actinide and ultra-heavy abundances in the local galactic ... - IOPscience

The Astrophysical Journal, 747:40 (14pp), 2012 March 1 C 2012.

doi:10.1088/0004-637X/747/1/40

The American Astronomical Society. All rights reserved. Printed in the U.S.A.

ACTINIDE AND ULTRA-HEAVY ABUNDANCES IN THE LOCAL GALACTIC COSMIC RAYS: AN ANALYSIS OF THE RESULTS FROM THE LDEF ULTRA-HEAVY COSMIC-RAY EXPERIMENT J. Donnelly1,2 , A. Thompson2 , D. O’Sullivan2 , J. Daly2 , L. Drury2 , V. Domingo3,4 , and K.-P. Wenzel3,5 1 Dublin Institute of Technology (DIT), School of Physics, Kevin Street, Dublin 8, Ireland School of Cosmic Physics, Dublin Institute for Advanced Studies, 31 Fitzwilliam Place, Dublin 2, Ireland 3 European Space Research and Technology Centre (ESTEC), Keplerlaan 1, Postbus 299, 2200 AG Noordwijk, The Netherlands Received 2011 March 24; accepted 2011 December 6; published 2012 February 13 2

ABSTRACT The LDEF Ultra-Heavy Cosmic-Ray Experiment (UHCRE) detected Galactic cosmic rays (GCRs) of charge Z 70 in Earth orbit with an exposure factor of 170 m2 sr yr, much larger than any other experiment. The major results include the first statistically significant uniform sample of GCR actinides with 35 events passing quality cuts, evidence for the existence of transuranic nuclei in the GCR with one 96 Cm candidate event, and a low 82 Pb/78 Pt ratio consistent with other experiments. The probability of the existence of a transuranic component is estimated as 96%, while the most likely 92 U/90 Th ratio is found to be 0.4 within a wide 70% confidence interval ranging from 0 to 0.96. Overall, the results are consistent with a volatility-based acceleration bias and source material which is mainly ordinary interstellar medium material with some recent contamination by freshly synthesized material. Uncertainty in the key 92 U/90 Th ratio is dominated by statistical errors resulting from the small sample size and any improved determination will thus require an experiment with a substantially larger exposure factor than the UHCRE. Key words: cosmic rays – ISM: abundances Online-only material: color figures

≈400 m2 sr days equivalent to 1.1 m2 sr yr. However, the charge assignments were distorted by the registration temperature effect (only discovered after these experiments; Thompson et al. 1979; O’Sullivan & Thompson 1980) and by other factors such as the unavailability of direct ultra-heavy calibration (accelerated beams of ultra-heavy ions had not yet been developed). Consequently, the actinide to subactinide ratio was overestimated (O’Sullivan et al. 1986) and the aggregation of the results was rendered particularly problematical by the different thermal histories of the various flights. A major step forward was the launch into Earth orbit of Ariel 6 and HEAO-3 in 1979 June and 1979 September, respectively. Ariel 6 employed a spherical gas scintillator and Cherenkov detector design with a geometry factor of ≈2 m2 sr at 55◦ orbital inclination and achieved an exposure of 635 m2 sr days after event selection, i.e., 1.7 m2 sr yr. (Fowler et al. 1987). Three actinide candidates were claimed. HEAO-3 featured an ionization chamber and Cherenkov counter detector array with ≈4 m2 sr geometry factor at 43.◦ 6 orbital inclination (Binns et al. 1985). No actinide candidates were detected during an exposure of ≈6 m2 sr yr. In both cases data acquisition ceased within about two years of launch. The Trek experiment (Westphal et al. 1998), using passive detector technology, was based on a 1.2 m2 array of glass track-etch detectors mounted outside the Mir Space Station in 1991 June and retrieved in 1994 January (first third) and 1995 November (remaining two-thirds). An exposure of ≈10 m2 sr yr was achieved in a 51.◦ 6 inclination orbit and six actinide candidates were found. The development by NASA of the retrievable, largely passive, Long Duration Exposure Facility (LDEF) orbiting vehicle provided a unique opportunity to expose very large collecting areas (>10 m2 ) of passive detectors for an extended period (>1 yr). The Ultra-Heavy Cosmic-Ray Experiment (UHCRE) was based on solid-state nuclear track detector (SSNTD) stacks using Lexan polycarbonate as the main detector element. As a

1. INTRODUCTION The primary motivation for investigating the charge composition of ultra-heavy (Z 70) Galactic cosmic rays (GCRs) is to search for signs of both nucleosynthesis and decay processes unique to these elements. For example, the actinide6 group (88 Z 103) and the elements of the platinum group (hereafter referred to as “Pt” and defined by 74 Z 80) are mainly synthesized by the r-process, while the elements of the lead group (hereafter referred to as “Pb” and defined by 81 Z 83) are mainly s-process products. The relative abundances of actinide elements can also act as nucleosynthetic clocks to infer the mean time since their nucleosynthesis (note that this is unrelated to the use of unstable spallation secondaries to infer the confinement time). Study of the chemical composition of the GCRs in the ultraheavy region has always been hampered by the very low fluxes involved and the difficulty of getting an adequate exposure factor (the product of geometry factor and time). The relative abundance of a typical ultra-heavy nucleus is ≈10−6 that of 26 Fe while actinides are a further two orders of magnitude less abundant. An early attempt involved a series of pioneering long-duration, heavy payload (several days, ≈1000 kg) balloon flights at intermediate latitudes from which the results were aggregated (see, for example, Fowler et al. 1977a, 1977b). The passive detectors employed were nuclear emulsion and/or polymer track detectors (mainly the latter). The accumulated exposure of eight flights at ≈3.8 g cm−2 atmospheric depth was 4 Also at Grupo de Astronomia y Ciencias del Espacio (GACE), IPL, University of Valencia, Valencia, Spain. 5 Also at President Kennedylaan 228, 2343 GX Oegstgeest, The Netherlands. 6 The International Union of Pure and Applied Chemistry recommends that the elements from actinium (charge 89) to lawrencium (charge 103) be referred to collectively as the actinoids, but the older term actinides is still allowed and will be used throughout this paper as the name more familiar to most physicists.

