Laser ablation and selective excitation directed to trace element analysis

'.

LASER ABLATION

Arm

SELECTIVE EXCITATIor~ DIRECTEU TO TRACE ELEMENT ANALYSIS

by

V. H. S. Kwon 9

1

':C. 1980

TECHNISCHE HOGESCHOOL DELFT

LUCHTV RT- EN AJI HEVAAf1TTECHNIEK Bm~..:v)" ~tEl(

Kluyverweg 1 - DELFT

...

LASER ABLATION AND SELECTIVE EXCrI'ATION DIRECTED TO TRACE ELEME:NT ANALYSIS

by

V. H. S. Kwong

'",

Acknowledgements

I wish to thank the members of my Thesis Committee, Drs. J. H. de Leeuw, J. B. French and R. M. Measures, for their comments and careful reading of my originaJ. manuscript. I am especially grateful to Dr. R. M. Measures, who has worked With me closely in the past few years in the preparation of this dissertation and for bis criticaJ. camments at numerous points in time. I am also grateful to N. Drewell for bis invaluable help throughout the course of this work.

Acknowledgements are also due to the support staff of the Institute for their cooperation.

No word here could adequately express my gratitude to my wife, Julia, for her unflagging support throughout this work.

The "'f'inancial assistance received from the U.S. Air Force Office of

Scien t'ific Research under grant No. AF-AFOSR 76-2902B, and the National Research Council of~Canäda is gratefully acknowledged.

Abstract

A Trace (Element) Analyser Based on Laser Ablation and Selectively Excited Radiation (TABLASER) is proposed as a new ultramicro-ultratrace technique for quantitative element analysis. Measurements of trace quan-tities of chromium in samples of NES standard reference material (steel), doped skim milk powder and doped flour have been undertaken. There is a linear

45°

slope for Log/Log plot dependence of signal versus concentration that extends at least up to 1.3% (concentration by weight) in the case of chromium. The detection limit for the current unoptimized system is in the ppm range which corresponds to the absolute detection limit of lO-13g • In this investigation, no chemical interference effects have been observed. This can be explained in terms of the atomization of all molecular compounds in the extremely high temperature attained during laser ablation, and the subsequent fast expans ion int" a low vacuwn region minimizing compound formation. Two physical interference effects can still be observed:

differential mass vaporization and inhomogeneous spatial and temporal dis-tribution of fast expanding analyte. Other physical interference effects, however, are either absent or minimized by the streaming motion of the neutrals. The differential Doppler shift between the atoms along the line of observation reduces self-absorption even at high analyte concentrations. Mie scattering is virtually eliminated by a judicious choice of a space-time window, effectively avoiding the particulate debris that expands into the low vacuwn region at a slower speed. TABLASER, besides being adaptable to in situ analysis, has three additional important areas of application: (1) the study of the expansion dynamics of neutral species resulting from laser-target-interaction, (2) the measurement of radiative lifetimes of atoms, and (3) ultratrace element water analysis.

1. 2.

3.

4.

CONTENTS Acknow1edgements Abstract INTRODUCTION

PRINCIPLE OF THE TABLASER

2.1 Under1ying Principle of the TABLASER 2.2 Attractive Features of the TABLASER

2 ~3 Rationale for Studying Atomic Chromium

THE TAB LASER FACnITY

3.1 Summary of Major Components of the TABLASER

ii iii 1

3

3

3

4

4

4

-3.1.1 Ablation Laser 5

3.1.2 Nitrogen Laser Pumped Dye Laser 5

3.1.3 Time De1ay Unit and Arrangement of Optical Axes 5

3.1.4 Geametry of Observation Window 6

3.1.5 Photomultip1ier 6

3.1.6 Ablation Chamber and lts Acces,?ories 7

3.1. 7 Vacuum System 7

3.2 Sample Preparation 3.3 Experimental Procedures

MATRIX EFFECTS AND THE TABLASER SUSCEPTIBnITY 4.1 C1assification of Interference Effects 4.2 Chemical Interference 7

B

9

9

10

4.2.1 Source Interference (Meta11oid Effect) 10

4.2.2 Cation-Anion Interference 10

4.2.3 Cation-Cation Interference 10

4.2.4 Oxide, Hydroxide and Carbide Formation, Etc. 11

4.3 Physical Interference 4.3.1 4.3.2

4.3.3

4.3.4 4.3.5 4.3.6

4.3.7

Spectral Line Interference Ionization Interference Se1f-Absodltion

. Particulate Scattering and Wall Scattering

Col1isional Q~ching and Mixing

Differential Vaporization

Spatial and Temporal Interference

11 11 11

la

12 13 13 13

4.4 The TABLASER Susceptibi1ity to Matrix Effects

4.4.1 Evidence Indicating Absence of Various Chemica1

~

13

Interference and Ionization Effects 13 4.4.2 Possib1e Physical lnterference Effects 15

4.4.2.1

4.4.2.2 4.4.2.3 4.4.2.4 4.4.2.5

Particulate Scattering and Wal1 Scattering

Se1f-Absorption

Co11isiona1 Mixing and Quenching Spectra1 Interference

Other Cohstraints of the TABLASER

4.4.3 Evidence Indicating Adverse Physica1 Interference 15 15 15 16 16 Effects 17

4.4.3.1 Differentia1 Vaporization Effect, Spatial, Temporal Interference Effect . 17

5. SIGNAL OPl'IMIZATION FOR THE TAB LASER AND lTS RANGE OF APPLlCATION

5.1 Detection Limit, Precision and Limi tation of the .

Present Faci1ity

5.2 Possib1e Improvement on Detection Limit.

5.2.1 Signal lmprovement by Increasing the Probing Volume

5.2.2 Signal Improvement 'by lncreasing the Efficiency of the Co11ection Optics .

5.2.3 Stray-Laser-Light and lts Limitations

5.3 Projected Detection Limit and Linear Dynamic Range of the Response Curve

5.4 Other Characteristics of the TABLASER

5.4.1 Possibi1ity of MUlti-E1ement Analysis 5 .4~~2 Spatial and Depth Reso1ution

5.4.3 Reduction of the Analysis Time

5.4.4 Possibi1ity of Ultratrace Element Water Analysis

6. PHENOMENOLOOICAL STUDY OF THE EXPANDING PLASMA/NEUTRALS AFTER THE ABLATION EVENT

6.1 Review of Various Studies on Laser-Induced-P1asma Expansion in Vacuum

6.2 Study of Laser Plasma/Neut ral Species Expansion by the TABLASER

6.3 lnterpretation of the Absence of Chemica1 and Some Physica1 Interference Effects in Terms of Free Expanding Ab1ation P1ume

6.4 Imp1ications of the Free Expansion Model

17 17 18 18 19 19 20 20 20 20 21 21 22 22 22 24 25 V

7.

