Increased sensitivity in laser microprobe mass analysis by using

May 15, 1987 - Laser ablation/ionization technique for trace element analysis. S. S. Alimpiev , M. E. .... G. Rothschopf , J. Zoller , R. Lewis , C. G...
1 downloads 0 Views 526KB Size
Anal. Chem. 1087, 59, 1303-1387 (22) Tan, R. W. I n Kinetics of Ion Molecule Reactlons; Ausloos, P., Ed.; Plenum: New York, 1979; p 271. (23) Kebarle, P.: Caldwell. G.; Magnera, T.; Sunner, J. Pure Appl. Chem. 1985, 57, 339. (24) Handbook of Chemistry and Physlcs, 24th ed.; CRC Press: Boca Ratnn., .Fl-. , 19AR-19Ad .- - - . - - .. (25) Caprioll, R. M. Anal. Chem. 1983, 55, 2387. (28) kncaster, G. M.: Honda, F.; Fukuda, Y . ; Rabalals, J. W. J. Am. Chem. Soc. 1979, 101, 1951. (27) Murray, P. T.; Rabalais, J. W. J. Am. Chem. Soc. 1981, 103, 1000. (28) David, D. E.: Magnera, T. F.; Tlan, R.; Stullk, D.; Michl, J. Nucl. Inst-

1383

rum. Methods Phys. Res., Sect. B 1986, 8 1 4 , 378. (29) Wong, S. S.; Roilgen, F. W. Nucl. Instrum. Methods Phys. Res., Sect. B 1986, 814, 436. (30) Roiigen, F. W., private communication in reviewer’s report.

RECEIVED for review October 16, 1986. Accepted February 2*1987. This work was by a grant from the Canadian Natural Sciences and Engineering Research Council.

Increased Sensitivity in Laser Microprobe Mass Analysis by Using Resonant Two-Photon Ionization Processes F. R. Verdun,’ G. Krier, and J. F. Muller* Laboratoire de Spectrometrie de Masse et de Chimie Laser, UniversitC de Metz, B.P. 794, 57012 Metz Cedex 1 , France

The coupling of a tunable dye laser to a laser microprobe mass analyzer (LAMMA), used to both atomize and Ionize matter, allows slgnlficant increases in sensitivity by using resonance lonizatlon properties (if we compare the results wlth those of a standard LAMMA 500 system). We present first the results obtained from studying metal-doped polymer thin sections and then discuss the matrix effects observed. Finally we will present the analysis of a steel alloy as an applkatlon of our coupling to more current samples.

Resonant ionization mass spectrometry (RIMS) is a relatively new technique that exploits both resonance ionization spectrometry (RIS) and mass spectrometry (MS) properties. RIS is a photoionization method in which atoms in the gas phase are ionized by the absorption of photons (produced by dye lasers) that energetically match transitions between quantum states of these atoms (1,Z). Hurst et al. (I) have proposed five basic photoionization schemes, according to the relative energy position of the intermediate quantum level(s) to the continuum, to obtain a very selective and sensitive ionization of a majority of the atoms. However, to be very selective, RIS needs the use of wavelengths in the visible range which requires in many cases, more than one order. The idea ww then developed to use lower wavelengths (UV range) to ionize atoms in the gas phase by a two-photon ionization process: (A ( w l , w l e) A+) (1)where two photons from the same pulse (1)are successively absorbed by the atoms which are then exited into their own ionization continuum. However, even though the advantage of increased sensitivity is obtained with this procedure, the selectivity of the method is strongly reduced because only one resonant step is involved. To counteract that, a mass spectrometer is generally used as the ion detector (3-6). Fassett et al. (6) have already published very convincing results on noticeable increases in sensitivity for about 50 elements photoionized by a well-tuned laser in the range of 260-355 nm, after a thermal vaporization step. Present address: Chemistry Department, Ohio State University, Columbus, OH 43210.

