A. Dragi}, et al.: The New Set-up in the Bel grade Low-Level and Cos mic-Ray ... Nuclea r Technol ogy & Ra dia tion Pro tection: Year 2011, Vol. 26, No. 3, pp.181-192 181

THE NEW SET-UP IN THE BELGR ADE LOW-LEVEL AND COS MIC-RAY LAB O RA TORY

by

Aleksandar DRAGI]1*, Vladi mir I. UDOVI^I]1, Radomir BANJANAC1, Dejan JOKOVI]1, Dimitrije MALETI]1, Nikola VESELINOVI]1, Mihailo SAVI]2, Jovan PUZOVI]2, and Ivan V. ANI^IN1

1In sti tute of Phys ics, Bel grade, Ser bia 2Fac ulty of Phys ics, Uni ver sity of Bel grade, Bel grade, Ser bia

Sci en tific pa per UDC: 543.428/.429:551.521.6:539.166 DOI: 10.2298/NTRP1103181D

The Belgrade un der ground lab o ra tory con sists of two in tercon nected spaces, a ground level labo ra tory and a shal low under ground one, at 25 meters of wa ter equivalent. The labo ra tory hosts a low-back ground gamma spectros copy sys tem and cos mic-ray muon de tec tors. With the re cently adopted digi tal data ac qui si tion sys tem it is pos si ble to si mul ta neously study in - de pend ent op er a tions of the two de tec tor sys tems, as well as pro cesses in duced by cos mic-ray muons in ger ma nium spec trom e ters. Char ac ter is tics and po ten tials of the pres ent ex per i men - tal setup, to gether with some pre limi nary re sults for the flux of fast neu trons and stopped muons, are re ported here.

Key words: un der ground lab o ra tory, gamma-ray spec tros copy, low-level mea sure ments, cos mic rays

IN TRO DUC TION num ber of years now, while the TL is still not func - tional. The low-level and cosm ic-ray labo ratory in Bel- Con tin u ous mea sure ments of the cos mic-ray grade is ded icated to the measure ment of low ac tiv ities muon flux by means of a pair of small plas tic scinti lla - and cos mic-ray (CR) muon com ponents . At the inter - tors 50 cm ´ 23 cm ´ 5 cm started in the GLL and UL sec tion of the two re search subject s, the study of back in 2002 and lasted for about 5 years. These mea- muon-induced background in gamma spectros copy is sure ments yielded the pre cise val ues of the in tegral CR of par tic u lar in ter est. The lab o ra tory adds to the list of muon flux at ground level and at the depth of 25 m.w.e. rel a tively shal low un der ground lab o ra to ries world- [2]. Dif fer ent anal yses of the time se ries of these mea- wide (see the re cent re view [1]). It is lo cated on the surem ents have also been perform ed [3, 4]. right bank of the river Danube in the Belgrade borough Signif i cant ef forts are being made to contai n of Zemun, on the grounds of the Insti tute of Physics . the low ra don con cen tra tion within the lab o ra tory. The ground level porti on of the labo ra tory (GLL), at The UL is com pletely lined with a her meti cally 75 meters above sea level (m.a.s.l), is situ ated at the sealed, 1 mm thick alu minu m foil. The ven tila tion foot of a verti cal loess cliff, about 10 meter s high. The sys tem maintai ns the overpressure of 2 mbar, so as to under ground part of the labo ra tory (UL), use ful area prevent radon dif fusion from the soil. Fresh air enter - 45 m2, is dug into the foot of the cliff and is ac ces sible ing the labo ra tory is passed through a two-stage fil- from the GLL via a 10 meter s long hori zon tal corri dor ter ing sys tem. The first stage is a mechan ical fil ter for which also serves as a pressure buffer for a slight dust re moval. The second one is a bat tery of coarse overpressure that is consta ntly maintai ned in the UL and fine char coal ac tive fil ters. The con cen tra tion of (fig. 1). The overbur den of the UL is about 12 meter s radon is kept at an aver age value of about 10 Bq/m3. of loess soil, equiva lent to 25 meter s of water . The con - Through out the years, cer tain in ter est ing be hav iors tainer, which is to accom modate the top labo ra tory of the said concen tra tion have also been reporte d [5, (TL), is sit u ated at the top of the cliff, just above the 6]. UL. The GLL and UL have been in some use for a The two lab o ra tory spaces have re cently been furnishe d with a new exper i m ental set-up which is now ready for rou tine measure ments. Here pre sented are some pre lim inary re sults of wider in ter est, ob - * Cor re spond ing au thor; e-mail: [email protected] A. Dragi}, et al.: The New Set-up in the Bel grade Low-Level and Cos mic-Ray ... 182 Nuclea r Technol ogy & Ra dia tion Protec tion: Year 2011, Vol. 26, No. 3, pp. 181-192

Fig ure 1. Cross-sec tion of the low-level and CR labo ra tory at IOP, Bel grade, 44°49'N, 20°28'E, ver ti cal ri gid ity cutoff 5.3 GV