LIFETIME MEASUREMENTS

Page 25 7.1 A Review of the Needs for Atomic Lifetime Measurements

and the Associated Problems of Same of the Current

Measuring Techniques 25

7.2 Attractive Features of the TABLASER Approach to Lifetime

Measuremen t 26

7.3 Measurement of the Lifetime T for the 7P2o State of

Chromium 27

8.

SUMMARY AND CONCLUSION 27

31 REFERENCES FIGURES APPENDIX 1: APPENDIX 2: APPENDIX 3: APPENDIX 4: APPENDIX 5: APPENDIX 6: APPENDIX 7: APPENDIX 8(a): APPENDIX 8(b): APPENDIX 9: APPENDIX 10: APPENDIX ll(a): APPENDIX ll(b):

RELATIONSHIP BETWEEN GROUND STATE ATOM DENSITY PRIOR TO LASER EXCITATION AND MEASURED INDUCED FLUORESCENT SIGNAL INTENSITY

ESTIMATION OF THE PEAK POWER OF THE NITROGEN LASER PUMPED DYE LASER AT 428.9 NM USING A HP 5082-4220 PHOTODIODE SATURATION IRRADIANCE FOR CHROMIUM 7P2o - 7S3 TRANSITION ESTIMATION OF THE DIAMETER OF THE DYE LASER PROBE BEAM AT THE OBSERVATION WINDQW

SIGNAL OPrIMIZATION BY ENLARGING THE EXCITATION VOLUME EFFICIENCY OF THE PRESENT COMBINED LIGHT DUMP

ABSOLUTE DETECTION LIMIT FOR AN IMPROVED SYSTEM

NEUTRAL CHROMIUM DENSITY ESTIMATION PLASMA DENSITY ESTIMATION

ESTIMATION FOR VARIOUS COLLISIONAL RELAXATION TIME IN

LASER INDUCED PLASMA

PLASMA TEMPERATURE ESTIMATION

COLLISIONAL MIXING AND Q,UENCHING OF EXCITED ATOMS

DIFFERENTIAL SHIFT OF ABSORPTION AND EMISSION LINE PROFILE DUE TO THE STRFA.MING MOTION OF THE ATOMS

1. INTRODUCTION

The applications of trace and ultratrace element analysis to low levels of concentration (ppm to sub-ppb) are numerous. The major applications include: the characterization of ultrapure materials tor advanced

technolo-gies (Zief and Speight, 1972); the establishment of meaningful toleranee ' limits for pollutants and toxic sub stances (Laitinen, 1971); the analysis of geological and cosmological samples (Keil,1972; Morrison, 1971); the determination of the distribution and quantity of trace constituents for effective use of ocean resources (Carpenter, 1972); the gathering of scientific evidence for forensic investigation (Moenke et al, 1973); the study of the physics and chemistry of semiconductors (Wittry, 1973); the elucidation of the role of trace metals in biological functions (Hall et al, 1974); and the determination of the mechanism by which heavy metals induce

. .,; toxicity (Chakrabarti, 1973). With the growing environmental concern, the demand for identification and quantification of pollutants from the low ppb to the sub-ppb levels is increasing. In fields such as biomedical research and semi conductor development, the sampled regions are of sufficiently small dimensions as to require ultramicro-ultratrace analytical techniques

(Chakrabarti, 1973).

A selection of various trace and ultratrace analytical techniques commonly used are outlined in Table 1. Among these, ion microprobe (IMMA) , emission lasermicroprobe (ELM) , and the laser microprobe mass analyser (LAMMA) possess both high sensitivity and ultramicro-samp:ting capabili ty (see Table 2).

The ion microprobe mass analyser (IMMA), developed by Liebl in 1967, has by far the highest analytical capability both in resolution and sensi-tivity (Heinrich and Newbury, 1975a; Hall et al, 1974). However, it

experienees severe matrix effects, has to operate in a high vacuum of 10-6 torr and involves extensive sample treatment especially for non-conducting materials (Heinrich and Newbury, 1975b). In addition, the interpretation of results is difficult because the limit of detection for different ele-ments varies over several orders of magnitude (Heinrich and Newbury, 1975a; Hall et al, 1973c).

Among the family of laser microanalysers (Fig. 1), the emission laser microprobe (Harding-Barlow et al, 1973) has been used extensively in the past decade for trace element analysis in a variety of matrix materials

(Hall et al, 1974b; Keil et al, 1973). This approach employs a pulsed laser to ablate a small quantity (10- 7 g) of material and relies on a spectroscopie analysis of the resulting plasma emission for a measurement of the trace consti tuents . This kind of laser microprobe has the virtue of simplicity, real time monitoring without sample preparation and can be used for simultaneous,multi-element analysis. However, its detection limit is hampered by the intense background radiation emitted by the ablated plasma (piepmeier and Osten, 1971; Treyl et al, 1971), and the interpretation of the spectral information is further complicated by various matrix effects (Margoshes, 1973; Marich et al, 1970;, Piepmeier et al, 1971; Piepmeier and Malmstadt, 1969). The sensitivity of ELM is improved by using an auxiliary spark discharge to enhance the line radiation fram the elements of interest

(Marich et al, 1970; Harding-Barlow et al, 1973). Unfortunately, spatial resolution has to be sacrificed and memoryeffects (Blackburn et al, 1968)

can be introduced by the auxiliary cross spark electrodes. Osten et al

(1973)

and Ishizuka et al

(1977)

caIDbined the laser microprobe with atomic absorption but little improvement in sensitivity has been achieved and concentrations are still restricted to 10 ppm or more depending on the specific element under investigation.

