Miniball: A Gamma-Ray Spectrometer WithPosition-Sensitive Ge Detectors For Nuclear Structure

Studies At REX-ISOLDE

J. Eberth, G. Pascovici, H.G. Thomas, N. Warr, D. Weißhaar∗, D. Habs, P. Reiter, P.Thirolf†, D. Schwalm, C. Gund, H. Scheit, M. Lauer∗∗, P. Van Duppen, S. Franchoo,

M. Huyse‡, R.M. Lieder, W. Gast§, J. Gerl¶, K.P. Lieb‖ and the Miniballcollaboration

∗Institut für Kernphysik, Universität zu Köln, D-50937 Köln, Germany†Ludwig-Maximilian-Universität München, D-85748 Garching, Germany

∗∗Max-Planck-Institut für Kernphysik, D-69029 Heidelberg, Germany‡Instituut voor Kern- en Stralingsfysica, University of Leuven, B-3001 Leuven, Belgium

§Institut für Kernphysik, Forschungszentrum Jülich, D-52425 Jülich, Germany¶Gesellschaft für Schwerionenforschung, D-64220 Darmstadt, Germany

‖II. Physikalisches Institut, Universität Göttingen, D-37073 Göttingen, Germany

Abstract. Miniball is a dedicated Ge detector array which has been developed for the investigation of rareγ decays at thenew radioactive beam facility REX-ISOLDE [1, 2] at CERN. The array is optimised for high full-energy peak efficiency andfor high granularity needed to perform Doppler corrections ofγ-rays emitted by fast moving nuclei. Miniball will finallyconsist of 40 six-fold segmented, encapsulated detectors which are clustered in eight cryostats with three detectors each andfour cryostats with four detectors, respectively. It is shown that from an analysis of the pulse shapes and of the amplitudes ofthe mirror charges in the adjacent segments the effective granularity of Miniball can be enhanced from 240 to≈ 4000. Theproperties of Miniball are compiled on the basis of experimental data. Examples of the first data measured with Miniball arepresented.

THE CHALLENGE OF RARE-ISOTOPE BEAMS

A major direction of research at present, is the development of rare isotope beam facilities, which typically accelerateradioactive isotopes separated on line after reactions produced by a primary beam (REX-ISOLDE, MAFF, SPIRAL) orproduce them directly via heavy-ion induced fragmentation reactions (MSU, RIKEN, GSI). Such facilities give accessto physics far from stability, such as probing the drip lines, halo nuclei and the study of effects like shell melting inneutron-rich nuclei, the T=0 proton-neutron pairing in selfconjugated nuclei up to100Sn and the detection of waitingpoints in the r- and rp-process during nucleosynthesis. Coulomb excitation and single nucleon transfer reactions[3]may be used to probe these nuclei.

Due to the very low beam intensities produced by such facilities, it is important to detectγ rays with high efficiencyas well as a large solid angle coverage. Consequently it is desirable to use large volume Ge detectors close to thetarget. In the current phase of Miniball, this is achieved using twenty-four 60% efficiency Ge detectors housed ineight triple cryostats. Finally Miniball will consist of 40 detectors where the additional detectors are clustered in fourcryostats with four detectors each. Furthermore, a high flexible frame has been constructed which makes differentsetups possible like an efficient 4π configuration or a 2π configuration with place for auxiliary detectors.

On the other hand, the use of inverse kinematics produces recoil velocities of the order of 5 % c, which results insignificant Doppler broadening. For this reason, it is important that the angle subtended by each detector should bekept to a minimum in order to have good angular resolution (high granularity) so we can perform a Doppler correction.

The requirements of high efficiency, high granularity and reasonable cost cannot be reconciled using conventionalGe detectors. Instead, Miniball has adopted a new approach, which is the application of position-sensitive Ge detectors.

Downloaded 14 Nov 2006 to 137.138.5.76. Redistribution subject to AIP license or copyright, see http://proceedings.aip.org/proceedings/cpcr.jsp

HV AC−coupled Core

segment signals

segments6 DC−coupled

segmented

capsulelid

Alucan

Ge crystal

FIGURE 1. The Miniball 6-fold segmented, encapsulated detector.

