1. Gas Pixel TRD/Tracker With the support of the TRT collaboration
F. Hartjes, M.Fransen, W. Koppert, S.Konovalov, S.Morozov, N. Hessey,
A.Romaniouk, M. Rogers, H. van der Graaf.
With the support of the TRT collaboration

2. Motivation.
A combination of a pixel chip and a gas chamber opens new opportunities for particle detectors.
Pixelization of the information from the particle track offers new possibilities to combine in one detector many features:
1. Precise coordinate measurements (well below diffusion limit). 2. Vector track reconstruction. 3. Very good multi-track resolution. 4. Powerful pattern recognition features. 5. L1 trigger possibilities for stiff tracks. 6. Enhanced Transition Radiation separation 7. dE/dX measurements.
TR clusters

3. What is the GasPixel detector
~ -500 V
Pixel chip
Mesh
Gas volume
Drift distance 16 mm
55 mm
~50 mm
Track image on the pixel plane

4. Potential applications: everywhere. Issues important for TRD:
But might be interesting for the Inner tracker in the collider experiments.
Detector shell occupy the outer radii of the Inner Tracker and contain at least two layers which can be interleaved with TR radiator
What is new?
A vector tracking !
Precision space points:
X ( f-direction) and Y (h – direction)
2. Vector direction f and h
3. Enhanced TR and dE/dx measurement function f
Issues important for TRD:
Limited space
Number of layers 1 or 2
Effective TR radiator needed
Minimum drift space (Rate constrains)
May not be optimal for TRD
Hope to work with each cluster

5. Particle ID: use TR and dE/dX information
Total
Cluster counting
Relativistic rise vs d-electron energy
B. Dolgoshein, NIM A326 (1993) 434
Number of primary
Collisions vs gamma
Number of primary collisions per cm of Xe
Factor of 1.36
ln g MC
Number of clusters above ~1 keV
Factor of 2
Pions
Plato Fremi : Mean 4
Min. ionising: Mean 2.7
Factor of 1.48
Detailed knowledge of the ionisation structure of the track gives more information for the particle ID.
Although track the length of the drift volume is the main parameter!
dE/dx, arb. units

6. X-ray absorption probabilities. Something more we may find useful.
Particle ID: TR registration: The main issue is to separate dE/dX from TR
Use the low:
X-ray absorption probabilities.
Something more we may find useful.
1+2 == 3 6% 3% 1%
TR clusters
Still there is dE/dX under the cluster
B. Dolgoshein, NIM A326 (1993) 434

7. First test beam studies: Set-up Exposed to PS 5 GeV particle beam
No detailed optimization done.
InGrid TimePix detector, 14 x 14 mm2, 256 x 256 pixels (pixel size 55 mm)
Two operation modes: time measurements and time-over-threshould measurements (amplitude)
Exposed to PS 5 GeV particle beam
Xe/CO2 (70/30) mixture.
L = 16 mm
0.05 mm V0 V1
L for TR absorption ~ 16 mm
TR-Radiator
Beam
p, el E
25o e-
Length - 19 cm
PP foils of 15 mm spaced by 200 mm h
Pixel plane
MC simulations studies were performed and compared with the test beam data using the same analysis tools.

8. Test beam studies. May 2008 test chamber: Operation parameters
For the gas mixture 70%Xe+30%CO2
Total drift time ~ 300 ns
Ion signal ~ 80 ns
Transverse diffusion sT ~240 mm/cm1/2
Longitudinal diffusion sL ~120 mm/cm1/2
May 2008 test chamber:
InGrid technology
Drift distance= 16 mm
Vdrift = 3800 V
Edrift= 2000 V/cm
Vamp ~470 V
Gas gain ~ 800 – 3500
Protection layer 30 mm
Amplification gap - 50 mm
Orientation: 25 degree to the beam and horizontal.
Gas Mixtures
Ar/CO2
Xe/CO2
He/Isobutane
DME/CO2
Mesh
Si prot. layer
Pixels
Charge
Electronics operating threshold: ~ 800 el.
Gas gain range: – 1800
Geometry + induced charge effect
Effective threshold > 1600 el.
OR > 1 primary electron.
Estimated efficiency for 1 primary electron was ~30%

9. Test beam studies (5 GeV). Comparison EXP data with MC
Particle tracking including all interactions + geometry - GEANT3
Ionisation losses PAI model
TR generation ATLAS TRT code
After all only 3 basic parameters were tuned to fit the MC model to the experimental data:
Diffusion coefficient - was chosen to match a track widths.
Threshold - tuned to satisfy number of fired pixels.
Energy scale - scaled using calibration factor.
Exp MC

10. Particle Identification.
Time-Over-Threshold representing a signal amplitude
was measured for each pixel
Exp. data
Amplitude (Arb. units Log scale)
Pion
Electron
On the hip map color represents the amplitude or ToT information
Amplitude (Arb. units Log scale)
Pion
Electron

11. Particle Identification
Methods
A total energy method
The amplitudes of all the pixels on track are summed.
2. Cluster method
Sum of the signals projected to the reconstructed track (binning is 55 mm).
Exp. data
rebin
Cluster
threshold
Cluster energy

12. Particle Identification: Total energy.
Test beam results
Particle Identification: Total energy.
Electrons
Pions
Distribution of a total ionization on the particle track for electrons and pions.
Comparison MC and Data

13. Particle Identification: Total energy.
Test beam results
Particle Identification: Total energy. MC
Exp
Distribution of a total ionization on the particle track for electrons and pions
Electron/pion separation.