tained during the testing period of the new equipm ent tec tor, of 10% ef ficie ncy, is put to use in the GLL. It is and the devel op m ent of the needed software. shielded with lead of the same ori gin, parts of a plumb - ing sys tem collected at a de moli tion site of an old housing es tate. The exact history of this lead is not EX PER I MEN TAL SET-UP known, but all the com ponents are known to be older than two half-lives of Pb-210. The equipm ent now consis ts of two al most iden- At the heart of the data ac quisi tion sys tem are ti cal sets of de tec tors and an a lyz ing elec tron ics, one two flash: ana log to digi tal convert er (ADC), flash an- set sit u ated in the GLL, the other in the UL. Each set is a log to dig i tal con verter (FADC), one in each labo ra - com posed of muon detec tors and a gamma spectrom e- tory, made by CAEN (type N1728B). These are versa - ter. tile instru m ents, capa ble of working in the so-called A pair of plas tic scintillator de tec tors is used for en ergy his to gram mode when per form ing as dig i tal CR muon measure m ents. One of them is a larger spec trom eter s or, in the oscillogram mode, when they 100 cm ´ 100 cm ´ 5 cm detec tor, equipped with four per form as dig i tal stor age os cil lo scopes. In both PMT direct ly coupled to the corners bevele d at 45°, modes, they sam ple at 10 ns inter vals into 214 channels made by Amcrys-H, Kharkov, Ukraine. The other, a in four inde pend ent inputs. The full voltage range is small 50 cm ´ 23 cm ´ 5 cm plas tic scintillator de tec - ±1.1 V. tor, with a sin gle PMT look ing at its lon gest side via a They are capa ble of oper at ing in the list mode, Perspex light guide ta pering to the di am eter of a PMT, when every ana ly zed event is fully re corded by the made by JINR, Dubna, Russia , and as sem bled local ly. time of its oc currence and its am pli tude. This enable s The smaller detec tor may serve as a check of sta bil ity the corre la ti on of events, both prompt and arbi tra rily of the muon time se ries obtai ned from the larger detec - de layed, at all four in puts with the time res o lu tion of tor, which is im portant for long term measure m ents. It 10 ns. Single and coin ci dent data can be or ganize d into can also be used (in co in cidence with the larger detec - time se ries within any inte gra ti on period from 10 ns tor) for measure ments of the lateral spread of par ticles up. The two N1728B units are synchro nize d, enabli ng in CR show ers. Plas tic scin til la tion de tec tors are also co in ci dence/cor re la tion of the events re corded in both em ployed for ac tive shielding of gamma spectrom e- of them. The flexi ble software encom passing all above ters. said off-line analyses is user-friendly and entirely In the UL, a 35% ef ficie ncy radio-pure p-type homemade. HPGe de tec tor, made by ORTEC, in its 12 cm thick cy - The usual dispo si ti on of FADC inputs is de- lin dri cal lead cas tle, is de ployed. An other HPGe de - scribed next. The preamplifier outputs of the PMT of A. Dragi}, et al.: The New Set-up in the Bel grade Low-Level and Cos mic-Ray ... Nuclea r Technol ogy & Ra dia tion Pro tection: Year 2011, Vol. 26, No. 3, pp.181-192 183 the larger detec tors are paired di ag o nally, the en tire de - mic rays on the spectra in low-level high-reso lu ti on tec tor thus engag ing these two inputs of the FADC. gamma-ray spec tros copy, it is used in co in cidence as Signals from these inputs are later coin cide d off-line the trigger for the CR-induced proces ses. These two and their am pli tudes added to produce the single spec- functi ons of the sys tem are perform ed si multa neousl y tra of these de tectors. This pro cedur e re sults in a prac - and do not in ter fere, as they are re alized by differ ent ti cally com plete sup pres sion of the un in ter est ing off-line analy ses of the same set of data. low-ener gy porti on of the background spec trum (up to some 3 MeV), mostly due to envi ron m ental radi a ti on, leaving only high-ener gy loss events due to CR muons TEST ING THE SYS TEM AND and EM showers that peak at about 10 MeV. The out - DE VEL OP ING THE SOFT WARE put of the PMT of the smaller de tec tor is fed to the third input. The fourth input is reserve d for the HPGe de tec - In order to test the perfor m ance of the digi tal tors. spec tros copy sys tem, a se ries of test measure ments In some in stances, aux il iary mea sure ments are with dif ferent count rates and dif ferent types of radi a - perform ed with a dif ferent defi ni ti on of the inputs of tion detec tors at the input of the FADC are perform ed. the data ac qui sition sys tem. For ex am ple, a (3 ´ 3)” One of the tests is de signed to cor respond to the NaI detec tor is used in the GLL to scan the response of real sit ua ti ons where neutrons create d by CR muons in the larger detec tor to CR as a functi on of the posi ti on of the lead shield pro duce cer tain ef fects in HPGe de tec- the in ter ac tion point. tors. Neutrons produced by muons in the vicin it y of In the UL, the HPGe detec tor is posi ti oned be- the de tec tor or the surround ing rock mass repre sent a neath the center of the larger detec tor (fig. 2). For the signif i cant source of background in ul tra-low back- purpose of measur ing low acti vi ti es, the large plasti c ground ex per i ments car ried out deep un der ground, de tector is used in anti co inci dence, as a cos mic-ray such as those searching for dark matter or double beta muon veto de tec tor. In order to study the ef fects of cos - decay . In the test, done at the GLL, Cf-252 was used as a neu tron source and the small plas tic scintillator as a trigger for neutrons. To disti nguish be tween the ef fects of fast and slow neutrons, some mate ri als com mon in neutron work, such as rubber ize d B4C, Cd sheets, par - af fin, lead and iron slabs, were placed around and in betwee n the source and the de tec tors. In addi ti on to the en vi ron men tal back ground, the HPGe spectra con sists of dif ferent feature s induced by slow and fast neutrons in the HPGe de tec tor and sur round ing ma te ri als. Results of measure ments are stored as a list of events rep re sented with their ampli tudes, time tags, des ig na tion of in put chan nels and some ad di tional in- form ati on (pile-up event or not, etc.). The time tag for every event is de ter mined by the mom ent of cross ing the set-trig ger ing level. In or der to min imize the am- pli tude walk, there is a possi bil ity to choose betwee n dif ferent types of triggers, termed here as sim ple, digi - tal or CFD. We have stuck to the dig ital trig ger which was found to work reli ably and, if neces sary , to the off-line correc ti on of the am pli tude walk [7]. The dis tri bu tion of time in ter vals be tween events in the trigger detec tor and the HPGe is deduced from the recorded data in off-line analy sis . There is no need to im plem ent the hardware tome-to-am pli tude convert er (TAC). For conve nienc e, in what follows we will re fer to this dis tri bu tion as the TAC spec trum. As an illus tra tion, the TAC spec trum between Fig ure 2. De tec tors in the un der ground lab o ra tory (UL). events in the plas tic scintillator and HPGe is pre sented The big plastic scintillator is posi tioned over the HPGe here (fig. 3).The prompt-time dis tri buti on is seen to be detec tor , seen in its lead shield ing. The small plastic about 90 ns wide, while the tail of de layed coin ci - scintillator is in the front up per right cor ner. A her met i - cally sealed, 1 mm Al lining cover ing the entir e UL, dences is discernable beyond approx. 100 ns upon the which en ables the dou bly fil trated ven ti la tion sys tem to prompt peak. The same time spectrum , off-line cor- sus tain an overpressure of 2 mbar and keeps the ra don rected for the am pli tude walk, accord ing to the proce - con cen tra tion at an av er age level of some 10 Bq/m3, is dure descri bed in [7], is present ed in fig. 3(b). The full also shown A. Dragi}, et al.: The New Set-up in the Bel grade Low-Level and Cos mic-Ray ... 184 Nuclea r Technol ogy & Ra dia tion Protec tion: Year 2011, Vol. 26, No. 3, pp. 181-192

