Resonance and Nonresonant Laser Ionization of Sputtered Uranium

Anal. Chem. 2003, 75, 3175-3181

Resonance and Nonresonant Laser Ionization of Sputtered Uranium Atoms from Thin Films and Single Microparticles: Evaluation of a Combined System for Particle Trace Analysis

Downloaded via STOCKHOLM UNIV on January 22, 2019 at 06:42:07 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Nicole Erdmann,*,†,‡ Maria Betti,*,‡ Felix Kollmer,§ Alfred Benninghoven,§ Carsten Gru 1 ning,†,| ⊥ ⊥ ⊥ ⊥ Vicky Philipsen, Peter Lievens, Roger E. Silverans, and Erno Vandeweert

Institute for Environment and Sustainability, European Commission Joint Research Center, T.P. 290, I-21020 Ispra (VA), Italy, Institute for Transuranium Elements, European Commission Joint Research Center, Postfach 2340, D-76125 Karlsruhe, Germany, Physikalisches Institut, Westfa¨lische Wilhelms-Universita¨t Mu¨nster, Wilhelm-Klemm-Strasse 10, D-48149 Mu¨nster, Germany, Institut fu¨r Kernchemie, Johannes Gutenberg-Universita¨t Mainz, D-55099 Mainz, Germany, and Laboratorium voor Vaste-Stoffysica en Magnetisme, Katholieke Universiteit Leuven, Celestijnenlaan 200D, B-3001 Leuven, Belgium

The resonance and nonresonant laser ionization of uranium atoms sputtered from thin metal films and individual micrometer-size uranium oxide particles, respectively, was studied to evaluate a new setup for the analysis of actinide-containing micrometer-size particles. Experiments using nonresonant (193-nm) ionization of atoms and molecules sputtered from micrometer-size uranium oxide particles have shown that the uranium detection efficiencies for sputtered neutral atoms are ∼2 orders of magnitude higher than for secondary ions. In uranium particles of 0.5-µm diameter, 6 × 106 atoms of 235U were easily detected and the isotopic ratio of 235U/238U ) 0.0048 ( 4.6% is in excellent agreement with the certified value. The use of two-color, two-step resonance ionization of the sputtered neutral uranium atoms from thin films was investigated. Several excitation schemes were tested, and a significant population of several low-lying metastable states after ion sputtering was observed. Autoionizing states for double-resonant ionization were determined, and the high selectivity of ionization schemes involving these autoionizing states was illustrated by comparing the flight-time distributions of different sputtered species obtained both by resonance and nonresonant multiphoton (355-nm) laser postionization. Ideally, the options for resonance as well as nonresonant ionization would be combined in a single setup, to obtain a large gain in sensitivity and selectivity. Thus, information about the main components as well as specific isotopic information of a trace element could be obtained from the same single particle. * Corresponding authors. E-mail: [email protected]; [email protected]. † Institute for Environment and Sustainability, European Commission Joint Research Center. ‡ Institute for Transuranium Elements, European Commission Joint Research Center. § Westfa ¨lische Wilhelms-Universita¨t Mu ¨ nster. | Universita ¨t Mainz. ⊥ Katholieke Universiteit Leuven. 10.1021/ac0264426 CCC: $25.00 Published on Web 06/03/2003