1

The Astrophysical Journal, 747:40 (14pp), 2012 March 1

Donnelly et al.

result of the Challenger disaster the LDEF mission was extended to nearly six years. This enabled the UHCRE to obtain a very much larger exposure factor than any other experiment (170 m2 sr yr at 28.◦ 5 orbital inclination). The UHCRE measured 35 actinide events, currently the only statistically significant (>10 events) uniform sample of actinide cosmic-ray nuclei. 2. EXPERIMENTAL DETAILS 2.1. The LDEF Mission The LDEF was designed and built by the NASA Langley Research Center to accommodate experiments for long-term exposure to the space environment. It was a low-cost, reusable, unmanned, and free-flying spacecraft designed to transport experiments into space via the Space Shuttle and be retrieved from Earth orbit on a later flight for analysis. The LDEF was deployed in Earth orbit on 1984 April 7 during the STS-41C Challenger mission. It was retrieved on 1990 January 9 by STS-32 (shuttle Columbia) after spending 69 months in space, 58 more than planned. LDEF deployment and retrieval occurred close to solar minimum and solar maximum, respectively. The orbit was almost circular (perigee 473.9 km, apogee 480.0 km) and at an inclination of 28.◦ 5. Just before retrieval, the orbital altitude had decayed to 328 km and the LDEF was only weeks from uncontrolled re-entry. The geomagnetic cutoff for ultra-heavy GCRs in this orbit was ≈1.5 GeV nucleon−1 or a dimensionless particle velocity of β ≈ 0.92 (Thompson et al. 1993). The LDEF was configured as a 12-sided, 4.3 m diameter, 9.1 m long aluminium open grid frame; see Figure 1(a) and Levine (1991). The largest experiment on the LDEF was the UHCRE which accounted for over 20% of the entire payload mass. 2.2. The Ultra-Heavy Cosmic-Ray Experiment The UHCRE was housed in experiment trays, 1.2 m2 in area, which were distributed around the surface of the LDEF. There were 16 UHCRE-designated trays, each containing three pressurized aluminium cylinders. Each cylinder contained four SSNTD stacks as shown in Figure 1(b). Thus, the experiment totalled 192 detector stacks which yielded a total collecting area of ≈10.2 m2 . When in orbit, the LDEF was three-axis gravity-gradient stabilized, with its main axis parallel to the local earth vertical, so that all peripheral experiment trays (including those of the UHCRE) viewed horizontally. Allowing for Earth shadowing, the total UHCRE exposure factor for high-energy ultra-heavy GCRs was 170 m2 sr yr. To provide both mechanical stability and thermal insulation, the detector stacks were embedded in sets of four inside cylindrical rigid polyurethane foam resin (Eccofoam) mouldings. These were placed inside the experiment’s 48 cylindrical aluminium pressure vessels (diameter 25.4 cm, length 117 cm, and wall thickness 0.24 cm; Thompson et al. 1979). These vessels contained a dry oxygen–nitrogen–helium mixture (20:70:10) at 1.0 bar, except for one, which was vented to space. Examination of the vented detector stacks revealed poor signal strength (probably due to a lack of oxygen) which rendered them unusable. Post-flight analysis revealed that no gas leakage had occurred in the sealed cylinders after 5.8 years in space. Thermal design was a priority because of the registration temperature effect (see Section 3.2). Passive temperature control and thermal decoupling maintained the detector stack temperatures at least 20◦ C cooler than the average temperature of the LDEF itself and largely independent of its temperature fluctuations (Thompson

Figure 1. (a) The LDEF spacecraft with the locations (shaded boxes) of UHCRE experiment trays. (b) A typical UHCRE-designated experiment tray. (c) A typical UHCRE stack configuration, in this case, type A.

et al. 1990). Tray temperature measurements were recorded for the first 390 days of the LDEF’s 2105 day flight. A thermal analysis of the spacecraft showed that the 16 experiment trays had mean temperatures between −15.5◦ C and −23◦ C, with an overall mean of −20◦ C. Standard deviations in tray temperature varied from 2.2◦ C to 6.8◦ C, averaging at 4.3◦ C (Sampair 1991). After retrieval, de-integration of the UHCRE hardware was carried out in a clean room at the Kennedy Space Center in early 1990. The UHCRE experiment trays were then shipped to the European Space Research and Technology Centre (ESTEC) in Noordwijk, the Netherlands, except for six cylinders which were sent for post-flight calibration at the Lawrence Berkeley Laboratory (LBL) Bevelac heavy-ion accelerator in California. Corresponding pre-flight calibration exposures employing the same cylinders had been carried out at LBL approximately 10 months prior to launch. At ESTEC the aluminium cylinders were opened and examined, and in 1990 August the detector stacks were transferred to the Dublin Institute for Advanced Studies (DIAS) for analysis. The detectors spent approximately eight months at NASA and ESTEC facilities at room temperature before being placed in long-term storage at DIAS at a temperature of −31◦ C. 2


Donnelly et al.

the actinide tracks were fully measured. Before imposing quality cuts, a total of 919 events were measured in the 58 stacks that were fully examined with an estimated further 1648 events in the remaining 104 stacks which were scanned for actinides giving a nominal total of 2567 events in the 162 stacks studied. Once removed from cold storage, the UHCRE stack detectors were etched in an aqueous sodium hydroxide (NaOH) solution of normality 6.25 ± 0.01 N saturated with Lexan etch byproducts. The solution also contained 0.05% Dowfax 2A1 surfactant which improved the surface quality of the etched Lexan plates. The solution continuously circulated about the detector plates at a temperature of 40.00 ± 0.01◦ C. Two etching parameters were measured. The bulk etch rate, VG , is the rate at which the detector surface is eroded by the etchant. The tracketch rate, VT , is the rate at which the latent cosmic-ray track is eroded by the etchant. The amount by which VT exceeds VG provides a measure of the ionization damage produced by the incident particle which can be related to its charge and velocity.

