion beam in one of the straight sections of ASTRID (see Fig. 1). As the wavelength of the laser light is "xed, the photon energy seen by the moving ions is controlled, via the Doppler shift, by tuning their velocity (i.e. tuning the energy of the ions). Neutral atoms produced by photodetachment or by collisional detachment in the rest-gas pass unde#ected through the magnetic lattice of the storage ring, and are recorded by a detector at the end of the straight section.
3. Cooling results giving a total circumference of 40 m (see Fig. 1). Ions are produced and accelerated to typically 150 keV in a small electrostatic accelerator before injected into the ring. In the ring the ions can then be accelerated, by drift tube acceleration [7], to a few MeV per nucleon, limited by the rigidity of 1.9 Tm of the dipole magnets. In the present experiment, D~-ions are produced in a duoplasmatron ion source, injected into the ring at 150 keV, and during 3.0 s accelerated to 1.6 MeV. After acceleration the beam is debunched and electron cooling is started. The ion-beam current after acceleration in the present experiment has been 0.1}0.2 lA. The lifetime of the ion beam is &2.5 s (calculated from the areas of Schottky peaks), and is determined by collisional detachment with the rest-gas of &3]10~11 mbar. The ASTRID electron cooler is located in one of the straight sections. It allows overlap with an electron beam with an energy of up to 2.0 keV over a length of 1 m. The electron beam can be adiabatically expanded by a factor of up to about 10. Typical electron currents are 2}5 mA. In the present experiment the electron energy is &450 eV, the
In Fig. 2 the result of electron cooling of a 1.6 MeV D~ beam is shown. As can be seen from the "gure initially there is a fast decrease in the velocity spread as the beam cools, and then a slow decrease, until a more or less constant minimum relative spread of &7]10~5 (FWHM) is obtained after about 3 s. The region with the slow decrease (&1.0 to &2.5 s) is interpreted as the result of an equilibrium between cooling and intra-beam scattering (IBS). As predicted by simple estimates [8], the decrease in the velocity spread is approximately proportional to N0.4, where N is the number of particles in the ring (N is closely following an exponential decay with a lifetime of &2.5 s). At the time where the minimum spread is obtained the number of particles in the ion beam is &1]107, for which Eq. (161) in Ref. [8] predicts a spread of 7]10~5, in agreement with our measurement. The minimum spread is interpreted as a temperature equilibrium between the ion beam and the electron beam (or very close to equilibrium). The longitudinal electron temperature can be extracted from the minimum spread to be 1 meV, which is close to the value expected from the adiabatic acceleration.
SECTION III.
152
J.S. Nielsen et al. / Nuclear Instruments and Methods in Physics Research A 441 (2000) 150}153
aim is to be able to fully quantify the properties of the M0N~ resonance (i.e. determine its width and 2 3 asymmetry). Furthermore, upper limits to the width of the second resonance ( M0N~) will be deter2 4 mined. It is expected to be very narrow (&2 leV) and hence serves as an independent measure of the ion-beam velocity spread. Finally, a search for the third resonance ( M0N~), which is predicted to exist 2 5 very close to the threshold, will be performed.
4. New Schottky data-taking system Fig. 2. The measured relative velocity spread (FWHM) as a function of time after acceleration has "nished and the beam has been debunched.
