MICROMEGAS chambers for hadronic calorimetry at a future linear collider
Prototypes of MICROMEGAS chambers, using bulk technology and analog readout, with 11 readout segmentation have been built and tested.
Measurements in Ar/iCH (95/5) and Ar/CO (80/20) are reported. The dependency of the prototypes gas gain versus pressure, gas temperature and amplification gap thickness variations has been measured with an Fe source and a method for temperature and pressure correction of data is presented.
A stack of four chambers has been tested in 200 and 7 muon and pion beams respectively. Measurements of response uniformity, detection efficiency and hit multiplicity are reported.
A bulk MICROMEGAS prototype with embedded digital readout electronics has been assembled and tested. The chamber’s layout and first results are presented.
Preprint typeset in JINST style - HYPER VERSION
MICROMEGAS chambers for hadronic calorimetry at a future linear collider
C. Adloff, D. Attié, J. Blaha, S. Cap, M. Chefdeville, P. Colas, A. Dalmaz, C. Drancourt, A. Espargilière††thanks: Corresponding author. , R. Gaglione, R. Gallet, N. Geffroy, I. Giomataris, J. Jaquemier, Y. Karyotakis, F. Peltier, J. Prast, G. Vouters
Laboratoire d’Annecy-le-Vieux de Physique des Particules (LAPP), Université de Savoie, CNRS/IN2P3,
9 Chemin de Bellevue, F-74941 Annecy-le-Vieux, France
Institut de recherche sur les lois fondamentales de l’Univers (IRFU), CEA, Saclay, France
Keywords: Micro-Pattern Gaseous Detectors (MPGD) ; MICRO MEsh GAseous Stucture (MICROMEGAS) ; Calorimeters; Large detector systems for particle and astroparticle physics.
Future linear colliders at Terascale energies, like the International Linear Collider (ILC) or the Compact LInear Collider (CLIC), will be the probes for new physics. Depending on Large Hadron Collider (LHC) results they will be able to get unprecedented measurements on Higgs physics but also on Super Symmetry and Standard Model extensions .
To obtain the unequaled jet energy resolution required to separate and induced jets (30%), one of the most promising analysis technique is based on Particle Flow Algorithms (PFA) [2, 3]. Jet energy is measured by combining the measurements of the track momentum from charged particles and of the calorimetric energy from neutrals. With this technique, the jet energy resolution is dominated by the error in hit assignment to clusters (so-called PFA’s confusion term), it is therefore mandatory to discriminate between charged and neutral clusters. This calls for highly segmented calorimeters with the the capability to produce narrow showers.
The high segmentation together with the large area to be instrumented (e.g. 3000 for the SiD hadronic calorimeter ) leads to a dramatic increase of the readout channel number and thus of the amount of data to handle and store. This can be balanced by using a Digital Hadronic CALorimeter (DHCAL) counting the number of hits rather than measuring the deposited energy. In a sampling DHCAL, a few options concerning the active layer are considered: scintillators tiles, GEMs, RPCs and MICROMEGAS [4, 5, 6].
The MICRO-MEsh GAseous Structure (MICROMEGAS) was introduced in 1996  as a fast signal, position-sensitive, radiation hard gaseous detector. MICROMEGAS consists of a conductive mesh held a few tens of micrometers above a segmented anode plane, defining the amplification gap, surmounted by a cathode defining the drift gap. An incident charged particle crossing the drift gap ionizes the gas. Using suitable voltage settings, the ionization electrons drift to the mesh, enter the high amplification field region where they produce further ionizations in cascade. The motion of charges in this region induces a signal on the mesh and on the anode plane.
Large area MICROMEGAS based detectors have already been produced (up to 4040 ), using the so-called bulk MICROMEGAS technique. Such a bulk MICROMEGAS is made by lamination of photoresistive film layers on the readout Printed Circuit Board (PCB), strongly encapsulating a stretched micro-woven mesh at a fixed distance from the PCB, and forming spacer pillars after photolithography . With this manufacturing technique, a bulk MICROMEGAS with anode pad on one side of the PCB and embedded readout electronics on the back side offers a compact and robust detector that can be produced by industry. MICROMEGAS is therefore a very appealing possibility to equip a DHCAL.
