First Direct Measurement of the ^{17}O(p,γ)^{18}F Reaction Cross-Section at Gamow Energies for Classical Novae
D. A. Scott, A. Caciolli, A. DiLeva, A. Formicola, M. Aliotta, M. Anders, D. Bemmerer, C. Broggini, M. Campeggio, P. Corvisiero, Z. Elekes, Zs. Fülöp, G. Gervino, A. Guglielmetti, C. Gustavino, Gy. Gyürky, G. Imbriani, M. Junker, M. Laubenstein, R. Menegazzo, M. Marta, E. Napolitani, P. Prati, V. Rigato, V. Roca, E. Somorjai, C. Salvo, O. Straniero, F. Strieder, T. Szücs, F. Terrasi, D. Trezzi
aa r X i v : . [ a s t r o - ph . S R ] O c t First Direct Measurement of the O(p, γ ) F Reaction Cross-Section at GamowEnergies for Classical Novae
D.A. Scott, A. Caciolli, , A. Di Leva, A. Formicola, ∗ M. Aliotta, M. Anders, D. Bemmerer, C. Broggini, M. Campeggio, P. Corvisiero, Z. Elekes, Zs. F¨ul¨op, G. Gervino, A. Guglielmetti, C. Gustavino, Gy. Gy¨urky, G. Imbriani, M. Junker, M. Laubenstein, R. Menegazzo, M. Marta, E. Napolitani, P. Prati, V. Rigato, V. Roca, E. Somorjai, C. Salvo, , O. Straniero, F. Strieder, T. Sz¨ucs, F. Terrasi, D. Trezzi (LUNA Collaboration) SUPA, School of Physics and Astronomy, University of Edinburgh, Edinburgh EH9 3JZ, UK INFN, Sezione di Padova, 35131 Padova, Italy INFN, Laboratori Nazionali di Legnaro, Padova, Italy Dipartimento di Scienze Fisiche, Universit`a di Napoli “Federico II”, and INFN, Sezione di Napoli, Napoli, Italy INFN, Laboratori Nazionali del Gran Sasso, Assergi, Italy Helmholtz-Zentrum Dresden-Rossendorf, Dresden, Germany Universit`a degli Studi di Milano and INFN, Sezione di Milano, Milano, Italy Dipartimento di Fisica, Universit`a di Genova, and INFN, Genova, Italy ATOMKI, Debrecen, Hungary Dipartimento di Fisica Sperimentale, Universit`a degli Studi di Torino, and INFN, Sezione di Torino, Torino, Italy GSI Helmholtzzentrum f¨ur Schwerionenforschung GmbH, 64291 Darmstadt, Germany MATIS-IMM-CNR at Dipartimento di Fisica e Astronomia, Universit`a di Padova, Padova, Italy Institut f¨ur Experimentalphysik III, Ruhr-Universit¨at Bochum, Bochum, Germany Osservatorio Astronomico di Collurania, Teramo, and INFN, Sezione di Napoli, Napoli, Italy Seconda Universit`a di Napoli, Caserta, and INFN, Sezione di Napoli, Napoli, Italy and INFN, Sezione di Milano, Milano, Italy (Dated: June 21, 2018)Classical novae are important contributors to the abundances of key isotopes, such as the ra-dioactive F, whose observation by satellite missions could provide constraints on nucleosynthesismodels in novae. The O(p, γ ) F reaction plays a critical role in the synthesis of both oxygen andfluorine isotopes but its reaction rate is not well determined because of the lack of experimental dataat energies relevant to novae explosions. In this study, the reaction cross section has been measureddirectly for the first time in a wide energy range E cm ≃ −
370 keV appropriate to hydrogenburning in classical novae. In addition, the E =183 keV resonance strength, ωγ =1.67 ± µ eV, hasbeen measured with the highest precision to date. The uncertainty on the O(p, γ ) F reaction ratehas been reduced by a factor of 4, thus leading to firmer constraints on accurate models of novaenucleosynthesis.
