DOI 10.1140/epja/i2013-13064-1

Regular Article Experimental Physics

Eur. Phys. J. A (2013) 49: 64 THE EUROPEANPHYSICAL JOURNAL A

New search for correlated e+e pairs in the decay of 241Am

R. Bernabei1,2,a, P. Belli1,2, F. Cappella3,4, V. Caracciolo5, S. Castellano5, R. Cerulli5, C.J. Dai6, A. dAngelo3,4,A. Di Marco1,2, H.L. He6, A. Incicchitti3,4, M. Laubenstein5, X.H. Ma6, F. Montecchia1,7, X.D. Sheng6, V.I. Tretyak8,R.G. Wang6, and Z.P. Ye6,9

1 INFN, Sezione Roma Tor Vergata, I-00133 Rome, Italy2 Dipartimento di Fisica, Universita` di Roma Tor Vergata, I-00133 Rome, Italy3 INFN, Sezione Roma, I-00185 Rome, Italy4 Dipartimento di Fisica, Universita` di Roma La Sapienza, I-00185 Rome, Italy5 INFN, Laboratori Nazionali del Gran Sasso, I-67100 Assergi (AQ), Italy6 IHEP, Chinese Academy, P.O. Box 918/3, Beijing 100039, China7 Dipartimento di Ingegneria Civile e Ingegneria Informatica, Universita` di Roma Tor Vergata, Italy8 Institute for Nuclear Research, MSP 03680 Kyiv, Ukraine9 University of Jing Gangshan, Jiangxi, China

Received: 15 February 2013 / Revised: 17 April 2013Published online: 29 May 2013 c Societa` Italiana di Fisica / Springer-Verlag 2013Communicated by E. Bellotti

Abstract. A new search for production of correlated e+e pairs in the decay of 241Am has been carriedout deep underground at the Gran Sasso National Laboratory of the INFN by using pairs of NaI(Tl)detectors of the DAMA/LIBRA set-up. The experimental data show an excess of double coincidences ofevents with energy around 511 keV in faced pairs of detectors, which are not explained by known sidereactions. This measured excess gives a relative activity = (4.70 0.63) 109 for the Internal PairProduction (IPP) with respect to the alpha decay of 241Am; this value is of the same order of magnitudeas previous determinations. In a conservative approach the upper limit < 5.5 109 (90% C.L.) can bederived. It is worth noting that this is the rst result on IPP obtained in an underground experiment, andthat the value obtained in the present work is independent of the live-time estimate.

1 Introduction

In the seventies experimental and theoretical investiga-tions of weak interactions in nuclei suggested the pres-ence of various electromagnetic phenomena, which modifythe basic decay processes. In the conventional perturba-tion treatment of the weak interactions these phenomenaare attributed to higher-order terms. These higher-orderterms, accompanying decay and shell-electron capture,are [1,2]: i) internal bremsstrahlung; ii) ionization and ex-citation of the electron cloud; iii) internal pair production(IPP). These higher-order eects are much less intensethan the rst-order processes; for example, in the case ofthe internal bremsstrahlung about one photon is producedper 100 decays, and the internal ionization process onlyoccurs about once per 104 decays. The rst estimate ofthe contribution of the IPP process in the decay wasperformed by Arley and Moller [3] and, independently, byTisza [4]; these studies were rened in the following years,and experimentally tested by Greenberg and Deutsch [5].

a e-mail: rita.bernabei@roma2.infn.it

In 1973 [2] and in the following years, the IPP pro-cess was also investigated in the decay, where this eectcould give additional information on the higher-order phe-nomena accompanying nuclear transformations. In ref. [2],a theoretical estimate of the e+e production due tobremsstrahlung emitted by particles in the 241Am de-cay was given: = Ae+eA = 1.2 109, where Ae+e andA are the e+e and activities, respectively. This esti-mate was in reasonable agreement with the experimentalresult, obtained in the same work by investigating dou-ble coincidences of the two 511 keV quanta produced bythe positron annihilation; the set-up did not allow the de-tection of the emitted e+e pairs. In particular, a sourceof 241AmO2 (deposited on top of Pt backing and coveredwith thin Au layer) was mounted between Ge(Li) 40 cm3and NaI(Tl) 80 cm3 detectors. The mechanical support ofthe source was made of a stainless steel backing (withoutdirect access of s to the stainless steel); Pb and V foilswere used to stop the emitted particles. The observationof two 511 keV quanta in coincidence between the Ge(Li)and NaI(Tl) gave for the probability of the e+e emissionthe value: = (3.1 0.6) 109. The authors excluded

Page 2 of 10 Eur. Phys. J. A (2013) 49: 64

Table 1. Experimental results and theoretical estimates of for the IPP process in decays. The superscript(a) identies the value derived from ref. [16] when adopting the procedure described in the text.

Source Experiment Theory

(109) Detectors Year Ref. (109) Year Ref.210Po 5.3 1.7 NaI(Tl)+Ge(Li) 1986 [10] 4.4 1978 [6]239Pu 7 9 NaI(Tl)+Ge(Li) 1986 [10] 2.2 1978 [6]241Am 3.1 0.6 NaI(Tl)+Ge(Li) 1973 [2] 1.2 1973 [2]

2.15 0.25 NaI(Tl)+Ge(Li) 1986 [10] 2.3 1978 [6]1.8 0.7(a) Plastics+Ge 1990 [16]4.70 0.63 NaI(Tl) pairs 2013 This work

(or took into account) various background processes whichcould imitate the observation, such as: i) e+e pair cre-ation by cosmic rays and high-energy s from natural ra-dioactivity; ii) high-energy external bremsstrahlung froms; iii) (,n), (,p) reactions on Pt, Pb, V, O; iv) excita-tion of high-energy nuclear levels in (, ) scattering; v)presence of 65Zn and other long-lived + emitters; vi) de-cay of 241Am to high energy levels of 237Np.