The position sensitivity is achieved by a combination of segmented Ge detectors and pulse shape analysis of thedetector signals[4]. Position sensitive detectors also are the ingredient of the new generation ofγ-ray trackingspectrometers planned in Europe (AGATA[5]) and the USA (GRETA[6]). The principle ofγ-ray tracking necessitates,in addition to the energy and time of eachγ-ray, the recording of each interaction point within the position-sensitivedetectors. The target is surrounded by a shell of 100-150 position-sensitive Ge detectors and utilising the kinematicsof the Compton scattering, powerful tracking algorithms reconstruct the track of theγ-ray to decide whether or nottheγ-ray was emitted from the target and if it was fully absorbed in the Ge shell. Such spectrometers will give severalorders of magnitude improvement in resolving power compared to the most powerful spectrometers today (Euroball,Gammasphere).

Miniball is the first fully operational spectrometer which puts the new technology of position-sensitive detectors intoaction and uses segmented detectors and pulse shape analysis to determine the two-dimensional position of the firstinteraction of aγ-ray for Doppler correction. As Miniball is not capable of a fullγ-ray tracking because the detectorslack segmentation in the depth, this spectrometer is dedicated to experiments with lowγ-multiplicity as performedat REX-ISOLDE. But the experience and results of Miniball give a good idea about the feasibility ofγ-ray trackingspectrometers as all the necessary technologies and techniques are employed for the first time.

SEGMENTATION AND PULSE-SHAPE ANALYSIS

Germanium detectors used inγ-ray spectroscopy usually have a quasi-coaxial geometry with a closed front end. Highvoltage is applied at the central electrode and signals can be detected at the outer electrode which is DC coupled to thepreamplifier. The crystals of the Miniball detectors are similar, but the outer electrode is cut into six segments, each ofwhich is DC coupled to its own field effect transistor (FET). A further signal from the core electrode is connected byAC coupling to a seventh FET (see figure 1).

When aγ ray is completely absorbed in the active volume of the Ge crystal, the signal on the core electrodeindicates the full energy, as with a conventional detector. Furthermore, by looking at the signals in the segments, wecan determine what fraction of that energy was lost in each segment and thereby obtain the segment in which theinteraction with the highest energy deposition (main interaction) occurred. By making the assumption that the main

Downloaded 14 Nov 2006 to 137.138.5.76. Redistribution subject to AIP license or copyright, see http://proceedings.aip.org/proceedings/cpcr.jsp

1000

2000

Time[ns]−400 4000

Time[ns]−400 0 400 800

2000

1000

Time[ns]

2000

1000

1000

1000

2000

2000

1000

2000

Time[ns] Time[ns]4000−400400−400 0

2000

Time[ns]8004000800

2000

1000

1000

−400 0 400

Seg 3

Seg 2

Seg 6

Seg 4

Seg 5

Seg 1

Seg 6Seg 3

Seg 5

Seg 4

3

2 6

Seg 1

Seg 2

4

1

5

Core 4

2 61

Am

plitu

de[A

DC

−un

its]

Core

53

FIGURE 2. Signals for two different interactions. In each case, the interaction takes place in segment 4, but in different places.Note the dependence of the rising edge of the core signal on the radius of the interaction and the asymmetry of the amplitudes ofthe transient signals induced in the neighbouring segments.

interaction is at the same position as the first interaction, we can use this information to refine our Doppler correction.This assumption turns out to be rather good at high and low energy and acceptable at intermediate energy.