2.3. Detector Specifications The UHCRE SSNTD stacks had a surface area of 26.5 cm × 20.0 cm and were composed of various combinations of sheets (or plates) of Lexan polycarbonate (each of thickness ≈250 μm), lead (≈500 μm), CR-39 (≈500 μm), and Tuffak polycarbonate (≈250 μm). The Lexan polycarbonate plates were used to record incident cosmic rays with Z/β > 65. Lexan (stoichiometric formula C16 H14 O3 , density 1.2 g cm−3 ) has a high registration threshold (Z/β > 60) making it ideal for studying ultra-heavy cosmic rays while filtering out more abundant lower-charge nuclei. Typically ≈70 Lexan plates were used per detector stack, and these plates were randomly selected from the total sample (of about 15,000) to assemble each stack. The lead plates were interleaved periodically to serve as velocity degraders and electron strippers, see Figure 1(c). Because the nuclear interaction length of fast ultra-heavy nuclei is 10 times greater in lead than in polycarbonate, the lead plates increased the probability that cosmic rays could lose significant energy in passing through the entire stack without suffering a nuclear interaction. Type A stacks had six lead absorbers, type B had four, and type C had a variable number and included CR-39 and Tuffak polycarbonate plates. The CR-39 and Tuffak plates were not used in the present work. In all, an average stack had a thickness of approximately 5.6 g cm−2 , the Lexan equivalent thickness for a 1.0 GeV nucleon−1 ultra-heavy ion being 4.1 g cm−2 (Thompson et al. 1979). The latent (unprocessed) damage trail in a track detector may be susceptible to thermal annealing, i.e., fading, if stored at a relatively high temperature. Some annealing has been observed in Lexan at temperatures 40◦ C for four weeks, but no detectable effects were seen for storage at 23◦ C for 27 weeks (Adams & Beahm 1981). Research at DIAS has shown that latent track stability in the UHCRE is excellent. Twelve UHCRE detector stacks were exposed to 0.96 GeV nucleon−1 92 U beams from the heavy-ion accelerator at the LBL, California, before and after the LDEF mission (1983 May and 1990 May, respectively). Using the pre-flight 92 U beam as calibration, the post-flight ions were successfully identified as having charge, Z = 92.8 ± 1.3 (Bosch 1995). This stability of track response in the UHCRE detectors over the seven-year mission confirms extensive earlier DIAS studies (see, e.g., O’Sullivan et al. 1987; Domingo et al. 1990) which saw no indication of long-term aging of latent tracks in Lexan, even when stored at room temperatures up to 25◦ C. Other research (Bosch 1995) has confirmed that long-term thermal evolution effects can be completely avoided by storing the track detectors at or below −25◦ C. Annealing effects and latent track stability under storage are thus not issues for the UHCRE, but they should not be confused with the registration temperature effect which is relevant (see Section 3.2).

2.5. Track-etch-rate Monitors (TERMs) The effects of random etch-tank variations on VT values were monitored with track-etch-rate monitors (TERMs). TERMs were plates from SSNTDs which had been exposed to heavy-ion accelerator beams (both an 79 Au beam at 10.1 GeV nucleon−1 and a 82 Pb beam at 158 GeV nucleon−1 were used). These plates were included with all actinide etches to serve as a standard reference. The resulting TERM tracks were measured after each actinide etch to obtain an estimate of the tank’s etching strength. No statistically significant variation of VT with etch-product concentration, etching time, or dip angle (the angle of the cosmic ray’s trajectory relative to the surface of the detector plate) was observed. However, despite all efforts significant variations did occur between different etches as discussed in Section 3.3. 2.6. Quality Cuts Quality cuts were applied to the raw sample of events. Events were eliminated from the sample if: 1. there was evidence of a nuclear interaction within the stack, 2. the event was not fully contained within the stack and entered or exited the edge of the stack, 3. the trajectory dip angle was shallower than 30◦ , that is the angle between the track and the stack normal was greater than 60◦ , and 4. the estimated charge was less than 70. Of the 919 events found in the 58 fully measured detector stacks, 868 events were of Z 70 and 637 of these survived the remaining quality cuts. In the 162 stacks that were examined for actinides, 35 good quality actinide events were found. The actinide measurements were individually corrected for VT variations during the etching procedure with data from TERMs (see Section 3.3).

2.4. Event Selection and Measurement Of the 192 detector stacks flown in the UHCRE, only 162 were used in the present work. Thirty stacks were rejected because they were either saturated with calibration beams from heavy-ion accelerators, had non-standard compositions, or were in the one cylinder vented to space. A total of 58 stacks were pseudo-randomly selected from the remaining 162 for full measurement of all events by optical microscopy. These provided a relative-abundance Z 70 charge spectrum, including some actinides. From the remaining 104 usable stacks, the so-called ammonia scanning method (Fleischer et al. 1975) was used to rapidly locate all high-ionization events and only

2.7. Fragmentation Corrections Fragmentation of incident GCR nuclei in the detector housing will reduce the measured abundance of all elements in the sample in a charge-dependent manner, and so distort the charge abundance spectrum. Thus, the elemental abundances must be adjusted upward to their pre-fragmentation levels outside the LDEF spacecraft. Fragmentation in the detector also produces spallation products which increase the measured flux of the lower charges and further distort the spectrum. 3


Donnelly et al.

There are five layers of the UHCRE apparatus which can cause fragmentation of incoming cosmic rays. The first is a layer of silver-coated Teflon ((C2 F4 )n , of thickness 0.0272 g cm−2 ) that covers each experiment tray. An aluminium cylinder (wall thickness 0.65 g cm−2 ) surrounds the Eccofoam mount (thickness 0.374 g cm−2 ) for the detector stacks. The stacks themselves consist of Lexan polycarbonate plates (average total thickness 2.10 g cm−2 ) interleaved with a varying number of lead plates (detected GCRs pass through an average of 5.6 lead plates with total surface density 3.2 g cm−2 ). The average dip angle was 54◦ . Obtaining the cosmic-ray abundances at the top of the detector requires accurate partial (σpartial ) and total (σtotal ) cross-sections for the appropriate projectile, projectile energy, and target. Nilsen et al. (1995) provide parameterizations of fragmentation cross-sections based on relativistic 36 Kr and 47 Ag nuclei in targets from hydrogen to lead. Equations (11) and (14) therein were used to calculate σtotal and σpartial , respectively. The nuclear interaction cross-sections for an 79 Au projectile ion were assumed to scale as A2/3 for other elements. The projectile’s kinetic energy was assumed to be 1.5 GeV nucleon−1 , close to the modal energy of cosmic rays in the UHCRE sample. In the case of this experiment, fragmentation varied the relative abundances of 79 Au and 92 U (for example) by 14%. This was mostly due to total cross-section values, as the partial crosssection generally contributes less than a 2% effect. The results of the correction can be seen in the Appendix.

2.9. Calibration and Charge Assignment Calibration of the UHCRE SSNTDs rests on an empirical relationship between the measurable track-etch rate VT and the ionization energy deposition J along the track of the form: VT = aJ b .

(1)

We assume the ionization energy deposition J to be given by the restricted energy loss (REL) model (Benton & Nix 1969), with the cutoff energy (ω0 ) set to 1 keV. The REL for fast, heavy ions in dense absorbers is a corrected Bethe–Bloch equation (Waddington et al. 1985). Most of the correction terms used in that equation are not relevant to the UHCRE detectors, with the exception of the density effect. An implicit assumption of the standard Bethe–Bloch formula is that the absorber is a dilute gas. However, in a dense solid-state track detector, the incident ion cannot be considered to act on one atom at a time (Ahlen 1980), so the medium can become polarized. The density effect correction (δ) accounts for such absorber polarization, which lessens the energy loss at high velocities (typically by ≈10% for the cosmic rays encountered by UHCRE). This gives an expression for dE J =− (2) d ξ ω VG where θ is the trajectory dip angle (subjected to a quality cut at 30◦ ; see Section 2.6). Detector efficiency drops as VT sin θ approaches VG , but the effect is negligible for θ > 30◦ . There is also an efficiency effect at low charges near the threshold for registration in Lexan (Z/β > 65) where the stacks only register sub-relativistic (and thus more highly ionizing) particles. Such sub-relativistic particles are however strongly suppressed by the geomagnetic cutoff so that there is an abrupt drop in detector efficiency below charge 65. This drop-off in detection efficiency is of course the very property which makes the detectors so suitable for studies of the ultra-heavy composition (see Section 2.3). The effect is negligible for Z 70. In conclusion, with the applied quality cuts, the instrumental acceptance can be taken as charge independent for the purpose of ultra-heavy composition studies.