It is clear that the main advantage of using the RIS properties is to obtain an increase in sensitivity which is particularly important in searching for trace elements. In this range of application, it is often very important to know the precise location in a sample where the trace element has been detected, so the idea has been to associate the RIMS and the “microprobe” techniques. Williams et al. (7) have shown some exciting results with a system in which the vaporization of a small area mm2) of the sample is obtained by a well-focused Nd:YAG laser (1064 nm) and in which an excimer laser is involved in the resonant ionization step. In this system, the resulting ions are analyzed by a time-of-flight spectrometer. Finally, Donohue et al. in a recent paper (8)described the use of an ion microprobe (2 pm of spatial resolution) to produce gas-phase atoms which were subsequently ionized by resonant processes with a tunable dye laser. However, with this system, the use of only one ionizing tunable dye laser in the visible range does not allow one to exploit RIS advantages with a great number of atoms. Our purpose is to examine the possibility of using the RIS advantages with only one UV-tunable laser used for both the vaporization and the resonant photoionization of a spot area in the range of a few square micrometers. That has been done by the coupling of a commercial laser microprobe LAMMA 500 with a dye laser (9-11).

EXPERIMENTAL SECTION The laser mass analyses were performed on a LAMMA 500 (laser microprobe mass analyzer, Leybold Heraeus, Germany) which analyzes thin samples (d 5 1 pm) in a transmission arrangement. In its commercialized version the ionization of the sample is induced by a Q-switched quadrupled NdYAG laser (266 nm, 15 ns), which is power controlled by a set of filters, and focused by a microscope on an area of 2-5 pm2. A visible He-Ne pilot laser beam, collinear with the ionizing laser, locates the sample area to be analyzed. Positive or negative ions are extracted into a time-of-flightmass spectrometer though an “einzel”lens. The common resolution (M/dM) of this instrument is about 800. The signals are stored in a 100-MHz transient recorder (2K memory, Biomation 8100) (12-14). In our modified system the NdYAG laser second harmonic (532 nm) pumps a TDL I11 dye laser (Quantel, France) and the dye laser output is frequency doubled (or frequency mixed with the NdYAG residual IR)before being focused onto the sample (Figure 1) (15). The dyes used in these experiments were the Rhodamines

0003-2700/87/0359-1383$01.50/00 1987 American Chemical Society

1384

ANALYTICAL CHEMISTRY, VOL. 59, NO. 10, MAY 15, 1987

I'il

4- -Di - % I

I

I

f

II

@-

Am

'Cav

AUYA

_---------

286.5 ?

7

8

I

I

I

EM

i FIgure 1. Coupling of the TDL 111 dye laser to the LAMMA 500 microprobe. (A) This work Am, ampllfler of the Nd:YAG laser; AI, first amplifier of the dye laser; A2, second amplirier of the dye laser; C, cavity of the dye laser (oscillator);Cav, cam of the W Y A G laser; DCC, doubling frequency crystal; DL (IR), delay line of residual IR beam; DM, dichroic mirror; EM, end mlrror; f, filter for stopping the visible wavelength (ex: UG 11); G, holographic network; ip, initial pathway in the standard configuration of LAMMA 500; KDPl, doubling frequency crystal; MCC, mixing frequency crystal; OM, output mirror; ROT, adustment by rotation of DCC and MCC crystals; SP, separating prism; VANY,adjustable voltage of the Nd:YAG amplifier;(-) 1064 nm Nd:YAG laser; (- -) 532 nm Nd:YAG laser; (- - -) visible dye laser A,, 573 nm; doubled visible dye laser, A, 286.5 nm; (-) mixed UV dye laser, A,, 225.7 nm. (B) Modification for combination of dye laser beam with 266-nm beam with variable delay. (.e.)

