Repositorio Institucional de Documentos

2013 72 Laura Seguí Iglesia Pattern Recognition in a High Pressure Time Projection Chamber prototype with a Micromegas readout for the 136Xe double beta decay Departamento Director/es Física Teórica Dafni, Theopisti Luzón Marco, Gloria Director/es Tesis Doctoral Autor Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA Departamento Director/es Laura Seguí Iglesia PATTERN RECOGNITION IN A HIGH PRESSURE TIME PROJECTION CHAMBER PROTOTYPE WITH A MICROMEGAS READOUT FOR THE 136XE DOUBLE BETA DECAY Director/es Física Teórica Dafni, Theopisti Luzón Marco, Gloria Tesis Doctoral Autor 2013 Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA Departamento Director/es Director/es Tesis Doctoral Autor Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA Pattern Recognition in a High Pressure Time Projection Chamber prototype with a Micromegas readout for the 136Xe double beta decay Memoria presentada por Laura Segu´ı Iglesia para optar al grado de Doctora en F´ısica en el Laboratorio de F´ısica Nuclear y Astropart´ıculas ´ Area de F´ısica At´omica, Molecular y Nuclear Departamento de F´ısica Te´orica UNIVERSIDAD DE ZARAGOZA Abril 2013 Agradecimientos En primer lugar me gustar´ıa dedicar esta tesis a la memoria de Julio Morales, que me permiti´o comenzar mi etapa investigadora en el grupo de F´ısica Nuclear y Astropart´ıculas de la Universidad de Zaragoza. Este trabajo no hubiera sido posible sin el cont´ınuo apoyo y gu´ıa de mis directoras Gloria Luz´on y Theopisti Dafn´ı (ǫυχαριστ ´ω). Quiero agradecerles su tiempo y dedicaci´on, as´ı como su consejo en las m´ultiples discusiones que hemos mantenido. A ambas les quiero agradecer el haberme ense˜nado a ser cr´ıtica con mi trabajo, a no dejar de buscar cabos sueltos y el haberme transmitido su valent´ıa y entusiasmo ayud´andome a formarme como f´ısica. Me gustar´ıa tambi´en agradecer a Igor G. Irastorza la confianza que siempre ha demostrado en mi trabajo, as´ı como el transmitirme su entusiamo por la F´ısica y estar siempre dispuesto a ayudarme. Tambi´en mi agradecimineto a Jos´e ´ Angel Villar por su constante int´eres a lo largo de estos a˜nos. Mi reconocimineto se hace extensivo a todos mis compa˜neros del departamento en especial a H´ector G´omez, por su disposici´on continua ante las numerosas dudas y problemas que han surgido durante la realizaci´on de esta tesis. Igualmente agradezco a Francisco Iguaz y Alfredo Tom´as su continua ayuda, en especial, con el software. Gracias tambi´en a Juan Castel y ´ Angel Lagraba porque sin ellos no hubiese sido posible la construcci´on del prototipo usado en esta investigaci´on. Quiero nombrar tambi´en especialmente a mi compa˜nera de despacho, Clara Cuesta, porque sin su apoyo y nuestras conversaciones arreglando el mundo este camino hubiera sido m´as largo. A ella y a Asun Rodr´ıguez quiero agradecerles los ratos de div´an. Al resto de compa˜neros mis sinceras gracias porque siempre hab´eis estado dispuestos a ayudarme. Me gustar´ıa agradecer a todo el experimento NEXT el haberme permitido realizar mi tesis dentro de la colaboraci´on. Tambien me gustar´ıa nombrar a los compa˜neros que he tenido durante mis estancias en otros centros. Especialmente mi agradecimineto a Esther Ferrer Ribas, por permitirme desarrollar mi investigaci´on en un gran centro como CEA de Saclay donde he mejorado mis conocimientos con los detectores Micromegas. Gracias tambi´en a Thomas Papaevangelou y a Iannis Giomataris por su amabilidad y consejo as´ı como al resto de mis compaeros en mis estancias en Saclay y en CAST. Por ´ultimo gracias a toda mi familia, en especial a Antonio Jos´e, Anne-Marie, ´ Alvaro e In´es, y a todos mis amigos por su apoyo y comprensi´on durante estos a˜nos. iii Contents Agradecimientos iii Preface 1 I Neutrino Physics and Motivation 3 1 Neutrino physics and double beta decay: theory and experiments 7 1.1 Neutrino Physics .................................... 8 1.1.1 Neutrino mass and oscillations . . . . . . . . . . . . . . . . . . . . . . . . 9 1.1.2 The seesaw mass mechanism and neutrino masses .............. 12 1.2 Double beta decay ................................... 13 1.3 The search of neutrinoless double beta decay . . . . . . . . . . . . . . . . . . . . 19 1.3.1 Techniques of detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 1.3.2 Experimental challenge . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 1.3.3 Sensitivity of an experiment . . . . . . . . . . . . . . . . . . . . . . . . . . 22 1.4 Experimental situation: Past, present and future experiments ........... 23 1.4.1 Past Experiments ................................ 24 1.4.2 Present situation ................................ 26 1.4.3 Micromegas for rare event searches . . . . . . . . . . . . . . . . . . . . . . 30 II Operation of Micromegas detectors focus on a ββ0νexperiment. 33 2 Phenomenology of particles in gases and gaseous detectors. Micromegas Detectors. 37 2.1 Ionization in gases ................................... 38 2.2 Transport of electrons in gases . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 2.2.1 Diffusion .................................... 41 2.2.2 Intrinsic uncertainties in gaseous detectors .................. 42 2.3 Operation of a gas detector and the avalanche multiplication region. Townsend coefficients ........................................ 45 2.4 Signal creation ..................................... 49 2.5 Energy resolution .................................... 50 2.5.1 Statistical contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 2.5.2 Other factors .................................. 52 2.6 Gaseous detectors .................................... 52 2.6.1 Time Projection Chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 2.6.2 Micromegas detectors ............................. 54 2.6.3 Micromegas Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 2.6.3.1 Bulk technology . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 2.6.3.2 Microbulk Technology . . . . . . . . . . . . . . . . . . . . . . . . 58 v Contents 3 NEXT-MM prototype: design and commissioning 61 3.1 The NEXT experiment ................................. 62 3.2 NEXT-MM ....................................... 64 3.2.1 Experimental setup ............................... 64 3.2.1.1 Vessel and internal components ................... 64 3.2.1.2 Gas system .............................. 66 3.2.1.3 Acquisition system .......................... 68 3.2.2 The Micromegas Detectors used ....................... 71 3.3 Operational tests with NEXT-MM prototype .................... 73 3.3.1 High pressure test ............................... 73 3.3.2 Vacuum and outgassing ............................ 73 3.3.3 High voltage tests ............................... 75 3.4 Main features of the detectors in Ar-isobutane: gain curves and first energy resolution estimations. ................................... 76 3.4.1 Electron transparency ............................. 78 3.4.2 Absolute Gain ................................. 78 3.4.3 Energy resolution ................................ 79 3.4.4 Cosmic Rays .................................. 80 3.5 Commissioning of the detectors and the electronics in NEXT-MM ........ 80 3.5.1 Bulk detector: First Results .......................... 81 3.5.2 Microbulk detector: First Results ....................... 82 4 NEXT-MM: First data with Xe-TMA 87 4.1 Analysis code ...................................... 88 4.2 First tracks ....................................... 89 4.2.1 Energy resolution results ........................... 95 4.3 Tracking with more than one detector ........................ 96 4.4 Trigger from an external source. ........................... 102 4.4.1 Energy resolution ................................ 105 4.4.2 Drift velocity .................................. 117 4.4.3 Preliminary attachment results ........................ 117 4.5 Summary ........................................ 119 III Physics potential of a ββ0νexperiment with Micromegas in a HP Xe TPC. 121 5 Background studies and discrimination algorithms 125 5.1 Simulation codes .................................... 126 5.1.1 Decay0 ..................................... 127 5.1.2 Geant4 ..................................... 127 5.1.3 Simulation of the Physical system and anylisis: RESTSoft ......... 128 5.2 Simulated Events .................................... 131 5.2.1 Signal ...................................... 131 5.2.2 Background ................................... 133 5.3 Simulated geometry .................................. 137 5.4 The topology of the events ............................... 138 5.4.1 Signal Events .................................. 138 5.4.2 Background ................................... 139 5.4.3 Deterioration of pattern recognition ..................... 142 5.5 Discrimination Algorithms ............................... 149 5.5.1 A figure of merit for background reduction . . . . . . . . . . . . . . . . . 149 5.5.2 Energy ...................................... 150 vi The main goal of the work presented here is to evaluate the physics potential of a neutrinoless double beta decay (ββ0ν) experiment based on a gaseous Xenon time projection chamber equipped with Micromegas. The ββ0νsearch is one of the hottest topics in particle and astroparticle physics nowadays because its measure could give basic information about neutrino nature, and intrinsic properties like its mass, and about the Universe. Oscillation experiments have proved that neutrinos are massive particles, opening the door to new physics. In this landscape, determination of the absolute mass of this particle has become a crucial point. Different experimental approaches are valid to obtain information about the neutrino mass: cosmological observations, direct measurements and ββ0νexperiment, but the latest is of special interest because its detection would lead to conclude that neutrino and antineutrino are the same particles, revealing its Majorana nature. Before summarizing the experimental studies performed, a general summary of the neutrino physics and properties which have motivated this work, is presented. In addition, past experiments and the results obtained are reviewed. Since only upper limits for the neutrino mass were reached, several experiments of the so-called new generation have been proposed claiming to be sensitive enough to explore lower mass regions thanks to the appearance of new technologies. A global view of these new experiments and techniques is also presented. 5 Chapter 1 Neutrino physics and double beta decay: theory and experiments Contents 1.1 Neutrino Physics .............................. 8 1.1.1 Neutrino mass and oscillations ...................... 9 1.1.2 The seesaw mass mechanism and neutrino masses . . . . . . . . . . . . 12 1.2 Double beta decay ............................. 13 1.3 The search of neutrinoless double beta decay . . . . . . . . . . . . 19 1.3.1 Techniques of detection .......................... 19 1.3.2 Experimental challenge .......................... 20 1.3.3 Sensitivity of an experiment ........................ 22 1.4 Experimental situation: Past, present and future experiments . . 23 1.4.1 Past Experiments .............................. 24 1.4.2 Present situation .............................. 26 1.4.3 Micromegas for rare event searches .................... 30 The neutrino was first proposed theoretically in 1930 by W. Pauli [1] to explain the continuous beta spectrum observed by J. Chadwick in 1914 and like this, preserve the energy, momentum and angular momentum. In 1933 E. Fermi proposed the name neutrino and one year later published the first theory of beta decay based on this particle [2], but it was not experimentally observed until 1956 in the Reines-Cowan experiment [3]. Reines was rewarded with the Nobel Prize in 1995. In 1962 L. M. Lederman, M. Schwartz and J. Steinberger showed the existence of more than one type of neutrinos by first detecting interactions of the muon neutrino at the Brookhaven National Laboratory [4] and they earned Nobel Prize in Physics in 1988. The third type of neutrino was discovered in 2000 by the DONUT collaboration at Fermilab [5]. Its existence had already been inferred by, both, theoretical consistency and experimental data from the Large Electron Positron Collider in CERN. 7 1. Neutrino physics and double beta decay: theory and experiments From the mid-60’s to 2002 remained unsolved the so-called solar neutrino problem. In the late 1960s, R. Davis and J. N. Bahcall were the first physicists to measure the flux of neutrinos from the Sun and detect a deficit in the Homestake Experiment.This deficit was confirmed twenty years later by the experiment Kamiokande II (in the Mozumi mine in Japan) [6]. In 2001 the SNO Experiment (Sudbury Neutrino Observatory, in Canada) gave the first evidence of neutrino oscillation that could solve this problem [7]. The neutrino oscillation concept was first proposed by B. Pontecorvo in 1957 [8] in a similar way as Kmeson oscillations. In 2005 Z. Maki, M. Nakagawa and S. Sakata developed this idea in the theory of neutrino flavour mixing and flavour oscillations [9]. The oscillations of neutrino can only be possible if neutrinos have mass. In 2002 Ray Davis and Masatoshi Koshiba won part of the Nobel Prize in Physics for experimental work that found the flux of solar neutrinos to be around a third of the number predicted by the standard solar model. The neutrino flavour oscillation has also been observed in neutrinos produced in the atmosphere or in reactors. As in the solar case, it causes a deficit in the number of detected neutrinos as proved by IMB [10] and in Kamiokande [6] in 1985 and was explained with neutrino oscillation theories in 1988 in the Super-Kamiokande [11,12] for atmosphere neutrinos. The deficit in reactor neutrinos was observed by KamLAND in 2002 [13]. The confirmation of neutrino oscillation entails that neutrino has mass, giving a rebirth of experiments looking for the neutrinoless double beta decay between nuclei and neutrino physics beyond the Standard Model. 1.1 Neutrino Physics The neutrino is an elementary particle with spin 1/2 and does not carry electric charge. Moreover, from the Standard Model (SM) it has zero mass. This is constrained by three independent reasons: there are only left-handed (LH) neutrinos, νL; there are only Higss doublets of SU(2)L and there are only renormalizable terms. In the SM these conditions apply and the three neutrino flavours, νe,νµand ντcan be distinguished by separate lepton numbers. In addition, as only LH neutrinos exist (and right-handed antineutrinos, RH), the chirality is not conserved, violating parity. However, from the experiment of oscillations it is known that neutrinos have mass 6= 0. To generate a Lorentz invariant neutrino mass term some of the previous conditions must be relaxed. A Dirac mass term may be generated if neutrino, νi, and anti-neutrino, ¯νi, have the same mass and different lepton number in the usual way as for fermions MD[¯νRνL+ ¯νc Lνc R].(1.1) The quantum states form a four-component object of mass mthat can be described by the Dirac equations 8 1.1. Neutrino Physics i(ˆσµ∂µ)ψR−mDǫψL= 0 (1.2) i(σµ∂µ)ψL−mDǫψR,(1.3) where ¯σµ=σ0~σ,σµ=σ0−~σ and (σ0~σ) are the Pauli matrices. The chirality states ψRand ψL are two-component spinors that couple to form a four-component bi-spinor quantum state. On the other hand, if there is no conservation of lepton number, as in a ββ0νdecay, Majorana [14] suggested an alternative description of massive fermions. A Majorana particle denotes a particle that is identical to its antiparticle, therefore, for neutrinos νi≡¯νi, while Dirac particles distinguish between the two. This characteristic implies that, mathematically, Majorana particles are two-component objects, ψR(mass mR) and ψL(mass mL) which obey independent equations i(ˆσµ∂µ)ψR−mRǫψ∗ R= 0 (1.4) i(σµ∂µ)ψL−mLǫψ∗ L= 0,(1.5) (1.6) where ǫ=iσy. The Majorana fields can be expressed in four-component notation: ψL(x) = −ǫψ∗ L(x) ψL(x)!, ψR(x) = ψR(x) ǫψ∗ L(x)!.(1.7) These spinors are autoconjugate states, ψc L(R)(x) = ψL(R)(x). The change in the chirality requires for the antiparticle to be equal to the particle and a massive one. The Lorentz invariant mass terms for a Majorana neutrino are ML[( ¯νL)cνL+ ¯νLνc L] (1.8) MR[( ¯νR)cνR+ ¯νRνc R].(1.9) Notice that these terms are only allowed for neutral particles. 1.1.1 Neutrino mass and oscillations Oscillations of neutrinos are a consequence of the presence of neutrino flavour mixing in vacuum. Therefore, neutrinos of a defined flavour (νl) have not to be necessary states of a definite mass (νi). The Lagrangian term of charged current used to describe the interactions of flavour neutrinos is given in the SM by 9 1. Neutrino physics and double beta decay: theory and experiments LCC(x) = −g 2√2X l=e,µ,τ ¯ νlL(x)γαlL(x)Wα(x) + h.c., (1.10) where W is the gauge boson mediating the weak interactions, ldenotes the lepton particles, νlL(x) are the LH component of the field of the neutrino νland γαthe Gamma matrices. The flavour neutrino field νlL(x) are linear combinations of the fields of three (or more) neutrinos νi with masses mi6= 0 νlL(x) = X i Uli νiL(x),(1.11) Ubeing a unitary 3 ×3 (N×Nif there exist N neutrinos families) mixing matrix called the Pontecorvo-Maki-Nakagawa-Sakata (PMNS) mixing matrix [8,9] U=    Ue1Ue2Ue3 Uµ1Uµ2Uµ3 Uτ1Uτ2Uτ3     = =    c12 c13 s12 c13 s13 eiδ −s12c23 −c12s23s13 eiδ c12c23 −s13s23s13 eiδ s23c13 s12s23 −c12c23s13 eiδ −c12c23 −s12c23s13 eiδ c23c13        eiα1/20 0 0eiα2/20 0 0 1     . (1.12) where cij = cos θij and sij = sin θij, being θij the mixing angles [0 −π/2] that takes into account the different origins of the oscillations (θ12 solar, θ23 atmospheric and θ13 from reactors). δ= [0−2π] is the Dirac CP phase that would be 6= 0 only if neutrino oscillation violates CP (this is expected but has not yet been observed). α1and α2are two Majorana CP violating phases only relevant if neutrinos are Majorana particles (can result in CP violation if their values differs from a multiple of π) and they do not affect oscillation experiments [15]. If experiment shows that the PMNS matrix is not unitary, a sterile neutrino or some other new physics is required. Equation 1.11 can be rewritten as |νii=X l Uli |νli(1.13) |νli=X i U∗ li |νii(1.14) The probability of the transition between an state νlto νl′in vacuum is given by P(νl→νl′) =  3 X i=1 Ul′ie−i ∆m2 ij L 2EU∗ li 2 ,(1.15) 10 1.1. Neutrino Physics where jis an arbitrary index 1,2 or 3, ∆m2 ij =|m2 i−m2 j|,Lis the propagation length and Eis the neutrino energy. This expression is obtained taking into account the ultra-relativistic nature of the neutrinos in the Schr¨odinger equation of the eigenstates [16]. In addition, to obtain this expression it has been assumed that the momentum of the initial and ending neutrino state is the same pi=pj[17]. The probability 1.15 depends on six parameters: two mass-squared differences, three mixing angles and a CP phase, δ. The oscillation probability of an antineutrino is the same as the one of the neutrino except for the mixing matrix Uthat is transformed to its complex conjugated U∗, implying that if Uis not real the oscillation probabilities are different. If CPT is conserved, the difference in the probability of oscillation between neutrinos and antineutrinos is an indication of CP violation. The first evidence of the oscillation process was observed by SNO [7] in solar neutrinos, and SuperKamiokande [11,12,18] in atmospheric neutrinos. Since these first measurements, some of the unknowns in previous equations have been measured or limited by different experiments dedicated to neutrino oscillations like SNO [7,19–22], KamLand [13], CHOOZ [23], K2K [24–26], MINOS [27], BOREXINO [28]. Recent results from Daya-Bay [29] give a non-zero value for θ13, the last of the angles to be determined. Combined with results from RENO [30], DoubleCHOOZ [31], T2K [32] and MINOS [33]θ13 = 0 is now excluded at more than 10σ. Two different neutrino mass spectra are compatible with existing data in the case of three massive neutrinos because there are only two independent neutrino mass squared differences. |∆m2 21|is usually identified with the smaller of the two neutrino mass squared differences. From the data it is known that |∆m2 21|is positive, so ν2is heavier than ν1(called like this for convenience), but nothing can be said about ν3. Therefore, the two scenarios are: 1. Normal spectrum or hierarchy (NH): m1< m2< m3, ∆m2 21 ≪∆m2 23 2. Inverted spectrum or hierarchy (IH): m3< m1< m2, ∆m2 21 ≪ |∆m2 13| When all three masses are significantly larger than ∆m2 32, the hierarchy is referred to as quasidegenerate (QD, m1∼ =m2∼ =m3), no matter which eigenstate is the lightest. In all the cases, one of the neutrinos have to have a neutrino mass of the order of 50 meV. In Figure1.1 the different scenarios are schematically shown. Figure 1.1: Sketch of the possible neutrino mass hierarchies: normal hierarchy (left) and inverted hierarchy (right) (image extracted from [34]). 11 1. Neutrino physics and double beta decay: theory and experiments 1.1.2 The seesaw mass mechanism and neutrino masses There are several ways to extend the Standard Model to introduce a mass mechanism that generates neutrino masses and also accommodates its small values naturally. If right-handed neutrinos exist, a Dirac mass can be generated as usual. However, this mass term would not explain easily the small values of the neutrino mass, because it requires fine-tuning of the Yukawa coupling (some extra-dimensions theory can accommodate this small Yukawa couplings). The most popular scenario is the see-saw mechanism [35], that goes beyond the Standard Model postulating the existence of a right-handed neutrino field inert under the electroweak interaction, and also requires a very large mass scale. This mass scale can be identified with the theoretically predicted scale of grand unification (∼1015 GeV). In the SM there is no need of right-handed neutrinos. Without them, neutrinos can acquire a mass by the existence of so-called irrelevant operators. These are operators of dimensions larger than four that require the existence of a cut-off. Otherwise, for an infinite cut-off, the mass of the neutrino would be zero. The cut-off cannot be the Planck scale, 1019 GeV, because it predicts a mass for the neutrino of the order of 10−5eV, not compatible with the observation from neutrino oscillation experiments. But, it is believed that the SM accommodates in a larger gauge group in the Grand Unified Theories (GUT) with a smaller cut-off. Very schematically, the consistency (absence of anomalies) of the GUT requires the existence of right-handed sterile neutrinos that, as are sterile under the SM gauge group (SU(3) ×SU(2) ×U(1)) acquire a mass of the order of the cut-off. If a sterile RH neutrino is introduced in the theory, the Lorentz invariant mass terms can appear in three different forms, one for the Dirac type 1.1 and two for the Majorana one 1.9. The most general Lagrangian can be written as L=−1 2¯νLcνc R MLMT D MDMR! νL νc R!+h.c. (1.16) For three neutrino families, the neutrino fields, ν, are three component vectors, and the elements of the mass matrix, M, are 3 ×3 matrices. In order to obtain the mass eigenstates we need to diagonalize the mass matrix, resulting in two generally non-degenerate eigenvalues for each flavour. In the see-saw mechanism we have MR≫MD≫ML∼0 (ML= 0 corresponds to the type I see-saw mechanism) 0MT D MDMR!