Noname manuscript No. (will be inserted by the editor) S.R. Elliott · B.H. LaRoque · V.M. Gehman · M.F. Kidd · M. Chen An Improved Limit on Pauli-Exclusion-Principle Forbidden Atomic Transitions Received: date / Accepted: date 2 1 0 2 Abstract WehaveexaminedtheatomictheorybehindrecentconstraintsontheviolationofthePauli Exclusion Principle derived from experiments that look for x rays emitted from conductors while a n largecurrentispresent.Wealsore-examinetheassumptionsunderlyingsuchexperiments.Weusethe a J results of these studies to assess pilot measurements to develop an improved test of the Principle. We presentanimprovedlimitof 1β2 <2.6×10−39 onthePauliExclusionPrinciple.Thislimitisthebest 9 2 to date for interactions between a system of fermions and a fermion that has not previously interacted ] with that given system. That is, for systems that do not obviously violate the Messiah-Greenberg x symmetrization-postulate selection rule. e - PACS 11.30.-j · 03.65.-w · 32.30.Rj l c u n 1 Introduction [ 2 Pauli’s original idea [46] for the exclusion principle was postulated to explain patterns in the periodic v table. Recently there has been interest in theories that might permit a small violation of the Pauli 8 ExclusionPrinciple(PEP).TheintroductionofRef.[11]providesaguidetotheliteraturewithregard 1 to both theory and experiment. 1 Messiah and Greenberg described a superselection rule regarding the symmetrization postulate 3 . (SP) in 1964 [39] by noting “In summary, for systems with a fixed number of particles, there is a 7 superselection rule between symmetry types which permits one to insert SP in the quantum theory in 0 aconsistentway.HoweverthepostulatedoesnotappearasanecessaryfeatureoftheQMdescriptionof 1 nature.”ThepaperbyAmadoandPrimakoff[2]useddifferentphrasingstating“Evenifsomeprinciple 1 : permittedsmallmixedsymmetrycomponentsinwavefunctionsthatareprimarilyantisymmetric,and v keptthemsmall,thesymmetricworldHamiltonianwouldonlyconnectmixedsymmetrystatestomixed i X symmetry states, just as it connects only antisymmetric states to antisymmetric states”. This argues that electrons or nucleons in higher orbits are forbidden from transitions to lower orbits regardless of r a S.R. Elliott · B.H. LaRoque · V.M. Gehman · M.F. Kidd Physics Division, Los Alamos National Laboratory, Los Alamos, NM 87545 E-mail: [email protected] M. Chen Department of Physics and Life Sciences, Lawrence Livermore National Laboratory, Livermore, California 94550 V.M. Gehman present Address: Physics Division, Lawrence Berkeley National Laboratory,Berkeley, CA 94720 B.H. LaRoque Present Address: Department of Physics, University of California, Santa Barbara, CA 93106 2 thePEPandalthoughsuchstudiesstilltestotherprohibitedprocesses(e.g.electronornucleondecay), they are not explicitly a test of the PEP. Subsequently a number of authors developed models that containedsmallviolationsofthePEP[33,27,28,24,43,44,25,26,17],butstillexperimentaleffortsmust confrontthisconstraint.TheexperimentofRambergandSnow[48]pioneeredatechniqueintendedto avoidtheMessiah-Greenbergsuperselectionrulebyintroducingnewelectronsintoasystem.Thesenew electrons would supposedly not have an established symmetry with respect to the electrons already contained within the system, thus avoiding the constraint. To accomplish this, Ramberg and Snow (RS) ran a high electrical current through a Cu conductor and searched for evidence of x rays emitted by a PEP-forbidden transition during a capture of an electron onto a Cu atom. Theparameter 1β2hasbecomecommonlyusedtodefinetheprobabilityforasymmetriccomponent 2 ofafermionsystemwavefunctioninamixedstateortheprobabilitythatwhenfermionsformastate, itissymmetric.However,ithasbeenshownthatsmallviolationsofthesymmetryprincipleareoutside the context of quantum field theory. The paper by Greenberg [26] gives a succinct summary of the theoreticalsituation.Asaresult,itislikelyanover-simplificationtocomparethisparameterasdeduced from different systems directly and a wide variety of tests of the PEP are warranted. For example, the recent effort by the DAMA group [10,13] resulted in a strong constraint, but it is subject to the Messiah-Greenberg superselection rule. In Table 1 we summarize the previous experimental results. These tests include looking for