Analyzing individual presolar grains with CHARISMA

doi:10.1016/S0016-7037(00)00082-6

Analyzing individual presolar grains with CHARISMA

MICHAEL R. SAVINA ,1,* M ICHAEL J. PELLIN,1 C. EMIL TRIPA,1,2 IGOR V. VERYOVKIN,1 WALLIS F. CALAWAY ,1 andANDREW M. DAVIS

2,3

1Materials Science Division, Argonne National Laboratory, Argonne, IL 60439, USA2Enrico Fermi Institute, University of Chicago, Chicago, IL 60637, USA

3Department of the Geophysical Sciences, University of Chicago, Chicago, IL 60637, USA

(Received April 10, 2002;accepted in revised form December 18, 2002)

Abstract—Isotopic analysis of heavy elements in individual stardust grains is important in testing andconstraining theories of stellar nucleosynthesis. These analyses are challenging in that the grains are verysmall, the largest being perhaps a few microns in diameter, and contain only trace concentrations of heavyelements, generally on the order of ppm. In addition, isotopic analysis requires the suppression of isobaricinterferences. We describe a unique instrument, based on resonant ionization mass spectrometry, that hassuccessfully characterized such grains for the past several years, and report on some recent upgrades thatsignificantly enhance the instrumental capabilities. The fundamental principles and operational details arediscussed, along with illustrative results and plans for future modifications.Copyright © 2003 Elsevier Ltd

1. INTRODUCTION

In 1995 we reported on a new instrument for isotopic anal-ysis of trace elements in small specimens, known as CHA-RISMA (Chicago-Argonne Resonant Ionization Spectrometerfor Mass Analysis) (Ma et al., 1995). Since then, CHARISMAhas demonstrated the unique ability to analyze the isotopiccomposition of trace elements in individual stardust grains(Nicolussi et al., 1997a,b, 1998a–c; Davis et al., 1998, 1999;Pellin et al., 1999a,b, 2000; Jennings et al., 2002; Savina et al.,2002, 2003; Tripa et al., 2002). The operating principle andbasic configuration of the instrument have remained the same:material is desorbed from the grain via a pulsed laser beam, theelement of interest is resonantly ionized with tunable lasers,and the ions are analyzed via time-of-flight (TOF) mass spec-trometry. During the past several years the instrument hasundergone several upgrades, however. New laser and dataacquisition systems allow data to be acquired with greatlyenhanced speed and accuracy. In this paper, we report on theseupgrades, as well as on some significant improvements to thebasic instrumental design planned for the near future. A com-panion paper in this issue reports on barium isotopes in indi-vidual presolar grains, which is the first analysis made with theupgraded instrument (Savina et al., 2003).

The study of isotopic compositions of heavy elements inindividual presolar grains presents a substantial analytical chal-lenge because the grains are quite small and the concentrationsof these elements are low. Silicon carbide, which is the best-studied phase to date, has a median grain diameter on the orderof a few hundred nm and rarely exceeds 5�m in diameter(Amari et al., 1994). Heavy element concentrations are gener-ally from a few to a few tens of ppm (Amari et al., 1995;Kashiv et al., 2001, 2002). Because of these considerations,initial isotopic studies of heavy elements were restricted toaggregates of many grains. Analyses of noble gases, barium,neodymium and samarium in grain aggregates using Thermal

Ionization Mass Spectrometry (TIMS) and Secondary Ion MassSpectrometry (SIMS) showed enhancements in s-process iso-topes, and indicated that Asymptotic Giant Branch stars werethe likely source of the majority of the grains (Ott et al., 1988;Ott and Begemann, 1990; Zinner et al., 1991; Prombo et al.,1993). For example, Ott and Begemann (1990) and Prombo etal. (1993) determined barium isotopic compositions in size-separated fractions of SiC by TIMS and found depletions(relative to 136Ba, compared to solar) in all isotopes except134Ba, in accordance with predictions fors-process nucleosyn-thesis. Zinner et al. (1991) used SIMS on SiC size separates toanalyze for barium, neodymium, and samarium and also foundisotopic compositions in accordance with s-process nucleosyn-thesis. In addition to indicating AGB stars as likely progenitorsfor most presolar SiC grains, these studies of trace metals alsoshowed increasings-process enhancement with decreasinggrain size.

The grain size effect, together with measurements of isotopiccompositions of carbon, silicon and nitrogen (see, for example,Zinner, 1998), showed that the isotopic compositions were notthe same from grain to grain and pointed to the need for traceelement analysis in individual grains. Studies of heavy ele-ments in individual grains provide information on individualstars, and permit direct comparison with stellar models for aspecific set of physical conditions, such as stellar mass andmetallicity. A companion article in this issue reports just suchan analysis of Ba in single presolar SiC grains (Savina et al.,2003).

2. ANALYTICAL REQUIREMENTS FOR INDIVIDUALGRAIN ANALYSIS

Single-grain trace element isotopic studies require high sen-sitivity, discrimination against isobaric interference, and spatialresolution on the order of the grain size, i.e.,�1 �m. WhileSIMS microprobe analysis is capable of determining the isoto-pic compositions of the major elements and some of the moreabundant minor elements in individual grains (e.g., Hoppe etal., 1994), a different technique is required for trace element

* Author to whom correspondence should be addressed ([email protected]).

Pergamon

Geochimica et Cosmochimica Acta, Vol. 67, No. 17, pp. 3215–3225, 2003Copyright © 2003 Elsevier Ltd

Printed in the USA. All rights reserved0016-7037/03 $30.00� .00

3215

isotopic analysis. To illustrate this point we consider barium asan example: the average Ba concentration in large (3.4 to 5.9�m) SiC grains from Murchison size separate KJH (Amari etal., 1994) is �40 ppm by weight (Amari et al., 1995), howeverhalf of the grains have less than �15 ppm Ba and more than aquarter have less than 5 ppm (Fig. 1). In fact, the histogram ofFigure 1 is artificially skewed toward high Ba concentrations,because 12 of the 60 data points are actually upper limits to theBa concentrations in the particular grains analyzed (Amari etal., 1995). Therefore, to obtain a reasonably representativesample of grains, the analysis must include grains with as littleas 5 ppm Ba. A 3 �m grain with 5 ppm Ba contains �106

barium atoms, distributed over 7 isotopes. The question thenarises: What fraction of these atoms must be detected to definethe isotopic composition of barium in this particular grain withsufficient precision to distinguish it from terrestrial materialand investigate the nucleosynthetic conditions in its parent star?