14. Particle Identification: Cluster counting
Test beam results.
Particle Identification: Cluster counting
Electrons
Pions
No clusters
29/368 – 7.9%
No clusters
711/ %
Cluster energy distribution on the particle track for electrons and pions

15. Particle Identification: Cluster counting
Test beam results
Particle Identification: Cluster counting
Cluster energy distribution on the particle track for electrons and pions
Electrons 1 cluster
175
Cluster energy (arb units)
Events
For PID additional CUT on the cluster
energy is applied,.
Electrons and pions
1 cluster
177
Electrons 2 clusters
Events
Only 1% of pions have 2 clusters
Maximum cluster energy (arb units)

16. Particle Identification
Rejection factor
Pion registration efficiency as a function of electron efficiency.
For 90% electron efficiency pion rejection factor is ~7 for the total energy method.
Total energy
Pion registration efficiency as a function of electron efficiency for the total energy method and Cluster counting method.
Cluster counting method has a bit larger rejection power.

17. Particle Identification Rejection factor with cluster counting method
Pion registration efficiency as a function of electron efficiency for 1 and 2 layers of the detector. Cluster counting method.
TRD with two detector layers (total thickness ~ 40 cm) allows to achieve rejection factor of ~ 50 for 90% electron efficiency.
The radiator is one of the biggest issues for the compact detectors.

18. TR generation and absorption (test beam set-up).
Particle ID.
TR generation and absorption (test beam set-up). MC MC
Average:
2.5 TR photons.
No TR photons:
12.5%
Number of photons at the radiator output
Number of detected photons in one layer per event (test beam chamber)

19. TR generation and absorption.
Particle ID.
TR generation and absorption. MC MC
20 keV
Spectrum TR absorbed in 1.6 cm thick chamber
Spectrum TR absorbed in 4 cm thick chamber
Number of detected photons in inserts.
For 4 cm chamber 6% electrons have no TR cluster

20. TR generation and absorption
Particle ID.
TR generation and absorption
For 1-2 layer detector the biggest problem is the compact radiator which should provide a soft photon spectrum.
Be - Li materials? MC
Polypropylene radiator
Beryllium radiator
Number of absorbed photons as a function of the pass in the gas
in 1.7 cm chamber (70%Xe + 30% CO2).

21. TR generation and absorption
Particle ID.
TR generation and absorption
Radiators: 17mm/100mm/9 cm MC
Polypropylene
Beryllium
Spectrum of absorbed TR for different types of radiator.
There is a way to get a compact radiator with good TR yield
But much more optimisation and test work is required.

22. Technique used for the tacking property studies.
How well we understand the results?
The answer is in MC simulations
which use the same technique.
Reminder:
Diffusion coefficient and threshold in MC were chosen
to satisfy a track widths and number of fired pixels
observed in test beam.
Only tack projection to the pixel plane is used
(no time information taken into account). Y X
« f » MC
What we can measure with the beam without reference detector? Represent the chamber as two independent interleaved units and measure parameters of two “tracks” (pseudo-tracks) reconstructed independently!
Track 1
Track 2
Chamber 1 Chamber 2
A real resolution is about ½ of the measured one with this method.

23. Results strongly depend on the gas properties.
Tacking.
Results strongly depend on the gas properties.
DME/CO2 (50/50),
19o incident angle
Xe/CO2 (70/30),
19o incident angle
Low diffusion gas
Electron NO TR
Electron with TR
Apart of detector geometry a track reconstruction accuracy depends on diffusion, single electron efficiency and also
on a presence of the TR clusters.

24. Test beam results: Tacking.
Low diffusion gas.
Particle
Chip
DME/CO2 (50/50), 20 mm drift,
Incident angle of 10o.
s = 0.6o
s = 11.5 mm
At 10o incident angle an angle measurement accuracy
of 0.6o for the track projection would mean:
At B=2 T this corresponds to 15% momentum measurement accuracy for Pt of 40 GeV
with one layer of the GasPixel tracker

25. Test beam results: Tacking.
Xe - mixture
Xe mixture: Pions, 5 GeV,, 55 mm pixel, 240 mm diffusion.
25o of incident angle
Intrinsic special accuracy of the gas pixel detector (MC and EXP).
Spatial accuracy
MC intrinsic
MC based on pseudo tracks
EXP
Effective electronics threshold (primary electrons)
Spatial accuracy (mm)
Angular resolution
Angular resolution (o)
MC based on pseudo tracks
MC intrinsic
MC: for the diffusion coefficient factor of 2 less
EXP
Effective electronics threshold (primary electrons)
Intrinsic angular resolution of the gas pixel detector. (MC and EXP).
30 mm space accuracy and 0.6 o angular resolution obtained in the experiment for Xe based mixture (without radiator) .
It is a bit worse (~55 mm, 0.9o) for electrons with TR clusters but effect still to be studied in details

26. Conclusions.
Gas pixel technology offers new possibilities for the tracking and particle ID detectors
Combining these detectors in one will certainly reduce tracking properties for the particles which produce TR but still still very much adequate to the electron tracking.
For the High Energy Physics Experiments data flow limitations require some data processing at the Front-End level. This also makes natural the use of these devices as L1 track trigger.
Definitely more test–beam and MC studies are needed to understand all the details and accumulate knowledge which would allow to produce a real detector.