Low-level detection of metal atoms by multiphoton ionization in a low

Perry R. Blazewicz, William B. Whitten, and J. Michael. Ramsey. Anal. Chem. , 1989, 61 (9), pp 1010–1013. DOI: 10.1021/ac00184a018. Publication Date...
0 downloads 0 Views 523KB Size
1010

Anal. Chem. 1989, 61, 1010-1013

(20) ManderJones, 6.; Trikojus, V. M. J . froc. R . SOC.N . S . w . L X V I , 313. (21) Monson, R. S . Advanced Organic Synthesis; Academic Press: New York, 1971; pp 43, 44. (22) Kolthoff, I.M.; Sandell, E.; Meehan, E.; Bruckenstein, S. Quantitative Chemical Analysis, 4th ed.; Macmillan Co.: London, 1969; pp 81, 278. 599. (23) Davies, C. W. J. Chem. SOC. 1938, 2093. (24) Carr, J. W.; Harrls, J. M. Anal. Chem. 1086, 5 8 , 626-631. (25) Lochmuller, C. H.; Colborn, S. A,. Hunnicutt, M. L.; Harris, J. M. Anal. Chem. 1983, 5 5 , 1344. (26) Ruzicka, J.; Hansen, E. H.; Ramsing, A. U. Anal. Chim. Acta 1982. 134. 55-74. (27) Stary,-J. The Solvent Extraction of Metal Chelates; Macmillan Co.: New York, 1964. (28) Adamson, A. W. Physical Chemistry of Surfaces, 4th ed.; John Wiley and Sons: New York, 1982; Chapter V. (29) Bockris, J. O'M; Reddy, A. K. N. Modern Electrochemistry; Plenum Publishing Corp.: New York, 1970; Vol. 2, Chapter 8. (30) Fernandez, M. S.;Fromherz, P. J. fhys. Chem. 1977, 8 1 , 1755. (31) Drummond, C. J.; Greiser. F.: Healy, T. W. Faraday Discuss. Chem. SOC. 1986, No. 81, 95. (32) Avnir, D. J. Am. Chem. SOC. 1987, 109, 2931.

(33) (34) (35) (36) (37) (38) (39) (40)

Weaver, M. R.: Harris, J. M., to be submitted for publication. Ohnesorge, W. E. J. Inorg. Nucl. Chem. 1967, 29, 485. Wolfbeis, 0.; Offenbacher, H. Sens. Actuators 1986. 9. 85-91. Kawabata, Y.; Tsuchida, K.; Imasaka, T.; Ishlbashi, N. Anal. S d . 1987. 3 , 7. Iler, R. K. The Chemistry of Silica; John Wiiey and Sons: New York, 1979. Bousse. L.; de Rooii. N. F.; BeraveM. P. I€€€ Trans. Electron Devicies 1983, ED-30, 1263. Fiat, D.; Connick, R. E. J. Am. Chem. SOC. 1968, 90, 608. Harned, H. S.; Owen, B. B. The fhysical Chemistry of €lectro&ilc Solutions, 3rd ed.; Van Nostrand-Reinhold: Princeton, NJ, 1958. Synovec, R. E.; Yeung, E. S. Anal. Chem. 1985, 57, 2162-2167. Poston, P. E.; Harris, J. M. Anal. Chem. 1987. 59, 1620-1626. Abramowitz, M.; Stegun, I. A. Handbook of Mathematical Functions; Dover: New York, 1972; p 10.

RECEIVED for review November 2, 1987. Resubmitted September 6, 1988. Accepted January 17, 1989. This research was supported in part by the Office of Naval Research, Dow Chemical U.S.A.