Resonant Ionization Mass Spectrometry (RIMS) techniqueshave demonstrated useful yields (atoms detected per atomconsumed) of up to 5% (Pellin et al., 1990). If we assume auseful yield of 1%, which is typical for CHARISMA in itscurrent configuration, we would expect to detect 10000 bariumatoms in the 3 �m grain referred to above. If the isotopiccomposition is solar, the 2� uncertainties as calculated fromPoisson statistics (see below) for the measured �xBa values(where �xBa � [(xBa/136Ba)SiC/(xBa/136Ba)std � 1] * 1000)would be �607‰ for 130Ba, �636‰ for 132Ba, �147‰ for134Ba, �106‰ for 135Ba, �93‰ for 137Ba, and �75‰ for138Ba. With this number of total counts, no useful informationcan be obtained for 130Ba and 132Ba, because of their low solarsystem abundance; the abundance of these p-process isotopes iseven lower in s-process-enriched grains (Ott and Begemann,1990; Prombo et al., 1993; Jennings et al., 2002). Given that the

remaining �xBa values vary by several hundred ‰ amongindividual presolar grains, this level of precision is sufficient toinvestigate the stellar nucleosynthesis of barium (Savina et al.,2002, 2003; Jennings et al., 2002; Lugaro et al., 2003).

In addition to possessing high sensitivity, the analyticalmethod must also discriminate against isobaric interferences.Elements such as iron, zirconium and molybdenum in presolargrains present stringent requirements (Nicolussi et al., 1997a,1998a,c; Tripa et al., 2002). Major iron isobars in SiC grainsare SiC2 and Si2, as well as 54Cr and 58Ni plus hydrocarbons,Al2, AlSi, CaO, TiC, and FeH. Zirconium isobars include92Mo, 94Mo and 96Mo, plus YH, ZrH, MoH and Ti2. Forbarium analysis, the technique must be able to either suppressor resolve 136Ce and 138Ce, plus BaH and surface contaminantssuch as organic molecules. Ion cyclotron resonance mass spec-trometers and high-resolution magnetic sector instruments arecapable of resolving some of these interferences, however theirlow sensitivities preclude single grain analysis. Since RIMSanalysis relies on gas-phase ionization by lasers tuned to elec-tronic transitions specific to the analyte of interest, the resonantnature of the ionization event ensures that unwanted species areleft as undetected neutrals, thereby suppressing isobaric inter-ferences.

When combined with a microfocus laser for desorbing atomsfrom the grains (see below), a RIMS instrument thus satisfiesthe analytical requirements of sensitivity, discrimination, andlateral resolution for isotopic analysis of trace elements instardust grains.

3. INSTRUMENTAL DETAILS

The schematic layout of the CHARISMA instrument is es-sentially unchanged from the previous reports (Ma et al., 1995;Nicolussi et al., 1996, 1997b), consisting of a desorption lasersystem with a Schwarzschild microscope objective, primary ionguns for sample cleaning and/or analysis, a tunable laser sys-tem, ion extraction optics, a reflectron-type TOF mass analyzer,and a data acquisition system. The major upgrades are in thelaser desorption, laser photoionization, and data acquisitionsystems, which operate at 1000 Hz, compared to 60 Hz previ-ously. The ionization lasers are solid state Ti:Sapphire systems,which are more reliable and easier to use than the old dyelasers. Taken together, the higher repetition rate and greaterpower stability of these lasers give rise to a system in whichexperimental conditions do not change on the timescale of asingle grain analysis. An analysis that required 500 min (andpossibly a dye change) using the old system can now be donein 30 min.

3.1. Desorption Laser System

The desorption laser is an intracavity frequency-tripled di-ode-pumped Nd:YAG (IB Laser model DiNY pQ), whichdelivers �600 �J of 355 nm light in 8 ns pulses at a 1 kHzrepetition rate. The UV photons are strongly absorbed in mostmaterials, and the system has very high beam quality, whichallows for tight focusing at the target. The laser has excellentpulse-to-pulse stability (power fluctuations are less than 1%) togive steady, controllable desorbed neutral yields. To achievesuch stability the laser must be run at full power. However,

Fig. 1. Histogram of log Ba concentration in presolar SiC grainsfrom the Murchison meteorite with nominal diameters between 3 and 5�m (Amari et al., 1995). The bars (left-hand scale) give the absolutenumber of grains in each concentration bin, while the line (right-handscale) gives the cumulative percentage of grains as concentrationincreases. Lighter bars denote grains for which only upper limits to theBa concentration are available. It should be noted that this is a non-representative sample of presolar SiC grains: nearly half of the grainsplotted are rare types (A, B, X, and Y), whereas mainstream grainscomprise �90% of all presolar SiC in Murchison (see Zinner, 1998, fora description of SiC grain types).

3216 M. R. Savina et al.

since the 600 �J output is far higher than the �10 nJ we findto be optimal for laser ablation of SiC at useful rates (Ma et al.,1995), power is attenuated in two steps: first with a half-waveplate and polarizer, and second with a variable neutral densityfilter (Fig. 2). The excellent pulse-to-pulse stability makes thislaser especially well suited for trace analysis, since in ion-counting mode it is important to achieve a high count rate (ionsper pulse) while ensuring that no (or very few) pulses containmore than one ion in any given time channel. Because thedesorption process is nonlinear in laser pulse energy, evenmoderate pulse-to-pulse fluctuations give rise to large varia-tions in the desorbed neutral flux. This can lead to countingerrors, because registered counts will not obey the Poissondistribution law. Furthermore, the larger the pulse-to-pulsefluctuations, the more likely the laser is to occasionally producerelatively powerful pulses that give a great many ions per timechannel, and cause serious undercounting of the major isotopes.Hence, desorption lasers must be run at very low averagepower, resulting in very low count rates, to avoid such prob-lems. Therefore, in addition to running at a higher repetitionrate than CHARISMA’s previous desorption laser system, thenew high-stability laser system allows the experiment to be runat a higher count rate, further reducing the time required toacquire spectra.

Figure 2 shows the desorption laser beam path, including the

Schwarzschild microscope, which is mounted inside the vac-uum chamber. The microscope has been described in detailelsewhere (Ma et al., 1995). It consists of two spherical mirrors,one concave and the other convex. It has a numerical apertureof 0.47, giving a magnification factor of 19 and a workingdistance of 8.1 cm from the sample. Because the Schwarzschildmicroscope is an obstructed-pupil optical element, the opticalaxis is blind everywhere except at the image and object planes.Therefore, holes are drilled through the mirrors along theoptical axis to allow ions to pass through into the TOF massanalyzer. This design allows for normal-incidence focusing ofthe desorption laser to achieve a circular, rather than elliptical,beam spot. In addition to focusing the desorption laser beam,the microscope also provides an optical image of the samplewith a lateral resolution of �1 �m.

3.2. Ionization Laser System

The tunable laser is a custom-built Photonics Industriesmodel TU-TR-S1/UV-S2 Ti:Sapphire system, a portion ofwhich is shown schematically in Figure 3. The actual systemhas two separate Ti:Sapphire cavities, both of which are lon-gitudinally pumped by a single Nd:YLF laser (Photonics In-dustries model GM30-527 P/S) split equally between the twocavities. The pump laser operates in the 2nd harmonic at 527 nm

Fig. 2. Schematic drawing of the desorption laser beam path. See discussion in text.