Table I PI 8.99

element

7.73

cadmium copper

7.10

molybdenum

line intensity 1.500 5.000

2.500 170 1.800

wavelength, nm (spectrum)

term combinations

228.8 (I) 324.75 (I) 327.39 (I) 311.21 (I) 313.26 (I)

4d'O 5s2'So 2 4d'O 5s' 5p' 'PI 3d" 4s' 2Sl/z 3d" 4p' 2P3p 3d'O 4s' 3d'O 4p' 2Pl,2 4d5 5s' IS3 4d4 5s' 5p' 7d4 4d5 5s' IS3 4d6 5p' 7P3

---

1961, No. 32. OFrom: Meagers, W. F.; Corliss, C. H.; Schrinbner, B. F. NBS Monogr. (U.S.)

590 and 610 (Exciton distributed by Optilas France). All the organometallic complexes used were prepared and purified by classical methods. The resin monomers (Polaron, France) were used without previous purification. The metal-doped resins (epoxy resin (Spurr's low-viscosity medium) and alradite) were prepared by mixing some organometallic complexes (previously dissolved in a small amount of warm sulfolan) in precise weights of monomer to obtain the desired concentrations. After polymerization, the doped resins were cut on a LKB microtome (0.2 pm thick) and desposited on TEM copper or nickel grids (3.2 mm diameter). The copper-doped albumin f i b s were prepared directly on the TEM grids by drying an aqueous albumin (distributed by Sigma and used without prior dialysis) solution containing precise quantities of CuClzsalt. (The irradiance range used was conditioned to obtain both a visible signal when analyzing with a laser wavelength out of resonance and no saturated signal when tuning laser on resonance. The optimal irradiance range always gave impact sizes between 2.5 and 1.5 pm2. The isotope ratio is correct (without signal saturation).) The alloy steel used was a standard sample produced by the IRSID Laboratory (Saint Germain en Laye, France) reference 108-1 containing 0.12% Cu, 2.92% Cr, 0.69% Mn and 0.538% Mo. Each point used to draw the different curves presented is

the result of an average of 50 spectra. The standard deviation was between 7% and 12%.

RESULTS AND DISCUSSION First we studied the evolution of the signal of Cd+ ion when analyzing, a t different wavelengths, some microcrystals of cadmium acetylacetonate complex (Cd(acacI2)on TEM copper grids. We saw a sharp increase of the Cd+ signal when the ionizing laser was tuned to the intense absorbing line of cadmium, 'So 'P, at 228.8 nm (Table I and Figure 2a). At that point resonance effects in the UV region when directly irradiating solid samples were foreshadowed. It is important to note that this effect has been found on a wide irradiance range (107-10'0W/cm2). Peaks other than those of Cd+ were ions generated by ion/molecule reactions in the microplasma. They are a LAMMA spectra characteristic (16). In order to examine this phenomenon in more detail, we prepared some thin sections of araldite polymer homogeneously doped with precise (Cd(acac)2)complex quantities. The wavelength variation around an intense allowed transition of cadmium is a very characteristic resonance phenomenon

-

ANALYTICAL CHEMISTRY, VOL. 59, NO. 10, MAY 15, 1987

A =

298

1385

nm

(zM-L)+

Cd+ 0.

d

Z

521 M"

x

b.

RZ-

I(Cd)

521

cd

II

I(L0)

- 1.35 228.8

-1

0

1

-0.5

0

*** 220.7

hi

++ ++

I

-0 2i7

291.5

-1

229

231

291 I

I

'I

293

Flgure 2. (A) Ionization wavelength effect on cadmium ionization yield when analyzing some I Acac)* complex crystals at the same irradiance (8 X lo7 W cm-*) and focal condttlons: (a) 50 ns at 298 nm and (b) 5 0 ns at 228.8 nm. Part.' J, d, and e show the extending spectrum at part a (20 ns). M is the mass peak of the complex: 310 amu for "*Cd and L is for the ligand (acetyl acetonate). (B)Cd' intensity vs. wavelength when analyzing araldite cadmium doped thin section (Irradiance, 6 X 10' W cm-2). The ratio, R , of the sum of the peak intensities at 11 1, 1 12, 113, and 114 amu for Cd isotopes to the sum of the peak intensities at 84, 91, 121, and 135 amu varies from 0.25 to 1.35 when wavelength used varies from 225.5 to 228.8 nm. I n the range of 290-295 nm, R is about 0.05.