(1.17) The light neutrino, after diagonalization, acquires a mass mν∼M2 D MR (1.18) 12 1.2. Double beta decay MDis of the order of the electroweak scale 102GeV, the observed neutrino mass is of the order of √δm2∼0.01 eV. This implies that the scale of the cut-off is MR∼1015 GeV, the GUT scale energy. In an effective way, the small value of the mass of the neutrino opens a window to the high energy physics and this can also address the baryonic observed asymmetry in the Universe [36]. Neutrino mass measurements They exist different ways that try to determine the exact value of the neutrino mass. Direct searches via Tritium βdecay Direct investigations do not rely on further assumptions on the neutrino mass type. The direct or kinematical approach to the search of the neutrino rest mass is based on the study of the relativistic energy momentum relation E2=p2c2+m2c4and in the energy and momentum conservation in weak semileptonic decays. The most sensitive direct searches for the electron neutrino mass up to now are based on the investigation of the electron spectrum of tritium β decay 3H→3He++e−+ ¯νe. The Mainz Experiment [37] reached a sensitivity of 2.3 eV at 95% C.L and the Troitsk [38] re-analysis put a limit for electronic neutrino mass of 2.05 eV at 95% C.L. The future experiment KATRIN [39], under construction, expects to have a sensitivity of 0.2 eV. A different approach to directly measure the electron neutrino mass is the use of cryogenic bolometers. In this case, the βsource can be identical to the βelectron spectrometer. Two isotopes are mainly under study: 187Re and 163Ho and the experiments, as MARE [40], are still in the early stage of development. Cosmological bounds Cosmology provides bounds on the sum of the neutrino masses. From the Cosmic Microwave Background (CMB) data of the WMAP experiment, combined with supernovae data and data on galaxy clustering it is obtained a limit Pim1≤0.3−1.3 eV at 95% C.L [41]. This limit is also marked in Figure 1.1 It follows from these measurements that neutrino masses are much smaller than the masses of charged leptons by at least 6 orders of magnitude. It is natural to suppose that this difference hides new physical mechanisms or processes beyond those predicted by the SM. Double beta decay This process will be explained in detail in the next section because it is the objective of this work. The detection of this nuclear process is proportional to effective neutrino electronic mass. 1.2 Double beta decay A double beta decay (ββ) is a weak second order nuclear transition between two even-even isobars. An initial nucleus (A, Z) decays to an isobar (A, Z + 2) through the emission of two 13 1. Neutrino physics and double beta decay: theory and experiments electrons. This transition is only possible in nuclei in which the βdecay is energetically forbidden or highly suppressed by a high angular momentum difference. A generic level diagram can be seen in Figure 1.2 (a). This process has been largely studied. The decay can exist in two modes (A, Z)→(A, Z + 2) + 2e−+ 2νe,(1.19) (A, Z)→(A, Z + 2) + 2e−.(1.20) The mode with the emission of neutrinos (ββ2ν) is a process allowed by the SM model and has been observed in different isotopes (summarized in Table 1.1). The neutrinoless mode (ββ0ν) has only been theoretically predicted and implies physics beyond the Standard Model (BSM) because the violation of the leptonic number is required. Moreover, it is only possible if the neutrino and antineutrino are the same particles, that is, if neutrino nature is Majorana. The Feynman diagrams of both decays are shown in Figure 1.3. Z+1 Z 0+ ββ 2+ 0+ 0+ Z+2 (a) 20 40 60 80 100 120 140 160 180 2000 2500 3000 3500 4000 48Ca(0.2%) 76Ge (7.8%) 82Se(9.2%) 96Zr (2.8%) 116Cd (7.5%) 130Te (33.8%) 150Nd (5.6%) 100Mo (9.6%) Transition energy (keV) Mass number 136Xe (8.9%) (b) Figure 1.2: (a) An scheme with a generic level diagram of a ββ decay (modified from [42]. (b) Plot of the mass and transition energy (Qββ in keV) for different isotopes in which the ββ0νdecay is more favourable experimentally due to its high Qββ value. Energies of the main background sources are also indicated in red (214Bi(β) = 3270 keV, 214Bi(γ) = 2447 keV, 208Tl(γ) = 2615 keV) (modified from [43]). Isotope T2ν 1/2y 48Ca (4.4±0.6 0.5)×19 76Ge (1.5±0.1) ×1021 82Se (0.92 ±0.07) ×1020 96Zr (2.3±0.2) ×1019 100Mo (7.1±0.4) ×1018 116Cd (2.8±0.2) ×1019 128Te (1.9±0.4) ×1024 130Te (6.8±1.2 1.1)×1020 136Xe (2.1±0.14(stat)±0.2(syst)) ×1021 150Nd (8.2±0.9) ×1018 288U (2.0±0.6) ×1021 Table 1.1: List of ββ2νdecay half-lifes measured in different isotopes, extracted from [44,45]. 14 1.3. The search of neutrinoless double beta decay purification, another requirement for a ββ experiment. The nuclear parameters for the isotope are also important. A nucleus with a higher phase-space and higher predicted nuclear matrix elements is more favourable to reach lower neutrino mass sensitivity as Equation 1.22 shows. In addition, an improvement of the nuclear theory may help to reduce uncertainties. Xenon gas fulfill most of these requirements. 2. A large amount of isotope. In order to develop the experiment in a real time, due to the low expected rate of ββ0νevents, a large amount of mass is required. To cover completely the degenerate hierarchy region of masses, an isotope mass scale between 10-100 kg is needed. In order to enter in the inverted hierarchy region the 1 ton scale is required. 3. High energy resolution. To discriminate the ββ0νpeak from the tail of the ββ2νmode and to reduce the background in the region of interest. It can be shown [49] that the expected number of counts of the ββ0νdecay over those due to ββ2νis R0ν/2ν=me 7Qββδ6 T0ν 1/2 T2ν 1/2 ,(1.25) where δ=∆E Qββ . From Equation 1.25 it is clear that candidates with a short ββ2νhalf-life (as Xenon) are favourable. Also, the better the energy resolution is, the better the discrimination between signal and background will be. For the next generation of experiments, an energy resolution of the order of 1% FWHM at Qββ is required. 4. Low background. It is a crucial point in a direct counting experiment because background can distort and even hide the awaited ββ signal and it is the only parameter that can be improved through the analysis of the detected events. The main components of the background are the cosmic radiation (muons, protons, gammas...) and the environmental radiation, mainly composed by the isotopes from the natural radioactivity chains of 232Th, 235U and 238U as well as some isotopes as 40K or 60Co. Of special interest is the Radon present in the air or that can be emanated from the materials. It is necessary to differentiate between external and internal backgrounds. To address the reduction of the background different mechanisms exist. (a) Underground placement of the detector to reduce the external cosmic contribution. The surface muons flux is 4 ×109m2y−1for an energy range between 0.01 MeV to 500 GeV [70]. This flux can be drastically reduced in an underground experiment, by the rock that acts as a shielding. (b) Passive shielding. The stopping power of the different materials for the different particles and energies can be used to design a passive shielding. Usually, lead, copper or water is employed to stop the high energetic gammas. For a more detailed discussion of the different materials see [17,71]. (c) Active shielding. It is used to reduce the background level by detecting the background events. Some of the most used techniques are: •Anti-muons veto. Even working in an underground facility sometimes the level of muons must be reduced to a lower value. The detection of the muons signal in two detectors (at different time) can allow for it. 21 1. Neutrino physics and double beta decay: theory and experiments •Anti-coincidence techniques. It is based on the fact that the background, mainly, will produce separated deposits of energy or will travel a longer distances than the electrons emitted in the ββ2νdecay. Therefore, if the detector is segmented, or if different detectors are used, anti-coincidence techniques can be useful to reject those events that interact in different parts of the setup. (d) Materials selection. For the inner components of the detectors it is very difficult or even impossible to develop passive shielding. Hence, a careful selection of the materials according to their radioactivity is required. A screening campaign has to be done by the experiments to control all the elements installed inside the detector. Special care has to be taken with those elements very close to the fiducial volume. (e) Discrimination Techniques. The differences between the signal and background events allow the development of analysis techniques to discriminate background. Once the materials are selected this is the only way to reduce background from the internal components to lower levels. An advantage in a gaseous time projection chamber is the possibility to reconstruct the track of the event and, based on pattern recognition, to carry out different analysis. The reduction of the background in a neutrinoless double beta decay of 136Xe in a gaseous TPC equipped with pixelized detectors is one of the objectives of this work and is covered in the third part of the thesis. Also it can be very useful the identification of the daughter nuclei, as in the proposition made by EXO [72] to tag the 136Ba daughter ion in the 136Xe decay, to eliminate all the backgrounds. In experiments without tracking capabilities, as GERDA, pulse-shape analysis is developed [73]. 1.3.3 Sensitivity of an experiment In order to compare different experiments and also to design a future one, a figure of merit can be defined to obtain a value of the expected sensitivity for the effective neutrino mass in a ββ0νexperiment. The half-life of the decay is the signal measured in a ββ0νexperiment that can be related to the effective electronic neutrino mass with expression 1.22. The half-life of a radioactive decay can be written as T0ν 1/2=ln 2 λ(1.26) where λis the decay constant that can be expressed as the activity, A, divided by the number of nucleus, N. The number of nucleus is N=NAv M·f W.(1.27) In this expression NAv is the Avogadro’s number, Wis the molecular weight of the ββ decaying isotope atoms, fis the isotopic abundance and Mthe mass of beta emitter. The activity can be expressed in terms of the detected signal, s, in a time tas 22 1.4. Experimental situation: Past, present and future experiments A=s ǫt (1.28) being ǫthe efficiency of the detector. Combining expression 1.27 and 1.28, Equation 1.26 can be rewritten as T0ν 1/2= ln 2 ·NAv M·f W ǫt s.(1.29) It is clear that, to obtain the best sensitivity, a detector that maximizes the signal counting rate over the background has to be built. The signal sensitivity is proportional to the statistical precision of the background determination. Since the number of background counts is a linear function of time, the sensitivity increases with the square root of time. In absence of a positive signal an upper limit to the expected signal can be given as, at 1σof confidence level, s < √b·∆E·M·t(1.30) where bis the background level in c keV−1kg−1y−1, ∆Eis the energy window (the energy resolution of the detector) in keV, Min this case has to be expressed in kg. With this expression a lower limit for the half-life of the decay can be obtained from 1.29 as T0ν 1/2>ln 2 ·NAv ǫ·f Wrǫt s.(1.31) From this expression the factors determinant are the mass of the beta emitter, the background level expressed in c keV−1kg−1y−1and the energy resolution of the detector. M×Tis usually defined as the exposure of the experiment, the sensitivity increases as the exposure increases until some point that the behaviour is asymptotic. Finally, to obtain a limit for the neutrino mass expression 1.22 is used on which variables related to the nucleus and its structure play an important role. 1.4 Experimental situation: Past, present and future experiments At the present, we are entering in a very exciting era for neutrinoless double beta decay experiments. Those proposed and others already taking data will cover the sensitivity of the Heidelberg-Moscow claim [74] and will also enter into the inverted hierarchy region of masses. Before explaining the present experiments, a review of the past ones and their main results has to be done. A more complete description of the different experiments is done in [42], [75] and in the recent [43]. 23 1. Neutrino physics and double beta decay: theory and experiments 1.4.1 Past Experiments Among the past experiments there were two based on germanium detectors: IGEX (International Germanium Experiment) [76], placed in the Canfranc Underground Laboratory (LSC) with around 6 kg of 76Ge enriched at 86% and the Heidelberg-Moscow (HM) experiment [77], placed in Gran Sasso (LNGS) with around 10 kg of isotope. Part of the HM collaboration claimed in 2001 an evidence of a ββ0νsignal [74]. This result was not accepted by some members of the community as well as from part of the collaboration [78], [79], [80] and for this reason one of the main goals for the future experiments is to cover the pointed region of masses. The result claimed in the last paper [81], with higher statistics and a different pulse-shape analysis, is a T0 ν= 2.3×1025 y that corresponds to a neutrino effective mass of hmνi= 0.44 eV. Apart from this result, the two experiments (IGEX and HM) are compatible and they put an upper limit on the half-life of the decay of the order of T0 ν= 1.5×1025 y (hmνi ≤ 0.33 −1.35 eV) [76,82]. The bolometric technique is, also, a suitable one for ββ0νsearches, and rare events in general, due to its high energy resolution [83,84]. From 2003 to 2008 the Cuoriccino experiment, proposed in 1994 (an extension of MIBETA, and a previous step for CUORE), operated with TeO2crystal bolometers with an isotopic abundance of 33.8% and a mass of the order of 40 kg in the LNGS. They improved the results obtained previously with the MIBETA experiment [85] and put an upper limit for the half-life of the 130Te isotope of T0 ν>2.8×1024 y that corresponds (with a 90% C.L) to a neutrino effective mass of hmνi ≤ 0.30 −0.71 eV, [86,87]. Up to now the mentioned experiments operated in the calorimetric approach. The best experiment of the external-source technique was NEMO3 (Neutrino Ettore Majorana Observatory) [88] operating in the Laboratoire Souterrain de Modane (LSM), in France. The detection technique is based on the reconstruction of the track of the particle bent in the presence of a magnetic field in Geiger wire cells. Also a calorimeter surrounding the tracking wires was installed with scintillator blocks connected to photomultipliers (PMT) to read the energy signal. It could measure different beta emitters at the same run installing foils of different isotopes in the middle of the detector that is divided into 20 sectors. The ββ2νmode was measured for the seven installed isotopes while for the ββ0νmode only upper limits were obtained being the most stringent one for the 100Mo with a half-life of T0 ν>5.8×1023 y (hmνi ≤ 0.61 −1.26 eV). Finally, it is important to mention the Gothard Xenon TPC Experiment [89] that ran in the 90s. It was the first double beta experiment using a gaseous TPC filled with the emitter gas, in this case 136Xe mixed with CF4at 5 bar. They only put limits for both ββ modes but proved that the tracking capabilities could reduce the background even with a modest energy resolution [90]. The idea of using a gaseous TPC was abandoned for few years due to drawbacks of the technique as the poor energy resolution ot its difficulty to scale to higher masses. Currently, it is a promising technique exploited by the EXO and NEXT experiments. The details of the previously mentioned experiments are listed in Table 1.2. 24 1.4. Experimental situation: Past, present and future experiments Experiment Isotope Qββ Mass Technique Efficiency Background in the RoI ERes hmνiref. and Laboratory (keV) (kg) (c keV−1kg−1y−1) at Qββ (eV) IGEX (LSC) 76Ge 2039 6 Ge diodes 90% 0.1 4 keV 0.33-1.35 [76] HM (LNGS) 76Ge 2039 10.9 Ge diodes 95% 0.17 4 keV 0.35-0.60 [77]* CUORICCINO 130Te 2528 40.7 TeO284% 0.1 8 keV 0.30-0.71 [86],[87] (LNGS) bolometers NEMO3 100Mo/82Se 7/1 Tracking 18% 0.1 11-14% FWHM 0.61-1.26 [88] (LSM) +calorimeter Gothard 136Xe 2457 3.3 Tracking 70% 0.01 164 keV >4.4×1023 y [89][90] Xe TPC +calorimeter Table 1.2: Summary of the main parameters relevant in a neutrinoless double beta decay. Here are mentioned the main past experiments that have been discussed in the text. In all cases a limit to the effective neutrino mass was obtained (excepting the claim by the HM collaboration mentioned in the text). In the last column are given the references for a complete description of the experiments and results. 25 1. Neutrino physics and double beta decay: theory and experiments 1.4.2 Present situation The main experiments currently taking data or proposed for the near future are briefly described. A full review of them can be found in [42,43] and references therein. In the present situation the main goals are to cover the sensitivity of the HM claim in the degenerate region of masses and to enter in the inverted hierarchy one (hmν≤50imeV). To fully investigate this region (from hmνi= 50 meV to 20 meV) the required background must be of the order of 1 c/y/ton and an excellent energy resolution is also mandatory, of the order of 1% FWHM. Moreover, the different tracking or analysis capabilities can decrease the background level and improve the expected sensitivity in those experiments with a limited energy resolution. At the end, one has to choose one characteristic over the others when designing a ββ0νexperiment. Along this section the different experiments will be briefly explained and in Table 1.3 are summarized the main features of the current experiments. Some of them as GERDA, EXO or KamLand-Zen are already taking data. Most of the proposed experiments are based on the search of the double beta decay in the same isotope as in the past but with a higher mass and new techniques. This is the case for GERDA and MAJORANA looking for the nuclear transition in 76Ge. GERDA (GERmanium Detector Array): It is a 76Ge experiment with Ge diodes arrays immersed in liquid Argon [91], [67] installed in LNGS. The experiment is divided in different phases. The first one started in November 2011 using 8 refurbished enriched diodes from HM and IGEX [92] (14.6 kg) with an energy resolution of 4.5 keV at 2039 keV. The background level was 0.1 c keV−1kg−1y−1. With the application of pulse-shape analysis (PSA) they expect to decrease it by one order of magnitude. In this first run they obtained the half-life for the ββ2ν mode [93] being of T2ν 1/2= 1.84+0.14 −0.10 ×1021 y. In the second phase they will operate 20 kg of broad-energy Germanium with a higher energy resolution and better PSA and also in this phase will take place the installation of a liquid Argon (LAr) scintillation veto. It will start the present year. MAJORANA: The Majorana project [94], [95] will be composed by Ge diodes operated in conventional copper cryostat having a mass of 1 ton of 76Ge. They will be located at Stanford Underground Facility (SURF) in the USA. The first phase, called MAJORANA Demonstrator [96], will consist of 40 kg of Ge and its main purpose is to prove that a background level of 0.001 c keV−1kg−1y−1can be obtained in the 1 ton extension. The proven energy resolution is 0.16% FWHM at Qββ. The MAJORANA and GERDA experiments will merge for the future 1 ton experiment combining the best results in both cases [97]; this corresponds to the third phase of GERDA. As extensions of past experiments there are the CUORE and the SuperNEMO experiments. CUORE (Cryogenic Underground Observatory for Rare Event): [98] is an extension of CUORICCINO with 988 TeO2bolometers to be installed in the LNGS. They will be arrange in 19 towers operating at 10 mK. They look for the ββ0νdecay of 130Te with a total mass of 200 kg. The expected energy resolution is 0.25% FWHM and the expected background is of the order of 0.02-0.01 c keV−1kg−1y−1, one order of magnitude better than in CUORICCINO. With these 26 1.4. Experimental situation: Past, present and future experiments background levels a sensitivity of T0ν 1/2≈9.7×1025 y is expected (50-130 meV). The first tower that will operate in LNGS along this year is known as CUORE-0. SuperNEMO [69,99]: As in NEMO3 it will follow a modular concept of 20 units with 5 kg of isotope each one and will be installed in LSM. The electrons emitted from the foil source placed in the middle of the detector will traverse the tracking module composed by Geiger cells and enter the calorimeter made with 600 channels of organic scintillator blocks coupled to PMTs. It improves the NEMO3 features having an efficiency of 30%, an energy resolution of 7.5% FWHM and could reduce the background to 0.01 c/kg/y. It will investigate mainly the 82Se double beta emission because is a good ββ0νemitter and part with the advantage of previous knowledge of backgrounds and analysis from NEMO previous phase. Other isotopes can be studied as 150Nd,96Zr and 48Ca. The predicted 5y sensitivity for the 82Se is 1026 y, which corresponds to a neutrino effective mass between 40-110 meV. At the moment, a dedicated radiopurity detector, BIPO [100], has been installed in Canfranc (LSC) to measure, with enough sensitivity, the contamination of the ββ source foils that will be installed in SuperNEMO. Another approach is the SNO+ experiment investigating the 150Nd isotope detecting the Cerenkov light in a scintillating media. SNO+: [101], [102] is an upgrade of the SNO (Sudbury Neutrino Observatory) solar neutrino experiment located at the SNOLAB in Canada. The idea is to fill the detector with Nd loaded liquid scintillator to investigate the 150Nd ββ decay. It is expected to be filled along this year with a mass of 43.7 kg and an energy resolution of 6.4% FWHM. After 3y the expected sensitivity is 1.3×1025 y (an effective neutrino mass between 120-410 meV). Finally 500 kg of enriched isotope will be used. The main drawback is the difficulty to enrich this quantity of 150Nd, while one of the advantages of this experiment is the low background. The expected sensitivity for the 500 kg run is T0ν 1/2= 1 ×1025 , that will allow to investigate the region for the effective neutrino mass from 55 to 180 meV. The next group of experiments are those looking for the emission in 136Xe. KamLand-Zen (Kamioka Large Anti Neutrino Detector Zenon): [103] is a modification of the KamLand detector located in the Kamioka mine in Japan. Xe gas is dissolved in an organic liquid scintillator contained in a nylon balloon. This balloon is suspended at the centre of the KamLand detector and it is surrounded by 1 kton of liquid scintillator contained in an outer balloon that acts as active veto. The enriched Xe is 300 kg, being 125 kg in the sensitive volume. It has an energy resolution of 10% FWHM. After a total live time of 112.3 days a value for the half-life of the ββ2νmode was obtained [104] of T2ν 1/2= 2.38 ±0.02(stat)±0.14(syst)×1021 y, completely compatible with the previous result by EXO. An extra background was observed in both regions ββ2νand ββ0ν. Studies point out that this contributions are due to the emission of unexpected isotopes when the Fukushima fallout [105] or to spallation of 136Xe on surface. The main issue now is to reduce this background, for example, recirculating the Xenon to reach the expected value that allows to explore sensitivities of the order of 1024 y for the ββ0νmode. Also they plan to increase the amount of Xenon up to 1 ton. Another proposal with Xe is BOREXINO with few tons of 136Xe [28] dissolved in a scintillator. 