In

1975,

Hillenkamp et al developed the laser microprobe mass analyser (LAMMA) by coIDbining the laser microprobe with a time-of-flight mass spec-trometer. This new micro-sampling trace element detection technique has a detection limit in the ppm range for most elements. However, this technique suffers from several disadvantages that tend to limit its range of applic-abi;ity • Besides the measurement having to be carried out in a vacuum of 10- torr, the sample has to be in the form of a thin section. This prevents in situ measurements in general. Interference between species limits the detection limit that can be achieved (Wechsung et al,

1978).

The ideal ultramicro-ultratrace element analyser should have the following characteristics:

1. No sample preparation, allowing in situ measurement,

2. High sensitivity and low detection limit, i.e., ppb and attogram,

3.

Freedom from matrix effects (chemical and physical) so that calibration is independent of the substrate containing the element,

4.

Linearity of response over wide dynamic range of concentrations •

•

5.

High selectivity for element of interest.

6.

Selective ultramicro-sampling capability.

7.

Depth profiling capability;

8.

Real-time analysis,

9.

Minimum variation in sensitivity between elements. 10. Simultaneous multi-element measurement possible.

11.

Capability of isotope ratio measurements.

Unfortunately, all of the ultramicro-trace (Chakrabarti,

1973)

analysis tech-niques mentioned above fall short of some or most of these characteristics. The purpose of this thesis is to introduce to the field of microscopic trace

element analysis a new and versatile approach: TABLASER - , Trace (Element) Analyser Based on Laser Ablation and Selectively Exci ted Radiation'. This approach combines the microselectivity and in situ capability of the laser microprobe with the high sensitivity of laser selective excitation and

saturation spectroscopy (Measures,

1968;

Piepmeier,

1972;

Omenetto et al,

1973;

Daily,

1977).

The high sensitivity associated with laser selective excitation spectro-scopy (or resonance fluorescence) has been demonstrated by several investigators

(Denton et al, 1971; Neumann et al, 1974; Kuhl et al, 1973; Smith et al, 1977; Weeks et al, 1978) in the past decade. Experiments with sodium have demonstrated that an atom density as low as 102 cm-3 can be detected by· resonance fluorescence (Fairbank et al, 1975). Saturation optical non-resonance emission spectroscopy (SONRES) (Gelbwachs et al, 1977; Hohimer et al, 1977) has a sensitivity 6f few atams cm-3. Furthermore, several studies have shoWn that narrow band, suitably tuned, high power laser radiation can

saturate atomic transitions (Kuhl et al, 1972; Omenettoet al, 1973; Pi epmei er , 1972; Daily, 1977). Laser saturation yields an upper limit to the magnitude of the induced fluorescent signal achievable with any given analyte concentra-tion, extends the range of linearity' as a function of concentration and makes the peak amplitude of fluorescent signal independent of both the quenching rate and laser intensity fluctuations (Measures, 1968). These features,are utilized in the TABLASER and when combined wi th' the advantages of the laser microprobe mentioned earlier suggest that the TABLASER could closely approach the ideal ultramicro-ultratrace element analyser outlined above.

2. PRINCIPLE OF TEE TABLASER

2.1 Underlying Principle of the TABLASER

The underlying principle of the TABLASER can be described in the following manner:

Laser ablation of the sample target generates a plume of dense high temperature plasma contaminated with particulate debris. The material within this plume rapidly expands into the low vacuum (10-3 - 10-

4

torr) of the ablation chamber. Af ter an appropriate time delay the outward streaming highly recombined fan of material is interrogated by a brief pulse of intense dye laser radiation tuned to saturate one of the strong resonance transitions of the element to be detected. The resulting resonance fluorescence or the direct-line fluorescence peak can be related to the concentration of the trace element within the solid target (Appendix 1) The basic concept of the TABLASER is illustrated in Fig. 2.

2.2 Attractive Features of the TABLASER

The TABLASER has the following attractive features:

1. Mie scattering from the particulate debris, almost invariably associated with laser ablation, is eliminated by the judicious choice of a space-time window that effectively exploits the differential expansion velocity between the particulates and the atomic species

2. Rapid electron ion recombination and collisional radiative de-excitation in the early phase of the expansion ensures that within the Mie-scattering-free window, a large fraction of the target material atoms reside in their ground state. This ensures that the resonance fluorescence, excited by the~P!obe dye laser, truly reflects the atom density of the constituent of interest and is insensitive to the conditions within the expanding fan of target material.

3. Plasma background emission within the space-time window chosen to mini-mize Mie scattering is essentially zero.

4. Free streaming of the atomic species into the modest vacuum minimizes the loss of atoms through the formation of molecular compounds thereby avoiding the chemical interference effects that plague most of the other trace element atomic fluorescent techniques.

5.

Saturation of the resonance transition by the dye laser ensures that the signal is maximized while susceptibility to dye laser fluctuation is minimized and insignificant preabsorption of the probe beam occurs (Brod et al, 1976).

2.3 Rationale for Studying Atomic Chromium

To demonstrate the capability of the TABLASER, chromium was used as the analyte • A diverse array of carefully prepared samples doped with chromium were used. These included NBS-SRM steel samples, ~pecially manufactured pellets of freeze-dried skim milk powder and flour samples. A detailed

des-cription of these and other samples will be given in the next chapter. The energy levels and transition of interest of Chromium are shown in Fig 3. In the present series of experiments , the low resonance state 7P2° f , was chosen to be excited selectively by tuning the dye laser to 428.9 nm.

The upper resonance state 7P40 _{was not used because of the proximity to the}

3Pl ' sta te thit may provide an addi tional channel to d eplete the laser exci ted state population.

Resonance fluorescence was used in this investigation because of the usually high value of the fundament al constants, i.e., A-values, charac-terizing the resonance fluorescent transition. Its radiance is significantly greater than for other transitions (Omenetto and Winefordner, 1978). The use of resonance fluorescence has one major weakness. It is difficult to differentiate between stray laser light and the resonance fluorescence spectrally. However, this problem can be eliminated by the direct-line-fluorescence (Appendix 1) .

