1. Results from the DIRAC Experiment at CERN DImeson Relativistic Atomic Complex
L. Tauscher, for the DIRAC collaboration
Frascati, June 10 , 2004
16 Institutes
Spokesman: Leonid Nemenov, Dubna
L. Tauscher, Basel

2. The pp scattering length
DIRAC measures the lifetime of the pp atom (of order s)
Lifetime due to decay of pp atom: strong pp  pp (99.6%)
el. magn. pp   (0.4%)
Lifetime linked to s-wave pp scattering lengths (a2-a0)
Annihilation from higher l-states forbidden / strongly suppressed, in practice, absent
Method is independent of QCD models and constraints
The aim of DIRAC is to measure t with an accuracy of 0.3 fs
Physics motivation: elementary quantity in soft QCD (similar to mp)
L. Tauscher, Basel

3. The p+p- atom
Produced in proton-nucleus collisions (Coulomb final state interaction)
Atomic bound state (Eb ≈ 2.9 KeV)
Strong interaction  I = 0,2
Annihilation, p+p-p0p0 (a0-a2),
t≈ 310-15 s
Relativistic atom (g ≈17) migrates more than
10 mm in the target and
encounters around 100’000 atoms
L. Tauscher, Basel

4. Excitation and break-up
In collisions atom becomes excited
and/or breaks up
Pions from break-up have
very similar momenta and
very small opening angle (small Q)
At target exit this feature is smeared by
multiple scattering, especially in QT
L. Tauscher, Basel

5. Principle of measuring the lifetime
Excitation and break-up of produced atoms (NA) are competing with decay
Break-up probability Pbr is linked to the lifetime t
(theory of relativistic atomic collisions)
p+p- pairs from break-up
provide measurable signal nA
Pbr is linked to nA :
Pbr = nA/NA
The number of produced atoms NA is not directly measurable
 to be obtained otherwise (normalization using background)
L. Tauscher, Basel

6. Background
Pairs of pions produced in high energy proton nucleus collisions are coherent,
if they originate directly from hadronisation or
involve short lived intermediate resonances.
Coherent pairs undergo Coulomb final state interaction and become “Coulomb-correlated”
They are incoherent,
if one of them originates from long lived intermediate resonances, e.g. h’s (non-correlated)
because they originate from different proton collisions (accidentals)
L. Tauscher, Basel

7. Coulomb correlated (C) background
Coherent p+p- pairs undergo Coulomb final state interaction
 enhancement of low-Q event rate with respect to non-correlated events.
Coulomb enhancement function (Sommerfeld, Gamov, Sacharov, etc):
The same mechanism leads also to the formation of atoms
number of produced atoms NA and the Coulomb correlated background (NC) are in a calculable and fixed relation (normalization):
NA = * NC (Q ≤ 2 MeV/c)
L. Tauscher, Basel

8. Signal and background summary
Pion pairs from atoms have very low Q
C background is Coulomb enhanced at very low Q
Multiple scattering in the target smears the signals and the cuts
DIRAC uses the C background as normalization for produced atoms
Intrinsic difficulty: multiple scattering in MC
measuring 2 tracks with opening angle of 0.3 mrad
L. Tauscher, Basel

9. DIRAC spectrometer
L. Tauscher, Basel

10. Timing
L. Tauscher, Basel

11. Monte-Carlo Generators: tailored to the experiment
Accidental background according to Nacc/Q  Q2 with momentum distributions as measured with accidentals
Non-correlated background according to NnC/Q  Q2 with momentum distribution as measured for one pion and Fritjof momentum distribution for long-lived resonances for second pion
C background according to NC/Q  fCC(Q)*Q2 with momentum distribution as measured but corrected for long-lived incoherent pion pairs (Fritjof)
Atomic pairs according to dynamics of atom-target collisions and atom momentum distribution for C background
Geant4: full spectrometer simulation
Detector simulation: full simulation of response, read-out, digitalization and noise
Trigger simulation: full simulation of trigger processors
Reconstruction: as for real data
L. Tauscher, Basel

12. Data from Ni taken in 2001
Best coherent data taking with full trigger and set-up
Typical cuts:
QT < 4 MeV/c
QL < 22 MeV/c
No of reconstructed events in prompt window : 570‘000
For analysis the accidental background in the prompt window is obtained by proper scaling and kept fixed.
L. Tauscher, Basel

13. Experimental Q and Ql distributions (Ni2001)
Fit MonteCarlo C and nC background
outside the A2p signal region (Q > 4 MeV, QL > 2 MeV)
simultaneously to Q and QL
L. Tauscher, Basel