© 2003 American Chemical Society

International governmental organizations for the control of nuclear material, such as the Euratom Safeguards Office (ESO) and the International Atomic Energy Agency (IAEA), have implemented new safeguards programs in order to strengthen the control compliance of nuclear facilities declarations and detect undeclared nuclear activities.1-3 One of the concerns is the characterization of microparticles containing actinides as well as fission products, so-called “hot particles”. Such particles are also of high interest for nuclear forensic analyses as well as for a risk assessment of contaminated areas.4-9 They have dimensions varying between several tens of nanometers up to several micrometers, and depending on their size and composition, they typically contain only femtogram to picogram amounts of actinides. Their characterization therefore requires high-sensitivity techniques. A very important piece of information here is the isotopic composition of the actinides contained because this can be related to the history of the material. Instrumental analytical techniques such as secondary ion mass spectrometry (SIMS) have been successfully applied for the localization as well as chemical and isotopic characterization of uranium- and plutonium-containing microparticles.8,9 However, as in many cases the microparticles contain a mixture of actinides, the problem of isobaric interferences (e.g., 238Pu/238U and 241Am/ 241Pu) arises. In this work, a new experimental setup for the analysis of such particles has been evaluated. The main objectives were to gain sensitivity in order to study even smaller (a few hundred (1) Donohue, D. L.; Ziesler, R. Anal. Chem. 1993, 65, 395A-368A. (2) Cooley, J. N.; Kuhn, E.; Donohue, D. L. Proc. 19th Annu. ESARDA Symp. On Safeguards and Nucl. Mater. Manage., EUR 17665 EN, CEC Joint Research Centre, Ispra, 1997; pp 31-35. (3) Donohue, D. L. Anal. Chem. 2002, 74, 28A-35A. (4) Po ¨llanen, R.; Ika¨heimonen, T. K.; Klemola, S.; Juhanoja, J. J. Environ. Radioact. 1999, 45, 149-160. (5) Salbu, B.; Krekling, T.; Oughton, D. H. Analyst 1998, 123, 843-849. (6) Bunzl, K. Analyst 1997, 122, 653-656. (7) Salbu, B. In Plutonium in the Environment; Kudo, A., Ed.; Elsevier: New York, 2001; pp 121-138. (8) Betti, M.; Tamborini, G.; Koch, L. Anal. Chem. 1999, 71, 2616-2622. (9) Tamborini, G.; Betti, M. Microchim. Acta 1999, 593, 1-7.

Analytical Chemistry, Vol. 75, No. 13, July 1, 2003 3175

Table 1. Key Features of Experimental Parameters for the Tests Performed To Evaluate a Combined Setup for Nonresonant (SNMS) and Resonance Ionization (SIRIS) of Sputtered Neutrals University of Mu ¨ nster, Germany SNMS ion gun laser system mass spectrometer additional features tests performed

K. U. Leuven, Belgium SIRIS

Ga+ liquid metal ion source at 30 keV spot size, ∼200 nm ArF (193 nm) excimer laser for nonresonant ionization reflectron-type time of flight image acquisition software operation in SIMS mode possible nonresonant ionization of atoms and molecules sputtered from single submicro-meter-size uranium oxide particles

nanometers) particles in the future and to overcome the problem of isobaric interferences. The proposed setup would be an instrument that combines nonresonant sputtered neutral mass spectrometry (SNMS) with resonance laser postionization of sputtered neutral particles (sputter-initiated resonance ionization spectroscopy, SIRIS). Such an instrument would be equipped with an ion gun, which spot size allows the investigation of submicrometer-size particles, one laser system for resonance and one for nonresonant postionization of the sputtered neutrals, and a time-of-flight (TOF) mass spectrometer. Since it is well known that after ion bombardment the neutral species are produced with much higher yields than the secondary ions, a large gain in sensitivity can be expected if these neutral particles are efficiently postionized with lasers.10 The capability of SNMS to detect cerium oxide particles as small as 75 nm has been shown.11 Resonance ionization spectroscopy (RIS) has been successfully applied to ultratrace analysis of plutonium from bulk samples,12 with detection limits down to a few femtograms of material. In these investigations, RIS has been demonstrated to overcome the problems of isobaric interferences due to its extremely high element selectivity, especially if more than one resonant transition is involved. In particular in the case of microparticles, for which no chemical separation can be performed, resonance ionization of material sputtered from individual grains would be a very elegant method to determine the isotopic composition of a single trace element without isobaric interferences. One possible application would be the isotopic analysis of plutonium traces in uranium particles, where the SIMS technique suffers from isobaric interferences (238U/238Pu, in some cases also 241Am/241Pu, since 241Am is the decay product of 241Pu). Very few studies have so far been performed on laser postionization of sputtered actinide material.13,14 No instrumental setup like the one suggested is commercially available at present. To evaluate the proposed setup, experiments to test the different (10) Kollmer, F.; Kamischke, R.; Ostendorf, R.; Schnieders, A.; Kim, C. Y.; Lee, J. W.; Benninghoven, A. In Secondary Ion Mass Spectrometry; SIMS XII; Benninghoven, A., Bertrand, P., Migeon, H. N., Werner, H. W., Eds.; Wiley & Sons: Chichester, U.K., 2000. (11) Kollmer, F. Dissertation, Westfa¨lische Wilhelms-Universita¨t Mu ¨ nster, 2001. (12) Erdmann, N.; Herrmann, G.; Huber, G.; Ko ¨hler, S.; Kratz, J. V.; Mansel, A.; Nunnemann, M.; Passler, G.; Trautmann, N.; Turchin, A.; Waldek, A.; Fresenius J. Anal. Chem. 1997, 359, 378-381. (13) Arlinghaus, H. F.; Spaar, M. T.; Thonnard, N.; McMahon, A. W.; Tanigaki, T.; Shichi, H.; Holloway, P. H. J. Vac. Sci. Technol. A 1993, 11, 23172323. (14) Goeringer, D. E.; Christie, W. H.; Valiga, R. E. Anal. Chem. 1988, 60, 345349.