Chromium was chosen in this investigation for several reasons. Its atomic transitions were directly accessible to the nitrogen laser pumped dye laser available in our laboratory It has an energy level structure conducive

to the series of experiments necessary to enhance the physical understanding of the laser induced expanding plasma/vapor; the radiative lifetimes of the excited state of interest have already been accurately documented (Measures, Drewell, Kwong, 1977; Bieniewski, 1976; Mar ek , 1973) In addition, it is an essential trace element in human and animal nutrition (Black, 1976); and as the toxic nature of chromium is bet ter understood, concern over chromium pollution of the environment has also increased (Wolf, 1978)

3. TEE TABLASER FACILITY

3.1 Summary of Major Components of theTABLASER

The TAB LASER has six major components (Fig. 4): a Q-switched ruby laser (A) for ablation; a nitrogen laser pumped dye laser (B) for the selective

excitation of the ablated atomie species; an ablation ehamber (C) that held the target sample turntable in a modest vaeuum.; a receiving optical arrange-ment that coupled the laser induced resonance fluorescence into a photo-detection system (D) through appr~riate speetral discrimination; a data acquisition and di~lay system (E); a delay unit

(i)

that controlled the triggering of the dye laser p~se relative to the rtiby laser to optimize the signals observed. A detailed. schematie and the aetual layout of the faciiities are presented in Fig. 5a, b.

3.1.1 Ablation Laser

The ablation laser was a Control Data Corporation ~RG 104A ruby laser which operated at 694.3 nm. It was passively Q-switched by Eastman Q-switched

solution Al0220, a stabiliz~dsolution o~ c~ytocyanine in acetonitrile. The ' '# typical output energy in the ablation experiments ranged from 10 mJ to 30 mJ

in a pulse of 20 to 30 ns. The shot-to~shot reprodueibility of the rtiby laser was around

±

15 to

±

3(Jl/o. •. ,The temporal profile of a ~ypical Q-switehed ruby laser pulse is shown in Fig. 6.

The ruby laser pulse was focussed on the sample surf ace by a lens system to a 100 ~m diameter whieh was estimatedon the basis of the crater diameter observed under an electron microscopè. ·Typical micrograms are shown in Fig. l4a, b. The intensity of the laser at the target surface was estimated to be in excess of 109 W em- 2 "

In the eariy phase .of the invëstigation, a Phase-R Co. model DL-2l00A flashlamp-pumped dye laser operated in broadband emission mode was used

(dye: Coumar~n 120; centre emission wavelength ~440 nm, bandwidth 4 nm). The output energy of the, ablation laser ranged from 10 mJ to 30 mJ in a pulse

close to 400 ns (FWHM). THe shot-to-shot reproducibility of the laser was around

±

5%. The laser was used to investigate the effect of long ablation laser pulses on the indueed fluorescent signal. The laser intensity at the target surface was estimated to be in exces~ of 107 Wjcm-2 •

3.1.2 Nitrogen Laser Purrwed Dye Laser

Both the nitrogen laser and the dye laser were designed and constructed in this laboratory. Detailed descriptions of their construction and design have been given elsewhere (Drewell, 1979). The spectral width of the dye laser was reduced by a beam expander-grating system (as described by Hänsch

(1972) but exclusive of polarizer and etalons). The spectral overlap between the dye laser and the line being excited was ehecked by splitting off a small fraction of the dye laser output and passing it through a Fabry-Perot inter-ferometer and comparing the subsequent interferomet;r-ic ring system with that produced with the prefiltered light of a hollow cathode chromium lamp

(WestinghouseWL-22934A). The peak power of the dye laser pulse was esti-mated to be about 1 kW (Appendix 2); its duration was close to 4 ns (FWHM)

and i ts spectral line width was better than 0.01 nm (Fig. 7).

3.1.3 Time Delay Unit and Arrangement of Optical Axes

The precise time delay betwee~ the moment of ablation and the arrival of the dye laser pulse at the preselected location in front of the sample surface was attained by triggering the nitrogen laser from a 556 Tektronix

oscilloscope. The input pulse to the 556 oscilloscope was derived from a photodiode that sampled a small fraction of the rtiby laser output.

The three optical axes (the rtiby laser, dye laser and the receiver opties) were arranged to be mutually orthogonal within the vacuum chamber. A better appreciation of the sampling, probing and observation geometry can be obtained from Fig. 8. This configuration is dictated by the following considerations: the projected entrance slit of the monochromator (observation window) had to be matched with the probing beam; a spatial separation between atomie species and the slow moving particulate debris had to be achieved; a scan of the vapour was required to facilitate the evaluation of the optimal spatial and temp or al probing location.

3.1.4 Geometry of Observation Wind ow

For the bulk of our observations, the.cross-section of the observation volume was 0.1 cm x 0.5 cm and was defined by the entrance slit of the

mono-chromator and the magnification of the collection f/4.5 lens used in the

system. The diameter of the dye laser beam was less than 0.03 cm and for most measurements, it was located about 0.55 cm in front of the target sample and

centred within the field of view of the monochromator. The small dimension and precise location of the dye laser beam was dictated by: (a) the high power density required to attain complete saturation of atomie transition discussed in Appendices 1 and 3; and (b) the need to eliminate the dynamic effect introduced by the high streaming velocity of the excited neutral species as they moved across the field of view of the receiver opties. The significanee of the observed lifetime is to be diseussed in Chapter 7. 3.1.5 Photomultiplier

Measurements were made on an ll-stage RCA developmental type C3l034 photomultiplier with a 4 mm x 10 mm 60ER responsephotocathode. The quantum efficiency at 428.9 nm was around 16%. The photomultiplier tube was wired for fast pulse response mode. The risetime of the t~e estimated on the basis of dark pulses was less than 3 ns at 1500 V. The PMT was encased in an aluminum housing mounted directly onto the exit port of the 8pex 170011 Czerny Turner monochromator with a resolution of 1 nm/mm.

