Research Collection
Doctoral Thesis
Design of an experiment to search for invisible decays of ortho- positronium in vacuum and ortho-positronium formation studies in mesostructured silica films
Author(s): Gendotti, Ulisse
Publication Date: 2010
Permanent Link: https://doi.org/10.3929/ethz-a-006250397
Rights / License: In Copyright - Non-Commercial Use Permitted
This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.
ETH Library Diss. ETHZ No. 14774 and DISS. ETH NO. 19067 ETHZ-IPP Internal Report 2002-05 August 2002
Design of an experiment to search for invisible decays of ortho-positronium in vacuum and ortho-positronium formation studies in mesostructured silica films
A dissertation submitted to the
SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZURICH
for the degree of
Doctor of Sciences
presented by
Ulisse Gendotti Dipl. Phys., ETH Zurich born 12.12.1981 citizen of Prato Leventina, TI Switzerland accepted on recommendation of Prof. Andre Rubbia, examiner Prof. G¨unther Dissertori, co-examiner
2010 ii Abstract
In this PhD thesis, the design of an experiment for the search for invisible decays of ortho-positronium (o-Ps) in vacuum is presented. Such decays could be interpreted as a signal for Mirror type Dark Matter. The goal is to reach a sensitivity in the branching ratio of Br(o Ps invisible) 10−7 to confront the annual modulation of the signal observed by DAMA/NaI− → and DAMA/LIBRA≃ (with 8.2σ significance) with Mirror Dark Matter scenarios. In case of a signal observation, the experiment would offer a unique and essential feature for a check of this signal: by changing the number of collisions of o-Ps in the vacuum cavity by a factor 2 the signal is scaled by the same factor while the background stays constant. In case of a null result, this search will provide an upper limit for the photon-mirror photon mixing strength about one order of magnitude better than presently derived from the Big Bang Nucleosynthesis. Based on the ETH Zurich slow positron beam, a special apparatus to select an appro- priate target for the Ps formation has been constructed. In some targets, fractions as high as 40 % of the positrons implanted in the film were converted into o-Ps and emitted from the surface. Moreover, measurements in a temperature range of 50-400 K were performed showing that the o-Ps yield is almost independent of the film temperature. Quantum mechanical effects related to confinement of o-Ps in the pore have been observed. In par- ticular, it has been experimentally demonstrated that the lowest possible emission energy of o-Ps is limited by the energy of its ground state in the pore confinement potential. These experimental results provide a solid ground for the future development of low temperature e+ o-Ps converters . →
iii iv ABSTRACT Zusammenfassung
In dieser Dissertation wird die Entwicklung und Planung f¨ur ein Experiment zur Suche nach unsichtbaren Zerf¨allen von Ortho-Positronium Atomen (o-Ps)im Vakuum vorgestellt. Solche Zerf¨alle k¨onnten als Signal f¨ur Dunkler Materie in einer Spiegelwelt interpretiert werden. Das Ziel ist es, eine Empfindlichkeit im Verzweigungsverh¨altnis von Br(o Ps invisible) 10−7 zu erreichen, um die von den DAMA/NaI und DAMA/LIBRA− Experi-→ ≃ menten beobachtete j¨ahrliche Variation ihres Signals (mit 8.2σ Signifikanz) mit Szenarien von Dunkler Materie in der Spiegelwelt vergleichen zu k¨onnen. Falls ein Signal beobachtet wird, bietet das vorgestellte Experiment eine einzigartige und wesentliche M¨oglichkeit, dieses Signal zu ¨uberpr¨ufen: indem man die Anzahl Kollisionen von o-Ps mit der Vaku- umkammer um einen Faktor 2 ¨andert, ¨andert auch das Signal um denselben Faktor bei gleich bleibendem Untergrund. Falls kein Signal beobachtet wird (Null-Resultat), liefert das Experiment eine obere Grenze f¨ur die St¨arke der Photon - Spiegelphoton Mischung, die etwa um eine Gr¨ossenornung besser ist, als was heute aus der Big Bang Nukleosynthese hergeleitet werden kann. Basierend auf dem Strahl von langsamen Positronen der ETH Z¨urich wurde ein spezieller Apparat konstruiert, um ein m¨oglichst gutes Target f¨ur die Positronium-Formation auszuw¨ahlen. In einigen Targets bildeten bis zu 40 % der im Film implantierten Positronen o-Ps, das von der Oberfl¨ache emittiert wurde. Zus¨atzlich haben Messungen im Temperaturbereich von 50 - 400 K ergeben, dass die o-Ps-Ausbeute nahezu unabh¨angig von der Filmtemper- atur ist. Quantenmechanische Effekte wurden beobachtet, die vom Einschluss der o-Ps- Atome in den Poren des Filmtargets herr¨uhren. Insbesondere wurde experimentell gezeigt, dass die niedrigste Energie der emittierten o-Ps-Atome limitert ist durch die Energie des Grundzustandas von o-Ps im Potential, das die Atome in den Poren einschliesst. Diese ex- perimentellen Resultate bieten eine solide Grundlage, f¨ur die Entwicklung von zuk¨unftigen e+ o-Ps Konvertern bei niedrigen Temperaturen. →
v vi ZUSAMMENFASSUNG Contents
Abstract iii
Zusammenfassung v
1 Mirror Matter as Dark Matter 1 1.1 PositroniumandDarkMatter ...... 2 1.2 Experimental evidence forMirrorDarkMatter ...... 7
2 Search for the invisible decay of o-Ps in Aereogel 9 2.1 Experimentaltechnique...... 9 2.2 Background estimation and dedicated engineering run ...... 17 2.3 Dataanalysis ...... 19 2.4 Results...... 20 2.5 Interpretation ...... 23
3 Design of an experiment to search for invisible decays of o-Ps in vacuum 25 3.1 Thesetup ...... 26 3.1.1 Theslowpositronbeam...... 27 3.1.2 o-Psproductiontarget...... 28 3.1.3 Thevacuumcavity ...... 29 3.1.4 Positrontaggingsystem ...... 32 3.1.5 Trigger efficiency and confidence level ...... 34 3.1.6 Thephotondetector ...... 39 3.2 Backgroundestimation ...... 39 3.2.1 Fastbackscattered o-Ps fromthecarbonfoil ...... 43 3.2.2 Backscatteredpositrons ...... 43 3.3 Sensitivity...... 45 3.4 Summary ...... 47
4 Study of positronium emission from porous silica films 55 4.1 Positrons and positronium in solids ...... 55 4.1.1 Positronium formation mechanism ...... 57 4.1.2 Thepick-offprocess...... 58
vii viii CONTENTS
4.1.3 Positronium diffusion and thermalization in porous films ...... 59 4.2 Dopplerbroadeningspectroscopy(DBS) ...... 61 4.3 Angular correlation of annihilation radiation (ACAR)...... 61 4.4 3γ/2γ technique...... 62 4.5 Positron annihilation lifetime spectroscopy (PALS) ...... 63 4.6 Positronium time-of-flight (Ps-TOF) ...... 63
5 The slow positron beam 65 5.1 Design and construction of a 22Na positronsourcechamber...... 65 5.1.1 Thepositronmoderator ...... 69 5.2 Thepositrontransportsystem...... 71 5.3 Thevacuumsystem...... 73 5.4 Positron and positronium tagging ...... 73 5.4.1 Pulsedbeammode ...... 73 5.4.2 Continuos beam and secondary electron trigger mode ...... 74 5.5 Beam profile monitoring using a MCP with a position sensitive screen . .. 77 5.6 Conclusions ...... 79