(3)

where ξ is the distance traveled in g cm−2 , dξ = ρdx, where ρ is the mass density of the medium and x is the linear distance. me c2 is the rest mass of the electron (0.511 MeV). Iadj is the mean ionization potential of the medium. This value is semi-empirical and determined by applying the Bragg rule of additivity to the stoichiometric components of the medium. Zeff is the effective charge of the ionizing particle. Zeff ≈ Z[1 − exp(−130β/Z 2/3 )]. Though derived for nuclear emulsions this relation can be used for other solids except for very heavy ions at very low velocities (Benton & Nix 1969; Pierce & Blann 1968). √ γ is the Lorentz factor 1/ 1−β 2 . δ is the density effect correction computed from the parameterization of Sternheimer & Peierls (1971). Detector stacks were exposed to 92 U and 79 Au calibration beams (used to calibrate for actinide and subactinide events, respectively) at two different temperatures. These exposures yielded the relationship between VT and the kinetic energy, E, and charge Z of the incident particle within the detector. The calibrations were carried out at various exposure temperatures ranging from −78◦ C to +18◦ C to allow adjustment for the registration temperature effect (see Section 3.2). Empirically derived range-energy tables for the ideal proton (Barkas & Berger 1964) with a suitable heavy-ion extension (Benton & Henke 1969) provided a correlation between the energy and charge of an ion and its residual range in a detector, R (i.e., the distance traveled before the ion stops). Range-energy (R, E) and energy-ionization (E, J ) values were calculated for all 70 Z 100 elements, providing a table of R, J, E relationships across a wide energy spectrum. These tables were

7 There is of course an implicit assumption here that all chemical species have essentially identical power-law spectra in rigidity or energy per nucleon, and while there is no reason to suspect that this is not the case (and it is observed to be the case for lighter species) it is an assumption for the ultra-heavy species.

4


Donnelly et al.

combined with the VT , E, Z measurements from calibration. In this way, R, J, E, VT tables were compiled for all ultra-heavy elements at all relevant temperatures. Linear interpolation was used to compensate for the registration temperature effect (see Section 3.2) in which VT varies with the temperature of the SSNTD at the time of track registration. A cosmic ray’s VT can be measured at various points along its trajectory through the detector (i.e., at various residual ranges, R). In theory, a comparison between the VT , R curve of the ion and those of the calibration particles then allows charge identification, exactly as in the classical residual-range versus ionization technique. However, if the cosmic ray penetrates the entire detector without coming to a halt, its residual range (distance to its stopping point) cannot be easily determined and another method must be used. This was the case with the UHCRE sample and so the gradient method was employed. The fractional etch-rate gradient, G, is defined as (Fowler et al. 1977a, 1977b) G=

1 d VT . VT d x

Figure 2. Veff -gradient calibration curves at −20◦ C with three measured cosmic-ray events illustrating the method of charge assignment. Events are identified by stack number and an event number within each stack; thus 39/4 is event 4 in stack 39. (A color version of this figure is available in the online journal.)

(4)

It is thus a measure of the rate of change of VT with respect to distance traveled by the ion. If two etch rates, VT 1 and VT 2 , are measured at two points separated by a distance x, then G can be estimated by the central difference formula G≈

2 VT 2 − VT 1 . x VT 2 + VT 1

at 960 MeV/u and the model then tested using 79 Au ions of 1.125 GeV nucleon−1 in another UHCRE stack as test events. The model assigned a nominal charge to these ions of Z = 79.1 ± 1.10, demonstrating that it could successfully operate from charge 92 down to 79. To test the energy coverage, UHCRE stacks were further exposed to ions from an 79 Au beam of lower (100 E 400 MeV nucleon−1 ) and higher (10.3 E 10.6 GeV nucleon−1 ) energies; these were identified by the model as charge Z = 78.3 ± 1.0 and 78.9 ± 1.0 ions, respectively (Bosch 1994). This successful identification implies that the values of the exponents a and b in the calibration are essentially energy independent in the range of energies encountered by the UHCRE (a few GeV nucleon−1 ). These tests demonstrate that the calibration model used is reliable from the actinide region to the 78 Pt–82 Pb region (across at least 13 charge units) and from energies as low as 100 MeV nucleon−1 to as high as 10.6 GeV nucleon−1 . All beam exposures were conducted with the detector stacks in their original experimental configurations (i.e., still sealed inside Eccofoam and their pressure vessels).

(5)

The second quantity to be defined is the effective track-etch rate, Veff , given by the harmonic mean of the two end point etch rates: 2VT 1 VT 2 . (6) Veff = VT 1 + VT 2 In an SSNTD stack of n plates, a cosmic-ray track will leave 2n etch cones and the overall mean value of G is best estimated by n

2(V2n+1−i − Vi ) 1 ¯ Wi , (7) G = n (V2n+1−i + Vi )xi i=1 Wi i=1 where xi is the separation between cones 2n + 1 − i and i, Wi ∝ xi2 is a weighting factor, and Vi is the track-etch rate of the cone i. Thus, the gradients of the two cone pairs furthest from the center of the detector are weighted preferentially. Similarly, the overall value of Veff in such a stack of detector plates is the generalized harmonic mean of the individual etch rates, 1 1 1 = . 2n i=1 Vi V¯eff

3. ERROR ANALYSIS Three major error sources contribute to the overall error in charge identification in the UHCRE detectors: intrinsic detector imperfection, temperature dependence of track registration, and varying etching conditions. The magnitudes of these errors are shown in Table 1, as is their total, summed in quadrature. Throughout this paper, the errors quoted in both text and figures correspond to 1σ .