Fig ure 3. (a) The time spectrum between events in the plastic scintillator and the HPGe; (b) the same time spec- trum, but with off-line time cor rected for the ampli tude walk

with at half maxi m um (FWHM) is now only 15 ns and de layed co in cidences, per haps, start as early as at ap - proxi m ately 20 ns after the prompt peak. Minding the geom etry of our set-up, we expect the ef fects induced by fast neutrons, both in the envi ron m ent and the Fig ure 4. The prompt, de layed (up to 1 ms), and ran dom co in ci dence spec tra of: (a) the 478 keV Dopp ler-wid ened HPGe de tec tor itsel f, to be within the prompt peak, line from the (n, a) re ac tion on 10B which appear s only in while those induced by thermalized neutrons should the delayed spectrum; (b) the 847 keV line from the (n, be found in the tail of delay ed coin ci dence s. n') in elas tic scat ter ing on 56Fe which ap pears only in the To il lus trate the complete sep a ra tion of the ef- prompt spectrum; (c) the 558 keV line which appear s in fects due to fast and slow neutrons, here pre sented (fig. both the prompt and de layed spec tra, prov ing that this line orig i nates partly from the usu ally as sumed ther mal 4) are the porti ons of the coin ci dent HPGe spectrum neu tron cap ture by 113Cd and, de pend ing on the hard - around the spectra l lines origi nat ing from dif ferent ness of the neu tron spec trum, in part, from the fast neu - pro cesses in duced by neu trons in dif fer ent ma te ri als, tron (n, n') reac tion on 114Cd gated with dif ferent por tions of the time spectrum – with the prompt peak, tail of delay ed events up to one cited in in elas tic neu tron scatter ing with the en ergy of micro sec ond and the flat porti on of random coin ci - the recoil ing nucle us. The one at 596 keV appears in the dences. Short com ments can be found in the capti on prompt spectrum , since the state at 596 keV in Ge-74 is under fig. 4. short-lived. The regu lar peak of this ener gy in the de - We will now briefly com ment on the two layed spectrum result s from the therm al neutron capture well-known structure s induced by neutrons in the by Ge-7, as is the case with the neighbor ing 609 keV HPGe de tec tor itsel f, the struc tures at 596 keV and 692 line stem ming from the same cap ture re action. If the keV. Their ap pear ance in the co in cidence spectra is de - neutron flux at the detec tor is high, some of the inten sit y picted in fig. 5. The trian gu lar distri buti ons result from of the ubiqui tous background line of 609 keV, usuall y the sum ming of the ra di a tions de pop u lat ing the state ex- en tirely at trib uted to 214Bi, is due to this pro cess. A. Dragi}, et al.: The New Set-up in the Bel grade Low-Level and Cos mic-Ray ... Nuclea r Technol ogy & Ra dia tion Pro tection: Year 2011, Vol. 26, No. 3, pp.181-192 185

mind when using the inte gral of this structure for fast neutron flux deter mina ti on. It appears that here the shape and the inten sit y of the dis tri buti on strongly de - pend on the height of the trigger ing level. The re coil pulse is prompt, while the corre spond ing 692 keV pulse follows the recoil with delay distri buted accord - ing to the de cay law with the half-life of 444 ns. When the trigger is higher than the height of the recoil pulse, 692 keV pulse sums prac ti cally com pletely with the re coil. When the trig ger is lower than the re coil, it will trigger the ADC, and this pulse, togethe r with the fol- lowing 692 keV pulse, will be reject ed by the pile-up re ject ing al go rithm. This is il lus trated in fig. 6 where the same porti on of the direct HPGe spectrum is pre - sented, with two dif ferent trigger ing levels . The width of the tri an gu lar struc ture ap pears pro por tional to the height of the trigger ing level. If this structure is to be used for quanti tati ve purposes, the safe height of the trigger ing level that may be recom mended is, thus,

