Construction of a Gas Electron Multiplier (GEM) Detector for Medical Imaging N. N. Mondal, S. Chattopadhyay, M. R. Dutta Mazumdar, A. K. Dubey and Y. P. Vioygi. Physics Group, Variable Energy Cyclotron Center (VECC) 1/AF, Bidhannagar, Kolkata 700064, India. Abstract: A prototype Gas Electron Multiplier (GEM) detector is under construction for medical imaging purposes. A single thick GEM of size 10 10 cm 2 is assembled inside a square shaped air-tight box which is made of Perspex glass. In order to ionize gas inside the drift field two types of voltage supplier circuits were fabricated, and array of 2 4 pads of each size 4 8 mm 2 were utilized for collecting avalanche charges. Preliminary testing results show that the circuit which produces high voltage and low current is better than that of low voltage and high current supplier circuit in terms of x-ray signal counting rates. 1. Introduction: Gas Electron Multiplier (GEM) detector system was first introduced by F. Sauli [1] at CERN in 1997. A GEM is a composite grid consisting of two metal layers separated by a thin insulator, etched with a regular array of pin hole (diameter of a few hundredth of mm). This sandwiched package is kept inside a non explosive gas (mainly inert gas) or mixture of gases with optimized proportion under normal temperature and pressure in general. A suitable potential difference is applied between the two copperplates, which generate drift field along the holes perpendicular to the plates. When energetic photons or charged particles enter into this drift region gas is ionized and create pair of ions. Fig. 1: A schematic diagram of a standard double GEM detector (not in scale). 2D image reconstruction is
possible by using the data X- and Y-grids. Electrons are accelerated by the drift field and ionize more electrons and those avalanched electrons cloud are accumulated on the anode plates, made of copper strips/pads on the printed circuit board (PCB) and electronics. Principle of GEM is similar to that of a photomultiplier tube (PMT). Single, double, triple or multiple GEM detectors are possible to develop in order to improve gain, better time and energy resolutions. A schematic view of a double GEM detector is depicted in Fig. 1. The upper plate of GEM is a beryllium window of a few µ m thicknesses through which photons or charged particles can easily pass without loss of significant energies. This layer is useful for gas confinement also. GEM1 and GEM2 are thick copper foils consists of many holes, each diameter is 50 60 µ m. Data are collected using the two grids (X and Y) and 2D image reconstruction can be done by image reconstruction algorithms. The advanced features of GEM based detectors are high counting rate, excellent spatial resolution, and good imaging capability, operation in electric/magnetic fields, large sensitive area, flexible geometry and low cost [2]. The remarkable features stated above of multi-gem structure are very attractive for the numerous applications [3]. GEM-based detectors like other micro-pattern devices, offer localization to around 40 microns. However, their unique feature is that two co-ordinates can be recorded on the pickup electrode. Both X- and Y-strips are at ground potential, essential when using high-density readout. Such two-dimensional localization, useful in particle tracking, is necessary for medical imaging. X-rays are efficient tool for diagnostics in medicine and industry and are widely used for biological, chemical and material science researches. In astro-particle physics x-rays provide significant information on the type of x-ray sources and the objects that scatter and affect the polarization of x-rays such as magnetic fields. Several important domains of x-ray imaging are medical diagnostics, industrial x-ray imaging, synchrotron radiation experiments and astro-particle physics. A comparison among the medical scanners with GEM can be found elsewhere [4]. 2. Materials and Methods: Fig. 2 Schematic diagram of a thick GEM foil. The GEM works in the proportional regime as an avalanche preamplifier, which means that the amplitude of the output signal is proportional to the initial number of ion pairs produced by the incident photon. The gap between the drift plane and the GEM top is called the drift region or conversion gap, and the lower space, where amplified electrons are collected by the readout plate (anodes), is named the induction region or collection gap (see Fig. 1.). Upon application of a potential difference between the GEM copper plates, a high electric field is developed inside the holes focusing the field lines between the drift electrode and the readout strips/plates. The field density in the amplifying channel can be varied either by increasing the potential difference between the upper and the lower electrodes. Or by reducing the diameter of the GEM holes. The length of the amplifying
channel for a single GEM grid is fixed by the thickness of the insulating foil. So far most standard GEM foils have an insulator thickness of 50 µ m. Along the field directions perpendicular to the axis of the holes the field strength is almost uniform in the center and increases sharply near the edges, particularly close to the copper-kapton interface. This determines how a high voltage could be applied to the GEM, i.e., the point of electrical breakdown (appears as a spark signal). The charge avalanche traverses the GEM hole mostly in the center, but a fraction of it could approach this extremely high field region on the edges of the hole due to diffusion and trigger a discharge. Also electron emission from the GEM electrodes in this area is possible. Fig.3 Voltage supplier circuits: (a) High voltage and low current, input voltage to the Mesh was supplied separately, (b) Low Voltage and high current supply across the GEM and Mesh. Both Mesh and GEM are kept in floating conditions. Two types of voltage supplier circuit are demonstrated in fig. 3. Total impedance of (a) is about five times higher than (b). For various input voltage in the circuit DV G across the GEM foil and corresponding currents are obtained for two different circuits. In Fig. 4, DV G versus current are shown. 4. Results & Discussion Using the 55 Fe RI source (a few µ Ci) we have tried to get signals and counting rates. In Fig.5 signals with and without sources are depicted. The amplitude of the noise level is about 2 4 mv and with source amplitude is about 6 10 mv without amplifier. The voltage divider circuit (fig. 3(a) was used for this testing purpose. Using more than 5 mv discriminator threshold much of the noise and other background is reduced but sparks and cosmic