3217Analyzing presolar grains with CHARISMA

with a pulse length of 200 ns at a repetition rate of 1 kHz. EachTi:Sapphire cavity is independently tunable over the wave-length range from 700 to 1000 nm via a gold-coated gratingthat functions as the end mirror. Each cavity is arranged in a “J”configuration to prevent the Nd:YLF pump energy from heat-ing the grating, which would broaden the output linewidth andcause wavelength instabilities, and to allow for reflection andrefocusing of the Nd:YLF pump beam back into the Ti:Sap-phire crystal (via the refocusing mirror in Fig. 3) to boost thepower output. A four-prism beam expander is used to fill thegrating, thereby narrowing the output linewidth. At a pumpenergy of 10 mJ (20 mJ split 50/50), each cavity produces 1mJ of IR energy in a 25 ns pulse with a 2 GHz linewidth.

The 2nd, 3rd, and 4th harmonics (2�, 3�, and 4�, respec-tively) of the Ti:Sapphire fundamental beam are formed innonlinear optical crystals as shown schematically in Figure 3.The fundamental beam is focused via a 150 mm lens into a 20mm lithium triborate (LBO) crystal. The exit beam containsboth the 2nd harmonic and residual fundamental. If only thefundamental and/or 2nd harmonic beams are required, they aresplit out of the beam path at the appropriate points (i.e., eitherahead of or immediately after the LBO crystal). To make the3rd harmonic, the fundamental and 2nd harmonic beams aremixed in an 8 mm �-barium borate (BBO) crystal as shown.Due to the walk-off induced by the LBO, the fundamental and2nd harmonic beams propagate along slightly different opticalpaths on exit from the LBO crystal, so they are split by mirrora, refocused separately, and combined on mirror b beforeinsertion into the BBO crystal. (Mirrors a and b reflect the 2nd

harmonic and transmit the fundamental.) The plane of polar-ization of the 2nd harmonic beam is rotated 90° with respect tothe fundamental on exit from the LBO crystal, so a half-wave

plate is used to correct the polarization before mixing the twobeams in the BBO crystal to form the 3rd harmonic. To form the4th harmonic, mirrors a and b in Figure 3 are removed and the2nd harmonic is frequency-doubled in the BBO crystal. What-ever the scheme, the final beam contains several colors, i.e., thedesired harmonic plus the residual of the fundamental and allthe lower harmonics. The beamsplitters at right in Figure 3 areused to isolate the colors of interest. (The beamsplitters aresimilar to mirrors a and b in that they reflect only the harmonicof interest and transmit all the others.) The beams are thencollimated and shaped separately before insertion into the in-strument. Overall conversion efficiencies (from the fundamen-tal) are on the order of 30% for the 2nd harmonic, 8% for the 3rd

harmonic, and 4% for the 4th harmonic.The total photoion signal, and therefore sensitivity, is af-

fected by many parameters (Pellin et al., 2001a,b) including theTi:Sapphire laser power and the size of the ionization volume.Unlike a SIMS instrument, in which ions are collected directlyfrom the sample surface, a RIMS instrument must extract ionsfrom a 3-dimensional volume above the sample and transmitthem to the detector. After the desorption laser pulse, the cloudof desorbed neutrals expands into a solid angle of 2� srad as itleaves the surface. To optimize the efficiency, one wants toionize the neutrals as near to the surface as possible before theyhave expanded into a large volume. However, because theionization laser power is sufficient to produce a number ofunwanted “spray” ions if it hits the sample, it must be keptsome distance above the sample surface, generally �0.2 to 0.5mm. Ion optical simulations (see below) show that the mostefficient ion extraction occurs in a cylindrical volume above thesample approximately 3 mm in diameter and �1 to 1.5 mmthick. Cylindrical lenses are used to collimate the resonance

Fig. 3. Schematic drawing of a tunable Ti:Sapphire laser cavity and frequency conversion scheme for producing theharmonics of the Ti:Sapphire fundamental. Lithium triborate (LBO) is used for 2nd harmonic (2�) generation, beta bariumborate (BBO) is used for 3rd and 4th harmonic (3� and 4�) generation. Flip mirrors and beam blocks (not shown) are usedto split out the fundamental and/or second harmonic beam if desired. Mirrors a and b are removed for 4th harmonicproduction as described in the text. All lenses are 150 mm focal length. Cylindrical lenses (not shown) are used to collimateand focus the beams before insertion into the vacuum chamber.


and ionization lasers at �1 � 3 mm as they cross 0.5 mm abovethe sample surface. Depending on the mass of the analyteelement, the Ti:Sapphire lasers are timed to fire from a fewhundred to �1500 ns after the desorption laser, giving thecloud of desorbed neutrals time to fly to the photoionizationvolume.

3.3. Mass Spectrometer

Two ion guns provide for sample cleaning and analysis. Thefirst is a Colutron 101 Q ion source, which provides 5 keV Ar�

ions at an angle of 60° to the sample normal in a 0.5 to 1 mmspot at a current of 1 to 2 �A. The second is an Ionoptika liquidmetal ion gun, which provides 25 keV 69Ga� ions at an angleof 60° from the sample normal in a spot focusable to 50 nm. Forgrain analysis, the guns are most often used to provide a steadysource of desorbed neutrals for tuning and aligning the laserand extraction optics. In addition, the Ar� beam is used toremove organic contamination on grain surfaces. Removal of�10 atomic layers is generally sufficient.

Presolar grains are deposited on a gold foil and pressed intothe gold with a sapphire platen in a manner described elsewhere(Nicolussi et al., 1998a). Samples and standards are mountedon standard 0.5-inch SEM stubs and transferred to a 4-positioncarousel inside the chamber. A 3-axis Burleigh 6000 Inchwormpiezoelectric positioning system is used to position the samplesfor analysis. In addition, one of the sample holders is equippedwith a second piezoelectric system capable of submicron mo-tion for fine positioning of the grains under the laser spot.

The ion extraction optics, reflectron time-of-flight mass an-alyzer, and microchannel plate detector are unchanged fromprevious reports (Ma et al., 1995; Nicolussi et al., 1996,1997b). The data acquisition system was upgraded to accom-modate the 1000 Hz repetition rate of the new laser systems. AnOrtec FastFlight digitizer is used to acquire and accumulate theion signal from the microchannel plate detector. Individual ionstrikes on the microchannel plate detector produce signals thatvary in magnitude from zero to 200 mV. A discriminatorbetween the detector preamplifier and the digitizer convertssignals greater than 30 mV (to discriminate against backgroundnoise) into 400 mV square wave pulses of 2 ns duration. Thedigitizer builds a time-of-flight spectrum by summing the dis-criminator output into 2 ns bins in a buffer that accumulates4096 spectra before uploading to a computer. Each uploadedsummation spectrum is recorded as a “ loop” on the computer,which converts the spectrum from voltage to ion counts andsums a user-determined number of loops into a final spectrum.In practice, grain spectra typically consist of 25 loops, for atotal of 102,400 laser shots per spectrum. Up to 10 or 15 suchintermediate spectra may be summed into one final grain spec-trum. At 1 kHz, each 25-loop spectrum is acquired in �2 min.A typical grain analysis therefore requires from 20 to 30 min,depending on the number of loops acquired.