6 Design and construction of the PALS and TOF detectors 81 6.1 Monte-Carlosimulation...... 81 6.2 The magnetic transport system in the target region ...... 82 6.3 ThePALSdetector ...... 83 6.3.1 3γ/2γ detection efficiency correction using MC simulations . . . . . 84 6.4 The positronium time-of-flight detector ...... 87 6.5 Cooling of the target using a cryocooler ...... 89 6.5.1 Position calibration of the TOF spectrometer at low temperatures. 91 6.6 TheDAQsystem ...... 92 6.7 PALSdataprocessing ...... 92 6.8 TOFdataprocessing ...... 93 6.9 Conclusions ...... 96
7 Analysis of the PALS spectra 97 7.1 Samplepreparation...... 97 7.2 Extrapolation of the pore size from the o-Ps lifetime ...... 99 7.2.1 Lifetime in sub-nanometer pores: Tau-Eldrup model ...... 99 7.2.2 Lifetime in mesoporous materials, the RTE model ...... 99 7.3 Determination of the lifetime in the pores ...... 101 7.3.1 Measured yield of the o-Ps emission into the vacuum and the o-Ps lifetimeinpores...... 107 7.3.2 Calculation of the pore size ...... 110 7.4 Conclusions ...... 111 CONTENTS ix
8 Analysis of the TOF data 113 8.1 Ortho-positronium thermalization in porous films ...... 114 8.1.1 Theclassicalmodel...... 114 8.1.2 Positronium cooling by phonon scattering ...... 117 8.2 Determination of the o-Ps energy from the TOF measurements ...... 120 8.2.1 Lifetime and detection efficiency correction and TOF peak analysis 121 8.2.2 Subtractionofthetargetcomponent ...... 123 8.2.3 Subtraction of the non-thermalized component ...... 124 8.3 Lowtemperaturemeasurements ...... 127 8.3.1 Comparison with the MC simulation and reconstruction of the total o-Psenergy ...... 131 8.3.2 Ps in thermodynamic equilibrium at a temperature T in the pore . 134 8.3.3 TheP32sample...... 137 8.3.4 Conclusions ...... 138
9 Conclusions 141
10 Future perspectives 143 10.1 1S-2S transition of ortho-positronium ...... 143 10.2PsandAntihydrogen ...... 144 10.2.1 Precise o-Ps decay rate measurement ...... 145 10.3 Bose-Einsteincondensationofo-Ps ...... 145 x CONTENTS List of Figures
1.1 History of the ortho-positronium lifetime measurements and of the calculations. 4 1.2 The double degeneracy between ortho-positronium mass eigenstates of or- dinary (o-Ps) and mirror (o-Ps′) is broken when a small mixing term is included...... 5 1.3 The points are the annual modulation observed by DAMA/NaI. The line is the prediction of this modulation for Mirror Matter [12]...... 8
2.1 Schematic illustration of the experimental setup: a) front view, b) top view. 10 2.2 Schematic illustration of the positron tagging system and the o-Ps formation target of the setup...... 11 2.3 Schematic view of the scintillating fiber with the 22Na source on the squeezed part in the center...... 12 2.4 Schematic illustration of the method to readout the energy deposition in the fiber...... 13 2.5 An example of the pulse shape of the FBGO signal, which is an overlap of signals from a positron and a photon, detected in the fiber and the BGO crystal, respectively. The short gate was used to measure the energy loss from positrons (shown not in scale)...... 15 2.6 The scatter plot of energy loss measured with the short versus long gate measured for the TBGO counter coupled with the front face to the plastic scintillator. Three regions corresponding to detection of a) pure positrons detected in the plastic scintillator, b) 511 keV photons contaminated by the positron energy deposited in the plastic scintillator, and c) pure photons ,i.e. without detection of positrons in the plastic scintillator are clearly seen. The ellipse indicates the one sigma countur of the 1.275 MeV peak with a barely seen tail from positrons detected in the plastic scintillator...... 16 2.7 The cuts applied to the variables. The numbers on the plot correspond to the variable defined in the text. Only the colored regions contribute...... 20 2.8 Zero energy peak measured using the EC photon as a trigger. The arrow shows the threshold at 80 keV...... 21 2.9 Distribution of time between the fiber trigger and one of the annihilation photons in the TBGO. An exponential fit (solid line) is performed in order to determine the fraction and the lifetime of o-Ps produced in the aerogel. . 22
xi xii LIST OF FIGURES
2.10 Spectrum of the sum of the total energy in the ECAL. The inset shows the magnified view of the low-energy region in logarithmic scale...... 23 2.11 a) Mass–charge parameter space for the o-Ps decay into milli-charged par- ticles, excluded with this experiment, b) Comparison of our results (the dashed region on the plot) with other experimental (SLAC [61] and previous o Ps invisible [55]) and astrophysical bounds (the plot was taken from [62]).− ...... → 24
3.1 Schematic view of the setup. The main parts of the system are: The bunched slow positron beam, the secondary electron tagging system, the target cham- ber and the γ detector...... 26 3.2 Timing between the secondary electrons detected in the MCP and the bunch- ing pulse of the beam for 1 and 2 keV implantation energy of the positrons. 28 3.3 Fraction of o-Ps emitted into the vacuum from the a selected sample as a function of the positron implantation energy measured in our setup. .... 30 3.4 (a) Ps is tagged by the secondary electrons produced by the positron that hits the converter surface (target). In the case where Ps escapes the detection region, it could mimic an invisible decay due to the strongly suppressed de- tection efficiency of the decay photons. (b) By closing the vacuum cavity around the o-Ps converter one obtains a region of highly uniform detection efficiency and the o-Ps leakage through the aperture is suppressed. The 15 nm carbon foil acts as a barrier for the o-Ps emitted from the converter but it is nearly transparent for the incoming positron beam and for the secondary electrons used for the trigger...... 31 3.5 Schematic of the target region and the carbon foil...... 32 3.6 Schematic of the positron tagging system for the experiment using the MCP signal for secondary electrons emitted from the target and from the thin carbon foil...... 33 3.7 Trajectories of the positron (blue) and secondary electrons (red) for the new design with the carbon foil...... 33 3.8 Simulated time distributions of the secondary electrons detected in the MCP. See text for details...... 34 3.9 Setup used for the measurements with the tagging system. In order to accel- erate the SEs from the target a voltage difference between the sample holder (green) and the carbon foil has to be applied. A ceramic ring electrically insulates the carbon foil holder and the target holder...... 35 3.10 Example of an event acquired with an oscilloscope. The first signal is pro- duced by the SE from the target and defines the START signal. The second pulse is produced by the SE from the CF 15ns after the START signal. The time window ω between 15 and 22 ns∼ (i.e. between 10 and 17 ns delay from the START signal) is defined in order to calculate the trigger efficiency and the fake trigger ratio. A Gaussian fit is used in order to determine the position of the pulses (green)...... 36 LIST OF FIGURES xiii
