Arxiv:1203.1369V1 [Nucl-Ex]

Home , Isotopes of fluorine

arXiv:1203.1369v1 [nucl-ex] 7 Mar 2012 n sindt h rt1 first the to assigned and gemn ihUD[ USD with agreement xiissignificant exhibits 25 ttsi urn stpshsbe eotd 8kVreso- keV 28 A from reported. decay been has nant isotopes fluorine in states oe acltos[ calculations Model lc h 1 the place nec fthe of each in two reported near systems possi neutron-rich become increasingly has explore it to increased, have facilities beam isotope iin ih orsodt 1 to correspond might sitions 10kV hssget httefis 1 first the that suggests This keV. 1100 ihtehigher-lying the with r nbte gemn ihosrain lcn h 1 the placing observation, with agreement better in are eiet[ periment e.[ Ref. rsn xeietltcnqe.A20 esrmn [ measurement 2004 A techniques. experimental present neutron-rich dripline. neutron the beyond and at those cluding wn otetasto rmpure from transition ye the 30 to over owing for research theoretical and experimental active ihsdu stps[ neutron- isotopes sodium of rich measurements mass from deduced first figurations p † ∗ - ta xiaineeg f44 e [ keV 4249 of energy excitation an at F 47 an France. Caen, Bec 14076 Henri Bvd CEA/DSM-CNRS/IN2P3, GANIL, Br address: Present Vancouver, Mall, Wesbrook [email protected] Canada. 4004 2A3, V6T TRIUMF, Columbia address: Present sd ni eetyol n esrmn fneutron-unbound of measurement one only recently Until h eto-ihrgo around region neutron-rich The h ev urn stpsrpeetsm ftemost the of some represent isotopes fluorine heavy The rs hl excitations. shell cross 3 eot o-nrytasto o rdce yUSD by predicted not transition low-energy a reports ] 4 2 .In ]. / ainlSprodcigCcornLbrtr,Michigan Laboratory, Cyclotron Superconducting National 2 γ N ASnmes 11.r 21.10.Pc 21.10.Dr, numbers: PACS oprsno hsrsl ihUD/SBselmdlpredic model shell USDA/USDB with result this of Comparison eto-non xie tt in state excited neutron-unbound hc h rudsaersnneof resonance state ground the which tt spiaiydmntdby dominated primarily is state + rytastosi aho the of each in transitions -ray Spyrou, 25 .Christian, G. 1 h rudsaeof state ground The 27 ∼ eateto hsc n srnm,Mcia tt Univer State Michigan Astronomy, and Physics of Department , t19 e.SP- ot al Shell Carlo Monte SDPF-M keV. 1997 at 25 26 ,atasto a bevda 777 at observed was transition a F, .F Grinyer, F. G. a sindt 1 a to assigned was F , 0ssesta a emaue with measured be can that systems 20 sd 27 .INTRODUCTION I. 3 7 5 eateto hsc n srnm,AgsaaClee Ro College, Augustana Astronomy, and Physics of Department 7 - 1 stps pcltn htteetran- these that speculating isotopes, F 26 ,wihalwfor allow which ], eateto hsc,CnrlMcia nvriy t Pl Mt. University, Michigan Central Physics, of Department eateto hmsr,Mcia tt nvriy atL East University, State Michigan Chemistry, of Department 1,2 f p , 5 rniincnre narcn ex- recent a in confirmed transition F 2 , .A h vial neste trare- at intensities available the As ]. ofiuainmxn.Additionally, mixing. configuration .J Strongman, J. M. 6 / 6 eateto hsc,CnodaClee ored Minne Moorhead, College, Concordia Physics, of Department 2 hl oe rdcin which predictions model shell ] ,2, 1, + / 4 eateto hsc,Hp olg,Hlad ihgn494 Michigan Holland, College, Hope Physics, of Department 2 xie tt.Ti si poor in is This state. excited ∗ 2, − sd .Frank, N. 28 † N ttsaiigfo proton from arising states omixed to a enosre sa non eoac 22 resonance unbound an as observed been has F .Luther, B. = pcrsoyo Neutron-Unbound of Spectroscopy / 2 0hsbe oi of topic a been has 20 + / sd sd 2 xie tt in state excited 8 3 - 1,2 27 − 25 .I h present the In ]. selcngrtos eew rsn ealdrpr nt on report detailed a present we Here configurations. -shell f p sd .Ash, S. ta xiaineeg f250 of energy excitation an at F , xie tt in state excited 26 6 .Thoennessen, M. 28 - , hl mixing, shell .Mosby, M. f p 27 a rtosre.Adtoal,w eottefis observ first the report we Additionally, observed. first was F N isotopes, F hl con- shell = ( 3 19 .Baumann, T. 20 / ) querel, 2 keV , + ars, itish 27 ble in- 3 at ,7 2, F ] .Mosby, S. 1,2 oaino aho hs eetr ln h emai is axis beam the along detectors The these cm. Fig. in 1044 of shown by each separated of scintillators plastic location of pair cm 227 a by and separated (CRDCs) position-sensi chambers of drift pair readout a cathode through passed beam the target, the emot 8 mg/cm 288 a onto beam a rdcdb rtaccelerating first by produced was ocluaeteba oiino h ecintre yray scint by upstream target The 1010 reaction was triplet. lator the quadrupole on the position through tracing beam the calculate to tt nvriy non ttsin states Unbound Michigan University. at State (NSCL) Laboratory Cyclotron Superconducting yncenrmvlfo emof beam a from removal nucleon by h SLculdccorn [ cyclotrons coupled NSCL the opnnswr elsprtdi nryls esstm of desired time versus The loss energy flight. in separated well were components 254 was tillator of transmission state ground unbound the of observation of the report we paper crmtcauiu eg tisscn mg on to point image second to its according fragments at disperse wedge aluminum achromatic pna11 mg/cm 1316 a upon 28 acu nbrlimrato eesn hog h A1900 the [ through sent separator were fragment reaction beryllium on calcium ryls signal of loss time ergy beam the calculate to used flight were signals these and nal, tt nvriy atLnig ihgn484 USA 48824, Michigan Lansing, East University, State 2 xeiethv enrpre narcn ril [ article recent a in reported been have experiment F .Warren, M. fe h 10,aqarpl rpe antfcsdthe focused magnet triplet quadrupole a A1900, the After h xeietwspromda h National the at performed was experiment The 28 .A DeYoung, A. P. n nubudectdsaein state excited unbound an and F in ed otecnlso htthe that conclusion the to leads tions ( t b ,2 1, iy atLnig ihgn484 USA 48824, Michigan Lansing, East sity, 0 ) ( diinly h agtsitlao eodda en- an recorded scintillator target the Additionally, . 22 .K Smith, K. J. nig ihgn484 USA 48824, Michigan ansing, aat ihgn489 USA 48859, Michigan easant, 0 kIln,Ilni 10,USA 61201, Illinois Island, ck 3 ) 1 I XEIETLDETAILS EXPERIMENTAL II. e bv h rudsaeof state ground the above keV 0 keV. h RCpsto esrmnswr used were measurements position CRDC The . .Weisshaar, D. µ ( 29 27 hc,adtedwsra “agt)scin- (“target”) downstream the and thick, m µ ∆ ea 2MVu h 10 nlddan included A1900 The MeV/u. 62 at Ne E .Ec cniltrrcre iesig- time a recorded scintillator Each m. , oa552 USA 56562, sota 4 28 29 b 2 .E Finck, E. J. 11 ) ecmoe prxmtl %o the of 2% approximately composed Ne F 9 ssoni Fig. in shown As . epouto agt rdcso the of Products target. production Be 3 USA 23, ,2 1, ,wihwstndt piiethe optimize to tuned was which ], 9 2 .Setup A. .Snyder, J. erato agt ptemof Upstream target. reaction Be 2 n .Wersal A. and 10 A 5 eeprmn in experiment he / .The ]. .Gade, A. Z 48 1,2 n mrv separation. improve and Ca 28 29 to fa of ation 27 27 ground F 20 e The Ne. A. , .Terslso the of results The F. 28 48 2 + h aiu beam various the , eepopulated were F ate impinged then Ca ,2 1, 27 o10MVuin MeV/u 140 to 2 F. 29 ebeam Ne 9 ]. tive il- 2