In 1978 the theoretical value of the probability of theprocess in 241Am was re-estimated to be [6]: 2.3 109.Further theoretical studies have also been carried out; infact, while the theory of the IPP in the decay seemedto be clear, in the decay it required more attention [2,69]. Essentially two modes have been suggested for theIPP process in decay: i) the particles are acceleratedin the Coulomb eld of the daughter nucleus and emitbremsstrahlung s, with energy above 2mec2, giving riseto e+e pairs; ii) the daughter nucleus originates in anexcited state, and the resulting quantum (real or virtual)is converted to a e+e pair. Both processes are possible,as pointed out in ref. [9]; the process i) proposed in refs. [2,6] gives a theoretical estimate1 109 (as mentionedbefore), whereas the process ii) discussed in refs. [7,8] gives 1012. Therefore, according to ref. [9] the dominantcontribution to the IPP in the decay of 241Am arisesfrom the mode i).

In 1986 a new dedicated experiment [10] was per-formed in order to investigate the result of ref. [2]. Inthis experiment, three emitters were studied: 210Po,239Pu, 241Am. Pb and Cu absorbers were used to stop and + particles; the 511 keV double coincidences be-tween a Ge(Li) 70 cm3 and a NaI(Tl) 785 cm3 detectors,placed in a low-background shield, were observed. For allthe sources, the process was observed with probabilities:(210Po) = (5.3 1.7) 109, (239Pu) = (7 9) 109,(241Am) = (2.15 0.25) 109, respectively. Pollutionof the 241Am source by 152Eu and 154Eu, and pollutionof the 239Pu by 241Am, 234mPa, 214Bi, 208Tl, 137Cs weremeasured, while in the 210Po source radioactive impuritieswere not identied. Moreover, any decay of 241Am to

1 For completeness we recall that in [7,8] the value for thisprocess was estimated to be 1020, but [9] has shown thatthis value was incorrect because it did not take into accountthe full transition energy of the particles.

the high-energy (> 1022 keV) excited states of 237Np wasexcluded (intensity < 109). The obtained results werein agreement with the rened theoretical expectations ofref. [6] (see table 1), and in particular the previous positiveresult [2] on 241Am was conrmed.

In the late eighties, some anomalous phenomena wereobserved in collision of heavy ions at GSI and in collision ofheavy ions with e+ [1114], and it was suggested the possi-ble existence of an hypothetical neutral X particle, whosedecay in e+e pair may give rise to the observed peaks inthe measured e+ and e spectra. In particular, in ref. [15]it was proposed a new model where the pair productionwould be due to a QED strong coupling phase: heavy nu-clei have a surface region with a strong static electric eldthat can be perturbed by the passage of charged particles(e.g., and ). The perturbation causes a uctuation inthe electric eld, that would produce a strong non-staticelectromagnetic eld which may be source of strong cou-pling vacuum; the vacuum bubbles can decay in positron-ium states, which are e+e bound states. In ref. [15] someexperimental tests were suggested for this model; one ofthem is the search for double coincidence events producedby positron annihilation. Thus, in this scenario the pro-duction of positronium states in decay of heavy nucleimay be a concurrent process with respect to the IPP one.However, since the IPP process is better stated, in thefollowing we will mainly refer the results to it.

In 1990 an experiment was performed [16] with theaim to test the model of ref. [15]. Two thin plastic scin-tillators registered e+e pairs emitted from an 241Amsource, and two Ge detectors of 198 cm3 each registeredtwo 511 keV quanta. The 241Am source was sandwichedby a pair of 10m Ti foils followed by two aluminized My-lar foils of 30m; two plastic scintillators were placed afterthem. During 541.7 h live time, the coicidences between 511 keV events in the two Ge detectors were 466, most ofthem were explained as due to background from + de-cay of 26mAl produced by 23Na(,n)26mAl from sodiumcontamination; the presence of sodium in the vicinity of241Am source was also conrmed by the observation ofthe lines at 1809 and 1130 keV emitted by the 26Mgproduced in the process: 23Na(,p)26Mg. When requir-ing energy releases in two plastics: E1 and E2 greater than30 keV, just 6 coincidences survive. Finally, only one eventremains when the last requirement E1 E2 as expected

Eur. Phys. J. A (2013) 49: 64 Page 3 of 10

by the model of ref. [15] is adopted; hence, an upperlimit on the probability of the production of a hypotheti-cal neutral particle with mass above 1.4MeV/c2 decayingin e+e, was set: < 1.5109 (95% condence level, C.L.).The previous experimental results regarding the IPP pro-cess [2,10] (and also the theoretical estimates) were missedin this work. However, in case the IPP process is insteadconsidered, the requirement E1 E2 should be released,and thus the useful number of 511 keV double coincidencesfrom ref. [16] is 6 events; considering this fact, one can de-rive a value: (1.80.7)109, compatible with previousobservations. It should be, however, noted that since theeciency in [10] was calculated by simulating the X decaywith mass of mX > 1.4MeV the derived value of hasto be considered approximate.

Thus, considering the present situation (see table 1), anew search for e+e pairs in the 241Am decay has beencarried out deep underground. The obtained results aredescribed in the following.