A crucial point in obtaining a stable mean longitudinal velocity of the cooled ion-beam has been the application of a small RF signal on electrostatic plates in the cooler, to &&shake-o!'' rest-gas ions trapped in the electron beam [9].1 The trapped ions change the potential in the electron beam, thereby changing the electron velocity and hence the cooled ion-beam velocity. The time scale for the ion production is long compared with the storage time, and a steady state is therefore never achieved. A constant velocity is, however, crucial for the experiment, due to the use of the Doppler tuning of the apparent laser frequency. The problem is cured by applying a sinusoidal RF signal with a frequency of 250 kHz and amplitude 3 V to one side of four sets of plates usually used as pickups for position measurements. The other side is grounded. This clearly reduces the neutralisation, and, more importantly, the time it takes to obtain a steady-state situation. With RF-shaking the drift after 2.0 s was below the velocity spread. The reduction in the velocity spread obtained by cooling has allowed the width of the "rst resonance ( M0N~) to be determined to 63 (13) leV [10]. Due 2 3 to time limitations and problems with the laser system, the uncertainty of the measurement is large. This will be improved in future beam times, and the
1 This cure was kindly suggested to us by Jacques Bosser, PS Division, CERN. See for instance Ref. [9]
To facilitate real-time diagnostics of cooling as well as detailed o!-line analysis, a simple and inexpensive system to analyse the Schottky signals [11] has been developed. Previously, the Schottky signals were analysed using a spectrum analyser (HP4195A). In free-running, however, this allowed only a few scans per injection at a fairly poor resolution. This could therefore only be used as a crude measure of the success of cooling. To obtain quantitative results the analyser had to be triggered allowing only one measurement at each injection. As the signal to noise in the Schottky signal is low many injections had to averaged, adding to the time consumption. The fundamental problem with the spectrum analyser is that it is Fourier limited on each point. A much better solution is to digitise the time signal and perform a Fast Fourier Transform to get the frequency spectra. In this way, the whole spectrum is obtained at once. To avoid the necessity of high sampling rates, mixing to an intermediate frequency is employed. This also drastically reduces the sample size (and thereby the computation time) for a given frequency resolution. In our setup the Schottky signal at about 4.5 MHz is mixed to about 18 kHz and low-pass "ltered before being digitised by a commercial PC data acquisition card (National Instruments PCI6023E) at 200 kS/s. Acquiring 32768 points gives a frequency resolution of 6 Hz and a new spectrum can be acquired every 0.16 s. On a modern PC (Pentium II 266 MHz) a FFT of 32768 points can, with commercial software (LabView), be done in about 100 ms, which means that the spectra can be recorded and analysed (position and width be extracted) real-time. A special LabView program has
J.S. Nielsen et al. / Nuclear Instruments and Methods in Physics Research A 441 (2000) 150}153
been written which reads the digitised signal, performs the FFT, and displays the Schottky spectra real-time. It furthermore allows averaging over a number of injections, after which the frequency spectra can be stored in a "le. A peak position and a peak width is extracted and displayed online from the averaged signals. The development time for the LabView application was about a week for two new LabView developers.
5. Conclusions We have described a recent experiment with electron cooling of D~ at the ASTRID storage ring. The relative longitudinal velocity spread has been reduced from &4]10~4 to &7]10~5 close to equilibrium with the electron beam. A crucial point for attainment of a stable ion-velocity is the application of a RF-signal to electrostatic plates in the cooler to &&shake-o!'' ions produced by ionisation of rest-gas atoms. The development of a much improved Schottky diagnostic system has been important for optimising electron cooling, in particular regarding the alignment of the ion and electron
153
beams. The experiment using the cold D~ beam has already bene"ted from the reduced velocity spread, and further progress is expected in the future.
References [1] S.P. M+ller, The Aarhus storage ring for ions and electrons ASTRID, Proceedings from 1993 Particle Accelerator Conference, Washington 1993, p. 1741. [2] L.H. Andersen, J. Bolko, P. Kvistgaard, Phys. Rev. A 41 (1990) 1293. [3] H.T. Schmidt, Ph.D. Thesis, As rhus University, 1994, unpublished. [4] See for instance D. Kella, P.J. Johnson, H.B. Pedersen, L. Vejby-Christensen, L.H. Andersen, Science 276 (1997) 1530. [5] P. Balling et al., Phys. Rev. Lett. 77 (1996) 2905. [6] H.H. Andersen et al., Phys. Rev. Lett. 79 (1997) 4770. [7] See for instance K. Abrahamsson et al., Nucl. Instr. and Meth. B 31 (1988) 475. [8] H. Poth, Phys. Rep. 196 (1990) 135. [9] A. Poncet, in: S, Turner (Ed.), CERN Accelerator School: Fifth Advanced Physics Course, CERN 95-06, 1995. [10] M.K. Raarup, Progress report, As rhus University, 1999, unpublished. [11] See for instance H. Koziol, Cern Accelerator School: Fifth General Accelerator Physics Course, CERN 94-01, p. 565.
