Goal of the project The main goal of this project was to realise the reconstruction of α tracks in an optically read out GEM (Gas Electron Multiplier) based Time Projection Chamber (TPC). Secondary goal was to initialise a series of systematic studies on the scintillation of α particles in Ar/CF 4 (80-20%) mixture. Track reconstruction is needed for primary scintillation studies as only tracks fully contained can be considered. A vetoing and trigerring logic was built for the TPC from NIM modules. TPC operation The signal of the tracks is recorded by a Photo Multiplier Tube (PMT) and a Charge-Coupled Device (CCD) camera. When an α particle is emitted by the decaying 220 Rn it creates scintillation which is recorded by the PMT. This is called the primary scintillation. The α particle also ionizes the gas. The electrons are drifted to a triple-gem stack using an electric field. The electrons are multiplied inside the GEMs by an order of 10 5. These avalanches also create scintillation light, called the secondary scintillation. This light is recorded by the CCD camera and the PMT as well. The image recorded by the camera is the two dimensional projection of the track. The waveform of the PMT is used to determine the orientation of the track in the direction perpendicular to the plane of the GEMs (Z direction). The time difference is measured between the primary scintillation and the beggining of the secondary scintillation. This corresponds to the drift time of the electrons created closest to the GEMs. Additionally the time difference between the primary and the end of the secondary scintillation is determined. This corresponds to the drift time of the electrons created furthest to the GEMs The origin of the track is derived using the position of the Bragg-curve. Raw data The PMT is read out using a 625Zi Lecroy oscilloscope. Both the camera and the oscilloscope is connected to the computer. A MATLAB code records the raw data and if needed it also recontructs the tracks automatically. As an example you can see an image and the waveform corresponding to the same track in the following figures. Calibration In order to obtain the real position of the track end points calculated from the PMT waveform the drift velocity of the electrons in the used gas mixture and applied electric field had to be measured. We have measured the time difference of the primary and the end of the CERN sum. stud. report 1 Gabor Galgoczi
Figure 2.1: Greyscale image taken of the secondary scintillation of α tracks Figure 2.2: Waveform of the PMT signal containing the primary and secondary scintillation secondary scintillation. We knew that the electrons created close to the cathod would take the most time to drift to the GEMs. In fig. 2.1 you can see the histogram of the time difference of the primary and the end of the secondary scintillation. The peak corresponds to the electrons created at the cathod. Therefore the drift velocity of the electrons is: v diff = 8.91cm 0.853µs = 1.04 107 cm s Figure 2.3: Maximal drift time distribution of a few thousand events The distances in the plane of the GEM were calibrated using the pitch size of the GEM. We used low binning and long exposure. This way we were able to observe the pattern of the holes in the pictures. We have plotted the intensity pixel value profile along a line of holes. The pitch size of the GEM is 140 µm. The conversion constant between the pixel distance and the real distance turned out to be 42 µm. This calibration was possible because the electric field pixel lines, which guide the electrons from the α track to the GEM are uniform. CERN sum. stud. report 2 Gabor Galgoczi
Figure 2.4: A high resolution image of the secondary scintillation with visible pattern of the GEM structure Figure 2.5: Grayscale value along a line of hole centers Track reconstruction Image analysis In order to reconstruct tracks we had to recognise individual tracks in images. First our algorithm calculates the histogram of the pixel values of the picture. A Gaussian is fit to the first part of the histogram, which corresponds to the background pixels of the image. Then a cut is introduced at 5σ. Figure 2.6: Greyscale image of an α track with a visible Bragg-peak Figure 2.7: Fitted Gaussian to the data points of the pixel value distribution corresponding to the background pixels The resulting image contains the pixels of all tracks in it and some other pixels as well, which mostly correspond to scintillation light scattered back from other parts of the detector. The so called floodfill method is used to distinguish between different tracks and to throw away pixels which doesn t belong to any of them. If there are too many identified tracks, then the CERN sum. stud. report 3 Gabor Galgoczi