3176 Analytical Chemistry, Vol. 75, No. 13, July 1, 2003

Ar+ ion gun at 15keV spot size, ∼3 mm 2 independent tunable lasers: 1 dye laser, 1 OPO for resonance ionization, pumped by Nd:YAG lasers at 532 and 355 nm,, respectively reflectron-type time of flight nonresonant ionization with Nd:YAG laser at 355 nm resonance ionization of uranium atoms sputtered from uranium metal layers

components separately were performed on two different facilities, the University of Mu¨nster, Germany, and the Katholieke Universiteit Leuven (K.U. Leuven), Belgium. The experiments performed and the results obtained are discussed in this paper. EXPERIMENTAL SECTION The key features of the respective systems and the studies performed are summarized in Table 1. While the system in Mu¨nster allowed for nonresonant ionization of single micrometersize uranium oxide particles due to the small spot size of the ion gun, no tunable laser system was available for additional resonance ionization studies. Therefore, all investigations involving resonance ionization were performed at the K.U. Leuven. Since the large spot size of ion gun in this setup did not allow for single-particle measurements, these studies were performed on sputter-deposited uranium layers as a proof of principle. Setup for Experiments with Nonresonant Ionization. Experiments with nonresonant ionization were performed at the University of Mu¨nster. The experimental setup is described in detail elsewhere.11 It consisted of a pulsed gallium liquid metal ion gun, an ionizing laser, and a gridless reflectron time-of-flight (RTOF) mass spectrometer. The instrument could be operated either in SIMS mode or in SNMS mode. Isotopically pure 69Ga+ ions with a kinetic energy of 30 keV and a static primary ion current of up to 1.5 nA were focused to a spot size of 200-300 nm. Primary ion pulse lengths could be varied between 500 ps and several microseconds, the longer pulses were typically applied for SNMS operation where lower pulse repetition rates were used (due to the necessity to match the data acquisition with the laser repetition rate). By scanning the finely focused ion beam over the surface, spatially and mass-resolved secondary ion images (chemical maps) were obtained. For ionization of the neutral species, light (193 nm) from an ArF excimer laser (Lambda Physik, Go¨ttingen, Germany) was focused into the neutral particle plume. The position of this focus was optimized to achieve minimum distance between the sample surface and the laser (∼200 µm) for high geometrical overlap. The laser was operated with a pulse repetition rate of 20 Hz, a pulse width of 20 ns, and a maximum pulse energy of 100 mJ. The photon energy of ∼6.4 eV is sufficient to ionize uranium (ionization energy 6.194 07(6) eV)15 in a single step. In SNMS mode, photoions were extracted into the TOF spectrometer by (15) Solarz, R. W.; May, C. A.; Carlson, L. R.; Worden, E. F.; Johnson, S. A.; Paisner, J. A.; Radziemski, L. J. Phys. Rev. A 1976, 14, 1129-1136.