forbidden transitions in atomic or nuclear systems as well as looking for atoms in Pauli-forbidden states. In Section 2, we discuss the concept of a new fermion and the assumptions underlying the various experimental results. Next in Section 3, we address atomic physics issues related to the capture of electrons in PEP violating processes and the impact on the derivation of limits. In Section 4 we summarize our experimental activities and our results. In particular, we find that Pb offers many advantages over Cu in a RS-style experiment. We describe searches for PEP-violating capture on atoms by electrons from three different origins. Finally we finish with some discussion. 2 The New Electron Conundrum When a fermion initiates an interaction with a system, it will form a total wave-function that is antisymmetric in the interchange of any two identical fermions in the system. However, if the PEP is violated,thenitispossiblethatonsomerareoccasions,theresultingwave-functionmaybesymmetric. Once the symmetry of the system is established, however, the Messiah-Greenberg superselection rule indicates that the transition probability between the two symmetry states is zero. Therefore to avoid this constraint, we must search for processes where a new fermion interacts with a system containing identical fermions and forms a symmetric state. How we define the term new in this context relates to the assumptions under which any given test of the PEP is done. The VIolation of Pauli exclusion principle (VIP) collaboration began a speculative discussion on this topic [9]. Here we expand on that discussion by classifying various levels of newness and state the underlying assumptions. This list is ordered in decreasing confidence that the fermion-system interaction is new. We define: – Type I interactions are between a system of fermions and a fermion that has not previously inter- acted with any other fermions. – Type II interactions are between a system of fermions and a fermion that has not previously interacted with that given system. – Type III interactions are between a system of fermions and a fermion within that given system. In the following paragraphs we expound on these definitions in the context of Table 1. – Type-I Primordial System Formation: Soon after the Big Bang, when the particle content of the Universe was just coming into being, all fermions would be new fermions. If some fraction of them formed symmetric states, they might have survived until the present epoch. The standard of this type of test is a search for anomalous nuclear state 5L˜i [52]. The survival of an anomalous atomic statewithinachemicalenvironmentovercosmictimescalesseemslesslikelyandwedon’tconsider those as examples of this Type. – Type-Ia Recently Created Fermions Interacting with a System: The original experiment along this line was that of Goldhaber and Scharff-Goldhaber [23] where they searched for PEP violating capture of 14C β rays onto Pb atoms. The fundamental point being that the β particles were 3 Table 1 AsummaryofpreviouslimitsonthePauliExclusionPrinciple.A˘indicatesanatomwheretheinner- most shell has 3 electrons instead of 2. A˜ indicates a nucleus with added nucleons in the ground state. The classification by Type is described in the text. e− refers to an electron that is part of a current, e− refers to I f an electron within the Fermi sea of a metal, and e− refers to an electron produced by pair production. pp Process Type Experimental Limit 1β2 limit Reference 2 Atomic Transitions β−+Pb→P˘b Ia 3×10−2 [23] e− +Ge→G˘e Ia 1.4×10−3 This Work pp e−+Cu→C˘u II 1.7×10−26 [48] I e−+Cu→C˘u II 4.5×10−28 [8] I e−+Cu→C˘u II 6.0×10−29 [9] I e−+Pb→P˘b II 1.5×10−27 This Work I e−+Pb→P˘b IIa 2.6×10−39 This Work f I →I˘+x ray III τ >2×1027sec 3×10−44 [49] I →I˘+x ray III τ >4.7×1030sec 6.5×10−46 [13] Nuclear Transitions 12C →12 C˜+γ III τ >6×1027y 1.7×10−44 [38] 12C →12 C˜+γ III τ >4.2×1024y [3] 12C →12 C˜+γ III τ >5.0×1031y 2.2×10−57 [11] 16O→16 O˜+γ III τ >4.6×1026y 2.3×10−57 [51] 12C →12 N˜ +β−+ν¯ IIIa τ >3.1×1024y [3] e 12C →12 N˜ +β−+ν¯ IIIa τ >3.1×1030y [11] e 12C →12 N˜ +β−+ν¯ IIIa τ >0.97×1027sec 6.5×10−34 [35] e 12C →12 B˜+β++ν IIIa τ >2.6×1024y [3] e 12C →12 B˜+β++ν IIIa τ >2.1×1030y 2.1×10−35 [11] e 12C →11 B˜+p III τ >8.9×1029y 7.4×10−60 [11] 23Na→22 N˜e+p