SECTION III.
Electron cooling of D~ at the ASTRID storage ring J.S. Nielsen*, S.P. M+ller!, L.H. Andersen", P. Balling", M.K. Raarup" !Institute for Storage Ring Facilities (ISA), University of As rhus, Ny Munkegade, Bygn 520, DK-8000 As rhus C, Denmark "Institute for Physics and Astronomy, University of As rhus, DK-8000 As rhus C, Denmark
Abstract A report of recent results on electron cooling of D~ at an energy of 1.6 MeV in the ASTRID storage ring is given. The longitudinal velocity spread has been reduced from &4]10~4 (FWHM) to &7]10~5 (FWHM) at a current of &0.1 lA. A drift in the mean velocity of the cooled beam has been reduced by application of a small RF signal on four sets of plates in the cooler. Initially, the velocity spread is found to decrease with ion current, indicating equilibrium between cooling and intra-beam scattering, whereas at later times (lower current) the velocity spread becomes constant, indicating equilibrium with the electron beam. To diagnose cooling, a simple system allowing to follow the frequency width and position of a Schottky harmonic on a sub-second time-scale, has been developed. The system uses a standard data acquisition card to digitize a down-mixed Schottky-signal and a FFT routine in Labview on a standard PC. The electron-cooled ion-beam is used for high-resolution vacuum ultra-violent spectroscopy of H~ and D~ in the region near the H(n"2) threshold. The velocity spread of the ion beam can be directly extracted from these experiments. ( 2000 Elsevier Science B.V. All rights reserved. PACS: 29.20.Dh; 29.27.-a; 41.75.Cn Keywords: Storage ring; Electron cooling; Negative ion-beam
1. Introduction The ASTRID storage ring [1] has now been in operation for more than 10 years, serving half the time as a 580 MeV Synchrotron Light Source, and half the time as a low-energy heavy-ion storage ring, primarily for atomic and molecular physics studies. The ring has from the start been equipped with an electron cooler [2], but apart from a single
test experiment with electron cooling of D` in the early 1990s [3] the electron cooler has only been used as a target for recombination studies [4]. However, recently an experiment performing VUV spectroscopy on H~ and D~ [5,6] has been limited in resolution by the Doppler broadening due to the velocity spread of the ion beam. An interest in cooling of H~ and/or D~ has therefore emerged. 2. Experimental setup
* Corresponding author. Tel.: #45-89422899; fax: #4586120740. E-mail address: [email protected] (J.S. Nielsen)
ASTRID consist of four pairs of dipole magnets, and four straight sections, each of 8 m length,
0168-9002/00/$ - see front matter ( 2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 9 0 0 2 ( 9 9 ) 0 1 1 2 5 - 0
J.S. Nielsen et al. / Nuclear Instruments and Methods in Physics Research A 441 (2000) 150}153
Fig. 1. Layout of the ASTRID storage ring together with the experimental setup.
151
current about 3.5 mA, and the expansion factor &6. The electron density in the cooler is about 4]106 cm~3. The details of the VUV spectroscopy experiment can be found in Ref. [5,6]. The experiment aims at observing and characterising the 1P0 dipole series of autodetaching resonances in H~ (and D~) located just below the H(n"2) threshold. In order to measure the photodetachment cross section, the ninth harmonic (118 nm) of a Nd:YAG laser is overlapped with the ion beam in one of the straight sections of ASTRID (see Fig. 1). As the wavelength of the laser light is "xed, the photon energy seen by the moving ions is controlled, via the Doppler shift, by tuning their velocity (i.e. tuning the energy of the ions). Neutral atoms produced by photodetachment or by collisional detachment in the rest-gas pass unde#ected through the magnetic lattice of the storage ring, and are recorded by a detector at the end of the straight section.