Low-Level Detection of Metal Atoms by Multiphoton Ionization in a Low-Pressure Flame Sampling Cell Perry R. Blazewicz, William B. Whitten, and J. Michael Ramsey* Oak Ridge National Laboratory, Analytical Chemistry Division, Oak Ridge, Tennessee 37831 -6142

We are uslng a low-pressure sampling cell to extract specles from an alrlacetylene analytical burner. Slngle-color multiphoton lonlzatlon by a pulsed dye laser Is used for the sensitive detection of atomic specles In the cell. The dye laser excites one of the low-lylng twcbphoton accessible states, and absorptlon of an additlonal photon efficiently ionlres the excled state. Excellent detectlon lknns are reported for sodlum, lithium, calclum, and copper (Na, 20 pptr (parts per trilllon); Li, 100 pptr; Ca, 7 ppb; Cu, 25 pptr). No slgnals were observed for neodymium or alumlnum, presumably due to the consumptlon of the atoms to form molecular specles in the cell. Potasslum shows a much poorer detection llmlt and a quadratic dependence of slgnal on concentration.

Resonance ionization spectroscopy (RIS) is a well-developed method for the sensitive detection of atomic species. Schemes have been developed for the efficient photoionization of most elements (1). Sensitivity has extended to the detection of a single atom of some species ( I , 2). Flame atomization using an analytical burner is a commonly used method for the spectroscopic study of elements contained in solution samples. We have recently demonstrated highresolution spectroscopy using intermodulated fluorescence (3, 4 ) and degenerate four-wave mixing (5,6)on flame-atomized species sampled in a low-pressure cell. The flame from the air/acetylene analytical burner is sucked through an orifice into a vacuum-pumped region where the gaseous species are interrogated by a laser beam. By placing a biased wire inside the low-pressure cell, we can sensitively detect ionization of atomic species via [Z + 11multiphoton ionization (MPI), i.e., by exciting two-photon resonances and then photoionizing them with an additional photon (7). Two-photon absorption

* Author

to whom correspondence should be addressed.

excites states of the same parity as the ground state. When absorption of an additional photon is energetically sufficient to cause ionization of the species, strong ion signals result if the laser is tuned to the two-photon resonance. Two-photon transitions are useful because they can be driven to saturation with common lasers and can be used for efficient sub-Doppler excitation (8). They also permit the study of many states at visible and near-UV wavelengths by using available lasers. Ionization detection is advantageous in that there is generally no background in the absence of the laser beam. Furthermore, it is often possible to convert essentially all excited states produced to ions, and the ions can be efficiently collected with only a biased wire, placing few constraints on the accessibility of the ionization region. The use of two-photon transitions in efficient ionization schemes has been discussed earlier ( I , 9,10) and applied to resonance ionization mass spectrometry of several metallic elements (7). Previous flame studies have demonstrated sensitive detection of molecular species by [ 1 + 11 resonant photoionization in atmospheric pressure flames (11, 12). Our experimental configuration allows the convenient study of samples in aqueous solution as is common in analytical experiments. At the same time, the low-pressure sampling technique provides a cleaner environment for a variety of spectroscopic methods. The sampling cell environment gives a reduction in the number of collisions, in stray light, in background ionization, and control over molecular density or even temperature. Ultimately, this sort of configuration is well-suited for Doppler-free spectroscopic methods with isotopic discrimination as we have shown with four-wave mixing (5).

EXPERIMENTAL SECTION Most of the details of the experimental arrangement of the burner and sampling cell have been given previously (4, 6 ) . The air/acetylene burner is of a type commonly used in analytical studies with the slot burner replaced by a cylindrical

0003-2700/89/0361-1010$01.50/0@ 1989 American Chemical Society

ANALYTICAL CHEMISTRY, VOL. 61, NO. 9, MAY 1, 1989 PREAMP

HV

70

6o

r

t

BOXCAR

v

A

c

-

Go+

~

--

Li

+

4s

50

COMPUTER

I

SAMPLE

4-1

BURNER

1011

4P

I

2D

Figure 1. A schematic diagram of the experimental apparatus.

