Micromegas TPC P Colas CEAIrfu Saclay MPGD Lectures

Micromegas TPC P. Colas, CEA/Irfu Saclay MPGD Lectures, SINP, Kolkata October 20 -22, 2014

OUTLINE PART I – operation and properties TPC, drift and amplification Micromegas principle of operation Micromegas properties Gain stability and uniformity, optimal gap Energy resolution Electron collection efficiency and transparency Ion feedback suppression Micromegas manufacturing meshes and pillars In. Grid “bulk” technology Resistive anode Micromegas Digital TPC Kolkata, October 20, 2014 P. Colas - Micromegas 2

OUTLINE PART II – Applications The CAST experiment The COMPASS experiment The KABES beam spectrometer The T 2 K ND-280 TPC The Large Prototype for the ILC Micromegas neutron detectors TPCs for Dark Matter search and neutrino studies Practical operation and use of Micromegas Kolkata, October 20, 2014 P. Colas - Micromegas 3

Electrons in gases : drift, ionization and avalanche Typical (thermic) energy of an electron in a gas: 0. 04 e. V E Mean free path l=ns (0. 4 mm at 1 e. V) Low enough electric field (<1 k. V/cm) : collisions with gas atoms limit the electron velocity to vdrift = f(E) (effective friction force) At higher fields ionization takes place (gain 10 V in 2 mm =50 k. V/cm) magboltz Kolkata, October 20, 2014 P. Colas - Micromegas 4

Cross-sections of most common quenchers follow the same kind of shape, but not all (noticeably, not He); Dip due to Ramsauer effect (interf. when e-wavelength~mol. size) Note : attachment Kolkata, October 20, 2014 P. Colas - Micromegas 5

Electrons in gases : drift, ionization and avalanche Thanks to the Ramsauer effect, there is a maximum drift velocity at low drift field : important for a TPC, to have a homogeneous time-to-z relation Typical drift velocities : 5 cm/ms (or 50 mm/ns) Higher with CF 4 mixtures Lower with CO 2 mixtures Kolkata, October 20, 2014 P. Colas - Micromegas 6

Attachment Ne = Ne 0 exp(-az) electron capture by the molecules a can be from mm-1 to (many m) -1 Attachment coefficient = 1 / attenuation length 2 -body : e- + A -> A- ; 3 -body : e- + A -> A*-, A *- B -> AB-, a a [A][B] Exemple of 2 -body attachment : O 2, CF 4 Exemple of 3 -body attachment : O 2, O 2+CO 2 Very small (10 ppm) contamination of O 2, H 2 O, or some solvants, can ruin the operation of a TPC Kolkata, October 20, 2014 P. Colas - Micromegas 7

Diffusion limits z resolution (typically 200 -500 m/√cm) Limits rf resolution at high z (“diffusion limit”) B field greatly reduces the diffusion w=e. B/me, t = time between collisions (assumed isotropic) wt = from ~1 to 15 -20 (note wt ~Vdrift B/E) Drift Kolkata, October 20, 2014 Langevin equation v(E, B) -> Ex. B effect P. Colas - Micromegas 8

Electrons in gases : drift, ionization and avalanche E At high enough fields (5 – 10 k. V/cm) electrons acquire enough energy to bounce other electrons out of the atoms, and these electrons also can bounce others, and so on… This is an avalanche In a TPC, electrons are extracted from the gas by the high energy particles (100 Me. V to Ge. Vs), these electrons drift in an electric field, and arrive in a region of high field where they produce an avalanche. Wires, Micromegas and GEMs provide these high field regions. Kolkata, October 20, 2014 P. Colas - Micromegas 9

t TPC: Time Projection Chamber electrons diffuse and Ionizing Particle drift due to the E-field electrons are separated from ions B E A magnetic field reduces electron diffusion Micromegas TPC : the amplification is made by a Micromegas Kolkata, October 20, 2014 Localization in time and x-y P. Colas - Micromegas y x 10

Micromegas: How does it work? Y. Giomataris, Ph. Rebourgeard, JP Robert and G. Charpak, S 1 NIM A 376 (1996) 29 Micromesh Gaseous Chamber: a micromesh supported by 50 -100 mm insulating pillars, and held at Vanode – 400 V Multiplication (up to 105 or more) takes place between the anode and the mesh and the charge is collected on the anode (one stage) stage Funnel field lines: electron transparency very close to 1 for thin meshes Small gap: fast collection of ions Kolkata, October 20, 2014 S 2/S 1 = Edrift/Eamplif ~ 200/60000= 1/300 P. Colas - Micromegas 11