2n

(8)

Two different data sets are required. First, the R – J – E – VT relations outlined above are used to generate theoretical Veff , G curves for incident ions of all relevant charges. Second, measured values of the cosmic-ray event’s VT and R are used to ¯ point for each event. The curve closest to the generate a (V¯eff , G) point indicates the charge of the incident ion. This is illustrated ¯ point for in Figure 2. It should be emphasized that the (V¯eff , G) each cosmic-ray event incorporates the information content of many (typically 32) independent track-etch rate measurements (i.e., signal strength measurements). The parameters a and b (in Equation (1)) were determined using data from a UHCRE stack exposed to a beam of 92 U ions

3.1. Intrinsic Errors Intrinsic detector errors are caused by inhomogeneities in the detector material and may also be caused by intrinsic fluctuations in the track formation process. Their effect was determined empirically by exposing detector material to a beam of ≈10.5 GeV nucleon−1 79 Au ions. Lead degraders in the detector stack reduced the energy of the beam to 2 GeV nucleon−1 . Samples of this exposed polycarbonate were then processed, yielding hundreds of individual cone-pair measurements. The resulting 5


Donnelly et al.

Table 1 Charge Assignment Error Contributions for Actinide and Pt–Pb Ions Expressed in Charge Units Z region

Error Sign

Intrinsica

RTEb

Detector Processing

Uncorrected ΔZ

Corrected ΔZ c

Actinides (Z 88)

+ −

1.1 1.2

0.6 0.6

1.0 1.0

1.6 1.7

1.3 1.3

Pt–Pb (70 Z < 88)

+ −

0.6 0.9

0.6 0.6

1.1 1.1

1.4 1.5

··· ···

Notes. a Actinide errors are upper limits, since the fractional error in measuring ΔV is smaller for these higher-ionization events. T b Pt–Pb errors from the registration temperature effect (RTE) are upper limits because this effect is smaller for lower-ionisation events. c After correcting actinide events for variations in detector processing using TERMs. Table 2 Statistical and Monte Carlo Calculated Systematic Errors for Various Abundance Ratios Based on the UHCRE Dataa Ratio

Systematic Error (%)

Statistical Error (%)

Total (%)

“Pt”/“Pb” Actinides/“Pt”

±9.1b ±4.5

Actinides/subactinides

±4.2

±11.2 +20.5 −17.5 +20.4 −17.4

±14.4 +21.0 −18.1 +20.8 −17.9

Notes. a “Pt” = (74 Z 80), “Pb” = (81 Z 83), “Subactinides” = (74 Z 87), and “Actinides” = (Z 88). b Note that the systematic error on this ratio is not merely a combination of the separate “Pt” and “Pb” errors; since these charge ranges are contiguous the ratio is twice as sensitive to charge-assignment errors at the common boundary. Figure 3. Cosmic-ray event assigned a charge of 79.0 with calibration curves for 79 Au, 77 Ir, and 81 Tl at various energies. (A color version of this figure is available in the online journal.)

point-spread function of charge assignment yielded a measure of the detector’s intrinsic error. 3.2. Temperature-induced Errors

all actinide etches using TERMs (Section 2.5). A three-year study discovered a maximum variation in VT between etches of approximately 10% from the mean. These data were used to adjust the VT measurements of all actinide events and correct for the effect of etching variations. This eliminated one contribution to the charge-assignment error of up to ±1.0 in the actinide region.

As indicated in Section 2.3, the influence of temperature on latent track response in the UHCRE can be restricted to the RTE. The strength of a latent ion track in an SSNTD is a function of the temperature of the stack at the time of registration. Higher registration temperatures reduce the latent track strength that results from a given ionization (Thompson et al. 1979; O’Sullivan & Thompson 1980). Furthermore, the magnitude of the RTE itself increases at higher ionizations (Thompson et al. 1986). Empirical studies of the RTE indicate that a variation in nominal charge of 0.14 ◦ C−1 in Lexan for uranium ions at 960 MeV nucleon−1 (Thompson & O’Sullivan 1984). Since the average cosmic ray in the UHCRE sample is of both lower charge and higher energy, this figure safely overestimates the magnitude of the RTE. Using this figure implies that temperature variations contribute a 1σ error on charge assignment of ±0.6 in the actinide region and considerably less in the (lower-ionization) 79 Pt to 82 Pb region. Calibration VT measurements were made at two different temperatures (see Section 2.9). Interpolation was used to adjust the calibration curves to the modal temperature of each UHCRE detector stack as given by the spacecraft thermal model.

3.4. Total Errors on the Abundances of Cosmic-ray Charge Groups To calculate the errors on the abundances of GCRs in important charge groups, a Monte Carlo simulation was used. Each event in the UHCRE sample was assigned a random error from a Gaussian distribution of standard deviation equal to the systematic charge-assignment error of the UHCRE (+1.4 −1.5 in the subactinide region and ±1.3 for the actinides). The Monte Carlo simulation generated hundreds of synthetic spectra and determined the mean abundance over these spectra for each charge bin in 70 Z 87. The standard deviation about this mean abundance was inferred and taken as the systematic error on the abundance of that bin. These systematic errors were combined with statistical errors (Gehrels 1986) to determine the total charge-abundance error on the various Z groups. The effects of these errors on the more important charge abundance ratios are presented in Table 2. The addition of systematic errors reveals that the total charge-assignment error (especially on actinide abundances) is dominated by statistical error (see Tables 2 and 5). Figure 3 provides a graphical illustration that charge differences of ±2 can be clearly distinguished. Figures 4

3.3. Detector Processing Errors The UHCRE detectors were all etched in an aqueous NaOH 6.25N solution at 40◦ C saturated with etch by-products. Bulk etch-rate (VG ) and track-etch rate (VT ) were monitored in each of the 40 separate chemical etches. Variations in VG contributed negligibly to the charge-assignment error. VT was recorded in 6


Donnelly et al.

Figure 6. LDEF subactinide data binned to the nearest nominal charge and corrected for systematic instrumental effects as described in the text. Note that the “abundances” have been normalized relative to the entire sample and that the data are oversampled. The lead and platinum peaks are clearly seen. The error bars indicate statistical errors only.

Figure 4. Cosmic-ray event estimated to be charge 82.3 with 78 Pt calibration curves at various energies. (A color version of this figure is available in the online journal.)

Figure 7. LDEF actinide data binned to the nearest nominal charge and corrected for systematic instrumental effects as described in the text. Note that the abundances have been normalized on the same basis as the subactinides for ease of comparison. The error bars indicate statistical errors only.

Figure 5. Cosmic-ray event estimated to be charge 82.3 with 82 Pb calibration curves at various energies. (A color version of this figure is available in the online journal.)