3.11 Measured delay of the SE from the carbon foil relative to SE from the START signal. The dashed line represents the data obtained by applying the same voltage on the carbon foil and on the sample (-1.2kV). The solid line is obtained by applying -1.2 kV on the carbon foil and -3.8kV on the sample. In this case the broadening of the peak at around 15 ns is mainly due to variations of the flight time of the SE from the target that produce the START signal. The peak at about 2 ns are due to the MCP deadtime (see text for details)...... 37 3.12 Charge (measured with 50 Ω termination) in pVs of the start signal versus the charge of the signal for events in the time window 12-17 ns. Top: the same voltage of 1.2 keV is applied to the target and the carbon foil. Bottom: a voltage of 3.8 keV is applied to the target and 1.2 keV to the carbon foil. 40
3.13 Top: trigger efficiency (boxes) and coincidence suppression factor κSE (cir- cles) as a function of the threshold set on the signal in the time window 12-17 ns. Bottom: trigger efficiency (boxes) and coincidence suppression factor κSE (circles) as a function of the cut on the charge (voltage integral) on the signals in the time window 12-17 ns...... 41 3.14 Cross section of the BGO calorimeter mounted around the beam pipe. The sphere is to show that the minimal BGO thickness around the target is of the order of 200 mm...... 42 3.15 Distributions of the energy deposited in the dead material surrounding the target region (target substrate, beam pipe and copper coil) from annihilation events in the target. The upper plot shows the results of the MC simula- tion for a 0.84 mm thick aluminum pipe. Below the distribution for the pipe construction with 0.04 mm aluminum and 0.800 mm carbon. The total number of simulated 2γ-events is 108 in both cases. The peaks at 511 keV and 1022 keV correspond to the total photo-absorption either of a single 511 keV photon or of both of them, respectively...... 48 3.16 Energy distribution of fast o-Ps originating from backscattered e+. The pick- off probability after a collision of o-Ps with the walls is taken to be zero for o-Ps energies smaller than the 6.8 eV while for higher energies it is 0.95. . 49 3.17 Simulated decay position of o-Ps in the vacuum cavity. The escaping proba- bility estimated with the MC simulation is 1 10−4...... 49 ≃ × 3.18 Positron backscattering coefficients (η+) versus positron kinetic energy [70]. 50
3.19 Positron backscatter coefficient (η+) from a thick carbon target (black squares) and a thin carbon foil (triangles) as a function of the positron kinetic en- ergy. The transmission probability, through the thin foil is also shown (open squares, right scale)...... 51 3.20 Top: simulation using EGSnrc. Energy spectrum of positrons backscat- tered from a 20 nm carbon foil for positron implantation energies in the 1-2 keV range. The cut-off energy is 300 eV. Bottom: angular distribution for backscattered positrons ...... 52 xiv LIST OF FIGURES
3.21 Escaping probability for positrons backscattered from the carbon foil as a function of the carbon foil voltage for different positron initial energies E0. 52 3.22 Backscattering coefficient η+ for the SiO2 target as a function of the positron incident energy...... 53 3.23 Escaping probability for the total number of incident positrons at the SiO2 target as a function of the implantation voltage for different positron initials energies E0...... 53
4.1 Calculated Makhovian implantation profile for 10 and 15 keV incident positrons on tungsten. The parameters are α = 3.6[µg/cm2] and n = 1.6. See text fordetails...... 57 4.2 Orthopositronium lifetime dependence as a function of the mean free path l, computed with the RTE model. The pore size a for a cubic box can be 3 estimated with a = 2 l [82]...... 59 4.3 (a) In closed porosity samples Ps remains trapped in the pore. (b) In inter- connected pore systems Ps can move between the pores. If it reaches the film surface it is emitted into the vacuum...... 60
5.1 Picture of the ETH Zurich slow positron beamline. The installation is about 6 meters long and 2 meters high...... 66 5.2 Schematic diagram of the pulsed slow positron beam apparatus ...... 66 5.3 Source and moderator chambers for the 380 MBq source. The moderator chamber (yellow) is equipped with an electron beam (green) used for the moderator annealing up to 2000oC...... 68 ∼ 5.4 Section of the source chamber. The extraction optics system consists of a grid placed at a distance of 3 mm from the moderator and a first acceleration tube...... 70 5.5 Left: finite element calculation of the magnetic transport system performed with the COMSOL multi-physics program. Right: Positron trajectories cal- culated with GEANT4...... 71 5.6 Correction coils produce a magnetic field which is perpendicular to the beam direction. They are used for the beam positioning...... 72 5.7 Beam position with respect to the center of the beam pipe by applying a current on the correction coil. The current ranges from 0 to 5 A with a step of 0.2 A...... 72 5.8 Bunching voltage at the first and second velocity modulation gaps. The time delay between the non-linear pulse and the pulse applied to the chopper grid is tuned in order to achieve the best compression factor...... 74 5.9 Top: trajectories of the positrons (blue) and secondary electrons (red) from the Geant4 simulation; E- and B-fields calculated with the COMSOL package for our existing setup. Bottom: simulation of the secondary electron trajec- tories from the sample to the MCP for the PALS and TOF spectrometers described in Chapter 6...... 75 LIST OF FIGURES xv