27 1. Neutrino physics and double beta decay: theory and experiments After Gothard, the option to use a Xe filled TPC is considered in the EXO and NEXT experiments. EXO (Enriched Xe Observatory) [72]: is a liquid Xe TPC that can measure both the scintillation light and the ionization in a plane of 250 large-area avalanche photo-diodes (LAAPDs). The first phase, called EXO-200, is already taking data in the Waste Isolation Pilot Plant (WIPP). It has 175 kg of Xe enriched to 80% 136Xe. It has an energy resolution of 3.9% FWHM in the region of interest and a background of 0.0015 c keV−1kg−1y−1. They were the first to obtain the half-life of the ββ2νmode [45] with a value of T2ν 1/2= 2.11 ±0.14(stat)±0.21(syst)×1021 y in complete agreement with the posterior result obtained by the KamLand-ZEN collaboration [104]. They also put a lower limit for the half-life of the ββ0νmode [106] of T0ν 1/2>1.6×1025 y. In future phases it will have a mass between 1 and 4 ton with a sensitivity of 1027 y; this extension is called nEXO [107]. NEXT (Neutrino Experiment with a Xe TPC): [68] will be installed in the LSC. It will operate with 100 kg gas Xenon at 10 bar and will measure the electroluminescent signal. More details are given in Chapter 3. This idea is implemented in the present work with the development of a high pressure TPC fill with gaseous Xe to look for its beta decays and the use of pixelized detectors as Micromegas to discriminate, based on topology, between signal and background. With these detectors an energy resolution of at least 1% is expected [108], [109]. Other proposals not mentioned here are summarized in Table 1.3 and are mainly in an R&D phase. These experiments are MOON [110,111], CANDLES [112], LUCIFER [113], COBRA [114], and AMORE [115,116] (that will look for Dark Matter also). See also [42] for other projects. In the future, in order to investigate the normal hierarchy region of neutrino masses (hmνi between 2-4 meV) it is needed to go to masses of beta emitter of 10 ton and background levels of 0.1-0.01 c/ton/y. 28 1.4. Experimental situation: Past, present and future experiments Experiment Isotope Mass Technique Background ERes %FWHM T0ν 1/2hmνiStatus . and Laboratory (kg) (c keV−1kg−1y−1) at Qββ (y) (meV) GERDA (LNGS) 76Ge 35 Ge diodes 0.02 0.25 2 ×1026 60 −200aIn progress MAJORANA (SURF) 76Ge 30 Ge diodes 4 c/t/y 0.16 1026 90 −300aIn progress MAJORANA-GERDA 76Ge 1000 Ge diodes 0.001 0.16 6 ×1027 10 −40bR&D CUORE (LNGS) 130Te 200 TeO2bolometers 0.01 0.25 1026 50 −130aIn progress SuperNEMO 76Se 100-200 Tracking + 0.01 7.5 2 ×1026 40 −110aInstallation first (LSM) calorimeter module, R&D EXO (WIPP) 136Xe 140 TPC ionization+ 0.0015 3.9 >1.6×1025 <140 −380cCurrent scintillation 136Xe 1000 Id. (+Ba tagging) ≥0.0015 1.6 8 ×1026 20 −55aR&D NEXT (LSC) 136Xe 100-150 Tracking + 0.0008 <1 5.9×1025 102 −129bIn progress calorimeter KamLand-Zen 136Xe 130 Liquid 6.8 2 ×1026 40 −110aCurrent (Kamioka) 136Xe 1000 scintillation 1027 18 −50aR&D SNO+ (Sudbury) 150Nd 43.7 Liquid 5 ×10−76.4 1.3×1025 120 −410bIn progress scintillation 500 idem 40 R&D COBRA (LNGS) 116Cd 117 CdZnTe semiconductor 0.001 1.9 >1026 50 R&D + tracker LUCIFER (LNGS) 82Se 10 ZnSe bolometers 0.0001 <1 2 ×1026 ≈100bR&D AMORE (Y2L) 100Mo 50 CaMoO4bolometers — 0.5 3 ×1026 20-60eR&D MOON 100Mo 480 Tracking+scintillator 20 c/ton/y 2.2 3 ×1026 45dR&D CANDLESS 48Ca 3200 Scintillation 1 c/ton/y 0.2 — 30dR&D (Kamioka) Table 1.3: Summary of the main neutrinoless double beta decay experiments at present with its main features and its status. References to each experiment can be found along the text. The effective neutrino mass range depends on the nuclear matrix element used. For this reason it is mentioned the reference from where these values were taken: (a) is [50], (b)[43], (c) is [106], (d) [51], (e)[115] 29 1. Neutrino physics and double beta decay: theory and experiments 1.4.3 Micromegas for rare event searches The forthcoming generation of neutrinoless double beta decay experiments will need to go to the few tons scale to reach sensitivities down to 10-20 meV and to fully investigate the inverted hierarchy region of masses. However, larger masses do not assure such sensitivity and it has to come with continuous improvements in energy resolution and in background reduction techniques [15]. Specially, the background has to be of the order of 10−5c keV−1kg−1y−1. The background level is up to a certain point related to the energy resolution that marks off the region of interest (RoI). Nonetheless, the required minimum energy resolution is limited by the intensity of the ββ2νmode, because it is needed to separate the signal peak from the tail of the ββ2ν distribution. In 136Xe a minimum 4.5% FWHM at 2458 keV is required at the 100 kg scale, while in the landscape of a 1 ton experiment it is needed to be at least 2.5% FWHM [15]. Within the TREX project [117] (TPCs for rare event searches), the use of Micromegas readouts in a gaseous TPC together with ultra-low background techniques to develop ideas for rare event searches is studied. TPCs were normally used for tracking rather than calorimetry. This was due mainly to the large volumes and times of exposure needed in those experiments making these detectors complex in terms of mechanics (wire tension), electrostatically (field under control over large volumes) and operation. In addition, the use of wire planes implied a poor energy resolution. The Gothard Xenon TPC Experiment was a pioneer experiment that showed the background rejection potential of a gaseous TPC, based on the characteristic topology of the ββ0νsignal (a long track with two high energy depositions at both ends) [118]. Though promising, the idea stopped for few years due to drawbacks of the technique as the low energy resolution and spatial resolution obtained. These perspective changed with the use of the previously invention of the Micropattern Gas Detectors (MPGD) in 1988 by A. Oed [119], when he demonstrated the first functional Microstrip Gas Chambers (MSGCs). From this moment different detectors arose replacing the traditional wires with readouts that use metallic strips or pads precisely printed with photolitograpic techniques. In general the MPGD can achieve higher spatial resolution, are simpler and more robust and also had an impact on the electronics. One of the most attractive MPGD to use in double beta decay searches (and rare event searches in general) is the MICRO MEsh GAseous Structure detector (Micromegas) [120]. The Micromegas detector also have a very good energy resolution and have shown a high stability, granularity and homogeneity in experiments as CAST [121], [122], [123], the first application of this detector in rare event searches (dark matter). The idea to use a TPC has already given very good results, as discussed previously, in the EXO experiment and it has been proposed for the NEXT experiment. In general, two approaches are promising: the detection of the electroluminescent signal (NEXT, EXO) or the use of a novel charge readout detector as Micromegas, option also considered by the mentioned experiments. The use of a Micromegas detector for the neutrinoless double beta decay searches was evaluated in [108]. One of the main drawbacks could be the stability and features of these detectors at high pressure. It has been proven that Micromegas have an energy resolution of around 3% FWHM with an alpha source up to 5 bar in pure Xenon [108], with the best result being 2.5% FWHM at 4 bar. With a detailed analysis removing edge effect this value is improved obtaining a 1.8% FWHM at 4 bar. Recent studies with a gamma source in Xe gas mixtures with thrimetilamine 30 Chapter 2 Phenomenology of particles in gases and gaseous detectors. Micromegas Detectors. Contents 2.1 Ionization in gases ............................. 38 2.2 Transport of electrons in gases ..................... 39 2.2.1 Diffusion .................................. 41 2.2.2 Intrinsic uncertainties in gaseous detectors ................ 42 2.3 Operation of a gas detector and the avalanche multiplication region. Townsend coefficients ....................... 45 2.4 Signal creation ............................... 49 2.5 Energy resolution ............................. 50 2.5.1 Statistical contributions .......................... 50 2.5.2 Other factors ................................ 52 2.6 Gaseous detectors ............................. 52 2.6.1 Time Projection Chamber ......................... 53 2.6.2 Micromegas detectors ........................... 54 2.6.3 Micromegas Technology .......................... 57 Radiation and particle detection is the main goal in particle physics experiments. It can be detected in a gas by the effect that they produce on it. The primary modes of interaction are ionization and excitation of gas molecules along their path. In the following we will focus on the ionization process. The electron cloud produced can be detected on detectors based on electron collection, as the Micromegas, which will be used in this work. The charged particles ionize the medium when interacting with the gas molecules releasing part of their kinetic energy to produce an e−-ion pair. The required energy depends on the medium but is of the order of few eV. The produced electrons can then be drifted by the application 37 2. Phenomenology of particles in gases and gaseous detectors. Micromegas Detectors. of an electric field and amplified in avalanches until reaching the readout plane where they are read as electrical pulses. These electrical signals could provide information about the original location and ionization density of every track segment. The choice of the medium is very important and depends on the purpose of the experiment. Here the processes will be described in general but we give more details for the case of Xenon gas. A major advantage of the use of gas instead of liquid is that the topological information reconstructed by a pixelized detector can be used to distinguish between signal and background. However, if a big amount of isotope is needed, as in a neutrinoless double beta experiment, it implies a big volume or working at high pressure. Regarding the gas, in the case of very penetrating radiation, heavy noble gases, as Ar or Xe [124], are advantageous because of their high stopping power. As a consequence the interest in these gases is increasing. The advantages that Xe presents for a ββ0νexperiment have been already pointed out in Section 1.3.2 and are its high Qββ (2458 keV), its slow half-life of the order of T2ν 1 2≈2×1021 y and that it easy to enrich. Xenon can also emit secondary scintillation light that could create additional ionization that, if it is detected, can produce a loss in proportionality. This effect can be reduced adding a small amount of a polyatomic gas that suppress the photon-induced effects. This is usually called a quencher. The basic properties of the gas change by adding a small concentration of some additive, specially interesting is the case when the secondary gas has an ionization potential smaller than the one of the base gas. If there exist long-lived or metastable excited states in the principal gas, then a Penning mixture is obtained. In those cases an improvement in the gain and energy resolution arises because an extra ionization appears through a collision with the metastable state that produces more ion-electron pairs. The increment in the electrons increases the gain and that fact reduces the statistical fluctuations in the total number of particles per energy that corresponds to an improvement in the energy resolution [71]. In particular, some studies have been carried out with Trimetylamine (TMA) showing this performance [109]. An extra advantage from this gas is that the diffusion coefficient is smaller. We will discuss in Chapter 5how this improves the topological reconstruction of the track compared with pure Xenon, where the diffusion is higher. In the first sections of this chapter the different processes that take place in a gaseous detector are described. As general references [71,124–127] have been used. Another important selection is the detector to be used which depends on the requirements for the expected signal. As emphasized previously in a neutrinoless double beta decay experiment it is required a very good energy and spatial resolution and a low background budget. All these characteristics are fulfilled in a Time Projection Chamber detector equipped with Micromegas [108,109]. In the second part of the chapter a general view of the different gaseous detectors is briefly presented to move on to explain in detail the Micromegas detectors and their different technologies. 2.1 Ionization in gases A charged particle passing through a gas creates both excited and ionized molecules along its path. The encounters with the gas molecules are purely random and they are characterized 38 2.2. Transport of electrons in gases by the mean free path, λ= 1/(nσI) where n is the density of electrons, and σis the effective cross-section area for the encounters with the gas, that in general will depend on the energy of the scattering. The number of encounters along a length Lfollows a Poisson distribution. Different ionization mechanisms exist. They can be distinguished between primary and secondary ionization. When an incident electron gives enough energy to the atom, a bound electron can be ejected, resulting in the ionization of the atom and in the formation of an electron-ion pair (e−-ion pair). The energy of the ejected electron depends on the incident electron energy and on its binding energy. If the energy carried away by the ejected electron is enough to produce secondary ionization, in the same manner as the primary, the process can continue. An avalanche of electrons is produced until the energy is less than the ionization potential of the atom. This process is sketched in figure 2.1 (a). As it is a probabilistic process, not all the electrons that have energy higher than the ionization potential of the atom produce subsequent ionization. The practical parameter is the number of e−-ion pairs produced along the trajectory of the particle. This quantity depends on the so-called W-value, the energy required to produce an e−- ion pair. It is determined experimentally and it is higher than the ionization potential because some energy is also usually lost due to excitation. Wdepends on the cross section, i.e. it depends on the gas (its composition and density) and on the nature and energy of the incident particle. However, experimentally, it has been found that it is independent of the incident energy above a few keV for electrons and a few MeV for alphas. In the case of Xenon, Whas been found to be 22 eV/(e−-ion pair). For more values in different gases reader is referred to [71] or [127], for example. Therefore, the number of e−-ion pairs can be obtained experimentally as the energy of the incident particle over the energy needed to create an ion pair, W, as shown in Equation 2.1. In the case of partial energy loss of the incident particle it can be computed using the stopping power dE dx as in Equation 2.2. N=E W(2.1) N=1 W dE dx ∆x(2.2) 2.2 Transport of electrons in gases The electrons and ions produced in a gas move randomly and are scattered by collisions with the gas molecules. They rapidly thermalize adopting a Maxwellian distribution in absence of an external electric field. At room temperature this thermalization occurs at 0.04 eV. The diffusion of the charge is isotropic because there is not a preferred direction of motion. Therefore, the diffusion in any direction can be described using a Gaussian distribution. 39 2. Phenomenology of particles in gases and gaseous detectors. Micromegas Detectors. + Incident particle Atom A A+ e- + A+ eElectron avalanche (secondary ionization) Primary ionization Electric field (a) Cathode Anode Drift Distance Ionization radiation E (b) Figure 2.1: (a) Effect of the passing of radiation through a gas. Primary e−-ion pairs are formed when interacting with the gas molecules. In the presence of an electric field a secondary ionization is produced leading to an avalanche of charge pairs. (b) Liquid-drop profile of the avalanche multiplication due to the fact that the electrons have smaller mass than the ions and then drift faster to the anode. Images modified from [127]. In the presence of an external electric field, E, the electrons gain an extra velocity between collisions due to the acceleration produced by the electric force. This extra velocity appears macroscopically and it is called drift velocity,vd, vd=eE mλ v=eE mτ, (2.3) where vis the velocity of the electron and λits mean free path, so τis the average time between collisions. eis the electron charge. The drift velocity depends on the pressure of the gas. Because the density of molecules in space changes with the pressure, the probability of collisions is modified. Then, the mean free path of the electron is inversely proportional to the pressure (P) and the electron energy is a function of the reduced field E/P. Therefore, the drift velocity is also a function of the E/P ratio and it is called reduced drift velocity. Expression 2.3 is deduced in [126] and is based on the classical kinetic theory of gases. A more rigorous statistical theory, it is the transport theory that is based on the solution of transport and energy conservation equations for free electrons under the influence of external fields [128].In the transport theory, an electron distribution function is introduced for the six dimensions space (positions and velocities) and appropriate differential equations are describing density and energy conservation. In the last decades, the drift properties and the diffusion in a large variety of gases and gas mixtures have been studied (experimentally and theoretically). In particular, a computer programme Magboltz [129] exists (and it is continuously updated with new data) to simulate electrons in gases giving drift velocity and diffusion coefficients with an accuracy better than 2%. In Table 2.1 values of the drift velocity and diffusion at 1 kV/cm for the drift field are summarized for the gases under study in this work. A drift field of 1 kV/cm is required to obtain the optimum gain of the detectors at 10 bar. The values have been obtained using Magboltz. In Figure 2.2 40 2.2. Transport of electrons in gases Reduced drift field (V/cm/bar) 10 2 10 3 10 4 10 5 10 sec)µ Drift velocity (cm / -2 10 -1 10 1 10 0% 0.5% 1% 2% 3% 4% 5% 10% Figure 2.2: Dependence of the drift velocity with the reduced drift field for pure Xe and XeTMA mixtures at different concentrations. The addition of a quencher increases the velocity of the gas. is shown the drift velocity in pure Xe and Xe-TMA at different concentrations obtained with Magboltz in function of the reduced drift voltage1. Gas P Drift velocity σLσT (bar) (cm/µs) (µcm/√cm) (µcm/√cm) Xe 1 0.11 960 2487 Xe 10 0.11 355 850 XeTMA (3.5%) 1 0.24 218 312 XeTMA (3.5%) 10 0.24 218 312 Ar 1 0.3 1050 800 Ar 10 0.25 312 1000 Ar-C4H10(2%) 1 3.2 35 370 Ar-C4H10(2%) 10 3.0 400 700 Table 2.1: Drift velocities and diffusion coefficients obtained using Magboltz for the gas mixtures used in this work. Results where obtain at Vd= 1 kV cm−1at 10 bar and 150 kV cm−1at 1 bar. 2.2.1 Diffusion Diffusion is one of several transport phenomena that occur in nature. In a gas the atoms and molecules are in constant thermal motion. The flux of particles per unit area and per unit time, ~ J, can be written as ~ J=n~vd−D~ ∇n, (2.4) where Dis the diffusion coefficient and nthe number density of carriers. The electrons and ions produced when a particle interacts with the gas molecules also take part in the random thermal motion and therefore have some tendency to diffuse away. Because of 1Simulations done by F. Iguaz 41 2. Phenomenology of particles in gases and gaseous detectors. Micromegas Detectors. their lower mass the electrons have in average a higher thermal velocity than the ions (of the order of cm/µs for electrons and 10−2cm/µs for ions, at room temperature) and their diffusion is much more pronounced. In absence of an external electric field the diffusion is isotropic and a point-like collection of free electrons will spread around the original point (we take here the origin) following a Gaussian spatial distribution whose width will increase with time. In any direction and integrating the relation 2.4 with vd= 0, it can be described by dN =N √4πDte−x2 4Dt dx. (2.5) This distribution represents the number of charges dN that can be found in a differential element dx at a distance xfrom the centrer of the initial charge distribution after a time t.Dcan be used to determine the standard deviation through the relation σ=√2Dt. (2.6) In the presence of an electric field the diffusion is no longer isotropic and in this case it is a tensor with two non-zero components: a longitudinal, DL, in the direction of the drift field and a transversal component, DT, orthogonal to the drift field. Integrating again the relation 2.4, now with the drift field term present, it is obtained n(x, y, z, t) = N 4πDTtp(4πDLt)exp −(x2+y2) 4DTtexp −(z+vdt)2 4DLt,(2.7) where the initial condition at t= 0 is n(x, y, z, 0) = Nδ(x)δ(y)δ(z) (with δrepresenting the Dirac-δdistribution). A detailed study can be found, for example in [130]. In general, the longitudinal component is smaller than the transversal one as can be seen in Table 2.1. In Figure 2.3 is plotted the obtained transversal and longitudinal diffusion coefficient in Xe and Xe-TMA mixtures using Magboltz2. It can be seen that the diffusion is less important when the drift field increases. Also, the addition of a quencher to the main gas reduces the diffusion coefficients. 2.2.2 Intrinsic uncertainties in gaseous detectors Ideally the measured current in any readout would be all the electrons coming from the generated e−-ion pairs. However, due to different processes, not all of them are collected and some losses of charges occur along the drift distance. These losses can be important in the gain and energy resolution of a detector and must be taken into account for precise measurements. In this section some of the main effects that lead to the uncertainties are discussed. Recombination of ions and electrons in gases 2Simulations done by F. Iguaz 42 2.2. Transport of electrons in gases Reduced drift field (V/cm/bar) 10 2 10 3 10 4 10 5 10 Long. diff. coef. (um bar^1/2 cm^-1/2) 2 10 3 10 0% 0.5% 1% 2% 5% 10% (a) Reduced drift field (V/cm/bar) 10 2 10 3 10 4 10 5 10 Trans. diff. coef. (um bar^1/2 cm^-1/2) 2 10 3 10 4 10 0% 0.5% 1% 2% 5% 10% (b) Figure 2.3: Longitudinal and transversal diffusion coefficients versus the reduced drift field for pure Xenon and Xe-TMA mixtures. It can be seen how the diffusion decreases with the addition of a quencher. The recombination process depends on the number of charge carriers and on the recombination coefficient. This process implies a decrease in the number density of e−-ion pairs , n, along the drift distance (along the time) that follows the equation dn dt =prn+n−,(2.8) where pris the recombination coefficient, usually of the order of 10−7cm−3s−1and n−(n+) are the number density of electrons (ions). Though this process is not usually very important, there is a general tendency to increase with the e−-ion pair concentration, and therefore, in the case of noble gas detectors the recombination rates are usually higher [124] than in other gases. This effect is reduced when applying an electric field between the cathode and anode plane. Experiments with Xe-TMA [131] show that recombination is reduced doping the Xe with TMA. In pure Xe the recombination coefficient was measured by [132] with densities between 0.05 and 1.7 g/cm3. It was found that it is practically independent of gamma ray energy until the low energy range (below 400 keV). Effect of contaminants or attachment The effect of impurities in the gas can produce the loss of some charge carriers, and consequently, a degradation in the detected signal. These contaminants are electronegative molecules that usually have several vibrational energy levels and, therefore, they are able to absorb electrons in a wide range of energies. When absorbing electrons they can form stable or metastable ions. Noble gases and most organic molecules can only form stable negative ions at collision energies of several eV, which are higher than energies reached during any electron drift. However, there are some molecules that are capable of absorbing electrons at much lower collision energies, these molecules are called impurities and are mainly oxygen, air and water. In Table 2.2 are shown some of the main impurities with its electron affinity. 43 2. Phenomenology of particles in gases and gaseous detectors. Micromegas Detectors. Molecule Electron affinity (eV) Negative Ion Electron affinity (eV) O20.44 O−1.47 C23.54 C−1.27 Cl22.38 Cl−3.61 OH 1.83 H−0.75 Table 2.2: Electron affinities of different molecules and ions from [127] Three different mechanisms to produce the capture of electrons by the contaminants are usually considered: 1. Radiative Attachment: the capture of an electron leaves the atom in an excited state. It occurs only in molecules with positive electron affinity. e−+X→X−∗ ,(2.9) X−∗ →X−+γ. (2.10) It has usually a small cross-section. 2. Dissociative Attachment: the molecule that captures the electron dissociate into simpler molecules emitting an electron with an energy smaller than the original one. e−+AB →AB∗+e−→A++B−+e−.(2.11) Not all the polyatomic molecules emit secondary electrons, they can also dissociate into stable molecules. e−+AB →AB−→A∗+B−.(2.12) 3. Three-body attachment: in this process the polyatomic molecule captures an electron and it transfers its excess energy to another molecule. e−+AB ↔AB−∗ ,(2.13) AB−∗ +X→AB−+X. (2.14) The energy released in this process is known as electron affinity. In order to compare the attachment in different gases, a probability of electron attachment can be defined p=1 τeνe ,(2.15) where τeis the mean lifetime of the electrons and νetheir collision frequency, both magnitudes can be determined experimentally. Then, expression 2.15 represents the probability of capture in a single collision. The values for the same gases are shown in Table 2.3. The extremely small capture lifetimes make that this probability increases and could be a serious problem for precise measurements. 44 2.3. Operation of a gas detector and the avalanche multiplication region. Townsend coefficients Gas τe(s) νe(s−1)p O27.1×10−42.2×1011 6.4×10−9 CO21.9×10−72.1×1011 2.5×10−5 H20 1.4×10−72.8×1011 2.5×10−5 Cl24.7×10−94.5×1011 4.7×10−4 Table 2.3: Probability of electron capture in a single collision, p, for different gases. pis defined in 2.15.τeis the capture lifetime of the electrons and νetheir collision frequency. The electron lifetime, τe, is the average time for an electron to be drifted before it is captured by an electronegative molecule and therefore, lost to generate the signal. It follows an exponential law N(t) = N0e −t τe,(2.16) where N0is the number of original electrons. Attachment Origins Attachment, as mentioned, is produced when the gas contains electronegative impurities. These molecules may be there because a low enough vacuum level had not been achieved before injecting the gas or they may appear once the gas is inside the chamber due to outgassing processes: •Real leaks: air and oxygen can reach the gas. •Internal or virtual leaks: volumes of air separated from the main one, by little tubes like screws or by additives. They are difficult to extract pumping. Therefore, the design of the different elements have to be made taking this effect into account and also, a possible solution is to make little holes in the conflictive elements that allows the gas flux. •Vaporization: release of impurities molecules that were trapped in the surface materials. •Permeation: through the detector walls. This effect is important in TPCs if they have many plastic O-rings. Apart from the real leaks, the other processes can be controlled. For example, selecting low outgassing materials. Also pumping and making bake-out cycles. Results of bake-out cycles and outgassing measurements are presented in Chapter 3for the designed prototype. 