PACS numbers: 26.20.Cd; 26.30.-k; 26.50.+x
Classical novae, a frequent phenomenon in our Galaxy,are explained as thermonuclear explosions on the sur-face of white dwarf stars accreting hydrogen-rich mate-rial from less evolved companions in binary star systems[1] and have been proposed as a key source of C, N, , O, and , F isotopes in the Universe [2]. In particu-lar, the short-lived radioisotope F ( t / = 110 min) mayprovide a signature of novae outbursts through the detec-tion of 511 keV γ -ray emission from positron-electron an-nihilation following its β + decay. Indeed, the observationof these γ rays by satellite missions could put constraintson current nova models [3]. Hydrogen burning of O isbelieved to play a key role on the destruction of O andon the formation of F, mainly through the compet-ing reactions O(p, γ ) F and O(p, α ) N. Thus, thethermonuclear rates of both reactions should be deter-mined with a high degree of accuracy directly in the en-ergy region of hydrogen burning in classical novae. Inthis Letter, we report on a four-fold improvement in the O(p, γ ) F reaction rate determination. We have mea- sured the O(p, γ ) F reaction cross section down to thelowest energies to date and within the Gamow window[4] for peak temperatures T = 0 . − . O(p, γ ) F reaction rate isdominated by a direct-capture (DC) reaction mechanismdespite the presence of two narrow resonances at E = 66and 183 keV above the proton threshold in F [5, 6](all energies are in the center-of-mass system, unless oth-erwise stated). In addition, non-resonant contributionsarise from the low-energy tails of two broad resonancesat E = 556 . . O(p, γ ) F reaction rate thus requires theaccurate knowledge of the individual energy dependenceof both resonant and non-resonant contributions.Early investigations by Rolfs [7] reported a constantDC contribution to the S-factor (S DC ) [4, 8] in agree-ment with the four lowest data points measured at en-ergies E = 280 −
425 keV by prompt γ -ray detection[7]. It was later questioned by Fox et al. [5] whetherthese data points were dominated by the DC process or [keV] γ E0 1000 2000 3000 4000 5000 6000 7000 c oun t s pe r k e V → R → R → R → R → R → R T l K B i B i B i B i B i N ) γ C ( p , F ) γ O(p, C ) γ B(p, O ) γα F(p, FIG. 1. Sample spectrum from prompt γ -ray measurements of the O(p, γ ) F reaction at the E = 183 keV resonance.Primary transitions are labelled as R → E x , while secondary transitions are labelled by their energies. Some of the mostprominent background transitions (dashed arrows) are also shown for comparison, together with a time-normalised backgroundspectrum (grey area). affected by the presence of the two broad resonance tails.Thus, in Ref. [5], the S DC was calculated using measuredspectroscopic factors and a realistic Woods-Saxon poten-tial and found to be up to a factor of ∼ E ≃ −
470 keV led to an S DC in agreementwith the predictions by Fox et al. [5] and still about afactor of 2 lower than in Ref. [7]. More recently, the to-tal (i.e., DC plus broad-resonance contributions) S factorwas measured at E = 260 −
470 keV using the DRAGONrecoil separator at TRIUMF [9] and found to be in fairlygood agreement with values by Ref. [6], but consistentlyhigher than the total S factor reported in Ref. [5]. Asfor the E = 183 keV resonance strength, only two val-ues exist to date: ωγ = (1 . ± . × − eV, as de-termined by a prompt γ -ray measurement [5, 10], and ωγ = (2 . ± . × − eV, as determined by the ac-tivation technique [11], disagree at the 95% confidencelevel. The origin of this discrepancy is not understood atpresent, but may be due in part to unobserved gammatransitions in Refs. [5, 10] and/or an inappropriate sub-traction of the DC component in either Refs. [5, 10] orRefs. [11]. Thus, the lack of experimental data at lowenergies and the largely unconstrained S DC -factor haveso far precluded the accurate determination of the ther-monuclear rate for this important reaction.Here, we report on the results of a new and improvedinvestigation of the O(p, γ ) F reaction using both acti- vation and prompt γ -ray detection techniques to addressa key source of discrepancy in existing results. Measure-ments were carried out at the 400kV LUNA accelera-tor [12] of the Laboratory for Underground Nuclear As-trophysics (LUNA) facility, which offers significant im-provements in sensitivity thanks to its low-backgroundenvironment [13]. A proton beam, with currents up to200 µ A on target, entered the target chamber through aliquid-nitrogen cooled copper-pipe biased to -300 V