A further increase in the granularity can be obtained by considering the pulse shapes of the signals on the electrodes.The rise time of the signal on the (positive) core electrode is determined by the collection time of electrons. Clearly, thecloser the interaction to the core, the faster the electrons will be collected and the faster the signal will peak. We definea quantity called the “time to steepest slope” for this signal, which is proportional to the radius of the interaction. Thesegmentation of the outer electrode provides an additional information. After an interaction in the crystal, electronsmove towards the core electrode and holes to the outer electrode. If the electrons reach the core before the holes reachthe outer electrode, there will be a net transient charge imbalance, which induces transient signals on the electrodes forthe adjacent segments. If the holes arrive at their electrode first, we get a similar transient on the adjacent segments, butof opposite polarity. The relative amplitudes of the transient on the left and right adjacent segments indicates whetherthe interaction took place nearer to one or the other (See figure 2).

Tests with a Miniball detector using a collimated source and a scanning table have shown that each segment canbe subdivided into about 16 pixels with this method giving two orders of magnitude better granularity compared tounsegmented detectors[7]. The price to be paid for this is that each of the 24 Ge detectors provides 7 signals, a totalof 168 channels for the data acquisition system to process for the Ge part alone. In order to place 21 signal channels

Downloaded 14 Nov 2006 to 137.138.5.76. Redistribution subject to AIP license or copyright, see http://proceedings.aip.org/proceedings/cpcr.jsp

Gain DAC

Offset DAC

Analog inputcircuitry

Good

Fast

FIFO1024

Filter

loca

l bus

RAM

24bit

CAMAC IEEE 1394

DMA

1 of 4 channels

NyquistFilter 12bit 40MHz

ADC

Slow

Data

Logic Lines:−GFLT (Master Trg)

DGF−4C

DSP

Fast FirewireCAMAC I

Host computer

4*32K

16 MWrd/s

−Fast MULT IN/OUT

−GSLT (Validation)

−BUSY/SYNCH for synchronous operation of several DGF moduls

2.5 MWrd/s

FIGURE 3. The block diagram shows the functional principle of the DGF-4C CAMAC module.

in each cryostat, smaller cold and warm preamplifiers had to be developed and much effort was needed to eliminatecrosstalk, microphonics and noise from these systems. Furthermore, it is no longer sufficient to acquire energies andtimes, but pulse-shape information is also needed, which means that conventional acquisition systems are inadequate.

DIGITAL ELECTRONICS

The solution to the acquisition problem was found in the form of a commercially available camac module, the DGF-4C supplied by the company XIA, Newark(CA), USA. This single width camac module provides four completespectroscopic channels, where the signal first passes through an analogue stage with a Nyquist filter and an amplifierwith software-controlled gain and offset, before being digitised by a 12 bit 40 MHz sampling ADC (newer versions are14 bit). From then on, all processing is numerical, first in a field programmable gate array (FPGA) for each channel,which performs filtering and triggering operations and then in a digital signal processor (DSP) for each module andwhich performs the pulse-shape analysis using DSP code developed specially for Miniball.

Times for events are calculated relative to the 40 MHz clock which is supplied from an external module and passedto each DGF-4C. Energies are obtained by filtering in the FPGA and then performing ballistic deficit correction in theDSP. Finally, some parameters to describe the pulse shape are determined in the DSP. The DSP packages up this datainto buffers and stores up to 8000 words of data in internal memory which can be read out over the camac bus usingfast camac level 1 at 2.5 Mwords/s. In order to overcome the limitations in transfer rates imposed by the CAMACstandard the board has an additional Firewire readout (IEEE 1394) with a maximum transfer rate of 16 Mwrds/s.On-board memory of 4×32k forγ-ray single spectra is available.

A multiplicity bus and readout with adjustable threshold and inputs for first- and second-level triggers make itpossible to incorporate the modules in complex coincidence systems. Each event in the DGF-4C is tagged with a timestamp and coincident events in different modules are identified by comparing the time stamps in the offline analysis.

The energy resolution obtained for the Miniball detectors with DGF-4C is comparable or better than the resolutionmeasured with conventional analogue spectroscopy amplifiers.