2.3 Operation of a gas detector and the avalanche multiplication region. Townsend coefficients A gas filled detector shows different regions of operation depending on the applied bias drift voltage. The main differences between these modes are the amount of charges produced in the ionization, their movement inside the detector and whether the output signal is proportional to 45 2. Phenomenology of particles in gases and gaseous detectors. Micromegas Detectors. the original energy of the particle or not. In Figure 2.4 the different regions are shown. In the following a brief explanation of each region is given focusing in the avalanche process. For more information the reader is referred to [127], for example. Figure 2.4: Regions of operation of gas filled detectors based on the applied bias voltage. The two curves correspond to two different energies of incident radiation. Figure extracted from [127]. In the absence of an external electric field, as it has been mentioned the electrons and ions produced by an external radiation interacting with the gas molecules are quickly recombined, as the voltage increases some of the ions can reach the detector. The recombination region refers to this region up to a voltage high enough to consider recombination negligible. The output signal at these voltages of operation is not proportional to the incident energy. As the voltage increases the detector enters in the ion chamber region where all the charges produced are recollected. The current measured is proportional to the energy deposited by the incident particle and it is called saturation current. The detectors designed to operate in this mode are called ionization chambers. Then, as the voltage is kept increasing the detector enters the so-called proportional region. In this region the produced e−-ion pairs have enough energy to produce additional pairs, process called secondary ionization, as can be seen in Figure 2.1 (a). The electrons, owing to their small mass, are the ones that cause most of the subsequent ionizations. This multiplication of charges is exploited in the proportional detectors to increase the height of the output signal being still proportional to the initial energy of the incident particle. The final number of pairs produced is proportional to the initial number of charges. In the next section the charge multiplication process is explained in detail. Avalanche multiplication An avalanche multiplication is a process that consists in a rapid multiplication of charges by the primary charges produced from the incident radiation. This multiplication occurs at higher voltages, of the order of few kV/cm, that also decreases the charge collection time of the detector. 46 2.6. Gaseous detectors In this section it is explained in detail the principle of operation of a TPC and of a Micromegas detector. In addition the main features and technologies of manufacturing these readouts are also explained. 2.6.1 Time Projection Chamber A revolution in charged particle tracking happened in 1974, with the Time-Projection Chamber (TPC) invention by David Nygren [146]. The main idea is to define a long drift distance volume where a constant electric field is applied. The electrons produced by ionising particles in this volume follow the electrical field lines towards the MWPC detectors, located at the end of the volume. Once the electrons are close to the anode they suffer a multiplication process and the avalanches are detected in the pads plane. The two coordinates of the track of the particle are obtained in the MWPC detector while the third one is obtained from the drift time measurement of the electrons to the anode plane. For a given constant gas mixture and a drift electric field, the electron drift velocity is also constant. Typical values are of the order of few cm/µs. In its first design, a cylindrical TPC with a central cathode and two anode planes at the end-caps was constructed. It was used for collider physics, specially it was designed for the e+-e−PEP-4 collider experiment at Stanford. Furthermore, a magnetic field was applied along the length of the cylinder, parallel to the electric field, to measure momentum from the curvature of the track produced by the magnetic field over charged particles. The actual designs can vary from this one while keeping the same philosophy. In the case of rare event searches a very large and dense volume is required in order to increase the interaction probability and no magnetic field is needed. In Figure 2.5 are shown the different process that take place in a TPC and that have been already mentioned in the previous sections: 1) the ionization of the medium by an incident particle, 2) the drift of the secondary electrons to the readout plane, 3) the amplification of the signal in the device, and finally 4) the creation of the signal. Figure 2.5: Sketch of the different process that a charge suffer while traversing a TPC. While the principle of operation looks simple, there are many constraints on its realisation, especially for the gas choice. The drift over a long distance can produce a loss in the spatial 53 2. Phenomenology of particles in gases and gaseous detectors. Micromegas Detectors. resolution due to attachment or diffusion in the gas. Therefore, the gas has to be carefully selected in function of its attachment, drift velocity and diffusion, as well as for their ionisation and multiplication capabilities. Finally, the applied electric field must be uniform and constant, for this reason field cages are constructed to assure its uniformity. 2.6.2 Micromegas detectors The Micromegas detector is a two-stage parallel-plate avalanche chamber invented by I. Giomataris in 1995 [120]. It consists of two parallel plates: a metallic micromesh suspended over an anode plane by insulator pillars. The gap is usually 50-100 µm. It is combined with a conversion-drift space; inside a TPC, the mesh and the cathode define this drift region. The mesh and the anode define the amplification region. The operating principle of a Micromegas is shown in Figure 2.6. Voltages are applied in the three electrodes of the detector. The electrons freed in the gas are drifted to the detector by an electric field of the order of 102−103V cm−1. Then they cross the mesh holes entering in the amplification region where an avalanche takes place producing detectable signals. The generated ions are collected in the grid and the electrons in the anode. Usually amplification fields of 104−105V cm−1are applied. In this configuration of fields the field lines from the drift region will enter through the holes of the mesh as is shown in Figure 2.7 assuring the transmission of the produced electrons into the gap. Also the fields are homogeneous inside both regions except a very little deviation in the openings of the mesh. The accumulation of positive ions on the insulator can also modify locally the field and therefore lower the gain. This was one of the main drawbacks in the previous MPGD and is surpassed in the Micromegas because the high field in the gap also assure the fast collection of the ion cloud in the mesh. Moreover, only a small part of the ions, inversely proportional to the electric field ratio [147], escape to the drift region, effect known as ion backflow. Also the fast evacuation of the ions, together with the high granularity of the detector give them high rate capabilities [148]. The detector gain depends directly on the size of the gap. The gain is defined in Equation 2.19. Substituting αfrom Equation 2.21 it is obtained M=exp(A P d e−B P d/V ),(2.36) where A, B are gas parameters, Pis the operating pressure and Vthe voltage. From this equation different aspects can be extrapolated. The gain increases as dincreases, until it reaches a maximum as is shown in [148]. The maximum multiplication value can be obtained differentiating the equation and it is d=V/B at 1 bar. In general the gain variations due to flatness defects between the mesh and the anode are minimized for d≈V/Bp. The variation of the gain with the gap size also follows the same relation. For higher pressures, smaller gap sizes will produce higher gains. 54 2.6. Gaseous detectors Gap: 30-100 μm Drift field Typical 102-103 V/cm Amplification field Typical 104-105 V/cm Incident particle eMicromesh Figure 2.6: Sketch of the different process that occurs in a Micromegas. In the conversion region, delimited by the cathode and mesh, the particles ionize the gas and the electrons are drifted to the Micromegas. Electric field of 102−103V/cm are usually applied. In the gap of the Micromegas an avalanche of electrons produces the amplification of the charge. In this region electric field of 104−105V/cm are usually applied. The ions are detected in the mesh and the electrons in the anode, that can be pixelized. Figure 2.7: Electric field lines in the Micromegas, image from [148]. The gap dimension also plays an important role in the performance of the Micromegas. The gap reduction from 100 µm to 50 µm reduces the signal rise time by a factor of 3. Therefore the signal in the mesh is decreased from 100 ns to ∼30 ns. This allows to work at moderate gains preventing sparks and also decrease the time the ions are collected. The drift velocity of the electrons is 100 times bigger, therefore the typical signals are of the order of 1 ns. These fast signals allow to reach time resolution of the order of 1 ns, in particular, in the KABES experiment [149] they 55 2. Phenomenology of particles in gases and gaseous detectors. Micromegas Detectors. obtained a time resolution of 0.60 ns. The electron transmission,T, is defined as the proportion of primary electrons (nD) that reach the amplification region and are therefore detected (nA) T=nA nD .(2.37) The absolute transmission is difficult to measure experimentally and as an approximation, signals are normalized to the maximum. For a given amplification field, the transparency is low at high drift fields because most of the field lines end up in the mesh. As the drift field decreases the transparency is close or equal to one along several values of the field ratio (known as the plateau of operation). In this configuration, most of the field lines go through the mesh holes. For very low drift fields, the transparency sharply decreases because the drift velocity of the electrons is very low and attachment starts to be more relevant. The transmission depends mainly on the geometry of the detector (size holes and pitch) and the ratio of fields, as mentioned. The Micromegas is the MPGD with the best energy resolution. The reasons are inherent to the avalanche process and some of them have been already mentioned as the spatial homogeneity of the gain or the minimal loss of charge before the avalanche. Also the gain is less dependent on environmental factors as temperature or pressure than in MWPC. Moreover, the ions are evacuated fast and efficiently. The relation with the avalanche can be seen from expression 2.35,where the energy resolution can be written in terms of the Fano factor and the fluctuation in the number of charges produced in the avalanche. FWHM = 2.35rW E(F+b).(2.38) In this expression Eis the energy of the incident particle, Wis the mean energy per e−-ion pair, Fthe Fano Factor and bis avalanche contribution. The energy resolution depends on the amplification field (the gain) and on the ratio of fields (transmission). It is stable when the electron transmission is maximum. With respect to the gain, the energy resolution deteriorates at lower gains and also at very high ones, because it is close to the spark limit and the field could not be very stable. Other factors considered non-fundamental can worsen the resolution but in principle they can be overcome (although with experimental difficulties). Some of them are the non-uniformity of the readout plane, problems of equalization of multiple channels or the attachment. In general, energy resolutions of 11% FWHM with a 5.9 keV 55Fe are routinely achieved in Argonisobutane mixtures for the new Microbulk Micromegas. Tests in different gases or pressures are not standardized but studies with different mixtures were obtained showing similar results [150]. In addition, values with an 241Am alpha source were obtained, showing an energy resolution of 1.8% FWHM at 5.5 MeV (with evidence of less than 1%) in Ar-isobutane [151]. Data with the same source have been obtained in pure Xe [108] with results of around 3% FWHM up to 5 bar. Recently, studies in Xe-TMA mixtures showed energy resolution of 7.6% (9.6%) FWHM 56 2.6. Gaseous detectors at 1(10) bar for the 22.1 keV 109Cd peak [109]. With respect to the spatial resolution, in most of the cases is only limited by diffusion, and values down to 11 µm were achieved [152]. 2.6.3 Micromegas Technology The construction of Micromegas detectors is in continuous innovation and development in close collaboration of the experimentalists with the manufacturer experts from CERN and CEA/Saclay. The original Micromegas [120] micromesh was a metallic grid made of thin electroformed Nickel (as the Frisch grid). The mesh was stretched and glued on a removable glass-fibre frame. The anode strips of gold-coated copper were printed by metal deposition techniques in a Kapton substrate. The first strips were 5 µm thick while thinner ones could be obtained by vacuum deposition. The pillars between the anode and the mesh were small cylindrical insulating spacers made of photo-imaginable resin, 100 µm thick and 150 µm diameter, and they were printed on top of the anode by a standard printed circuit technique. The two frames were screwed together. By applying a voltage between them, the mesh was pulled down by the intense electric field obtaining the flatness and parallelism of both surfaces and the gap size was defined by the pillars height with an accuracy better than 10 µm. Subsequently, another type of mesh was developed [153]. It was based on etching techniques. The raw material was a double-sided Kapton foil of 50 µm stretched on a frame. The holes and pillars pattern was done applying two lithographic masks at both sides of the Kapton. Kapton was partially removed defining the pillars spacers. The high accuracy of the process allows to print on a 5 µm copper grid with 25 µm opening holes and a pitch of 50 µm. Since then, efforts have been focused on producing the amplification region from a single piece to avoid the delicate operation of screw the anode to the mesh.Reducing, at the same time, the mass amount of materials. Two different detectors were developed with this “all-in-one”fabrication techniques: the bulk and microbulk Micromegas. 2.6.3.1 Bulk technology In the bulk technology [154] a woven wire mesh is used instead of the usual electroformed micromesh. Different materials are available (Fe, Cu, Ti, Ni, Au) in rolls 40 ×40 m that are commonly produced making this option inexpensive and that can cover large surfaces. They are also robust for stretching and handling. In a bulk Micromegas the anode plane carrying the copper strips, the mesh and a photoresitive film with the thickness of the gap are laminated together at high temperature to form a single object. The pillars are produced by photolitograpic methods etching the photoresistive film. The fabrication process is sketched in Figure 2.8. The main advantages of this technique are the uniformity and robustness, lower capacity (that implies a low noise) and easy fabrication (they do not need a support frame). It allows the construction of large area detectors, also curved ones.They have shown acceptable energy resolution (18% FWHM at 5.9 keV) and good maximal gains (of the order of 104in Ar-isobutane). The 57 2. Phenomenology of particles in gases and gaseous detectors. Micromegas Detectors. mesh thickness could be a disadvantage in some applications, while the ideal gap size depends on the application, for example bigger gaps are an advantage at low pressure. Figure 2.8: Fabrication process of the bulk Micromegas detector [154]. 2.6.3.2 Microbulk Technology In the microbulk fabrication [151] the raw material is a thin flexible poliamide foil with a thin copper layer on both sides. It is usually glued on top of a rigid substrate that provides the support and usually is were the strips or pixels are made by photolitography. Then a thin photoresistive film is laminated on top of the Kapton and it is isolated by UV light to produce the mask. The copper is removed by standard lithographic obtaining the holes of the mesh. Two approaches were made to obtain the gap space. In one the polyamide is etched and partially removed in order to create thin pillars below the copper mesh. In the other approach, the polyamide is completely removed except the points where small pillars are created. In order to achieve this an additional spot is formed during the insulation process producing a Copper spot of 200 µm after the litographic process. By controlling the etching duration the polyamide can be completely removed except under these Copper spots. A sketch of the process is shown in Figure 2.9. With this process there are no space connections under the mesh and each avalanche takes place under the mesh holes decreasing the fluctuations in the process. This property, together with the fact that the homogeneity of the gap is higher than in previous methods, allows to achieve the best energy resolution among the MPGD detectors. The pillars are under the shadow of the copper pads, avoiding any dead space. 58 •Kapton foil (50 μm), both side Cu-coated (5 μm) •Construction of readout strips/pads (photolithography) •Attachment of a single-side Cu-coated kapton foil (25/5 μm) •Construction of readout lines •Etching of kapton •Vias construction •2nd Layer of Cu-coated kapton •Photochemical production of mesh holes •Kapton etching •Cleaning Figure 2.9: Fabrication process of the bulk Micromegas detector. Another important advantage of a microbulk Micromegas is that it can be extremely light and most of the raw material is Kapton and Copper, two materials that can achieve the best levels of radiopurity [155]. Indeed, the first radiopurity study of Micromegas [156] shows that current microbulk detectors contain levels at least as low as 57 ±25 µBq/cm2for 40K, 26 ±14 µBq/cm2 for 238U and <9.3µBq/cm2for 232Th (the big values of the errors make them compatible with a lower limit due to the Minimum Detectable Activity of the germanium detectors used). Moreover, the contamination probably comes from the treatment of the materials used and can be then reduced with a careful selection of them. Next steps are focused on the identification of the contaminating steps in the fabrication and find alternatives, as well as to make a more precise measurement. In this chapter a brief study of the phenomenology of particles in gaseous detectors, emphasizing the Micromegas detectors which are the ones with the better features for a neutrinoless double beta decay experiment. In the next chapter the description of the TPC constructed and the first results with different Micromegas are presented. 59 Chapter 3 NEXT-MM prototype: design and commissioning Contents 3.1 The NEXT experiment . . . . . . . . . . . . . . . . . . . . . . . . . . 62 3.2 NEXT-MM ................................. 64 3.2.1 Experimental setup ............................. 64 3.2.2 The Micromegas Detectors used ..................... 71 3.3 Operational tests with NEXT-MM prototype . . . . . . . . . . . . 73 3.3.1 High pressure test ............................. 73 3.3.2 Vacuum and outgassing .......................... 73 3.3.3 High voltage tests ............................. 75 3.4 Main features of the detectors in Ar-isobutane: gain curves and first energy resolution estimations. . . . . . . . . . . . . . . . . . . . 76 3.4.1 Electron transparency ........................... 78 3.4.2 Absolute Gain ............................... 78 3.4.3 Energy resolution .............................. 79 3.4.4 Cosmic Rays ................................ 80 3.5 Commissioning of the detectors and the electronics in NEXT-MM 80 3.5.1 Bulk detector: First Results ........................ 81 3.5.2 Microbulk detector: First Results ..................... 82 In the NEXT collaboration, a detector prototype with Micromegas readouts has been constructed and commissioned in the group of the University of Zaragoza. In this chapter, first, the NEXT experiment and the operation of the different prototypes are presented. A separate section is dedicated to describe the NEXT-MM prototype, based on Micromegas readouts. Its setup is explained summarizing all the work done for the commissioning. Finally, the first operational results are presented as well as the first data taken to commission the Micromegas detectors. The analysis program for the data acquisition and the first results regarding energy resolution and tracking obtained in Xe-TMA will be deeply described in the next chapter. 61 3. NEXT-MM prototype: design and commissioning 3.1 The NEXT experiment The NEXT experiment (Neutrino Experiment with a Xenon TPC) [68] aims at measuring the neutrinoless double beta decay of 136Xe using a high pressure (HP) gaseous electroluminescent time projection chamber (TPC). It will operate with 100 kg of Xe (90% enriched to 136Xe) at 10 bar with the possibility to go up to 15 bar, that means a mass of Xe of around 150 kg. The NEXT experiment is a broad collaboration with Spanish groups together with international groups from Europe and America. It will be installed in the Canfranc Underground Laboratory (LSC [157]) under the Spanish Pyrinees, more specifically under the Tobazo mountain that provides an overburden of 2450 m.w.e. The use of Xe has several advantages for the double beta decay searches. The first advantage is its high Qββ value of 2458 keV [158]. Therefore, the ββ0νsignal is separated from most of the emissions presented in the natural radioactive chains. Recently, the ββ2νhas been measured by [45], [104] and it has a slow half-life of the order of T2ν 1 2≈2×1021y. As we can see in Equation 1.25, if the ββ2νmode is considered as background for the ββ0νmode, it is better if its half-life is short. Another advantage is its high abundance, of 9% in 136Xe; also, as it is a gas, is easy to enrich to higher levels. In terms of its nuclear structure it has a high phase-space factor and favourable nuclear matrix elements. NEXT combines two of the technological advantages in the ββ0νdecay searches. One is its promising good energy resolution, of around 1% FWHM at Qββ [159,160]. On the other hand, a gaseous TPC can provide topological information of the track of the event using a pixelized detector. This allows to develop very efficient discrimination algorithms to separate between signal and background events. In order to evaluate the power of rejection in a pixelized TPC based on pattern recognition, Montecarlo simulations can be performed. This is one of the main points of this work and results are presented in Chapter 5. The NEXT detection principle is based on the SOFT concept: Separated Optimized Energy Function from Tracking: it measures separately the energy and performs the reconstruction of the path of the event. In the experimental setup there will be two planes of detectors, as sketched in Figure 3.1. At one side of the field cage is placed a photomultiplier plane (PMTs) with which the energy will be measured, and on the other side a Silicon photomultiplier’s plane (SiPMs) having enough pixelization to reconstruct the path followed by the event. In the PMTs plane is also determined the start-of-event time, to obtain the third dimension of the track. The physical process and detection concept in NEXT is the following. Any particle interacting in the gas (signal or background) will produce a first scintillation light that will be detected in the PMTs plane and will give the start-of-event, t0. As explained in Chapter 2, then an ionization process starts. The electrons produced are drifted by an electric field to an electroluminescence region. It consists of a gap of the order of few mm delimited by the SiPMs plane and two meshes, where a high voltage is applied to obtain an electric field of the order of few kV/cm in the gap. With these electric fields the atoms of the gas when excited by the electrons produce secondary scintillation light. This secondary light is measured both, in the PMTs plane and in the SiPMs 62 3.2. NEXT-MM For the mesh, the signal is read by a pre-amplifier (PA), in this work, a Canberra 2004 PA, which also allows to supply the operational voltage to the Micromegas through the bias line. The PA output is sent through an RC low-pass filter to reduce the electronic noise. The preamplified signal is a measurement of the total energy of the event registered in the mesh, so it could be used as input in an amplifier module or used as trigger signal for the pixels acquisition. The amplified signal can be recorded by a multi-channel analyser (MCA) to obtain the spectrum of the mesh. For the pixelized anode, the signals of each pixel are independently read using a reduced version of the T2K experiment electronics [171], that is based on the chip AFTER [172]. The imprint of the pads in the Micromegas is connected through a limande to the inner side of the feedthrough. Outside, another limande cable connects to an interface card designed on purpose to connect flexibly to a FEC card. The connection between the limandes, Micromegas and feedthrough is done using special high density contact Samtec connectors. In the case of a bulk detector there exist four points to do the connection to the pixels in the Micromegas and in the case in which the surface is covered with 4 microbulk detectors each one is read independently. Therefore, in both cases, the connection of the full surface of pixels is done through 4 of the following chains: (inner limande + outer limande + interface) to the FEC card. These FEC cards are connected to a FEM (Front-End Mezzanine) that concentrates the different signals and, through a Full Duplex Gbit optical, linked to a DCC (Data Concentrator Card) card RS232. The DCC is connected by Ethernet to the PC where the DAQ interface is installed. The FEM card has a capacity to read 1728 channels divided in connection to 6 cards through 4 ERNI connectors each one with 72 channels per ERNI (6 ×(72 ×4) channels can be read independently). A schematic view of the chain is shown in Figure 3.7 and a picture of each element is presented in Figure 3.8. The acquisition trigger for the pixels is given through the DCC card. It is a positive TTL signal. Several possibilities have been used for the trigger of the pixels. Most of the data are taken in self-trigger with the mesh signal. In this case the mesh signal is sent to a Low Level Discriminator (LLD), either from the PA or from the amplifier. The signals that go beyond a given voltage threshold are sent to a NIM-TTL converter and then to the DCC (the trigger must be a positive signal, if it is necessary, it is switched in the NIM-TTL module). Other options are possible, as for example, to send an external trigger signal, if we want to measure in coincidence with another setup. This option is explained in more detail in Section 4.4. 