14. Residuals in Q and Ql Comments:
Q and QL provide same number of events  background consistent
Signal shapes well reproduced
L. Tauscher, Basel

15. Normalization e0.615kNC (Qrec ≤ Qcut) PBr = nA / NA
Fraction of atomic pairs with Qrec ≤ Qcut:
e = nA(Qrec ≤ Qcut)/nA (from MonteCarlo)
Number of atomic pairs:
nA = nA(Qrec ≤ Qcut) / e
Fraction of C background with Qinit ≤ 2 MeV contained in the measured C background with Qrec ≤ Qcut
k = NC(Qinit ≤ 2 MeV ) / NC (Qrec ≤ Qcut) (from MonteCarlo)
Number of produced atoms
NA = 0.615kNC (Qrec ≤ Qcut)
nA(Qrec ≤ Qcut)
_______________________________________________________________________
e0.615kNC (Qrec ≤ Qcut)
PBr =
L. Tauscher, Basel

16. Break-up from Ni2001 e0.615k = 0.1383 Strategy:
Use MC shapes for background from Monte Carlo
Use MC shapes for the atomic signal
Fit in Q and QL simultaneously
Require that background composition in Q and QL are the same
Result:
nA = 6560 ± 295 (4.5%)
NC = ± 3561 (1.0%)
NC (Q<4 MeV/c) =
e0.615k =
PBr = ± 0.023stat (5.1%)
L. Tauscher, Basel

17. Lifetime t = 2.85 [fs] PBr = 0.447 ± 0.023stat + 0.44stat - 0.38stat
L. Tauscher, Basel

18. Systematics Systematics from: Normalization (C vs. nC determination)
Cut (Qcut) for PBr determination
Multiple scattering simulation
Signal shape simulation
Many others
L. Tauscher, Basel

19. Lifetime with systematics
+ 0.48
t = [fs]
- 0.41
L. Tauscher, Basel

20. Dual target technique Method to get rid of uncertainties from
normalization, multiple scattering and shape
Two targets:
standard single layer Ni-target
multi-layer Ni-target with the same thickness as the standard target, but
segmented into 12 equally thick layers at distances of 1.0 mm.
Both targets have the same properties in terms of:
production of secondary particles by the beam,
correlated and uncorrelated backgrounds (Nback),
produced atoms (NA),
integral multiple scattering,
Measuring conditions
But: break up Pbrm is smaller than Pbrs because of enhanced annihilation.
L. Tauscher, Basel

21. Normalization-free determination of t
Nm(Q) = Pbrm NA SA(Q) + Nback(Q) Ns(Q) = Pbrs NA SA(Q) + Nback(Q)
SA(Q) : normalized Monte Carlo shape function for the atomic break up signal
Signal shape: Ns(Q) - Nm(Q) = (Pbrs - Pbrm)NA SA(Q)
Background shape: Ns(Q) - r Nm(Q) = (1-r) Nback(Q) = (1-r) B(Q)
r = Pbrs / Pbrm
B(Q) = w BC(Q) + (1-w) BnC(Q)
B: Monte Carlo shape function for the background, normalized to the no. of background-events
t from r (independent of normalization)
L. Tauscher, Basel

22. Rate combination for signal (2002) (preliminary)
Ns(Q) - Nm(Q) = (Pbrs-Pbrm)NA = 825 ± 140 events
signal shape well reproduced
L. Tauscher, Basel

23. Rate combination for background (2002) (preliminary)
r = 1.86 ± 0.20stat
Background shape from MonteCarlo is consistent also at low Q
L. Tauscher, Basel

24. Lifetime single/multilayer (2002) (preliminary)
Result:  = [fs]
L. Tauscher, Basel

25. Number of Atomic pairs (approx.)
outlook
Number of Atomic pairs (approx.)
Pt1999
24 GeV
Ni2000
Ti2000
Ti2001
Ni2001
Ni2002
20 GeV
Ni2003
Sum
Sharp selection
280
1300
900
1500
6500
2000
2600
16600
Loose selection
(high background)
27000
Full statistic probably sufficient to reach the goal of 10% accuracy
L. Tauscher, Basel

26. DIRAC has achieved a lifetime measurement with 16% accuracy
Conclusion
Using a subset of data
DIRAC has achieved a lifetime measurement with 16% accuracy
Systematic errors have been studied
Normalization consistency in Q, QL
Cut uncertainties
Multiple scattering
Shape uncertainties
Many others
Systematic errors < statistical errors
Full statistics sufficient to reach 10% accuracy (st = 0.3 fs)
L. Tauscher, Basel