event is thrown away because we can not be sure which track the waveform belongs to. A curve is fitted to the points of the track in order to determine the coordinates of the track in the image. Figure 2.8: Binary image produced by introducing a cut from the distribution of pixel values Figure 2.9: Pixels coloured with red correspond to the identified track Waveform analysis The signal of the PMT is read out by an oscilloscope. In order to digitize both the primary and the secondary scintillation with a sufficient resolution, the signal is read out by two channels with different voltage resolution. First our algorithm recognises the secondary scintillation peak in the small resolution waveform. Then it determines the two times when the value of this signal was 10 % of the maximum. As the peak is very sharp these times can be considered as the beggining and end of the secondary peak. Afterwards the high resolution waveform is divided into two parts when the secondary scintillation starts. Only the first part is considered by the algorithm while it searches for a primary peak. The time difference between the primary and the beggining of the secondary scintillation defines the distance in the Z direction between the GEM and the closest point of the track. The time difference between the primary and the end of the secondary scintillation defines the farthest point of the track. CERN sum. stud. report 4 Gabor Galgoczi
Figure 2.10: Waveform corresponding to the an α track, which was parallel to the GEM plane Figure 2.11: Waveform corresponding to the an α track, which was almost perpendicular to the GEM plane Direction of the track The beggining of the track corresponding to its origin, the 220 Rn is identified by the place of the Bragg peak. In the images, the Bragg peak is clearly distinguishable. In the PMT waveforms the secondary scintillation peak is assymetrical. If the absolute maximum happened earlier then the Bragg peak was closer to the GEMs than the origin of the track. If the absolute maximum happened earlier that indicates that the origin of the track was closer to the GEMs. In the images, the pixel values were projected onto the fitted line of the track. The place of the maximum pixel values is clearly distinguishable this way (fig. 2.13). The number of pixels projected to each point along the fitted line slightly increases around the Bragg peak. Figure 2.12: Number of pixels projected to the fitted line of a track Figure 2.13: Pixel value projected to the fitted line of a track Reconstructed tracks The image analysis algorithm derives the coordinates of the end points of the track in the plane of the GEMs. The PMT waveforms provides the coordinates of these points in the CERN sum. stud. report 5 Gabor Galgoczi
perpendicular direction to the GEMs plane. Matching up the position of the Bragg peak in the image and the PMT waveform our algorithm derives the direction of the track as well. The origin of the track is the black dot (fig. 2.14.). Figure 2.14: A reconstructed track inside the field shaper rings Track length distribution The track reconstruction algorithm calculated the length of the tracks as well. It assumed the tracks to be straight. The distribution of the track length is broad (fig. 2.15). The reason is that the length of an α track is comparable to the size of the field shaper rings. Therefore less than half of the tracks are fully contained. Selecting these tracks for primary scintillation studies is vital. Figure 2.15: Track length distribution, the peak was fitted by a Gaussian with a polynomial background Figure 2.16: Track length distribution of a monoenergetic α beam A Gaussian curve with a cubic polynomial was fitted to the peak of the track length distribution. The cubic polynomial corresponds to events, which were not fully contained. The events above 50 mm belong to an event when two α tracks originated from the same point in the opposite direction. This seemed to be one long track for the algorithm. CERN sum. stud. report 6 Gabor Galgoczi
A GEANT4 simulation was built in order to validate the measurements. The peak of the simulated distribution is very close to the one from the measurements. The full width at half maximum is wider because there are α particles created with two different energies while the simulation was done using monoenergetic particles. CERN sum. stud. report 7 Gabor Galgoczi