Figure 6 shows that in the subactinide region, the most significant features of the charge spectrum are prominent 78 Pt and 82 Pb peaks followed by a fall in abundance beyond the 82 Pb peak (as elements immediately heavier than 83 Bi are quite unstable). Figure 7 shows that in the actinide region there are plausible 90 Th and 92 U peaks plus a possible transuranic presence, including one 96 Cm candidate event (estimated Z ≈ 95.9). The 96 Cm event was processed and measured twice, using a large number of different detector plates on the second occasion. Both independent attempts yielded exemplary measurements and concurring Z-assignment results. In view of its interest detailed plots of the data for this one event are shown with predictions for uranium (Figure 8), for plutonium (Figure 9), and for curium (Figure 10). The 96 Cm candidate is also shown in Figure 11 which illustrates charge separation at 2.0 GeV/u, the best-fit energy. Figure 12 shows the candidate and three other actinide events plotted with Veff -gradient calibration curves. In this context, it should be noted that the UHCRE stack ensemble is basically an ionization detector. Since there is no independent velocity determination, there may be difficulty in distinguishing slower lower-charge events from faster higher-charge events. One can only measure the change in ionization rate along the available

and 5 demonstrate the ability to separate platinum (78 Pt) from lead (82 Pb). 4. RESULTS 4.1. Charge Abundance in the Cosmic Rays The actinide (88 Z 103) and subactinide (70 Z 87) cosmic-ray events were selected in slightly different ways. The latter were derived from the 58 stacks in which all ultra-heavy nuclei events were measured (a total of 637 events, including 8 actinides, survived the quality cuts in these stacks). Actinide candidate events were measured in all 162 detector stacks which represented the entire examined detector area of the UHCRE (35 actinide events survived the quality cuts, including the 8 detected in the fully scanned subset). These data were corrected for fragmentation within the detector. The result is the relative elemental abundances of the Z 70 GCRs at the top of the Earth’s atmosphere (Figures 6 and 7 which show statistical errors only). The subactinide data are tabulated in the Appendix along with details of the actinide events. 7


Donnelly et al.

Figure 8. Cosmic-ray event 126/16 with 92 U calibration curves at various energies. It is clearly impossible to achieve a satisfactory fit with uranium. (A color version of this figure is available in the online journal.)

Figure 10. Cosmic-ray event 126/16 with 96 Cm calibration curves at various energies. This is clearly a very good fit to the data and the preferred interpretation. (A color version of this figure is available in the online journal.)

Figure 11. Cosmic-ray event 126/16 with 92 U, 94 Pu, and 96 Cm at 2 GeV/u. This illustrates charge separation for a given energy per nucleon and graphically demonstrates the fit to curium at 2 GeV/u. (A color version of this figure is available in the online journal.)

Figure 9. Cosmic-ray event 126/16 with 94 Pu calibration curves at various energies; the fit at 1.5 GeV/u is not good, but cannot be totally excluded. (A color version of this figure is available in the online journal.)

path length of a few g cm−2 within a detector stack, in other words the gradient in ionization rate, G. It is clear that this difficulty increases as one approaches relativistic velocities and the gradient tends to zero. This is graphically illustrated for the 96 Cm candidate in Figure 12 and in Figure 9 where a lowprobability 1.5 GeV/u94 Pu fit is shown.

The cosmic-ray 82 Pb/78 Pt abundance ratio is a key indicator of whether GCR seed nuclei are fractionated by first ionization potential (FIP) or volatility during acceleration (See Meyer et al. 1997; Ellison et al. 1997; Meyer et al. 1998; Ellison et al. 1998). In the lower part of Table 4, the UHCRE’s results are compared with those reported from other experiments and those derived from meteoritic and photospheric analysis. It is convenient to compare the ratios of charge groups, such as “Pt” (74 Z 80) and “Pb” (81 Z 83). CI-chondrite meteoritic data were obtained from the following sources: A&E (Anders & Ebihara 1982), A&G (Anders & Grevesse 1989), and Lodders (Lodders 2003). Cosmic-ray data were obtained from combined HEAO-3 and ARIEL-6 spacecraft results (Binns et al. 1989) and the Trek experiment (Westphal et al. 1998). The measured GCR composition must differ from that at the source due to spallation during propagation through the interstellar medium (ISM). The maximum likely range at the source of the key “Pb” to “Pt” ratio (taking propagation factors

4.2. Abundance Ratios Abundance data from a variety of sources, including this work, are compiled in Tables 3 and 4. Note that while the three sets of solar system abundances are all quoted relative to silicon, this is impossible in the case of the GCR data sets which each have an essentially arbitrary normalization. Thus the absolute values should not be compared, but the ratios of the charge groups can be compared and the most significant of these are presented in the last three rows of Table 4. The errors quoted are total errors, i.e., both statistical and systematic. 8


Donnelly et al.

Figure 13. “Pb”/“Pt” abundance ratios from various sources. Systematic and statistical error bars are shown on experimental data (errors on the value recommended by Lodders are negligible), except for the Trek datum which includes only statistical errors (systematic errors are small in this case). The triangular points are upper limits at the source, assuming minimum and maximum propagation changes to the ratio (factors of 1.3 and 2.6, respectively, for the lower and upper triangles). (A color version of this figure is available in the online journal.)

Figure 12. Cosmic-ray event 126/16 and three other actinide events with Veff –G calibration curves. This again illustrates the quality of the charge assignment of event 126/16. (A color version of this figure is available in the online journal.) Table 3 Solar System Reference Abundances Charge

A&E (1982)

A&G (1989)

Lodders (2003)

70 71 72 73 74 75 76 77 78 79 80 81 82 83

0.243 0.0369 0.176 0.0226 0.137 0.0507 0.717 0.6600 1.37 0.186 0.52 0.184 3.15 0.144

0.2479 0.0367 0.151 0.0207 0.133 0.0517 0.675 0.661 1.34 0.187 0.34 0.184 3.15 0.144

0.2484 0.03572 0.1699 0.02099 0.1277 0.05254 0.6738 0.6448 1.357 0.1955 0.4128 0.1845 3.258 0.1388

90 92

No stable nuclides 84 Z 89 0.0335 0.0335 0.0090 0.0090

from Meyer et al. 1997) is shown in Figure 13. The UHCRE results indicate that at the 1σ level, the “Pb”/“Pt” ratio in the cosmic rays at Earth is at most 32% of that found in CI-chondrite source material. By assuming a commonly accepted propagation change (×1.65, see, e.g., Binns et al. 1989) a comparison between the “Pb”/“Pt” ratio in the GCRs and that found in CI-chondrite material can be made. With this assumption, the UHCRE results show that at the 1σ level, the “Pb”/“Pt” ratio in the cosmic rays at the Earth is at most just 52% of that found in CI-chondrite source material. Even using the most extreme propagation factor (×2.6) the “Pb”/“Pt” ratio in the cosmicray source material is still at most 82% of the minimum level found in CI-chondrite material. Clearly 82 Pb is depleted in the GCR. Interpreting this value as a low s-process abundance is problematic, given the lack of any such effect in the first and second s-nuclei and r-nuclei peaks. Depending on propagation conditions the FIP model (e.g., Havnes 1971) predicts a GCR source enhancement of between 1.3 and 2.6 times solar. Since FIP-biased source material would have a “Pb”/“Pt” ratio slightly