forsecondary electrons suppression. The target was directlywater cooled with de-ionised water. Targets were pre-pared by anodization of Ta backings (0.3 mm thick disks)in isotopically enriched water (66% in O and 4% in O). Full details on target preparation and characteri-sation have been reported in [14]. The target thicknesswas closely monitored for signs of degradation during in-tense proton-beam bombardment by regularly measuringthe thick-target yield profile [4] of the narrow isolated res-onance at E = 143 keV in O(p, γ ) F [14]. Prompt γ rays from the O(p, γ ) F reaction ( Q =5606.5 ± ◦ with re-spect to the beam axis, with the detector’s front face par-allel to the target surface. Energy calibration, full-energypeak efficiency, and total efficiency were determined tak-ing into account corrections for true-coincidence sum-
150 200 250 300 350 400 450 500510152025 present (primary) present (activation) Newton et al., 2010 Rolfs, 1973 Hager et al., 2012 present best fit Fox et al., 2005 Newton S ( k e V b ) E cm (keV) FIG. 2. Total astrophysical S factor as a function of center-of-mass energy for the O(p, γ ) F reaction. Filled symbolsrefer to the activation (circles) and the prompt γ -ray (squares)measurements in the present study. The solid line is the bestfit to our data, while the dotted line labeled “Newton” iscalculated using the DC contribution from Ref. [6] and theresonance parameters from Ref. [5] as used in Ref. [20] (seetext for further details). ming as described in Ref. [15]. Further details on thedata analysis not reported here will be presented in aforthcoming publication [16].The O(p, γ ) F reaction proceeds by populating sev-eral states in F, leading to a complex decay scheme.Measurements were taken at the 183 keV resonance andat several off-resonance energies. At LUNA it was possi-ble to observe and identify a large number of the γ -raytransitions from the entrance channel to F intermediateor ground state (the primaries ) and subsequent direct ormultistep decay of excited states to the F ground state(the secondaries ). A sample γ -ray spectrum showing thequality of our data is given in Fig. 1. Independent deter-minations of the total S-factor were carried out for all pri-mary and secondary transitions. The main primary tran-sitions to F states at E x = 937 , , , , O abundance (3%).The E =183 keV resonance strength ωγ was deter-mined using the thick-target yield approach [4]. Priorto our investigation the E =183 keV resonance had beenobserved to decay to only two states at E x =1080 and937 keV in F, with branching ratios of 60% and 40%,respectively [18]. However, several additional transi-tions were identified in our study, as shown in Fig. 1.Both the resonance strength and the branching ratiosof all observed transitions were treated as free param-eters in a global χ fit to the experimental counts asin Ref. [17]. For the three most dominant transitions,R → →
937 and R → ± ±
2% and 12 ± E =183 keV resonance strength was ωγ =1.70 ± ± µ eV. (Note that for mea-surements at the resonance, the target thickness uncer-tainty is only relevant for the DC subtraction and thecommon uncertainties reduce to 6.6%).An independent determination of both resonant andnon-resonant contributions to the O(p, γ ) F total crosssection was also obtained with the activation technique,i.e. by detecting the 511 keV γ rays from the positronannihilation following the β + decay of F. Targets wereirradiated for several hours to saturate the F activ-ity, using the same setup described earlier but withoutlead shielding. Loss of F from the target during irra-diation was estimated to be below 1% at all bombardingenergies according to GEANT4 Monte Carlo simulations.Activation spectra were recorded every 20 minutes over aperiod of several hours using the low-background facilitySTELLA (SubTErranean Low Level Assay) [19]. The ab-solute efficiency of the detector was measured with a cal-ibrated Sr source which emits a single γ -ray line at 514keV allowing for a direct determination of the efficiencyat the required energy, free from summing corrections.The laboratory background was negligible compared tothe count rate of the activated targets. The irradiatedtargets were placed on top of the detector crystal and aTa absorber, in direct contact with the activated target,was used to fully stop the emitted positrons. Our bestvalue of t / = 109 . ± . F half-life was ob-tained as an average of several fits to experimental dataat different energies and found to be in excellent agree-ment with the literature value t / = 109 . ± .