Figur e 5. The spec trum of prompt, de layed, and ran dom co in ci dences (prac ti cally neg li gi ble) con tain ing: (a) the structur e at 596 keV from inelas tic fast neutr on scatter - ing on Ge-74 in the prompt spec trum, and the line of the same en ergy from slow neutr on captur e on Ge-73, in the de layed spec trum; (b) the structur e at 692 keV from the inelas tic scat ter ing of fast neutr ons on Ge-72, this time in the delayed spectrum, due to the finite lifetime of 444 ns at the ex cited state of 692 keV

The struc ture at 692 keV, however , appears in the de layed spec trum since the ex cited state of this en - ergy in Ge-72 is com para ti vely long-lived, with a half-life of 444 ns, which is long in com pari son to the time res olu ti on of our system . As a dem onstra tion of the ca pa bil i ties of the sys tem, we de ter mined this half-life by setting the software gate to encom pass the whole tri an gu lar struc ture in the co in ci dent Ge spec - trum, thus produc ing a time spectrum corre spond ing to this condi ti on, thanks to which the fit produced a sat isfac tory value of 447(25) ns for the said half-life. This parti cu lar struc ture has been studied in de- tail many times in the past, since 692 keV ra di a tion is Fig ure 6. The 692 keV dis tri bution in the di rect spec trum pure E0, detect able with 100% ef fi ciency, which is with the trig ger ing level set at 50 keV (a) and at 20 keV why the in te gral of the trian gu lar struc ture is a re liable (b) measure of the fast neutron flux at the posi ti on of the detec tor [8-12]. These studies were perform ed with an a log spec tros copy sys tems where the in te gra tion perhaps about 50 keV, when the loss of the counts consta nts are long and the recoil s invari ably sum up within the structure is expect ed to be negli gible. with the 692 keV pulses. In digi tal spec troscopy sys - Af ter the sys tem sat is fac to rily passed all the tems, however , there is one im portant caveat to keep in tests, we started some pre lim inary measure ments of A. Dragi}, et al.: The New Set-up in the Bel grade Low-Level and Cos mic-Ray ... 186 Nuclea r Technol ogy & Ra dia tion Protec tion: Year 2011, Vol. 26, No. 3, pp. 181-192 the type that we plan to per form in the long run in the LOW-LEVEL MEA SURE MENTS future , and it is of these re sults that we report in what fol lows. Fu ture ap pli ca tions of the low-back ground gamma spectro scopi c system include the study of rare nu clear pro cesses, mea sure ments of en vi ron men tal ra - PLANNED MEA SURE MENTS dio ac tiv ity and radiopurity of ma te ri als. The cyli ndri cal lead shielding of the 35% ef fi- We plan to conti nu ously collec t the background ciency ra dio-pure HPGe ORTEC de tector , with a wall spec tra of all de tec tors in both lab o ra to ries, even tu ally thickness of 12 cm and an overall weight of 900 kg, was changing only their mutual spati al arrange m ents. The cast local ly out of scratch plumbing re trieved after the spec tra of the large scin tilla tors will stretch up to a cou - dem oli tion of some old hous ing. The inte gral back- ple of hundreds of MeV, so as to include all multi ple ground rate in the region from 50 keV to 3 MeV is about CR and shower events, while the one corre spond ing to 0.5 cps. The lines of Co-60 are absent in the background the HPGe de tec tor will go up to about 30 MeV, in order spec trum, while the line of Cs-137 with the rate of to in clude all pos si ble nu clear ra di a tions in duced by 1×10–4 cps starts to ap pear sig nif icantly only if the mea- CR radi a ti ons. Each event is to be recorded in a sepa - surem ent time approache s one month. Fukushima ac - rate list, in ac cor dance with the time of the occur rence tiv ities, though strongly pres ent in our in let air filter of its trigger (with the res olu ti on of 10 ns) and by its sam ples, did not enter the background at observ able am pli tude (in 32 k channels ). By off-line analy ses of lev els, in spite of the great quan tities of air that we pump this data we ex pect to di rectly ob tain: into the UL to maintai n the overpressure, and it seems – the con tin u ous time se ries of cos mic-ray in ten sity that the double air filter ing and double buffer door sys- (muon plus elec trom agneti c – EM, com ponents ) tem, along with stringent radi a ti on hygiene measures , is in large and small plas tic scin tilla tors at ground capa ble of keep ing the UL clean in cases of global acci - level, as well as those gener ate d under ground, den tal con tam i na tions (see e. g. [13]). – decoherence curves of cos mic-ray co in cidence No sig na tures of en vi ron men tal neu trons, nei- counts and co in ci dence spec tra at dif fer ent sep a - ther slow nor fast, are present in direct background rati ons of the said detec tors, be it at ground level or spec tra. The rates of some char acter istic back ground the under ground one, the idea being to first define and, af ter wards, sepa rat e muon from EM com po - nents, Ta ble 1. Count rates in some back ground lines. Rotes are – the spectrum of the HPGe de tec tor in coin ci dence given in counts per sec ond (cps) and antico inci dence with the large plasti c Energy [keV] Count rate [cps] scintillator posi ti oned right above it, in the under - 186.2 (Ra-226) 2.4×10–4 ground, and 351.9 (Pb-214 1.1×10–3 – as well as the signa ture s of the soft com ponent of –4 583.1 (Tl-208) 6.6×10 EM showers in the spectrum of the unshie lded NaI 609.3 (Bi-214) 1.1×10–3 detec tor, taken in coin ci dence with the plasti c 911.1 (Ac-228) 4.5×10–4 detectors. 1460.8 (K-40) 3.5×10–3 Since all above men tioned measure ments are –3 spec tral, we hope to exploi t this feature to some advan - 2614.5 (Tl-208) 1.1×10 tage, even in the case of rather feature less spec tra of plas tic scin tilla tors. With the help of MC simu la tion pro grams (mainly CORSIKA and GEANT4), we ex- pect to discri m inate the signa ture s of CR muons from those of electro m agneti c showers to some degree. If all the measure ments are per formed con tin u - ously, we es timate that, to gether, both set-ups will pro - duce about 1 TB of data per year, all of which would be kept per manently for later anal yses. To il lus trate the po ten tial of these measur e - ments, we will now briefly report on some preli m inary re sults ob tained dur ing a test ing pe riod, ap prox i - mately year long. We will first briefly dis cuss the per - form ance of low-level measure m ents and then those per tain ing to CR mea sure ments. Figur e 7. The back ground spec trum of the HPGe de tec - tor in its lead cas tle in the UL and the part of the spec - trum coin ci dent with the large scintillator posi tioned right above it A. Dragi}, et al.: The New Set-up in the Bel grade Low-Level and Cos mic-Ray ... Nuclea r Technol ogy & Ra dia tion Pro tection: Year 2011, Vol. 26, No. 3, pp.181-192 187 lines are listed in tab. 1. With the large plas tic scintillator current ly posi ti oned rather high over the de tector top, at a ver tical dis tance of 60 cm from the top of the lead castl e, in order to allow for the placing of volu m inous sources in front of the verti cally ori- ented detec tor, the off-line reduc ti on of this inte gral count by the CR veto condi ti on is about 18% (see fig. 7). Up to a factor of two might be gained if the veto de- tec tor were to be posi ti oned at the closest possi ble dis- tance over the HPGe de tector . This agrees well with simple es timates of the rate of events sus cep tible to the veto condi ti on [14]. The veto spec trum contai ns all events, prompt as well as those with delay s of up to 10 ms, which is why be sides the conti nuum it contai ns only the lead X-rays and the anni hi la ti on line. As we shall see, the se lection of de layed events only re veals some other de tails in this spec trum. Since for the time being we are not able to im prove on the intri nsic back- ground of our de tec tor, when an a lyz ing the an a lyt i cal powers of our system , at present, we do not insis t on the low er ing of sta tis ti cal er rors which de pend on background levels solely and are dif ficult to re duce fur ther with available means, but rather empha size its sta bil ity due to the low and control led radon concen - tra tion in the lab o ra tory. This is es sen tial, es pe cially in NORM measure m ents, and makes our sys tem virtu - ally free of sys tem atic error s as compared to sys tems which oper ate in envi ron m ents where ra don is not con- trolled and where the reduc ti on of post-radon back- Figur e 8. The spectra of two diag o nals of the large plas tic ground acti vi ti es is achieved by flushing the detec tor detec tor in the UL. Note that the peak of charged parti cle cavit y with liquid nitrogen vapor , where the transie nt energy losses of 10 MeV, which corr espo nds to channel regimes during sample changes and possible 320, is not sepa rated from the low-en ergy tail of deposition of radon progenies [15] may introduce Compton scat tered en vi ron men tal gamma ra di a tions. systematic uncertainties which are difficult to When summed, in co in ci dence they pro duce the spec - trum from fig. 9 estimate.