0.03512 0.009306

Table 4 Comparison of Abundance Compilations from Various Sources Solar System (SS)a Charge Groupb

A&E (1982)

A&G (1989)

HS “Pt” “Pb” Actinides

0.48 3.64 3.48 0.04

0.46 3.39 3.48 0.04

Galactic Cosmic Ray (GCR) Lodders (2003)

This Work

Binns et al. (1989)

Westphal et al. (1998)

0.48 3.46 3.58 0.04

0.15 ± 0.02 0.65 ± 0.04 0.16 ± 0.02 0.016 ± 0.003

1.9 5.4 1.6 0.13

··· 135 36 6

0.30 ± 0.081 0.0241+0.022 −0.010

0.27± 0.05 0.0444+0.027 −0.018

0.0186+0.018 −0.010

0.035+0.021 −0.014

Key abundance ratios “Pb”/“Pt” Actinides/“Pt”

0.9553 0.0117

1.0267 0.0125

1.0338 ± 0.1166 0.0128 ± 0.0014

0.25 ± 0.04 0.025 ± 0.005

Actinides/Subactinides

0.0060

0.0062

0.0063 ± 0.0006

0.020 ± 0.004

Notes. a The solar system abundances are all normalized, as is conventional, to an Si abundance of 106 . They are based on meteoritic studies but Lodders also includes data from studies of the solar photosphere. b HS = heavy secondaries 70 Z 73 (these are mainly spallation products from the Pt peak), “Pt” = (74 Z 80), “Pb” = (81 Z 83), “Subactinides” = (74 Z 87), and “Actinides” = (Z 88).

9


Donnelly et al.

Figure 14. Actinide/“Pt” abundance ratios from various sources. Systematic and statistical error bars are shown on experimental data (errors on the value recommended by Lodders are negligible), except for the Trek datum which includes only statistical errors (systematic errors are small in this case). The triangular point and dotted error bars denote the Trek value excluding its two unassigned actinides. Note that all experimental data are values observed in the solar system and have not been corrected for interstellar propagation. (A color version of this figure is available in the online journal.)

Figure 15. Likelihood contours (at the 90%, 70%, and 50% level) from the Monte Carlo analysis for a three-component composition (90 Th, 92 U, and 94 Pu). The black dot indicates the location of the maximum likelihood composition in this ternary diagram.

higher than solar (≈1.6), the UHCRE value is actually lower than that predicted by the FIP model. This is shown graphically in Figure 13. In Table 4, the ratios of actinides to the Pt group [(Z 88)/(74 Z 80)] and to the subactinides [(Z 88)/(74 Z 87)] are shown. Both the actinide / “Pt” and actinide/subactinide ratios measured in the UHCRE are (like those of HEAO-ARIEL and Trek) higher than typical solar system values, although the effect of GCR propagation on these ratios is uncertain. The same data are shown graphically in Figures 14 and 17.

numerical measure of how far the spectrum differed from the mean. The same test was then applied between the actually measured spectrum and the mean sample. 7. The fraction of the simulated data set (i.e., synthetic spectra) that lies further from the mean than the real (i.e., empirically measured) data set is calculated. For example, if it is 5% of the total sample then to a 95% confidence level, the real data are incompatible with the original hypothetical composition, in the sense that only 5% of the simulated spectra deviate as much from the mean spectrum as the actual measurement.

5. MONTE CARLO LIKELIHOOD ANALYSES

The above procedure was repeated for various hypotheses of

To quantify the significance of these various abundance indicators a Monte Carlo technique was used to estimate likelihood intervals. This allowed a straightforward inclusion of both systematic and statistical errors which is essential because, despite the very large exposure factor and determined effort to control systematic effects, the total sample is still quite small and the charge resolution is not perfect. In addition these simulations were used to test the significance of the detection of one transuranic event. The procedure was as follows. 1. Assuming the actinide (Z 88) spectrum to consist only of 90 Th, 92 U, and 94 Pu nuclei, a hypothetical composition was chosen (e.g., 70% 90 Th, 28% 92 U, and 2% 94 Pu). 2. Using this hypothesis, a Monte Carlo derived primary spectrum was created, with a sample size equal to that of the measured UHCRE sample (i.e., 35 events). 3. Each event in this sample was fed through a Gaussian error generator to simulate a set of 35 “measured” Z values forming a synthetic spectrum including measurement errors. 4. This process was repeated and 1000 such synthetic spectra assembled. 5. The Z values for all synthetic spectra were totalled to create a mean sample. 6. A Kolmogorov–Smirnov8 test was applied between each of the synthetic spectra and this mean sample to obtain a

90 Th, 92 U, and 94 Pu content in the primary cosmic rays. The like-

lihoods derived for every hypothesis allow us to determine the most probable composition of the CR spectrum within suitable confidence intervals (or contours). To test the technique, artificial “measured” spectra (each assumed to have a certain 90 Th content) were randomly generated. The Monte Carlo analysis successfully identified the most likely original composition from these spectra (Donnelly 2004), demonstrating that the method is both robust and unbiased. When applied to the UHCRE data, the maximum likelihood Th– 90 92 U cosmic-ray composition was inferior to that obtained for the maximum likelihood 90 Th–92 U–94 Pu composition, due to the extra degree of freedom in the latter case. Figure 15 depicts in a ternary diagram the likelihood of various hypothetical GCR compositions being consistent with the measured data when a 90 Th, 92 U, and 94 Pu content is assumed in the cosmic rays. The elongation of the likelihood contours is easily understood by noting that mean charge over all events is well determined and essentially fixed. Thus, denoting the fraction of uranium as [U], etc, 90[Th] + 92[U] + 94[Pu] ≈ const, (9) from which it follows that [U] ≈ const − 2[Pu].

(10)

8

Kolmorogov–Smirnov was chosen as a simple, robust, and non-parametric goodness-of-fit metric ideally suited to this application rather than the more conventional parametric goodness-of-fit statistics such as χ 2 .

In effect this is just saying that two uranium events are rather similar to a superimposed thorium and plutonium event, and thus 10


Donnelly et al.