2 The experimental set-up and the datataking conditions

The results presented in the following have been obtainedthrough dedicated measurements with some pairs of theNaI(Tl) detectors of the DAMA/LIBRA set-up [1722],whose description, radiopurity and main features are dis-cussed in details in refs. [17,19,22]. We just recall thatthe sensitive part of this set-up is made of 25 highly ra-diopure NaI(Tl) crystal scintillators (5 rows by 5 columnsmatrix); each NaI(Tl) detector has 9.70 kg mass and asize of (10.2 10.2 25.4) cm3. The bare crystals are en-veloped in Tetratec-Teon foils and encapsulated in ra-diopure OFHC Cu housings. In each detector two 10 cmlong special quartz light guides also act as optical windowson the two end faces of the crystal and are coupled to twolow-background photomultipliers working in coincidenceat single photoelectron threshold. The data taking consid-ered here has been performed in the new DAMA/LIBRAconguration after the upgrading occurred in fall 2010when all the PMTs were replaced by new ones withhigher quantum eciency, specially developed by HAMA-MATSU; details can be found in ref. [22]. The light re-sponse in this experimental conguration is typically 610 ph.e./keV depending on the detector. The detectorsare housed in a sealed low-radioactivity copper box in-stalled in the center of a low-radioactivity Cu/Pb/Cd-foils/Polyethylene/Paran shield; moreover, about 1mconcrete (made from the Gran Sasso rock material) almostfully surrounds (mostly outside the barrack) this passiveshield, acting as a further neutron moderator. The copperbox is maintained in HP nitrogen atmosphere in slightlyoverpressure with respect to the external environment; itis part of the threefold-level sealing system which excludesthe detectors from environmental air. For more details see,e.g., refs. [17,19,22].

In the particular measurements dedicated to thepresent study an 241Am source was placed in the mid-dle of each used NaI(Tl) pair for a total of 6 sources used.

Table 2. Experimental results of the rst running period:i) D1-D2 identies the NaI(Tl) detectors in each pair accord-ing to the numbering given in ref. [18]; ii) A is the activity ofthe used 241Am source at time of measurements (error given bythe producer: 3.5%); iii) n1 is the number of the double coin-cidences measured in the (465557) keV energy window during1.29 d of live time in presence of the 241Am sources; iv) n2is the number of double coincidences measured in the sameconditions during 24.6 d live time without sources; v) s is theexcess rate of double coincidences derived according to the sta-tistical procedure given in the appendix. The upper limits areat 90% C.L. See text.

D1-D2 A n1 n2 s

(kBq) (T1 = 1.29 d) (T2 = 24.6 d) (counts/d)

1-2 34.8 5 23 2.9+2.01.56-7 40.8 14 26 9.8+3.22.79-10 31.7 10 25 6.7+2.72.211-12 34.6 1 28 < 2.4

16-17 29.5 5 22 3.0+2.01.519-20 29.4 10 17 7.0+2.72.2Total 200.8 45 141 29.2 5.2

The sources have been produced by CEA-DAMRI (nowa-days called CERCA-LEA) and are identied as AM241-EGSA15. These sources are made of a thin circular layerof active AmO2 ( 3mm) heat sealed by two thin Mylarfoils. They have been stored deep underground for 10 y;their present mean activity is 35 kBq.

The detailed decay scheme of 241Am can be foundin ref. [23]. The energy spectrum measured by theDAMA/LIBRA detectors, when irradiated by the 241Amsources, has been accurately reproduced by a suitableMonte Carlo simulation, which takes into account the de-cay scheme of 241Am [23] and the activities of the sources.It is worth noting that about 99.91% of the 241Am totalrate is due to events below 90 keV. Thus, in order to reducethe acquisition dead time in our measurements the energythreshold for each detector has been set at 90 keV toexclude most of the events from the 241Am which are outof the region of interest for the present investigation. Theaccuracy of the Monte Carlo simulation allows us to becondent with the standard procedures used for the deter-mination of the data acquisition live time. The electronicchain was modied with respect to the usual one [17] byexcluding the 500s blocking time and by using eithertwelve detectors (six pairs) in the rst running period orsix detectors (three pairs) in the second one. An improve-ment of a factor 2 has been achieved as regards the ratioof the live time over elapsed one in the second dedicatedrun, where only three pairs have been used (those withthe lower number of coincidences in the rst run).

In both running periods each pair has the detectors ofthe other used pairs as anticoincidences; this implies twodierent coincidence backgrounds for a same pair in thetwo dierent running periods (see tables 2 and 3). Finally,the energies of the acquired events were read out from theADCs (see, e.g., ref. [17]).

Page 4 of 10 Eur. Phys. J. A (2013) 49: 64

3 Experimental results

The experimental procedure is based on the detection ofthe two s from the positron annihilation through thestudy of double coincidences in pairs of faced highly ra-diopure NaI(Tl) detectors.

3.1 The rst dedicated run

As mentioned above, for practical reasons in this data tak-ing six pairs of faced highly radiopure NaI(Tl) detectors(i.e. 12 of the 25 available) of the DAMA/LIBRA set-uphave been used. The activity, A, of each used 241Am sourceis given in table 2; the total live time was 1.29 d.

In the data analysis an energy window (465557) keV practically 2 around 511 keV has beenconsidered for the coincidences for both the detectors ofa pair. Since no decay of 241Am to an excited level of237Np with energy larger than 0.8MeV has been observed,to our knowledge and in the decay scheme of 241Am [23]2,a contribution of 241Am decay in the applied coincidenceenergy windows is not expected. Hence, a possible excesswith respect to the data obtained when running withoutsources is due to the sources but not directly to the known241Am decay scheme [23].

The obtained results are summarized in table 2, wherethe number of the double coincidences (n1) measured inthe (465557) keV energy window during 1.29 d of livetime (T1) in the presence of the 241Am sources are com-pared with those (n2) recorded in the same energy windowduring 24.6 d live time (T2) without any source.