The present work was carried out in the framework of the Calorimeters for ILC Experiments collaboration (CALICE)  with the aim to test the bulk MICROMEGAS technology for DHCAL. For extensive characterization, measurements were performed with bulk MICROMEGAS prototypes with external analog readout electronics boards equipped with GASSIPLEX chips . The feasibility of a compact detector has been studied by building a bulk MICROMEGAS with an embedded digital readout chip called DIRAC .
2 Experimental Setup
2.1 Analog Readout Prototypes
Each prototype consists of a bulk MICROMEGAS chamber with a 3 drift gap and a 128 amplification gap. The drift cathode is a 5 thick copper foil fixed on a 75 thick Kapton film. The whole is glued on a 2 thick steel plate, forming the device’s lid. The steel cover plate is part of the absorber and therefore would not contribute to the HCAL active layer’s thickness. The 3 drift gap is ensured by a 3 thick resin frame enclosing the chamber and providing the gas inlet and outlet (see photography on figure 1, right).
The bulk uses an industrial micro woven stainless steel mesh made of 30 diameter wires interwoven at a pitch of 80. The mesh is held by 128 high, 300 diameter pillars laid out on a square lattice with a 2 pitch. The anode plane consists of 0.980.98 pads spaced every 200 lying on the detector’s PCB. The PCB is a 4 layers class 4, 1.6 thick. The 1 pattern made of a pad and the free space around will be denoted hereafter as a 1 area where ’’ will be the unit area symbol.
Four prototypes with analog readout have been built. Three of them have a 616 active area (96) and the last one is four times larger with a 1232 active area (384). The mesh voltage of the small chambers is supplied through a dedicated pad while a 4 sidelong mesh voltage pad is used for the large chamber (see figure 1). In the following, the three small chambers will be denoted CH0, CH1 and CH2 and the large one, CH3.
2.2 Readout System
The anode analog readout was provided by 16-channel ASICs called GASSIPLEX . Boards111CEA DAPNIA Board N613V, 96 channels, 6 GASSIPLEX chips 0.7 v3., each equipped with 6 GASSIPLEX chips, were mounted on the side of the chambers. One board was used for each of the small chambers and four for the large one.
GASSIPLEX chips, when triggered, gather the signal from every channel and build one single multiplexed differential output with a nominal conversion factor of 3.6 and a peaking time of 1.2. The multiplexed signal from GASSIPLEX boards was digitized by CAEN V550 10-bit ADCs (VME modules) sequenced by a CAEN V551B C-RAMS sequencer VME module. Data were then collected by the computer through an optical VME/PCI bridge. A Labview based software, called CENTAURE , was used for online monitoring and data recording. A very similar readout system was used for the CAST experiment and is described in . The global conversion ratio of the GASSIPLEX-based readout chain was measured to be 4.690.25 (Analog to Digital Unit). Its r.m.s. over all 672 channels is 2.5%.
For the X-ray study, only one chamber was used with a different readout system, based on the mesh signal. The mesh signal was read out by an ORTEC 142C charge preamplifier linked to its corresponding amplifier/shaper. The calibration constant of the mesh readout chain was precisely measured to be 2.1990.026.
A detailed note about the calibration of both readout chains is available in .
3 X-ray Study
X-ray tests using a 5.9 Fe source have been performed to measure the basic performance of the prototypes. For each prototype, the steel lid is drilled on a few locations to allow X-ray injection through the cathode and Kapton foil. The electron collection efficiency and the gas gain in Ar/iCH (95/5) and Ar/CO (80/20) were deduced from the Fe K photopeak value given by a fit of of three gaussian functions to the Fe spectrum (figure 2, left). Gain measurements were used to predict gain dependency versus pressure, temperature and amplification gap thickness variations. Those predictions are confronted to direct measurements in section 3.4.