COS MIC-RAY MEA SURE MENTS

Muon spec tra and the time series

During the com mission ing of the large plasti c de tectors, we tested the re sponse of these de tectors to CR muons and their sta bil ity over a prolonged pe riod of time. Cer tain re sults of these pre limi nar y stud ies are pre sented here. In fig. 8, we pres ent the spectra of the two di ago - nals of the large plas tic scintillator in the UL. Con trary to the situ a ti on in the GLL, the peak of charged parti cle ener gy losses sit uate d at about 10 MeV (due to both muons and elec trons from EM showers), is not fully sepa rat ed from the low-ener gy tail of Compton elec- trons in the UL, be cause of gamma-ray in ter actions Fig ure 9. The sum spectra of two diag o nals of the large (both envi ron m ental and from EM showers). plastic de tec tors in the UL and GLL. For com pari son, the spectra are normal iz ed for the peaks to coin cide. Figure 9 pres ents the coin ci dent sum spec tra of Channel 650 now cor re sponds to the muon energy loss of the two diag o nals . Ener gy spectra now contai n only 10 MeV. The in te gral of this peaked dis tri bu tion is taken the well-de fined peak of charged par ticles en ergy as the first ap prox i ma tion to the CR muon count by the losses. The offset s are not im posed and occur sim ply large de tec tors A. Dragi}, et al.: The New Set-up in the Bel grade Low-Level and Cos mic-Ray ... 188 Nuclea r Technol ogy & Ra dia tion Protec tion: Year 2011, Vol. 26, No. 3, pp. 181-192

be cause at low en er gies there are no co in cident events. As the sim ula ti ons dem onstra te, this is so becaus e sin - gle Compton electrons do not produce enough light to trig ger both di ag o nals. The major it y of events that produce this peaked distri buti on are due to CR muons that pass through the de tector . We thus form the time se ries of this spec trum in tegrated over dif fer ent time in ter vals. As an ex am - ple, in fig. 10 we present the time se ries of this count in 30-minute inter vals, both in the UL and in the GLL, for a period of 16 days in March 2010, and in one-hour in - ter vals for the period start ing with July 26, 2010. The data are not cor rected ei ther for at mo spheric pressure or temperature. The two se ries ap pear highly cor re lated, the am- pli tude of the modu la ti on of this count in the UL is about 1.8%, while the corre spond ing one in the GLL is about 3.5%. At these in te grat ing times, this is al ready suf fi ciently sta tis ti cally sig nif i cant, even in the UL. Pre vi ous mea sure ments at the same lo ca tion with the small de tec tors yielded re sults for the muon flux of 1.37(6)×10–2 per cm2s in the GLL, and of 4.5(2)×10–3 per cm2s in the UL [2].