Both the estimate of the coincidence excess rate in theconsidered energy window, s, and the limit (90% C.L.)value have been calculated according to the statisticalprocedure described in the appendix. The obtained s val-ues are compatible; therefore, a cumulative analysis hasbeen performed (last row of table 2). The cumulativedata give the result: s = (29.2 5.2) counts/d, that is(4.9 0.9) counts/d/pair.

3.2 The second dedicated run

The second dedicated run has been performed using thethree detectors pairs/sources which had, in the rst dedi-cated run, the lowest values of s; these pairs are: 1-2, 11-12and 16-17 (see table 2). The experimental live time in thissecond dedicated run was 2.63 d.

The results are summarized in table 3. Also in thiscase, the obtained s values are well compatible; thus, thecumulative analysis is performed obtaining: s = (15.7 2.7) counts/d, that is (5.2 0.9) counts/d/pair.

2 One should note that in the list of 241Am quanta [24]there are unplaced s with low intensity and higher energies;however, even the most energetic one, having E = 1014.5 keV(intensity 6.4 1010), has not enough energy to create e+epair.

Table 3. Experimental results of the second running period.See the caption of table 1 for a description of the given quan-tities. Here the data taking live time with 241Am sources is2.63 d (T1), and without them is 24.6 d (T2). See text.

D1-D2 A n1 n2 s

(kBq) (T1 = 2.63 d) (T2 = 24.6 d) (counts/d)

1-2 34.8 19 29 6.0+1.81.6

11-12 34.6 15 33 4.3+1.61.4

16-17 29.5 17 29 5.3+1.71.5

Total 98.9 51 91 15.7 2.7

3.3 The cumulative results

The excess rates per pair in the region of interest duringthe two running periods are compatible. Thus, the com-patibility and the independence between the rst and thesecond dedicated runs allow us to perform a combinedanalysis of the two data sets.

In g. 1a) the coincidence scatter plot E2 vs E1 of theconsidered pairs of faced NaI(Tl) detectors is shown forthe cumulative data collected in the presence of the 241Amsources during 3.92 d live time. The red box denes the re-gion with E1 and E2 in the energy window (465557) keV(2 around 511 keV); the spot at 511511 keV can berecognized. This excess is not present in the data mea-sured in absence of the sources, g. 1b), and in the modelof the 241Am decay according to the scheme of ref. [23],g. 1c). Most of the other coincidence events outside thered box of g. 1a) are instead well described by the scat-ter plots in gs. 1b) and 1c). Finally, g. 1d) shows themodel scatter plot for the case of positrons annihilationswithin the sources; the 27% of the positrons annihilationsgive events that are contained in the red box. The scatterplots in g. 1b), c) and d) are suitably normalized (see therespective captions).

The analysis of the double coincidences allows us topoint out the presence of 243Am (alpha decaying with half-life 7370 y) and of 154Eu (mainly beta decaying with half-life 8.601 y) isotopes in the sources. Their presence can bejustied by side reactions in the production proceduresof the 241Am sources. In fact: i) the 241Am is producedbombarding 239Pu nuclides with neutrons; ii) the 243Amis produced by neutron captures on 241Am; iii) the 154Euis produced by induced ssions of 239Pu. Moreover, fromthe study of the experimental cumulative energy spectrumwith multiplicity 1 shown in the following, the pres-ence of 233Pa isotope can be pointed out; it belongs to the241Am chain and it is the only daughter of 241Am able toproduce a not negligible contribution of s above 100 keV.

The 154Eu can be mainly identied by the detection ofa of 123 keV in coincidence with s of 1005 keV (this oc-curs in the 8.2% of the 154Eu decays), 1274 keV (15.8%),1596 keV (0.8%). Hence, selecting a slice in the doublecoincidence scatter plot of g. 1a) with E1 in the 100

Eur. Phys. J. A (2013) 49: 64 Page 5 of 10

0

200

400

600

800

0 200 400 600 8000

200

400

600

800

0 200 400 600 800

0

200

400

600

800

0 200 400 600 800

a) E1(keV)

E2(

keV

)

b) E1(keV)E

2(ke

V)

c) E1(keV)

E2(

keV

)

d) E1(keV)

E2(

keV

)

0

200

400

600

800

0 200 400 600 800

Fig. 1. (Color online) Coincidence scatter plots of the used pairs of faced NaI(Tl) detectors: a) experimental scatter plotmeasured during the cumulative 3.92 d live time in the presence of the 241Am sources; b) experimental scatter plot withoutsources (the number of displayed events is normalized to the same time as the case a)); c) model scatter plot of the 241Amdecay [23] obtained by simulation (the number of displayed events is normalized to the same time as the case a) and applyingthe sources activities given in tables 2 and 3); d) model scatter plot of the positrons annihilations within the sources obtainedby simulation (the number of events in the red box is 100 times that of a)). The red boxes dene the region with E1 and E2 inthe energy window (465557) keV (2 around 511 keV); the spot at 511511 keV can be recognized in a), while it is absent inb) and c). See text.

150 keV energy window, the spectrum of E2 is producedand reported in g. 2a). The best t3, which includes thecontributions of the set-up and of the 154Eu decay model(also see g. 2b)), provides (1.700.08)104 154Eu decays.This corresponds to an average activity of 154Eu equal to(12.6 0.6)mBq in each source; thus, the expected num-ber of coincidences with E1 and E2 in the energy window(465557) keV (2 around 511 keV) during 3.92 d livetime is (8.3 0.4) events, rather low compared with the96 measured events (see tables 2 and 3).