Beam tracking detectors Modular Neutron CsI(Na) Array Array (MoNA) Quadrupole triplet (CAESAR) Sweeper magnet

29Ne (62 MeV/u) 9 Ionization Be reaction target chamber (288 mg/cm2) 93 cm93 cm 71 cm 5 mm scintillator 150 mm scintillator Timing scintillators Charged fragment x x tracking detectors y z y z

FIG. 1. (color online) Diagram of the experimental setup.

put was indicative of the fragment energy. 20 Due to the size and complexity of the setup, separate 32 Mg data acquisition (DAQ) systems were used for MoNA and 102 15 29Ne Sweeper-CAESAR. Events in each DAQ were recorded with Wedge Fragments a timestamp, allowing coincidences to be reconstructed off- line. Although run separately, the triggering of each DAQ was 10 10 controlled by a shared logic module, which allowed for trig- [arbitrary] b ger conditions involving both subsystems. To reduce dead- E Δ 5 time, the experiment required coincidences between MoNA 1 and the 5 mm scintillator located at the back of the Sweeper box. CAESAR detected γ rays in coincidence, but they were 0 not a required trigger condition. 80 85 90 95 100 105 110 tb [ns] B. Data Analysis FIG. 2. Energy loss versus ﬂight time of the secondary beam, show- ing its three components: 29Ne, 32Mg, and various lighter species (“wedge fragments”) produced in the A1900 wedge. 1. Charged Particle Separation

The charged particle measurements allowed for event-by- beam, and the remainder was composed of 32Mg (87%) and event isotope identification after making a variety of correc- various lighter species produced in the aluminum wedge. tions to the data. The first step was to identify the various el- A diagram of the experimental setup is shown in Fig. 1. ements reaching the end of the Sweeper using measurements The experiment consisted of three subsystems, each used to of energy loss and total energy. Energy loss (∆E f ) was ob- measure a different type of reaction residue potentially re- tained from the ionization chamber signal. The fragment time 27,28 sulting from the breakup of neutron-unbound states in F: of flight (t f ) and charge output of the 150 mm scintillator (Q) neutrons, γ rays (from feeding to bound excited states in the were each used as an independent indicator of total energy. daughter), and residual charged particles. Neutrons were de- Figures 3(a) and 3(b) show the element separation in ∆E f -Q tected in the Modular Neutron Array (MoNA) [12], which and ∆E f -t f , respectively. In the final analysis, events were measured their time of flight, position, and the amount of light required to fulfill conditions in both parameter spaces. deposited. γ Rays were detected in the Caesium Iodide Array For a given element, isotopes were separated by construct- (CAESAR) [13], which measured their total energy and time ing a corrected time of flight parameter (tc) indicative of A/Z. of flight. Charged particles were first deflected 43◦ by the The corrections to the time of flight accounted for the vary- Sweeper magnet [14]. They were then detected in a pair of ing paths taken through the Sweeper, and the primary indica- CRDCs, an ionization chamber measuring energy loss, and tors of this path length were the dispersive position (x) and two plastic scintillators. The front face of each scintillator dispersive angle (θx) of the fragment as it exited the magnet. was 40cm×40cm, and each was coupled to four photo-tubes. Additionally, a variety of other parameters (c.f. Table I) were The upstream scintillator was 5 mm thick and recorded a time found to correlate with the time of flight of a given isotope signal; this signal was combined with the time output of the and were included in the corrections. Due to the lack of focus- target scintillator to determine the fragment time of flight. The ing elements, as well as non-homogeneities in the Sweeper’s downstream scintillator was 150 mm thick, and its charge out- magnetic field, it was necessary to consider three-dimensional 3