3.1 Electron Collection Efficiency
The ratio between amplification and drift electric fields affects the mesh transparency to electrons (or collection efficiency) by contracting the field lines so that the electrons are mostly driven through the center of the mesh’s holes and reach the amplification gap. Figure 2 (right) shows the variation of the Fe peak value versus the field ratio in Ar/iCH (95/5) and Ar/CO (80/20). The amplification field was kept constant while the drift field varied to set the field ratio. The mesh voltage was 420 in Ar/iCH (95/5) and 570 in Ar/CO (80/20).
The curves displayed on figure 2 (right) show a maximum at a field ratio of about 150 – 200 for both gas mixtures. The measurements reported in the following were performed at ratios within this range.
A possible explanation for the decline at high field ratio observed in figure 2 (right) is the attachment of some primary electrons in the drift region by electronegative impurities (e.g. oxygen, water vapor). For a constant amplification field, a higher ratio means a lower drift field and consequently the primary electrons tend to have less energy. Since the attachment cross section of some impurities peaks at low energy (e.g. 0.1 for oxygen ), a lower drift field can lead to a higher attachment probability.
3.2 Gas Gain
The amplification gap gain, so-called gas gain, is determined through a fit of three gaussian functions to the Fe spectrum (figure 2, left) assuming 230 and 209 primary electrons in the Ar/iCH (95/5) and Ar/CO (80/20) mixtures respectively and using the mesh readout calibration constant (2.19). Keeping the drift field at 150, a set of measurements at 954 in the Ar/iCH mixture, 961 in Ar/CO and at a temperature of 293 with various voltage settings gave the gain curves displayed in figure 3.
A much higher maximum gain (410) can be achieved in Ar/iCH than in Ar/CO. Also the mesh voltage, for a given gain, is 180 lower in Ar/iCH.
3.3 Method for Pressure and Temperature Correction
3.3.1 Gas Gain Model
The gas gain of the chambers is given by the exponential of the average number of primary ionizations from a single avalanche initiating electron. In a uniform field, this number is given by the first Townsend coefficient, denoted , multiplied by the amplification gap thickness, denoted :
The Townsend coefficient can be parameterized by the Rose and Korff formula :
with , and , where is the Boltzmann constant. Equation 3.3.1 unlights the dependency of the gas gain versus pressure (), temperature () and amplification gap thickness (). Those dependencies can be derived from and expressed as:
In practice, it is convenient to apply one single correction for pressure and temperature variations using the coefficient:
A correction is applied by multiplying the gain by the correction factor given by:
where stands for , , or .
3.3.2 Application to Ar/iCH and Ar/CO
The gain dependencies on P, T, P/T and g can be predicted from a gain curve by adjusting the constants and on the measured trend via formula 3.3.1. The fits gave and in Ar/iCH and and in Ar/CO. The dependencies calculated using the formulae 3.3.1 – 3.3.1 are gathered in table 1.
The predicted values of , and in Ar/CO are compared to direct measurements in the next section.
3.4 Environmental Study in Ar/CO (80/20)
3.4.1 Experimental Conditions
During two weeks the amplitude of some 10 pulses from Fe quanta conversions in Ar/CO 80/20 were recorded, enabling a precise monitoring of the detector gain as a function of time. In parallel, gas pressure and temperature were also recorded. The mesh voltage was set to 570 at which a gain of about 510 was measured (see figure 3). The drift field was kept at 100. The Ar and CO gas flows were equal to 0.97 and 0.24, yielding a total flow of 1.21. The mean pressure was 959.7 and the mean temperature was 298.2. The temperature was controlled with the help of an air conditioner and the gas pressure fluctuated according to the atmospheric pressure variations.
3.4.2 Pressure Corrections
During part of the run, the gas temperature was maintained around 298 to examine the gas gain dependency on pressure only. Figure 4 shows the Fe peak value versus pressure recorded at a temperature . A linear behavior is observed and fitted with a slope which relates to the coefficient according to:
where is the average Fe peak value over the fitted range. With and , computation gives:
which is consistent with the predicted value (section 3.4.1).
