GRETINA experiments with fast beams at NSCL Dirk Weisshaar, NSCL @
Layout of a fast-beam experiment at NSCL… …involving the S800 spectrograph Reaction target 100-400mg/cm2 Identification and beam transport 80-100MeV/nucleon 100-106 pps Reaction product identification S800 spectrograph A1900 fragment separator Production target Beam energy from cyclotrons 100-150MeV/nucleon Length ~80m (260 ft) (linear distance)
(Find) GRETINA in S3
GRETINA/S800 data acquisition NSCL DAQ framework: Run Control Online Analysis (s800) event data traffic start/stop start/stop S800 data S800 data online data GRETINA DAQ + GEB * S800 DAQ data sample GRETINA + S800 data Online analysis Storage of production data on GRETINA disk array *Gretina Event Builder
GRETINA/S800 triggering/synchronization Downscaled, 12.5MHz GRETINA clock Gamma-ray trigger S800 focal plane trigger ~250ns latency GRETINA DAQ S800 Trigger Logic S800-GRETINA Coincidence window: 600ns Trigger signal issued ~900ns after particle detection in S800 S800 DAQ Important note: If GRETINA runs into deadtime, no busy signal or the like is provided. Trigger live out
GRETINA timing GRETINA leading edge timing GRETINA ‘t0’ timing: 600 ns This timing quality is available as prompt signal (~200 ns latency) for a trigger logic. GRETINA ‘t0’ timing: This timing quality is available after decomposition, i.e. in the data analysis. Coincidence with 1.3 MeV γ ray of 60Co source energy energy 600 ns 600 ns
T0 timing quality at low energies Ge-Ge 1.17-1.33MeV: 14ns 100keV 200keV ~40% ~15% Coincidence with 1.3 MeV γ ray of 60Co source 300keV 400keV 400 ns ~5% <5% T0 comes from decomposition fit. This means that a significant amount of low-energy events were fitted badly by decomp. On the other hand, for events > 400keV fits do well in the regard.
Singles efficiency: Treat GRETINA as independent crystals (and ignore scattering and sum-peak effects)
I love low-activity sources! Here, 60Co at 0.3uCi does <0.2% randoms (time gated) This is as close as you get to a ‘single gamma GEANT’ (Note: angular correlation does have few % impact) Gretina gated on 1332keV in LaBr:Ce Thought on tracking: Assume average detector fold 5 and 10kHz per detector, i.e. crystal livetime 0.9 (otherwise energy pile-up). Only 60% chance that all 5 detectors deliver ‘good’ information, especially interaction points for tracking. In other words on 40% of the events, tracking works on an incomplete data set LaBr:Ce
P/T calorimeter and add-back Calorimeter for 1173keV P/T = 0.394 Calorimeter for 1836keV P/T = 0.355 SE 511 DE ‘wrong’ fep subtraction Note: 1.53/1.42 = 93%, i.e. addback recovers 90+% of calorimeter efficiency. (at low multiplicity)
Doppler-shift reconstruction γ rays of 28Si at v/c = 0.38 in GRETINA S800 36Ar 28Si θ Ge v/c ≈ 0.4 ϴ [rad] in GRETINA energy [keV] (laboratory frame) (Obvious) requirements for a spectrometer: Doppler-shift correction Spatial resolution Lorentz Boost Detection efficiency at forward angle …and GRETINA is a perfect match beam
FWHM after Doppler-shift reconstruction full GRETINA FWHM at 1.78MeV: 16keV, 0.9% Measured FWHM of 0.9% translates (naively) in a spatial resolution of σ=1.8mm Energy [kev]
FWHM correlated with angle Doppler Broadening is driven by the uncertainty of: spatial resolution of γ detection velocity of emitter direction of emitter beam spot size Measured FWHM vs. angle defines an effective angle resolution of 14 mrad σ ≈ 1.2 mm FWHM [keV] of 1779keV line Δθ Δβ Beam spot Polar angle [rad]
Comparison with SeGA Meaning: ‘Good’ Doppler-correction we did already before tracking
That’s GRETINA’s advantage over SeGA 64Ge γ-γ, produced from 65Ge on 9Be at v/c=0.4 plain singles tracked* * Tracked and provided by I-Yang Lee Reduction of Compton background by tracking allows – for the first time – gamma spectroscopy with fast beams with spectral quality comparable to arrays with anti-Compton shield, i.e. GammaSphere.
Spectral quality using add-back 6+ 3991 4+ 2754 2+ 1368 0+ 24Mg
Remarks Tracking arrays are beautiful devices for in-beam spectroscopy with fast beams! FWHM of the Doppler-reconstructed peaks is dominated by the position resolution (thin target) Nice testing ground of decomposition (and tracking) ‘Alternative’ (and simple) methods like add-back give already good results in those low-multiplicity data