1. SAS TRD Possible TRD configurations for PID up to the TeVs energies fig.s for this talk taken by: B.Dolgoshein Transition radiation detectors -NIM A326(1993) N.Giglietto, C.Favuzzi, M.N.Mazziotta and P.Spinelli Transition radiation detectors for particle physics and astrophysics -Nuovo Cimento 5-6 (2001) p.spinelli CERN 2/10/215 1. Some history (I hope not too much, but it is necessary) 2. Conventional configurations : what problems ? 3. An almost unconventional idea: but not yet proved…

2. In TRDs the faster radiating particle emits TR, observed as a pulse height distribution in the detector overcoming that of the slower one thresholds for TR emission  TH = 2.5 d 1  p Slower particle pulse heights TR saturation  TH = 1.2  TH (d 2 / d 2 ) 1/2

3. The TR thresholds for e/  pairs are really far apart ! not the same for  /k or k/p pairs…

4. 90% cut on electron distribution  particle to identify 90% electron acceptance a few % pion contamination  particle to reject Harris et al. 1971 p.h. distributions are skewed!

5. 2% @90% 4% @90% 6% @90% Better averaging on more sets e.g. 4 sets (5 Gev/c) pions electrons Likelihood technique improves performances

6. Cluster counting method Dolgoshein, Fabjan et al. 1981 the distributions become nearly poissonian !! therefore : 1% @ 90% acceptance means ≈ 3.6  0.1% means ≈ 4.4 

7. Dolgoshein et al. HELIOS TRD (1986) 8 sets L = 70cm The best so far… << 10 -3

8. good pion/electron separation  better than 10 -3 but what about pion/kaon/proton identification?

9. European Hybrid Spectometer CERN (1980) 20 modules (Mylar-Xe) L= 3.5 m What about  /p or  /k separation ? average over 20 modules

10. … a few % contamination @ 90% acceptance for any pair, but note: no k/p separation quoted…

11. what Dolgoshein did in 1981 with cluster counting?

12. ≈ 5%  contamination into K sample but with just L ≈ 70 cm !

13.  k becomes ≈ 1% with 24 sets, L = 1.32 m but note: k/p separation is not quoted … What happened later at higher energies?

14. Fermilab E 715 (1983 ) @ 250 GeV/c  contamination = 6.10 -4 @ 92.5% e - acceptance - 12 sets - L=3.6 m ≈ 2 clusters

15. Fermilab E 769 @250 GeV (1991) 24 sets - L = 2.79 m  contamination =2% @ 87% p acceptance simulation at @ 500 GeV/c k(2%)  (98%)  contamination = 3% @ 90% k acceptance kaonspions k/p separation ability not quoted…

16. radiator C-fibers straw tube detector read out electronics 16 sets TRD for PID (  /p @280 GeV) CERN NA57 Bari Group very compact, L= 90 cm p contamination = 1% @ 90%  acceptance

17. Possible TRD configurations for PID up to the TeVs energies three configurations based on polyethylene radiators: 100-200 foils 15  - 300  gap 100-200 foils 50  - 1mm gap 100-200 foils 100  - 2 mm gap detectors: gaseous or solid state (see next talk) We discuss their PID capability (with a naive procedure ) e.g. starting from 100 TR photons (clusters) collected on the average at saturation; then: MC simulation, Poisson distribution assumed for photons, rejection calculation

18. …some normalization problems : I think we over-estimate the TR photons at saturation for thicker radiators from our MC

19. -we assume that the collected photons are ≈ 2 per TRD set at best  we need 50 sets ( each with 200 foils radiator) to get 100 TR photons - 200 ?

20. The thresholds move to higher energies we assume that the collected photons are ≈2 per TRD set at best - 200 ?

21. The TR thresholds still move at higher energies The proton shows up at 1 TeV and saturates at 5 TeV we assume that the collected photons are ≈2 per TRD set at best - 200 ?