3.4.3 Temperature Corrections
Data recorded during a period with temperature variations of a few kelvins have been corrected for pressure variations using from section 3.4.2 and formula 3.3.1. The corrected Fe peak value, , is given by:
where is the raw peak value, and is plotted in figure 5. A linear fit was performed and its slope gave the coefficient through . With and , computation leads to:
which is consistent with the value predicted in section 3.4.1.
3.4.4 Corrections using the Ratio of Pressure over Temperature
The evolution of the Fe peak value along the whole data set versus the ratio of pressure over temperature is plotted in figure 6. A straight line was adjusted on the points and its slope gave the coefficient through . With and , its value is:
which is within the error range of the value predicted in section 3.4.1.
3.4.5 Conclusion of the Study
Direct measurement of the coefficients , and showed good agreement with the gas gain model prediction (table 2). Using those coefficients, the Fe peak value has been corrected for pressure and temperature variations according to:
The result of those corrections are gathered in figure 7. The Fe peak value is very scattered before applying any correction. The successive corrections for pressure and temperature leads to a major improvement of the Fe peak value regularity. The direct correction using the ratio of pressure over temperature is also valuable. The correction yielding the strongest reduction of the distribution r.m.s. is the one based on C because the temperature was controlled during the data acquisition and showed limited variations.
Those results validate the gas gain model and the method for environmental corrections of the data.
3.5 X-Ray Test Conclusion
The mesh collection efficiency should be maximum for a field ratio around 200 in Ar/iCH. Hence for later measurements, as the mesh voltage will be around -420, the cathode voltage will be kept 50 below.
In a given gas mixture, the gas gain depends primarily on the gas number density and the amplification gap thickness. The density relates to the ratio P/T which, in Ar/CH (95/5), impacts on the gain according to -236% . Typical values of this ratio are around with variations of the order of 10 leading to some 2 – 3% gas gain fluctuation. If the correction factor is mainly below 10%, the error margins make the corrections uncertain so they shall be applied only if a significant amount of data need a correction factor above 10%.
The amplification gap size determines the distance over which an electron avalanche develops and the amplification field for a given mesh voltage. It strongly impacts on the gain. The bulk planarity is better than 5, but a gap variation of 1 should result in a change of the gas gain of 6.6% in Ar/CH (95/5). Therefore the mesh irregularities are expected to play a major role in the detector’s gain disparity. Smaller variations are predicted in Ar/CO (80/20) due to a milder dependence of the first Townsend coefficient on the electric field.
4 Measurements with Particle Beams
4.1 Experimental Layout
The detector stack was set up placing the small chambers in the front followed by the large one at the rear (figure 8). The distance between each chamber was 10. Three scintillator paddles were placed in front of the stack, the trigger signal was provided by the triple time coincidence of their output. Two of them were 832 and the last one had the exact dimensions of the small chambers (616).
A common pre-mixed Ar/iCH (95/5) gas was used and the voltage applied on the prototypes’ meshes were -420, -420, -430 and -410, for CH0 to CH3, respectively. These voltage values where set as a trade off between a high gain and a spark rate below about one per hour. The drift voltages were set 50 below the meshes’ ones so that the fields ratios were always corresponding to the maximum collection efficiency (see section 3.1).
4.2 Particle Sources
4.2.1 CERN/SPS, H2 beam line
The Super Proton Synchrotron (SPS) delivers a 400 proton beam for fixed target experiments. Data were taken with a secondary 200 negative muon beam. The beam was available during a 9 spill period every machine cycle. Each machine cycle lasted for 33 during night time and 48 during day time. The beam was intense enough to saturate the acquisition rate at about 130 events. Data were recorded during August 2008 beam test session.