3. Cooling results giving a total circumference of 40 m (see Fig. 1). Ions are produced and accelerated to typically 150 keV in a small electrostatic accelerator before injected into the ring. In the ring the ions can then be accelerated, by drift tube acceleration [7], to a few MeV per nucleon, limited by the rigidity of 1.9 Tm of the dipole magnets. In the present experiment, D~-ions are produced in a duoplasmatron ion source, injected into the ring at 150 keV, and during 3.0 s accelerated to 1.6 MeV. After acceleration the beam is debunched and electron cooling is started. The ion-beam current after acceleration in the present experiment has been 0.1}0.2 lA. The lifetime of the ion beam is &2.5 s (calculated from the areas of Schottky peaks), and is determined by collisional detachment with the rest-gas of &3]10~11 mbar. The ASTRID electron cooler is located in one of the straight sections. It allows overlap with an electron beam with an energy of up to 2.0 keV over a length of 1 m. The electron beam can be adiabatically expanded by a factor of up to about 10. Typical electron currents are 2}5 mA. In the present experiment the electron energy is &450 eV, the
In Fig. 2 the result of electron cooling of a 1.6 MeV D~ beam is shown. As can be seen from the "gure initially there is a fast decrease in the velocity spread as the beam cools, and then a slow decrease, until a more or less constant minimum relative spread of &7]10~5 (FWHM) is obtained after about 3 s. The region with the slow decrease (&1.0 to &2.5 s) is interpreted as the result of an equilibrium between cooling and intra-beam scattering (IBS). As predicted by simple estimates [8], the decrease in the velocity spread is approximately proportional to N0.4, where N is the number of particles in the ring (N is closely following an exponential decay with a lifetime of &2.5 s). At the time where the minimum spread is obtained the number of particles in the ion beam is &1]107, for which Eq. (161) in Ref. [8] predicts a spread of 7]10~5, in agreement with our measurement. The minimum spread is interpreted as a temperature equilibrium between the ion beam and the electron beam (or very close to equilibrium). The longitudinal electron temperature can be extracted from the minimum spread to be 1 meV, which is close to the value expected from the adiabatic acceleration.
SECTION III.
152
J.S. Nielsen et al. / Nuclear Instruments and Methods in Physics Research A 441 (2000) 150}153
aim is to be able to fully quantify the properties of the M0N~ resonance (i.e. determine its width and 2 3 asymmetry). Furthermore, upper limits to the width of the second resonance ( M0N~) will be deter2 4 mined. It is expected to be very narrow (&2 leV) and hence serves as an independent measure of the ion-beam velocity spread. Finally, a search for the third resonance ( M0N~), which is predicted to exist 2 5 very close to the threshold, will be performed.
4. New Schottky data-taking system Fig. 2. The measured relative velocity spread (FWHM) as a function of time after acceleration has "nished and the beam has been debunched.