ANCILLARY DETECTORS

In addition to the Ge detectors, Miniball is equipped with a double-sided silicon strip detector (DSSSD)[8] in thetarget chamber for detecting charged particles. The readout for the 160 channels of this detector is performed usingconventional analogue electronics, so techniques had to be developed to match the data from the analogue electronicsto the timestamped digital data. This was achieved by using an additional DGF-4C to acquire the same logic signalthat was used to generate the ADC/TDC gates for the analogue electronics. This provided the absolute time, relativeto the 40 MHz clock, of the gate, and the relative times between the events in the DSSSD and the ADC gate were

Downloaded 14 Nov 2006 to 137.138.5.76. Redistribution subject to AIP license or copyright, see http://proceedings.aip.org/proceedings/cpcr.jsp

100 200 300 400 500 600 700 800 900

0

200

100

40 50 60 70 80 90 keV

1000

2000

3000

keV

905

197

87

52

46

874

941

87

FIGURE 4. Cut at the 1473keV transition in160Dy populated by theβ -decay of160Er. The cutout on the upper right shows thelow energy part of this spectrum to demonstrate the low threshold at 25keV. The full range of the spectrum is 4MeV.

1700 1800 1900

1400 1600 18001000800

80000

60000

40000

20000

1200

50

40

30

20

10

900850 950 1000 1050

100000

100

200

300

400

+2

+2

4+

160Dy

1930

keV

Gate at 879keV

728keV

Gate at 1930keV

966keV880keV

966

728

1930

879

[keV][keV]

FIGURE 5. The cut at the 879keV transition shows that the 1930keV transition has just 0.05% of the statistic of the parallel728keV transition (see level scheme on the right). Nonetheless a cut at the 1930keV transition gives a coincidence spectrum ofgood quality. The 1930keV transition in160Dy was previously unknown.

recorded with the TDCs. Both the ADCs and TDCs, like the DGF-4Cs, were capable of buffering the data, so thatmany events could be acquired during each beam pulse. Due to the time structure of the REX-ISOLDE linac operatingat 50 Hz, efficient acquisition was obtained by forcing a readout at the end of each beam pulse, so the system was readyto acquire again when the next beam pulse occurred. In Miniball experiments, only the core electrodes were used togenerate triggers, with the segment electrodes always being read out whenever the corresponding core triggered.

Miniball also has a parallel-plate avalanche counter which can be used as a trigger or for beam diagnostics and canbe read out in single-particle or current-measurement modes. The PPAC has 25 strips in the x- and 25 strips in they-direction to provide information about the direction of the beam particle after interaction. This is needed to resolvethe kinematics of the reaction for the Doppler correction[9].

Downloaded 14 Nov 2006 to 137.138.5.76. Redistribution subject to AIP license or copyright, see http://proceedings.aip.org/proceedings/cpcr.jsp

0 100 200 300 400 5000

50

100

150

200

250

300

350

400 d(25

Na,26

Na), E=2.25 MeV/u,doppler corrected

Energy [ keV ]

coun

ts

82

151

234

FIGURE 6. Doppler-corrected data for the d(25Na,26Na) measurement using a deuterated polyethylene target. The peaks labelledwith their energies are the only transitions previously placed in this nucleus. Theγ-rays were measured in coincidence with theemitted protons detected with the segmented silicon detector.

RESULTS

We performed in-beam measurements in inverse kinematics at the Cologne tandem accelerator to test the capabilityof the position sensitivity for correcting the Doppler broadening. A Miniball Cluster was positioned at 90o in respectto the beam axis and in a distance of 11cm to the target. Theγ emitting recoil nuclei had a velocity of v/c=5.6%.The analysis of a 2.17MeV-transition shows that with a FWHM of about 35 keV before Doppler correction, we canimprove this to around 15 keV using just the segment hit pattern and to 10 keV using the full pulse-shape analysis.The main contribution to the latter resolution is the kinematics of the reaction and was found experimentally to be7.8keV in good agreement with the calculated value of 7.5keV of a Monte-Carlo simulation. This means for the targetdetector distance of 11cm, used in the measurement, that an effective detection angle of 5.9o with segment informationand 3.1o with pulse shape analysis was achieved. This in-beam result confirms the measurements with the collimatedsource that in fact the pulse shape analysis enhance the granularity of a segment by a factor of 16. Thus the granularityof the complete Miniball phase II with 40 detectors is improved from 240 given by the segmentation to 4000 withpulse shape analysis.