4. RESONANT IONIZATION

Resonant ionization spectroscopy (RIS) is a mature field,with compilations of schemes for a variety of elements avail-able in the literature (see, for example, Saloman, 1994). To beuseful in terms of sensitivity and accurate isotope measure-

ment, efficient RIS schemes must take into account effects suchas ionization rate, isotope shifts, power broadening, and Dopp-ler broadening. We consider these in turn.

Figure 4 shows a RIS scheme for a hypothetical atom forpurposes of illustration. Because laser pulses have finite tem-poral widths, and desorbed atoms have finite residence times inthe laser beam path, it is necessary to achieve a high ionizationrate to achieve a high ionization fraction, i.e., ions produced perneutral atom irradiated. See Wucher (2001) for a general dis-cussion of ionization processes). To do so, atoms must bepumped from the excited intermediate state (state A in Fig. 4)to the ionization continuum appreciably faster than they decayto lower-lying electronic states (state B in Fig. 4), which aretransparent at the laser frequency and hence inaccessible untilthey decay back to the ground state. Such low-lying states canbe avoided by strongly saturating the resonance transition (�1 inFig. 4) with high laser intensity. At high intensity, stimulatedemission from the intermediate state to the ground state dom-inates spontaneous decay to the lower-lying electronic states, sothat atoms are cycled between the ground and intermediatestates. (This is the well-known Rabi oscillation.) Thus, noappreciable population builds up in dark states, and nearly allatoms are available for ionization by the ionization laser (�2 inFig. 4).

High laser intensity also leads to power broadening of theatomic transitions, which has two important consequences: itleads to excitation and ionization of atoms which are Doppler-shifted outside the laser linewidth (Wright et al., 1981), and itcan ensure that measured isotope ratios are unaffected byisotope shifts. We first consider Doppler broadening. Becauselaser-desorbed atoms leave the sample surface with a distribu-tion of speeds and angles, the desorption process leads toDoppler broadening of the atomic transitions which is generallylarge compared to the laser linewidth. If the desorbed atoms areapproximated as a Maxwellian gas, the Doppler broadening ofan atomic transition is given by

Fig. 4. RIS scheme for a hypothetical atom, illustrating the groundstate (G), intermediate state (A), low-lying electronic state (B), reso-nant transition (�1), and ionizing transition (�2). Isotope splitting isshown for the intermediate state only.


�FWHM� c

�1 �1

1 � 2�2ln2� kT

mc2� (1)

where c is the speed of light, is the wavelength of thetransition, k is Boltzmann’s constant, T is the temperature of thegas, and m is the mass of the atom. Figure 5 shows Dopplerwidths for a variety of temperatures at a resonant transitionwavelength of 300 nm, along with the 1.15 GHz FWHMlinewidth of the Ti:Sapphire laser at this wavelength (3rd har-monic). Because the Doppler width is always larger than thelaser linewidth, many atoms would go undetected if the laserpower was insufficient to substantially power-broaden theatomic transition.

Power broadening is also important in preventing isotopeshifts from affecting measured isotope ratios. Figure 4 showsshifts for three isotopes for the transition from the ground tointermediate state A of a hypothetical element. If these shiftsare large compared to the laser and transition linewidths, themeasured isotope ratios can vary significantly as the laserfrequency is scanned across the transition (Wunderlich et al.,1992, 1993). We have previously shown that calcium can showdramatic variations in RIMS-measured isotope ratios as a func-tion of resonance laser frequency due to large isotope shifts(Nicolussi et al., 1997b). Plotted in Figure 6 are isotope ratiosfor even-numbered calcium isotopes (desorbed from Ca metal)as measured by Nicolussi et al. (1997b). (We do not considerthe odd isotope 43Ca here because the nuclear hyperfine split-ting induced by the 7/2 spin of the nucleus complicates theanalysis.) The isotope shifts for the 4s2 (1S) 3 4s5p (1P)transition at 239.856 nm used in that study are not known,however the 4s2 (1S) 3 4s4p (1P) calcium transition isotopeshifts relative to 40Ca are 0.40, 0.76, and 1.50 GHz for 42Ca,44Ca, and 48Ca, respectively (Brandt et al., 1978). We expectthat the 4s2 (1S)3 4s5p (1P) transition isotope shifts should bewithin 10 to 20% of these values, based on the behavior seen in

other calcium transition series (Lorenzen et al., 1983). Theseshifts become important when the atomic transition lines arenot appreciably power- or Doppler-broadened. The dye laserused in the calcium study had a linewidth of 1.9 GHz, and thepower was kept low to avoid non-resonant ionization of inter-fering species, so the transition was not strongly power-broad-ened.

The data in Figure 6 was fit to a model convoluting thegaussian laser line with Voigt line shapes for the calciumtransitions (i.e., assuming both power- and Doppler-broadenedtransitions). In the model, the desorption temperature (andhence, Doppler linewidth) of the atoms, the Lorentzian line-width due to power broadening, the transition isotope shiftsrelative to 40Ca, and the absorption cross sections of the variousisotopes were fit to the data. The best fit is shown in Figure 6,and shows excellent agreement between theory and observationfor a desorption temperature of 1960 K (the boiling point of Cametal is 1717 K), transition isotope shifts of 0.332, 0.728, and1.475 GHz, and cross sections (relative to 40Ca) of 1.05, 1.06,and 0.983 for 42Ca, 44Ca, and 48Ca, respectively. (The “crosssections” calculated here are not strictly speaking correct, sincethey implicitly include an instrumental mass fractionation ef-fect (see below), which could not be accurately calculated forcalcium.) The FWHM of the transitions ranged from 2.8 to 3.1GHz. Because of the relatively low power, the Lorentzian(power broadening) contribution to the linewidth was only 0.6GHz. Wavelength control was therefore crucial to obtainingreproducible calcium isotope ratios.

Light elements (Z �20) experience isotope shifts dueprimarily to differences in nuclear masses, whereas nuclearvolume effects dominate the isotope shifts in heavier elements(Z � �60) (Sobel’man, 1972). Because the two shifts haveopposite signs, they tend to cancel one another in elements ofintermediate mass such as barium. Thus one expects isotopeshifts to be less important for barium than for calcium. As weshall see, this is indeed the case. The resonant ionizationscheme used for Ba analysis in individual SiC grains reportedin the companion paper in this issue is 6s2 (1S) 3 6s7p (1P0)

Fig. 5. Doppler broadening of atomic transitions at 300 nm as afunction of atomic mass for desorption temperatures of 1500, 2500, and3500 K. Also shown is the linewidth (FWHM) of the 3rd harmonic (3�)of the Ti:Sapphire laser. The laser and atomic line widths are given infrequency units rather than wavelength to make the ordinate wave-length-independent.

Fig. 6. RIMS-measured calcium isotope ratios as a function ofresonance laser detuning from Nicolussi et al. (1997b), with fits to amodel assuming Doppler and power broadening. Model parameters aregiven in the text.