NEUTRONS AND STOPPED POS ITIVE MUONS IN THE UNDER GROUND LAB O RA TORY

During the testing period, we have accu m ulat ed some six months of data-taking in the under ground labo ra tory . The background spec trum of the HPGe de -

Fig ure 11. The por tion of the back ground of the HPGe spec trum co in ci dent with the large plas tic de tec tor with de lays in the range of 1 to 5 ms, af ter 187 days of mea sure - Fig ure 10. The time se ries of the CR muon count of the ment time. It shows the an ni hi la tion line which is due to large plas tic de tec tor in the UL – (a) and (b) graphs – and the de cays of pos i tive muons stopped in the lead cas tle, GLL (c) and (d) graphs for the pe riod start ing with and the tri an gu lar struc ture at 692 keV, which is due to March 12, 2010, av er aged to half hour inter vals, as op- inelas tic scat ter ing of fast neutr ons on 72-Ge, the neu- posed to the period start ing with July 26, 2010, av er ag - trons orig i nating mostly from dir ect fast muon inter ac - ing to one hour in ter vals. It is ev i dent that the tions with nu clei and cer tainly less from cap tures of mod u la tion in the two lab o ra to ries is cor re lated and that stopped neg a tive muons. The thresh old in this spec trum the am pli tude of the mod u la tion in the UL is roughly half is suf fi ciently high to leave this last struc ture un scathed A. Dragi}, et al.: The New Set-up in the Bel grade Low-Level and Cos mic-Ray ... Nuclea r Technol ogy & Ra dia tion Pro tection: Year 2011, Vol. 26, No. 3, pp.181-192 189

of lead, a com mon envi ron m ent in most measure m ents of low ac tiv i ties. The second feature of this spec trum is the anni hi - lati on line. The gate put on this line gives the TAC dis- tri buti on present ed in fig. 12(b), where the fit through the tail of de layed co in cidences yields the mean life of 2.24(9) ms. This jus tifies the as sump tion that these events are due to the decay s of stopped posi ti ve muons. We fur ther as sume that the source of these delayed annihilations is ho mo ge neously dis trib uted through out the vol ume of the lead castl e and use GEANT4 to find the over all de tec tion ef fi ciency. From the in ten sity of these delay ed annihilations, we then obtai n that the num ber of stopped posi ti ve muons per kg of lead per –4 sec ond equals 3.0(5)×10 mstop/kgs. We have also been able to es tim ate the number of stopped muons in the large plas tic scin tilla tors them - selves, both in the GLL and the UL (e. g. [16]). For this

Fig ure 12. Time dis tri butions of events that be long (a) to the struc ture of 692 keV, the slope of which yields 500(50) ns for the half-life of the state of this en ergy in 72-Ge and, (b) that of the an ni hi la tion line, which yields 2.24(9) ms for the mean life of the muon tec tor contai ning the coin ci dence s with the large plas- tic scintillator from the de layed tail of the cor re spond - ing TAC distri buti on in the region of 1 to 5 ms, shows at this sta tis tics only two in ter est ing fea tures (fig. 11), though some more seem to emerge, but still insuf fi- ciently sig nif icant. The first is the al ready dis cussed tri angu lar struc ture at 692 keV already discusse d, which orig inates from the in elas tic (n, n’) scatter ing ex cit ing the first ex cited state of the sta ble Ge-72 within the Ge de tec tor itself. In this case the trig ger was suf ficie ntly high, and ac cording to our finding, the inten sit y of the 692 keV distri buti on can be reli ably used for the es ti mate of the fast CR-induced neutron flux at the posi ti on of the detec tor. To verify this, we appli ed the software gate to this structure and obtai ned the TAC distri buti on present ed in fig. 12(a). Although Fig ure 13. The dis tri butions of short time in ter vals be - the sta tis tics is poor, the fit through the tail of de layed tween succes sive counts in large plas tic scin tilla tors which sit on the dis tri bution of time in ter vals be tween coin ci dence s yields the half-life of 500(50) ns, which suc ces sive counts with a long ex po nen tial con stant that com pares well with the known value of 444 ns. Using cor re sponds to the to tal counting rate and appears flat on the expres sion from ref. [8], we obtai n the value of this scale. The decay constan t of short time inter val dis- 4(1)×10–7 cm–2s –1 for the flux of neu trons of CR ori gin tri bu tions, how ever, equals that of a muon mean life. with ener gies over 1 MeV. This refers to the flux at the Note that the first two life times are missing, due to the dead time of the sys tem, which in this par tic u lar case depth of 25 m.w.e., (see e. g. [16]) within roughly a ton equals 4 ms A. Dragi}, et al.: The New Set-up in the Bel grade Low-Level and Cos mic-Ray ... 190 Nuclea r Technol ogy & Ra dia tion Protec tion: Year 2011, Vol. 26, No. 3, pp. 181-192 purpose, we looked into the dis tri buti on of time inter - en ables the dis en tan gle ment of the sig na tures of these vals be tween the succes sive counts of these de tectors. radi a ti ons. As an il lustra tion, fig. 14 presents the spec- The gross structure of this dis tri buti on is nicely expo - trum of the large plas tic de tec tor in the UL, in co in ci - nen tial, cor re spond ing to the av er age CR count ing rate dence with the small de tec tor, at their small est pos si ble and to an av er age time in ter val be tween the counts, re - sepa ra ti on. This is to be com pared with the direct spec - cip ro cal to the said rate. At short time inter vals, how - trum of the large detec tor, as pre sented in fig. 9. Re- ever, the distri buti on strongly depart s from this shape flecting the structure of the EM com ponent at the (fig. 13). given loca ti on, not only the inten sit y, but also the It is again expo nen ti al, now with the time con- stant of 2.16(4) and 2.22(3) ms in the GLL and UL, re - spec tively, re pro duc ing sat is fac to rily the muon mean life. This suggest s that these events origi nate from muons that both stop and decay within the detec tor. Minding that the fiducial vol ume for this kind of signa - ture has not been esti mated, the inten sit y of this expo - nen tial dis tri bu tion, tak ing in ac count the miss ing events due to the 4 ms long dead time, now gives the es - ti mate for the lower limit of the num ber of muons that stop in 5 cm of plasti c per square meter per second, at –2 2 ground level as 6×10 mstop per m s, and at a depth of –2 2 25 m.w.e. as 1.52×10 mstop per m s. It is in ter est ing to com pare those figures with the result s obtai ned re- cently at Gran Sasso [18, 19].