As shown in Fig. 3(e), plotting e(x,θ ) versus t reveals iso- TABLE I. Correction factors used for isotope separation. To calcu- x f tope bands in two dimensions. From here, an initial corrected late the corrected time of flight, we take the sum of each factor multi- plied by its corresponding parameter and then add this sum to t . The time of flight parameter was calculated by projecting onto the f axis perpendicular to the bands. The time of flight corrections symbols x(y) and θx(θy) respectively refer to the dispersive (non- dispersive) position and angle of the charged fragment as it exits the were then further refined by iteratively removing any correla- Sweeper. The symbol xtrgt(ytrgt) denotes the beam’s dispersive (non- tions between t f and the parameters listed in Table I. dispersive) position on the reaction target. The remaining symbols The final corrected time of flight (tc) for fluorines produced are all introduced in the text. from 29Ne is shown in Fig. 3(f). By fitting this spectrum with Parameter Correction Factor the sum of five Gaussians constrained to have equal width, we determined the 26F-27F cross-contamination to be approxi- x (mm) −5.0595 × 10−2 mately 4%. The factors used in constructingthe corrected time x2 (mm2) −8.97 × 10−4 of flight are listed in Table I, and it should be noted that the x3 (mm3) −3.0 × 10−6 θ −2 most important corrections (in addition to those for x, θx, and x (mrad) +8.0 × 10 2 θ 2 2 −5 their higher order combinations) are those for y and xtrgt. x (mrad ) −1.0 × 10 θ 3 3 −6 x (mrad ) +2.0 × 10 −4 xθx (mm mrad) −1.5 × 10 θ 2 2 −6 2. Decay Energy Calculation x x (mm mrad ) −6.0 × 10 2 2 −6 x θx (mm mrad) −2.0 × 10 2θ 2 2 2 −7 x x (mm mrad ) +1.4 × 10 The decay energy of the breakup of unbound states was y2 (mm2) +1.0 × 10−3 calculated using invariant mass analysis. In Euclidian coor- θ −3 y (mrad) −3.0 × 10 dinates, the decay energy Ed is expressed as −2 xtrgt (mm) +1.7 × 10 −3 ytrgt (mm) +4.0 × 10 E m2 m2 2 E E p p cosθ m m (3) −1 d = q f + n + f n − f n − f − n, tb (ns) +1.0 × 10 Q (arb.) +1.3 × 10−3 where m m E E and p p refer to the mass, energy, ∆E (arb.) +4.0 × 10−3 f ( n), f ( n), f ( n) and momentum of the charged fragment (neutron), respectively, and θ is the openingangle between the two decay products. Charged fragment inputs to Eq. 3 were determined us- θ correlations between t f , x, and x, as well as non-linearities, in ing a partially inverted COSY transformation matrix [16, 17], determining the appropriate corrections. Because of superior which operated on the measured position and angle behind the statistics, the time of flight corrections were determined for Sweeper and the x position of the beam on target. The trans- 32 fluorine elements produced from the Mg beam. These same formation returned the energy and angle at the reaction target, corrections were then used to separate the isotopes of interest, as well as the track length and the target y position. 29 fluorines produced from Ne. The neutron input to Eq. 3 was calculated from time of θ As shown in Fig. 3(c), a three-dimensional plot of t f -x- x flight and position measurements in MoNA using relativis- displays isotope bands. For the purpose of time of flight cor- tic kinematics. The trigger logic was designed such that the θ rections, it is useful to reduce the x- x phase space into a sin- stop for each MoNA time digitizer channel was provided by a θ gle “emittance” parameter e(x, x), as this will allow for cor- delayed signal from the target scintillator. Thus the recorded rections to the flight time to be made in a straightforward way. time signals were a measurement of neutron time of flight (t ). θ θ n To determine e(x, x), the t f -x- x scatter-graph was profiled To calibrate the raw digitizer signals, a linear slope and offset θ by dividing the x- x phase space into small regular rectangu- were applied to each channel. The slopes were determined lar regions and finding the mean t f for each region [15]. This from a pulser run; relative offsets between MoNA bars were profile plot is shown in Fig. 3(d), with the grayscale level rep- determined from cosmic-ray muon tracks; and an overall off- θ resenting mean t f . From here, the location of x as a function set was set from the travel time of prompt γ rays. Vertical of x was fit along the lines of constant t f in the profile. As and lateral positions in MoNA were assumed to be at the cen- shown by the curve in Fig. 3(d), the location of these lines ter of the interaction bar, and the horizontal position was cal- was well-described by a second order polynomial, culated from the time difference between signals measured on either end of the bar. In the case of multiple interactions within MoNA, the earliest hit with t > 40 ns was used in the f (x)= ax2 + bx + c [mrad], (1) n analysis. The cutoff of 40 ns was chosen to eliminate any ran- with a = 0.010391 mrad/mm2 and b = 0.84215 mrad/mm. dom first hits that arrived too early to be prompt neutrons. The final constant c can take on any value; it only causes the A plot of the neutron time of flight to the front face of curve to shift to a different line of constant t . MoNA is presented in Fig. 4, for three conditions: ungated f 26 Once f (x) was determined, e(x,θx) was constructed simply (including all incoming beam components), F produced 29 27 29 as from Ne, and F produced from Ne. In the ungated plot, the peaks from prompt neutrons and γ rays are clearly identi- fiable on top of a random flat background consisting primar- e(x,θx)= θx − f (x). (2) ily of room background γ rays and cosmic-ray muons. When 4