3 The contribution from 241Am decay is negligible in thisenergy window.

As concerns the 243Am contribution, it decays into239Np, which decays with half-life 2.36 d in the long-living isotope 239Pu. The model scatter plots of the 239Npand of the 233Pa decays are depicted in g. 3. The formercan explain the structures at 110 keV in coincidencewith 220 and 280 keV, present in the experimentaldata shown in g. 1a). From the decay schemes of thesetwo isotopes, one can see that no coincidence is expectedin the region of interest for the present study.

In order to estimate the amount of the aforementionednuclides, g. 4 shows the cumulative experimental energyspectrum (black histogram) with multiplicity 1 mea-sured during 3.92 d live time, after the subtraction of the

Page 6 of 10 Eur. Phys. J. A (2013) 49: 64

a) E(keV)

Cou

nts/

20 k

eV

0

5

10

15

20

25

30

35

750 1000 1250 1500 1750b) E1(keV)

E2(

keV

)

0

500

1000

1500

0 500 1000 1500

Fig. 2. (Color online) a) The distribution of E2, requiring that E1 is in the 100150 keV energy window (see g. 1a)), is shownfor the cumulative 3.92 d live time in the presence of 241Am sources (solid black histogram) and for 24.6 d without the sources,suitably normalized (dotted histogram blue). The presence of the 154Eu peaks is clearly evident. The dashed curve (red) isthe result of the best t, that includes the contributions of the set-up (dotted histogram blue) and of the 154Eu decay model.The contribution from 241Am decay is negligible in this energy window. b) model scatter plot of the 154Eu decay obtained bythe simulation. The number of plotted events is arbitrary. The red box denes the region with E1 and E2 in the energy window(465557) keV (2 around 511 keV).

a) E1(keV)

E2(

keV

)

0

200

400

600

800

0 200 400 600 800b) E1(keV)

E2(

keV

)

0

200

400

600

800

0 200 400 600 800

Fig. 3. (Color online) Model scatter plots of the 239Np (daughter of 243Am) decay (a) and of the 233Pa (in the 241Am radioactivechain) decay (b), obtained by the simulation. The red boxes dene the region with E1 and E2 in the energy window (465557) keV(2 around 511 keV). The numbers of plotted events are arbitrary.

contribution of the set-up derived from the measurementswithout the 241Am sources. This energy spectrum is t-ted in the energy interval (195770) keV by the modelgiven by the sum of the mentioned components: 241Am,239Np, 233Pa, and 154Eu. The latter contribution is xedto 1.70 104 154Eu decays, as previously estimated. Forcompleteness we note that the contribution of the 511 keV

events, observed in the coincidence scatter plot of g. 1a),is negligible and is not included in the t. The result ofthe t is shown in g. 4 together with the four mentionedcomponents. A slight excess remains around 900 keV; itmay be ascribed, for example, to a possible trace contam-ination (of order of 3.4102 ppt) of 234mPa belonging tothe 238U chain.

Eur. Phys. J. A (2013) 49: 64 Page 7 of 10

233Pa

239Np

154Eu

241Am

E(keV)

Cou

nts/5

keV

10

10 2

10 3

10 4

10 5

250 500 750 1000 1250

Fig. 4. (Color online) Experimental cumulative energy spec-trum (black histogram) with multiplicity 1, measured dur-ing 3.92 d live time, after the subtraction of the contributionof the set-up derived from the 24.6 d measurements withoutthe sources. The simulated contributions of 241Am (green his-togram), 239Np (blue histogram), and 233Pa (purple histogram)are reported normalizing their numbers of decays to their bestt values. The simulated contribution of 154Eu (light-blue his-togram) is depicted for 1.70 104 154Eu decays, see text. Thered histogram is the sum of the four contributions. A slightexcess remains around 900 keV; see text.

The best-t values for the total number of decaysof 241Am, 239Np, and 233Pa are (4.717 0.011) 1010,(1.228 0.007) 106, and (2.81 0.04) 105, respec-tively; thus the average activity of 241Am in each sourceis (34.88 0.08) kBq, compatible with the nominal ac-tivities of the sources4 reported in tables 2 and 3. Sincethe 243Am-239Np system is at equilibrium, the estimatedaverage activities of 243Am and 239Np in each sourceare both equal to (908 5)mBq; this corresponds to(4433) ppm of 243Am contamination in the sources. The233Pa has instead an average activity in each source equalto (208 3)mBq; this value allows us to estimate the ageof the sources, t, from their production. In fact, assum-ing that all the sources were produced at the same timeand only 241Am isotopes (N0 nuclides) were present, atthe present time one can write: N241Am = N0et/ andN237Np = N0(1 et/ ), where: i) N241Am is the presentnumber of 241Am nuclides, given by the best t valueof the average activity multiplied by ; ii) N237Np is the

4 A small discrepancy ( 5%) between measured and nomi-nal values of the activity can be explained taking into accountthe error given by the producer (3.5%) and the uncertaintyon the estimated live time (few percent). Anyway, the resultsin the following are largely independent of the live time (seelater).

E(keV)C

ount

s/25

keV

0

10

20

30

40

50

60

70

80

500 750 1000 1250

Fig. 5. (Color online) Cumulative energy distributions mea-sured by one detector of the pair when a coincident event withenergy in the window (465557) keV (2 around 511 keV) iscollected in the faced detector; both the detectors of the pairand all the used pairs contribute to the histograms. Solid his-togram: experimental data in the presence of the 241Am sourcesfor the cumulative 3.92 d live time. Dotted histogram (blue):model of the expected background obtained by summing thecontributions of the 24.6 d data without sources and of the154Eu decays, both suitably normalized. The presence of the511 keV peak in the source data is well evident. The red dashedcurve is the result of the best t (in the 300750 keV region),that also includes a model of background and the contributionof the positrons annihilations inside the sources. The contribu-tion from 241Am decay is negligible in this energy window. Seetext.

present number of 237Np daughters, given by its activity(at equilibrium with 233Pa) multiplied by the mean life of237Np; iii) = 623.5 y is the mean life of 241Am [24]. Byusing these data, the age of the sources can be estimatedas: t = ln(1+

N237NpN241Am

) = (18.10.3) y, well in agreementwith the specications of the sources, also considering theproduction time of the sources material.