Kolkata, October 20, 2014 P. Colas - Micromegas 12

Small size => Fast signals => Short recovery time => High rate capabilities A GARFIELD simulation of a Micromegas avalanche (Lanzhou university) micromesh signal strip signals Electron and ion signals seen by a fast (current) amplifier In a TPC, the signals are usually integrated and shaped Kolkata, October 20, 2014 P. Colas - Micromegas 13

Gain of Ar mixtures measured with Micromegas (D. Attié, PC, M. Was) Kolkata, October 20, 2014 P. Colas - Micromegas 14

Gain Compared with the “simple” picture, there are complications : -due to photon emission (which can re-ionize if the gas is transparent in the UV domain and make photo-electric effect on the mesh). This increases the gain, but causes instabilities. This is avoided by adding a (quencher) gas, usually a polyatomic gas with many degrees of freedom (vibration, rotation) to absorb UVs -due to molecular effects : molecules of one type can be excited in collisions and the excitation energy can be transferred to a molecule of another type, with sufficiently low ionization potential, which releases it in ionization (Penning effect) : e. A -> e. A* A*B ->AB+e Kolkata, October 20, 2014 P. Colas - Micromegas 15

Gain uniformity in Micromegas The nicest property of Micromegas • Gain (=e ad) • Townsend a increases with field • Field decreases with gap at given V • => there is a maximum gain for a given gap (about 50 m for Ar mixt. and 100 m for He mixt. ) Kolkata, October 20, 2014 P. Colas - Micromegas 16

Gain stability Very good gain stability (G. Puill et al. ) Optimization in progress for CAST <2% rms over 6 months Kolkata, October 20, 2014 P. Colas - Micromegas 17

Max Chefdeville et al (NIKHEF/Saclay) + Twente Univ. • This leads to excellent energy resolution 11. 7 % @ 5. 9 ke. V in P 10 That is 5% in r. m. s. obtained by grids postprocessed on silicon substrate. Similar results obtained with Microbulk Micromegas Kα escape line Kβ escape line 13. 6 % FWHM Gap : 50 μm; Trou, pas : 32 μm, Ø : 14 μm – with F = 0. 14 & Ne = 229 one can estimate the gain fluctuation parameter q Kolkata, October 20, 2014 Kβ removed by using a Cr foil P. Colas - Micromegas 11. 7 % FWHM 18

2007 MM 1_001 prototype Gain uniformity measurements Y- vs-X 55 Fe source illumination 404 / 1726 tested pads Gain ~ 1000 7% rms @ 5. 9 ke. V AFTER based FEE Average resolution = 19% FWHM @ 5. 9 ke. V Kolkata, October 20, 2014 P. Colas - Micromegas 19

Gain uniformity MM 1_001 prototype Inactive pads (Vmesh connection) 55 Fe source near module edge 55 Fe source near module centre Gain uniformity within a few % Kolkata, October 20, 2014 P. Colas - Micromegas 20

MM 0_007: gain uniformity 487 / 1726 tested pads Vmesh = 350 V 7. 4 % rms @ 5. 9 ke. V Average resolution = 21% FWHM @ 5. 9 ke. V Kolkata, October 20, 2014 P. Colas - Micromegas 21

MM 1_002 : gain uniformity and energy resolution Measured non-uniformities (%) Bopp micromesh AFTER ORTEC amplifier : 12 pads / measurement 21% FWHM @ 5. 9 ke. V 5. 6 1. 4 4. 1 4. 7 1. 0 1. 4 3. 0 3. 9 1. 6 0. 0 4. 4 0. 6 2. 8 5. 2 4. 4 2. 8 0. 8 3. 8 5. 8 1. 0 2. 2 1. 9 RMS = 3. 3% Kolkata, October 20, P. Colas - Micromegas 22

Transparency Collection efficiency reaches a plateau (100%? ) at high enough field ratio Micromesh Gantois Bopp pitch ( m) 57 63 ( m) 19 18 Operation point of Micro. Megas detectors in T 2 K is in the region where high micromesh transparencies are obtained Kolkata, October 20, P. Colas - Micromegas 23

