1. Particle visualization and identification with Timepix detectors
Jess Flores
Stony Brook University
Mentors: Jim Pinfold, Albert de Roeck, Tom Whyntie

2. Project #1 goals
To use information from Timepix detectors to analyze patterns of charge. This will aid in the identification of different particle species.
Sensor is bump-bonded onto an application0soecific integrated circuit (ASIC). As a result, the pixels have a pitch (width) of 55 microns and form a 256x256 grid with a sensitive area of 1.98 cm^2.

3. Timepix detector
The Medipix2 device was originally used to count single X-ray photons for medical applications.
With the development of the Timepix mode, the Timepix detector was used as a charged particle detector as this mode offered the time-of-arrival info about incident photons.
The reverse bias p-n junction of the silicon-based detector creates a charge-sensitive volume of space-time that generates a signal in the presence of ionizing radiation.
The pixelated mature of the patterns formed by charge deposited in the depletion zone readily facilitates the visualization of incident radiation.
When a penetrating heavy ion traverses the Si and produces a core of charge carries surrounded by a halo of carriers associated with the track structure, the diffusion of the collected charge by the time it reaches the Medipix2 pixels is generally cone shaped with the charge from the most distant part of the track being disbursed the widest and that from nearest to the pixel contact remaining closest to the track itself.

4. Previously on the last episode…
First algorithm : had difficulties in differentiating SW, brancher and crossover clusters.
Wanted to come up with additional cluster properties to better differentiate between cluster types
Brancher
Process:
Could be a
Decay
Crossover
Detector could
Be reading the
Tracks left by
2+ particles as
A single particle.
Looper
Beta
Radiation.
Slug
Heavy
Ions and/
Or nuclear
Fragments.
Straight Wiggly (SW)
Fast particles losing KE
To Coulomb interactions
With the lattice.
Boxy
Possibly
Caused by
Neutron-
Detector
interactions

5. Cuts based on cluster properties
Cluster properties: linearity, radius, density, inner pixel fraction, etc
Figure: Cuts to classify clusters.

6. But why friend zone some clusters?
Size < 20
Just noise
Not interesting
Size > 20

7. Cuts based on cluster properties
Does it contain bulk pixels or not?
Compact/linear pixel distributions expected to contain at least one pixel entirely surrounded by other pixels.
The shorter more compact boxys and slugs were isolated from the longer and thinner, or looping clusters. Boxys and slugs contain such bulk pixels, but loopers do not.
Figure: Cuts to classify clusters.

8. Cuts based on cluster properties
How to tell boxys and slugs apart?
The unweighted radius of a cluster was defined as the distance from the unweighted cluster centre to the furthest pixel.
The density of a cluster was defined as the ratio of the area occupied by pixels contained within this radius and the total area of the circle with an equivalent radius.
A cluster was considered slug- or boxy-like if the density ρ > 0.1.
Figure: Cuts to classify clusters.

9. Cuts based on cluster properties
How to tell boxys and slugs apart?
Aspect ratios of slugs was significantly different from 1 while that of boxys approached unity.
Figure: Cuts to classify clusters.

10. Cuts based on cluster properties
How to sort lower density clusters?
When the topological distribution of pixels was graphed in x-y plane, branchers and crossovers were not described by single-valued continuous function.
For multiple pixels with the same horizontal x-coordinates, these clusters could be separated from straight and curved clusters if we required a spread of more than 3 pixels in the plane having vertical y-coordinates.
Figure: Cuts to classify clusters.

11. Results
87% efficient!
A lot better than the 35% efficiency from last time.

12. Trip pics