(307.247 nm, 3�), followed by ionization at 883.472 nm.Figure 7 shows saturation curves (ion yield vs. laser power) forbarium using this scheme. The curves are constructed by hold-ing the power of one laser at the maximum while varying thepower of the other. The solid lines in Figure 7 are least-squaresfits to the behavior expected for an ideal resonant ionizationprocess (Wucher, 2001). The resonant transition is well satu-rated, while the ionization transition is approximately 85%saturated. This fit, while useful in determining the saturationbehavior, is an approximation and does not take into accountDoppler and power broadening effects, which are importantconsiderations in maximizing the sensitivity and understandingisotope shift effects such as those seen in the Ca data.

In the case of Ba, it is possible to estimate the temperature,and hence Doppler broadening, of the desorbed atoms. Thetemperature of the desorbed atoms can be approximated bymeasuring their flight time from the sample surface to theionization volume. This is done by measuring the optimal delay

time between the desorption and ionization lasers: that is, thedelay time that maximizes the ion signal. The number densityof atoms at a distance r above the surface and at an angle �from the surface normal at the desorption laser spot after thelaser pulse is given by:

dNr,�,�� Q2cos�

�r3

�4

2kT/m�2 e�m�2/ 2kTd�, (2)

where Q is the total number of atoms desorbed, v is the velocityof the atom, k is Boltzmann’s constant, m is the mass of theatom, and d� is the volume element (Goodman and Wachman,1976; Arlinghaus et al., 1989). In this case, T is the temperatureof the atoms leaving the surface. Setting v � r/t and Q � 1, andintegrating over the extraction volume gives an expressionrelating normalized ion signal to delay time. Eqn. 2 was inte-grated assuming a cylindrical extraction volume 3 mm indiameter and 1 mm thick starting 0.5 mm above the surface.For barium desorption from BaTiO3, the measured optimaldelay time was 1.3 �s, giving a gas temperature of �3500 K.Uncertainties in the optimal delay time and the distance fromthe surface to the ionization volume lead to uncertainties in thecalculated temperature which we estimate to be on the order of�800 K. This leads to an uncertainty in the Doppler broadeningof �15%. From Eqn. 1, the Doppler FWHM of the bariumtransition at 307.247 nm is 3.6 � 0.4 GHz, considerably largerthan the 1.15 GHz Ti:Sapphire linewidth. The largest isotopeshift for the barium 6s2 (1S)3 6s7p (1P0) resonant transition is0.190 GHz (Eliel et al., 1983), well within both the Dopplerwidth of the transition and the laser linewidth. Thus we wouldnot expect the isotope ratios measured on the standard todeviate significantly from the accepted values by virtue of theisotope shifts. Figure 8 shows the �138Ba values as a functionof the resonance laser wavelength (a frequency scale is alsogiven), and shows that this is indeed the case; �138Ba isconstant over a wide range of resonance laser wavelengths. Theother Ba isotopes all showed similar behavior.

Figure 8 has two other significant aspects. First, ions aredetected more than 40 GHz from resonance. Unlike in thecalcium study, isobaric interference was not a problem, and thefull laser power was used to maximize the barium ion signal.Applying the same Voigt line shape model to the barium datayields a power-broadened FWHM on the order of 20 GHz,compared to 0.6 GHz for Ca. Thus the saturation behavior ofthe barium resonance process shown in Figure 8 is due in partto strong power broadening, which allows excitation of atomsthat would otherwise be Doppler-shifted far out of resonance.Second, the measured isotope ratio is not solar; the mean�138Ba value is �50‰. A small part of the difference is due tomass fractionation effects arising from the fact that the heavierisotopes have slightly longer flight times to the ionizationvolume. The mass fractionation can be calculated from Eqn. 2by taking the ratio of the number atoms of each isotope in theextraction volume at the time of the ionization laser pulse. Inthis way, the mass fractionation for �138Ba was calculated to be�8‰. The remaining difference of �42‰ must thereforereflect the relative cross sections of the 136Ba and 138Ba elec-tronic transitions. Table 1 shows the measured isotope ratios forall barium isotopes, along with the calculated mass fraction-ations and their uncertainties. The mass fractionation for the

Fig. 7. Saturation curves for the (a) resonance and (b) ionizationsteps of the barium resonant ionization scheme described in the text,constructed by varying the pulse energy of one laser while holding theother at its maximum. For the resonance curve, the ionization laserenergy was 550 �J per pulse; for the ionization curve, the resonancelaser power was 54 �J per pulse. Beam cross sections were �1 � 3mm.


calcium isotopes discussed above could not be accurately cal-culated because the smaller masses, larger relative mass differ-ences between isotopes, and shorter delay times lead to largeuncertainties in the calculation when uncertainties in the exactposition of the ionization beam are taken into account. Themass fractionation effect is therefore implicitly included in thecross sections calculated for calcium.

The Ti:Sapphire laser frequency can be easily maintainedwithin �3 GHz of the resonant frequency, so when transitionsare power-broadened as in Figure 8 the measured isotope ratiosare stable and spectroscopic and mass fractionation effects canbe normalized out of the data. Standards are run each day andthe counting statistics uncertainties are given by

� � � 1000�� 1xNgrain

�1

yNgrain�

1xNstd

�1

yNstd, (3)

where xNgrain, etc. are the numbers of atoms of isotopes x andy counted in either the grain or the standard.

We now present an example of a RIMS measurement of Baon a single SiC grain. Figure 9 compares the barium RIMSspectra of a terrestrial sample of BaTiO3 and an individual SiC

grain isolated from the Indarch meteorite (Jennings et al.,2002). The grain spectrum was acquired in 3 � 105 laser shots.Many more laser shots, �1.5 � 106, were summed to obtainthe BaTiO3 standard spectrum to accumulate a large number ofion counts and reduce the standard contribution to the error inthe �-values (Eqn. 3). In both cases the count rate in the highesttime channel (the maximum of the 138Ba peak) was kept verylow, about one count in 600 laser shots, to avoid undercountingand to ensure that the process obeyed the Poisson statistical lawused to calculate the error. Close examination of Figure 9shows that the grain spectrum is distinctly nonterrestrial. Note,for example, the depletion in 135Ba compared to the terrestrialmaterial. The �xBa/136Ba values and 2� errors for this grain are�888 � 161‰ for 130Ba, �469 � 381‰ for 132Ba, �51 �125‰ for 134Ba, �666 � 43‰ for 135Ba, �382 � 57‰ for137Ba, and �266 � 50‰ for 138Ba. These measured �-valuesclearly demonstrate the extraterrestrial origin of the grain. Acompanion article in this issue discusses barium isotope datameasured in a number of individual presolar SiC grains fromthe Murchison meteorite with the CHARISMA instrument(Savina et al., 2003). These grains, as well as similar grainsfrom the Indarch meteorite (Jennings et al., 2002), all showclear s-process nucleosynthesis signatures.

Comparison of the isotopic compositions of barium, stron-tium, zirconium, and molybdenum in individual presolar grainsobtained on CHARISMA (Nicolussi et al., 1997a, 1998a–c;Davis et al., 1998), with nucleosynthetic models (Gallino et al.,1997, 1998; Busso et al., 1999; Lugaro et al., 2003) stronglysuggest that mainstream SiC grains originate in AGB stars of�1.5 to 3 solar masses. These studies provide the most strin-gent tests of the theories and constrain stellar models of s-process nuclesosynthesis.