In the next phase; a gated l4-stage RCA 7265 photomultiplier tube with a 2 inch diameter 8-20 photocathode coupled with a narrow band interference filter (at 431 ± 5 nm) were used. This arrangement lays the groundwork for future multi-element analysis and the elimination of the monochromator to optimize the optical efficiency of the system. Two Keithley Instruments Model 242 regulated power supplies and provided the operating voltage ranging between 1300 V and 2200 V. The RCA 7265 PMT was gated between the photocathode

and the first dynode to avoid overloading, and possibly permanently damaging the dynode chain by the burst of intense scattered rtiby laser radiation during ablation. The design of the gated system is similar to that proposed by

De Martin (1967). The variabIe delay that operated the gate pulse generator enabled the gated PMT to look at various portions of the plasma emission. By properly synchronizing the gating with respect to the firing of the N2 laser, induced fluorescence can be monitored. The outputs of both photodetecting systems were displayed separatelyon two single-beam oscilloscopes (Tektronix 7704 and 475) .

3.1.6 Ablation Chamber and Its Accessories

The ablation chamber is constructed from a modified four port Corning glass cross, and was sealed from the atmosphere by four specially made

aluminum flanges and end platES cambination moun ted onto the rims of the ports (see Fig.

9).

The ruby laser focussing lens system an~ the receiver opties of the photodetection system were installed onto two of the end plates mounted on the adjacent ports. The vacuum seal between various parts was maintained by O-ring and flange system (see Fig.

9).

The sample mount could be adjusted from the outside by an aluminum rad passing through one of the end plates

(see Fig. 10a). The çhamber could be filled wi th any desired buffer gas through a val ve system installed on one of the end plates. This also allowed the pressure in the chamber to be precisely controlled so that tests with vari ous buffer gases cotUd be carried out during the ini tial phase of the

... investigation.

Two light traps were installed; one attenuated the dye laser beam af ter it passed through the interaction zone while the other reduced the amount of wall scattered dye laser radiation that fell within the field of view of the receiver opties. Owing to the limited space inside the reaction chamber, the latter light trap was an inverted cone design (Fig. 11) and was mounted onto the aluminum end plate facing the receiver optical system. The dye laser beam was attenuated by a crude but relatively effective light-dump made of black feltpositioned on the floor of the chamber intercepting the beam. No pre-caution was made to eliminate the transmission of laser light along the glass walls and the reflection from the reflective surfaces of the aluminum flanges. The scattered dye laser signal was observed to be of the order of 1 ppm Cr signal intensity for flour matrix.

3.1.7 Vacuum System

The ablation chaIDber was evacuated by a CVC PMC 2C diffusion pump equipped with a water cooled baffle valve and backed by a Welch 1402 rotary pump. A liquid nitrogen cold trap situated between the pumps and the ablation chamber effectively prevented any contamination of the system with oil (Balzer-oil 71) from the diffusion pump. The vacuum in the system was measured with a CVC model GPH-100A ionization gauge equipped with a GPH-OOl ion gauge head (25 m to 10-4 J.1.m) .

The pressure of the background gases which were used in some tests were monitored roughly by a CVC Pirani gauge GBllO equipped with a GPOOl head

(1-50 J.1.m, 50-2000 J.1.ffi). With such an arrangement, background pressure over a wide dynamic range could be monitored.

3.2 Sample Preparation

Stock solution I (K2CT.207) and stock solution II [Cr(N03)~J were both prepared from re agent grade salts using 10 mg Cr/l m~ of solut10n. Dilutions

of 10, 100, 1000 were made from each stock solution. 20 m~ of each diluted solution was mixed with 20 g (dry weight) of Carnation skim milk powder to generate samples: SlO, SlOO, SlOOO containing K2Cr207 and SC10, SC100, SC1000 containing Cr(N03)3. Their concentrations are outlined in Table 3. Twenty m~

of diluted stock solution I was also mixed with 20 g (dry weight) flour powder to generate F10, F100, F1000. In addition to the above samples, two samples of

of 20 IIJ.E 100 times diluted stock solutions I· and II were separately mixed with 20 g (dry weight) of Carnation skim rnQlk powder, 200 rog of anhydrous reagent grade copper sulphate (CuS04) ,was added to one, and 200 rog potassium sulfate

(K2S04) was added to the other, generating two additional samples: SlOO.CuS04 (containing 100 times more by weight of CUS04 than its chromium content) and SC100.K2S04 (containing 100 times more by weight of ~S04 than its chromium content) . In all cases; the mixtures were thoroughly blended into paste form. The samples were frozen before undergoing the freeze-drying process to avoid foaming. Af ter the drying process which took about 48 hours, the solid samples were broken down and grounded into a fine powder in an ,aluminia mortar to

ensure homogeneous mixing. Sample powder was then pelleted into small tablets and mounted onto the sample holders and kept dry for later use. In order to ensure the accuracy of this sample preparation, samples SlO, SlOO and SlooO containing 10 ppm, 100'ppm and 1000 ppm, respectively, of chromium were checked by neutron activation technique. The result confirmed the calculated value

as presented in Table 4. This test, however,could not evaluate the homogeneity of the samples.

Steel samples obtained from NBS were also used to test our new technique. These were designated SRM 665, SRM 664, SRM 663, SRM 662 and SRM 661 containing 0.007%,0.06%,.1.31%,0.3% and 0.69% of chromium by weight respectively. Since these standards are designed for microanalysis such as electron and laser

probe analysis, we assumed that the homogeneity of these standard samples was sufficient for our measurements.

3.3 Experimental Procedures

The typical procedure and sequence of events corresponding to the laser ablation and selective excitation of trace element can be described as follows (see Fig. 4):

The samples were mounted onto the target turntable (Fig. lOb) which could be positioned from the outside of the ablation chaIDber. A series' of tests were carried outby rotating the target support without altering the optical geometry. This was crucial in the case of amplitude measurement.

Care was also taken to ensure no cross contamination between samples of extreme concentrations during sample mounting and sampling. Owing to the differences in the thickness of the sample pellets, the position of each sample when in line with the ruby laser axis, was independently adjusted by the horizontal adjustment of the control knob outside the ablation chamber. lts position with respect to the observationwindow was measured.