DECOHERENCE CURVES Fig ure 15. The decoherence curve in the GLL, reflect ing AND SPEC TRA the lat eral profile of EM show ers on the surface ; the nce de notes num ber of co in ci dent events In test mea sure ments, co in ci dence spec tra be - tween large and small de tectors at differ ent sep ara - tions betwee n the two were recov ered in off-line anal- yses, both in the GLL and the UL. These are predom inantl y the spectra of EM showers, as seen by re spec tive de tec tors. The com par i son with their di rect spec tra, which at lower en er gies are com posed mostly of the sig na tures of en vi ron men tal gamma ra di a tions and at higher ener gies of the signa ture s of CR muons,

Fig ure 16. The decoherence curve in the UL, reflect ing the lat eral profile of EM show ers at the equiva lent depth of 25 m.w.e; the nce denotes num ber of coin ci dent events

shape of the spectrum changes with de tector sep ara - tion, and the differ ences be tween these co in cidence spec tra in the GLL and the UL re flect the dif fer ence in the com posi ti on of showers on the surface and under - ground. The sim plest in te gral char ac ter is tic of the shower profil e is present ed by the inte gral of this coin - ci dence spec trum as a func tion of de tec tor sep a ra tion, Fig ure 14. The coin ci dent spectrum of the large plastic sometim es re ferred to as the decoherence curve. Fig - detec tor in the UL, with the small de tec tor at the smallest sep a ra tion be tween them. Com pare this with the di rect ure 15 shows the decoherence curve at the GLL, fig. 16 spectrum present ed in fig. 9 the same in the UL. A. Dragi}, et al.: The New Set-up in the Bel grade Low-Level and Cos mic-Ray ... Nuclea r Technol ogy & Ra dia tion Pro tection: Year 2011, Vol. 26, No. 3, pp.181-192 191