6 6 102 (a) 102 (b)

4 4 10 10 [arbitrary] [arbitrary] f f E E 2 2 Δ Δ 1 1

0 0 0 2 4 6 8 10 45 50 55 60 Q [arbitrary] tf [ns]

100 58 (d) 56 (c) Fit along lines of constant tf 50 54 52 [ns] f 0 t 50 [mrad] x θ 48 46 -50 40 0 θ -40 -100 44 x [mrad] 0 40 -80 -80 -40 -100 -500 50 100 x [mm] x [mm]

(f) (e) 26F 10 120 0 100 8 80 6 27 ) [mrad] F x

Counts 60

, θ -100 x ( 4 e 40

2 20

-200 0 0 45 50 55 40 42 44 4648 50 52 tf [ns] tc [arbitrary]

29 FIG. 3. a) Element separation in ∆E-E. b) Element separation in ∆E-tb. Panels (a) and (b) are both gated on incoming Ne and have the fluorine events circled. c) Three-dimensional plot of tb vs. dispersive position and angle after the Sweeper for fluorine fragments produced from the 32Mg beam. The various bands correspond to different isotopes of fluorine. d) Profile of the three-dimensional plot in (c), including the emittance parameter, e(x,θx), that is determined by fitting lines of constant tb with a second order polynomial. e) Plot of the e(x,θx) parameter determined in (c) versus tb, demonstrating isotope bands in two dimensions. f) Final corrected time of flight (tc) for fluorine elements produced from the 29Ne beam, with 26F and 27F indicated. requiring coincidences with 26,27F, the flat background is es- 3. γ-Ray Measurements sentially eliminated, and the prompt neutron peak dominates the spectrum. CAESAR was calibrated using a variety of standard γ-ray sources [18]. Although a large magnetic shield was placed between it and the Sweeper, CAESAR was still subject to significant fringe fields (on the order of 3 mT) which affected the response of its photo-tubes. To account for this, the array was 5

for the correction, the detection point taken as the center of the ﬁrst interaction crystal, and the emission point assumed 30 (a) n to be the center of the reaction target. To reduce background from random coincidences, only events falling within a spe-

] cific time window were included in the final analysis. Because -3 20 of electronic effects (walk in the leading-edge discriminators used for timing), the time window was implemented as a two- dimensional cut on time of flight versus Doppler-correcteden- ergy. 10 Counts / ns [10Counts / ns γ

C. Modeling and Simulation 0 -50 0 50 100 150 200 Resonant states were modeled by a Breit-Wigner line-shape with an energy dependent width derived from R-Matrix theory [19]. The equation for the line-shape is

80 (b) AΓ (E;Γ ) σ(E;E ,Γ ,ℓ)= ℓ 0 , 0 0 ∆ Γ 2 1 Γ Γ 2 [E0 + ℓ (E; 0) − E] + 4 [ ℓ (E; 0)] 60 (4) where A is an amplitude, E0 is the central resonance energy, Γ 40 0 parameterizes the central resonance width, ℓ is the orbital angular momentum of the resonance, and Γℓ and ∆ℓ are given

Counts / (5 ns) by 20

Γ γ2 ℓ (E)= 2Pℓ (E) 0 0 (5) -50 0 50 100 150 200 ∆ γ2 ℓ (E)= −[Sℓ (E) − Sℓ (E0)] 0 .

The Pℓ and Sℓ functions in Eq. 5 are related to the spherical Bessel Functions, Jℓ(ρ), and their derivatives: 40 (c) ρ 2 2 P = / Fℓ + Gℓ r=a (6) 30 S = ρ F F′ + G G′ / F2 + G2 , ℓ ℓ ℓ ℓ ℓ ℓ r=a πρ 1/2 ρ ℓ πρ 1/2 with Fℓ = ( /2) Jℓ+1/2 ( ) and Gℓ = (−1) ( /2) 20 ρ J−(ℓ+1/2) ( ).

Counts / (5 ns) In addition to resonant states, a non-resonant background is 10 expected in the 27F →26 F + n decay energy spectrum, resulting from the decay (via emission of a neutron with Ed . 3 28 0 MeV) of high-lying continuum states in F to high-lying 27 26 -50 0 50 100 150 200 states in F that subsequently feed the ground state of F. 26 tn [ns] The F fragment can then be detected in coincidence with the first neutron, giving rise to the background distribution. This background was modeled as a Maxwellian distribution FIG. 4. Neutron time of flight to the front face of MoNA, for the of beam velocity neutrons, following conditions: a) ungated; b) in coincidence with 26F; c) in coincidence with 27F. ε Θ ε Θ3 −ε/Θ f ( ; )= Aq / e , (7) calibrated with the Sweeper set to the rigidity at which the ex- with the temperature Θ a free parameter. This model provides periment was performed. Furthermore, to account for poten- a good fit to the observed non-resonant data and has been em- tial hysteresis effects, a recalibration run using a 88Y source ployed in number of other invariant mass measurements, for was taken any time the field of the Sweeper was changed dur- example [20–24]. ing the experiment. Broadening due to experimental resolution and acceptance For γ-ray events depositing light in multiple crystals, the was accounted for in a Monte Carlo simulation of the exper- total deposited energy was calculated using an “add-back” iment. In the simulation, the kinetic energy of the incoming 29 technique [13]. The in-beam data were Doppler corrected; Ne beam was modeled as a Gaussian with E0 = 62.1 MeV/u 6 and σE = 1.72 MeV/u, clipped at E < 64.5 MeV/u. The beam angle and position were also modeled as Gaussian with σx = σ σ σ 104 11 mm, θx = 4.0 mrad, y = 9.0 mm, and θy = 1.1 mrad. 60 1σ Additionally, the dispersive angle and position were given a L 103