A crucial point in obtaining a stable mean longitudinal velocity of the cooled ion-beam has been the application of a small RF signal on electrostatic plates in the cooler, to &&shake-o!'' rest-gas ions trapped in the electron beam [9].1 The trapped ions change the potential in the electron beam, thereby changing the electron velocity and hence the cooled ion-beam velocity. The time scale for the ion production is long compared with the storage time, and a steady state is therefore never achieved. A constant velocity is, however, crucial for the experiment, due to the use of the Doppler tuning of the apparent laser frequency. The problem is cured by applying a sinusoidal RF signal with a frequency of 250 kHz and amplitude 3 V to one side of four sets of plates usually used as pickups for position measurements. The other side is grounded. This clearly reduces the neutralisation, and, more importantly, the time it takes to obtain a steady-state situation. With RF-shaking the drift after 2.0 s was below the velocity spread. The reduction in the velocity spread obtained by cooling has allowed the width of the "rst resonance ( M0N~) to be determined to 63 (13) leV [10]. Due 2 3 to time limitations and problems with the laser system, the uncertainty of the measurement is large. This will be improved in future beam times, and the
1 This cure was kindly suggested to us by Jacques Bosser, PS Division, CERN. See for instance Ref. [9]
To facilitate real-time diagnostics of cooling as well as detailed o!-line analysis, a simple and inexpensive system to analyse the Schottky signals [11] has been developed. Previously, the Schottky signals were analysed using a spectrum analyser (HP4195A). In free-running, however, this allowed only a few scans per injection at a fairly poor resolution. This could therefore only be used as a crude measure of the success of cooling. To obtain quantitative results the analyser had to be triggered allowing only one measurement at each injection. As the signal to noise in the Schottky signal is low many injections had to averaged, adding to the time consumption. The fundamental problem with the spectrum analyser is that it is Fourier limited on each point. A much better solution is to digitise the time signal and perform a Fast Fourier Transform to get the frequency spectra. In this way, the whole spectrum is obtained at once. To avoid the necessity of high sampling rates, mixing to an intermediate frequency is employed. This also drastically reduces the sample size (and thereby the computation time) for a given frequency resolution. In our setup the Schottky signal at about 4.5 MHz is mixed to about 18 kHz and low-pass "ltered before being digitised by a commercial PC data acquisition card (National Instruments PCI6023E) at 200 kS/s. Acquiring 32768 points gives a frequency resolution of 6 Hz and a new spectrum can be acquired every 0.16 s. On a modern PC (Pentium II 266 MHz) a FFT of 32768 points can, with commercial software (LabView), be done in about 100 ms, which means that the spectra can be recorded and analysed (position and width be extracted) real-time. A special LabView program has
J.S. Nielsen et al. / Nuclear Instruments and Methods in Physics Research A 441 (2000) 150}153
been written which reads the digitised signal, performs the FFT, and displays the Schottky spectra real-time. It furthermore allows averaging over a number of injections, after which the frequency spectra can be stored in a "le. A peak position and a peak width is extracted and displayed online from the averaged signals. The development time for the LabView application was about a week for two new LabView developers.
5. Conclusions We have described a recent experiment with electron cooling of D~ at the ASTRID storage ring. The relative longitudinal velocity spread has been reduced from &4]10~4 to &7]10~5 close to equilibrium with the electron beam. A crucial point for attainment of a stable ion-velocity is the application of a RF-signal to electrostatic plates in the cooler to &&shake-o!'' ions produced by ionisation of rest-gas atoms. The development of a much improved Schottky diagnostic system has been important for optimising electron cooling, in particular regarding the alignment of the ion and electron
153
beams. The experiment using the cold D~ beam has already bene"ted from the reduced velocity spread, and further progress is expected in the future.
References [1] S.P. M+ller, The Aarhus storage ring for ions and electrons ASTRID, Proceedings from 1993 Particle Accelerator Conference, Washington 1993, p. 1741. [2] L.H. Andersen, J. Bolko, P. Kvistgaard, Phys. Rev. A 41 (1990) 1293. [3] H.T. Schmidt, Ph.D. Thesis, As rhus University, 1994, unpublished. [4] See for instance D. Kella, P.J. Johnson, H.B. Pedersen, L. Vejby-Christensen, L.H. Andersen, Science 276 (1997) 1530. [5] P. Balling et al., Phys. Rev. Lett. 77 (1996) 2905. [6] H.H. Andersen et al., Phys. Rev. Lett. 79 (1997) 4770. [7] See for instance K. Abrahamsson et al., Nucl. Instr. and Meth. B 31 (1988) 475. [8] H. Poth, Phys. Rep. 196 (1990) 135. [9] A. Poncet, in: S, Turner (Ed.), CERN Accelerator School: Fifth Advanced Physics Course, CERN 95-06, 1995. [10] M.K. Raarup, Progress report, As rhus University, 1999, unpublished. [11] See for instance H. Koziol, Cern Accelerator School: Fifth General Accelerator Physics Course, CERN 94-01, p. 565.
SECTION III.