05 min.The only contaminant to the spectra was found to bethe short-lived N β + emitter. No longer-lived positronemitters were observed after several half-lives. Fromthe number of observed disintegrations at off-resonanceenergies, the total S factor was determined followinga standard procedure described in Ref. [4]. Resultsfrom the activation analysis are shown in Fig. 2. Inaddition to the statistical errors shown, data are af- TABLE I. Reaction rate for the O(p, γ ) F gs reaction as afunction of temperature. Temperature lower limit recommended value upper limitT (GK) N A < συ > (cm mol − s − )0.040 2 . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × − . × +00 . × +00 . × +00 . × +00 . × +01 . × +01 . × +01 . × +01 . × +01 . × +01 . × +02 . × +02 . × +02 . × +02 . × +02 . × +02 fected by an independent systematic uncertainty of 2.9%(sum in quadrature of uncertainties from backscatteringlosses (1%), N contamination (1%), detector efficiency(2.5%)), and by a 7.6% uncertainty in common with theprompt γ -ray data (see above). The E =183 keV reso-nance strength was derived from measurements at ener-gies on the plateau of the thick-target yield after subtrac-tion of a DC contribution of about 18%, which results in a3% DC-correction uncertainty on ωγ . The obtained res-onance strength, ωγ =1.65 ± ± µ eV,is in excellent agreement with the result of the prompt γ -ray measurements. The weighted average of the res-onance strength values from both measurements gives ωγ =1.67 ± µ eV.To obtain an overall total S factor, the activation andthe prompt γ -ray data sets were analyzed in a commonfit procedure. Unfortunately, inclusion of other data sets(e.g. from Ref [9]) in our fit was not possible sincethese seem to be affected by a mixture of systematicand point-to-point uncertainties that prevent a commonand self-consistent analysis of all data sets. Followingthe phenomenological approach of Refs. [5, 20], the fit-ting function included contributions from the two broadresonances at E =557 and 667 keV, with energy depen-dence of the involved partial widths described by the T [GK] -1
10 1 L UN A 〉 v σ 〈 / 〉 v σ 〈 FIG. 3. (Colour online) O(p, γ ) F reaction rate ratio asa function of temperature. The (red) solid line is the ratiobetween the rate from a recent compilation rate (Ref. [20])and the present value. Hatched and shaded areas represent a1 σ uncertainty on present and previous rate, respectively. Inthe relevant nova temperatures T = 0 . − . broad-resonance formalism [4], as well as a constant DCterm. The weak energy dependence of the DC compo-nent was neglected in the present analysis, as in previ-ous works [5, 20], since its effect is beyond the precisionof our data. Finally, preliminary calculations [16] indi-cate that interference effects between resonant and DCcontributions are negligible and have not been included.A proper treatment of the systematic uncertainties [21]required the introduction of three scaling factors, onefor each data set to account for non-common system-atic uncertainties, and one for the common uncertain-ties. Their values were then obtained simultaneously ina modified χ fit (for a similar approach see Ref. [22]).Additional free parameters in the fit were the DC con-tribution and the partial Γ γ width of the 557 keV res-onance. The former parameter was left unconstrainedwhile the latter was treated as a free parameter weighedby a Gaussian probability distribution with expectationvalue and standard deviation given by the literature valueΓ γ, = 0 . ± .
13 eV [23]. The influence of the secondbroad-resonance on the fit quality was extremely smalland, thus, Γ γ, was fixed to the literature value [23].The best fit ( χ = 13 . γ, = 0 .