69 3. NEXT-MM prototype: design and commissioning Limandes MMs Detector FT FT FT FT NEXTT2K interface FEC FEC FEC FEC FEM DCC Optical Fiber or Etehrnet Optical Fiber PA LA Digitizer Mesh NEXTT2K interface NEXTT2K interface NEXTT2K interface Inside TPC ERNI Connectors LLD NIMTTL Trigger OR Figure 3.7: Conceptual sketch of the acquisition system of the prototype. In black the chain for the pixels signal to the T2K electronics. In red the mesh signal is shown, that can also be used as trigger for the pixels (blue line). (a) (b) (c) (d) (e) (f) Figure 3.8: Pictures of the different elements of the electronic chain to read the pixels. (a) Flat cable, or limande, to transport the pixels signals. (b) Samtec connector that is used between the Micromegas and the limande and between the different limandes. (c) feedthrough that transport the signal from inside to outside the vessel with the two limandes connected. (d)(e) Interface cards used that connects to the FEC card. (f) FEM card connected with the FEC cards in a Faraday cage. 70 3.2. NEXT-MM 3.2.2 The Micromegas Detectors used The detectors used in NEXT-MM are Micromegas detectors. As it was explained in detail in Section 2.6.2, they have an amplification gap between a mesh and an anode plane of the order of few µm where a high electric field produces an avalanche of electrons that can be detected in both sides. Two different technologies were used in NEXT-MM, the bulk and the microbulk. Bulk detector The first detector installed inside NEXT-MM was a bulk Micromegas. It had a diameter of 30 cm, with a gap of 50 µm, mesh holes diameter of 35 µm and a pitch of 100 µm. It covered all the sensitive surface of the prototype as shown in picture 3.9 (a). The anode was pixelized in 1152 independent pixels, as it could be seen in the design presented in Figure 3.11, of 0.8×0.8 cm2 each one. (a) (b) (c) (d) Figure 3.9: Micromegas used (a) Bulk detector (b) Microbulk detector. Four equal detectors have been designed in order to cover all the NEXT-MM surface. (c) Installation of two of the microbulk detectors. They are placed over a copper plate, the external ring not only fixes the detector but also connect the rim lines to a voltage a bit higher of the mesh to avoid dead space (see text for more details). (d) Installation of the four microbulk detectors. Microbulk detector Up to today limitations exist in the fabrication procedure for a Microbulk, and only detectors with a diameter up to 20 cm can be done. For this reason, to cover all the active surface of the detector, four identical sectoral detectors, with the shape of a quarter of a circle, were manufactured (Figure 3.9 (b)). They are made with the microbulk technology with a radius of 15 cm, and a gap of 50 µm, mesh holes diameter of 35 µm and a pitch of 100 µm. The tolerances 71 3. NEXT-MM prototype: design and commissioning achievable in the assembly of the neighbouring modules can be well below the mm. However, no dead zone is allowed in the readout if we want to keep high energy resolution. This is achievable by means of the rim concept. In each detector, around the mesh, an independent strip of about 100 µm is engraved in the same process of the mesh manufacturing. This strip “surrounds”the mesh and if powered independently with a voltage a bit higher than that of the mesh, the field lines of the drift region (and therefore, the drifted electrons) in the border of the surfaces of each detector are pushed towards the sensitive surface. In addition, the deformation of the drift lines is much less than the pixel size, so there is no consequences on the topology information. We can see a simulation done with COMSOL of this effect in Figure 3.10. Then border effects in the field lines and dead space between the detectors are avoided. Figure 3.10: Comsol simulation (2D-approximation) of the rim effect described in the text for the drift lines around the boundary between two microbulk detectors. In this simulation the rim (a 200 µm metallic strip placed at 200 µm from the active region of the Micromegas) was placed at -255 V while the mesh was with a voltage slightly lower (-250 V). The Micromegas anode is grounded and the drift field was of 100 V cm−1. It can be seen how the drift lines are gently pushed into the active area of the detectors, while if the rim is at 0 V would have fallen on the dead area between Micromegas. Dimensions are expressed in mm being the overall distortion below this scale. The four detectors in NEXT-MM are placed on top of a copper plate grounded. Four copper pieces, as 4 quarter rings, are used to screw the Micromegas to the plate using PEEK screws. It is done in the insulator part of the detector. In this way movements when placing inside the prototype or when pumping the detector are avoided. Moreover they help to unify the surface of the detectors. These copper pieces are electrically connected between them and connected to the rim line described before. Therefore, applying a voltage to the ring the field in the boundary of the detectors is unified. A picture of the plate with two microbulk detectors installed and one with the four is shown in Figure 3.9 (c) and (d)). 72 3.3. Operational tests with NEXT-MM prototype Figure 3.11: Design of the pixels distribution and routing for the pixelized anode of the Bulk Micromegas (a). In (b) detail of the routing at the level of one of the four connectors prepared for the signal readout. The routing of the microbulk Micromegas used and described in the text, is equivalent to one quarter of the bulk design. 3.3 Operational tests with NEXT-MM prototype Different operational tests were done before starting the acquisition phase in order to check the performance of the setup and the capability of it to reach the values of HP, vacuum and HV needed to operate with the detector in Xenon. The results are presented in the next subsections. 3.3.1 High pressure test The vessel is constructed to be able to hold up to 15 bar. A pressure test in pure Argon was done in order, not only to verify the specifications assured by the manufacturer, but also to test the sealing, the capability of the different feedthroughs and valves installed and the capability of the gas system described in Section 3.2.1.2 to hold up to more than 10 bar. The pressure of the system as well as the temperature were monitored for 11 days after placing 11 bar of pure Ar in the chamber. In Figure 3.12 the evolution of the pressure and the temperature during these days is shown; the small oscillations of the pressure are correlated with variations in the temperature so if the ratio P/T ×Tmean is plotted (Figure 3.12 (b)) the pressure is constant during all the tests, taking into account the uncertainties in the measurement of P and T. Moreover, the best linear fit of the recorded data limits the possible leak of the system to be <5.9×10−4mbar×l×s−1, value that is limited by the accuracy of the high pressure nanometer. 3.3.2 Vacuum and outgassing Another important parameter is the vacuum that the system can be pumped down before filling it with gas. A good vacuum assures the system has no leaks and allows to keep the purity of the gas as high as possible, even if a recirculation through filters is done. In addition, the outgassing of some of the impurities trapped in the inner materials can also degrade the purity of the gas, as mentioned in (2.2.2). Therefore, the vacuum reached could change after the installation of new elements inside the vessel. 73 3. NEXT-MM prototype: design and commissioning (a) (b) Figure 3.12: (a) Evolution of the pressure (P) and temperature (T) during a HP test done with the vessel fill with Ar at 11 bar. The variations of the pressure are related with environmental changes in the temperature. (b) Evolution of the ratio P/T normalized by the mean temperature along time. A variation of less than 1% is observed. However, the tendence to decrease shows a possible leak rate below 5.9×10−4mbar×l×s−1and is conditioned by the accuracy of the high pressure manometer. A way to accelerate this outgassing and reduce it before injecting gas inside the vessel is to do a heating process. This process is called bake-out and consists in heating up the system in vacuum and keep pumping it at a high temperature. During the heating the emanation process is increased and the impurities are released from the materials more quickly. After several hours the heaters are switched off but not the pumping. Usually the pumping continues at least until the temperature goes back to ambient temperature. The outgassing is calculated as the difference in pressure over time and volume (∆P×V olume/∆t) and is expressed in mbar l s−1. In order to heat the vessel, several resistors are attached to the outer wall of the vessel. They are surrounded by an insulator coat to keep the heat, we can see a picture of the device in Figure 3.13. The heater operation is centralized in a programmable control unit that allows to go up to 200 ◦C. This temperature is enough for our system but, at the end, is limited by the thermal resistance of the inner components, specially the Micromegas detector which is not convenient to heat up to more than 150 ◦C This process was first done with the vessel empty and was repeated after the installation of any new element in the vessel. The first elements to be placed were those with higher heating capacitance allowing to make the first bake-out cycle at higher temperatures. Before the first bake-out cycle, the empty vessel was pumped with all the metallic seals installed in order to obtain a reference value for the vacuum that could be reached. After 95 hours of pumping a pressure of 7.8×10−7mbar was obtained, which is in good agreement with the vacuum certification from the manufacturer. Results for the different outgassing rates after the installation of different elements are summarized in Table 3.2. The main parameters are: the time the system is heated and pumped simultaneously (heating time) and the temperature and the time the system is pumped after the heating system is switched off (pumping time). Even if the cycles are not comparable (since the 74 3.3. Operational tests with NEXT-MM prototype (a) (b) Figure 3.13: (a) Bake-out system: different resistors are connected to the vessel to warm it up. It is surrounded by an insulator coat. The temperature can be controlled with a programable unit that fix it (b). temperature and times are not equivalent) the obtained results show that the longer the bakeout time is for the common components (like the vessel or the field cage), the better outgassing rates are obtained, even if more new components have been installed. This phenomenon leads to think that regular bake-out cycles could be necessary to keep or even improve the outgassing rates reached in the prototype. The last bake-out cycle was done with the full setup installed that corresponds to the detector, the field cage with the resistors and the cirlex protection and all the necessary internal cables and the required feedthroughs described in 3.2.1.1. For the full system the outgassing rate obtained is 4.5×10−7mbar×l×s−1, a value that seems enough to assure the purity conditions of the gas, specially if the recirculation of the gas is done. Components Heating Heating Pumping P0Outgassing time(h) T( ◦C) time(h) (mbar) (mbar×l×s−1) Empty Vessel 140 180 26 7.4×10−74.7×10−7 Field Cage(F.C.) 100 155 17 3.8×10−73.4×10−7 F.C. + Resistors 113 150 6 9.8×10−85.9×10−8 F.C. + Resistors+ Bulk MMs 94 160 144 6.1×10−83.2×10−8 Full Setup 112 150 11 7.6×10−74.5×10−7 Table 3.2: Summary of the bake-out cycles carried out in NEXT-MM prototype. In the table are indicated the components in the vessel (see text for explanation of full setup), and the main variables as well as the obtained outgassing rate measured for each case. 3.3.3 High voltage tests Another important point is to check the application of HV in the drift region. The optimal drift field for the electrons at 10 bar is of the order of ∼1 kV cm−1, therefore, in order to reach this electric field, the voltage that should be applied to the cathode at 10 bar is 35 kV, taking into account the drift distance. A HV source is connected to the cathode through the HV cable and 75 3. NEXT-MM prototype: design and commissioning feedthrough described in Section 3.2.1.1. The possible problem is that discharges could appear in any point between the cathode, field cage, resistors and the vessel, if a good insulation is not performed. Some protections were already installed like the cirlex screen between the field cage and vessel (see Figure 3.3) or Kapton scotch to cover any sharp point and avoid such discharges. A break point is defined as the voltage when the current begins to be higher than expected (based on the total impedance of the field cage), which indicates that some derivation to mass, mainly to the vessel, is happening. To study the breakpoint in the system, the electric current passing through the field cage has been monitored while the voltage was increasing. This test has been carried out for different pressures, up to 8 bar, using pure Argon. Results are shown in Figure 3.14. In Figure 3.14 (left) the evolution of the current with respect to the applied voltage is plotted for each pressure and, in Figure 3.14 (right), the curve for the different break points obtained in pure Ar for different pressures is shown. The trend of the points show that at 12 bar, 35 kV can be reached in pure Ar. In the case of Xenon the break point is approximately 2 times better [173] than for Ar, which allow to think that these voltages could be reached at pressures higher that 4 −5 bar. Other elements where a voltage has to be supplied is in the Micromegas detector. For the Micromegas a standard SHV feedthrough is used that allows to supply up to 3.5 kV, being much more that the operational voltage for the MMs (usually between 100 −500 V, depending on the pressure and gas). Hence, the ramping up of the voltaje in the Micromegas does not present a problem. Figure 3.14: High voltage tests. In the left it is shown the current evolution for each voltage and each pressure and in the right it is shown the break point evolution for each pressure. It can be seen how a higher pressure the break point occurs at higher voltages. Also it fits with the theoretical line. All data were taken in pure Argon. 3.4 Main features of the detectors in Ar-isobutane: gain curves and first energy resolution estimations. Before installing the detectors in NEXT-MM, a full characterization is done in a smaller chamber specifically designed to develop fast tests of the Micromegas detectors. It has a lateral length of 18 cm and is made of stainless steel of 1 cm thickness. Inside the field cage, it is installed a field 76 3.4. Main features of the detectors in Ar-isobutane: gain curves and first energy resolution estimations. cage in the same way as in NEXT-MM but smaller (16 cm) and without the cirlex protection because smaller voltages are applied to the cathode. A picture of the detector and its field cage is shown in Figure 3.15 (a) and (b). There are two possible calibration positions for the sources centred in two of the detectors as it can be observed in Figure 3.15 (b). This chamber is not expected to work at HP because it is not leak tight; for this reason Xe is not used. The performance of the Micromegas detectors was studied in Ar-isobutane (Ar-iCH4) at 2%. This gas is a good point to start checking the detectors because the Micromegas are usually tested on it, and a lot of data exists to compare the behaviour of the detectors with the standard ones before installing in NEXT-MM. The gas was in a continuous flow of 5 l/h and at 1 bar; data taking starts after few hours of circulation to assure the purity of the gas inside the chamber. Particle Energy (keV) Intensity (%) X-Ray 3.15 2.64 X-Ray 21.99 29.5 X-Ray 22.16 55.7 X-Ray 24.91 4.76 X-Ray 24.94 9.2 X-Ray 25.45 2.30 γ88.04 3.61 Table 3.3: Summary of the emissions from the 109Cd. It is shown the nature of the emission, its energy and intensity. All the tests were done with a 109Cd calibration source placed in one of the designed positions. In Table 3.3 are summarized the main emissions from this source. The four microbulk to be installed in NEXT-MM were characterized two by two. They are labelled as MM1, MM2, MM3 and MM4. Unfortunately, MM3 behaved strange and no peaks were recorded. However, when installed in NEXT-MM it recovered. The main problem in this small chamber was the noise level for the detectors. An extra problem comes from the limitation in the drift voltage, sparks start at around 3000 V. The drift voltage was applied using a Spellman HV power supply and the mesh voltages using a CAEN N471A module. The signal amplified in the Micromegas is read from the mesh using a CANBERRA 2005 PA. The PA output is fed to a CANBERRA 2022 amplifier module with a shaping time of 8 µs. Subsequently it is fed to a multichannel analyser AMPTEK MCA-8000A that produces the pulse height distribution proportional to the signal amplitude. In this part of the work the data used are the signals from the mesh. The recorded spectrum with the MCA (in ASCII) is converted into a ROOT file. It is then analysed using a C++ macro. In this analysis, the 22.1 keV gamma peak is used to obtain the gain and energy resolution. The procedure is as follows. First, a Gaussian fit is done in the peak. Then a background is estimated fitting to a line. The final fit is done convoluting two Gaussian, one at 22.1 keV and another at 25 keV, with the linear background. In the second fit the parameters from the first Gaussian are used. To fit the 25 keV the amplitude is calculated in function of the amplitude at 22 keV by the relative intensity of the two gammas. The errors obtained with the analysis in the peak determination are less than 2% and in the FWHM of the peak less than 0.1%. 77 3. NEXT-MM prototype: design and commissioning During the characterization, the pixels are grounded connecting them to the acquisition explained in 3.2.1.3 and the pixels signals are not used for the characterization of the detectors but the data registered were used to obtain the 2D information of the events. MM pixel plane – Limande-FEC-FEM FEM to DCC to PC Preamplifier MMs plane (a) (b) Figure 3.15: Chamber specifically designed to test microbulk Micromegas at 1 bar before installing them in NEXT-MM. (a) Setup. The Micromegas detectors are placed perpendicular to the floor. They are fed with a CANBERRA pre-amplifier and the signal from the mesh is obtained from the bias output.(b) Field cage made by copper rings. The two designed calibration position in the cathode at 16 cm from the detectors can be seen. 3.4.1 Electron transparency First of all, the electron transmission curves were obtained. The procedure is to vary the drift voltage at a fixed mesh voltage to obtain the transparency of the mesh to primary ionization. In Figure 3.16 is shown the relative electron transmission as a function of the drift to amplification field ratio. Data were obtained in the plateau where the transmission is maximum. At higher drift voltages the electron transmission is expected to decrease because the drift field lines start to end in the mesh instead of passing through the holes. However, this region could not be reached because sparks started appearing at the drift early. The results are in concordance with previous results with a microbulk in Ar-isobutane at 2% [150]. 3.4.2 Absolute Gain The evolution of the peak position with the mesh voltage gives the gain curve. For each voltage in the mesh, the drift voltage was selected to be operating in the plateau and always with the same ratio of drift-to-amplification fields. In order to calculate the absolute gain from the position of the peak in the spectrum, it is necessary to take into account different factors from the electronic chain. Firstly, the amplification gain introduced by the pre-amplifier (FP A). It depends on the model used, in this case the CANBERRA 2005 PA was in the position where a factor of 9 mV/e−was applied (FP A). It can be interpreted as the conversion factor convert charge (e−) to mV. Then, the amplifier factor (FA) has also to be taken into account. The peak in the MCA is given in mV/channel. The number of charges created by the incident particle is calculated using the 78 Entries 5929 / ndf 2 χ 9.815 / 9 Constant 8.1± 301.1 Mean 2.326e+002± 2.778e+005 Sigma 162.8± 5998 ADC_charge 50 100 150 200 250 300 350 400 3 10× Number of events 0 50 100 150 200 250 300 Entries 5929 / ndf 2 χ 9.815 / 9 Constant 8.1± 301.1 Mean 2.326e+002± 2.778e+005 Sigma 162.8± 5998 Figure 3.25: Distribution of the charge recorded in pixels, in acquisition units, in a microbulk detector installed in NEXT-MM. It was obtained in Ar-iCH4at 2% at 1 bar and with an 241Am source. The peak corresponds to the alpha emission of the source at 5.5 MeV. The distribution at the left is due probably to a bad recollection of the charge because the connectivity was low. The distribution at higher energies is due to pile-up. Time bin 0 100 200 300 400 500 ADC amplitude 0 500 1000 1500 2000 2500 3000 Pos X (number of pixel) -20 -15 -10 -5 0 Pos Y (number of pixel) -20 -15 -10 -5 0 0 500 1000 1500 2000 2500 Time bin 0 100 200 300 400 500 ADC amplitude 0 500 1000 1500 2000 2500 Pos X (number of pixel) -20 -15 -10 -5 0 Pos Y (number of pixel) -20 -15 -10 -5 0 0 500 1000 1500 2000 2500 Figure 3.26: Example of the tracks obtained looking at the events in the peak shown in 3.25. Left images: Registered pulses for each active pixel and right: XY projection for the activated pixels. 85 Time bin 0 100 200 300 400 500 ADC_Charge 0 500 1000 1500 2000 2500 3000 3500 triggered channels physchannel 513 col 17 row 3 physchannel 510 col 17 row 2 physchannel 512 col 17 row 4 physchannel 513 col 17 row 3 physchannel 530 col 15 row 4 physchannel 531 col 15 row 3 physchannel 532 col 15 row 6 physchannel 533 col 15 row 5 physchannel 553 col 13 row 2 Pos X (number of pixel) 0 2 4 6 8 10 12 14 16 18 20 Pos Y (number of pixel) 0 2 4 6 8 10 12 14 16 18 20 0 500 1000 1500 2000 2500 3000 Figure 3.27: Example of the tracks obtained looking at the events at higher energies than the peak showed in 3.25. There exists pile-up of two alphas of 5.5 MeV. Number of pixels 0 2 4 6 8 10 12 14 16 0 500 1000 1500 2000 2500 3000 3500 Figure 3.28: Distribution of the number of pixels for the run with a microbulk detector installed in NEXT-MM. It was obtained in Ar-iCH4at 2% at 1 bar and with an 241Am. The expected length of an alpha of 5.5 MeV in Ar at 1 bar is 4.5 cm in accordance with the obtained distribution peaked at 5 pixels (each one of 0.8×0.8cm2). 