Figure 16. LDEF data on the GCR composition compared to the ultra-heavy solar system abundances as given by Lodders (2003) plotted with the same normalisation. Note the striking depletion of 82 Pb in the GCR.

the confidence contours are roughly ellipses elongated along this line of constant mean charge. The most likely GCR actinide composition is found to be 69% 90 Th, 26% 92 U, and 5% 94 Pu. Within 70% likelihood contours, the 92 U/90 Th ratio is 0.38 with a maximum value of 0.96 and a minimum of zero (formally, the zero value would require a physically unreasonable abundance of plutonium and a pure thorium plutonium composition). There appears to be less uranium than thorium in the cosmic rays, but even with the LDEF’s large collecting power one is fundamentally limited by sampling statistics in making such statements (with a total number of actinide events of 35 and a 26% uranium composition √ only about nine uranium events are expected; the sampling n fluctuations in the number of uranium events are thus already of order ±3 before any other errors are included). Further Monte Carlo simulations using the most likely 90 Th, 92 U, and 94 Pu composition demonstrated that there is an approximately 18% chance of one of the lower-charge events being binned as 96 Cm. However, this falls to only 4% if we assume a 90 Th- and 92 U-only composition. In this sense, the presence of transuranic elements has been demonstrated at the 96% confidence level.

(≈0.57), and the cosmic rays expected to arise from superbubble interiors (1.2 ± 0.4), though not with superbubble interiors themselves (1.4 ± 0.4). It does not match most predictions of r-process yields nor those of ejecta accumulated from multiple supernovae inside superbubbles over a 50 Myr period (both 2.3). These estimates are from Lingenfelter et al. (2003) who also discuss the actinide to “Pt” ratios expected, but unfortunately use a slightly different definition of the platinum group (75 Z 79). Using this definition (in this paragraph only) gives an actinide/“Pt” ratio for the LDEF data of 0.032, in broad agreement with other observations. This is higher than in the present ISM (0.014 ± 0.002) and similar to that of the protosolar medium (≈0.023) and the interior of superbubbles (0.029 ± 0.005). Once propagated to the source, however, it is considerably higher. This unusually high value could be indicative of a recent r-process event. A major caveat here is that the exact effect of propagation on this ratio is unknown. The observational values are therefore best considered as lower limits. The presence of transuranic events in the UHCRE spectrum is intriguing as the half-lives of these elements are short (81 and 15.6 Myr for 94 Pu and 96 Cm, respectively). 96 Cm is the first longlived actinide to undergo significant decay so its presence in GCRs would mean that the measured GCR actinide abundances contain a recently synthesized component. Taken alone, the measured 92 U/90 Th ratio implies (with large errors) that the cosmic-ray source material is relatively old (>108 yr have elapsed between nucleosynthesis and acceleration). However, Monte Carlo simulations indicate that the probability that the UHCRE data contain at least one transuranic event is about 96%. These nuclei have relatively short half-lives and should not be present if the cosmic-ray-seed nuclei are old. The implication is that the cosmic-ray source material could be an admixture of a small amount (less than 10%) of freshly nucleosynthesized matter (containing transuranics) in what is otherwise rather normal well mixed and old interstellar material (material with a low 92 U/90 Th ratio and standard abundances). Obviously, more primary GCR Z-abundance measurements (especially of the actinides) are required. An important point here is that currently the statistical errors on measurements

6. DISCUSSION AND SUMMARY The UHCRE data set is the largest sample of ultra-heavy cosmic rays studied to date and includes the largest uniform sample of cosmic-ray actinides (35 events). The actinides are well separated from the subactinides due to the unstable nature of the elements immediately heavier than 83 Bi. There are some striking features in the data. The experiment has demonstrated, in agreement with other experiments, that the “Pb”/“Pt” abundance ratio is decidedly low in the GCRs (0.25 ± 0.04) compared to the best estimates from solar and meteoritic material (1.03 ± 0.12), see Figures 13 and 16. Even assuming a very severe propagation effect on this ratio (×2.6), the GCR value is a mere 0.66 ± 0.10. This could be indicative of a volatility-based acceleration bias. At 0.38+0.58 −0.38 , the GCR 92 U/90 Th ratio is compatible with that of the present-day ISM (0.27 ± 0.04), the protosolar medium 11


Donnelly et al.

Table 5 Oversampled Data from the UHCRE Normalized to the Entire Z 70 Sample Element

Charge

Before Fragmentation Correction “Abundance”

Yb Lu Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Po At Rn Fr Ra Ac Th Pa U Np Pu Am Cm Bk

70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97

0.036 0.038 0.044 0.047 0.056 0.060 0.078 0.110 0.150 0.116 0.075 0.058 0.063 0.034 0.013 0.006 0.002 0.000 0.0021 0.0008 0.0045 0.0017 0.0033 0.0012 0.0004 0.0000 0.0004 0.0000

After Fragmentation Correction

1σ Statistical Error + − 0.009 0.009 0.010 0.010 0.011 0.011 0.013 0.015 0.015 0.013 0.013 0.011 0.012 0.009 0.006 0.005 0.004 0.003 0.0014 0.0011 0.0018 0.0013 0.0016 0.0012 0.0009 0.0008 0.0009 0.0008

0.007 0.008 0.008 0.009 0.009 0.010 0.011 0.013 0.015 0.013 0.011 0.009 0.010 0.007 0.004 0.003 0.001 ... 0.0009 0.0005 0.0013 0.0008 0.0011 0.0007 0.0003 ... 0.0003 ...

“Abundance” 0.030 0.034 0.041 0.044 0.053 0.057 0.077 0.110 0.154 0.119 0.078 0.061 0.067 0.037 0.013 0.007 0.002 0.000 0.0023 0.0008 0.0052 0.0018 0.0039 0.0014 0.0005 0.0000 0.0005 0.0000

1σ Statistical Error + − 0.008 0.008 0.009 0.010 0.010 0.011 0.012 0.015 0.016 0.014 0.013 0.012 0.012 0.010 0.007 0.005 0.004 0.003 0.0015 0.0011 0.0021 0.0015 0.0019 0.0014 0.0011 0.0008 0.0011 0.0008

0.006 0.007 0.008 0.008 0.009 0.009 0.011 0.013 0.016 0.014 0.011 0.010 0.011 0.008 0.005 0.003 0.001 ... 0.0010 0.0005 0.0015 0.0009 0.0013 0.0008 0.0004 ... 0.0004 ...

the subactinide elements the actinides in the GCR show a modest enhancement of about a factor of three (see Figure 17). 2. Our current knowledge of the relative actinide abundances is mainly limited by statistical error and not by chargeassignment errors in the detectors. Within 70% confidence contours, the most likely 92 U/90 Th ratio is 0.38 with a maximum value of 0.96 and a minimum of zero. Although with high statistical errors this low value relative to fresh r-process material suggests that the GCR matter has had sufficient time between nucleosynthesis, acceleration, propagation, and finally detection for significant radioactive decay of the shorter-lived actinides to occur (>108 yr). 3. There is evidence that suggests a 94 Pu component in the cosmic rays and one 96 Cm candidate event. The probability of the existence of a transuranic component is 96% which is indicative of a component of freshly synthesized matter (certainly