-ln 102 correlation of θx/x = 0.0741 mrad/mm. The angle and position of the incoming beam were determined from position 40 101 measurements in the two CRDC detectors upstream of the re- 0.2 0.3 0.4 0.5 0.6 action target. The beam energy was determined by comparing E0 [MeV] measured and simulated distributions in the two downstream 20 Counts / (270 keV) CRDC detectors for runs where the reaction target was re- moved. The 9Be(29Ne,27,28 F) reactions were treated in the Goldhaber Model [25] including a small friction term [26] to 0 degrade the beam energy by 0.6%. The transport of charged 0 0.5 1 1.5 2 2.5 3 fragments through the Sweeper was simulated using a third Decay Energy [MeV] order COSY transformation matrix, produced from measurements of the Sweeper’s magnetic field [27]. FIG. 5. (color online) Measured relative energy spectrum for 26F+n The resolution of charged particle position and angle mea- coincidences. The filled squares with error bars are the experimental data, the dashed red curve is the result of a 380 keV resonant surements was modeled as Gaussian, with σpos = 1.3 mm and σ 0 8 mrad. These resolutions were determined from simulation, the shaded grey curve is a simulation of the Maxwellian ang = . Θ data taken with a tungsten mask shadowing the CRDC de- non-resonant background ( = 1.48 MeV), and the solid black curve is the sum of the resonant and non-resonant models, with a reso- tectors. The primary acceptance cut concerning the charged δ nant/total fraction of 33%. The inset is a plot of the negative log- particles was the requirement that they pass through the = likelihood as a function of central decay energy, with each point min- ±150 mm active area of the downstreamCRDC. Neutron time imized with respect to all other free parameters. of flight resolution was modeled as Gaussian with σ = 0.3 ns, and the neutron x-position resolution was modeled as a sum of two Laplacian functions: bution and an ℓ = 2 Breit-Wigner resonance, using the technique outlined in Section II C. In the fit, the resonance energy −|x/σ1| −|x/σ2| e e E , resonance width Γ , Maxwellian temperature Θ, and res- p1 · + (1 − p1) · , (8) 0 0 2σ1 2σ2 onant/total fraction f , were all allowed to vary freely. In order to extract E the parameter of interest, a profile log-likelihood σ σ 0, with 1 = 16.2 cm, 2 = 2.33 cm, and p1 = 53.4%. The was constructed by scanning a range of E values and plotting σ σ 0 form of Eq. 8 and the parameters 1, 2, and p1 were de- the negative log-likelihood (−ln[L]) minimized with respect termined from shadow bar measurements and GEANT3 sim- to the other free parameters (Γ0, Θ, and f ). This profile like- ulations [28]. As mentioned, the neutron y and z positions lihood curve is displayed in the inset of Fig. 5, and it reaches were assumed to be at the center of the detection bar, result- a clear minimum at E0 = 380 keV. The nσ confidence inter- ing in a uniform uncertainty of ±5 cm. The overall resolution 2 28 27 vals were determined from the ln[Lmax/L] ≥ n /2 limits. As and acceptance for the decay of F into F + n has already indicated on the figure, the 1σ and 2σ confidence intervals been presented in Ref. [9], and the corresponding shapes are were ±60 keV and +130 keV, respectively. The best-fit values essentially identical in the case of 27F∗ breakup. −120 of the other parameters were determined to be Γ = 10 keV, Due to the low statistics of the present data set, an unbinned 0 Θ = 1.48 MeV, and f = 33%. The simulated best fit curves maximum likelihood technique was used for parameter esti- are superimposed on the data in Fig. 5, with the dashed red mation [29]. This technique involves forming a small range, curve representing the 380 keV resonance, the shaded grey R around each experimental data point and then summing i, curve the Maxwellian background, and the solid black curve the number of weighted Monte Carlo points that lie within their sum. A comparison between simulation and data is also the volume. To marginalize systematic errors resulting from shown for selected intermediate parameters (neutron time of Monte Carlo fluctuations and non-linearities within the R , the i flight, fragment kinetic energy, neutron-fragment opening an- generated model sets were made large ( 3 106 events), and ∼ × gle, and neutron-fragment relative velocity) in Fig. 6. the volume size was chosen to be small (0.05 MeV). The presumption of ℓ = 2 decay is based on a pure single- particle model in which the least-bound neutron resides in the 0d shell. In reality, configuration mixing and shell evolu- III. RESULTS AND DISCUSSION 3/2 tion could lead to significant contributions from decay with other orbital angular momenta. Separate analyses using ℓ = 1 27 A. F Excited State and ℓ = 3 resonances yield results that do not differ signifi- cantly from the ℓ = 2 case. The lack of sensitivity to ℓ val- The black squares in Fig. 5 show the measured decay en- ues is largely due to experimental resolution, which is lim- ergy spectrum of 26F + n coincidences. As mentioned previ- ited primarily by uncertainty of the reaction position within ously, we expect a non-resonant contribution in the 26F + n the 9Be target. The width of the measured resonance is al- data, so they were fit with the sum of a Maxwellian distri- most completely determined by experimental response, over- 7