The excess of double coincidence events with energyaround 511 keV in faced pairs of detectors seen in g. 1a)can be analyzed to quantify the eect searched for. Forthis purpose, g. 5 shows the cumulative energy distribu-tion (solid histogram) measured by one detector of the pairwhen a coincident event with energy in the window (465557) keV (2 around 511 keV) is collected in the faceddetector; obviously, both the detectors of the pair and allthe used pairs contribute to the histogram. As mentioned,the cumulative live time in presence of 241Am sources is3.92 d. In the same plot, the dotted histogram representsthe expected background obtained by summing the con-tributions of the set-up (derived applying the same dataselection to the data acquired in absence of the 241Amsources) and of the 154Eu decays. The presence of a peak

Page 8 of 10 Eur. Phys. J. A (2013) 49: 64

at 511 keV is clearly evident. Thus, a t of the spectrumhas been performed in the region 300750 keV by includ-ing5: i) the measured coincidences rate of the set-up inthe absence of the sources; ii) the determined coincidencesrate of 154Eu decays; iii) the known contribution fromthe 241Am decay, which is however negligible, as the de-cay scheme is known [23]; iv) a model of the backgrounddescribed by a straight line (this includes other possiblecontributions in the region of interest due to the sources,as, e.g., possible 234mPa); v) the simulated spectrum dueto positrons annihilations within the sources. Thus theonly free parameters are the straight line and the num-ber of positrons annihilations within the sources. The t(2/d.o.f. = 3.7/14 = 0.26, where d.o.f. is the number ofdegrees of freedom) conrms the excess of double coinci-dences from 511 keV -rays due to positrons annihilations.The obtained number of positrons generated within thesources is: N = (222 30).

Considering the N positrons generated within thesources as related to IPP process, the value can be de-rived according to the relation

=Ae+e

A=

N

N, (1)

where N is the total number of 241Am decays dur-ing the running periods, whose best t value is N =(4.717 0.011) 1010 as previously estimated; it is worthnoting that the value is independent of the precise eval-uation of the live time of the data taking. Hence, the bestestimate of is (4.700.63)109, substantially compati-ble with the previous observations and with the theoreticalexpectations (see table 1). The coincidence excess in theenergy region of interest is at 7.4 C.L.

With the obtained estimate N = (22230) counts, wecould derive also the cautious limit for the IPP process < 5.5 109 (90% C.L.).

In the following, we have performed an investigation onthe possible processes, other than e+e pair productionin 241Am decay, which totally or partially mightaccount for the observed coincidence events.

4 Investigation on possible coincidencebackground from the sources

Let us summarize some relevant arguments:

Any possible contribution to double coincidences fromthe used detectors (as, e.g., from high-energy s in-teractions) has been taken into account in the adoptedprocedure (see above).

In the decay chain 241Am . . . 209Bi, + emittersare absent [24].

In the decay chain 241Am . . . 209Bi, the con-tribution of high-energy s (with energies larger than1.022MeV, that is able to produce e+e pairs) is neg-ligible. In particular, from the best t values of the

5 The 239Np and 233Pa decays give no contribution in thisbidimensional energy window, as mentioned before.

241Am and 233Pa activities given above, one can ob-tain for the daughters activities: 208mBq for 237Np,8.1Bq for 233U and 4.6 nBq for 229Th and its daugh-ters. Taking into account these activities and safelyassuming that each is converted in a e+e pair,the contribution to from high-energy s in the241Am decay chain can be estimated as in the fol-lowing: i) 1110 keV in 233U decay [25] can provide < 2 1018; ii) lines at 1046, 1100, 1119 and1328 keV in 213Bi decay [26] can provide < 21017, < 3 1016, < 7 1017 and < 5 1019,respectively; iii) lines at 1239 and 1567 keV in 209Tldecay [27] can provide < 11017 and < 31015,respectively.

No decay of 241Am to an excited level of 237Np withenergy larger than 0.8MeV has been observed to ourknowledge and in the decay scheme of 241Am [23]6.Thus, no direct contribution from the 241Am decay

scheme [23] to double coincidences in the energy windowsof interest is expected.

Remaining possible side processes are discussed in thefollowing.

4.1 Possible -induced reactions

We recall that the used sources are made of a thin layer ofactive AmO2 ( 3mm) heat sealed by two thin Mylarfoils, and have been stored underground for 10 y.

Thus, the elements that can give rise to (,n) or to(,p) reactions are the stable or long-lived isotopes of C,O (that are present in the source materials) and, if par-ticles may survive to the Mylar foils, also N (gas surround-ing the sources when they are inserted in the set-up) andCu (housing of the NaI(Tl) detectors).

Considering the energy of the particles emitted bythe 241Am, the possible (,n) reactions are: 13C(,n)16O,17O(,n)20Ne, 18O(,n)21Ne; none of the produced iso-topes is a + emitter. On the other hand, possible (,n)reactions on Ne produced by the O isotopes (,n) reac-tions: 20Ne(,n)23Mg, can give + emission but the en-ergy threshold is 8.7MeV, that is higher than the energyof the particles emitted by 241Am.