(cf.Figure 2b) (9-11). Thus an increase in Cd+ signal intensity (therefore in sensitivity) of a factor of -5 was obtained for this sample if we compare the Cd+ signal intensity at 226.7 nm (wavelength close to the most strongly allowed transition of Cd) and at 228.8 nm (resonance) (cf. Table I). To test the reproducibility and possible matrix effects on the observation previously described, we prepared other thin

sections or films of different kinds of polymers, araldite, epoxy resin (Spurr), and albumin, doped with copper acetylacetonate complex. To be free of interferences from the ions possibly emitted by the TEM support sample grids (when the laser is used near the net of the grids), nickel grids were systematically used during this study.

1386

ANALYTICAL CHEMISTRY, VOL. 59, NO. 10, MAY 15, 1987

e 1120

312

32)

326

328

))(I

nD

Figure 3. (A) Comparison of the resonant effect when analyzing different copper (isotope 63) doped matrixes in same irradiance (1.2 X 1OB W cm-') and focal condltions. For epoxy resin sample, electronic saturation effects are responsible of the modification of the ratio I ,/I2(during all the experiments, the isotope ratios of copper atoms have been respected). (B) Evolution of C d signal (isotope 65)at lower irradiance (4 X 10' W cm-') (to be free of electronic saturation effects) vs. the wavelength during the analysis of a copperdoped epoxy resin thin section. Each point corresponds to an average of 20 laser impacts and the signal intensity is normalized in comparison with the signal at 324.75 nm which is taken as 100% (reference point). Dotted lines correspond with two other independent experiments on copper-doped thin sections (0.2 pm) (irradiance (6 f 2) X 10' W cm-'). The effect of the laser polarization (vertical after the doubled harmonic of dye beam) is difficult to estimate at the laser impact point. Table 11. Copper-Doped Samples epoxy resin

(Spurr)

sample type

concn in ppm @%u+ intens in V," 324.75 nm 63Cu+intens in V," 330 nm detection limit: 330 nm (ppm);

araldite albumin

20 7.4 0.15 2

100

100

2.71 0.1 11

0.39 0.07 17

0.04

0.4

3

SIN = 2

detection limit: 324.75 nm (PPm); SIN = 2

"The isotope distribution is as follows: 63Cu, 69%; 66Cu,11%. The results obtained with these samples confirmed the observation, described above; thus when the wavelength of the laser beam was tuned to the more allowed transitions of copper 2S1,2 2Pzi3and 2 S l j 2 2Plj2a t 324.75 and 327.39 nm, respectively, we observed noticeable increases of the Cu+ ion signal. It should be noticed that the absorption of two photons of 324.75 or 327.39 nm cannot promote the copper atoms to a level higher than its continuum; however, it certainly promoted them so close to the ionization protential that the energy required is obtained either by collisions in the microplasma or by autoionization mechanisms (Rydberg levels). Nevertheless great differences in yield increase should be mentioned (cf. Figure 3 and Table 11), when analyzing different kind of matrixes, even though the same quantities of matter were analyzed (same spot area destroyed for each sample and same laser focus (17, 18)). As shown in Table 11, the increase in sensitivity can vary from a factor of about 50 for the copper included in a particular epoxy resin (Spurr) to 5 when included in albumin films.