III τ >7×1024y 10−54 [12] 127I →126 T˜e+p III τ >9×1024y 10−54 [12] 23Na→22 N˜e+p III τ >5×1026y 2×10−55 [13] 127I →126 T˜e+p III τ >5×1026y 2×10−55 [13] neutron emission from Pb III τ >1.0×1020y [37] 12C →11 C˜+n III τ >3.4×1030y [11] 16O→15 O˜+n III τ >1.0×1020y [37] 16O→15 O˜+n III τ >3.7×1026y [4] 12C →8 B˜e+α III τ >6.1×1023y [4] Na/I →N˜a/I˜→X III τ >1.7×1025y 1.5×10−53 [21] Nuclear Reactions 12C+p→12 C˜+p(cid:48) II dσ(51o)<40fb/sr [40] dΩ 12C+p→9 B˜+α II dσ(51o)<56fb/sr [40] dΩ Anomalous Nuclear Structures 5H˜e/4He I <2×10−15 [41] 5L˜i/6Li, 6Li→6 B˜e+e−+ν¯ III <2.1×10−15 <5×10−33 [41] e Anomalous Atomic Structures B˘e/Be III <9×10−12 [34] 12C˘/12C III <2.5×10−12 [6] 20N˘e/20Ne III <2×10−21 [42] 36A˘r/36Ar III <4×10−17 for H-like ions [42] Search for [1s2s]1 S state in He III =(0.2±5.0)×10−6 [19] antisym 0 Neutrino Statistics 100Mo→100 Re+2β−+2 bosionic ν¯ Ia sin2χ<0.6 [5] e Astrophysics and Cosmology Solar burning and p-p bound state IIa <1.6×10−15 [47] Primordial nucleosynthesis and 5L˜i I 5L˜i/6Li<8×10−18 <2×10−28 [52] Supernova neutrons and anomal. nuclei Ia O˜/O <10−18 <10−17 [7] Neutrino stat. and primordial nucleosyn. I 4He production [20] Thermal evolution of the Universe I <10−15−10−17 [28] 4 electrons that had no previous interaction with the target atoms. That experiment was the best example of this type prior to this work, although the limit is not very restrictive. Such recently- created fermions could also be produced by pair production or nuclear reactions. – Type-II Distant Fermions Brought to Interact with a System: The Ramberg-Snow experiment is the classic example of this approach with the best previous limit by VIP. The current electrons through a conductor were assumed to have no previous contact with the target and therefore the experiment searched for PEP-violating transition x rays from that conductor. This assumption has some subtlety, however. The power supplies used by these experiments use AC current from modern power grids. Hence the electrons that comprise the current originate from the conduction electrons within the target sample and circuitry that joins the sample to the power supply. The electrons are recirculated through the power supply. Therefore one might assume the fraction of these electrons that are within the target to be less new than electrons that were originally part of the other circuitry. Using a battery to produce the electrons might be more in line with this category of new fermions, however it would be difficult to sustain large current for an extended time with that technique. Corinaldesi [18] suggested that the anti-symmetry of half-integral spin particles under exchange is not a kinematic principle but rather the time-dependent consequence of interactions among the particles and a newly formed system may undergo PEP-violating transitions, whose probability decreases in time. Shimony [50] proposed an experiment to test this hypothesis using crossed Ne ion and electron beams. Although this proposal fits our definition of a Type II experiment, the added time-dependence is a twist. – Type-IIa Nearby Fermions Brought to Interact with System: The electrons in the Fermi sea of a conductor will interact with a specific atom in that conductor rarely. Because the time scale for a given electron to interact with a given atom is long, one could argue that each interaction is a legitimatenewtestofthePEP.TheRamberg-Snowstyleexperimentscanallbeanalyzedthisway. AlthoughtheFermiseaelectronsandtheatomicsystemelectronsarebothsystemsthathavebeen established as antisymmetric, the interaction of the two can be assumed to be a new interaction. – Type-III Stable System Transition: The Reines-Sobel experiment, of which the DAMA result is the best to date, defines this model as an system of electrons in an established symmetry state. A search is conducted for a spontaneous PEP-violating transition of one of the electrons. There are also many examples of experiments looking for similar processes in nuclei. All violate the Messiah- Greenberg superselection rule. – Type-IIIaStableSystemTransitionwithParticleTransmutation:Thesearchforβdecaysthatcan only occur if the PEP is violated is an interesting case. Such decays take place in a localized anti- symmetric nuclear system, however the nucleon created in the final PEP-violating state changes charge. This begs the question: Is this a freshly created fermion interacting with a system with which it has had no prior contact? We argue that the answer is no, because since the fermions never leave the nucleus, the PEP test is on-going within a localized system. 