The alignment of the ruby laser and the target sample was carried out by first aligning the He-Ne laser beam with the ruby laser axis. The location illuminated by the He-Ne laser would be the position sampled by the ruby laser. The ni trogen laser pumped dye lase r was collima ted by a len's near i ts exi t . The dye laser beam was fil tered spatially by a .series of apertures to ensure

that only .the central homogeneous portion of the laser beam was used to probe the ablated material. It was then focussed by alm focal length lens mounted on an adjustable optical stand. The arrangement permitted the direction of

the laser beam to be controlled. The dye laser beam was deflected down into the ablation chaIDber by a 90° prism as shown in Fig. 8.

After aligning the three optical axes, the ablation chaIDher was pumped,

down for 24 hours by the diffusion-rotary pump combination • The long pump-down

time was due to the low conductance of the vacuum tubing used to couple the chamber to the vacuum pump system. The pressure inside the chamber was checked with the vacuum system sealed off from the pwnps to ensure zero pressure

gradient between the chaIDber and the vacuum gauges.

The spectral tuning of the nitrogen pumped dye laser was checked at the beginning, during and at the end of each experimental run. The line width of

the laser was estimated to be less than 0.01 nm. The fine structure sometimes observed in the Fabry-Perot rings was eliminated by a careful adjustment of the beam expander. There was no apparent spectral drift during the experimental run.

The calibration of the ruby laser was carri,ed out prior to each experiment. A fraction of the laser beam was deflected out of the main beam and monitored by either a photodiode connected to a Waveform Digitizer (LeCroy WD2000)

or a Ballistic thermopile (model 100TRG Hadron) connected to an energy meter (TRG model 102 Hadron). The WD2000 allowed the temporal profile of the laser to be displayed by a 20 point array. Since nei ther the laser energy meter nor

the WD2000 could discriminate between a single laser pUlse or a sequence of such pUlses, spurious values of energy for a single laser pulse might result. To avoid this problem, a fraction of the pbotodiode output was sent to a pulse counter where multiple laser pulses would be revealed. Experimental evidence suggested that the chances of having multiple pulses were less than one in fifty within the first 100 shots for a fresh batch of Q-swi tched

solution. The energy displayed by the energy meter, the pulse height and the pulse width revealed by the WD2000 and the number of laser pulses per shot recorded by the counter were used for the energy calibration.

The ruby laser was externally triggered and its beam focussed onto the target surface by a system of lenses. Part of the laser beam used for power monitoring was intercepted by a photodiode in order to initiate a delay

trigger pulse (from the 556 Tektronix oscilloscope) used to trigger the

thyratron of the nitrogen laser. This time delay was adjusted so that the dye laser pulse arrived at the interaction zone at a preselected time relative to the moment of ablation. A fraction of the laser induced fluorescence captured by the receiver optical system was focussed onto the photocathode of the PMT

af ter being spectrally filtered by either a monochromator or interference filters. The output of the PMT was displayed on the screen of a Tektronix 7704 or 475 oscilloscope. The signal was photographed by a camera and re-tained for data analysis. In the case of the fluorescence decay time measure-ments, a Tektronix 4501 calculator with

a

4662 interactive digital plotter was used to digitize and analyse the photo-recorded decay curves.

4. MATRIX EFFECTS AND THE TABLASER SUSCEP:rIBILITY 4.1 Classification of Interference Effects

The spectral line intensity of a given element can be influenced by the chemical and physical nature of the sample. These effects are generally

'_ known as matrix effects • Since the intensity of this line fluorescence is .

of ten used to reflect the analyte concentration, such matrix effects can be very detrimental to the quantification capability of analytical fluorescent techniques.

The kinds of interference (observed in various other fluorescent tech-niques) that might also affect the TAB LASER may be classified into two categories: (a) Chemical interference and (b) physical interference. (a) Chemical interference is that caused by the difference in the chemical states of the analyte in the solid phase or the chemical affinity of the analyte and the concomitants in the vapor phase. These effects include:

1. Source interference,

2. Cation-anion interference,

3. Cation-cation interference,

4. Oxide, hydroxide, carbide formation.

(b) Physical interference is that caused by the physical properties of the sample in both the solid and vapor phases or produced by the physical processes involved in atomization, excitation or emission. These effects include:

1. Spectral line interference,

2. Ionization interference,

3. Self-absorption,

4. Collisional quenching and mixing,

5. Particulate scattering and wall scattering,

6. Differential vaporization,

7. ,Spatial and temporal distribution of analyte in the probing region.

4.2 Chemical Interference

4.2.1 Source Interference (Metalloid Effect)

The fluorescent intensity may dep end on the anion associated with the analyte ion. The major mechanism responsible for variation in signal level could be due to the differences in the dissociation energy of the source compound (Schrenk, 1975; Mavrodineanu, 1965).

4.2.2 Cation-Anion Interference

In this instance the analyte element forms a stable molecule with anion and thus will not be available for the excitation process. Anions report~d to_cause nevere decreases in the observed fluorescent signals are Cl-, S04'

N03' Si04 ' etc. (Czobik et al, 1977; 'Demers et al, 1968; Smeyers-Verbeke, 1977).

4.2.3 Cation-Cation Interference

Mutual interference between cation-cation have been observed in several investiga tions (Gilbert, 1964; Schrenk, 1975). Interference of this type invariably decreases the signal intensi ty of the analyte element. However,

their mechanisms are not well understood (Schrenk, 1975). One possible mechanism may be complex compound formation (Gilbert, 1964).

4 .2 .4 Oxide, Hydroxide and Carbide Formation , Etc.

As the analyte salt or complex molecule.~ dissociate in the atomization process, a variety of species are produced and a variety of reactions can occur among themselves and with the surrounding environment. One of the major reactions is oxide formation (Boumans, 1966). The presence of oxygen induces the formation of stable oxides with free analyte atoms and tends to permanently remove a substantial number of the analyte atoms from selective excitation. Some of the examples are magnesium oxide (Schrenk, 1975), oxides of chromium in flame (Reeves et al, 1973; Haraguchi and Winefordner, 1977). Besides oxide formation, however, hydroxide and carbide formation, etc., have also been known to occur in ~lame atomic fluorescent technique (Kirkbright et al,

1974; Schrenk, 1975). Chemical matrix effects can, however, be suppressed in a high temperature environment (Kirkbright et al, 1974; Boumans, 1966) where compound formation is kept to a minimum.