5.10 Flight time spread introduced in the transportation of SEs from the sample to the MCP for a flight distance of 90 cm. This time distribution was measured using two different MCPs. One at the target region detecting e+ and one in the usual position for the secondary electron tagging. The voltage on the target MCP substrate was 2.4 kV. The timing shows a spread of 1.9 ns FWHM. 76 5.11 Top: principle of work of the MCP phosphor screen. Bottom: achieved im- provement in the beam profile before (left) and after (right) the optimization of the magnetic transport system. The external bright circle around the beam spot is produced by light reflection of the metallic ring holding the screen. See text for details...... 78 5.12 Left: picture of the MCP substrate. Right: picture of the phosphor screen. . 78 5.13 The solenoid around the MCP chamber is used to keep the value of the magnetic field at the MCP position similar to the one in the beam pipe. This is done in order to reproduce, at the MCP position, the same beam properties as in the beam pipe. FEM calculations have been performed. In the lower plot the calculated magnetic field along the beam pipe is shown. . 79
6.1 The solenoid is wound directly on the beam pipe in order to decrease the magnetic field at the PMTs region. An additional coil is used at the interface between the 500 mm diameter coil and the solenoid on the beam pipe (7 cm diameter) in order to make the transition of the magnetic field smoother. . 83 6.2 Top: picture of the PALS detector (left) and a 3D CAD drawing of the BGO calorimeter (right). Bottom: Schematic drawing of the PALS setup. .... 85 6.3 Lifetime measured in a single BGO crystal. The apparent lifetime measured depends on the distance of the crystal from the target. The detection proba- bility at larger distances is enhanced for long living o-Ps...... 86 6.4 Ratio of the 3γ/2γ detection efficiency as a function of the o-Ps emission energy for three different thresholds applied to the BGOs: 200, 300 and 400 keV...... 87 6.5 3D view and picture of the TOF detector...... 88 6.6 Analysis of the 511 keV peak height as a function of the collimator position. With this scan the 0-position (target) and the slit width can be monitored. . 89 6.7 Working principle of the cryocooler. See text for details...... 90 6.8 Cryocooler cold finger and sample holder. FEM calculations have been done in order to estimate the contraction of the cryocooler assembly during cooling down...... 90 6.9 Analysis of the 511 keV peak as a function of the collimator position at room temperature and after 1 hour of operation of the crycooler (45 K). The scan is performed with a 0.1 mm step...... 91 6.10 DAQ scheme used for the PALS and TOF detectors...... 93 xvi LIST OF FIGURES
6.11 The energy spectra of the 4 BGO crystals are shown. The calibration of the PALS energy spectra by fitting the 511 keV peak from the 2 γ decay.The pedestal is given when no energy is deposited in the BGO. The distribution in the energy range 0
7.1 The porous materials are produced by mixing a porogen to a matrix material. At the end of the film preparation the porogen is evacuated by heating the sample to 4500C...... 98 7.2 (a) Tau-Eldrup model: in the central portion of the spherical pore with ra- dius R+∆R the annihilation rate is λ =0 (the self annihilation is neglected) while in the interaction region it is equal the to averaged spin annihilation λS+3λT rate λA = 4 . (b) RTE model: the annihilation rate in the central region λS +3λT of the pore is λT while in the interaction region it is λA = 4 . .... 102 7.3 o-Ps lifetime as a function of the pore size calculated with the RTE model with δ =0.18 nm assuming cubic pores (dotted line) and infinite rectangular channel pores (solid line) ...... 102 7.4 RTE model lifetime calculations assuming cubic pores at 0, 300 and 800K for δ = 0.18 nm. The RTE model is calibrated by demanding that at 0 K (i.e. ground state) the TE model is reproduced for pore sizes in the sub- nanometers range...... 103 7.5 (a): Uncapped and (b) capped porous silica film: o-Ps is forced to remain in the film. In interconnected pore systems, this is done in order to study the internal properties of the film by avoiding that o-Ps emitted from the surface biases the values of the lifetime related to the annihilation in the pores. . . 104
7.6 Left: measured o-Ps lifetime components, (τ2,I2) , (142ns,I3) as a function of the porogen content. Right: o-Ps-escape model calculations...... 106
7.7 Yield of o-Ps emitted into the vacuum (Yv, circles) and total yield of Ps (Itot = I2 + I3, squares) for F (upper plot) and C samples (lower plot) at room temperature and at 50K...... 108 7.8 o-Ps lifetime in the film determined with the escape model. Values of 54 ns for the C sample and 76 ns for the F sample have been measured. In the C sample, for implantation energies above 7 keV, positrons start to be implanted in the glass substrate. In the F sample below 6 keV the calculated lifetime using the escape model is biased by surface effects...... 109 LIST OF FIGURES xvii
8.1 Thermalization curves for similar pore structures coated with different chem- icals. The points represent the experimental data recorded at different positron implantation energies. The solid lines are the fit with the Eq. 8.2 where the effective mass MS is left as a free parameter. The effective mass ratio be- tween the two chemical groups used in the preparation of F38 and m-F38 has been observed to be constant for α ranging from 1 to 4...... 116 8.2 Minimal energy of Ps as function of the pore size for cubic box (circles) and square channel (dotted line) geometries. The minimal energy of o-Ps in a square channel is smaller than in the cubic box of the same side length a. 119 8.3 Comparison between the energy loss curves obtained from the classical model and the Ps-phonon interaction model. Figure from [86]. See the text for details.120
8.4 Schematic view of the collimator and of the target region: dc is the distance between the target surface and the center of the collimator slit and wc is the collimator width. In the TOF spectrum the Ps annihilation in the target, whose γs penetrate the lead collimator and produce a signal in the BGO, contribute to the target-component...... 121 8.5 Measured (raw) TOF spectra at 300K for implantation energies ranging from 0.7-7 kV. Top: F sample. Bottom: C sample...... 122