Scintillation studies Atmospheric pressure The primary scintillation of different gas mixtures at different pressure regimes is a topic, which is intensively studied. The TPC we are using was built in order to study this topic. The primary scintillation peak was integrated. The result was divided by the single photon response. (See subsection 5.2. for details.) We took data for 20 hours, which included 80000 events. 7000 tracks were reconstructed succesfully. The center in the Z direction of all of these tracks can be seen in fig. 3.1. About half of the tracks az very close to the cathode. The possible reason for this is that the Po ion created by the Rn decay drifts to the cathode and α decays there. The avarage number of photons in fig. 3.2. Figure 3.1: Primary scintillation light versus the center of the track in Z direction Figure 3.2: Avarage number of photons produced by tracks with different centers in Z direction For the primary scintillation studes we have choosen the tracks which had the following parameters: Were 45-50 mm long Had primary scintillation Primary scintillation peak was maximum 1 µs before secondary (in ambient pressure) This left 1871 events. CERN sum. stud. report 8 Gabor Galgoczi
Figure 3.3: Primary scintillation light versus the center of track with a length of 45-50mm in Z direction Figure 3.4: Avarage number of photons produced by tracks with a length of 45-50mm and different centers in Z direction Summary The reconstruction of α tracks in an optically read out GEM (Gas Electron Multiplier) based Time Projection Chamber (TPC) was realised as the main goal of this project. A series of systematic studies on the scintillation of α particles in Ar/CF 4 (80-20%) mixture were realised. A vetoing and trigerring logic was built for the TPC from NIM modules. CERN sum. stud. report 9 Gabor Galgoczi
Appendix Vetoing system for the TPC PMT Discriminator Timer Camera shutter out Coincidence Matlab Matlab Arduino output Reset Timer Arduino input Timer Timer Coincidence Osci. trigger Camera triggering Figure 5.1: Triggering and vetoing system of the TPC, red arrows represent veto Single photon peak We measured the number of photons recorded by the PMT during the primary scintillation of α particles. For this we derived the single photon response of the PMT. We measured the integral of PMT waveforms. We set the trigger treshold low in order that most of the waveforms corresponded to noise and the events when an electron is extracted by the electric field from the first dinode. The former events have the same signal as the single photons. If the avarage light yield of the PMT too high, then it starts to saturate. This would make primary scintillation studies unreliable. We wanted to be sure that we avoid this case. Therefore we repeated the same measurement using different settings. Regarding saturation the most important is the secondary scintillation as this produces the majority of photons reaching the PMT. No GEM Voltage res. [ mv ] Trigger [mv] BW [MHz] Sampling [GHz] Peak place [pwb] div 1 Off 2-0.6 200 10 55.59 ± 1.07 2 Off 2-0.6 200 5 56.59 ± 1.27 3 On 2-0.6 200 10 57.00 ± 1.05 4 On 5-0.8 200 10 57.81 ± 0.74 Table 1: Single photoelectron response of the PMT with different settings CERN sum. stud. report 10 Gabor Galgoczi
We fitted a Gaussian with an exponential background. The latter one corresponds to the tail of the distribution of the noise. The position of the center of the peak did not change. Thus we proved that the PMT did not saturate during the measurements. Figure 5.2: Histogram of waveform integrals with the single photon response peak, setting no. 1 Figure 5.3: Histogram of waveform integrals with the single photon response peak, setting no. 2 Figure 5.4: Histogram of waveform integrals with the single photon response peak, setting no. 3 Figure 5.5: Histogram of waveform integrals with the single photon response peak, setting no. 4 CERN sum. stud. report 11 Gabor Galgoczi
GEM photon yield calibration The photon yield decreases approximately exponentially by increasing the pressura and keeping the applied voltage on the GEMs. Therefore in order to be able to take images of the secondary scintillation, we had to determine the voltages needed to have the same photon yield. The photon yield of x-rays produced by an Fe-55 (12 MBq in 2006) source was used as reference. The oscilloscope avaraged the secondary scintillation signal of 1000 x-rays. The pressure was increased to 1.3 bar. We measured the voltages needed to have the same secondary scintillation signal. Afterwards the pressure was lowered by 50 mbar and the necessary voltages were derived. We repeated this process until we reached 1 bar. Pressure [mbar] V 1 [V] V 2 [V] V 3 [V] Amplitude [au] I 1 [µa] 1000 1570 380 400 117 ± 30 80 1050 1600 388 414 115 ± 28 82.25 1100 1640 385 420 115 ± 26 84.5 1150 1670 400 430 116±27 86.2 1200 1710 410 437 116±28 88.3 1250 1750 430 450 115±30 89.65 1300 1780 435 460 117±30 91.3 Table 2: Voltages needed to measure the same PMT waveform amplitude of the secondary scintillation of X-rays The photon yield at the initial voltages decreased approximately exponentially by increasing the pressure (fig. 5.6.). Figure 5.6: Photon yield of the secondary scintillation with different pressures applied CERN sum. stud. report 12 Gabor Galgoczi