4.3 Physical Interference

4.3.1 Spectral Line Interference

Spectral line interference may be due to one or a combination of the following:

(1) Line-line interference: this arises when concomitant elements have one or more spectral lines that are too close to the emission line of the analyte to be resolved (Schrenk, 1975).

(2) Line.band interference: in this instance, the spectral line of the analyte is overlapped by one or more spectral bands of molecules and

radicals formed during the recombination process (Sbhrenk, 1975).

(3) Line-background interference: under these circumstances, the spectral line ~f the analyte is overshadowed by the continuous or line background radiation of the plasma medium containing the analyte.

4.3.2 Ionization Interference

The presence of a low ionization element such as sodium, potassium,cesium or rubidium (Na-5.14 eV, K-4.34 eV, Cs-3.89 eV, Rb-4.18 eV) can shift the equilibrium point 'of the ionization process of an analyte possessing a higher ionization potential. This effect can best be seen by considering the Saha equation:

(Boumans, 1966) where n

i

=

ion density,

n e

=

(Üectron dens:ity,

u.

_J.

=

ionization potential, k

=

Boltzmp.nn constant, h

=

Pl-anck con~tant , !..

-m _e mass of e _~

T

=

ele ctrm{ teriiper a ture of the plasma,

,. e

z+/z

-

parti tipn 'rat:î:o for

'

J.on

.

and neutral.

Clearly the increase 'in electron density ne idue to the ionization of any low

lP element{s) decreases the ni/n ratio, thus effectively shifting the equili-brium point to the formàtion of more neutrals. The distribution among'different

states in the neutral is governed by Boltzmann distribution, therefore, the increase in neutral population will invariably increase the ground state popula-tion available for selective excitapopula-tion assuming that the perturbapopula-tion on the temperature due to the presence of low lP element is insignificant such.as in the case of flame (Kornblum et al,

1977).

Ahrens and Taylor

(1954)

suggested thatplasma temperature mayalso be lowered by elements with low lP to the point that the more refractory metals are not vaporized efficiently or that non-metal _may form compound with metallic elements to redUce spectral :Lntehsities of free at.om lines.

4.3.3

Self-Absorption

Self-absorption is one type of spectral int'erference that commonly oc;curs in atomic fluorescent measurements. It is especially noticeable' at high analyte concentration. The photons emitted by theexcited atom ensemble are reabsorbed by adjacent atoms of the same species. In the absence of quenching, all the photons are re-emi.tted and the processes Of absorption and isotropic re-emission may be repeated several times before the photons arrive at the Photodetector. The rate of arrival of these photons will accordingly be reduced at a finite distance from the excited atom ensemble due to the finite lifetime of the exci ted state of the absorbing species. Consequently-, ' the peak fluorescent

intens~ty seen by the photodetecting device will appear to be lowered, thus

giving rise to an impression of lower analyte concentration. Detailed analyses of the effects of spectral line intensity by the imprisonment of radl.ation have been carried out by ~lnes

(1927)

and Holstein

(1951).

Self-absorption effects

the lineari ty of the c'alibration curve when peak measurements are made. This could introduce errors to the analyte quantification if complex corrections are not carried out.

4.3.4

Particulate Scattering and Wall Scatteririg

One type of physical interference in atomic resonance fluorescence is particulate-Mie-scattering caused by solid particulates that resulted from

incomplete vaporization or condensation of refractory oxides ,and carbides (Green,

1976;

Kirkbright,

1974).

During laser ablation, the unvaporized dToplets of molten powdery or fibrous materials may be ejected from the interaction zone.

These can cause severe Mie scat tering and are detrimental to the analysis if they are not eliminated or in some way avoided.

Stray dye laser photons scattered into the receiving optics fram the chamber wall will also be detrimental to the analysis and in particular , the ultimate sensitivity achievable if they are not carefully suppressed. Similar to particulate-Mie-scattering, spectral discrimination is ineffective since the laser emission and the resonance fluorescent lines overlap each other.

4.3.5

Collisional Quenching and Mixing

The presence of foreign molecules such as oxygen, nitrogen, etc., induces compound formation and provides an additional channel to deplete the excited states through quenching and mixing collisions (Piepmeier,

1971;

Svoboda,

1972;

Okabe,

1976;

Green,

1976).

These effects may invariably affect the accuracy of the measuremen t.

4.3.6

Differential Vaporization

In the case of laser ablation, it has been reported by several investigators (Baldwin,

1970, 1973;

Whitehead,

1968,

I shiztika ,

1973;

Morton et al,

1973)

that the amount of material ablated is different for different matrices. This can be

attributed to the difference in the mechanisms of laser target interaction. Ready

(1971)

has revealed some of the mechanisms at some length. Certain

physi-cal characteristics of the sample such as crystal orientation (Kirchheim,

1976),

boiling point, thermal conductivity, thermal capacity, heat of vaporization, play key roles in determining the rate of vaporization (Margoshes,

1973).

Fur-thermore, ene important consideration is the fraction of energy going directly into vaporization compared to that invested in creating particulates. Since

the amount of analyte vaporized is directly related to the amount of host material ablated, differential vaporization may require calibration curves for different types of materials.

4.3.7

Spatial and T~oral Interference

Spatial and temporal inhomogeneities of atom density have been observed in flame fluorescence analysis as a result of chemical, physical and dynamic effects in the probing region. The spatial and temp or al probing locations must be precisely determined to minimize or eliminate these effects (Kirkbright et

al,

1974;

Treytl et al,

1971;

Bratzel et al,

1969).

4.4

The TABLASER Susceptibility to Matrix Effects

In the following, we shall present evidenceto show that the combination of laser ablation and laser selective saturation spectroscopy (TABLASER)

operated in low vacuum

(10-3

to

10-4

torr) enables us to eliminate most of the deleterious effects outlined above. Possible solutions to the remaining

adverse effects will also be discussed.

4.4.1

Evidence Indicating Absence of Various Chemical Interference and

~ Ionization Effects

To evaluate the susceptibility of the TABLASER to source, cation-anion, cation-cation and ionization interference effects, several tests were undertaken.