applying a voltage pulse to the extraction electrode several tens of nanoseconds after the laser pulse. After passing the gridless reflector, the ions were detected with a detection unit consisting of a 10-keV postacceleration region, a channel plate, a BaF2 scintillator, and a photomultiplier. The analog signal was digitized with a digitizer PC-card (Signatec, type DA-500). Instrument control and data acquisition were operated from a PC. The samples for SNMS were prepared by depositing monodisperse uranium oxide particles16 with different 235U enrichment (0.5 atom % depleted, 1 and 10 atom % 235U indicated as enriched) and size (1 and 0.5 µm) on 2 × 1 cm2 graphite foils. By scanning electron microscopy, it was possible to evaluate the population density of the particles. To facilitate the search for particles in the SNMS instrument, samples were chosen that showed a particle density of several particles per 60 × 60 µm2 area. Setup for Experiments with Resonance Ionization. All SIRIS experiments were performed at the K.U. Leuven. The experimental setup and procedures are described in detail in ref 17. The apparatus for the sputtering studies reported here consisted of an ultrahigh vacuum chamber in which an ion gun directed Ar+ ions with a kinetic energy of 15 keV onto a centrally located target foil at 45° incidence. The ion pulse length could be varied from a long pulse mode (∼1 µs) for the spectroscopy measurements down to short pulses of the order of 200 ns, which were typically used for flight-time distribution measurements. The plume of sputtered particles was intersected, parallel to the foil, by two overlapping laser beams from a pulsed optical parametric oscillator (OPO) pumped by a frequency tripled (355 nm) Nd: YAG laser and a pulsed dye laser system pumped by an independent Nd:YAG laser operating at its second harmonic (532 nm). These tunable laser systems generated linearly polarized laser light with laser pulses of ∼6 ns in the wavelength range from 225 to 1600 nm and pulse energies from 4 mJ in the UV up to 50 mJ in the visible range. The laser wavelengths were measured with a wavemeter (ATOS, “Lambdameter”). The atoms sputtered in a polar angle interval of ∼10° around the surface normal were photoionized and subsequently detected in a RTOF mass spectrometer. This polar angle is determined by the combination of the geometry of the ionization volume (defined by the focal condition andseventuallysthe overlap of the laser beams) and the reduced acceptance angle of the mass spectrometer. TOF spectra were preamplified, recorded with a digital oscilloscope, and finally transferred to a PC for further analysis. As the Ar+ ion gun spot size was ∼3 mm in diameter, this system did not allow for the analysis of single micrometer particles. Therefore, samples for resonance ionization investigations were prepared by sputter deposition18 of 0.5-µm-thick layers of depleted metal uranium on 5 × 5 mm2 Ta foils. These were transported under vacuum conditions. However, slight oxidation of the uranium metal occurred. RESULTS AND DISCUSSION Nonresonant Ionization of Ion Beam Sputtered Uranium Atoms from Single Micrometer-Sized Particles. The search for particles was performed using the image acquisition (mapping) (16) Erdmann, N.; Betti, M.; Stetzer, O.; Tamborini, G.; Kratz, J. V.; Trautmann, N.; van Geel, J. Spectrochim. Acta B 2000, 55, 1565-1575. (17) Vandeweert E.; Lievens P.; Philipsen V.; Bastiaansen J.; Silverans R. E. Phys. Rev. B 2001, 64, 195417. (18) Gouder, T. Surf. Sci. 1997, 382, 26-34.

Figure 1. Time-of-flight mass spectra and mass-resolved images (inserts) (acquired for mass 238) of monodisperse uranium oxide particles obtained by sputtered neutrals mass spectrometry: (a) 1-µm size, 10% 235U and (b) 0.5-µm size, depleted uranium (certified to contain 0.492% 235U).

mode of the instrument. Areas (60 × 60 µm2) on the sample were scanned with the ion beam, and spatially resolved secondary neutral particle intensity distributions were recorded for uranium and uranium oxide masses. Once a particle was located, due to the high intensity of the signal on mass-to-charge ratio m/z 238 (U+) and 254 (UO+), concentrated in a small spot, the scanned area was reduced to ∼5 µm2 containing the particle, and the material was successively consumed. The mass spectrum in the mass range for uranium and its oxides was recorded and integrated. Simultaneously, the spatial intensity distributions for masses of interest (m/z 235, 238, 251, 254 and 267, 270) were acquired in the image acquisition mode. In Figure 1, mass spectra in the mass range of atomic uranium, from a 1-µm enriched uranium particle (containing 10% 235U) and a 0.5-µm depleted uranium particle are shown, together with the respective image of the spatial intensity distribution, acquired on atomic mass 238. These mass spectra resulted from consuming only a fraction of the respective particle. The strongest signals were always observed on the atomic mass, while UO+ and UO2+ peaks were also present. For high laser intensities, peaks at m/z 117.5 and 119, which are due to doubly charged uranium, could also be observed. A fraction of these, as well as probably a nonnegligible fraction of the singly charged uranium ions can be attributed to laserinduced dissociation of uranium oxides at high laser intensities. This is confirmed by the observation that the oxide peaks in comparison to the singly and doubly charged uranium peaks show a slightly decreasing trend with increasing laser power. However, a clear quantification of this fraction was not possible during this Analytical Chemistry, Vol. 75, No. 13, July 1, 2003