3 Atomic Theory The theory of the capture of a free electron onto an atom via a PEP violating process has not been studied well in the literature. To quantitatively describe this process requires an estimate of the prob- ability that it will be captured (P ), a description of the cascade process and transition branching cpt ratios as the electron proceeds toward the ground state, and finally the energies of x rays emitted dur- ing this cascade. The experimental searches for PEP violation in this report all relate to the capture of electrons by an atom. In this section therefore, we examine these atomic physics issues upon which our derived limits depend. 3.1 The Capture Probability When an electron collides with an atom, the probability that it will be captured (P ) was assumed cpt by RS to be greater than 10% [48]. The VIP collaboration used the same estimate so results could be 5 compared.P ,however,likelydependsontheatomicnumberofthetargetatom.Tobetterunderstand cpt the capture cross section, we considered previous calculations of muon capture and direct radiative recombination. The PEP-violating capture of an electron is analogous to the capture of a muon, if the muon mass was that of an electron. Previous estimates of the muon capture cross sections use the approximation that the muon mass is much greater than the electron mass and use classical equations of motion for the muon [30]. These assumptions will not be valid for particles with mass equal to that oftheelectron.Incontrast,directradiativerecombination(DRR)crosssectionscalculatedforelectron capture on ions holds more promise. A modified Kramer’s formula [36] has been shown to effectively reproducetheDRRcrosssection.Theaccuracyofthisapproachhasbeeninvestigatedandverified[53] to low electron energies applicable to the Fermi sea in a metal. To estimate P , we use the modified Kramer’s formula of Ref. [36] and make two assumptions. cpt First, we assume that this formula and its expression for the effective Z is a reasonable approximation for a neutral atom. (Z = 1(Z +Z ), where Z is the ionization state of the atom and is equal eff 2 ion ion to zero for a neutral atom.) Second, we calculate the total cross section by summing over all atomic levels (that is for all n), instead of only summing over open shell levels. This latter point simply states that a PEP-violating transition can be to any of the atomic shells. In our analysis, we search for 2-1 transitions, therefore we calculate the partial cross section by summing over n≥2. The cross section is then given by: σ = (cid:88) 8√π α5 Ze4ff (1) D 3 3n3 K(K+En) n≥2 Z2 where E =(13.6 eV) eff is the binding energy of level n, and K is the incident electron energy. n n2 In Pb (Cu) the Fermi energy is 9.47 eV (7.0 eV) [31], and Eqn. 1 gives 1.2×10−18cm2 (1.9× 10−19cm2). One can compare this to the cross section (σ ) for an interaction between a conduction e electron and an atom. The mean free path (µ) for an current electron in Pb (Cu) is 2.34×10−7 cm (3.91×10−6 cm)andisdeterminedbytheresistivityofthemetalanditsFermienergy.Usingvaluesof µ and the atomic density in the metal from Ref. [31], a cross section can be estimated for Pb (Cu) as σ =1.3×10−16 cm2 (3.0×10−18 cm2). The ratio (σ /σ ) of these two cross sections is an estimate e D e of P with the result that P = 0.009 (0.058) for Pb (Cu). cpt cpt 3.2 The Cascade When an electron is captured, it cascades through the energy levels eventually emitting a K x ray as α it reaches the ground state. Although higher order transitions such as K are possible, it is estimated β that these transitions would have a reduced intensity as is seen in muonic x rays and in x ray emission during electron capture on ions. In the VIP analysis, the Cu K line is not resolved into the K and α α1 K components and the corresponding forbidden lines blend into a lone peak 300 eV lower in energy α2 (see Table 2). However, the VIP analysis did not correct for the possible emission of a forbidden x ray