Natural suppression of ion backflow S 1 THE SECOND NICEST PROPERTY OF MICROMEGAS Electrons are swallowed in the funnel, then make their avalanche, which is spread by diffusion. The positive ions, created near the anode, will flow back with negligible diffusion (due to their high mass). If the pitch is comparable to the avalanche size, only the fraction S 2/S 1 = EDRIFT/EAMPLIFICATION will make it to the drift space. Others will be neutralized on the mesh : optimally, the backflow fraction is as low as the field ratio. S 2 This has been experimentally thoroughly verified. Kolkata, October 20, 2014 P. Colas - Micromegas 24

Feedback : theory and simulation Hypothesis on the avalanche Periodical structure Gaussian diffusion Avalanche Resolution 2 s l Kolkata, October 20, 2014 P. Colas - Micromegas 25

ion backflow calculation Sum of gaussian diffusions 2 D Kolkata, October 20, 2014 3 D P. Colas - Micromegas 26

Theoretical ion feedback Results 500 lpi (sigma/l=0. 25) Kolkata, October 20, 2014 1000 lpi (sigma/l=0. 5) P. Colas - Micromegas 1500 lpi (sigma/l=0. 75) 27

Ion backflow (theory) Kolkata, October 20, 2014 P. Colas - Micromegas 28

Ion backflow measurements X-ray gun Vdrift I 1 (drift) Primaries+backflow Vmesh I 2 (mesh) I 1+I 2 ~ G x primaries One gets the primary ionisation from the drift current at low Vmesh One eliminates G and the backflow from the 2 equations The absence of effect of the magnetic field on the ion backflow suppression has been tested up to 2 T P. Colas, I. Giomataris and V. Lepeltier, NIM A 535 (2004) 226 Kolkata, October 20, 2014 P. Colas - Micromegas 29

Ion backflow measurements A new technique to make perfect meshes with various pitches and gaps has been set up (In. Grid at Twente) and allowed theory to be thoroughly tested (M. Chefdeville et al. , Saclay and Nikhef) rms avalanche sizes are 9. 5, 11. 6 and 13. 4 micron resp. for 45, 58 and 70 micron gaps. The predicted asymptotic minimum reached about s/pitch ~0. 5 is observed. Red: data Blue: calculation In conclusion, the backflow can be kept at O(1 permil) : does not add to primary ionisation (on average) Kolkata, October 20, 2014 P. Colas - Micromegas 30

Gain and spark rates E. Mazzucato et al. , T 2 K 95 m 128 m Threshold = 100 n. A The T 2 K/TPC will be operated at moderate gas gains of about 1000 where spark rates / module are sufficiently low (< 0. 1/hour). TPC dead time < 1% achievable. Kolkata, October 20, 2014 P. Colas - Micromegas 31

Discharge probability in a hadron beam 2. 5 mm conversion gap 100 µ amplif. gap Number of discharges per hadron Ne-C 2 H 6 -CF 4 gain ~ 104 P = 10 -6 <Z> ~20 <Z> ~14 <Z> ~10 Future, pion beam: -remove CF 4 -lower the gain Note that discharges are not destructive, and can be mitigated by resistive coating Kolkata, October 20, 2014 -increase the gap to compensate D. Thers et al. NIM A 469 (2001 )133 P. Colas - Micromegas 32

MESHES Many different technologies have been developped for making meshes (Back-buymers, CERN, 3 M-Purdue, Gantois, Twente…) Exist in many metals: nickel, copper, stainless steel, Al, … also gold, titanium, nanocristalline copper are possible. Chemically etched Laser etching, Plasma etching… Electroformed Wowen Deposited by vaporization 200 mm PILLARS Can be on the mesh (chemical etching) or on the anode (PCB technique with a photoimageable coverlay). Diameter 40 to 400 microns. Also fishing lines were used (Saclay, Lanzhou) Kolkata, October 20, 2014 P. Colas - Micromegas 33

The Bulk technology Fruit of a CERN-Saclay collaboration (2004) Mesh fixed by the pillars themselves : No frame needed : fully efficient surface Very robust : closed for > 20 µ dust Possibility to fragment the mesh (e. g. in bands) … and to repair it Used by the T 2 K TPC under construction Kolkata, October 20, 2014 P. Colas - Micromegas 34