+ 30 (a) 3 (2 ) (?) 380 keV + 2 (1 ) 20 [MeV]

* +

E 1 (1/2 ) - 10 (1/2 ) 0 (5/2+)

0 26F + n 27F 40 60 80 100 t [ns] n FIG. 7. Summary of experimentally known levels in 26,27F, including the present observation of an unbound excited state in 27F at 2500 keV, decaying to the ground state of 26F. The shaded grey boxes 40 (b) around the various levels indicate the total uncertainty in their place- ment relative to the 27F ground state. The dashed lines surrounding 30 the presently observed 2500 keV level represent the uncertainty on 20 the decay energy only, and the total uncertainty also includes that of the 27F 1n separation energy. All bound excited information is from π 10 [3], ground state energies are from [30], and ground state J are from [31]. 0 50 60 70 T [MeV/u] f shadowing any differences that might arise from varying the 30 ℓ value. Contribution from ℓ = 0 decays might also be possi- (c) ble, but such decays cannot be separated from the Maxwellian background since the resolved lineshape of the two models 20 is very similar for small absolute scattering lengths (|as| . 5 fm). A scattering state near threshold (larger |as|) is clearly not present since the data display no enhancement at low de- 10 cay energy. Only two counts were observed in CAESAR in coincidence 0 with 26F + n (Eγ = 760 and 1180 keV). In the case of 100% 0 2 4 6 8 10 branching to a bound excited state in 26F, roughly 50 counts θ [degrees] would be expected in CAESAR, based on the approximate γ- ray detection efﬁciency of 30% [13]. Thus the observation γ 40 (d) of only two rays in CAESAR indicates that the presently observed decays feed the ground state of 26F, allowing for 30 an unambiguous assignment of the observed resonance to an excited state in 27F. The most recent mass measurements of 20 26,27F[30] place the 26F ground state 2120(210) keV above the ground state of 27F, so we assign the presently observed 10 380(60) keV resonance to a 2500(220) keV excited level in Counts / (0.5 cm/ns) Counts / (0.5 degree) Counts / (MeV/u) Counts / (2 ns) 27F. Fig. 7 presents this newly observed level along with the 0 26,27 -4 -2 0 2 4 other measured states in F[3, 4, 30, 31].

vn - vf [cm/ns] To interpret our observations, we have performed shell model calculations using the USDA and USDB interactions [32], which operate in the traditional sd model space (0d5/2, FIG. 6. (color online) Comparison of 26F n simulation and data + 1s1/2, and 0d3/2 for both protons and neutrons). The calcula- for a) neutron time of flight; b) charged fragment kinetic energy; c) tion results are compared with experiment in Fig. 8. As seen neutron-fragment opening angle; d) neutron-fragment relative veloc- in the figure, each calculation predicts three or more states in ity. The filled black squares are the experimental data, and the solid the same energy region as our observation. Extending the cal- black, dashed red, and shaded grey curves depict the same simulation components as in Fig. 5. culations to include p f shell components would only compli- cate the situation since opening up the model space increases the available number of excited state configurations. The assignment of the observed resonance to a specific state is not possible because the reaction (1p-1n removal) does not pref- erentially populate one state over the others. 8

3.5 + 5/2 + 3 + 3/2 9/2 + (a) ? + 9/2 15 2.5 7/2 + 3/2+ 7/2 2 1/2+ 1.5 1/2+ 10 (MeV) * + E 1 1/2 5 0.5 1/2- Counts / (2 ns) 0 5/2+ 5/2+ 5/2+ 0 40 60 80 100

Experiment USDA USDB

£ ¤ [ns]

27 FIG. 8. Measured excited levels in F compared to predictions of (b) the USDA and USDB shell models. The ﬁrst two measured excited 20 states are both from [3], while the state at 2.5 MeV is from the present 15 work. 10

5 62 Counts / (MeV/u) 30 2σ

L 61 1σ 0

-ln 50 60 70 60 20 Tf [MeV/u] 59 0.1 0.2 0.3 15

E0 [MeV] (c) 10 Counts / (270 keV) 10

0 0 0.5 1 1.5 2 2.5 3 5 Decay Energy [MeV] Counts / (0.5 degree) 0 FIG. 9. (color online) Relative energy spectrum for 27F + n coinci- 0 2 4 6 8 10 dences. The ﬁlled squares with error bars are the experimental data, θ [degrees] the dashed red curve is the the 220 keV resonance simulation, and the dotted blue curve is the 810 keV simulation. The solid black curve is the sum of the 220 keV and 810 keV resonances, with the relative (d) contribution of the 220 keV resonance at 28%. The inset shows the 20 proﬁle log-likelihood as function of the lower resonance energy.