3177

study. Whether the uranium signal is due to photodissociation of uranium oxide or photoionization of the sputtered atomic uranium directly does not mean any drawback for SNMS applications; however, it may be an important issue for resonance ionization. To estimate the SNMS detection sensitivity, a 0.5-µm particle, consisting of depleted uranium, was consumed completely during a measurement time of 21 min. The small amount of only ∼6 × 106 atoms of 235U contained (∼2.4 fg) was detectable in the mass spectrum with a signal-to-noise ratio of ∼2. From this single measurement, the intensity of all ions contributing to the uranium signal (i.e., U+, UO+, UO2+, and U2+) was summed and a 235U/ 238U isotopic ratio 235

U ) 238 U

∑( ∑(

U+ + 235UO+ + 235U2+ + 235U2+)

235

U+ + 238UO+ + 238U2+ + 238U2+)

238

of 0.0048 ( 4.6% was determined. This is in excellent agreement with the certified value of 0.00492. The reproducibility and precision of this result, however, would have to be studied from repeated measurements of different particles. To directly estimate the gain in sensitivity by using SNMS instead of SIMS, the same sample was interrogated with the two methods sequentially. For the same dose of primary ions, SNMSand SIMS-detected ions were integrated and the numbers compared. In the SNMS mass spectrum, the highest intensity was observed from the U+ signal, whereas for SIMS, the peak on UO+ was the most intense signal. On the U+ masses, the ion signal counted from photoionized uranium atoms was 200 times higher than that from sputtered secondary uranium ions. For uranium monoxide, UO+, the number of photoions obtained by SNMS was still 10 times higher than the corresponding SIMS signal. This demonstrates the high increase in sensitivity that can be obtained by using SNMS instead of SIMS for the analysis of uranium-containing samples. From the results discussed above, we estimate that the determination of the isotopic composition in enriched uranium particles, down to ∼200-nm diameter in size, with 235U enrichment of more than 10%, would still be possible by SNMS. Such particles are of high interest for nuclear safeguards analysis, to discover undeclared nuclear enrichment activities through the analysis of released submicrometer particles. Resonance Ionization of Uranium Atoms after Ion Sputtering. To gain not only in sensitivity but also in selectivity against other elements contained in the same sample, resonance ionization of the sputtered atoms can be used. Here, only the fraction of material present in atomic form after sputtering is available for resonance ionization. Furthermore, it is well known, that metastable states even with excitation energies as high as 4.3 eV above the ground state can be populated to a degree much higher than could be explained by a thermal population distribution.17,19,20 Although the mechanisms governing the population on particular states are still under debate, we should be aware that such a distribution of sputtered neutrals over a large number of metastable configurations can significantly lower the atom reservoir initially available for highly state-selective techniques including (19) Vandeweert, E.; Philipsen, V.; Bouwen, W.; Thoen, P.; Weidele, H.; Silverans, R. E.; Lievens, P. Phys. Rev. Lett. 1997, 78, 138-141. (20) Berthold, W.; Wucher, A. Phys. Rev. Lett. 1997, 97, 2181-2184.

3178 Analytical Chemistry, Vol. 75, No. 13, July 1, 2003

Figure 2. Scan of the excitation laser wavelength for two-color, twostep resonance ionization of uranium atoms after ion sputtering. The second laser was ionizing the atoms nonresonantly into the ionization continuum. Several transitions starting from the ground state and the lowest-lying metastable states of uranium can be identified.