capillary head burner for greater sampling efficiency. The samples are aqueous salt solutions aspirated into the flame through the burner. The sampling cell consists of a copper block with a hollow vertical channel bored through it. The upper part of the block has a horizontal channel passing near three sides for the circulation of cooling water through the cell. The lower part of the block has two perpendicular, horizontal bores through it giving a center-access portal in each face. Access to the center of the cell is through these four portals. In the present arrangement, three of the portals have metal tube extensions sealed into them with rubber O-rings with windows sealed onto the tubes. The fourth portal has a thin aluminum plate sealed over it with an O-ring. A high-voltage feed-through in the plate has a short length of nichrome wire attached to it, which projects slightly into the interior cavity of the cell. Biasing of this wire allows efficient collection of electrons or ions produced by MPI in the cell. An aluminum plate containing a small aperture, typically 0.5-1.0 mm, is attached to the bottom face of the cell with an O-ring. The top of the cell is connected by a pipe fitting to a large mechanical vacuum pump. With the cell positioned ca. 1-2 cm from the analytical burner, some of the flame gases are pumped through the sampling orifice and through the cell at low pressure, typically 5-10 Torr. A schematic arrangement of the experimental apparatus is shown in Figure 1. The second- or third-harmonic output of a pulsed Nd:YAG laser (Quanta Ray DCR2A) at 532 or 355 nm, respectively, is used to pump a dye laser (Quanta Ray PDL-1). The output pulses of the dye laser of ZlO-ns duration, 1-10 mJ/pulse a t 10 Hz, are admitted to the sampling cell through a window. For sensitive detection, the laser beam is focused in the cell with a 1-m focal length lens. The cell is biased negatively with respect to the burner head by about 100 V to prevent electrons produced by thermal ionization in the flame from entering the cell. A positive bias voltage of 200-400 V on the collection wire in the cell gives efficient collection and avalanche amplification of electrons produced by multiphoton ionization inside the cell. When the dye laser is tuned to a two-photon transition of an atomic species in the flame, copious signal is generally observed with solution concentrations of a few parts per million. The electrons collected on the biased wire are detected with a charge-sensitive preamplifier (EG&G Ortec 142 PC) and typically averaged over 10 laser shots with a boxcar averager (Stanford Research Systems). The signal is then collected and displayed on a laboratory,computer.

RESULTS AND DISCUSSION An energy-level diagram corresponding to our experimental scheme is shown in Figure 2. The laser is tuned to a twophoton transition and it takes one more photon to photoionize the excited state. In addition, in the alkali metals, two-photon

Figure 2. An energy level diagram for the [2

+ 11 MPI experiment.

MPI DETECTION OF TWO-PHOTON TRANSITION TO No 4 d

n

578.9 570.0 570.7 570.6 WAVELENGTH (nrn)

Figure 3. Line shape of the two-photon transition to the sodium 4d state as a function of power density. The approximate power densities in the ionization region are (a) 4 X lo6, (b) 5 X lo8, and (c) 2 X lo8

Wlcm2. Table I. MPI Detection of Atoms Produced in Flames

two-photon state wavelength, nm lowest concn measd estimated detection

Na

Li

587.7 400 pptr 20 pptr

4s 571.1 1 ppb 100 pptr

4d

Ca

cu

5s 5s 600.1 463.5 7 ppb 250 pptr 7 ppb 25 pptr

excitation of the lowest s and d states places the one-photon virtual level near the lowest p state, giving a large resonant enhancement to the two-photon cross section. The transitions used to detect sodium, lithium, calcium, and copper are given in Table I. Scans of the laser wavelength over the transition to the sodium 4d state are presented in Figure 3. The effects of increasing power density on the signal can be seen in the successive scans. High power density c a m s a broadening and tailing of the absorption profiie to the red due to the ac Stark effect on the levels. This ac Stark effect is inherent to the efficient two-photon transition since it arises from the very presence of the state near the one-photon virtual state which gives the two-photon transition a large oscillator strength. A useful description of the ac Stark effect in atomic MPI is given in ref 13. Table I shows the results for limits of detection for the species studied. Sodium, lithium, calcium, and copper all give sensitive detection a t or well below the part-per-billion level using the listed transitions. Linear concentration dependence

1012

ANALYTICAL CHEMISTRY, VOL. 61, NO. 9, MAY 1. 1989 LI

4s

0.01 mg/L LiCO,

Figure 4. Two laser scans over the Li 3s-4s two-photon transition: left, aspirating a distilled water blank: right, a 0.01 mg/L LiCOBsolution ( = I ppb Li).