-

-

At first glance, the resonance phenomena observed and presented here suggest that it takes place in the gas phase. As a matter of fact, there is no particular reason for molecules in the solid state to give a very sharp increase in ionization yield, since it is well-known that the photon absorption in the solid phase is a broad wavelength band of more than few nanometers in width. However, the matrix effects obtained when studying the copper-doped samples show that the ionization process obtained with our system is not governed only by the well-known RIS theory. In fact, it seems that in the irradiance range used here, 109-1011W cm-2, the first photons of the ionizing laser pulse induce the formation of a phase containing primarily neutral atoms and a very small amount of atomic ion. A typical value, obtained from the local thermodynamic equilibrium plasma model (LTE), using the Saha-Eggert equation, is n+/n = 10-5-10-3,in the range of 4900-6500 K (19). (n+/nis the ion to neutral ratio of the species under study.) During the first step (sample atomization) the irradiating laser wavelength value is not critical (solid phase) but, on the other hand, sample atomization yield is very matrix dependent. So, each type of sample in the Same energy and focal conditions has its own atomization yield or has ita own fillup of the "atom reservoir" to be photoionized by photon resonance processes. Evidently, after this stage follows the second step where the other photons of the laser pulse induce the classic two-photon process on the atomized matter when the laser is tuned to a resonant transition. What appeared unusual with our system, was that the curves were particularly broad, 0.4-0.7 nm (Figures 2 and 3). Since these values were too large to result only from Doppler effect, the only parameters responsible for this broadening

ANALYTICAL CHEMISTRY, VOL. 59, NO. 10, MAY 15, 1987

R Figure 4. Evolution of Mo+ Intensity (Isotope 92) vs. the wavelength variation, when studyln some alloy steel fragments (standard 108-1) ( i i a a n C e (2-2.5) X 10$, W cm-2). The two bars at 31 1.21 and 313.26 nm represent the relative intensity of the emission lines of molyWenum (Table I, Meggers, W. F.; Corliss, C. H.; Schrihbner, B. F. NBS Monogr.

( U . S . ) 1061, No. 32).

could be either the laser focal effect or the pressure of the plasma during the second step of the proposed mechanism. However since it is well-known (16) that the high field effect (focal effect on the laser) generally induces both a splitting of the band and a Stark shift of the allowed transitions not observed during our experiment, this parameter does not seems to be responsible for the broadening obtained. So the only parameter responsible for this broadening could be the pressure of the plasma during the second step of the mechanism proposed previously (18). But in many cases when a pressure broadening takes place, a change in line position is expected (pressure shift comparable to the pressure broadening). Nevertheless other experimental data will be necessary to confirm this explanation. After having prepared samples, we studied the ionization of a steel alloy (standard 108-1) and, more particularly, the evolution of Mo+ when varying the ionizing wavelength around the allowed transitions of molybdenum. Having deposited some small metal fragments on a TEM grid previously covered with a noninterfering thin polymer film (Formvar thickness 0.01 pm), we recorded (Figure 4) the different Mo+ intensities vs. the irradiating wavelength. Here also a beautiful resonant effect was obtained with an increase in sensitivity of about five between 311.21 nm (out of resonance) and 313.26 nm (resonance) (18).

CONCLUSION The results presented above, are very encouraging and show how a simple modification of a commercial instrument makes it possible to take advantage of RIS processes for solid analysis. The ionization mechanism in two steps (atomization and photoionization) proposed here has already been detailed when discussing the preliminary results of the ionization of polycyclic aromatic hydrocarbons (PAH) irradiated at different wavelengths (10, 11,17). This is in accordance with the literature and particularly with the model proposed by Egorov et al. (20). Also very important the increases in ionization yield cited here are valid for a difference of only 2 nm between the resonant wavelength of the atom and the wavelength out of resonance. The increases are even more important if we compare these results to those obtained by the conventional LAMMA technique (266 nm). For example the limit of detection of copper in epoxy resins is 20 ppm at 266 nm (far from

1387

the stronger allowed transition) (17,21) but is 0.05 ppm at 324.75 nm (15,17). This particularity is of special interest in biology. However, it must be noted that the nature of the matrix is an important factor. It seems that the use of the epoxy resin (of the Spurr type) is the most appropriate for giving prominence to resonance effects. As for organic molecules, resonance effects have been observed with relatively volatile molecules. These effects do not appear if the molecules are slightly or not at all volatile or if their network energy is quite high (10). However, the use of very short wavelengths (Le., 225 nm) allows the desorption of organic molecules (especially HAP) a t much lower levels (0.005 pJ instead of 0.1 pJ) (10). This particularity has been used advantageously in the analysis of microparticles in two successive steps: (1)desorption of organic materials at very low energy and (2) identification of the mineral support at high energy (with the same particle) (21).