Moreover, considering the energy threshold, theonly possible (,p) reactions are: 14N(,p)17O and15N(,p)18O, both giving stable isotopes of O, and63Cu(,p)66Zn and 65Cu(,p)68Zn both giving stable iso-topes of Zn.

A further source of background could be due to e+epairs produced by photons from (,n) and (, ) reac-tions on the stable nuclei of C and O; considering themeasured cross sections [28,29], a cautious upper limit onthis contribution can be derived: (,n),(,)e+e 1011.

In conclusion, possible signicant contribution of +emitters and of high-energy s from interactions onknown materials can be excluded; a similar study wasperformed in ref. [2] where such possibilities were also ex-cluded.

6 Upper limit on intensity of such decay was obtained inref. [10] as < 109.

Eur. Phys. J. A (2013) 49: 64 Page 9 of 10

4.2 241Am ssion fragments and possible clusterdecays

The spontaneous ssion of 241Am has been measured inseveral works (see ref. [30] and reference therein) and therecommended lifetime is: f = (1.7 0.4) 1014 y. Con-sidering that the spontaneous ssion activity is given by:N241Am/f , and that not all the ssion fragments give riseto e+e emission, an upper limit: fissione+e 3.6 1012can be derived. Thus, the spontaneous ssion can be ex-cluded as source of the observed coincidence excess.

Another possibility may be the induced ssion. Fromexperimental studies [31,32] the more probable fragmentsmasses in case of 241Am ssion induced by thermal neu-trons (the one with higher cross section) are about 135and 105. Various nuclei with an excess of neutrons can becreated with a cumulative yield up to 7%, high Q values(up to 8MeV) and small T1/2s [33]. In their decay, deexci-tations s with high energies can be converted into e+epairs and imitate the observed eect. Thus, let us esti-mate the probability to have a ssion from neutrons in theexperimental conditions. The environmental ux of ther-mal neutrons inside the used multi-ton multi-componentshield7 has been measured to be: < 1.2 107 cm2s1(90% C.L.) [17]; the maximum cross-section for neutronsis 40 barn [35,36,28], and the 241Am nuclei in a sourceare about 6.9 1014. From these quantities a ssion ratefrom environmental thermal neutrons surviving the shieldcan be derived to be: < 1014 s1; on the other hand,neutrons produced in this possible ssion process have avery low probability to induce further ssions (P 107).Thus this process is too inecient to account for the ob-servations.

In addition to the environmental ux, neutrons canalso be produced in the (,n) reaction on O isotopes; inparticular, 17O and 18O which are in the mylar and inthe AmO2 with a total concentration of 3.5 1015 cm2.However, considering a mean source activity of 35 kBq anda mean cross section for the process of 0.2 barn for 17O(natural abundance = 0.038%) and 0.5 barn for 18O ( =0.2%) [28], a neutron production rate: 1.3107 s1, isderived with a neutron mean energy of order of MeV. Theexpected ssion rate from these neutrons ( 2 barn [28])is of order of 1016 s1. Thus, in conclusion also this ssionprocess is too inecient to explain the observed excess.

Finally, another possible process is the clusters de-cay of 241Am. The emission of 34Si is considered asthe most probable one in the cluster decay of the241Am : 241Am 34Si + 207Tl, with theoretical half-lives:3.2 1017 y [37] and 5.6 1018 y [38]. Both the cre-ated fragments are unstable. In the decay of 207Tl, onlythe ground state and few excited levels of 207Pb withEexc < 1022 keV are populated; their de-excitation can-not produce e+e pairs. However, in the decay chain 34Si

7 The ux of thermal neutrons outside the shield at LNGS is: 106 cm2s1 [34], and the adopted neutron shield is madeof 1m concrete from the same Gran Sasso rock, 10/40 cmparan/polyethylene, 1.5mm Cd. In addition, there are 10 cmof OFHC Cu and 15 cm of Pb [17].

(T1/2 = 2.77 s, Q = 4601 keV) 34P (T1/2 = 12.43 s,Q = 5374 keV) 34S [24] much more energy is re-leased, and the de-excitation of high-energy levels in 34Pand 34S leads to emission of e+e pairs. The best currentexperimental limit on the decay: 241Am 34Si + 207Tl,is T1/2 > 5.8 1017 y [39]; the limit for other clustersemission is > 8.6 1016 y [30]. In conclusion, the pos-sible contribution of 241Am cluster decay is negligible( < 51015) and cannot account for the observed eect.

5 Conclusions

The experimental data show an excess of double coinci-dences in the 511511 keV energy region which are notexplained by known analysed background processes as dis-cussed above. Moreover, any sizeable contamination of thesources themselves by isotopes + emitters is not presentto our knowledge.

Thus, this measured excess gives a a relative activity = (4.70 0.63) 109 for the Internal Pair Production(IPP) with respect to alpha decay of 241Am; this value isof the same order of magnitude as the previous determina-tions [2,10,16] obtained by using dierent setups, sources(with dierent features and producers) and experimentalapproaches. In a conservative approach the upper limit < 5.5 109 (90% C.L.) can be derived.

For completeness, we mention that, if one considersthe exotic frame hypothesized in ref. [15], one would get value or limit of the same order of magnitude as thosegiven above.

Further investigations are foreseen in the future withdierent dedicated sources and set-up.

Finally, it is worth noting that this is the rst resulton IPP obtained in an underground experiment (in theprevious ones the sea-level cosmic-rays background wasestimated and subtracted), and that the value obtainedin the present work is independent on the live time esti-mate (which is dicult to estimate with very high accu-racy in similar experiments because of the relatively highintensities of the 241Am sources).

Appendix A.

In order to estimate a possible presence of a coincidenceexcess in the region of interest in runs with presence of241Am sources with respect to runs without sources, thefollowing procedure has been adopted.