that is the analog of the K . This would be a modest correction to their efficiency. β 3.3 The X-Ray Energies If a new electron makes a Pauli-forbidden transition in an atom, one would expect an x-ray emission similar to the K transition in the host material. However during this process the K shell contains α 2 electrons, unlike a commonplace K transition, and therefore the energy is shifted down somewhat α due to the additional shielding of the nuclear charge. The energies of these transitions were calculated with an estimated accuracy of a few eV and are given for a few key elements in Table 2. These results are based on the Dirac-Hartree-Slater model with Breit interaction and QED corrections. These are relativistic jj configuration average calculations that include relaxation effects by performing separate self-consistent field calculations for initial and final states. The algorithm used to calculate these transition energies was modified to allow 3 electrons in 1s shell. (See Ref. [8] for an independent estimate of the size of this shift for Cu atoms for which the calculation by one of us (M.C.) gives a similar result.) The estimated energy of the Pauli forbidden K transition in Pb is then 71.6 keV and α2 would appear just below the normal 72.8-keV Pb x ray. 6 Table 2 The atomic transitions resulting from violation of the Pauli Exclusion Principle, indicated by the columnlabeledforb..Forreference,theallowedtransitionenergiesarealsoquoted(allow.).EnergiesareineV. Transition Cu Ge Pb forb. allow. forb. allow. forb. allow. 1s - 2p K 7741 8047 9543 9886 73713 74961 3/2 α1 1s - 2p K 7723 8027 9516 9854 71652 72798 1/2 α2 2p - 3s 738 953 8920 3/2 2p - 3d 873 951 1131 1221 12241 12611 1/2 3/2 2p - 3d 856 931 1104 1189 10180 10448 3/2 3/2 2p - 3s 755 981 10981 1/2 2p - 3d 856 931 1104 1190 10276 10550 3/2 5/2 4 The Experiments In this work we investigate improving upon the RS technique by using Pb instead of Cu as the conductor. Pb has a higher resistivity which leads to more electron-atom collisions. It also produces higher-energy x rays, which are less attenuated by self-shielding, and populate spectra in a region of lowerrelativebackground.Finally,theincreasedseparationbetweentheK emissionfromPbandthe α PEP forbidden transition also results in a lower background under the searched-for peak. However, a result of the use of Pb is that the various possible PEP forbidden transitions are well separated in energy and do not blend. Hence the efficiency must be considered for each specific transition within the search. Therecentuseofp-type,point-contactGedetectors(PPC)fordarkmatteranddoublebetadecay searches[1]provideanopportunityforPEP-forbiddentransitionstudies.Thesedetectorshaveamuch lower capacitance than the more commonly used semi-coax design and hence have excellent resolution at low energies even in sizable detectors. As a result, line features due to x-ray emission are well resolved.Thispermitsasearchforx-rayemissionduetoPEP-forbiddentransitions.Inourexperiment we use a PPC built by ORTEC [45] as a prototype detector for the Majorana Demonstrator [22, 29]. This detector is 53.7 mm long and 66.5 mm in diameter. It has a 3-mm diameter contact and a bevel on the edge of the contact end that is a 6 mm by 6 mm right triangle. The dead layer thickness (0.97±0.03 mm) was determined by source studies similar to those described in Ref [16]. In this section, we consider two searches for PEP-violating capture on Pb and one on Ge. The first of the Pb experiments is a RS-style, Type II experiment using Pb instead of Cu. The second is an analysis of the same data but considering all the free electrons in the conductor as the interacting fermions. This is a Type IIa experiment by our nomenclature. Finally, we look at electrons from pair production capturing on Ge atoms; a Type Ia experiment. Each of these three searches is described in turn in subsections below. 