10 B. 28F Binding Energy Counts / (0.5 cm/ns) 28 As discussed in Ref. [9], the measured F decay energy 0 is best described as a sum of two independent ℓ = 2 Breit- -4 -2 0 2 4

Wigner resonances, with the lower resonance at 220(50) keV vn - vf [cm/ns] (Γ0 ≡ 10 keV), the upper resonance at 810 keV (Γ0 ≡ 100 keV), and the lower resonance composing 28% of the total area. As with 27F, the width of each measured resonance was FIG. 10. (color online) Comparison of 27F + n simulation and data dominated by experimental resolution, making sensitivity to for a) neutron time of flight; b) charged fragment kinetic energy; c) the resonance ℓ value minimal. An ℓ = 0 scattering state was neutron-fragment opening angle; d) neutron-fragment relative veloc- excluded based on incompatibility with the measured data, ity. The filled black squares are the experimental data, and the solid black, dashed red, and dotted blue curves depict the same simulation and a non-resonant Maxwellian background was not expected components as in Fig. 9. since 28F was populated directly by one-protonknockout from 29Ne. The measured 28F decay energy spectrum is presented in Fig. 9, along with the best fit two-resonance simulation and the profile log-likelihood curve. Additionally, Fig. 10 shows a data-simulation comparison for neutron time of flight, 9

IV. CONCLUSIONS AND OUTLOOK 1.5 USDA USDB In conclusion, we have used the technique of invariant mass 1 spectroscopy to make the ﬁrst determination of the 28F binding energy at 186040(200) keV. Additionally, we have ob- [MeV] th 0.5 served a neutron-unbound excited state in neighboring 27F,

BE with 2500(220) keV excitation energy. - 0 27 exp Interpretation of the F state in terms of shell model pre-

BE dictions is difficult due to the large number of levels predicted -0.5 near 2500 keV and uncertainty in the reaction mechanism used to populate 27F∗. The level structure of 27F is relevant -1 9 10 11 12 13 14 15 16 17 to a variety of open questions in nuclear physics, including the transition from pure sd to mixed sd-p f neutron config- Z urations and its associated consequences (such as the large FIG. 11. (color online) Difference between experimental and theoret- oxygen-fluorine dripline shift of six or more neutrons [34]). ical (USDA, USDB) binding energies for N = 19 isotones, 9 ≤ Z ≤ Additionally, it has been suggested [3, 35] that proton p-sd 17. The error bars on the data points represent experimental errors cross-shell excitations could play a role in the structure of only. The blue dotted and red dashed bands represent the respective low-lying 27F excited states, possibly in tandem with sd- fp 170 and 130 keV RMS deviations of USDA and USDB interactions. shell breaking on the neutron side. As such, it would be inter- Experimental values, save for Z = 9 which is from the present work, esting to revisit unbound excited states in 27F experimentally, are taken from [30] if reported there; otherwise they are from the using a direct reaction mechanism that can selectively pop- 2003 Atomic Mass Evaluation [33]. Figure reproduced from Ref. ulate specific states. Possible reactions include one- or two- [9]. proton knockout (from 28Ne or 29Na) and 26F(d, p) in inverse kinematics. The measured 28F binding energy indicates a low-Z bound- ary of the island of inversion at N = 19. It would be interesting to further explore this mass region to see if this trend con- tinues. Extension of the present technique to lighter N = 19 isotones (Z ≤ 8) would be very difficult, if not impossible, since they are all unbound by three or more neutrons [36– fragment kinetic energy, neutron-fragment opening angle, and 40]. However, a similar technique could potentially be used neutron-fragment relative velocity. No γ rays were recorded in the N = 20 isotonic chain by performinga direct mass mea- 29 in CAESAR in coincidence with 27F+n, and around30 would surement of bound F. For this purpose, the precision obtain- be expectedin the case of 100%branchingto excited 27F. This able with time-of-flight techniques at current in-flight radioac- indicates that the observed resonances feed the 27F ground tive beam facilities would likely be sufficient. Such a mea- state. surement would be particularly interesting since the SDPF-M Monte Carlo Shell Model predicts 29F to have a very large intruder occupation of 91.5% (62.7% two-particle, two-hole excitation and 28.8% four-particle, four-hole) [41]. Measuring The present observation of the 28F ground state as a its mass would provide the first experimental data on 29F for 220(50) keV unbound resonance can be combined with the comparison with theory and help to better explain the evo- 27F mass measurement of Ref. [30](27F atomic mass excess lution of shell structure in the low-Z (< 10) region around equal to 24630(190) keV) to calculate the 28F binding energy N = 20. as 186040(200) keV. By comparing measured binding energies with the predictions of the UDSA/USDB shell model, which does not allow for mixing between sd and p f shell con- ACKNOWLEDGMENTS figurations, it is possible to qualitatively determine the contribution of p f shell “intruder” components in the ground state The authors thank the NSCL operations staff for provid- of a given nucleus. Such a comparison is shown in Fig. 11 for ing a high-quality beam throughout the experiment and the N = 19 isotones with 9 ≤ Z ≤ 17. As seen in the figure, the NSCL design staff for their efforts in the construction of a agreement is very good for the heavier isotones closer to sta- magnetic shield for CAESAR. We are also grateful to B. A. bility (Z ≥ 13), while it becomes dramatically worse for the Brown and A. Signoracci for their assistance with shell model isotones with 10 ≤ Z ≤ 12 which lie within the island of in- calculations. Finally, we thank the NSCL Gamma Group and version. At Z = 9, the good agreement between USDA/USDB MoNA Collaboration for their effort in setting up and support- and experiment is dramatically recovered, indicating that in- ing the experiment. Funding for this work was provided by truder components play a minimal role in the ground state the National Science Foundation under grants No. PHY-05- structure of 28F. This suggests the existence of a low-Z bound- 55488, No. PHY-05-55439, No. PHY- 06-51627, No. PHY- ary (or “shore”) of the island of inversion beginning at Z = 9. 06-06007, No. PHY-08-55456, and No. PHY-09-69173. 10