From the end of last year until March 2002, Miniball was used in a configuration consisting of 18 detectors mountedin 6 cryostats, for experiments with radioactive decays and stable beams at Cologne University. In figure 4 a cutspectrum on the 1473keV transition in160Dy is shown to demonstrate the quality and the low background. Furthermore

TABLE 1. Specification of Miniball for different setups. Resolutions of 1MeVγ-rays are given for different recoil velocities and detection angles.1.3MeV PPh gives theefficiency at 1.3MeV and∆Θr is the resolution of the effective detection angle.

configuration target distance 1.3MeV PPh ∆Θr β ∆E90o ∆E30o

Phase I 7cm 15% 5.7o 5% 5.3keV 3.3keV18 detectors 15% 14.9keV 8.9keV

9.5cm 9.4% 4.2o 5% 4.1keV 2.8keV15% 11.1keV 6.7keV

12cm 6.3% 3.3o 5% 3.5keV 2.5keV15% 8.9keV 5.5keV

Phase II 11cm 16.4% 3.6o 5% 3.7keV 2.6keV40 detectors 15% 9.6keV 5.9keV

13.5cm 11.5% 2.9o 5% 3.3keV 2.4keV15% 7.9keV 4.9keV

16 cm 8.5% 2.5o 5% 2.9keV 2.3keV15% 6.8keV 4.3keV

18.5cm 6.5% 3.4o 5% 2.7keV 2.2keV15% 6.0keV 3.9keV

Downloaded 14 Nov 2006 to 137.138.5.76. Redistribution subject to AIP license or copyright, see http://proceedings.aip.org/proceedings/cpcr.jsp

FIGURE 7. The photo shows Miniball at CERN 2002. The target chamber is surrounded by eight Miniball-triple detectors. Asilicon detector and a parallel-plate avalanche counter are mounted inside the chamber for particle measurements.

one notes the low energy threshold of 25keV for a full energy range of 4MeV which enables the detection of X-raysfrom 160Dy in coincidence to theγ-rays. In the second example (figure 5) a cut spectrum at 879keV is shown. In theblow-up the 1930keV transition is shown which has 0.05% of the statistic of the strong 728keV transition. Cutting onthe 1930keV transition still gives a good coincidence spectrum. The corresponding cutout from the level scheme of160Dy is also shown in this figure.

From the experimental data we are able to compile the specifications of the Miniball array. Table 1 shows total-absorption efficiencies for different target-detector distances and the energy resolution of a 1MeVγ-ray expected fordetection angles of 30o and 90o and recoil velocities of 5% and 15% of the velocity of light. The effective detectionangle∆Θr for different target distances are computed from the measured value at 11cm for the 2.17MeVγ-ray wherebythe measured value is scaled down for a 1MeVγ to take the lower position sensitivity for lowerγ-energies into account.From the results one sees that the experiments at REX-ISOLDE with a typical recoil velocity of 5%c can be performedwith an average resolution of 2.7-2.9keV which is close to the intrinsic resolution of the Ge-detector. Even at highrecoil velocities of 15%c the resolution is one order of magnitude better than the resolution obtained by scintillationdetectors.

Since Easter, Miniball has been in CERN with 24 six-fold segmented detectors in eight cryostats and has beenused several times this year for the study of light neutron-rich radioactive isotopes (Na and Mg) following Coulombexcitation and transfer reactions[10]. Analysis of the data is ongoing and the Mg experiments will be performedmid-September 2002. A preliminary spectrum of the d(25Na,26Na) measurement is shown in figure 6 and a photo ofMiniball at CERN is shown in figure 7.

Downloaded 14 Nov 2006 to 137.138.5.76. Redistribution subject to AIP license or copyright, see http://proceedings.aip.org/proceedings/cpcr.jsp

Following the planned upgrade of REX-ISOLDE to 3.1 MeV/u[11] it will be possible to study radioactive nuclei upto A = 150 at CERN.