These preliminary results are a first approach and a potential for an increased analytical sensitivity in laser microprobe mass analysis. The present difficulty now is the impossibility to separate the steps of the vaporization and ionization processes. Many ways of investigation are being explored, particularly the recombination of the 266 nm wavelength (NdYag quadrupled) with the dye laser beam after an optical delay (Figure 1B). The ionization mechanisms of solids, matrix effects, or metal-ligand bonds effects in organometallic compounds will be studied with better accuracy.

LITERATURE CITED Hurst, G. S.; Payne, M. G.; Kramer, S. D.; Young, J. P. Rev. Mod. Phys. 1070, 51, 769. Young, J. P.; Hurst, 0. S.: Kramer, S. D.; Chen, C. H. Anal. Chem. 107% 51, 1050A. Donohue, D. L.; Young, J. P.; Smith, D. H. I n t . J . Mass Spectrom. Ion Phvs. 293. . .,. - . 1982. .- - -, 43. .- , .- . Fasset, J. D.; Travis, J. C.; Moore, L. J.; Lyttle, F. E. And. Chem. 1083. 55, 765-770. Fassett, J. D.; Powel, L. J.; Moore, L. J. Anal. Chem. 1084, 56, 2228-2233. Moore,L. J.; Fassett, J. D.; Travis, J. C. Anal. Chem. 1084, 56, 2770-2775. Williams, M. W.; Beekman, D. W.; Swan, J. B.; Arakawa, E. T. Anal. Chem. 1984, 56, 1348-1350. Donohue, D. L.; Christie, W. H.; Goeringer, D. E.; McKown, H. S. Anal. Chem. 1085, 57, 1193-1197. Muller, J. F.; Verdun, F.; Krier, G.; Lamboule, M.; Gondouin, S.; Tourmann, J. L.; Muller, D.; Lorek, s. C . R . Acad. Scl., Ser. 2 1084, 299, 1113-1118. Muller, J. F.; Krier, 0.; Verdun, F.; Lamboule, M.; Muller, D. Int. J . Mass. Specwom. Ion &oc. 1085, 64, 127-138. Verdun. F.; Muller, J. F.; Krier, G. Laser Chem. 1085, 5 , 297. Vogt, H.; Heinen, H. J.; Meler, S.; Weschung, R. Fresenius’ 2.Anal. Chem. 1081, 308, 195-200. Hillenkamp, F.; Unsold, E.; Kaufmann, R.; Nitsches, R. Appl. phvs. 1075, 8 , 341-346. Kaufmann, R.; Hillenkamp, F.; Weschung, R. Med. Prog. Techno/. 1070, 6 , 106. Krier, G.; Verdun, F.; Muller, J. F. Fresenius’ 2.Anal. Chem. 1085. 322, 379-382. Muller, J. F.; Ricard, A.; Ricard, M. Int. J . Mass. Spectrom. Ion Process 1084. 62, 125. Verdun, F.; Krier. G.; Muller, J. F. 10th International Mass Spectrometry Conference, Swansea. U.K., 1985. Verdun, F. Thesis, University of Metz, France, 1985. Furstenau, N. Fresenlus’ 2.Anal. Chem. 1081, 308, 201-205. Egorov, S. E.; Letokov, V. S.; Shibanov, A. N . Chem. Fhys. 1084, 85, 303. Verdun, F. R.; Muller, J. F.; Klein, F.; Sowa, L.; Petit, P. Pollut. Atmos. 1086, NO. 112, 257-263.

RECEIVED for review February 12,1986. Resubmitted August 25,1986. Accepted February 9,1987. Financial support for

this work was received from the Minisere de I’Environnement. F. Verdun thanks Laboratoire d’Etudes et de ContrBle de YEnvironnement Sidgrurgique (LECES) for the financial support of his thesis.