8.6 Positronium emission energy < Ez > as a function of the implantation en- ergy at a target temperature of 300K. The triangles correspond to the energy obtained from the data by subtracting the target component (TC subtraction) and for the squares the target and the non-thermalized component were sub- tracted (TC+NTC subtraction). For both data sets the lifetime and detection efficiency correction factor was applied. The inserts show the plateaus with the low-energy part of the spectra cut off...... 125 8.7 Top: TOF spectra at 3,4 and 5 keV positron implantation energy. The ep- ithermal tails on the left of the TOF-peak decrease at larger implantation energy. Bottom: Makhovian stopping profile for different positron implan- tation energies. At higher energies the profiles are broadened. A fraction of positrons is always implanted near the surface, resulting in non-thermalized o-Ps to be emitted into the vacuum...... 126 8.8 Calculated fraction of positrons stopping at a depth smaller than L (open squares) and larger than L (black squres) obtained from Makhovian stop- ping profiles. L is the minimal implantation depth to produce thermalized positronium. Top: calculations for the C sample, bottom calculations for the F sample. The density and the thickness of the film are taken into account. 128 8.9 TOF spectra from the C sample (upper two plots) and the F sample (lower two plots): The solid histogram in panels (a) and (c) correspond to the measured (raw) TOF spectra, the diamonds represent the target component, and the crosses the non-thermalized plus the target component. Panels (b) and (d) show the corrected TOF spectra attributed to thermalized o-Ps decays.129 xviii LIST OF FIGURES
8.10 o-Ps mean emission energy < Ez > as a function of the positron implan- tation energy for the C (dashed line) and the F sample (solid line) at 300 K, after the subtraction of the target and non-thermalized component (cor- responds to the TC + NTC curves of Fig 8.6) The C sample shows a faster thermalization curve because of the smaller pore size. As expected, the mini- mal energy of the F sample is smaller than the one observed in the C sample. The insert shows the plateaus with the low-energy part cut off...... 130 8.11 Upper plot: Mean Ps energy < Ez > for implantation energies higher than 2 keV at 50 K and 300 K for the C sample. Lower plot: Mean Ps energy as a function of the implantation energies higher than 4 keV at 50 K and 300 K for the F sample...... 130 8.12 Upper plot: comparison between the data of the C sample at 6 keV and the MC simulating mono-energetic Ps emitted isotropically from the film surface. Lower plot: comparison between the data of the C sample at 6 keV and the MC simulating mono-energetic Ps emitted perpendicular from the film surface. In both cases, the measured target and non-thermalized components was added to the MC simulations...... 133 8.13 The correction parameter ξ between the total o-Ps emission energy < E0 > and the parallel component < Ez > obtained from MC simulations...... 134 8.14 Positronium mean energy < E0 > as a function of the mesoporous film temperature. Those results are obtained at 6 keV for the C (triangles) and 10 keV for the F sample (squares). The solid lines are the results of a fit of Eq.8.32 to the data with the pore side lengths a,b,c left as free parameters. The dashed lines were obtained fitting with Eq.8.32 with a single free side length free a = b = c (cubic box pores)...... 136 8.15 Sketch of a multilayer sample with combined porosity. (a) The interface between two layers of the same sample type acts as a blocking barrier for o-Ps (b) For two layers, one with a small pore size (e.g. a C sample), and the other with a large pore size (e.g. P32), the blocking effect might be suppressed...... 139
10.1 BEC transition temperature vs confining cavity radius assuming 108 positro- nium atoms...... 146 List of Tables
2.1 Comparison between expected (from MC) and measured background level for the different background sources in the engineering run and expected back- ground level for the final setup (see text for details)...... 18 2.2 Definition of cuts and the remaining fraction of events after the cut is applied. 21
3.1 Estimated flight time for SE emerging from CF (1.2 kV) and from the sample 3.8 kV for a distance of 80 cm. The expected delay time is 17 ns...... 35 3.2 Summary of the expected background level for the different background sources. 46
5.1 Beam intensity for the different 22Na positron sources that have been used since 2006. In the last row the efficiency of the moderator to emit almost mono-energetic, low energy positrons is reported...... 67
7.1 Film thickness (Z), calculated o-Ps lifetime in the pores τf , and escape rate kv at50Kand300K...... 108 7.2 Pore size calculated with the RTE model using the value of τf at300K. . . 110 8.1 Comparison of the pore sizes obtained from the PALS measurements and from the fit of the Ps mean emission as a function of the sample temperature (see Fig. 8.14) for both cubic box (BOX) and rectangular pores (RECT). Minimal energy < E0 > of Ps is taken at T=50 K. The errors are the combined statistical and systematic error...... 137
xix xx LIST OF TABLES Chapter 1
Mirror Matter as Dark Matter
The Dark Matter problem provides one of the strongest indications for the existence of physics beyond the Standard Model (SM). In the past years, many cosmological observa- tions accumulated support for the existence of Dark Matter (see e.g. [1] for an excellent review); these include galactic rotational curves [2] and gravitational lensing [3, 4]. The very recent confirmation by the DAMA/LIBRA [5] experiment of the DAMA/NaI [6] ob- servation of an annual modulation signal with a 8.2σ significance provides the first direct experimental observation of the existence of non-baryonic Dark Matter in our galactic halo. Supersymmetry is considered as one of the most attractive extensions of the SM, mainly because it provides an elegant solution to the hierarchy problem [7] and enables the cou- plings of the SM to evolve on a common scale in GUT theories [8]. Supersymmetric- particles could be viable candidates for Dark Matter (even though that’s not a compelling condition for their existence) if one forces R-parity to be conserved, leading superpartners to annihilate or be created in pairs and, therefore, the lightest supersymmetric partner (LSP) to be stable. However, the new results of DAMA/LIBRA give us a hint that probably standard neu- tralino models are not the solution of the Dark Matter puzzle. In fact, such an explanation would contradict other higher threshold experiments like CDMS/Ge [9], CDMS/Si [10] and XENON10 [11]. Among many Dark Matter candidates that have been discussed, one of the most promis- ing, which could reconcile the null results of those experiments with DAMA, are Mirror particles [12]. In fact, Mirror baryons are naturally dark, stable and massive. Currently, it seems that this concept could explain in a natural way that the visible and Dark Matter densities in the universe are of the same order of magnitude (ΩB =0.044 and ΩDM =0.26) [13, 14]. Furthermore, if Mirror Matter is present in our universe, it would mean that parity (spatial-inversion) is an unbroken symmetry of nature. Mirror Matter was originally discussed by Lee and Yang [15] in 1956, after their discov- ery of parity violation (for an excellent recent review on this subject see [16]). In order to save parity conservation they suggested, that the transformation in the particle space corresponding to the space inversion x x should not be the usual transformation P but PR, where R corresponds to the transformation→ − of a particle (proton [15]) into a re-