4.1 Current Through Lead The result of the VIP experiment is based on the RS concept and provides the best previous limit on Type I or II experiments. The VIP effort improved on the RS limit by using higher currents and lower background [9]. Inourwork,aPbcylinder1.15mmthickwithaninnerdiameterof11.25cmsurroundsthedetector. The length of the Pb cylinder (D) is 8.89 cm. The ends of this Pb cylinder are attached to Cu rings with conductive epoxy and these rings provide electrical contact to the Pb. Figure 1 shows the key aspects of our experimental setup. The detector was surrounded by 5 cm of Cu and 5 cm of Pb as a shield. The experiment was conducted in a basement laboratory at 2260 m with minimal overburden. ThecurrentthroughthePbconductorwas110Aat≈0.5V.Thesystemwascurrentcontrolledand the actual voltage varied a few per cent with temperature. We collected 254 (258) hours of current-on (current-off)data.ThespectrawereacquiredusingORTECNIMelectronicsreadoutusingtheORCA data acquisition software [32] and are shown in Fig. 2. The FWHM of the lines in this energy region is about1.15keVforthedetectorusedinthiswork.Thedetectorresolutionatlowenergiesismoderately sensitive to electronic noise. The width of the Pb x-ray peaks increased by 7% when the current was on. This effect is also seen in the noise wall at low energies, which increases from 750 eV to 1000 eV 7 Fig. 1 A photograph of the experimental setup with much of the shielding removed. The Pb conductor with its Cu contacts is shown surrounding the Ge detector. 100 ) V 80 e K 6 K a1 0 60 a2 . 9 d/ 40 3-1 / 4-1 ( s 20 t n u o 0 C 10 s t n u o 0 C f f O - -10 n O 65 70 75 80 85 90 Energy (keV) Fig. 2 Thespectraofthedataobservedwiththecurrentonandoff(top)andadifferencespectrumbetween datatakenwiththecurrentonandthattakenwiththecurrentoff(bottom).Thetwospectrainthetoppanel are very similar. The thick bar indicates the forbidden-transition region of interest. The four peaks are x rays originating from fluorescence of the Pb. with the current on. This small change in the spectrum does affect our analysis of the PEP forbidden peak as it increases the background in the region of interest and weakens the deduced constraint on 1β2. It also explains the structure in the difference spectrum of Fig. 2. 2 For easier comparison of our Pb data to the VIP Cu data, we sum the K and K lines, but α1 α2 excludeK andaccepttheadditionalefficiencypenalty.Toestimatethefractionofcapturedelectrons β that emit this K x ray, we assume that the forbidden emission spectrum mirrors that of the allowed α2 emission. In the data, ∼70% of the Pb x rays are K with the remaining being 3-1 or 4-1 transitions. α (See Fig. 2.) 8 Table3 Asummaryofthecontributionstothex-raydetectionefficiency.(cid:15) istheproductofthe3efficiency tot factors given in the table. Line Energy (cid:15) (cid:15) (cid:15) (cid:15) x ROI BR tot K 71.6 KeV 0.0072 0.834 0.23 0.0014 α2 K 73.7 keV 0.0072 0.811 0.47 0.0027 α1 Table 4 Asummaryofthedetectionratesinthetwopeaks.Columns4and5givetheratesperhourwiththe current on (R ) and off (R ) respectively. The differential rates (δR) given in the final column include the on off x-raydetectionefficiencies.Notethatanexcessisfoundbetweenthecurrentonandcurrentoffconfigurations. This excess is due to electronic noise originating from the power supply and is discussed in the text. Energy Counts On Counts Off R (/h) R (/h) (δR) (/h) on off K 71.6 keV 12503 12858 49.22± 0.44 49.84±0.44 -443±448 α2 K 73.7 keV 12995 12539 51.16±0.45 48.60±0.43 935±228 α1 weighted Average 652±203 Because the two√PEP violating x rays are near the Pb Kα lines, we chose our regions of interest to minimizetheratio, B/(cid:15) ,whereB isthebackgroundwithinthewindowand(cid:15) istheefficiency ROI ROI factorduetothefractionofthelineshapecontainedwithintheregionofinterest.Inaflatbackground spectrum,asymmetricwindowofwidth2.8σ istheoptimumregion-of-interestchoice.Forourdatawe calculate the optimum window taking into account the location and width of the neighboring peaks, whose tails can increase B if the region of interest is chosen too wide. The deduced region of interest (see Fig. 2) for K is 73.395 - 74.066 keV with an efficiency of 0.811 and for K is 71.295 - 72.002 α1 α2 keVwithanefficiencyof0.834.Therelativeprobabilityofthesetwotransitionsisestimatedtobe(cid:15) BR = 0.47 and 0.23 respectively. The efficiency for detecting x rays from the Pb ((cid:15) ) was determined by x simulation using the MaGe [14] framework developed by the Majorana and GERDA collaborations. Thesimulationwasverifiedbycomparingtosourcemeasurementsusingγ raysfrom241Amand133Ba placed both inside and outside the Pb sheet. The simulation matched