5. FUTURE MODIFICATIONS

New laser and data acquisition systems have greatly reducedanalysis times, from several hours when using the previous dyelaser system to 20 to 30 min with the Ti:Sapphire system, whilemaintaining the high sensitivity required for individual grainanalysis. Future modifications of the instrument are aimed atenhancing the sensitivity of the technique by improving thephotoion extraction and transmission of the ion optical system.

The CHARISMA ion optical system was modeled using theSIMION computer code (Scientific Instrument Services, Inc.,Ringoes, NJ) (Dahl, 1995). The ion extraction system is shownin Figure 10. To physically accommodate the Schwarzschildmicroscope and ion guns, conical electrodes are used to extractand focus the photoions, as shown in the cross sectional view

Fig. 8. Measured �138Ba for BaTiO3 standard as a function of theresonance laser detuning. Error bars are 2�. The horizontal line indi-cates the mean of measured values. The shaded area indicates theDoppler FWHM of the resonant transition at a temperature of 3500 K.

Table 1. Measured barium isotopic compositions from BaTiO3 standard, ratioed to 136Ba and expressed as permil deviations from terrestrialvalues.a

�130Ba �132Ba �134Ba �135Ba �137Ba �138Ba

Measuredb 21 � 54 0 � 54 13 � 11 �47 � 7 �53 � 5 �50 � 2Mass fractionationc 22 � 20 15 � 13 8 � 6 4 � 3 �4 � 3 �8 � 6

a Terrestrial barium isotopic ratios are: 130Ba/136Ba � 0.01347; 132Ba/136Ba � 0.01290; 134Ba/136Ba � 0.30779; 135Ba/136Ba � 0.83931;137Ba/136Ba � 1.42907; and 138Ba/136Ba � 9.12884 (Lewis et al., 1983).

b Based on total counts from 116 separate measurements, with 2� uncertainties calculated from counting statistics as given in Eqn. 3.c Calculated from Eqn. 2 assuming T � 3500 � 800K.


of Figure 10a. These electrodes shape the extraction field abovethe sample as shown in Figure 10b. Photoions are formedanywhere from �0.5 to 1.5 mm above the surface, and areaccelerated by the field into the time-of-flight mass spectrom-eter. This situation is distinct from SIMS instruments, in whichions are formed in a plane essentially at the sample surface. Inthe SIMS case, all ions have nearly the same energy within afew eV. In the RIMS case, the ions are born in a three-dimensional volume above the surface, and thus sample differ-ent regions of the extraction field. Figure 10b shows the fieldgradient above the target, which imparts an energy spread to thephotoions and limits the efficiency of ion transmission to thedetector. The SIMION model shows that only �8% of photo-ions formed in the 3 � 3 � 1 mm volume are transmitted to thedetector. In fact, it is this energy spread that limits the size ofthe ionization volume to 3 � 3 � 1 mm. Coupling the ionoptical modeling with the laser-desorbed neutral distribution ofEqn. 2 predicts (for barium) a maximum useful yield of �1%for this design. A substantial modification of the ion extractionand focusing system has been developed (Veryovkin et al.,2001) that substantially reduces the potential gradient in frontof the target. The new system will thus increase the volumefrom which ions can be extracted and transmitted efficiently tothe detector. The overall ion transmission should be more than

90%, which coupled with improvements in the detector designcould raise the useful yield to as much as 30%.

The higher useful yield will make possible new experimentsaimed at providing a better understanding of presolar grainsand their creation processes. Naturally, the precision of theexisting types of measurements will improve, thanks to thegreater number of ion counts obtained from the same sizegrains. Since the error in the measurement scales with thesquare root of the number of counts (Eqn. 3), analyses made onthe new instrument, with its potential 30-fold enhancement inuseful yield, will have uncertainties �5.5 times (�30) smallerthan the current ones. Elements with very low abundance inpresolar grains, owing either to low stellar abundance (forexample, the rare earth elements) or high volatility (and con-sequent incomplete condensation in refractory SiC grains), canbe studied for the first time.

In addition, it will be possible to characterize much smallergrains. Until now, only the largest grains have been amenableto analysis, restricting the studies to a subset of the grain size

Fig. 9. Barium RIMS spectra of (a) BaTiO3 standard and (b) anindividual 2 �m SiC grain from the Indarch meteorite (Jennings et al.,2002).

Fig. 10. (a) Cross sectional view of the CHARISMA ion opticalsystem. The distance from the sample surface to the first electrode is 10mm. (b) Potential energy surface of CHARISMA ion extraction regionas calculated using SIMION, showing the electrical potential in theionization/extraction region in front of the target.


distribution. As noted above, the majority of grains are muchsmaller than one micron, and generally do not contain enoughatoms of the elements of interest to be analyzed on the currentinstrument. Since the volume of a grain scales with the cube ofthe diameter, the new design will allow the analysis of grainsroughly 3 times (3�30) smaller than is currently possible.Given that our current lower limit is �1 to 2 �m (Jennings etal., 2002), the smallest grains amenable to analysis will shrinkto �0.33 to 0.66 �m. Since the mass distribution of MurchisonSiC grains peaks at �0.6 �m (Amari et al., 1994), the modifiedinstrument will be able to sample grains much more represen-tative of the overall population.

Furthermore, large grains can be depth-profiled, that is thecomposition can be probed as the laser burns its way from theshell to the core of the grain. Such studies can yield insightsinto the chemistry and physics of grain formation and residencein the interstellar medium. For example, some presolar graphitegrains are known to contain TiC crystals ranging up to 0.5 �min diameter, as well as smaller zirconium and molybdenumcarbide crystals (Bernatowicz et al., 1996). Carbides of muchless abundant elements such as hafnium, tungsten, and tantalumare expected to condense along with TiC (Lodders and Fegley,1999), and may be detected using the modified instrument. Itmay be possible with the new instrument to measure isotopiccompositions of major and minor elements in such subgrainsand possibly relate them to the composition of the surroundinggraphite.

Perhaps most promising from an astrophysical standpointwill be the ability to analyze for multiple elements in a singlegrain. While single grain multielement isotopic analysis hasbeen successful in several cases (Nicolussi et al., 1998a,c;Pellin et al., 2000), it has been difficult to perform becausereducing error bars to an acceptable level for any single ele-ment has often meant consuming most or all of a grain. Suchanalyses will now be routine because a much smaller portion ofa grain will have to be consumed to achieve acceptable preci-sion. In addition to finding correlations among isotopes of asingle element, correlations between isotopes of different ele-ments will reveal stellar evolutionary and nucleosynthetic pro-cesses in much greater detail. Furthermore, since growinggrains preferentially deplete the gas surrounding them of re-fractory elements (Lodders and Fegley, 1997, 1999), the mix ofrefractory and volatile elements in any given grain can giveinformation on the pressure and temperature conditions duringgrain condensation.