86 Chapter 4 NEXT-MM: First data with Xe-TMA Contents 4.1 Analysis code ................................ 88 4.2 First tracks ................................. 89 4.2.1 Energy resolution results ......................... 95 4.3 Tracking with more than one detector . . . . . . . . . . . . . . . . . 96 4.4 Trigger from an external source. . . . . . . . . . . . . . . . . . . . . 102 4.4.1 Energy resolution ..............................105 4.4.2 Drift velocity ................................117 4.4.3 Preliminary attachment results ......................117 4.5 Summary .................................. 119 In the previous chapter, the NEXT1-MM prototype was described and the operational tests and the first results to commission the detectors were presented. Energy resolution is a crucial point in a neutrinoless double beta decay experiment. Nevertheless, is the background rejection power that can be improved with a good pattern recognition. In this chapter first results will be presented regarding the tracking of the detector, composed by four different microbulk, independently read. Data was taken in a Xe-TMA mixture. Xe-TMA has a low diffusion coefficient, as discussed, making the topological reconstruction clearer. In addition, previous tests [109] with this gas in a smaller prototype have shown promising results up to 10 bar; it is interesting to corroborate this behaviour at longer drift distances. Moreover, first results regarding energy resolution have been also obtained. The structure of the chapter is as follows. In the first section a description of the analysis code is done. In the next sections, results with different setups and conditions are presented. First tracks and energy resolution studies are delivered. Next, first results about attachment and drift velocity are also obtained. Finally, some remarks for future work are discussed. 87 4. NEXT-MM: First data with Xe-TMA 4.1 Analysis code The code used to analyse the T2K acquisition files is composed by different methods and libraries based on C++ and ROOT called T2K electronics library. In the following the main elements are described. This code have been developed in [174], where a more detailed description is given and also it is shown its application to the CAST [121] detectors is presented. This code can be used with any pixelized detector, readout with the AFTER chip. It is based on the recording of pulses coming from different channels, as result of the detection of the interaction of a particle in the sensitive volume. In Figure 4.1 a flux diagram of the different methods and classes is shown. The output file of the T2K DAQ is converted into a ROOT file. The amplitude pulses of each of the activated pixels in an event are stored in ROOT histograms. Also each pulse has asociated an electronic channel. A program called acq2root manages this conversion through the manager. To do this conversion different classes are used. T2KReader directly leads with the binary output files of the T2K DAQ, as the set-up distinguishes between the electronics configuration and the detectors architecture, the auxiliary class T2KConfig allows to manage this through a configuration file where the number of cards or channels are indicated. Also the type of readout is mentioned. The events are stored as a T2KrawEvent; it is a low level general holder class. It stores the information in a ROOT TObject. It reconstructs the pulses for each electronic channel in an histogram and keeps the event time. It also stores the electronic channel number and two coordinates to place the pulse in the 2D readout. An associated class, ReadoutDecoding, implements the relationship between the electronic channel and the physical position for a particular detector and cabling configuration. Different decoding maps are implemented as those for NEXT-MM (with the different cabling used), the one for the test chamber and also for CAST. New decodings can be implemented if necessary.At this low-level a 2D visualization of the events also exist and can be used as an off-line browser. T2KTools is a set of functions to handle T2KrawEvents, using if necessary the readout configuration. Some of its functions allow to obtain the maximum amplitude of the pulses or are drawing functions. In general they are the basis for the analysis programs to be written. Then, the events are converted into a TRestPhysEvent; it is a description of the event in terms of its energy and position. It is a pixelized event. The zcoordinate is obtained converting the time information into length units using the drift velocity of the gas. To obtain the charge of the event an integration of all the channels that pass a threshold is done in a method called ReconstructEventAsPhys implemented in the class Tt2kManager. A calibration in energy must be done later correcting electronic effects. This class also has a 3D drawing method. Different variables are stored, like the number of pixels, for the analysis. Once the events are reconstructed in a TRestPhysEvent, also in a ROOT Tree, different analysis programs can be written that access easily the event information. SpectrumByFec is a program to develop 1D selection criteria on the observables (number of pixels, energy and position) and can separate them in function of the Micromegas that is read (in this case we are reading up to 4 different FEC, one per Micromegas). SelectTracks allows to select events that fulfil different conditions and to visualize and save their 2D tracks. 88 4.2. First tracks The purpose of the Tt2kManager class is also to act as a bridge-class to RESTSoft package. This package will be explained in Chapter 5and contains the main libraries to handle the simulation output from GEANT4 and implement the discrimination algorithms to do the analysis. T2K acq file acq2root T2KRaw Event ROOT file TRestPhys Event raw2phys ROOT file •Amplitude Integrated •Position x,y,z •3D map •Pulse amplitude in histograms •2D Map T2KReader Readout Decoding T2KTools T2kManager Online Browser SpectrumByFEC SelectTracks 1D Cuts REST Analysis Analysis Offline Browser Reconstruct EventAsPhys Figure 4.1: Flux diagram of the methods and classes for the T2K electronics library. The acquisition file is converted into a ROOT file transforming each event in a T2KRaw event that consists in a collection of histograms, one per each pixel activated, that represents its pulse amplitude. It is done with the program acq2root and uses the libraries T2KReader and Readout decoding to associate a position xy to the electronic channel. It also uses the methods contained in the class T2KTools. In a second step the event is converted into a TRestPhys Event for which the charge is obtained integrating the amplitude of all the binning of the histograms over a threshold. At this stage also the 3D positions of the events are calculated. Finally, different analysis programs are used that allow to select events that fulfil several conditions. 4.2 First tracks The first studies were done at 1 bar of Xe-TMA (3.5%) with two calibration sources inside the chamber. One of the sources was a 241Am source placed in the centre of the cathode. The other was a 57Co placed in one lateral wall of the field cage, between two rings, at around 8 cm from the Micromegas plane. A sketch of the setup is shown in Figure 4.2. With two different sources, having emissions of multiple energies, high phenomenology in terms of different track-lenghts is expected. On the other hand, the placement of a source close to the detector minimizes the risk of detecting no events due to possible problems with the gas purity. Table 4.1 summarizes the expected emissions from the two different sources with the energy, intensity, origin and estimated range at 1 bar of Xe based on the CSDA [175] calculus for each one. The 241Am source was placed shielding the alpha emission (that is used for trigger with a Silicon diode as will be discussed in detail in Section 4.3) and only the gammas and X-Rays will reach the drift region. 89 4. NEXT-MM: First data with Xe-TMA Origin Particle Energy (keV) Intensity (%) Range CSDA at 1 bar (cm) 241Am X-Ray 13.9 9.6 0.2 X-Ray 16.8 2.5 0.2 X-Ray 17.0 1.5 0.2 X-Ray 17.7 5.7 0.3 X-Ray 17.9 1.4 0.3 X-Ray 20.8 1.4 0.4 γ26.4 2.4 0.6 γ59.5 35 2.0 57Co γ14.5 10 0.2 γ122.1 85 8.0 γ136.5 10 8.0 Table 4.1: Summary of the emissions from the two sources placed inside NEXT1-MM. The nature of the emission is shown, its energy and intensity as well as the expected CSDA range of the electrons in Xe-TMA at 1 bar. 8 cm 38 cm Micromegas Plane 241Am 57Co Figure 4.2: Sketch of the first setup in which the calibration sources and their positions are shown. In addition, the escape peaks in Xe have to be taken into account. In Table 4.2 are summarized the Xe X-Rays energies. The incident particle can lose energy when exciting a Xe atom, losing the binding energy. Therefore, its final energy deposited will be less than the initial one. As shown in Table 4.2 the Kalpha shells are very close and are difficult to distinguish. Therefore, we can say that the two X-Ray can be detected with 4 keV of difference if the energy resolution of the detectors were good enough. As there are events with 60 and 122 keV coming from the sources, two escape peaks are expected: one at around 90 keV from the higher energy gammas of the 57Co source and another at 30 keV of the 60 keV gamma from the 241Am source or from the Xe characteristic X-Ray. The data here commented corresponds to just one of the four sectors of the readout, MM2, that is the one closest to the 57Co source. The mesh was at 275 V (55 kV cm−1amplification field) and the drift at 5900 V (drift field of 155 V cm−1). The signal from the mesh and from the pixels were recorded. The mesh signal was used to trigger the pixels acquisition. In Figure 4.3 is shown the recorded raw spectrum of the pixels after around 8 h of measurement. As mentioned, the charge has been obtained integrating the pulses that have an amplitude higher than a threshold 90 4.2. First tracks Shell Energy (keV) Kα129.4 Kα229.7 Kβ33.6 Lα14.1 Table 4.2: Summary of the Xe characteristic X-ray energies. imposed through the configuration file and which is based on the pedestals. It is calculated in ADC charge per µs. Along this data-taking the time window was 80 µs sampled in 512 time bins. Up to now, the only selection applied is to reject those events that saturate the electronics. 241Am (Kα,β scape) ~ 30 keV 241Am (γ) 60 keV 57Co (Kα,β scape) ~ 90 keV 57Co (γ) 122 & 132 keV 57Co (γ) 14 keV Figure 4.3: Raw spectrum reading the pixels of the MM2 microbulk detector obtained with two calibration sources, 57Co and 241Am, placed in NEXT1-MM (see text for more details). Although no energy calibration is available for these data, the assignment of the energy of the peaks is straightforward because of their position, since they correspond to well separated known photon energies, and is given in Figure 4.3. However, we can consider another characteristic in order to increase our confidence: the length of their corresponding tracks. First of all is plotted the distribution xy of each peak, as is shown in Figure 4.4. It can be observed how the population in the 14 keV peak is closer to the position of the 57Co source. The tracks coming from 241Am and the more energetic ones are contained in the centre of the readout. This leads to think that selecting a smaller fiducial region, the peaks could be observed clearer because we are avoiding boundary effects that could produce that some charge was detected in another Micromegas or would fall outside the sensitive region. This is discussed in more detail in the next section. The range for each energy is shown in Table 4.1 and in Figures 4.5 and 4.6 different examples of tracks selecting events in each peak are shown. It can be seen how the length is in agreement with the expected CSDA range. It has to be reminded that the pixel size in these Micromegas is 0.8×0.8 cm2. 91 4. NEXT-MM: First data with Xe-TMA 14 keV 30 keV 60 keV 90 keV 120-140 keV >140 keV 57Co Figure 4.4: xy distributions of the triggered pixels for each of the observed peaks and higher energy regions. 92 4.2. First tracks Time bin 0 100 200 300 400 500 ADC_Amplitude 0 100 200 300 400 500 600 700 800 (a) Pos X (number of pixel) -20 -15 -10 -5 0 Pos Y (number of pixel) -20 -15 -10 -5 0 0 100 200 300 400 500 (b) Time bin 0 100 200 300 400 500 ADC_amplitude 0 500 1000 1500 2000 2500 3000 (c) Pos X (number of pixel) -20 -15 -10 -5 0 Pos Y (number of pixel) -20 -15 -10 -5 0 0 500 1000 1500 2000 2500 (d) Time bin 0 100 200 300 400 500 ADC_amplitude 0 500 1000 1500 2000 2500 (e) Pos X (number of pixel) -20 -15 -10 -5 0 Pos Y (number of pixel) -20 -15 -10 -5 0 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 (f) Figure 4.5: Example of events recorded in the pixel plane on the Micromegas MM2 for the 57Co source. (a-b) 14 keV peak, (c-d) 90 keV peak and (e-f) 122 keV peak. Left images represent the registered amplitude for the active pixels and right plots are the xy projection of the events in the pixelized plane. 93 4. NEXT-MM: First data with Xe-TMA Time bin 0 100 200 300 400 500 ADC_amplitute 0 200 400 600 800 1000 1200 1400 1600 1800 2000 (a) Pos X (number of pixel) -20 -15 -10 -5 0 Pos Y (number of pixel) -20 -15 -10 -5 0 0 200 400 600 800 1000 1200 1400 1600 (b) time bin 0 100 200 300 400 500 ADC_amplitude 0 500 1000 1500 2000 2500 3000 3500 4000 (c) Pos X (number of pixel) -20 -15 -10 -5 0 Pos Y (number of pixel) -20 -15 -10 -5 0 0 500 1000 1500 2000 2500 3000 3500 (d) Figure 4.6: Example of events recorded in the pixel plane on the Micromegas MM2 for the 241Am source. (a-b) 30 keV peak and (c-d) 60 keV peak. Left images represent the registered amplitude for the active pixels and right plots are the xy projection of the events in the pixelized plane. 94 4.3. Tracking with more than one detector Time bin (160 ns/bin) 0 100 200 300 400 500 ADC_amplitude 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 (a) Pos X (number of pixel) -20 -15 -10 -5 0 5 10 15 20 Pos Y (number of pixel) -20 -15 -10 -5 0 5 10 15 20 0 200 400 600 800 1000 1200 1400 1600 1800 (b) Pos X (number of pixel) -20 -15 -10 -5 0 Pos Y (number of pixel) -20 -15 -10 -5 0 5 10 15 20 Time bin (160 ns/bin) 100 150 200 250 300 350 400 (c) Figure 4.14: Detection of two 30 keV events. (a) Pulses in the activated pixels. (b) xy projection of the track. (c) 3D reconstruction with a relative zposition. 101 4. NEXT-MM: First data with Xe-TMA 4.4 Trigger from an external source. A Silicon diode (Si) was installed in the cathode, as shown in Figure 4.15, in such a way so as information about the total zposition of the event in the TPC can be extracted. The Si was enclosed in a Teflon tube to avoid sparks with the cathode at HV. It was also closed with a cylindrical piece of Teflon that can hold the 241Am source. Then, the source was placed between the Si and the lower part of the Teflon. The Teflon tube was positioned in the cathode, in the central hole. The 241Am is an alpha emitter with an energy of 5.5 MeV. The source is deposited on a thin layer of aluminium. Then, the alpha emission escapes just from the source side pointing to the Si diode, while gammas and X-rays are distributed on both sides. Therefore, the Si detector is triggered with the alpha emissions. In 36% of the times this alpha emission is produced with a gamma of 60 keV. This emission can occur in all directions and, approximately, the half of the times will enter the drift region and therefore, may be detected in the Micromegas plane. A schematic view of the process can be seen in Figure 4.16. If the trigger is done with the signal from the Si, information about the origin of the events can be extracted from the data. In this run the 57Co source was taken out because the rate of random coincidences was of the order of the expected coincidence with the Si making more difficult this study. Data were recorded making a coincidence between the Si and the mesh signal. (a) (b) Figure 4.15: (a) Image of the Si diode installed in NEXT1-MM and used to trigger the T2K acquisition. (b) Placement of the Si in the cathode. It is covered by Teflon to avoid sparks with the metal surfaces. In the bottom part of the coverage is placed the radioactive source (241Am). In order to acquire in coincidence with the Si, different electronic modules and delays are needed. The electronic chain used is sketched in Figure 4.17. First of all, the Si is powered at 50 V with an ORTEC 142C preamplifier by the bias line while the signal is read. Then it is passed to a Linear Scaling Amplifier (LA/SCA) and converted into a logical signal. In the case of the Micromegas they are fed with a CANBERRA 2004 preamplifier. The operational voltages for these tests were 270 V. Then, the signals are passed to a linear amplifier. The output is sent to a CAEN N844 Low Level Discriminator (LLD). In order to make an OR between the mesh signals, the CAEN N455 Quad Coincidence Logic Unit is used. Both, the logical Si signal from the LA, and the output of the OR from the Micromegas are fed to a Dual Timer model CAEN N93B. The Si signal pulse width is stretched by a time window equivalent to at least all the drift region. Using the end-marker of the pulse results in delaying the Si pulse. The Micromegas pulses are 102 4.4. Trigger from an external source. α 5.5 MeV ϒ (36%) 60 keV Si diode Cathode (HV) E 241Am Figure 4.16: Drawing of the setup with the Si diode. An 241Am source is placed in the cathode. It emits an alpha particle of 5.5 MeV that can only escape to the upper part of the chamber where it is detected in the Si detector. This emission, the 36% of the times takes place with a gamma of 60 keV. As it can be emitted isotropically in all directions approximately half of the times it will enter the drift region and will be detected in the Micromegas plane. widened more than the Si to ensure that we are detecting all the relevant events. Finally, the end point from the Si and the Micromegas signal are passed again to the coincidence module and the AND output is used to trigger the pixels acquisition. The data presented were recorded using two Micromegas: MM1 and MM4. They are covering half of the surface of the anode. Data were recorded in recirculation mode with a rate of 0.8 - 1 Hz. The working pressure was 1.03 bar in a Xe-TMA mixture at 3.5%. In Figure 4.18 (b) can be observed the xy projection of all the events detected in the pixel plane of the detectors. In Figure 4.18 (a) it is shown the charge spectrum obtained in each detector, integrating the charge for all the channels above an amplitude threshold. In order to generate the spectra for each detector, the events considered have all their active pixels in the same detector. Two peaks are expected in the spectrum: one at 60 keV from the gamma detection and another at around 30 keV from the escape peak in Xe, as commented in previous section. The number of detected events is approximately the same in both detectors while the gain is a factor 1.12 higher in MM4 than in MM1. From previous measurements a difference up to two was observed between the two detectors. The observed distribution has a higher 30 keV escape X-Ray than the 60 keV. This is corroborated from what is obtained in a simulation of the setup performed using GEANT4 in [176] with just one detector active. 75% of the times the events will come accompanied by a K-shell emmision [124] and the mean free path of the 30 keV energy is 20 cm. Therefore, some 30 keV events produced by the 60 keV that would fall in other detector would be measured increasing its probability with respect to the 60 keV. In Figure 4.19 (a) the obtained spectrum from the simulation can be seen. In Figure 4.20 can be seen the time distribution in each Micromegas. The position of the trigger in the time window trigger was at 230 µs; as we are using it to trigger the acquisition, and it will detect the alpha while the gamma has to go across all the drift distance to be detected on the Micromegas plane, all the relevant information is pretrigger. In order to detect events in all 103 4. NEXT-MM: First data with Xe-TMA Si PA LA/SCA DT 200 μs AND End Point MM4 PA LA LLD -20 mV MM1 PA LA LLD -30 mV OR T2K DAQ Trigger DT 220 μs Output Figure 4.17: Sketch of the electronic chain used to trigger the Micromegas detectors in coincidence with the Si diode. The Si diode detects an alpha. Its signal then is the input in a Dual Timer that enlarge it and gives the end-point of the window. It has to be enlarged at least by the drift distance to be able to see all the chamber. The mesh signals, after a PA, LA are fed to a Low Level Discriminator (LLD). The two of them have more or less the same S/N threshold that is imposed to be around -25 mV. Then, this logical signals are fed to a dual timer which output is an OR between the two mesh. Finally an AND between the output of the OR and the end-point of the Si is used to trigger the acquisition. the drift region the Si diode signal was delayed by 190 µs. This delay was obtained taking into account the expected drift velocity of Xe-TMA in a drift field of 150 V cm−1from the Magboltz simulation shown in Figure 2.2. It is enlarged to be sure that events in all the drift region are detected. Therefore, the 0 position in the chamber (the Micromegas plane) has to be at 40 µs. The cathode could be assigned to the point in which the distribution falls down to the flat level again. This distribution has also be confirmed from the simulation of the setup as is shown in Figure 4.19 (b). It is not completely exponential, as expected for the mean free path of the particles. It fits to an exponential plus a constant line because there exist two contributions to take into account: on one hand the mean free path of the 60 keV gamma and the 30 keV that is around 20 cm, hence there is more probability that they interact in the upper part of the TPC, and on the other hand, the solid angle is much bigger close to the cathode because the source is not collimated. 104 4.4. Trigger from an external source. ADC_charge 0 5 10 15 20 25 30 35 6 10× Number of events 50 100 150 200 250 300 350 400 450 MM1 raw spectrum MM4 raw spectrum (a) Position X (number of pixel) -150 -100 -50 0 50 100 150 Position Y (number of pixel) -160 -140 -120 -100 -80 -60 -40 -20 0 (b) Figure 4.18: (a) Raw charge spectra obtained for the active detectors (MM1 in blue and MM4 in black) in 1 bar of Xe-TMA (3.5%). The 60 keV gamma peak from the 241Am source can be seen and, also, the 30 keV from the scape peak of Xe. A difference in gain of 1.2between the two detector exists. (b) xy projection of the total sum of charge of the active pixels. Event energy (keV) 0 10 20 30 40 50 60 70 80 90 100 Number of events 0 1000 2000 3000 4000 5000 (a) Z mean position (mm) 0 50 100 150 200 250 300 350 400 Number of events 0 1000 2000 3000 4000 5000 6000 7000 (b) Figure 4.19: Results from a simulation performed with GEANT4 of the setup, extracted from [176]. In (a) is represented the energy spectrum showing that the 30 keV particles has a higher probability than the 60 keV gamma because just one detector is active. In (b) the zposition distribution is plotted showing the same shape that the obtained with the data in Figure 4.20. 4.4.1 Energy resolution In Figure 4.18 (a) the background from cosmic muons, alphas and electrons from the natural isotopes can also be seen. Different selection criteria have been defined at this level to obtain a cleaner spectrum and therefore, a better energy resolution. They can also be useful to understand background. From the topological information of the events, differences between the events from the source and background are observed, as for example the number of triggered pixels. As seen in previous sections, events of 30 keV and 60 keV have a length of less than 10 pixels. The different selections defined are applied sequentially. The first one is to reject those events on which at least one pixel has an amplitude that saturates the electronics (as applied in Section 4.2). With this selection, not only noise events are rejected, but also alpha events (see Figure 4.10). 