22. Can we roughly estimate the PID capabilities in a simple way? naively by means of the cluster counting method

23. Kaon proton k/p separation at 1 TeV - 50 TRD sets 15  radiator configuration total cluster number 90% acceptance cut proton contamination ≈10 -5

24. 16 sets TRD (15  ) + 16 sets TRD (50  ) + 16 sets TRD (100  ) very good (<< 10 -3 )  /p separation in the 0.2-5 TeV range 10 -3

25. very good (<< 10 -3 )  /k separation in the 0.2-2.5 TeV range 10 -3

26. Poor K/p separation ≈ 2 % at best in the 1 - 3.5 TeV range …but 10 -2

27. 15  TRD – 16 x (6 cm radiator + 2 cm Xe)  L = 1.3 m (0.17X 0, 0.08 I ) 50  TRD – 16 x (20 cm radiator + 5 cm Kr ? )  L = 4 m (0.56X 0, 0.26 I ) 100  TRD –16 x (40 cm radiator + 10 cm Xe? )  L = 8 m (1.15X 0, 0.55 I ) namely  L tot = 13,3 m – 1.88X 0 – 0.9  I …huge and heavy TRD like a pre-shower detector! poor k/p separation if we choose a 12 + 12 + 48 configuration 

28. still good  /p separation in the 0.2 - 5 TeV range

29. still good  /k separation in the 0.25 -3 TeV range

30. no encouraging change: we get now 2% at best at 5 TeV

31. Conclusions…. Every configuration is “almost -conventional” but poses some serious problems: -on the detectors : thick gaseous (kripton or xenon gas) or solid state (silicon)? see next talk -on the performances: adequate k/p separation is never achieved at most energies -on the layout: too long configurations and too high radiation and interaction lengths !! Unless we apply to an old idea of Boris Dolgoshein…

32. M.Deutschmann et al. Partice identification using the angular distribution of transition radiation N.I.M. 180 (1981) 409-412 Single surface  = 6000 Single foil  = 2000 50-70 cm  ≈ 1/    x ≈  mm for  = 10.000 -1000 beware:  mult. scatt. ?

33. Regular radiator - 50 foils  =1000 The TR photon angular distribution exhibts sharp peaks around  1/  because of interfering effects (from theory!!) (note: Coulomb scattering is ≈1  rad for TeV hadrons << TR angle )

34. Pions and electrons @ 3.5 GeV detected in two drift chamber at 50 cm distance from the radiator. Note: most of the hit smearing in both cases is due to multiple scattering (10 times greater than TR angle)

35. Can we envisage a “miniaturized” ring imaging TRD = RITRD? now we have more advanced pixel detectors ! (see next talk) -we can collect with 10 sets radiator/pixel detector ≈ 20 TR photons (better than a conventional RICH) to overlay on a unique frame to reconstruct a ring -conventional 15  foil radiators to let any hadron to radiate + 1 m “espansion distance” in helium  L ≈ 10 m, still long, but X 0 and  will be negligible! -pixel size 50  x 50  ? (spatial resolution optimized by centroid calculation) -the momenta, namely the rings radii per each kind of particle, are fixed by the calorimeter: at 1 m of espansion distance  R p = 1mm @  = 1000 (1 TeV proton) or R k = 0.5mm @  = 2000 (1 TeV kaon)

36. 3 TeV Kaon - proton angular separation (10 TRD sets) Now we separate angular distributions instead of cluster distributions

37. 5 sets, L = 5,3 m, R p/k  15%

38. 10 sets, L = 10.6 m, R p/k  2.5%

39. 15 sets, L = 16 m R p/k  0.3%

40. final conclusions We need to explore with appropriate MC calculation the performances of these solutions (see nex talk) We need to carry out real tests on a beam asap

42. 500 GeV - 5 TeV : good separation (<< 10 -3 rejection) for  /k and  /p poor separation of k /p (restricted just at 4 -5 TeV ≈ 10 -3 rejection) if we take 50 sets, each 40 cm + 10 cm (Xe),  L = 25 m (note: 2.5X 0, 1.5 I ) we assume 100 TR photons at saturation 10 -3 contamination

43. Still good p/p separatio

45. 100 GeV - 1 TeV : good separation (<< 10 -3 rejection) for  /k and  /p poor separation of k/p (restricted just at 0.5 -1.2 TeV ≤ 10 -3 rejection) if we take 50 sets, each : 6cm +1,5cm Xe  L = 3.75m (note: 0.5X 0, 0.25 I ) we assume 100 TR photons at saturation 10 -3 contamination

46. good separation (<< 10 -3 rejection) for  /k (0.3 TeV - 2.5 TeV) and  /p (0.3 TeV- 5 TeV) poor separation for k/p (restricted just at 2.5 - 4 TeV ≤ 10 -3 rejection) if we take 50 sets, each : 20 cm + 5 cm Kr  L = 12.5m (note: 1.25 X 0, 0.75 I ) we assume 100 TR photons at saturation 10 -3 contamination

47. The p/k separation is problematic at 10 -3