70 eV and a DCcomponent of S DC = 4 . ± . c act = 1 . c prim = 0 . c com = 0 . O(p, γ ) F reaction rate was cal-culated using the formalism described in Ref. [24] as anincoherent sum of contributions from two narrow reso-nances (the E = 183 keV one from the present work,the E = 66 keV one from the literature [23]) and thecombined contribution of the broad resonances and thedirect capture, this latter obtained as numerical integra-tion of the present S-factor best fit (solid curve in Fig.2). The resulting reaction rate is tabulated in Table Iwith lower and upper limits given by the 68% confidencelevel. A comparison between our rate and that from arecent compilation [20] (Fig. 3) shows an improvementof a factor of 4 in the present rate uncertainty.On the basis of our improved rate, we have exploredpreliminary implications on the abundances of key iso-topes produced by classical novae. For example, ac-cording to Ref. [25] a 25% (1 σ ) uncertainty in the O(p, γ ) F reaction rate leads to a 40 −
50% varia-tion on the calculated yields of O, F and F at T = 0 . − . O, F and F abundances reduces to less than10% and puts firmer constraints towards more accuratenucleosynthesis calculations in novae events. The effectsof our revised rate on the computation of detailed novamodels will be discussed in a forthcoming paper.In summary, the O(p, γ ) F reaction cross section hasbeen measured for the first time in the relevant energyregion of hydrogen burning in classical novae. The astro-physical S factor has been determined both by means ofprompt γ -ray and activation measurements and resultsfrom the two approaches were found to be in excellentagreement. In addition, we have obtained the most accu-rate determination of the E =183 keV resonance strengthto date as ωγ =1.67 ± µ eV. The reaction rate uncer-tainty was reduced by a factor of 4 leading to improvedconstraints on classical novae nucleosynthesis.The authors wish to thank the INFN mechanical work-shop, electronics and chemical laboratories of LNGS, aswell and Dr H. Costantini and Dr C. Mazzocchi for theircontributions to the early stages of the experiments. AC acknowledges financial support by Fondazione Cassa diRisparmio di Padova e Rovigo. ADL acknowledges fi-nancial support by MIUR (FIRB RBFR08549F). Finan-cial support by OTKA K101328, NN83261, DFG (BE4100/2-1), and NAVI is also gratefully acknowledged. ∗ Corresponding author: [email protected] [1] B. Paczy´nski et al. , Acta Astron. 15 (1965) 197[2] J. Jos´e and M. Hernanz, Ap. J. 494 (1998) 680[3] M. Hernanz et al. , Ap. J. 526 (1999) L97[4] C. Iliadis,
Nuclear Physics of Stars (Wiley-VCH, 2007)[5] C. Fox et al. , Phys. Rev. C 71 (2005) 055801[6] J.R. Newton et al. , Phys. Rev. C 81 (2010) 045801[7] C. Rolfs, Nucl. Phys. A 217 (1973) 29[8] The astrophysical S factor is defined as S( E ) = σ ( E ) E exp(2 πη ), with σ the cross section, η the Som-merfeld parameter, and E the interaction energy [4].[9] U. Hager et al. , Phys. Rev. C 85 (2012) 035803[10] C. Fox et al. , Phys. Rev. Lett. 93 (2004) 081102[11] A. Chafa et al. , Phys. Rev. Lett. 95 (2005) 031101, andPhys. Rev. Lett. 96 (2006) 019902(E)[12] A. Formicola et al. , Nucl. Instr. Meth. A 507 (2003) 609[13] H. Costantini et al. , Rep. Progr. Phys. 72 (2009) 086301[14] A. Caciolli et al. , (2012) accepted for publication inEPJA; http://arxiv.org/abs/1210.0327.[15] B. Limata et al. , Phys. Rev. C 82 (2010) 015801[16] LUNA Collaboration, Phys. Rev. C (2012) in preparation[17] G. Imbriani et al. , Eur. Phys. J. A25 (2005) 455[18] D. R. Tilley et al. , Nucl. Phys. A595 (1995) 1[19] C. Arpesella, Appl. Radiat. Isot. 47 (1996) 991[20] C. Iliadis et al. , Nucl. Phys. A 841 (2010) 31[21] G. D’Agostini et al. , Nucl. Instr. Meth. A 346 (1994) 306[22] D. Sch¨urmann et al. , Phys. Lett. B 711 (2012) 35[23] C. Iliadis et al. , Nucl. Phys. A 841 (2010) 251[24] R. Longland et al. , Nucl. Phys. A 841 (2010) 1[25] C. Iliadis et al.et al.