Acknowledgments—This paper was originally presented orally at asymposium honoring the career of Robert N. Clayton in June 2001 andis dedicated to him. We thank Associate Editor Ulrich Ott and threeanonymous reviewers for their helpful reviews. We also thank CristineJennings for providing presolar SiC grains. This work was supported bythe Department of Energy, BES-Materials Sciences through ContractNo. W-31-109-ENG-38 and by the National Aeronautics and SpaceAdministration through grants 9510 (AMD) and 4493 (RNC) and anunnumbered grant to MJP.

The submitted manuscript has been created by the University ofChicago as Operator of Argonne National Laboratory (“Argonne” )under Contract No. W-31-109-ENG-38 with the U.S. Department ofEnergy. The U.S. Government retains for itself, and others acting on itsbehalf, a paid-up, nonexclusive, irrevocable worldwide license in saidarticle to reproduce, prepare derivative works, distribute copies to thepublic, and perform publicly and display publicly, by or on behalf ofthe Government.

Work supported by the U.S. Department of Energy, BES-MaterialsSciences under contract W-31-109-ENG-38 and the National Aeronau-tics and Space Administration through grants NAG5-4493 and 9510.

Associate editor: U. Ott

REFERENCES

Amari S., Lewis R. S., and Anders E. (1994) Interstellar grains inmeteorites: I. Isolation of SiC, graphite, and diamond: Size distribu-tions of SiC and graphite. Geochim. Cosmochim. Acta 58, 459–470.

Amari S., Hoppe P., Zinner E., and Lewis R. S. (1995) Trace-elementconcentrations in single circumstellar silicon carbide grains from theMurchison meteorite. Meteoritics 30, 679–693.

Arlinghaus H. F., Calaway W. F., Young C. E., Pellin M. J., GruenD. M., and Chase L. L. (1989) High-resolution multiphoton laser-induced flourescence spectroscopy of zinc atoms ejected from laser-irradiated ZnS crystals. J. Appl. Phys. 65, 281–289.

Bernatowicz T. J., Cowsik R., Gibbons P. C., Lodders K., Fegley B. Jr.,Amari S., and Lewis R. S. (1996) Constraints on stellar grainformation from presolar graphite in the Murchison meteorite. Astro-phys. J. 472, 760–782.

Brandt H.-W., Heilig K., Knockel H., and Steudel Z. (1978) Isotopeshift in the Ca I resonance line and changes in mean-square nuclearcharge radii of the stable Ca isotopes. Z. Phys. A 288, 241–246.

Busso M., Gallino R., and Wasserburg G. J. (1999) Nucleosynthesis inasymptotic giant branch stars: Relevance for galactic enrichment andsolar system formation. Annu. Rev. Astron. Astrophys. 37, 239–309.

Dahl D. A. (1995) SIMION 3D version 6.0. 43rd ASMS Conference onMass Spectrometry and Allied Topics, Atlanta, Georgia. p. 717.

Davis A. M., Nicolussi G. K., Pellin M. J., Lewis R. S., Clayton R. N.(1998) Heavy element isotopic compositions of single circumstellargrains from meteorites: Direct measurement of nucleosynthesisproducts from individual stars. In Nuclei in the Cosmos V (eds. N.Prantzos and S. Harissopulos), pp. 563–566. Editions Frontieres.

Davis A. M., Pellin M. J., Lewis R. S., Amari S., and Clayton R. N.(1999) Light and heavy element isotopic compositions of main-stream SiC grains (abstract 1976). Lunar Planet. Sci. 30, 1976.

Eliel E. R., Hogervorst W., Olsson T., and Pendrill L. R. (1983) Highresolution laser spectroscopy of low-lying p-states in Sr I and Ba I.Z. Phys. A 311, 1–6.

Gallino R., Busso M., and Lugaro M. (1997) Neutron capture nucleo-synthesis in AGB stars. In Astrophysical Implications of the Labo-ratory Study of Presolar Materials (eds. T. J. Bernatowicz and E.Zinner), AIP Conference Proceedings, Vol. 402, pp. 115–153. AIP.

Gallino R., Arlandini C., Busso M., Lugaro M., Travaglio C., StranieroO., Chieffi A., and Limongi M. (1998) Evolution and nucleosynthe-sis in low-mass asymptotic giant branch stars. II. Neutron captureand the s-process. Astrophys. J. 497, 388–403.

Goodman F. O. and Wachman H. Y. (1976) Dynamics of Gas–SurfaceScattering. Academic Press.

Hoppe P., Amari S., Zinner E., Ireland T., and Lewis R. S. (1994)Carbon, nitrogen, magnesium, silicon, and titanium isotopic compo-sitions of single interstellar silicon carbide grains from the Murchi-son carbonaceous chondrite. Astrophys. J. 430, 870–890.

Jennings C. L., Savina M. R., Messenger S., Amari S., Nichols R. H.Jr., Pellin M. J., and Podosek F. A. (2002) Indarch SiC by TIMS,RIMS, and NanoSIMS (abstract). Lunar Planet. Sci. 33, 1833.

Kashiv Y., Cai Z., Lai B., Sutton S. R., Lewis R. S., Davis A. M.,Clayton R. N., and Pellin M. J. (2001) Synchrotron x-ray fluores-cence: A new approach for determining trace element concentrationsin individual presolar SiC grains (abstract). Lunar Planet. Sci. 32,2192.

Kashiv Y., Cai Z., Lai B., Sutton S. R., Lewis R. S., Davis A. M.,Clayton R. N., and Pellin M. J. (2002) Condensation of trace ele-ments into presolar SiC stardust grains (abstract). Lunar Planet. Sci.33, 2056.

Lewis R. S., Anders E., Shimamura T., and Lugmair G. W. (1983)Barium isotopes in Allende meteorite: Evidence against an extinctsuperheavy element. Science 222, 1013–1015.

Lodders K. and Fegley B. (1997) Condensation chemistry of carbonstars. In Astrophysical Implications of the Laboratory Study of


Presolar Materials (eds. T. J. Bernatowicz and E. Zinner), AIPConference Proceedings, Vol. 402, pp. 391–423.

Lodders K. and Fegley B. (1999) Condensation chemistry of interstel-lar grains. In Asymptotic Giant Branch Stars (eds. T. Le Bertre, A.Lebre and C. Waelkens), pp. 279–289. Symposium 191. Interna-tional Astronomical Union.

Lorenzen C.-J., Niemax K., and Pendrill L. R. (1983) Isotope shifts ofenergy levels in the naturally abundant isotopes of strontium andcalcium. Phys. Rev. A 28, 2051–2058.

Lugaro M., Davis A. M., Gallino R., Pellin M. J., Straniero O., andKappeler F. (2003) Isotopic compositions of strontium, zirconium,molybdenum, and barium in single presolar SiC grains and asymp-totic giant branch stars. Astrophys. J., 593 (1), 486–508.

Ma Z., Thompson R. N., Lykke K. R., Pellin M. J., and Davis A. M.(1995) New instrument for microbeam analysis incorporating sub-micron imaging and resonance ionization mass spectrometry. Rev.Sci. Instrum. 66, 3168–3176.