1 2 CHAPTER 1. MIRROR MATTER AS DARK MATTER
flected state in the mirror particle space. After the observation of parity non-conservation, Landau assumed [17] that R=C, i.e. he suggested to identify antiparticles with the Mirror Matter but then CP must be conserved, which we know is not the case. The idea was further developed by A. Salam [18], and was clearly formulated in 1966 as a concept of the mirror universe by Kobzarev, Okun and Pomeranchuk [19]. In their paper, they have shown that ordinary and Mirror Matter can communicate predominantly through gravity and proposed that the Mirror Matter objects can be present in our universe. Since that time, the concept of Mirror Matter has found many interesting applications and developments. In the 80’s, it has been boosted by superstring theories with E E′ 8 × 8 symmetry, where the particles and the symmetry of interactions in each of the E8 groups are identical. Hence, the idea of Mirror Matter can be naturally combined in these models [20]. Nowadays, Mirror Matter models exist in two basic versions. The symmetric version, proposed earlier, was further developed and put into a modern context by Foot, Lew and Volkas [21]. The asymmetric version was proposed by Berezhiani and Mohapatra [22]. In the following we will concentrate on the symmetric model since it is the most inter- esting from a Dark Matter perspective and it could provide, as it will be discussed later, an experimental signature related to positronium. In the symmetric mirror model, the idea is that for each ordinary particle, such as the photon, electron, proton and neutron, there exists a corresponding mirror particle of exactly the same mass as the ordinary particle. R-parity interchanges the ordinary particles with the mirror particles so that the properties of the mirror particles completely mirror those of the ordinary particles. For example, the mirror proton and mirror electron are stable and interact with the mirror photon in the same way in which the ordinary proton and electron interact with the ordinary photons. The mirror particles are unlikely to be produced in laboratory experiments just because they couple very weakly with the ordinary particles. In the modern language of gauge the- ories, the mirror particles are all singlets under the standard G SU(3) SU(2)L U(1)Y gauge interactions [21]. The mirror particles interact with a set≡ of mirror⊗ gauge particles,⊗ so that the gauge symmetry of the theory is doubled, i.e. G G (the ordinary particles are ⊗ singlets under the mirror gauge symmetry) [21]. Parity is conserved because the mirror particles experience V + A (i.e. right-handed) mirror weak interactions while the ordinary particles experience the usual V A (i.e. left-handed) weak interactions. Ordinary and mirror particles interact with each− other predominantly by gravity.
1.1 Positronium and Dark Matter
Few years after the discovery of the positron by Anderson [23], Mohorovicic predicted that the positron and the electron could form a bound state [24]. Ruark proposed to name this system positronium (Ps) and gave a qualitative discussion of the spectroscopic structure of the Ps-atom [25]. The spins of the electron and of the positron in the positronium 1 can combine to give either a singlet spin ground state S0, called para-positronium (p- 3 Ps) or a triplet ground state S1, ortho-positronium (o-Ps). For a positronium system 1.1. POSITRONIUM AND DARK MATTER 3
with orbital angular momentum l and total spin S C-parity is C = ( 1)l+S and P-parity − is P = ( 1)l+1. For para-positronium and ortho-positronium in the ground state we P C − −+ P C −− have Jp−Ps = 0 , Jo-Ps = 1 . For other positronium states the quantum numbers are 2S+1 1 3 3 3 3 3 −− n LJP C = 2 S0−+ , 2 P0++ , 2 P1++ , 2 P1+− , 2 P2++ , 2 S1 , .... The detailed calculation of the decay rate for the spin singlet state was done by Wheeler (1946) [26] and Pirenne (1947) [27]. The more difficult calculation for the lowest order decay rate of the triplet spin state was first obtained correctly by Ore and Powell [103] in 1949. The first experimental detection and the measurement of the lifetime of the ortho-positronium, which confirmed the calculation of Ore and Powell, was performed by Deutsch (1951) [29, 30]. A lot of effort has been undertaken to determine the basic properties of o-Ps like lifetime, decay modes, spectroscopy, etc. In particular, the measurement of the o-Ps lifetime caught much atten- tion. The history of these measurements and the theoretical calculations is very interesting and at the same time controversial. Fig. 1.1, summarizes all these years of theoretical and experimental efforts to determine the lifetime after Deutsch’s first detection. The mea- surements performed by the Michigan group in the late eighties did not agree with theory [31]. This long standing discrepancy was called “the ortho-positronium lifetime puzzle” and ignited much experimental and theoretical activity devoted to its clarification. These are: (1) new direct lifetime measurements by the Tokyo group which did not confirm the discrepancy [34] (2) new theoretical calculations by Adkins et al. [32] including higher order terms improving the theoretical precision well below experimental errors, however, confirming early theoretical estimates (3) searches for “exotic” decay modes which could explain the lifetime discrepancy at the cost of new physics (violation of basic conservation laws with decays into 1 photon, 2 photons; anomalous rate in 5 photons; millicharged particles; new bosons, ...) (4) exotic suggestions for disappearance mechanisms (mirror worlds, extra dimensions). The experiments for the determination of the decay rate are not trivial; in fact, in order to form positronium one requires matter (electrons) and con- sequently some corrections are needed to get the value of the lifetime in vacuum. Many effects can introduce systematic errors, e.g. the Stark effect, magnetic quenching, positron- ium escaping the formation cavity, time dependence of the pick-off rate (see Section 4.1.2), fast backscattered positrons, etc. It wasn’t until 2003, when the Michigan researchers [33] obtained the newest results that agreement with the theory and consistency with the Tokyo result [34] was finally achieved. The lifetime puzzle seems to be solved, however, the precision in the experimental result is 200 times worse than the theoretical uncertainty.