4.2.2 CERN/PS, T9 beam line
The Proton Synchrotron (PS) delivers a 28 proton beam for injection in SPS and CERN’s East Area’s Fixed target experiments. Data were recorded with a secondary 7 positive pion beam. The beam was available during one to three 0.4 spill periods every machine cycle. A machine cycle lasted for a variable time around 40. The beam was also intense enough to saturate the acquisition rate. Data were recorded during November 2008 beam test session.
4.3.1 Environmental and Noise Conditions
During the data acquisition, the atmospheric pressure and the gas temperature were monitored. Using , a gain correction factor was computed using formula 3.3.1 and found always below 10% with an r.m.s. below 4% (see figure 9). In accordance with section 3.5 environmental corrections are sufficiently small and considered to be negligible. Moreover, in a digital detector those corrections could not be applied at all. Therefore, in the aim of a DHCAL the results given here will remain uncorrected.
The GASSIPLEX pedestals were periodically aligned at 20 on the V550 ADC modules. They were measured to be at this value with 2% r.m.s. variations over all channels through the whole data set (see figure 10, left). The pedestal sigmas were obtained from a gaussian fit and showed an average value of 1.5 corresponding to 0.3 or 2000 (figure 10, right). These figures demonstrate very good noise conditions.
4.3.2 Event tags
All channels were recorded without threshold. A hit is defined by applying an off-line threshold equal to 1.5 (7 above pedestal). Two types of events were selected for the analysis: the Platinum and the Golden events.
Platinum events: an event is tagged as platinum by requiring one single hit in each of the four chambers. Those events are used for gain and pedestal studies since they ensure a very low noise hit contamination.
Golden events: a golden event is selected by requiring one single hit in three out of the four chambers. Those events are used for efficiency and multiplicity studies.
4.4 Gain Distribution Measurement
For every channel, a Landau function was fitted on the data from platinum events (see figure 11 (left)) and its Most Probable Value (MPV) was defined as the detector’s global gain for charged particles (conversion, mesh amplification and electronics amplification). The resulting values are mapped in figure 12.
The most probable deposited charge, averaged over all channels, is 22.6, 22.9, 24.5 and 17.5 for CH0 to CH3 respectively. The relative gain distribution of all the channels is shown in figure 11 (right) having an r.m.s. of 11.25%. Since the electronics gain distribution has a very low r.m.s., this value is expected to be mainly due to drift and/or amplification gaps non-uniformity.
4.5 Efficiency Measurement
In order to measure the efficiency of a given chamber, a sub range of golden events was selected by requesting three aligned hits in the three other chambers to define a particle track. A safety threshold of 12.5 was applied for the three reference chambers’ hits to completely avoid taking noise hits into account. In each processed event, a hit was searched in a 33 area centered at the intersection between the extrapolated reconstructed particle track and the chamber plane. The resulting efficiencies are mapped in figure 13 and their distribution is shown in figure 14.
The hit background was estimated by counting the number of hits outside the 33 area. This number, normalized to the 33 area, was subtracted to measure the final efficiency (Table 3). Thanks to the very low threshold, three chambers show an excellent efficiency, larger than 97%. CH2 shows a lower efficiency (91%). It might be due to the lower tension of its mesh or to the broader pedestals of its electronics.
|Chamber||Raw efficiency||Noise hit fraction||Noise corrected efficiency|
In a DHCAL, such a low threshold (1.5) may not be achievable. Thus the same study was carried out for each chamber varying the threshold from 1.5 to 200. The dependency between efficiency and threshold is unlighted in figure 15.
A steep drop of efficiency with threshold is observed. The efficiency is about 70% at a threshold of 20 and drops below 10% for thresholds higher than 100. This has strong concequences on the electronics noise requirements.
4.6 Multiplicity Measurement
Still using golden events, the number of hits in a 33 area around the pad expected to be hit was counted. The multiplicity is computed as the mean of this number over all the processed events (Table 4) is corrected with the same method as for the efficiency (section 4.5).
|Chamber||Raw multiplicity||Noise corrected multiplicity|
The four chambers showed a noise corrected multiplicity between 1.06 and 1.10, which is a benefit for a PFA algorithm. The behavior of multiplicity versus threshold was also studied and is illustrated in figure 16.