over many orders of magnitude is found. Figure 4 shows two scans over the transition to the Li 4s state. The spectrum on the left was obtained by aspirating distilled water through the burner. The spectrum on the right results from aspirating an aqueous solution of 0.01 mg/L of LiCO, (1ppb Li) through the burner. When a distilled water sample is run, the clear presence of the lithium transition is due to the unavoidable concentration of lithium salts in the burner head and on the cell. This occurs within a few minutes of aspirating the solutions through the burner, even when they are quite dilute. Here, the detection of 1 ppb of lithium represents a 30% increase of the signal above this background. The signal is calibrated in such a spectrum by measuring the peak heights above base line for both the standard and the blank and taking the difference as being due to the measured sample. The total species engendering either peak is then the ratio of the corresponding peak height to the difference. The limit of detection is then estimated by assuming that a reduction in the signal and, hence, the concentration by the signal-to-noise ratio yields a detectable signal. The signalto-noise ratio is defined as the resonant peak height divided by the base line or nonresonant fluctuation. The dominant noise in the spectrum consists of fluctuations in ion production due to uneven aspiration and changing flame characteristics of the burner. The base line generally represents a lower level of nonresonant ionization of the metal species and, to a much lesser extent, some thermal ionization. To maximize the detection of the flame species a t low concentration, a large orifice ( d = 1.5 mm) is used which samples most of the flame gas. This gives the maximum number of species for detection and, for the detection of copper, for example, maximizes the ratio of the number of atoms from the flame that are detected to those from other sources such as the cell walls. However, the use of the larger orifice also increases the noise in the cell due to thermal ions from the tip of the flame, which penetrates slightly into the cell. For solution concentrations of analyte species of 1ppb, we estimate the density of analyte atoms in the cell to be of the order of lo6 atoms/cm3. This gives ca. lo3 atoms in the focused beam. These numbers are consistent with the calculated number of electrons collected by the biased electrode. The detection of A1 and Nd by [2 + 11 MPI yielded no signal. Cooling of the flame gas as it expands into the cell probably results in formation of molecular oxide species of these reactive metals, removing them from detection. Further studies of this problem are planned. The detection of potassium via s and d states gives a different behavior. The signal shows a quadratic dependence on species concentration, and the signal vanishes a t a much higher concentration, ca. 1 ppm. The same behavior is seen for several states of cesium which have been studied. None

of the Rydberg states of sodium (n = 4-15) show other than a linear concentration dependence of the signal. This may indicate that the states of heavier alkali metals have lower photoionization cross sections in the visible region and that they ionize predominantly by other mechanism such as energy pooling (14-18). Further investigation of these mechanisms is under way. It should be stated that the mechanisms that are operative in these experiments are the same as in most MPI cell experiments and differ from laser enhanced ionization (LEI) in flames (19). In LEI the goal is generally to populate a highlying excited state via transitions induced by one or more lasers. The atom then ionizes thermally, due to the high frequency of energetic collisions in the flame. In contrast, photoionization will often predominate in our experiments unless another mechanism can become similarly favorable. Factors important in the ionization processes include species concentration, collision frequency, thermal energy, and energetics for ionization. Several of these can be varied considerably in the cell experiment. The dependence of the ionization signal on laser power is of interest. For studies such as these in which a total of three photons must be absorbed to ionize the atoms, a cubic power dependence is expected. If the two-photon excited state ionizes by some collisional mechanism, then a quadratic power dependence might be expected instead. Under experimental conditions, power dependence seldom yields such conclusive information. For all the systems studied here, we find approximately a linear power dependence over the range in which we see signals. The focused power densities in our experiments are on the order of 109-10'0 W/cm2. This is roughly enough to saturate the enhanced two-photon step as well as the photoionization step. This leads to a rather complicated power dependence as diffraction fringes, and lengthening along the focus of the high-intensity region with increasing power must be considered. In addition, the ac Stark effect shifts the level out of resonance during high-power spikes in the pulse, giving anomalous intensity dependence (20). Our crude power dependence studies show only slightly greater than linear dependence in low-power, unfocused experiments and even sublinear dependence for focused laser beams. Complete saturation is not attained in these experiments, however, as ionization via the sodium 5s and 4d states yield different signal levels. Four our experimental conditions, it is not possible to differentiate ionization mechanisms by a quadratic vs cubic power dependence. Additional studies of other types may clarify these. In summary, we have demonstrated the sensitive detection of atomic species in a rather simple apparatus by [2 + 11 MPI in a low-pressure sampling cell. The detection in a lowpressure environment allows the possibility of doing subDoppler studies with isotopic discrimination while still using the conventional atomization source. The use of the standard flame atomization source facilitates the use of solution samples and rapid sample interchange. The limits of detection achieved in this way are comparable to those found in LEI (19). For sodium and copper, the most detectable species studied, detection was limited by a constant background of the species. The loss of analyte neutral atoms through the formation of molecular species during the expansion in the cell is a serious interference problem. Replacement of the flame with a rare gas plasma may reduce this problem. Total ion (electron) collection is efficient but naturally does not provide the discrimination of resonance ionization mass spectrometry. This leads to difficulties in the detection of trace amounts of one species in solutions containing large amounts of another easily ionizable species. The background due to thermal ionization and nonresonant photoionization