Para-positronium decays predominantly into two photons and its lifetime is 125 ps. Due to the phase-space and additional α suppression factors, as compared with the singlet 1 (1 S0) state, the decay time of o-Ps is 142 ns with 3 annihilation photons. Because it can be studied for a longer time, the o-Ps decay rate gives an enhancement factor 103 ≃ in sensitivity to an admixture of new interactions, which are not accommodated in the Standard Model [35]. Glashow realized that the o-Ps system provides a sensitive way to search for Mirror Mat- ter [36]. Glashow’s idea is that if a small kinetic mixing between ordinary and mirror photons exists [37], it would mix ordinary and mirror ortho-positronium, leading to maxi- 4 CHAPTER 1. MIRROR MATTER AS DARK MATTER ) -1 s
µ 7.3 Theory 7.25 Vacuum Gas Powder 7.2
7.15 o-Ps Decay Rate ( 7.1
7.05
1965 1970 1975 1980 1985 1990 1995 2000 Year
Figure 1.1: History of the ortho-positronium lifetime measurements and of the calculations. mal ortho-positronium - mirror ortho-positronium oscillations (see Fig. 1.2). Since mirror o-Ps′ decays predominantly into three mirror photons these oscillations would result in o Ps invisible decays in vacuum. Photon-mirror photon kinetic mixing is described by− the interaction→ Lagrangian density
µν ′ L = ǫF Fµν , (1.1) µν ′ where F (Fµν) is the field strength tensor for electromagnetism (mirror electromag- netism). Together with the Higgs-Mirror Higgs quartic coupling λφφ† φ′φ′† , these are the only renormalizable and gauge invariant terms that can be added to the SM Lagrangian. The effect of ordinary photon - Mirror photon kinetic mixing is to give the mirror charged particles a small electric charge [21, 37, 36]. That is, they couple to ordinary photons with charge 2ǫe1. Ortho-positronium is connected via a one-photon annihilation diagram to its mirror version (o-Ps′) [36]. This breaks the degeneracy between o-Ps and o-Ps′ so that the vacuum ′ ′ energy eigenstates are (o-Ps + o-Ps )/√2 and (o-Ps o-Ps )/√2, which are split in energy by − ∆E =2hǫf, (1.2) where f = 8.7 104 MHz is the contribution to the ortho-para splitting from the × < 1Note, that the direct experimental bound on ǫ from searches for ‘milli-charged’ particles is ǫ 10−5 [61]. ∼ 1.1. POSITRONIUM AND DARK MATTER 5
o−Ps
o−Ps’
o−Ps+ o−Ps ∆E o−Ps’ o−Ps−
Figure 1.2: The double degeneracy between ortho-positronium mass eigenstates of ordinary (o-Ps) and mirror (o-Ps′) is broken when a small mixing term is included.
one-photon annihilation diagram involving o-Ps [36]. Assuming a mixing strength of ǫ = 4 10−9 (as suggested by the DAMA results), one obtains an energy splittingof∆E =2.9 ×− × 10 12 eV. Thus, the interaction eigenstates are maximal combinations of mass eigenstates which implies that o-Ps oscillates into o-Ps′ with a probability:
′ P (o-Ps o-Ps ) = sin2 ωt, (1.3) → where ω =2πǫf. ′ The simplest case of o-Ps o-Ps oscillations in vacuum [36] leads to an apparent → increase in the decay rate because the mirror decays are not detected. The number of o-Ps N satisfies − N = cos2 ωt e ΓSM t exp[ t(Γ + ω2t)], (1.4) · ≃ − SM eff where ΓSM is the Standard Model decay rate of o-Ps [38, 39, 33]. Thus Γ ΓSM (1 + 2 ≈ ω /ΓSM ) leads to a branching ratio of:
2(2πǫf)2 Br(o-Ps invisible)= 2 2 . (1.5) → ΓSM + 4(2πǫf) The above calculation is not applicable to an experiment performed with a cavity confining the positronium, because in this case the collision rate is not zero and the loss of coherence due to the collisions must be included in the calculation [40, 41]. Assuming the collision rate is much larger than the decay rate Γ Γ [41], the evolution of the coll ≫ SM number of ortho-positronium states, N, satisfies:
dN Γ N Γ Nρ, (1.6) dt ≃ − SM − coll 6 CHAPTER 1. MIRROR MATTER AS DARK MATTER
′ where the second term is the rate at which o-Ps oscillates into o-Ps (whose subsequent decays are not detected). In this term, ρ denotes the average oscillation probability over the collision time. That is,
t t − ′ ′ ′ − ′ ′ ′ ρ Γ e Γcollt sin2 ωt dt Γ e Γcollt (ωt )2dt , (1.7) ≡ coll ≃ coll Z0 Z0 where we have used the constraint that the oscillation probability is small, i.e. ωt 1. As ≪ long as t 1/Γcoll, then ≫ 2ω2 ρ 2 (1.8) ≃ Γcoll is a reasonable approximation for a cavity experiment. Thus, substituting the above equa- tion into Eq.(1.6), we have
2ω2 2ω2 Γeff Γ + =Γ 1+ . (1.9) ≃ SM Γ SM Γ Γ coll coll SM Therefore, the branching ratio for o Ps invisible confined in a cavity can be written as − →
2ω2 Br(o-Ps invisible)= 2 (1.10) → ΓSM (ΓSM Γcoll +2ω ) and, therefore, the limit on the mixing strength, for an experiment with the o-Ps confined in cavity, can be expressed as:
1 Br − → Γ Γ ǫ = o Ps invisible SM coll . (1.11) 2πf 2(1 Br − → ) s − o Ps invisible ′ Note that the probability P (o-Ps o-Ps ) can also be affected by an additional splitting of o-Ps and o-Ps′ states induced by→ an external electric or magnetic field. This is similar to the phenomenon of n n [42] or muonium to antimuonium oscillations [43] in various environments: − 2 ′ ∆E Br(o-Ps o-Ps )= . (1.12) 2 2 2 → 2(∆E + ∆ +ΓSM ) Where ∆ is the breaking of the degeneracy due to the external fields. Only the second order Stark shift contributes to the positronium in the ground state. The energy shift can be calculated in the same way as for hydrogen and is given by: 1 ∆= α ǫ E2, (1.13) −2 0 0
3 where α0 = πa0 is the polarizability of o-Ps (a0 =0.1 nm is the Bohr radius of positronium), ǫ0 is the vacuum permittivity and E is the electric field. For the electric field of 3-5 kV/cm that we plan to use in the experiment this contribution (∆ = 1.26 10−18 eV) is 6 orders × 1.2. EXPERIMENTAL EVIDENCE FOR MIRROR DARK MATTER 7
of magnitude smaller than the energy splitting of Eq.(1.2) and, thus, its effect on the oscillation probability is negligible. A magnetic field affects only the triplet state with quantum number m=0. For the m= 1 states, the Zeeman effect is zero because the magnetic moments for positron and ± electron are opposite. The energy contribution for m=0 can be calculated with [?, zem]
A~2 A~2 2 ~2 2 2 ∆E = + ( ) + γ1 B0 (1.14) − 2 r 2 25 −1 −2 where A =4.92 10 eV s is the hyperfine splitting constant of the ground state, γ1 = 10 −1×−1 2.81 10 T s the gyromagnetic ratio, B0 the magnetic field. For the magnetic field × − of 100 Gauss applied to guide the positrons, one gets an energy shiftof∆=1.6 10 9eV. This is 3 orders of magnitude bigger than what is expected from Eq.(1.2) and,× therefore, the oscillation probability for m=0 states will be affected. This will result in a suppression of the expected signal that should be corrected by a factor of 1/3.