The Bulk technology Kolkata, October 20, 2014 P. Colas - Micromegas 35

The T 2 K TPC has been tested successfully at CERN (9/2007) 36 x 34 cm 2 1728 pads Pad pitch 6. 9 x 9 mm 2 Kolkata, October 20, 2014 P. Colas - Micromegas 36

T 2 K TPC (beam test events) Kolkata, October 20, 2014 P. Colas - Micromegas 37

Resistive anode Micromegas • With 2 mm x 6 mm pads, an ILC-TPC has 1. 2 106 channels, with consequences on cost, cooling, material budget… • 2 mm still too wide to give the target resolution (100 -130 µm) Not enough charge sharing, even for 1 mm wide pads in the case of Micromégas (s avalanche ~12µm) Kolkata, October 20, 2014 P. Colas - Micromegas 38

Solution (M. S. Dixit et. al. , NIM A 518 (2004) 721. ) Share the charge between several neighbouring pads after amplification, using a resistive coating on an insulator. The charge is spread in this continuous network of R, C M. S. Dixit and A. Rankin NIM A 566 (2006) 281 SIMULATION MEASUREMENT Kolkata, October 20, 2014 P. Colas - Micromegas 39

25 µm mylar with Cermet (1 MW/□) glued onto the pads with 50 µm thick dry adhesive Cermet selection and gluing technique are essential Drift Gap Al-Si Cermet on mylar MESH Amplification Gap 50 m pillars Kolkata, October 20, 2014 P. Colas - Micromegas 40

A point charge being deposited at t=0, r=0, the charge density at (r, t) is a solution of the 2 D telegraph equation. Only one parameter, RC (time per unit surface), links spread in space with time. R~1 MW/□ and C~1 p. F per pad area matches µs signal duration. (r) Q (r, t) integral over pads mm Kolkata, October 20, 2014 P. Colas - Micromegas ns 41

Another good property of the resistive foil: it prevents charge build-up, thus prevents sparks. Gains 2 orders of magnitude higher than with standard anodes can be reached. Mesh voltage (V) Kolkata, October 20, 2014 P. Colas - Micromegas 42

Reminder of past results • Demonstration with GEM + C-loaded kapton in a X-ray collimated source (M. S. Dixit et. al. , Nucl. Instrum. Methods A 518 (2004) 721) • Demonstration with Micromegas + C-loaded kapton in a X-ray collimated source (unpublished) • Cosmic-ray test with GEM + C-loaded kapton (K. Boudjemline et. al. , to appear in NIM) • Cosmic-ray test with Micromegas + Al. Si cermet (A. Bellerive et al. , in Proc. of LCWS 2005, Stanford) • Beam test and cosmic-ray test in B=1 T at KEK, October 2005 Kolkata, October 20, 2014 P. Colas - Micromegas 43

The Carleton chamber Carleton-Saclay Micromegas endplate with resistive anode. 128 pads (126 2 mmx 6 mm in 7 rows plus 2 large trigger pads) Drift length: 15. 7 cm ALEPH preamps + 200 MHz digitizers Kolkata, October 20, 2014 P. Colas - Micromegas 44

Kolkata, October 20, 2014 P. Colas - Micromegas 45

4 Ge. V/c + beam, B=1 T (KEK) Effect of diffusion: should become negligible at high magnetic field for a high t gas Kolkata, October 20, 2014 P. Colas - Micromegas 46

The 5 T cosmic-ray test at DESY 4 weeks of data taking (thanks to DESY and T. Behnke et al. ) Used 2 gas mixtures: Ar+5% isobutane (easy gas, for reference) Ar+3% CF 4+2% isobutane (so-called T 2 K gas, good trade-off for safety, velocity, large wt ) Most data taken at 5 T (highest field) and 0. 5 T (low enough field to check the effect of diffusion) Note: same foil used since more than a year. Still works perfectly. Was ~2 weeks at T=55°C in the magnet: no damage Kolkata, October 20, 2014 P. Colas - Micromegas 47

The gain is independent of the magnetic field until 5 T within 0. 5% Kolkata, October 20, 2014 P. Colas - Micromegas 48

Pad Response Function Kolkata, October 20, P. Colas - Micromegas 49

Residuals in z slices Kolkata, October 20, P. Colas - Micromegas 50

• Resolution = 50 µ independent of the drift distance Analysis: Ar+5% isobutane Curved track fit B=5 T P>2 Ge. V f < 0. 05 Kolkata, October 20, 2014 P. Colas - Micromegas 51