Anal. Chem. 1989, 6 1 , 1013-1016

of the major species obscures the signal of interest. This can degrade limits of detection by several orders of magnitude and, thus, can severely limit detectability in real samples. The applicability of the technique will depend upon the details of which species are to be discriminated, and a t what levels. In the most general terms, the greatest problems are posed by the detection of a trace of a species of much higher ionization potential than that which is present in excess (such as zinc in seawater or biological fluid). If the excess species ionizes by nonresonant two-photon ionization, its signal will tend to mask the resonant signal. In such a case, only an experiment involving a mass spectrometer or photoelectron spectrometer is plausible. When the ionization potentials are more similar there is much less interference as resonant photoionization is several orders of magnitude larger than the nonresonant process, with proper choice of resonance. Two-color experiments should allow even more sensitive and selective detection of many species.

1013

(5) Ramsey, J. M.; Whitten, W. B. Anal. Chem. 1987, 59, 167-171. (6) Nolan, T. G.; Koutny, L. B.; Blazewlcz, P. R.; Whltten, W. B.; Ramsey, J. M. Appl. Spectrosc. 1988, 42, 1045-1048. (7) Apei, E. C.; Anderson, J. E.; Estler, R. C.; Nogar, N. S.; Miller, C. M. Appl. Opt. 1987, 26, 1045-1050. (8) Levenson, M. D. Introduction to Nonlinear Laser Spechoscopy; Academic Press: New York. 1982. (9) Lucatorto, T. 8.; Clark, C. W.; Moore, L. J. Opt. Common. 1984. 4 8 , 406-410. (10) Engleman, R.; Keller, R. A.; Mlller, C. M.; Nogar, N. S.; Paisner, J. A. Nucl. Instrum. Methods Phys. Res., Sect. B 1987, 826, 448-451. (11) Mallard, W. G.; Miller, J. H.; Smyth, K. C. J. Chem. Phys. 1982, 76, 3483-3492. (12) Smyth, K. C . ; Mallard, W. G. J. Chem. Phys. 1982, 7 7 , 1779-1787. (13) Bagratashvili, V. N.; Ionov, S. I.; Mishakov. G. V.; Semchisen. V. A.; Masalov, A. V. J. Opt. SOC.Am. B: Opt. Phys. 1987,B4, 129-132. (14) Marr, G. V.; Wherrett, S. R. J. Phys. B 1972, 5 , 1735-1743. (15) Koch, M. E.; Collins, C. B. Phys. Rev. A 1978, 79, 1098-1105. (16) Cheret, M.; Lindinger, W.; Barbler, L.; Deloche, R. Chem. Phys. Left. 1982, 88, 229-232. (17) Cheret, M.; Barbier, L.; Linger, W.; Deloche, R. J. Phys. 8 1982, 75, 3463-3477. (18) Barbler, L.; Cheret, M. J. Phys. B 1983, 76, 3213-3228. (19) Travis, J. C. Anal. Chem. 1987, 59, 909-914. (20) Morellec, J.; Normand, D.; Petite, G. Phys. Rev. A 1976, A M , 300-31 1.