After a quick fall, the multiplicity rises slowly and then decreases at high threshold. At very low threshold, almost all pads receiving charge are seen and the maximum multiplicity is measured. With increasing threshold the multiplicity decreases dramatically down to 1.03 – 1.04 at 30 – 40. Above this value, mainly events of particles with large energy deposit are considered. These particles likely produce -rays leading to some ionization far from the track and hence to a higher multiplicity. Above 150, as a consequence of the decreasing detection efficiency with threshold, the multiplicity declines again as expected.
As a conclusion, the efficiency and hit multiplicity of the MICROMEGAS prototypes are potentially excellent (97% efficiency and 1.06 multiplicity at a 1.5 threshold) according to the requirements of a DHACL active layer. More technical details about the beam test analysis are available in .
5 Embedded Digital Readout Prototype
A compact detector is compulsory for the DHCAL active layer. Therefore, a process has been developed to reach a bulk MICROMEGAS with embedded readout chip. The DIRAC chip has been chosen for the first prototype. This was also an opportunity to test this R&D chip in real conditions.
5.1 Dirac Asic
DIRAC  is a 64-channel digital ASIC intended for the readout of gaseous detectors like MICROMEGAS, GEM, GRPC. Its design is highly specific to ILC DHCAL requirements. It is based on low cost technologies, offers a low power consumption thanks to power pulsing synchronized to the ILC clock. It provides two operative modes (high gain in MICROMEGAS/GEM mode and low gain in RPC mode), a scale of three thresholds with an 8-bit precision and a fully digital serial interface.
5.2 Prototype Layout
The prototype consists of an 88, 6 layers, class 6 PCB, equipped with a single DIRAC ASIC. A mask was fixed on the PCB’s ASIC side in order to avoid the embedded electronics from being squashed during the lamination of the mesh. The same bulk layout as the one described in section 2.1 was used. The anode segmentation was made of 11 pads spaced every 500.
5.3 Beam Test Result
The first operative test with bulk MICROMEGAS with embedded electronics has been carried out and the electronics proved to have survived the lamination process by showing the beam profile displayed in figure 17. This profile was obtained in a 200 muon beam, with a 19 threshold, in Ar/iCH (95/5), at a mesh voltage of 410 and a drift voltage of 460. The data acquisition was auto-triggered by the ASIC. A raw hit multiplicity of 1.1 has been determined. Further measurements will be performed with a stack of several embedded DIRAC MICROMEGAS detectors. This test is a proof of principle for the bulk MICROMEGAS with embedded electronics and for the DIRAC ASIC functionality.
Bulk MICROMEGAS prototypes with analog readout have been tested with X-rays and CERN’s particle beams.
The gas gain dependency on pressure, temperature and amplification gap thickness variations has been calculated from gain curves in Ar/iCH (95/5) and Ar/CO (80/20). As expected from the steeper gain curve, the values obtained in the Ar/iCH mixture are significantly higher. This difference makes Ar/CO more stable against pressure and temperature variations than Ar/iCH making this gas mixture interesting despite the lower gain it provides. In Ar/CO (80/20), these calculated values were confronted with measurements showing good agreement. A method for gain correction based on those dependencies has been presented.
Four chamber were placed in 200 muon and 7 pion beams. The gain distribution of 384 channels (a 384 area) showed an 11% r.m.s. variation. The efficiency and the hit multiplicity were measured and their dependency versus threshold was studied. An efficiency near 97% at a 1.5 threshold was obtained and a hit multiplicity as low as 1.03 has been found at 20.
The first bulk MICROMEGAS with embedded readout electronics have been built, tested and proved to be functional, which validates the fabrication process of a compact MICROMEGAS and also the DIRAC ASIC performances.