SIRIS. Nonresonant laser probing techniques such as multiphoton ionization do not suffer from this apparent drawback. However, the efficiency of most resonance ionization schemes is in general several orders of magnitude larger compared to nonresonance ionization using the same laser pulse power. The RIS studies were performed using samples of sputterdeposited metal uranium targets. As the uranium metallic layers were oxidized on the surface during transport, a typical mass spectrum contained mass peaks that corresponded to U+, UO+, and UO2+. Subjecting the sample surface to a continuous Ar+ bombardment for a short period (30 s) is known to remove the upper oxide layers.17 This sputter cleaning procedure resulted in a strong increase of the atomic uranium signal and was repeated periodically during the course of the experiment. Resonant transitions to more than 10 different intermediate levels,21 starting from the atomic ground state, were tested. From these, a great variation in the ion signal was observed, probably due to strong variations in the transition rates, which will make a more detailed analysis and comparison of different excitation schemes necessary to find an optimum scheme. Population of Ground State versus Metastable States. To estimate the fraction of uranium particles available in the atomic ground state after sputtering, the population of ground and metastable states after sputtering (with the first and second excited levels located 620.2 (≡ 0.08 eV) and 3800.8 cm-1 (≡ 0.47 eV) above the ground state) was studied by RIS. The excitation laser wavelength was scanned in the vicinity of two strong transitions from the ground state (Figure 2). Here, the energy fluence of the excitation laser was reduced to ∼20 mJ/cm2 per pulse so that resonant transitions were saturated but nonresonant ionization, due to absorption of a second photon from this laser, was suppressed. A second laser was used to ionize the atoms, nonresonantly, into the continuum above the ionization threshold. Several resonances, which could be assigned to transitions starting from metastable levels above the ground state21 and leading to (21) Blaise, J.; Wyart, J. F. International Tables of Selected Constants 20 (Tables Internationales de Constantes), Pergamon Press: Paris, 1992.

Figure 3. Ionization spectrum showing a large number of odd parity autoionizing states in the continuum near the first ionization threshold of U I. The excitation laser was fixed to pump a transition from the ground state to an intermediate state at E ) 30 795 cm-1 (J ) 5). Arrows indicate the total energies in the two-color, two-step ionization schemes, with and without an autoionizing state, which are compared with respect to their overall ionization efficiency in Figure 4.

different, tabulated, intermediate states, were observed. From their intensity, it is obvious that the first and the second excited levels are substantially populated upon ion beam sputtering, while the strongest resonant enhancements still started from the ground state. An abundant population on a large number of metastable states implies that, due to the high state selectivity of RIS, the reservoir of detectable (i.e., ground-state) atoms can easily be reduced to less than 50% of the sputtered particles compared to nonresonant photoionization.19 However, this can be compensated by the fact that the overall ionization efficiency of resonance ionization schemes is generally very high, and commonly available laser intensities are largely sufficient to saturate the resonant processes, which is often not the case for nonresonant multiphoton ionization. Autoionizing States. The RIS efficiency and selectivity can be further enhanced when a second tunable laser is utilized to devise a resonant ionization step. Although this in general makes the overall laser system more costly and complicated, such doubleresonant ionization schemes allow one to strongly increase the selectivity of the RIS process by suppression of the isobaric interferences from nonresonantly ionized species. Ionization through an autoionizing state allows the reduction of the high laser powers that are generally required to drive this step into saturation, due to a cross section that is usually ∼2-3 orders of magnitude higher compared to nonresonant ionization into the ionization continuum. This means that the ionizing step can be saturated at much lower laser powers already, strongly reducing the high background from other nonresonantly ionized species. Thus, these double-resonant, two-color, two-step ionization schemes strongly increase the overall sensitivity and selectivity of the RIS process.17,22 Knowledge of the spectral features of such autoionizing states is, however, virtually nonexistent for most elements, except for a (22) Lievens, P.; Vandeweert, E.; Thoen, P.; Silverans, R. E. Phys. Rev. A 1996, 54, 2253-2259.