105 4. NEXT-MM: First data with Xe-TMA T2K Trigger Zmax cathode (38 cm) Micromegas plane (0cm) Si delay 190 µs Random coincidences Figure 4.20: zdistribution in the two active detectors. Eventually some 30 keV events would also been rejected because they can deposit all the energy in just one pixel and saturate it, but the probability is around 1%. It does not happen with the 60 keV events since they spread on, at least 2 pixels. Secondly, the observable ∆zis defined as the difference between the maximum and minimum z position in an event. The expected length from the 60 keV gamma is of the order of 10 cm. If the event is completely vertical that would give the maximum ∆z. On the other hand, vertical muons, that would be difficult to differentiate because they will leave all their energy in few pixels, will have a longer ∆z, of the order of all the drift length. Also transversal muons and electrons will have a longer ∆zthan events coming from the source. In Figure 4.21(a) is shown the distribution of ∆zfor both detectors, MM1 and MM4, and the distribution of ∆zversus the energy in Figure 4.21(b). For both Micromegas the same distribution is obtained. In 4.21(b) can be observed the distribution of the 30 keV and 60 keV. The events populating the region with a high charge and a low ∆zcould be alphas or electrons. At higher ∆zthere are electrons and muons. A selection for the events with a ∆zsmaller than 10 cm (80µs) is done. To consider only real coincidences and to avoid random ones, as we have a longer acquisition window, the net selection is to consider only events in the drift region. As explained previously the drift region goes from z= 40 µs to 150 µs. However, there are still random coincidences from the background that populate mainly the 30 keV peak as can be seen in Figure 4.22 where it is shown the spectrum obtained when selecting events with a zbetween 0 and 30 µs (out of the Si diode window). The effect of the random coincidences is bigger at lower z’s where the probability of interaction is smaller. The next step is to perform a selection in the number of pixels. In Figure 4.23 is shown the distribution of the number of pixels selecting events in each peak. A selection of events with a pixel number between 1 and 12 is performed. As mentioned, vertical muons also touch few pixel but they are discriminated with the ∆zvariable. In Figure 4.22 (b) and (c) are shown 106 4.4. Trigger from an external source. the resulting energy spectrum and zdistribution selecting events with more than 12 pixels, that justify this selection criteria. The resulting spectra after each cut are shown in Figure 4.24 (a) for MM1 and (b) for MM4. In Tables 4.4 and 4.5 are shown the obtained efficiency and energy resolution fitting each peak over a flat background. The fit to obtain the energy resolution is made over a flat background, therefore, the energy resolution only changes slightly with them. The values obtained in MM1 extrapolates to an energy resolution of 2% FWHM at 2458 keV and less than 2% FWHM in MM4, proven, already an improvement with results shown in Section 4.2.1. The last step is to define an xy region to avoid boundary effects. Different regions are studied as in Section 4.2. Mainly two of them are evaluated in this subsection, shown in Figure 4.25. For each detector the most activated region has been chosen. The xand yposition of each event is calculated as the mean xand yposition of all the pixels. In a first step the selection is done using that the mean position has to be in the selected region. s)µz (∆ 0 10 20 30 40 50 60 70 80 Number of events 0 500 1000 1500 2000 2500 3000 3500 4000 (a) s)µz (∆ 0 10 20 30 40 50 60 70 80 ADC_charge 0 10 20 30 40 50 60 70 80 6 10× (b) s)µz (∆ 0 5 10 15 20 25 ADC_charge 0 10 20 30 40 50 60 6 10× (c) Figure 4.21: (a) Distribution of the ∆zvariable defined as the difference between the maximum and minimum zposition of the event for the two detectors: MM1 in black, MM4 in blue. (b) ∆zversus energy for the MM1 (same distribution obtained for MM4). In (c) there is a zoom of (b) showing clearr two distributions corresponding to the two detected particles (∼30 keV and 60 keV). ADC_charge 0 10 20 30 40 50 6 10× Number of events 0 5 10 15 20 25 30 35 40 (a) ADC_charge 0 10 20 30 40 50 60 70 80 6 10× Number of events (normalized) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 (b) s)µz position ( 50 100 150 200 250 Number of events 0 5 10 15 20 25 (c) Figure 4.22: (a) Charge spectrum obtained for both detectors in the region of random coincidences (z < 30 µs). (b) Energy spectrum obtained selecting long events (with more than 12 pixels). It can be seen how the structure of peaks disappeared dominated by the cosmic background. In (c) the zdistribution is flat when selecting events with more than 12 pixels. In the three plots the black-line corresponds to MM1 and the blue-line to MM4. The spectrum in Figure 4.26 is the resulting one after all the mentioned criteria for MM1 and in Figure 4.27 for the MM4 considering the more restrictive xy region. The fit is done in two 107 4. NEXT-MM: First data with Xe-TMA Number of pixels 0 2 4 6 8 10 12 14 16 18 Number of events 0 200 400 600 800 1000 1200 (a) Number of pixels 0 5 10 15 20 25 30 Number of events 0 50 100 150 200 250 300 (b) Figure 4.23: Distribution of number of pixels for each peak: (a) selecting events in the 30 keV peak and (b) in the 60 keV peak. Events with less than 12 pixels were kept in the analysis. ADC_charge 0 5 10 15 20 25 6 10× Number of events (normalized) 0 1 2 3 (a) ADC_charge 0 5 10 15 20 25 30 6 10× Number of events (normalized) 1 2 3 4 (b) Figure 4.24: Effect of the discrimination criteria on the charge spectrum for MM1 (a) and MM4 (b). The spectra are shown normalized and the criteria applied until now are the same for both Micromegas. They are: blue-line raw spectra, black-line: rejecting saturated pixels, red-line: applying δz < 80 µs, magenta-line: selecting events in coincidence with the Si (∆z between 40-150 µs, and finally, selecting events with less than 12 pixels (soft green line). Selection 30 keV peak 60 keV peak Surviving events (%) %FWHM Surviving events (%) %FWHM Raw spectrum 18.13 ±0.62 13.16 ±1.04 Rejecting pixel 100 18.13 ±0.62 90 12.98 ±1.03 that saturate ∆z < 80 µs 83 18.19 ±0.62 86 13.12 ±1.07 zselection (40-150 µs) 69 18.18 ±0.63 76 13.79 ±1.22 Number of pixels (1-12) 67 18.20 ±0.63 73 13.79 ±1.17 Table 4.4: Efficiency and energy resolution obtained applying sequentially the different selections explained in the text for the MM1. steps: one first Gaussian in each peak to determine the peak position and to fix the range for the fit at ±2.5σ. The second Gaussian is defined over a flat background left as a free parameter, hence, it is different in each case. In the case of the MM1 (Figure 4.26) the scaling of the energy 108 4.4. Trigger from an external source. Selection 30 keV peak 60 keV peak Surviving events (%) %FWHM Surviving events (%) %FWHM Raw spectrum 13.51 ±0.52 11.99 ±1.42 Rejecting pixel 100 13.51 ±0.52 84 11.75 ±1.53 that saturate ∆z < 80 µs 99 13.56 ±0.53 82 11.4±1.44 zselection (40-150 µs) 86 13.44 ±0.55 69 13.31 ±2.57 Number of pixels (1-12) 75 13.44 ±0.55 66 13.11 ±2.39 Table 4.5: Efficiency and energy resolution obtained applying sequentially the different selections explained in the text for the MM4. 1 2 (a) 2 1 (b) Figure 4.25: Distribution of the charge of the pixels in both Micromegas under study (a) MM4 and (b) MM1. The different selected regions are shown in black rectangles, labelled as 1 and 2 as along the text. resolution is correct. In the case of the MM4 (Figure 4.27) it is not and the relative error is quite large, for the 60 keV peak. It might be because in this detector there were more dead pixels and the statistics was lower. Entries 2405 ADC_charge 0 5 10 15 20 25 30 35 6 10× Number of events 0 20 40 60 80 100 Entries 2405 MM1 Fit at 29 keV 0.78±Res.(%FWHM) = 12.18 MM1 Fit at 60 keV 1.49±Res.(%FWHM) = 9.25 Figure 4.26: Energy spectrum obtained in MM1 of the 16% final selected events in region 1 and that contains the 16% of the events. The fit is done at two stages over a flat background. In order to completely reduce the background events fully contained in a xy region were selected calculating the minimum and maximum xand yposition of an event. The same two regions than 109 4. NEXT-MM: First data with Xe-TMA Entries 2765 ADC_charge 0 5 10 15 20 25 30 35 6 10× Number of events 0 20 40 60 80 100 Entries 2765 MM4 Fit at 29 keV 0.89±Res.(%FWHM) = 11.91 MM4 Fit at 60 keV 5.02±Res.(%FWHM) = 14.31 Figure 4.27: Energy spectrum obtained in MM4 for the 13% of teh events passing all the selection criteria in region 1. The fit at 60 keV has a large error. The fit in each peak is done in tww steps over a flat background. before have been considered. The resulting spectra are shown in Figures 4.28 and 4.29 for MM1 and MM4 respectively and in the two regions considered. It must be noted that in the called 30 keV peak there is expected to have the contribution of three peaks: one centred at 29 keV from the scape peak of 29.7 and 29.4 keV from Xe and another at 26.4 keV from the 33.6 escape peak. Also at 26 keV is expected a gamma emission of the 241Am with an intensity of 2.4%. In order to perform a two Gaussian fit it is interesting to know the relation between the two peaks. From the simulation shown in spectrum 4.19 we can conclude that in this geometry and with this gas at these conditions the ratio is around 0.3. The fit in the 30 keV region is done in two steps: first of all a Gaussian is defined in a wide range around the peak. With its mean position is defined the range of the second fit (between ±2.5−3σ). Secondly, a sum of two Gaussian is defined in this range, one with free parameters for the 29 keV and the second one (fit to the 26 keV) using the sigma of 29keV peak, a relative position of 0.89 (4 keV of difference) and we leave free the amplitude but controlling that the value is around the 0.3 ratio obtained from simulations. This sum of Gaussian is plotted in red (having a χ2value above 1). Finally, the separated Gaussian fits are in magenta and blue. The 60 keV has also been fitted in a similar way. As can bee seen in the smaller region for the MM4 this peak almost disappears. This could be because in region 1 around the 15% of the pixels are dead. This effect is more important in the 60 keV peak because the tracks are longer than the ones of 26-29 keV. Therefore, we will evaluate the energy resolution for the MM4 in the wider region 1. In Table 4.6 the results obtained for each Micromegas detector are shown. An extrapolation to the expected energy resolution at Qββ can be done using both peaks (at 30 keV and 60 keV). The relation of the energy and the resolution is a quadratic one defined as FWHM2=a∗E+b, (4.1) where aand bare free parameters that depends on the detector and electronics. At high energy 110 4.4. Trigger from an external source. 4.4.2 Drift velocity Xe-TMA is not a very common mixture but in the last years it has emerged because it can improve the gain and diffusion of the base gas [109]. Recently, the Magboltz program has implemented its cross-sections. In this section we will try to obtain a drift velocity from our measurements than can be compared with the results obtained with Magboltz. In Figure 4.20, as the trigger is done in coincidence with the detectors, it is possible to extract information of the drift time. Therefore, as the beginning of the chamber is at 40 µs and comparing with simulations, we have concluded that the cathode is where the distribution falls down to a flat value again, in this case at ∼150 µs, the drift time is 110 µs. Moreover, if only events in the 30 keV peak are selected, the resulted distribution in shown in Figure 4.35, it can be appreciated slightly the change at 40 µs. This drift distance corresponds to 38 cm when the Microbulk has been installed, therefore the velocity of the gas is around 0.35 cm µs−1. In Figure 2.2 are shown different results for the drift velocity in a Xe-TMA mixture with different proportions obtained with Magboltz. The data discussed here had a concentration of TMA around 3.5% and the drift field was 150 V cm−1at 1 bar, therefore, from the Magboltz results, a value of 0.25 cm µs−1is extrapolated for the velocity in this mixture, 0.1 cm µs−1smaller than the obtained value. However some facts have to be mentioned. First, in a previous measurement done in a smaller TPC the same behaviour [177] was observed: a small increase of the velocity at this drift field with respect to the values from Magboltz. In addition, it has to be noted that previous measurement recirculating through the SAES filter show that it can modify slightly the concentration of TMA along time, either absorbing TMA or expelling it again. Therefore, as we did not measure the concentration of TMA routinely (because of technical problems) it could change at some point. Also, there could be uncertainties in the determination of the cathode plane because the end point of the zdistribution is not so well define than in the simulations because, some effects are not implemented like the attachment or others related with the electronic chain (integration of the charge,...). However, they should not come a dominant effect. In addition to these effects, it has to be taken also into account that TMA cross-sections and diffusions have been recently implemented in Magboltz and with data only from simulations. Therefore, it could exist also some uncertainties. 4.4.3 Preliminary attachment results The attachment, as explained in Chapter 2, produces a loss in the charge detected due to electronegative impurities present in the gas. This charge loss implies a decrease in the gain and in the proportionality of the charge detected to the energy of the particle, therefore it can worsen the resolution. The effect of the attachment has also a dependence with the zposition being more important close to the cathode because the electron has to cross all the drift distance and the probability to be absorbed by some impurities is higher. If the zposition is represented versus the charge, a slightly negative slope can be observed at higher z’s as it is shown in Figure 4.36 (a) (b) for the Micromegas MM1 and MM4 respectively. 117 4. NEXT-MM: First data with Xe-TMA s)µz mean position ( 0 50 100 150 200 250 Number of events 0 10 20 30 40 50 60 70 80 (a) s)µz mean position ( 0 50 100 150 200 250 Number of events 0 20 40 60 80 100 (b) Figure 4.35: zdistribution of the events in the 30 keV peak in MM1 (a) and MM4 (b). In (a) the change in the distribution at 40 µs can be observed while in (b) it is less clear. In order to obtain the drift velocity the interval from 40to 150 µs was considered. In order to obtain an estimation of the attachment the electron mean-lifetime, τe, defined in 2.16 can be calculated. The expression 2.16 can be rewritten in terms of the amplitude or charge, A as A=A0e−t/τe.(4.2) If τeis large enough with respect to ta Taylor expansion can be done, and fit the exponential to a line, with slope equal to A0/τ.In order to estimate τea region centred in the peak is chosen, as marked in Figure 4.36, and the mean charge is calculated in zslices of 10 µs each one. In Figure 4.37 is shown the distribution of the mean charge obtained in each slice for the 30 keV peak in the regions defined in Figure 4.36. We decide to do the study only in the 30 keV peak because the energy is deposited in just one cluster of pixels with a better defined zposition. The zgoes from 40 to 150 µs but in order to determine A0in the plot there is an x-axis translation of x−40 µs. As can be seen in Figure 4.37 the mean charge distribution around the 30 keV region with respect to the mean zposition shows two different behaviours: a plateau at low zand a slope with negative slope at higher z. At lower z, close to the Micromegas plane, we are dominated by random coincidences, as is shown in Figure 4.35. Also from this Figure we can see that the signal detected in coincidence with the Si starts to be a 10% dominant over the random coincidences at 100 µs. Therefore a fit is done for the last points obtaining a mean electron lifetime of 1.08 ±0.12 ms in MM1 and 1.80 ±0.25 ms in MM4. For the NEXT-MM drift distance of 38 cm an electron lifetime several times the drift time (of 110 µs) is desired, and with the current setup it is achieved. However, in the case of a bigger TPC that can handle 100 kg, the drift distance will increase up to 1.4 m, hence τehas to be several time 428 µs if operating with this Xe-TMA mixture, and even more for the future step with 1 ton of mass the drift distance will doubled. All this is supposing a knowledge of the t0of the event. The requirements are 118 4.5. Summary (a) (b) Figure 4.36: 2D plot of the position zof the events versus their energy (in logarithmic scale) in MM1 (a) and MM4 (b). In the black rectangle is indicated the region used to calculate the attachment (see text for details). s)µz mean position ( 0 20 40 60 80 100 Number of events 7000 7200 7400 7600 7800 8000 8200 8400 8600 8800 9000 3 10× (a) s)µz mean position ( 0 20 40 60 80 100 Number of events 8000 8200 8400 8600 8800 9000 9200 9400 9600 9800 3 10× (b) Figure 4.37: Mean charge distribution calculated in different zslices in the regions marked in Figure 4.36 for the 30 keV peak in the different Micromegas (a) Result for MM1 and (b) for MM4. In both cases, at lower zwe are dominated by the effect of random coincidences and it is difficult to conclude something, However, from 60 µs, a decrease in the gain is observed that can be associated to attachment. more restrictivet if t0is not determined, demanding gain variations of the order of probably 1% along the drift distance. 4.5 Summary The NEXT-MM prototype is routinely taking data. Here, first setups and results in Xe-TMA at 1 bar have been discussed. One of the objectives of the prototype has been fulfilled, as it was to prove the operation of the detectors in a long drift distance TPC. First results regarding attachment and purity of the gas show the capability to work with this setup. Moreover, another 119 important point was the topological reconstruction of tracks. First tracks were obtained in a highphenomenology setup (with two gammas calibration sources) identifying the expected length for each peak in one detector. The next step was to take data with more than one Micromegas active. First electron tracks and muons crossing different detectors have been shown. In addition, first energy resolution estimations are promising for a neutrinoless double beta decay experiment, extrapolating to around 1.5% FWHM at Qββ with a first analysis. In addition, it was the first time that microbulk Micromegas of these dimensions (sectors with a radius of 15 cm covering a surface with a diameter of 30 cm) have been fabricated and tested. First results of the state-of-the art of them show the high quality of these detectors and are very promising. The slight differences found between them can be reasonably improved with the progress of the manufacturing process. In conclusion, it have been proven the capability to perform calorimetry and tracking using the pixels plane. The next steps are clearly focused in two directions. First, regarding the setup, the startedup of the RIM to improve the field lines from the drift field to the mesh, can make a clear improvement in recollection and, hence, in the energy resolution. Also, further steps are to go to higher pressures and to test the detectors in pure Xenon. Secondly, efforts have to be put to improve the analysis. One of these steps, is to reconstruct the events in different clusters or tracks and analyse them separately, in such a way as in the REST code does for the simulated event. We will discuss this code in detail in the next part of the work. Also important would be to correct the energy by the pixels gain, that can vary from one part to another (however, variations less than 1% are expected) and by the Micromegas gain. 120 Part III Physics potential of a ββ0ν experiment with Micromegas in a HP Xe TPC. 121 The use of a Xenon gaseous TPC as calorimeter for the neutrinoless double beta decay search of 136Xe has only recently been considered competitive, after the new developments with MPGD. The Gothard’s TPC [89] was pioneer on this approach showing the enhanced capabilities of background discrimination based on the topology of the events reconstructed in the TPC. This idea has been recently resuscitated by the NEXT (Neutrino Experiment with a Xenon TPC) collaboration [68]. The state-of-the-art Micromegas readouts has demonstrated its features for the ββ0νsearches as its excellent energy resolution [108,109] and their high radiopurity [156]. Micromegas detectors combined with a gaseous TPC can fulfil the requirements for an experiment in terms of background rejection and energy resolution. 136Xe is considered a good ββ0νisotope regarding their nuclear structure, that it is easily enriched, and its high Qββ. In this part of the work it is studied the different expected backgrounds in a high pressure TPC for the 136Xe ββ0νsearch. A full simulation has been done to generate the particles and their interactions inside the TPC. Also it has been simulated the physical processes that occurs in a TPC with a pixelized detector: ionization, diffusion (two different diffusion coefficients have been studied) and pixelization of the charge. Specially, the topological differences between background and signal events are studied and are used to develop discrimination algorithms that allow to decrease the background to the requires level. Finally, preliminary values of the background and sensitivity for a possible 100 kg HP Xe TPC experiment equipped with Micromegas have been obtained. These algorithms have been also used to evaluate the sensitivity for a 1 ton extension of the experiment. 123 Chapter 5 Background studies and discrimination algorithms Contents 5.1 Simulation codes .............................. 126 5.1.1 Decay0 ...................................127 5.1.2 Geant4 ...................................127 5.1.3 Simulation of the Physical system and anylisis: RESTSoft . . . . . . . 128 5.2 Simulated Events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 5.2.1 Signal ....................................131 5.2.2 Background .................................133 5.3 Simulated geometry ............................ 137 5.4 The topology of the events . . . . . . . . . . . . . . . . . . . . . . . . 138 5.4.1 Signal Events ................................138 5.4.2 Background .................................139 5.4.3 Deterioration of pattern recognition . . . . . . . . . . . . . . . . . . . 142 5.5 Discrimination Algorithms . . . . . . . . . . . . . . . . . . . . . . . . 149 5.5.1 A figure of merit for background reduction . . . . . . . . . . . . . . . 149 5.5.2 Energy ....................................150 5.5.3 Track selection ...............................150 5.5.4 Topology Selection .............................153 5.5.5 Fiducial rejection ..............................160 5.5.6 Summary ..................................160 5.6 Background Rejection . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 5.6.1 High diffusion gas: Pure Xenon ......................161 5.6.2 Low diffusion gas: Xenon-TMA. .....................162 5.6.3 Comparison between the two diffusions . . . . . . . . . . . . . . . . . 163 5.6.4 Surviving events ..............................165 5.7 Energy Resolution studies . . . . . . . . . . . . . . . . . . . . . . . . 168 125 5. Background studies and discrimination algorithms 5.8 Background level .............................. 173 5.9 Sensitivity for the neutrino effective mass . . . . . . . . . . . . . . 178 The sensitivity in a neutrinoless double beta decay experiment, depends on several factors like the background level, the detection efficiency of the expected events or the energy resolution of the detectors. All these factors are related to the intrinsic characteristics of the experimental setup. To evaluate them and, specially, to have a good knowledge of the background, it is important to develop precise Monte Carlo simulations that take into account the geometry of the setup as well as the generation of the charge in the gas. The understanding of the background can also help to develop discrimination algorithms to distinguish between background and signal events. A gas time projection chamber (TPC), combined with a pixelized Micromegas detector, offers a very high spatial resolution [148] which can help to identify events: for example, in the case of a muon it will be a long straight line crossing all the detector or an alpha event in a high pressure gas chamber will be a short track with a high concentration of charge as shown in previous chapter. The characterization of the signals and background expected in the experiment before the data-taking starts is an important point in order to know the expected background rate and to develop analysis tools. In this chapter, a description of the simulation codes used to generate events and to study their interactions is presented as well as a discussion of the main backgrounds concerning an experiment searching for ββ0νin Xenon. The topology of the signal and background events is studied. Then the chapter continues with a description of the selection criteria applied, several of them based on the topology of the events. In the next section the algorithms developed are applied on the simulated backgrounds and signal events. Two different conditions of diffusion are considered and the rejection power and efficiency are computed in both cases. In addition, the nature of the surviving background events and rejected signal is studied. Moreover, different energy regions are considered and the rejection factor on them obtained. In the last section, the background level produced with the surviving events is calculated based on measured activities of materials and it is also obtained the expected sensitivity to the effective neutrino mass with these background levels, efficiency and different energy resolutions. 