[1] C. Thibault, R. Klapisch, C. Rigaud, A. M. Poskanzer, [19] A. M. Lane and R. G. Thomas, Rev. Mod. Phys. 30, 257 (1958). R. Prieels, L. Lessard, and W. Reisdorf, Phys. Rev. C 12, 644 [20] F. De´ak, A. Kiss, Z. Seres, G. Caskey, A. Galonsky, and (1975). B. Remington, Nuc. Instr. Meth. in Phys. Res. A 258, 67 [2] E. K. Warburton, J. A. Becker, and B. A. Brown, Phys. Rev. C (1987). 41, 1147 (1990). [21] A. Schiller et al., Phys. Rev. Lett. 99, 112501 (2007). [3] Z. Elekes et al., Phys. Lett. B 599, 17 (2004). [22] C. R. Hoffman et al., Phys. Rev. C 83, 031303 (2011). [4] M. Stanoiu et al., Phys. Rev. C 85, 017303 (2012). [23] C. R. Hoffman et al., Phys. Rev. Lett. 100, 152502 (2008). [5] B. H. Wildenthal, Prog. Part. Nuc. Phys. 11, 5 (1984). [24] A. Spyrou et al., Physics Letters B 683, 129 (2010). [6] B. A. Brown and B. H. Wildenthal, Annu. Rev. Nuc. Part. Sci. [25] A. Goldhaber, Phys. Lett. B 53, 306 (1974). 38, 29 (1988). [26] O. Tarasov, Nuc. Phys. A 734, 536 (2004). [7] Y. Utsuno, T. Otsuka, T. Mizusaki, and M. Honma, Phys. Rev. [27] N. Frank, Spectroscopy of Neutron Unbound States in Neutron C 60, 054315 (1999). Rich Oxygen Isotopes, Ph.D. thesis, Michigan State University, [8] N. Frank et al., Phys. Rev. C 84, 037302 (2011). East Lansing, Michigan (2006). [9] G. Christian et al., Phys. Rev. Lett. 108, 032501 (2012). [28] W. A. Peters, Study of Neutron Unbound States Using the [10] F. Marti, P. Miller, D. Poe, M. Steiner, J. Stetson, and X. Y. Modular Neutron Array (MoNA), Ph.D. thesis, Michigan State Wu, AIP Conference Proceedings 600, 64 (2001). University, East Lansing, Michigan (2007). [11] D. J. Morrissey, B. M. Sherrill, M. Steiner, A. Stolz, and [29] D. Schmidt, R. Morrison, and M. Witherell, Nuc. Instr. Meth. I. Wiedenhoever, Nuc. Instr. Meth. in Phys. Res. B 204, 90 in Phys. Res. A 328, 547 (1993). (2003). [30] B. Jurado et al., Phys. Lett. B 649, 43 (2007). [12] T. Baumann et al., Nuc. Instr. Meth. in Phys. Res. A 543, 517 [31] G. Audi, O. Bersillon, J. Blachot, and A. H. Wapstra, Nuc. (2005). Phys. A 624, 1 (1997). [13] D. Weisshaar, A. Gade, T. Glasmacher, G. Grinyer, D. Bazin, [32] B. A. Brown and W. A. Richter, Phys. Rev. C 74, 034315 P. Adrich, T. Baugher, J. Cook, C. Diget, S. McDaniel, (2006). A. Ratkiewicz, K. Siwek, and K. Walsh, Nuc. Instr. Meth. in [33] G. Audi, A. H. Wapstra, and C. Thibault, Nuc. Phys. A 729, Phys. Res. A 624, 615 (2010). 337 (2003). [14] M. Bird, S. Kenney, J. Toth, H. Weijers, J. DeKamp, [34] Y. Utsuno, T. Otsuka, T. Mizusaki, and M. Honma, Phys. Rev. M. Thoennessen, and A. Zeller, IEEE Trans. Appl. Supercond. C 64, 011301 (2001). 15, 1252 (2005). [35] M. Kimura and N. Furutachi, Phys. Rev. C 83, 044304 (2011). [15] In practice, this was done using the TH3::Project3DProfile [36] O. Tarasov et al., Phys. Lett. B 409, 64 (1997). method of the ROOT data analysis package. For more informa- [37] H. Sakurai et al., Phys. Lett. B 448, 180 (1999). tion, see Chapter 3 of the ROOT users guide: http://root. [38] M. Langevin, E. Quiniou, M. Bernas, J. Galin, J. C. Jacmart, cern.ch/download/doc/3Histograms.pdf. F. Naulin, F. Pougheon, R. Anne, C. Dtraz, D. Guerreau, [16] K. Makino and M. Berz, Nuc. Instr. Meth. in Phys. Res. A 558, D. Guillemaud-Mueller, and A. C. Mueller, Phys. Lett. B 150, 346 (2006). 71 (1985). [17] N. Frank, A. Schiller, D. Bazin, W. Peters, and [39] D. Guillemaud-Mueller et al., Phys. Rev. C 41, 937 (1990). M. Thoennessen, Nuc. Instr. Meth. in Phys. Res. A 580, 1478 [40] M. Fauerbach, D. J. Morrissey, W. Benenson, B. A. Brown, (2007). M. Hellstr¨om, J. H. Kelley, R. A. Kryger, R. Pfaff, C. F. Powell, [18] The sources and their respective γ-ray lines (in keV) were: and B. M. Sherrill, Phys. Rev. C 53, 647 (1996). 133Ba (356), 137Cs (662), 22Na (1275), 88Y (898, 1836), 56Co [41] P. Fallon et al., Phys. Rev. C 81, 041302 (2010). (517, 846, 1771, 2034, 2598, 3272).