couple of model systems. To search for odd parity autoionizing states in uranium, the wavelength of the excitation laser was kept fixed to a strong transition from the ground state 5L6o to an intermediate state at E ) 30 795.4 cm-1 (J ) 5) (324.73 nm), and the ionization laser wavelength was scanned above the ionization potential. Figure 3 shows part of the ionization spectrum obtained. A large number of autoionizing states were observed. The saturation behavior of the two-color schemes was studied. Figure 4 shows the photoion signal in dependence of the laser power for the excitation and the ionization step. For the ionization step, both ionization through an autoionizing state and nonresonant ionization starting from the same first excited level were investigated. The corresponding total energies of the two schemes used are indicated by the arrows in Figure 3. It is obvious that the excitation transition is easily saturated with the laser power available. Incorporating an autoionizing state allows saturation of the overall ionization process at only a fraction of the laser energy fluence needed for the scheme employing a nonresonant ionization step. Figure 5 shows the mass spectra of uranium and its oxides, obtained under two different conditions: (a) using the doubleresonant scheme (excitation step, 324.6 nm at 7 mJ/cm2 per pulse; ionization step, 512.9 nm at 50 mJ/cm2 per pulse), from a sputtercleaned surface and (b) using nonresonant ionization, from a slightly oxidized surface. For nonresonant ionization, the powerful laser light of the Nd:YAG pump laser was used (7 J/cm2 per pulse). With a wavelength of 355 nm, which corresponds to a photon energy of 3.5 eV, the nonresonant ionization of uranium from the ground as well as from the low-lying metastable states would be a two-photon process. From Figure 5, one can clearly observe the strong increase in the photoion yield of U+ by using the double-resonant ionization scheme. However, a direct quantification is not possible since the degree of oxidation of the surface was different for the two measurements. It was observed during the experiments that the sputter yield for atomic uranium Analytical Chemistry, Vol. 75, No. 13, July 1, 2003

3179

Figure 4. Saturation behavior of (a) excitation and (b) ionization steps for two-color, two-step resonance ionization as a function of the laser pulse energy fluence (in mJ/cm2). For ionization, the saturation behavior including a transition into an autoionizing state (dots) and a nonresonant ionization step (triangular symbols) are compared.

depended on the degree of oxidation of the surface, which differed from sample to sample, as well as during the course of the experiments for the same sample. A more detailed quantitative analysis therefore went beyond the scope of these first experiments but would be a topic for further investigation. Flight-Time Distributions. The selectivity of the doubleresonant, two-color, two-step ionization scheme was demonstrated further by comparing the flight-time distributions of the sputtered particles obtained by resonance and nonresonant multiphoton (355 nm) laser postionization upon sputtering of an oxidized sample. These flight-time distributions can be obtained by measuring the photoion signal obtained from these species as function of the delay time t between the impact of a short ion pulse on the sample and the ionizing laser pulses.17 For a fixed distance between the target and the ionizing laser beams, the velocity of the atoms and molecules can be calculated. When the sample-to-ionization volume is known, such experiments enable to obtain information on the kinetic energy distribution with which the particles are ejected during the sputter event. The normalized flight-time distributions of U and UO obtained using nonresonant (355 nm, ∼5 J/cm2) and two-color, two-step resonance laser ionization are shown in Figure 6. In the case of nonresonant ionization only (upper panel of Figure 6), the distribution for U and UO are indistinguishable within the accuracy of the experiment. The fact that the distribution for U follows the same trend as the one for UO proves that a large fraction of the detected atom signal originates from photodissociated molecules by interaction with the intense nonresonant photon field in the ionization volume, 3180 Analytical Chemistry, Vol. 75, No. 13, July 1, 2003

Figure 5. Mass spectra of uranium and its oxides, obtained by (a) two-color double-resonant ionization from a sputter-cleaned metal uranium surface and (b) nonresonant (355-nm) ionization from a slightly oxidized uranium surface.