5.1 Simulation codes A full Monte Carlo simulation has been performed to produce the events in a TPC filled with Xe gas at 10 bar. The interactions and processes that the events suffer in the gas, as well as the physical processes that occur in a TPC and the analysis of the events, are implemented in three logical blocks. 1. Generation of signal and background events in a gas •Diffusion studies with Magboltz and event generation with Decay0. •Simulation of the geometry and the interactions of the particles with Geant4. 126 5.2. Simulated Events 5.2.2 Background The other group of events to simulate is the background. In chapter 1, the main components that contribute to the background in a ββ0νexperiment are described in general. Before starting any simulation, it is useful, first of all, to analyse which contributions can be neglected and which ones cannot, in order to optimize the simulation-time. A factor that can help to decide which isotopes are desired to be simulated is the energy window or region of interest (RoI) of the experiment. Anything well below the Qββ energy can be neglected. It is not the same for particles with higher energies because, through different processes, they can lose part of their energy and end up in the energy window. The RoI considered in the main parts of the work, as commented, is from 2400 to 2500 keV (that represents, roughly, an energy resolution of 2% FWHM at Qββ). Another natural selection is to divide the contributions for the background into internal and external. Internal embraces any contribution coming from the detector itself as it could be the construction materials, the electronics or the ββ emitter. External includes any contribution coming from the environment, as the walls of the laboratory or the radon in the air. The external component produced by gammas is reduced using passive shielding. For muons an active veto can be used. In the case of the internal contribution, there are not only gammas but also electrons and alpha particles depositing energy in the sensitive volume. Internal contaminations The materials for the detector setup and shielding contain impurities of radioactive elements. Natural radioactivity decay chains of 232Th and 238U and the isotopes of 40K and 60Co are the most important ones. 40K emits a gamma of 1460 keV, and is only relevant for the ββ2νmode. 60Co can appear by cosmogenic activation of the copper. It consists of a βemission of 318 keV and two gammas of 1173 and 1332 keV. Different physical processes can allow these three particles to deposit energy in the RoI but their contribution can be easily rejected through the analysis as shown in [138], due to the fact that they deposit their energy with a large spatial separation producing a multitrack event. In general, alpha decays are easily identified and rejected because they will leave all their energy in small tracks close to the walls. 232Th Chain In the natural decay chain of 232Th the only emission that can deposit energy in the RoI is the 208Tl produced by the 212Bi decay. In Figure 5.4 its energy levels are shown; the 99% of the times it decays emitting a gamma with an energy of 2615 keV, higher than the Qββ value but that can deposit an energy in the RoI by several processes that will be discussed in detail later on. 238U chain The main isotope from this chain than concerns in the search of the 136Xe ββ0νis the 214Bi. In this case, several gamma emissions, summarized in Table 5.2, can deposit energy in the RoI. The levels with an intensity higher than a 1% are shown in Figure 5.5. Among them, there is a 133 5. Background studies and discrimination algorithms 208Tl 3.05 min 3.05 β3.1% 3961 keV 24.3% 3708 keV 3475 keV 21.8% 3198 keV 51% 2615 keV 208Pb Q = 4.99 MeV 6.3% 21.6% 1.8% % 84.5% 12.4% % 100% 2615 keV 0.03% Figure 5.4: Decay levels of the 208 Tl. The diagonal arrows are βdecays and the vertical arrows are γdecays. The percentages indicate the decay probability. beta emission with a transition energy of 3272 keV and a gamma emission of 2447 keV combined with a beta emission of 824 keV. Also important is the 222Rn since it decays 88% of the times into 214Bi also. Energy (keV) Intensity(%) Energy (keV) Intensity(%) 2447.9 1.57 2921.9 0.01 2694.7 0.03 2978.9 0.01 2769.9 0.03 3053.9 0.02 Table 5.2: Gamma emissions of the 214Bi isotope with energies higher than 2.4 MeV. 214Bi 19.7 min β2.8% 2447 keV 4.3% 2118 keV 1847 keV 8.2% 609 keV 214Po Q = 3.27 MeV 46.1% 1.7% 5.7% 2204 keV 2.2% 2017 keV 17% 1764 keV 1730 keV 17.8% 2.9% 1543 keV 1378 keV 7.4% 18.2% 164 µs Figure 5.5: Decay levels of the 214Bi. The diagonal arrows are βdecays and the vertical arrows are γdecays. The percentages indicate the decay probability. 134 5.2. Simulated Events Double beta decay The source of the ββ0νdecay, the 136Xe has also some contaminants from 232Th and 238U. These activity have been measured in the Gothard experiment being at the level of 5×10−12 g232Th/g and 9×10−12 g238U/g. Xenon can also be activated cosmogenically, 136Xe(n,γ)137Xe by neutron capture. 137Xe has a beta emission the 67% of the times with an energy of 4173 keV and the 30% of times a beta plus gamma emission of 3173 keV and 455 keV respectively. The initial activity, A0, is defined A0(kg−1d−1) = R1−e−λtexp ,(5.6) where Ris the production rate, λis decay constant and texp is the exposure time. As its decay half-life is very short (3.8) minutes, the initial activity is equivalent to the production rate. With this half-time its contribution is negligible. For the ββ2νmode, its contribution to the ββ0νcan be calculated with Equation 1.25. Taking into account the recent results presented in [45] and [104] for the T2ν 1 2 and the limits for the T0ν 1 2 the fraction of ββ2νin the RoI of the ββ0νmode is 2.2×10−9. Hence it can be neglected with a 2% FWHM at Qββ. External contaminations Among the possible external contaminations, the muons can be clearly reduced when working underground. Specifically, in the Underground Laboratory of Canfranc (LSC, 2450 m.w.e) the flux is reduced by 5 orders of magnitude as shown in Figure 5.6, where different underground laboratories are compared. Figure 5.6: Graphic where the flux of muons measured in different underground laboratories is shown. In the case of the Canfranc Underground Laboratory this flux is of the order of 10−3m−2s−1. The other main contribution comes from the natural radioactivity chains of contaminants present in the laboratory rocks. As they are generated outside the setup, only energetic gammas can reach the detector. Although their contribution maybe higher than the internal one, a passive 135 5. Background studies and discrimination algorithms shielding, usually made of lead or water, can reduce it to levels below those of the internal components. Studies about which thickness is needed in each case to have a background level of the order of few 10−4c keV−1kg−1y−1, have been performed. Simulation of gammas coming from the laboratory walls have been carried out. An external flux of 0.13 γ/cm2/s (from measurements in the LSC [187]) has been considered. Then a water shielding surrounding the detector was simulated, considering three different thicknesses: 30, 50 and 100 cm.Without shielding and using as input the 0.13 γ/cm2/s flux a background of 10 c keV−1kg−1y−1is produced after the application of the rejection criteria that are explained later. In order to obtain a rejection factor of 106, as from internal contaminants (see Section 5.6), a 300 cm thickness of water is needed. Another option is to use radiopure lead. Taking into account that it has a density 12 times higher than water it can be calculated that, to obtain the same rejection factor, a thickness of lead of around 25 cm is needed. It is also important the Radon present in the air because it decays into 214Bi. A way to reduce its contribution is to purge the volume between the shielding and the detector with Nitrogen. Summary After these discussions, the simulated events to be considered in this work are: •ββ0νevents: signal events generated with Decay0 and homogeneously emitted inside the fiducial volume of the Geant4 geometry previously defined. •208Tl: the complete decay chain of this isotope using Decay0 file as input for the Geant4 geometry. It has been simulated for the different geometrical elements discussed in Section 5.3. •214Bi: the complete decay chain of this isotope using Decay0 file as input for the Geant4 geometry. It has been simulated for the same geometrical elements as for 208Tl. The number of simulated events are summarized in Table 5.3. They have been analysed assuming two diffusions, high diffusion as in pure Xenon or Argon, and a low diffusion like in ArgonIsobutane or in a Xenon mixture with, for example, Trimethilamine (TMA). The number of simulated events in all cases is more than one million with the intention that for all the estimations and calculus done with them, statistical relative errors, smaller than the 1%, are obtained. Events (×107)ββ0νVessel Field Cage Readout Cathode 208Tl 214Bi 208Tl 214Bi 208Tl 214Bi 208Tl 214Bi Pure Xe 0.12 1.7 19.7 1.6 6.11 7.4 28.1 1.1 2.7 XeTMA 0.12 1.9 22.9 0.8 23.1 1.5 1.5 1.1 2.7 Table 5.3: Number of simulated events from different parts of the internal setup to study the internal background. Signal events have been launched isotropically in the fiducial volume. Both, signal and background, have been studied for two diffusions: pure Xenon (high diffusion) and in a low diffusion Xenon mixture. 136 5.4. The topology of the events 5.3 Simulated geometry The designed geometry and the materials have been selected according to several factors as the required background level or the requirement to work at high pressure (HP), that implies mechanical restrictions. Also the experience with the NEXT-MM prototype has helped to define different elements. The simulation has been performed to be easily scalable in case some of the dimensions need to be changed. The vessel has been chosen to be made of Copper. Copper has shown to have a high degree of radiopurity [165] and it can hold high pressure (HP). The chosen thickness is 3 cm to increase the auto-shielding effect and to be able to work at 10 bar. The simulated end-caps have a spherical geometry to allow the work at high pressure. The dimensions of the vessel are: a length of 1.5 m and an inner diameter of 1.6 m. There is a distance between the vessel and the field cage of 5 cm in each side; this space is filled with the desired gas. The field cage is based on the design used in NEXT-MM. It consists of a Teflon cover, to avoid the possible sparks, with Cu rings embedded on it. The length is the same as for the vessel and the inner diameter is 1.39 m. There are a total of 134 rings with a thickness of 1 mm and they are separated by 1 cm. A cathode is placed on top of the field cage, with the same outer radio and thickness as the rings. It is also made of Cu and has an 88% of transparency. At the bottom of the field cage is placed the readout. The sensitive volume, where the interactions are registered, is the volume defined between the field cage, the cathode and the readout, and is filled with Xenon gas at 10 bar. The volume is 2.24 m3that means 124 kg of Xe at 10 bar and at 20 ◦C. Two flanges and two endcaps at both sides of the vessel have been simulated to enclose it. A sketch of the design is shown in Figure 5.7. Images of the resulting geometry simulated with G4 and visualized with the software vrml are in Figure 5.8. The background events are simulated from the volume of the lateral vessel and field cage and from the surface of the cathode and readout plane. Previous studies simulating from the endcaps and feedthroughs have shown that the contribution is much smaller because of the solid angle and distance than from these volumes or surfaces closer to the fiducial volume. Figure 5.7: Artistical view of the simulated vessel. 137 5. Background studies and discrimination algorithms (a) (b) Figure 5.8: Images of the simulated geometry obtained using Geant4. (a) General view of the different parts: in red the vessel with its semi-spherical end-caps, in green the field cage and in magenta the fiducial volume. (b) Detail of the Cu rings simulated for the field cage. In this image the Teflon is not shown. 5.4 The topology of the events In a gaseous TPC detector the path that a particle follows along the medium (track) can be determined with a pixelized detector. This allows to do pattern recognition with a high accuracy and it gives the possibility to define some discrimination algorithms to reject background, a crucial point to be able to detect such a rare process as neutrinoless double beta decay (can be seen in Equation 1.31). Moreover, the topology of the events depends on the gas (diffussion coefficients) and on the working pressure. Higher pressures and diffusion may complicate track detail recognition. As described before, an event is a set of energy depositions produced in the gas when a particle (signal or background) interacts with its molecules. Each one of the energy deposits, as mentioned in 5.1.3, is converted into charge, is drifted and spread according to diffusion parameters and, finally, pixelized. After these processes, the energy depositions produce a set of 3D pixels. Depending on the initial particles, initial positions and gas conditions (pressure and diffusion) this collection of pixels will have a characteristic topology. In this Section these features, that make them unique and that can help to differentiate between a ββ0νsignal and background, will be commented. For simplicity, only the contaminants from the background events simulated that can generate a signal in the RoI will be discussed. Previous studies were carried out to better understand the phenomenology of the events and to fit with a higher accuracy the discrimination methods. These results are also presented, as for example, what is the effect of a charge threshold in the readout, or the effect of diffusion in pattern recognition. 5.4.1 Signal Events Aββ0νevent will have a very characteristic topology: a long track with two high accumulations of charge where the electrons emitted in the process are recombined with the gas. The length of the track depends on the operational pressure. A 136Xe ββ0νevent at 10 bar is expected to be up to 30 cm long. Extra short tracks produced by Bremsstrahlung radiation or other secondary processes can appear (it will be discussed in more detail in Section 5.4.3). An example of a 138 5.4. The topology of the events typical ββ0νevent obtained in pure Xe at 10 bar can be seen in Figure 5.9. In this example one electron has an energy of 1.7 MeV while the other has 0.8 MeV. The track length is of the order of 20 cm due to warping. Thanks to the software it is possible to obtain the 3D reconstruction of the event (5.9 (a)) and the projection of the track in the XY plane of the readout (as in 5.9 (b)). In the plots the different colours represent the deposited charge per pixel (in number of electrons), and the scale is shown on the right of the xy plot. Position X (mm) 280 300 320 340 360 380 400 420 Position Y (mm) 510 520 530 540 550 560 570 580 590 600 610 Position Z (mm) 100 110 120 130 140 150 160 170 (a) X-axis (mm) 280 300 320 340 360 380 400 420 Y-axis (mm) 510 520 530 540 550 560 570 580 590 600 610 0 2000 4000 6000 8000 10000 12000 (b) Figure 5.9: An example of a neutrinoless double beta decay in 136Xe obtained with the simulation. The work conditions are a pixel of a 1 cm2and at 10 bar in pure Xe. The two blobs can be seen at both ends of the track. In this example the energy of the electrons was 1.7 MeV and 0.8 MeV, respectively, and all the energy of the event (2.45 MeV) was registered in the fiducial volume. In (a) the 3D view is shown and (b) is the projection of the track in the XY plane. Hence, a signal event will have three characteristics that can help us to discriminate between background and signal: •Most of the signal events will have a single track. In the case of secondary emissions the final event have at least a long main track with most of the energy. •The longest track ends in two high energy deposits, due to the Bragg absorption of the electrons, called blobs. •The events are isotropically distributed in the fiducial volume. 5.4.2 Background Background events are generated by contaminants in the materials or in the gas itself and may deposit energy in the RoI. It is important to study, not only the initial particles but also the physical processes that they suffer to understand the final topology of the events. As discussed in Section 5.2 the only isotopes that are simulated are 208Tl and 214Bi. Here the different physical processes and the expected topology of such events are presented. 208Tl events: photon interactions As discussed previously the main contribution from this isotope is its gamma of 2615 keV. The different photon interactions are: 139 5. Background studies and discrimination algorithms 1. Photoelectric absorption. The energy depositions are far from the RoI if the energy resolution is below a 5% FWHM at 2458 keV. But, in around the 15% of the cases, the photoelectric absorption comes with a secondary energetic electron radiation (by Bremsstrahlung process). This electron usually has an energy of the order of 100-200 keV. Two effects can arise: •The electron escapes from the fiducial volume and the final event will have an energy in the RoI. In principle it would be discriminated when looking for two blobs. •More than one Bremsstrahlung electrons of this energy occur. Then, one can leave the detector and another can produce a secondary fake blob at one end of the track that mimics a ββ0νtrack. 2. Compton Scattering inside the TPC. Electrons with an energy up to 2381 keV and photons from 233 keV to 2615 keV are generated. The three more dangerous cases are: •A photon that interacts again close to the main track. It can occur if a low energy gamma is emitted with an energetic electron that produces the main track. If the lower energy gamma interacts again inside the detection volume the event will have two tracks. To be in the RoI part of the energy has to leave the detector. In this case, it will happen in most of the case due to Bremsstrahlung radiation of the energetic electron. In principle, this case should be easily rejected except in the case where the interaction happens too close to the first electron track. •The emission of a lower energy electron plus an energetic gamma. In this case the electron can produce a secondary blob close to the main track being difficult to reject. If the gamma escapes from the fiducial volume, then the final event would be outside the RoI, or it can suffer a secondary Compton scattering where just part of the energy leaves the detector and, therefore, the event would be in the RoI. In this case the event will be a multi-track event with at least one larger second track far from the first one. 3. Compton interaction in materials outside. A lower energy photon enter the fiducial volume and deposits its energy by photoelectric effect. Small Bremsstrahlung emissions interacting close to the track may mimic blobs In conclusion, for the 208Tl most of the events will consist in several tracks. Events with one single long track depositing its energy in the RoI are a photoelectric absorption with a radiation lost or Compton or multi-Compton events. An example of each case can be seen in Figure 5.10. Figure 5.10 (a) and (b) represent a photoelectric event of 2615 keV, and in this case the event is a single-track but with just one blob at one end; (c) and (d) correspond to a multi-Compton event. 214Bi events Different gammas from the 214Bi have an energy higher than 2.4 MeV as summarized in Table 5.2. As mentioned, the most important one is the gamma of 2447.8 keV. It can deposit all its energy by photoelectric absorption and would be completely contained in the RoI unless an energy resolution better than 0.5% FWHM at Qββ is reached. In principle they would be single track 140 5.4. The topology of the events Position X (mm) 100 150 200 250 300 350 Position Y (mm) 120 140 160 180 200 220 240 260 Position Z (mm) 480 500 520 540 560 580 600 (a) Position X (mm) 100 150 200 250 300 350 Position Y (mm) 120 140 160 180 200 220 240 260 0 1000 2000 3000 4000 5000 6000 7000 8000 (b) Position X (mm) -600 -500 -400 -300 -200 -100 0100 Position Y (mm) -200 -100 0 100 200 300 Position Z (mm) 160 180 200 220 240 260 280 300 320 340 360 380 (c) Position X (mm) -600 -500 -400 -300 -200 -100 0 100 Position Y (mm) -200 -100 0 100 200 300 0 2000 4000 6000 8000 10000 (d) Figure 5.10: 208Tl events registered in a pixelized detector in a TPC fill with Xe at 10 bar. (a) 3D reconstruction of the track that corresponds to an event that has deposited its energy due to photoelectric effect (energy of 2.615 keV) (b) xy projection of the event in the pixelized readout. In this case the event is a single-track but with just one blob at one end. (c) and (d) similar plots for a multi-Compton event. events with only one blob at one one of the two ends. However, if a secondary emission via Bremsstrahlung radiation deposits all its energy in the detection volume and has such a high energy as to produce a blob close to the main track, then, the track would be misidentified as a signal track, as in the case of the 208Tl. Other gammas with higher energies can deposit their energy in the RoI due to Compton scattering, as explained in the case of the 208Tl, but their intensity is too low to be relevant. However, the full chain has been simulated. Together with the gamma emission, in the 214Bi decay, two beta emissions can be produced: •Electrons with a continuous beta spectrum with its transition Qββ at 3272 keV. This case has an intensity of 18%. An electron can easily be confused with a signal event if a secondary blob appears. But the only ones that can reach the RoI are those emitted from the surfaces close to the fiducial volume and in these cases they will leave energy near the walls and can be rejected using a veto volume. •The second electron beta spectrum has a maximum energy at 2663 keV and a gamma emission of 824 keV. Different processes can allow these two particles to leave energy in the 141 5. Background studies and discrimination algorithms RoI but they can be easily rejected, either using a veto volume or rejecting events with more than one energetic track. In conclusion, 214Bi can produce multi-track events, as in 208Tl, or single track with just one blob. This single track can come from the beta emission or from the gamma at 2457 keV. Examples obtained can be seen in Figure 5.11, (a) and (b) are for a gamma that deposits all its energy inside the chamber via a photoelectric effect and (c) and (d) represent the track produced by an electron coming from the walls that deposits an energy of 2300 keV. In both cases the event has just a single blob. Position X (mm) -440-420-400-380-360-340-320-300-280 -260 Position Y (mm) 200 250 300 350 400 Position Z (mm) 1000 1020 1040 1060 1080 1100 (a) Position X (mm) -440 -420 -400 -380 -360 -340 -320 -300 -280 -260 Position Y (mm) 200 250 300 350 400 0 1000 2000 3000 4000 5000 6000 7000 (b) Position X (mm) 100 120 140 160 180 200 220 240 260 Position Y (mm) 450 500 550 600 650 Position Z (mm) 950 960 970 980 990 1000 1010 1020 1030 1040 1050 (c) Position X (mm) 100 120 140 160 180 200 220 240 260 Position Y (mm) 450 500 550 600 650 0 1000 2000 3000 4000 5000 6000 (d) Figure 5.11: 214Bi events registered in a pixelized detector in a TPC filled with Xe at 10 bar. (a) 3D reconstruction of the track that corresponds to an event that deposits its energy by the photoelectric absorption of the 2447 keV gamma. (b) XY projection of the event in the pixelized readout. (c) and (d) similar plots for a beta emission coming from the walls. In both cases the events have just a single blob. 5.4.3 Deterioration of pattern recognition Effect of the energy threshold per pixel All readouts have a lower threshold in the energy (charge) that can be detected. A study was performed to understand the effect that this threshold in the pixels has on the efficiency to detect electrons. Different charge thresholds were imposed in order to consider a pixel active or not. The range of the study is from 0 to 10 keV. The effect on signal events and background events was studied. In Figure 5.12 is shown the efficiency for signal events as a function of the pixel 142