LITERATURE CITED (1) Hurst, G. S.; Payne, M. G.; Kramer, S. D.; Young, J. P. Rev. M o d . Phys. 1979, 57, 767-819. (2) Hurst, G. S.; Nayfeh, M. H.; Young, J. P. Appl. Phys. Left. 1977, 3 0 , 229-23 1. (3) Hurst, G. S.; Nayfeh, M. H.;Young, J. P. Phys. Rev. A 1977, A75, 2283-2292. (4) Whltten, W. B.; Koutny, L. B.; Nolan, T. G.; Ramsey, J. M. Anel. Chem. 1987, 59, 2203-2206.

RECEIVED for review November 17,1988. Accepted February 9, 1989. P.R.B. acknowledges a Postgraduate Research Trainee Fellowship from Oak Ridge Associated Universities. This research was sponsored by the U S . Department of Energy, Office of Energy Research, under Contract DE-ACOB840R21400 with Martin Marietta Energy Systems, Inc.

Quantitative Analysis of Low Molecular Weight Polar Compounds by Continuous Flow Liquid Secondary Ion Tandem Mass Spectrometry Tao-Chin Lin Wang,* Ming-chuen Shih, and Sanford P. Markey

Laboratory of Clinical Science, National Institute of Mental Health, 9000 Rockville Pike, Bethesda, Maryland 20892 Mark W. Duncan

Intramural Research Program, National Institute of Neurological Disorders and Stroke, 9000 Rockville Pike, Bethesda, Maryland 20892

Quantitative analyses of low molecular weight ( 100-200) polar compounds [ 1-methyl-4-phenyipyridine (MPP'), 2amlno-3-(methylamino)propanoic acid (synonyms, @-(methylamino)+-alanine or BMAA), and tryptophan] were conducted on a triple-stage quadrupole mass spectrometer configured for contlnuous flow liquld secondary ion mass spectrometry ionization (CF L-SIMS). I t is shown that quantlflcatlon by CF L-SIMS at subnanogram sensltivlty can be preclse (correlation coefficients > 0.99), accurate, speclflc, and routine for compounds not measurable by static L-SIMS. Successful analyses, however, are strongly dependent upon the stability of the film formed by the mobile phase on the probe tip. I n our system, film stability is affected by mobile phase composltlon and flow rate, ion source and probe tip temperature, probe-tip and capillary alignment, film thlckness, and sample composltlon.

INTRODUCTION The techniques of fast atom bombardment (FAB) (1,2)and liquid secondary ion mass spectroscopy (L-SIMS) (3) have

been widely accepted for their ability to analyze polar organic analytes. Effective ionization of polar compounds in both techniques is attributed to the use of low volatility, viscous solvents as a liquid matrix ( 4 , 5 ) . Recently, Ito et al. (6) and Caprioli et al. (7) introduced a refinement to FAB and L-SIMS which offers a number of potential advantages. Rather than external application of sample to the probe tip and reinsertion into the ion source, this approach, now known as continuous flow (or dynamic) FAB or L-SIMS, conveniently introduces the analyte into a continuous stream of mobile phase containing the low volatility, viscous matrix and applies it, via capillary tubing, directly to the surface of the probe tip in the ion source (6-10). This technique combines the convenience of direct liquid introduction characteristic of most chromatographic techniques with the power of FAB or L-SIMS ionization technique. While continuous flow FAB or L-SIMS is finding applications in many areas of qualitative and quantitative analyses, we were interested in applying this technique to the accurate quantification of low molecular weight polar biologicals with minimum chemical modification and sample work-up. In contrast to static FAB or L-SIMS, it was reasoned that CF techniques would permit an increase in signal to noise ratio in the low mass range where analyte

This article not subject to U S . Copyright. Published 1989 by the American Chemical Society