The digital electronics proved to be reliable and attained energy resolutions comparable to or better than withanalogue electronics. The timing resolution is limited in real time by the 25 ns sampling period, though this is quiteclose to the intrinsic timing resolution of such large-volume detectors. The software includes a constant fractionalgorithm but does not yet interpolate the zero crossing to fractions of the 25 ns tick for the listmode data. Offlineanalysis of traces acquired with the same modules show that interpolation of the zero crossing yields timing resolutionsequivalent to the intrinsic timing resolution of Ge detectors. Tests with a60Co source have shown that a time resolutionof 15ns between two Miniball detectors can be achieved with an energy threshold of 25keV.

LONG-TERM STABILITY AND FUTURE PLANS

The first of the Miniball detectors have been used in experiments for over two years now and have been shown to bereliable. The use of the same encapsulation technology, as used for the Euroball detectors, has proved essential to thesuccess of the project. Firstly, none of the capsules which passed the initial tests have failed, and have not, therefore,had to be returned to the manufacturer (Eurysis). At the same time, the cold part of the electronics remained accessible,which is necessary due to the complexity of a system with 21 cold FETs in a very limited space. Problems such ascrosstalk and microphonics etc. typically occur in the cold part and sufficient performance can only be achieved if thispart is accessible when the detector is warm. Furthermore, the AC-coupled FET of the core electrode is particularlyvulnerable to discharges of the high voltage. Whenever such problems occurred, it was always possible to open up thecryostat and replace the FET without breaking the vacuum of the capsule containing the Ge crystal and without havingto send the detector back to Eurysis. It was always possible for experienced personnel to perform on-site maintenance.

Future plans for tracking arrays such as AGATA and GRETA will have more highly segmented detectors than theMiniball clusters, requiring an even denser packing of the electronics. Experience with the Miniball detectors hasclearly shown that with a large number of FETs which have to be closer to the high voltage than for conventionaldetectors, reliability can only be achieved through encapsulation of the Ge crystal. Furthermore, the proximity of agreat many channels results in many constraints in order to avoid crosstalk problems.

The results of the pulse shape analysis have shown that the position sensitivity of a Miniball detector is limited bythe geometry of the detector and the lack of a higher segmentation especially in the depth but not by the quality of thedetector signals itself. Thus, this gives a very optimistic view for the feasibility of futureγ-ray tracking spectrometers.

ACKNOWLEDGMENTS

Many members of the Miniball collaboration have contributed to the work presented here. Their effort is gratefullyacknowledged. The work is supported by the BMBF under contract numbers 06 OK 862 I, 06 OK 958, 06 LM 868(I) and 06 GÖ 851, by the Flemish Fund for Scientific Research, Belgium and by the European Community undercontract No. TMR ERBFMRX CT97-0123.

REFERENCES

1. Habs, D.,et al., Prog. Part. Nucl. Phys.38, 111 (1997).2. Habs, D.,et al., Nucl. Phys.A 616, 29c (1997).3. Lenske, H., and Schrieder, G., Eur. Phys. J.A2, 41 (1998).4. Eberth, J.,et al., Prog. Part. Nucl. Phys.38, 28 (1997).5. Lieder, R.M.,et al., Nucl. Phys. A682, 279c (2001).6. Deleplanque, M.A.,et al., Nucl. Instr. Meth.A430, 292 (1998).7. Eberth, J.,et al., Prog. Part. Nucl. Phys.46, 389 (2001).8. Davinson, T.,et al., NIM A454, 350 (2000).9. Cub, J.,et al., NIM A453, 522 (2000).10. Scheit, H.,et al., ISOLDE proposal, CERN/ISTC 99-20, ISC/P 114.11. Kester, O.,et al., ISOLDE proposal, CERN-INTC-2002-009; INTC-P-152.

Downloaded 14 Nov 2006 to 137.138.5.76. Redistribution subject to AIP license or copyright, see http://proceedings.aip.org/proceedings/cpcr.jsp