We would like to thank Bruno Chauchaix, Lau Gatignon and Dragoslav-Laza Lazic for their very precious help while on the test areas. We are also grateful to Didier Roy for his hard work on the CENTAURE software for data acquisition and online monitoring. We wish to thank Rui de Olivera and the CERN TS-DEM group for the bulk production and their collaboration in the development of the prototype with embedded electronics. We also thank Philippe Abbon for the production of the 613V GASSIPLEX boards. A special thank also for Kostas Karakostas, Glenn Cougoulat, Richard Hermel and Jean Tassan for their contribution at the begining of the project and Denis Fougeron for designing the analog prototype PCBs.
-  ILC Global Design Effort and World Wide Study, ILC Reference Design Report, Physics at the ILC (2007).
-  J.-C. Brient, Improving the jet reconstruction with the particle flow method: An introduction, in the Proceedings of 11th International Conference on Calorimetry in High-Energy Physics (Calor 2004), Perugia, Italy, March 2004.
-  Stephen R Magill, Innovations in ILC detector design using a particle flow algorithm approach, New Journal of Physics 9 (2007) 409, doi:10.1088/1367-2630/9/11/409.
-  Sergej Schuwalow, Calorimetry at the ILC detectors, Nuclear Instruments and Methods in Physics Research A 598 (2009) 258 – 263.
-  SiD, Letter of Intent, March 2009.
-  ILD Concept Group, The International Large Detector Letter of Intent, March 2009.
-  I. Giomataris, Ph. Rebourgeard, J.P. Robert, G. Charpak, MICROMEGAS: a high-granularity position-sensitive gaseous detector for high particle-flux environments, Nuclear Instruments and Methods in Physics Research A 376 (1996) 29 – 35.
-  Yuri Kudenko, The near neutrino detector for the T2K experiment, Nuclear Instruments and Methods in Physics Research A 598 (2009) 289 – 295.
-  I. Giomataris, R. De Oliveira, S. Andriamonje, S. Aune, G. Charpak, P. Colas, G. Fanourakis, E. Ferrer, A. Giganon, Ph. Rebourgeard, P. Salin, MICROMEGAS in a bulk, Nuclear Instruments and Methods in Physics Research A 560 (2006) 405 – 408.
-  https://twiki.cern.ch/twiki/bin/view/CALICE/WebHome
-  J.C. Santiard, K. Marent, The Gassiplex0.7-2 Integrated Front-End Analog Processor for the HMPID and the Dimuon Spectrometer of ALICE, 5 Conference on Electronics for LHC Experiments, Snowmass, CO, USA, 20 – 24 Sep 1999, pp.431 – 435.
-  R. Gaglione, H. Mathez, DIRAC: A DIgital Readout Asic for hAdronic Calorimeter, Nuclear Science Symposium Conference Record, 19 – 25 Oct. 2008 Page(s):1815 – 1819.
-  D. Roy, Manuel utilisateur, CENTAURE version 21lv 8.5 http://www-subatech.in2p3.fr/ electro/infoaq/CENTAURE21lv85/pdfdoctxt/centaure21lv85.pdf
-  T. Geralis,G. Fanourakis, Y. Giomataris, K. Zachariadou, The data acquisition of the MICROMEGAS detector for the CAST experiment, Nuclear Science Symposium Conference Record, IEEE, 19 – 25 Oct. 2003, 3455 – 3459 Vol.5.
-  C. Adloff, M. Chefdeville, A. Espargilière, R. Gaglione, Calibration of electronic readout chains for MICROMEGAS chambers, LAPP Technical note, HAL : in2p3-00400295, 2009.
-  C. Adloff, M. Chefdeville, A. Espargilière, R. Gaglione, Environmental study of a MICROMEGAS detector, LAPP-TECH-2009-03.
-  Magboltz cross sections, http://rjd.web.cern.ch/rjd/cgi-bin/cross
-  F. Sauli, CERN yellow report 77-09.
-  C. Adloff, A. Espargilière, Y. Karyotakis, MICROMEGAS Beam Test, 2008 - Analysis & Results, LAPP-TECH-2009-04.