background remain the same. In Fig.6 a counting rate measurement circuit is depicted. Many sparks signals come at a time with highest amplitude are easily countable separately from the original signals. We have found the cosmic ray signals come out with higher amplitude than the x-ray signals. Amplitudes of x-ray signals are also varying and the average amplitude is about 7 mv without further amplification. The signal and the spark counting rates with corresponding DV G are depicted in Fig.7. It is very difficult to get pure x-ray signals out of sparking from this system, because nothing is optimized yet. The amplitude of the spark signals have no limits, therefore no discriminator threshold level can be set in order to get rid off it. Figure 4. Voltage across the single thick GEM and currents in the circuit (fig.3(a)) are shown. Signals of x-rays are obtained only in the sparking area. 3. Testing of a prototype GEM detector: It is quite known from many other experiments that avalanche electron gain increases with increasing DV G and more of it is required to obtain same gain of pure high Z inert gases. Optimized mixture of gases, e.g., Ar:CO 2 (70:30), Xe:Ar(95:5), Ar:CH 4 (90:10) etc. show better gain at lower DV G than pure gases at atmospheric pressure. We have used Ar:CO 2 (70:30) gas mixture to obtain better gain, special, time and energy resolutions [CBM report]. Gain or the detection efficiency depends on many parameters of the detector such as geometry, size, shape and materials of GEM foil, cathode mesh, pads, electronic accessories, gas, mixture of gases, drift, ionization and transfer distances, and the voltage across the Mesh and GEM. For the test of this prototype GEM we used two types of voltage divider circuits (see fig. 3) and obtained rare signals which are depicted in Fig. 5.
Fig. 5: Signals of 5.9 kev x-rays and noises are shown. The voltage divider circuit (see fig.3(a) was used to process these signals. 4. Results & Discussion Using the 55 Fe RI source (a few µ Ci) we have tried to get signals and counting rates. In Fig.5 signals with and without sources are depicted. The amplitude of the noise level is about 2 4 mv and with source amplitude is about 6 10 mv without amplifier. The voltage divider circuit (fig. 3(a) was used for this testing purpose. Using more than 5 mv discriminator threshold much of the noise and other background is reduced but sparks and cosmic background remain the same. In Fig.6 a counting rate measurement circuit is depicted. Many sparks signals come at a time with highest amplitude are easily countable separately from the original signals. We have found the cosmic ray signals come out with higher amplitude than the x-ray signals. Amplitudes of x-ray signals are also varying and the average amplitude is about 7 mv without further amplification. The signal and the spark counting rates with corresponding DV G are depicted in Fig.7. It is very difficult to get pure x-ray signals out of sparking from this system, because nothing is optimized yet. The amplitude of the spark signals have no limits, therefore no discriminator threshold level can be set in order to get rid off it. A lot of sparks almost half of the signals come out due to high DV G which is not recommended because of the safety reason of the detector. Sparks generally appear due to dust particles inside the holes or some other cracks or distortion among the copper layers and insulator. The rate of sparks increases due to avalanche diffusion of electrons. We also design another type of circuit [see
fig. 3(b)] for low DV G and high current supply. We increase current upto 700 µ A at DV G =1000 V. Fig. 6: Signal processing and counting circuit with Phosphor Digital Oscilloscope, Model No. DPO 7254, Tektronix. A lot of spark signals were seen without any x-ray signal. It seems to us that second type of circuit is not suitable for thick GEM. Our applied GEM foil is initially damaged a few areas and after cleaning we used it for testing purpose. Thickness of the top layer of the detector was 5 mm and drilled with 2 mm hole where 55 Fe source was kept and drift gap was 1 cm. Therefore x-ray exposure area was very limited and a very few ionization was occurred and less number of electrons were detected by the small size pad. Fig. 7. Counting rates are shown with different voltage drops across the single thick GEM foil.
Hence it is difficult to conclude that the geometry of the system, circuits or foil itself which are not providing efficient signals. In order to test this detector high intense x-rays and γ-rays with various energies are required to optimize DV G, current, geometry, gain etc of the system. Conclusion Advantages of GEM detector are its flexible shape, size, spatial resolution and above all cost. A prototype GEM detector is constructed and tested roughly. A lot of improvement in design, circuits and electronics are required in order to get pure signals of x-rays, good detection efficiency and measurement of spatial, energy and time resolutions. Preliminary test show that high DV G and low current supply circuit is better than low DV G and high current supply circuit for getting much counting rate of x-rays. Further improvements of the system are going on and the fruitful results are commensurate soon. Acknowledgements: Authors are pleased to acknowledge the sincere help of scientific assistants. One of the authors is pleased to acknowledge the fellowship support from the VECC and grateful to Dr. D. K. Srivastava for encouraging the project. Reference: 1. F. Sauli NIMA 386 (1997) 531-534 2. S. Bachmann et.al. NIMA 471 (2001) 115 3. A. Buzulutskov, NIMA 494(2002) 148-155 4. N. N. Mondal et.al., Annuual report 2009, VECC, Kolkata, India.