Note that the two curves can not be satis fac to rily ro pean Cos mic Ray Sym po sium, Košice, Slovakia, fit ted with quite the same type of functi onal de pend- 2008, pp. 368-373 [5] Udovi~i}, V., et al., Ra don Problem in an Under - ence, so that the width of the dis tri buti ons on the sur- ground Low-Level Lab o ra tory, Ra di a tion Mea sure - face and under ground cannot be di rectly com pared. ments 44 (2009), 9-10, pp. 1009-1012 The much narrower distri buti on under ground is a re- [6] Udovi~i}, V., et al., Ra don Time-Se ries Anal y sis in sult of the harder CR muon spec trum and of the dif fer- the Un der ground Low-Level Lab o ra tory in Bel grade, ent radi a ti on and at tenu a ti on lengths, as well as of the Ser bia, Ra di a tion Pro tec tion Do sim e try, 145 (2011), 2-3, pp. 155-158 ge om e try of the shower-pro duc ing me dium. The ra tio [7] Puzovi}, J., Ani~in, I., An Off-Line Method for the Cor - of the inten si ti es of the dis tri buti ons on the surface and rec tion of Am pli tude Walk in Lead ing-Edge Tim ing, under ground is roughly twice the rati o of CR muon in- Nucl. Instr. and Meth. in Phys. Res., A572 (2007), 2, pp. ten si ties at two lo ca tions. Full in ter pre ta tion awaits 926-928 better sta tis tics. [8] Škoro, G., et al., En vi ronmen tal Neu trons as Seen by a Ger ma nium Gamma-Ray Spec trom e ter, Nucl. Instr. and Meth. in Phys. Res., A316 (1992), 2-3, pp. 333-336 [9] Fehrehbacher, G., Meckbach, R., Paretzke, H. G., CON CLU SION Fast Neu tron De tec tion with Ger ma nium De tec tors: Com pu ta tion of Re sponse Functions for the 692 keV We have pre sented some pre lim inary re sults for In elas tic Scat ter ing Peak, Nucl. Instr. and Meth. in Phys. Res., A372 (1996), 1-2, pp. 239-245 muon and neutron fluxes at ground level and under - [10] Fehrehbacher, G., Meckbach, R., Paretzke, H. G., ground spaces of the Belgrade low-level and CR labo - Fast Neu tron De tec tion with Ger ma nium De tec tors: ratory , obtai ned during the com mission ing of the new Unfold ing the 692 keV Peak Re sponse dor Fission equip ment con sist ing of two scin til la tion de tec tors, a Neu tron Spec tra, Nucl. Instr. and Meth. in Phys. Res., HPGe de tec tor, and a dig i tal spec tros copy sys tem A377 (1997), 2-3, pp. 391-398 [11] Ataç, A., et al., Discrim i na tion of Gamma Rays Due based on two CAEN N1728B units. The main advan - to In elas tic Neu tron Scat ter ing in AGATA, Nucl. Instr. tage of the present set-up is that it enable s com plex and Meth. in Phys. Res., A607 (2009), 3, pp. 554-563 mea sure ments in volv ing rou tine low ac tiv ity mea- [12] Jovan~evi}, N., et al., Neu tron In duced Back ground surem ents with modest means, along with some inter - Gamma Activ ity in Low-Level Ge Spec troscopy Sys- esting re search work re lated to cos mic-ray physics. tems, Nucl. Instr. and Meth. in Phys. Res., A612 (2010), 2, pp. 303-308 We find that the result s obtai ned in this test ing phase [13] Adži}, P., et al., The Back ground Gamma-Ray Study justi fy the planned program of measure m ents and that be fore and af ter the Chernobyl Ac ci dent, En vi ron - the fu ture im prove ment on sta tis tics will con trib ute ment In ter na tional, 14 (1988), 4, pp. 295-297 not only to the qualit y of the result s al ready obtai ned, [14] Bikit, I., et al., Ve to ing Tech niques in Ra dio ac tive but furthemore with increa sed sensi ti vit y new result s Con tam i na tion Re search, in Ra dio ac tive Con tam i na - tion Re search De vel opments (Eds. N. K. Henshaw, C. are also ex pected to emerge. S. Alleyne), Nova Sci ence Pub lish ers, Hauppauge, N. Y., USA, 2010, pp. 67-99 [15] Guiseppe, V. E., et al., A Ra don Prog eny De po si tion ACKNOWL EDG EMENT Model, arXiv 1101.0126v1 [nucl-ex] 30 Dec, 2010 [16] Ryazhskaya, O., Neu trons from Cosmic-Ray Muons Un der ground, 20th Eu ro pean Cos mic Ray Sym po - The present work was funded by the Minis try of sium, Lisbon, 2006 Edu ca ti on and Science of the Repub li c of Serbia, un - [17] Coan, T., Liu, T., Ye, J., A Com pact Ap pa ra tus for der the Project No. 171002. The Bel grade Labo ra tory Muon Life time Measure ment and Time Dila tion bears the name of “Dr. Radovan Antanasijevi}”, in Dem on stra tion in the Un der grad u ate Lab o ra tory, Am. honor of its early de ceased founder and first di rec - J. Phys., 74 (2006), 2, pp. 161-164 tor. [18] Bonardi, A., et al., On a Mea sure ment of At mo spheric Stopping Muons and Neu tron Fluxes, 31st ICRC, Lodz, Po land, 2009 [19] Bonardi, A., Study of Cos mic Ray Neu trons, Ph. D. REF ER ENCES thesis, Uni ver sity of Torino, Torino, It aly, 2010

[1] Niese, S., Un der ground Lab o ra to ries for Low-Level Ra dio ac tiv ity Mea sure ments, in: Anal y sis of En vi - ronmen tal Radionuclides (Ed. P. Povinec), Elsevier, Amster dam, 2008, pp. 209-239 Recei ved on Septem ber 27, 2011 [2] Dragi}, A., et al., Mea sure ment of Cos mic Ray Muon Ac cepted on De cem ber 6, 2011 Flux in the Bel grade Ground Level and Un der ground Lab o ra to ries, Nucl. Instr. and Meth. in Phys. Res. A591 (2008), pp. 470-475 [3] Dragi}, A., et al., Com par a tive Study of Power Spec - tra of Ground and Shal low Under ground Muon Data, Int. Jour nal of Mod ern Phys ics A, 20 (2005), pp. 6953-6955 [4] Dragi}, A., et al., Pe ri odic Vari a tions of CR Muon In- ten sity in the Pe riod 2002-2004, Pro ceed ings, 21st Eu- A. Dragi}, et al.: The New Set-up in the Bel grade Low-Level and Cos mic-Ray ... 192 Nuclea r Technol ogy & Ra dia tion Protec tion: Year 2011, Vol. 26, No. 3, pp. 181-192

Aleksandar DRAGI], Vladi mir I. UDOVI^I], Radomir BAWANAC, Dejan JOKOVI], Dimitrije MALETI], Nikola VESELINOVI], Mihailo SAVI], Jovan PUZOVI], Ivan V. ANI^IN

NOVA OPREMA U BEOGRADSKOJ LABORATORIJI ZA MEREWE NISKIH AKTIVNOSTI I KOSMI^KOG ZRA^EWA

Beogradska laboratorija sastoji se od dva laboratorijska prostora, jednog na povr{ini i jednog podzemnog, na dubini od 25 metara vodenog ekvivalenta. Detaqno su opisani i ilustrovani potencijali ovih laboratorija za merewe niskih aktivnosti i za kontinuirano merewe mionske i elektromagnetne komponente kosmi~kog zra~ewa, kao i za studije procesa koje ova zra~ewa indukuju u germanijumskim spektrometrima sme{tenim u niskofonskim podzemnim laboratorijama. Sva ova merewa se izvode simultano, novim sistemom za digitalnu spektroskopiju, a podaci se zapisuju doga|aj po doga|aj, i analiziraju posle zavr{enih merewa. Tako|e su prikazani preliminarni rezultati koji su u fazi testirawa opreme dobijeni za fluks brzih neutrona i zaustavqenih miona u povr{inskoj i u podzemnoj laboratoriji.

Kqu~ne re~i: podzemna laboratorija, gama spektroskopija, merewa niskih aktivnosti, ...... ,...... kosmi~ko zra~ewe