the measurements to 10% at all pointsclosetotheROI.For71.6keVxraysemitteduniformlyfromthePbcylinder(cid:15) wasdetermined x to be 0.0072. The contributions to the x ray efficiency are summarized in Table 3. Thenumberofcountsobservedwiththecurrenton(off)aregiveninTable4.Thedifferentialrates (δR) are given by (R -R )/(cid:15) . We then use the weighted average of the δR results for the two on off tot lines in the determination of upper limit on the number of events that could be due to PEP violation. The previous work of RS and VIP used a 3-σ upper limit for the number of excess counts with the current on and for direct comparison we do the same here. It is clear that the positive excess shown in Table 4 is due to the electronic noise with the current on and is not a PEP-violation effect. Therefore, we emphasize here that we base our limit on the 3σ variation from the excess to be as conservative as possible.ExplicitlywecalculatetheupperlimitbasedonN =((652+3×203)/h)(254h)((cid:15) )=2307. 3σ x We have incorporated the difference in live time in the subtraction. We include the factor (cid:15) in this x expression so N can be compared directly to VIP and RS. As a result, this factor also appears 3σ explicitly in Eqn. 2. (cid:80) The number of new electrons introduced into the metal is given by N = (1/e) I∆t, where e new is the electron charge, I is the current passed through the conductor during time ∆t and the sum is overallmeasurementperiods.Thenumberofinteractions(N )byanindividualelectrontransversing int the metal is given by D, where D is the distance through the conductor the electrons travel. For Pb, µ N = 3×107. The 3-sigma upper limit on 1β2 from our work is then given by: int 2 1 N β2 < 3σ =1.5×10−27. (2) 2 N (cid:15) P N new x cpt int Table 5 summarizes the parameters for the 3 experiments of the RS genre. The reference for the latest VIP result [9] does not provide full detail for their preliminary results of the underground work as was done for their above ground studies. The underground runs include nearly a half year of current-on/off data. 9 Table 5 Acomparisonoftheexperimentalparameters.ThevaluesfortheVIPaboveground(AG)workand underground (UG) work are quoted separately. The value for N was not given by VIP (AG) and the value 3σ givenisourestimatebasedontheirlimitfor 1β2.Inthefinalcolumn,thelimitson 1β2 includeourestimates 2 2 ofthenewvaluesoftheRamberg-SnowandVIPlimitsbasedonresultsofourwork.Specifically,weusedour values of P for these results. cpt Project N (cid:15) P N N 1β2 new x cpt int 3σ 2 RS 9.7×1025 0.0029 0.058 6.4×105 300 <2.9×10−26 VIP (AG) 2.2×1026 0.01 0.058 2.3×106 219 <7.7×10−28 VIP (UG) 3.5×1027 0.01 0.058 2.3×106 ∼500 <1.1×10−28 This Work 6.29×1026 0.0072 0.009 3.8×107 2307 <1.5×10−27 Table 6 Free electron analysis of the violation of PEP. As above, we estimated the limit on the number of x-rays detected in VIP-UG to be 500 based on their result. The paper itself did not provide that number directly. Experiment N (/cm3) V (cm3) v (cm/s) Nfree×Nfree N3σ 1β2 e f int new (cid:15)tot 2 VIP-UG 8.41×1022 1.2 1.57×108 1.03×1044 5×104 8.4×10−39 This Work 1.33×1023 36.1 1.83×108 6.88×1045 1.64×105 2.6×10−39 4.2 Free Electrons in a Metal The current through a conductor in a Ramberg-Snow style experiment is comprised of electrons from the circuit itself. Since the electrons originate from the conductor, one should consider whether the current is necessary. One aspect of metal conductors is that there are a large number of free elec- trons unlike insulators such as NaI or semi-conductors like Ge. The interaction of these free electrons with atoms in the metal can avoid the Messiah-Greenberg superselection rule under a specific set of assumptions. A specific free electron in the metal interacts with a specific atom very rarely. The time frame is longenoughthatonemightassumeeachsuchinteractionisanewpossibilitytotestthePEP.Thatis, the electron-atom system does not remember their previous interaction. For a given electron, the time between interactions is 1.3×10−15 s (2.5×10−14 s) for Pb (Cu). With approximately an Avogadro’s number of atoms in a sample, the time between collisions between a given electron and a given atom is tens to hundreds of years. If one analyzes the data ignoring the current and instead considers