Nicolussi G. K., Pellin M. J., Lykke K. R., Trevor J. L., Mencer D. E.,and Davis A. M. (1996) Surface analysis by SNMS: Femtosecondlaser postionization of sputtered and laser desorbed atoms. Surf.Interface Anal. 24, 363–370.

Nicolussi G. K., Davis A. M., Pellin M. J., Lewis R. S., Clayton R. N.,and Amari S. (1997a) s-Process zirconium in presolar silicon carbidegrains. Science 277, 1281–1283.

Nicolussi G. K., Pellin M. J., Calaway W. F., Lewis R. S., Davis A. M.,Amari S., and Clayton R. N. (1997b) Isotopic analysis of Ca fromextraterrestrial micrometer-sized SiC by laser desorption and reso-nant ionization mass spectroscopy. Anal. Chem. 69, 1140–1146.

Nicolussi G. K., Pellin M. J., Lewis R. S., Davis A. M., Amari S., andClayton R. N. (1998a) Molybdenum isotopic composition of indi-vidual presolar silicon carbide grains from the Murchison meteorite.Geochim. Cosmochim. Acta 62, 1093–1104.

Nicolussi G. K., Pellin M. J., Lewis R. S., Davis A. M., Clayton R. N.,and Amari S. (1998b) Strontium isotopic composition in individualcircumstellar silicon carbide grains: A record of s-process nucleo-synthesis. Phys. Rev. Lett. 81, 3583–3586.

Nicolussi G. K., Pellin M. J., Lewis R. S., Davis A. M., Clayton R. N.,and Amari S. (1998c) Zirconium and molybdenum in individualcircumstellar graphite grains: New isotopic data on the nucleosyn-thesis of heavy elements. Astrophys. J. 504, 492–499.

Ott U., Begemann F., Yang J., and Epstein S. (1988) s-Process kryptonof variable isotopic composition in the Murchison meteorite. Nature332, 700–702.

Ott U. and Begemann F. (1990) Discovery of s-process barium in theMurchison meteorite. Astrophys. J. 353, L57–L60.

Pellin M. J., Young C. E., Calaway W. F., Whitten J. E., Gruen D. M.,Blum J. D., Hutcheon I. D., and Wasserburg G. J. (1990) Secondaryneutral mass spectrometry using three-colour resonance ionization:Osmium detection at the p.p.b. level and iron detection in silicon atthe � 200 p.p.t. level. Phil. Trans. R. Soc. Lond. A 333, 133–146.

Pellin M. J., Davis A. M., Lewis R. S., Amari S., and Clayton R. N.(1999a) Molybdenum isotopic composition of single silicon carbidegrains from supernovae (abstract). Lunar Planet. Sci. 30, 1969.

Pellin M. J., Nicolussi G. K., Davis A. M., Lewis R. S., and ClaytonR. N. (1999b) Heavy element isotopic abundances from individualcircumstellar grains isolated from meteorites: Nucleosynthetic sig-

natures of individual stars. 217th ACS National Meeting, AmericanChemical Society, Anaheim, Calif. NUCL-034.

Pellin M. J., Calaway W. F., Davis A. M., Lewis R. S., Amari S., andClayton R. N. (2000) Toward complete isotopic analysis of individ-ual presolar silicon carbide grains: C, N, Si, Sr, Zr, Mo, and Ba insingle grains of type X (abstract). Lunar Planet. Sci. 31, 1917.

Pellin M. J., Savina M. R., and Calaway W. F. (2001a) RIMS analysisof stardusttrace, isotopic analysis of individual micron-sized SiCgrains. In Resonance Ionization Spectroscopy 2000, Vol. 584, pp.112–122. AIP Conference Proceedings.

Pellin M. J., Calaway W. F., and VeryovkinI. V. (2001b) Laser post-ionisation for quantitative elemental analysis. In ToF-SIMS: SurfaceAnalysis by Mass Spectrometry (eds. J. C. Vickerman and D.Briggs), pp. 375–416. IM Publications and Surface Spectra Limited.

Prombo C. A., Podosek F. A., Amari S., and Lewis R. S. (1993)s-Process Ba isotopic compositions in presolar SiC from the Murchi-son meteorite. Astrophys. J. 410, 393–399.

Saloman E. B. (1994) A resonance ionization spectroscopy/resonanceionization mass spectrometry data service. V. Data sheets for Ga,Mn, Sc, and Tl. Spectrochim. Acta 49B, 251–281.

Savina M. R., Tripa C. E., Pellin M. J., Davis A. M., Clayton R. N.,Lewis R. S., and Amari S. (2002) Isotopic composition of barium insingle presolar silicon carbide grains (abstract). Lunar Planet. Sci.33, 1962.

Savina M. R., Davis A. M., Tripa C. E., Pellin M. J., Clayton R. N.,Lewis R. S., Amari S., Gallino R., and Lugaro M. (2003) Bariumisotopes in individual presolar mainstream silicon carbide grainsfrom the Murchison meteorite. Geochim. Cosmochim. Acta, thisissue.

Sobel’man I. I. (1972) Introduction to the Theory of Atomic Spectra.Pergammon Press.

Tripa C. E., Pellin M. J., Savina M. R., Davis A. M., Lewis R. S., andClayton R. N. (2002) Fe isotopic composition of presolar SiC main-stream grains (abstract). Lunar Planet. Sci. 33, 1975.

Veryovkin I. V., Calaway W. F., Moore J. F., King B. V., and PellinM. J. (2001) An ultra-sensitive time-of-flight mass spectrometer forquantitative surface analysis. 14th Annual Workshop on SecondaryIon Mass Spectrometry, Scottsdale, Ariz., pp. 53–55.

Wright R. B., Pellin M. J., and Gruen D. M. (1981) Velocity distribu-tion of sputtered Zr atoms as determined by laser induced floures-cence spectroscopy. Surf. Sci. 110, 151–178.

Wucher A. (2001) Laser Post-Ionisation: Fundamentals. In ToF-SIMS:Surface Analysis by Mass Spectrometry (eds. J. C. Vickerman and D.Briggs), pp. 347–374. IM Publications and Surface Spectra Limited.

Wunderlich R. K., Hutcheon I. D., Wasserburg G. J., and Blake G. A.(1992) Laser-induced isotopic selectivity in the resonance ionizationof Os. Int. J. Mass Spec. Ion Proc. 115, 123–155.

Wunderlich R. K., Wasserburg G. J., Hutcheon I. D., and Blake G. A.(1993) Laser-induced isotopic effects in titanium resonance ioniza-tion. Anal. Chem. 65, 1411–1418.

Zinner E. (1998) Stellar nucleosynthesis and the isotopic compositionof presolar grains from primitive meteorites. Annu. Rev. EarthPlanet. Sci. 26, 147–188.

Zinner E., Amari S., and Lewis R. S. (1991) s-Process Ba, Nd, and Smin presolar SiC from the Murchison meteorite. Astrophys. J. 382,L47–L50.


Documents

Analyzing individual presolar grains with CHARISMA