Let us consider that n1 is the number of coincidenceevents measured in the region of interest in presence ofthe source during a live time T1. The conditional proba-bility to measure n1 coincidence events in the presence ofa signal S = sT1 (s is the signal rate, see text) and of abackground rate b is given by

P (n1 | S + bT1) = e(S+bT1)(S + bT1)n1

n1!. (A.1)

Page 10 of 10 Eur. Phys. J. A (2013) 49: 64

The b parameter can be estimated by performing arun in absence of sources, where n2 coincidence eventsare collected in a live time T2. In fact, the conditionalprobability to obtain n2 given a background rate b is

P (n2 | bT2) = e(bT2)(bT2)n2

n2!. (A.2)

To determine the probability distribution function (p.d.f.)of S, L(S), the procedure of ref. [40] can be followed. Ifthe prior probability of b is a uniform function in (0,),the posterior p.d.f. given n2 can be written by usingthe Bayes theorem,

f(bT2 | n2) = AP (n2 | bT2), (A.3)while the posterior p.d.f. for S given n1 and the bparameter is

f(S | n1, bT1) = AP (n1 | S + bT1), (A.4)where A and A are normalization constants. Since b is anuisance parameter, one can marginalize with respect toit obtaining

L(S) = 0

f(S | n1, bT1)f(bT2 | n2)d(bT2)

= A 0

e(S+bT1)(S + bT1)n1ebT2(bT2)n2d(bT2), (A.5)

where A is a normalization constant such that 0

L(S)dS = 1. (A.6)

These integrals can be analytically solved and the ob-tained results are

L(S) = AeSn1

k=0

(n1k

)Sn1kk

(k + n2)!( + 1)n2+k+1

, (A.7)

where = T1/T2. The constant A is given by

A =(1 + )n2+1

n1!n2!n1

k=0

(n2 + k

n2

)(

+ 1

)k . (A.8)

Finally, according to Wilks theorem, the quantity2 ln(L(S)) follows a 2 distribution [41]; thus to obtainthe best estimate for S and its interval of condence, thesame procedure used for the 2 distribution can be fol-lowed.

References

1. B.G. Pettersson, in Alpha-, Beta-, and Gamma-Ray Spec-troscopy, edited by K. Siegbahn (North-Holland, Amster-dam, 1966) p. 1569.

2. A. Ljubicic, B.A. Logan, Phys. Rev. C 7, 1541 (1973).3. N. Arley, C.Mller, Kgl. Danske Videnskab. Selskab, Mat.-

Phys. Medd. 15, 9 (1938).4. L. Tisza, Phys. Z. Sowjetunion 11, 245 (1937).5. J.S. Greenberg, M. Deutsch, Phys. Rev. 102, 415 (1956).6. K. Pisk et al., Phys. Rev. C 17, 739 (1978).7. E. Presnajderova et al., Z. Phys. A 291, 283 (1979).8. V. Chudy et al., Czech. J. Phys. B 31, 1311 (1981).9. B.A. Logan et al., Z. Phys. A 305, 295 (1982).

10. J. Stanicek et al., Nucl. Instrum. Methods B 17, 462(1986).

11. T. Cowan et al., Phys. Rev. Lett. 56, 444 (1986).12. H. Tsertos et al., Z. Phys. A 326, 235 (1987).13. W. Koenig et al., Z. Phys. A 328, 129 (1987).14. E. Berdemann et al., Nucl. Phys. A 488, 683c (1988).15. M. Inoue et al., Mod. Phys. Lett. A 5, 309 (1990).16. T. Asanuma et al., Phys. Lett. B 237, 588 (1990).17. R. Bernabei et al., Nucl. Instrum. Methods A 592, 297

(2008).18. R. Bernabei et al., Eur. Phys. J. C 56, 333 (2008).19. R. Bernabei et al., Eur. Phys. J. C 67, 39 (2010).20. R. Bernabei et al., Eur. Phys. J. C 62, 327 (2009).21. R. Bernabei et al., Eur. Phys. J. C 72, 1920 (2012).22. R. Bernabei et al., JINST 7, P03009 (2012).23. M.S. Basunia, Nucl. Data Sheets 107, 2323 (2006).24. R.B. Firestone et al., Table of Isotopes, 8th edition (John

Wiley, New York, 1996) CD update, 1998.25. E. Browne, J.K. Tuli, Nucl. Data Sheets 109, 2657 (2008).26. M.S. Basunia, Nucl. Data Sheets 108, 633 (2007).27. M.J. Martin, Nucl. Data Sheets 63, 723 (1991).28. http://spes.jaea.go.jp.29. http://www.oecd-nea.org/janis/book/book-alpha.pdf.30. M.J. Martin, Nucl. Data Sheets 106, 89 (2005).31. N. Shinohara et al., J. Nucl. Sci. Technol. 36, 232 (1999).32. E.A.C. Crouch, At. Data Nucl. Data Tables 19, 417 (1977).33. http://ie.lbl.gov/fission.html.34. P. Belli et al., Nuovo Cimento A 101, 959 (1989).35. M. Calviani et al., J. Korean Phys. Soc. 59, 1912 (2011).36. F. Belloni et al., Eur. Phys. J. A 49, 2 (2013).37. O.A.P. Tavares et al., Phys. Scr. 76, 375 (2007).38. M. Bhattacharya, G. Gangopadhyay, Phys. Rev. C 77,

027603 (2008).39. K.J. Moody et al., Phys. Rev. C 36, 2710 (1987).40. O. Helene, Nucl. Instrum. Methods 212, 319 (1983).41. K. Nakamura et al., J. Phys. G 37, 075021 (2010).