1.2 Experimental evidence for Mirror Dark Matter
At present, there is some experimental evidence that Mirror Matter could exist, coming from cosmology as well as from the neutrino physics [45]. Foot discussed implications of the DAMA experiment for Mirror Matter-type Dark Matter, which is coupled to ordinary matter through the interaction of Eq.(1.1) [46, 12]. It has been shown that the annual modulation signal (see Fig. 1.3) measured by the DAMA/NaI experiment [6] can be explained by mirror matter-type Dark Matter if the photon-mirror photon mixing strength is in the region of: ǫ 4 10−9. (1.15) ≃ × Those results have been very recently confirmed by the DAMA/LIBRA experiment with a 8.2σ significance [5]. Mirror Matter is a promising candidate that can reconcile the null results from the other, higher threshold experiments like CDMS/Ge [9], CDMS/Si [10] and XENON10 [11]. Interestingly, this value of ǫ is also consistent with all other known experimental and cosmological bounds, including SN1987a2 [48] and the standard Big Bang Nucleosynthesis (BBN) bound [49]. It is also in the range of naturally small ǫ-values motivated by grand unification models [13]. To note is that, to be sensitive to an invisible decay with the ǫ constrain from the DAMA results, one should measure the lifetime rate with a precision of 0.1ppm. This value id 3 order of magnitude smaller than the current precision which is 150 ppm ??. Such an high precision in the lifetime measurements exclude precise QED tests as a way to probe for o-Ps invisible decays. If ǫ is as large as in Eq.(1.15), the branching ratio Br(o-Ps invisible) for the invisible decay of ortho-positronium in vacuum can be found with Eq.(1.10)→ and is of the order: Br(o-Ps invisible) 2 10−7. (1.16) → ≃ × 2The SN1987a limit ǫ < 10−9.5 obtained in Ref. [48] is actually much weaker. For a more detailed discussion of this and other constraints see Ref. [47] 8 CHAPTER 1. MIRROR MATTER AS DARK MATTER
Figure 1.3: The points are the annual modulation observed by DAMA/NaI. The line is the prediction of this modulation for Mirror Matter [12].
For comparison, the BBN limits [49] deduced from the successful prediction of the primor- dial 4He abundance are ǫ< 3 10−8 (1.17) × and Br(o-Ps invisible) < 10−5 (1.18) → respectively. Given the indications for the mirror world coming from Dark Matter [46] and the neutrino physics anomalies [45, 51], as well as the intuitive expectation that nature could be left-right symmetric, it is obviously important to determine experimentally whether ortho-positronium is a window to the mirror world or not. Recently, Barbieri et al. [52] proposed to search for mirror particles at LHC. Chapter 2
Search for the invisible decay of o-Ps in Aereogel
The first experiment to search for invisible decay channels of o-Ps was performed by Atoyan et al. [53]. Their result on Br(o Ps invisible) < 5.3 10−4 (90% C.L.) excluded this channel as a possible explanation− of the→ o-Ps lifetime anomaly× (for a recent review see e.g. [54]). This search was repeated by Mitsui et al. who found an upper limit for the branching ratio Br(o Ps invisible) < 2.8 10−6 (90% C.L.) [55]. Furthermore, they could place a limit on the− existence→ of milli-charged× particles and on the photon mirror-photon mixing. This result was corrected in [40] by taking into account the suppression factor for the mixing due to the presence of matter. Motivated by DAMA results and by a possibility to search for extra-dimensions, we performed an experiment with an improved sensitivity to search for positronium decays into invisible final states. This search was the subject of another PhD thesis [56] in which the detailed descritpion of the experiment can be found. In this chapter, a concise description of the apparatus and the experimental technique are given and the results are presented. This is important because this search served as the basis to design the experiment for the search of o-Ps invisible decays in vacuum (Chapter 3).
2.1 Experimental technique
The experimental signature for a o Ps invisible decay is the apparent disappearance − → of the energy 2me expected in ordinary decays in a hermetic calorimeter surrounding the o-Ps formation target. The readout trigger for the calorimeter is produced by tagging the stopping of a positron in the target with high efficiency. The schematic illustration of the detector setup is shown in Figure 2.1 (the detailed description of the experimental technique and setup can be found in [56]).
Positrons were produced from a 22Na source with an activity of 30 kBq. The 22Na has a half life of 2.6 years and has a Q-value for the nuclear transition to≃21Ne of Q =2.842 MeV.
9 10 CHAPTER 2. SEARCH FOR THE INVISIBLE DECAY OF O-PS IN AEREOGEL
Aerogel Target
XP 2020