rather than being sputtered off in the atomic form directly. This was also observed during the SNMS studies in Mu¨nster. In the lower panel of Figure 6, flight-time distributions of the same species are shown but now a double-resonant ionization scheme was employed to ionize ground-state atoms with a maximum of only 50 mJ/cm2 per pulse. At such low laser energy, still a sizable number of UO molecules was ionized. From the clear differences in both flight-time distributions shown in the lower panel of Figure 6, we conclude that in this case the laser pulse energy is too low to induce photodissociation of the oxide to a substantial degree and that the largest fraction of photoionized uranium indeed stems from two-step laser ionization of directly sputtered atoms, which have a kinetic energy distribution different from the molecular ions. The flight-time distributions shown in the upper and lower panels of Figure 6 cannot be directly compared as they are due to different conditions in both experiments. A difference in the flight path leads to a shift in the peak position of the distribution while the different ionization volume geometry contributes to the difference in width of the distributions. Especially this last parameter is difficult to evaluate. We therefore opted not to make any assumptions to convert the directly measured flight-time distributions into kinetic energy distributions. As already indicated, further and detailed studies are necessary to actually quantify which fractions are sputtered off an oxidized surface in the form of uranium atoms or uranium oxide molecules. CONCLUSION It was demonstrated that the use of SNMS instead of SIMS strongly enhanced the sensitivity for the detection of uranium in

Figure 6. Flight-time distributions of sputtered U (closed symbols) and UO (open symbols) ions obtained by (a) nonresonant multiphoton (355 nm) and (b) double-resonant laser ionization.

uranium oxide microparticles; a gain in sensitivity of ∼2 orders of magnitude was determined by direct comparison of the two methods. With only 6 × 106 atoms of 235U, the isotopic composition in a 0.5-µm size depleted uranium particle was measured with good precision and it can be estimated that higher (> 10%) 235U enrichments would be detectable in particles as small as ∼200 nm. Thus, the use of SNMS would make the investigation of much smaller particles possible than is currently possible with SIMS. Investigations of resonance ionization have shown that twocolor, two-step ionization schemes that include an autoionizing state can be successfully applied to detect small amounts of ground-state uranium atoms sputtered from surfaces involving minimal sample treatment. An abundant population on a large number of metastable states combined with the high state selectivity of RIS implies that the reservoir of detectable (i.e., ground-state) atoms would probably be reduced to less than 50% of the sputtered particles compared to nonresonant photoionization.19 However, the overall ionization efficiency of such doubleresonant ionization schemes is so high that the laser intensities

can be attenuated with several orders of magnitude to obtain the same (or even higher) ion signal intensities as in a nonresonant multiphoton experiment. Isobaric interferences can thus be largely avoided and photofragmentation of molecular species can be strongly reduced. A more quantitative assessment of the efficiency of the RIS process would have to be made in the future. During the SNMS as well as the RIS experiments a rather large fraction of sputtered neutral particles was present in the form of oxides. Thus, the role of laser-induced dissociation of UO+ and UO2+ molecules will have to be studied as this will have a strong effect on the total RIS efficiency. Also, matrix effects on neutral fractions sputtered from different uranium compounds, as have been observed by other groups for the sputtering from solids,14 would have to be investigated. If the options for resonance and nonresonant ionization are combined in a single setup, information about the main components (by SNMS) as well as specific isotopic information of a trace element (by SIRIS) can be obtained from a single particle. This could, in the case of particles bearing actinides, be applied to the detection of trace amounts of plutonium in uranium (or in mixed uranium plutonium oxide) particles, for example. In this case, the SNMS analysis could supply information about uranium isotope ratios, the SIRIS about the plutonium isotopes without isobaric interferences (238Pu/238U and 241Am/241Pu). The large number of intermediate states available for uranium as well as for plutonium and the large number of autoionizing states observed so far should encourage research into the design of one-color, two-step schemes in which the ionization transition is in near-resonance with an autoionizing state.22 Such schemes allow one to design a conceptually much simpler (and thus more economical) experimental setup incorporating only one tunable laser system, without trading in on the detection efficiency. ACKNOWLEDGMENT From the Institute for Transuranium Elements, we thank O. Stetzer for his help in the uranium particle preparation, H. Thiele for the SEM measurements, and T.Gouder and F. Miserque for the preparation of the sputter-deposited metallic uranium targets. This work was financially supported by the Belgian Fund for Scientific ResearchsFlanders (FWO), the Flemish Concerted Action Research Program (GOA), and the Interuniversity Poles of Attraction Program (IAP) - Belgian State, Prime Minister’s OfficesFederal Office for Scientific, Technical and Cultural Affairs. E.V. is a Postdoctoral Fellow of the FWO. Received for review December 19, 2002. Accepted March 27, 2003. AC0264426

Analytical Chemistry, Vol. 75, No. 13, July 1, 2003

3181