Resolution = 50 µ independent of the drift distance ‘T 2 K gas’ Kolkata, October 20, 2014 P. Colas - Micromegas 52

Average residual vs x position Before bias correction After bias correction ± 20 m Kolkata, October 20, 2014 P. Colas - Micromegas 53

• B=0. 5 T • Resolution at 0 distance ~50 µ even at low gain Gain = 2300 Gain = 4700 Neff=28. 8± 2. 2 Neff=25. 2± 2. 1 At 4 T with this gas, the point resol° is better than 80 µm at z=2 m Kolkata, October 20, 2014 P. Colas - Micromegas 54

Further developments • Make bulk with resistive foil for application to T 2 K, LC Large prototype, NSW, etc… • For this, several techniques are available: resistive coatings glued on PCB, serigraphied resistive pastes, photovoltaïc techniques Kolkata, October 20, 2014 P. Colas - Micromegas 55

Principle of the digital TPC Micromegas Cathode Ionizing particle Gas volume amplification system (MPGD) + Every single ionization electron is detected with an accuracy matching~50 theµm 80 k. V/cm avalanche size -> maximal information, ultimate resolution - + - Time. Pix chip Kolkata, October 20, 2014 P. Colas - Micromegas 56

Time. Pix/Micromegas CERN/Nikhef-Saclay Fenêtre pour sources X Capot 6 cm Fenêtre pour source b Cage de champ Mesh Micromegas Puce Medipix 2/Time. Pix Kolkata, October 20, 2014 P. Colas - Micromegas 57

Timepix chip 65000 pixels (500 transistors each) + Si. Prot 20 μm + Micromegas 55 Fe Ar/Iso (95: 5) Mode Time z = 25 mm Vmesh = -340 V Kolkata, October 20, 2014 P. Colas - Micromegas 58

Si. Prot: protection against sparks NIKHEF Timepix chip + Si. Prot 20 μm + Micromegas Introduce 228 Th in the gas to provoke sparks 6. 3 Me. V 228 Th 220 Rn 6. 8 Me. V 2. 5× 105 e 2. 7× 105 e- Ar/Iso (80: 20) Mode TOT z = 10 mm Vmesh = -420 V Kolkata, October 20, 2014 P. Colas - Micromegas 59

SPARKS, but the chip’s still alive NIKHEF Timepix chip + Si. Prot 20 μm + Micromegas 228 Th 220 Rn Ar/Iso (80: 20) Mode TOT z = 10 mm Vmesh = -420 V Kolkata, October 20, 2014 P. Colas - Micromegas 60

A few ‘historical’ Micromegas First Chinese Micromegas, with fishing lines (Zhang Xiao. Dong, Lanzhou) One of the Tunis boxes Japanese copy of the Saclay box (T. Matsuda, K. Fujii, KEK) Kolkata, October 20, 2014 P. Colas - Micromegas 61

Box in Kolkata (copy from Saclay’s) First Bulk in Aachen

Practical use of Micromegas Kolkata, October 20, 2014 P. Colas - Micromegas 63

‘ Choose your material For first tests of a detector, a power supply with current limitation is preferred. Set the current limitation at 500 n. A for instance. The CAEN N 471 A is ideal for testing, though not very precise. They have 2 chanels, you can use one for the mesh and one for the drift cathode. Check your gasbox for gas-tightness : must bubble down to 1 l/h. Before connecting the electronics, ‘cook’ your detector (see next slide). Preamp: use a protected fast preamp (for instance ORTEC 142 series) and an amplifier-shaper (0. 5 or 1 microsecond peaking time), for instance ORTEC 472 or 672. Hunt noise (microphonic noise, radiated noise, noise from the grounds) Kolkata, October 20, 2014 P. Colas - Micromegas 64

‘Burning’ or ‘cooking’ your detector To make the detector stable for further operation, it must be ‘cooked’ : raise the voltage slowly to 550 -600 V (50 micron gap) or 800 -900 V (128 micron gap), step by step, to the level where it starts sparking. This has to be done in air It consists of burning small dusts (mostly fibres). A relatively high (ionic) current (200 -250 n. A) can remain. It will decrease after circulation of the gas and go down to 0(1 n. A). A detector which stands its voltage in air will always work in gas. Kolkata, October 20, 2014 P. Colas - Micromegas 65