free electron collisions, a much improved constraint on PEP violation is found. The expression from Eqn. 2 can now be written, 1 N 1 β2 < 3σ (3) 2 (cid:15)tot PcptNnferweeNifnrtee where Nfree and Nfree are given by int new v Nfree =∆t f (4) int µ Nfree =N V new e where N is the free electron density, V is the volume of the sample and v is the Fermi velocity of e f electrons in the metal. The factor µ is the time between electron-atom collisions. vf In Table 6 we have calculated such a limit from our data using the sum of the current off and current on spectra. We added the two spectra, found the total number of counts in the two windows and used the square root of the number of counts as estimate of the 1-σ uncertainty. The results from Table 6 are very much more restrictive than for the Ramberg-Snow approach. AlthoughtheVIP-UGexperimenthasalowerbackgroundandamuchlongerruntime,ourPbsample hasamuchlargervolumeandthetimebetweencollisionsismuchshorterresultinginamorerestrictive limit. 10 Table 7 A summary of the limits from the DEP-SSE and SEP-MSE analyses. Peak Counts Echo Peak 1β2 2 Region Counts (3 σ limit) DEP-SSE 656999 92100 <0.0014 SEP-MSE 957525 236160 <0.0015 keV24000 keV35000 ounts/2202000000 ounts/30000 C C 18000 25000 16000 14000 20000 12000 10000 15000 8000 10000 6000 4000 5000 2000 20080 2085 2090 2095 2100 2105 0 1575 1580 1585 1590 1595 1600 1605 1610 Software Energy [keV] Software Energy [keV] Fig. 3 The left panel shows a spectrum histogram near the single escape peak resulting from exposing our detector to a Th source. The higher curve is the whole data set and the lower curve shows the spectrum after implementing a waveform analysis to eliminate single site energy deposits within the crystal. The line is a fit throughthedatatodeterminepeakstrength.Therightpanelshowsasimilarplotoftheregionnearthedouble escape peak, however the lower curve now excludes multiple site energy deposits. 4.3 Electrons from Pair Production If a γ ray reacts by pair production in the Ge detector, the electron produced is new to any atomic system and may violate the PEP as it slows down. If so it will capture and cascade to the K shell emitting 10.6 keV of energy that will sum with the initial energy deposit. Hence one can search for echos to the double escape (DEP) and single escape (SEP) peaks in the spectrum. We exposed the Ge detector described above for 3 weeks to a Th source and then searched for the peakechosrelatedto the DEP and SEPfrom the 2.6-MeV γ ray from 208Tl. For the DEP, bothof the annihilation γ rays escape the detector and the PEP-violating low-energy x-ray emissions are part of thesinglesiteenergydeposit(SSE).Incontrast,theSEPisbyitsnatureamultiplesiteenergydeposit (MSE). Though counts in the full energy peak (FEP) can result from pair production events in which theannihilationgammaraysdonotescape,mostarisefrommultipleComptonscatters.Therefore,we do not consider the FEP in this analysis. So for each the DEP and SEP energy regions we search for a peak that is 10.6 keV above the pair production features. The spectrum near each of these peaks is shown in Fig. 3 and in neither case is there any indication of any peak echo. In this figure, the SEP (DEP) region of the spectrum is shown before any analysis cuts and after a cut to select for MSE (SSE)depositevents.Becausethelow-energyxraysareemittedinternaltothecrystaltheefficiencyis effectively 100% and the ratio of the events in any peak echo to the pair production peak is a measure of the violation of PEP. Any uncertainty in the efficiency or deviation from 100% would cancel in the ratio. All events were digitized and we could analyze the event waveforms to select out SSE and MSE events using an analysis similar to Ref. [15]. In neither case was any evidence seen and we estimated the upper limit on the existence of such a peak to be the square root of the number of counts in a window that has a width defined by the primary peak width. Table 7 summarizes the results of this Type Ia experiment. 5 Discussion Theseexploratoryexperiments,conductedabovegroundwithadetectorinsideacommercialcryostat, were able to produce competitive limits in 1β2 and one result that is the best current limit for our 2 definedTypeIIinteractions.ThekeyadvantagesofourapparatusbeingtheuseofPbandaP-PCGe