These included a comparison of the signal responses from several samples with the same concentration of chromium, albeit that the chromium was in two different chemical states, i.e., Cr3+ and Cr2072-. Furthermore, two of the samples were prepared with excess (100:1) amount of copper sulphate or potassium sulphate to produce the four different skim milk powder samples indicated in Table 3. The use of two different chromium salts allows the source interferenc§ effect to be evaluated. Potassium and copper were chosen as added cation contaminants to study the vulnerability of this technique to cation-cation interference. The use of potassium with its ionization potential (K-4.34 eV, Cr-6.76 eV) should reveal the severity, if any, of the change in the sampling efficiency of the refractory metals due to change in the plasma temperature as suggested by Ahrens et al (1954). Sulphate was chosen as the added concomitant anion because of i ts severe cation-anion interference reported widely in the atomic

fluorescent literature" (Kirkbright, 1974; Schrenk, 1975; Smeyers, Verbeke et al,1978). In this way we could determine if the TABLASER calibration was prone to major variation depending upon the chemical camposition of the target material.

Table 5 indicates that there is no significant variation of the induced fluorescent signal (a) when two different chromium compounds were used; (b) when two different chemical"states of chromium were used; and (c) when 1% by weight of an additional compound, such as copper sulfate or potassium sulfate, was added t.6 the sample. The slightly lower signal intensity for SlOO.CuS04 was mainly due to the lower rUby laser energy used in that particular series of measurements .

These results suggest the fOllowitlg three major points: (1) the laser induced fluorescent signal is observed to be independent of the two chemical states or the two chemical forms of the chromium used within the target; (2) the sampling efficiency of the ablation laser appear to be insensitive to the presence of ether elements including those with a significantly lower ionization potential (concentrations higher than 1% by weight were not used because this might alter the physical properties of the substrate) ; (3) the added cation·.and

anion did not appear to induce chemical removal of the atomic chromium or induce spectral interference to the induced fluorescent signal. Furthermore, the

observed immunity of tllls technique from cation-cation interference and carbide formation may be suggested by the linearity of the log/log plot for the NBS-SRM steel standards where the chemical composition varied over two orders of magni-tude for individual impurity elements such as carbon, manganese, silicon and molybdenum (see Table 6) .

Since one of the ways to eliminate chemical matrix effe cts is high tempera-ture (Kirkbright,1974), it is worth estimating the approximate temperatempera-ture of the laser generated plasma. To achieve this, the mean speed of the neutral

chromium atoms was monitored and was estimated to be around 5 x 105 cm s-l

which corresponded" to 6.5 ~ . . Since a sUbstantial amount of the energy invested in the plasma 'was lost in the early phase of expansion through plasma emission, it is reasonable to assume that themean plasma temperature during laser ablation for the heavy species (such as atoms and ions) was somewhat inexcess of 105 K (Appendix 10). At such high temperatures, complete atomization can safely be assumed and molecular association is highly unlikely. Furthermore, the outward

streaming motion of the neutrals and ions (discussed in detail in Chapter 6) in their subsequent expansion phase into the low vacuum region drastically reduces the possibility of any interaction between them, thus minimizing com-pound formation between neutral chromium and the other constituents such as osygen, carbon, hydrogen, etc.

4.4.2 Possible Physical Interference Effects

4.4.2.1 Particulate Scattering and Wall Scattering

One weakness of the ablation process lies in its tendency to contaminate the fan of atamized material with particulate debris. This can take the form of molten droplets, as observed in the case of the steel samples, or small fragments of the surface material for powdery stibstences. In the latter case, an explosive buildup of pressure just beneath the surface of the irradiated target is suggested as the mechanism for spraying this debris into the ablation cloud (Whitehead,1968). The presence of these particulates could severely limit the sensitivity of this approach since Mie scattering could easily

over-shadow the weak laser-induced resonance fluorescence. Fortunately, the consider-able mass difference between the particulates end the atomic species leads to an appreciable velocity differential which creates a space-time window above the target which is free of Mie scattering. The clear separation between the resonance fluorescence from the chromium atoms end the leading edge of the Mie scattering signal from the particulates for two very different substrates is seen in Figs. 15 and 16. Time resolved techniques on atomic absorption end emission have been reported by Gough (1976) and Laqua (1978).

Our ability to discriminate between resonence fluorescence and Mie-scattering and/or stray-dye-laser-light (wall scattering) stems from our use of a short

pulse of probing radiation and from our intimate knowledge of the radiative

lifetime for the transition being pumped. Typical output traces of the resonance fluorescence, with and without Mie and/or wall scattering, are presented in Figs. 12a and 12b respectively. It is quite clear that when Mie and/or wall scattering is appreciable, the pulse width of the observed signal approaches that of the dye laser pulse as seen in Fig. 12c. When the signal arises exclusively from laser induced resonance fluorescence, the radiative lifetime of the excited state appreciably extends the duration of the signal. Further discussion on wall

scattering (stray light scattering) is given in Section 5.2.3 of Chapter 5. 4.4.2.2 Self-Absorption

The effect of self-absorption manifests itself in two forms: (1) en apparent reduction in the slope of the growth curve; (2) an extension of the decay time TI, beyond the natural lifetime T, of the excited state (in the

absence of mixing end quenching collisions). The degree of self-absorption is a function of three parameters: n, the effective density of the absorbing species; L, the absorption length; and rr, the resonance photon absorption cross-section. The relationship between the apparent decay constant T, and the parameters have been worked out by Milnes (1927) end Holstein (1951). Once self-absorption sets in, the calibration curve will become nonlinear end deviate from a straight line (Svoboda et al, 1972; Green, 1976; Weeks et al, 1978). The response

curves (Fig. 13a) are all linear with a 45° slope on the log/log scale for skim milk, flour and NES, SRM steel samples to a concentration of at least 1.3%

chromium. These results suggest that the effect of self-absorption is insigni-ficant at the concentration levels carried out in the current investigation. A more thorough discussion of this subject and its implications will be given in Chapters 6 and 7.

4.4.2.3 Collisional Mixing and Quenching

The effect of collisional mixing/quenching takes two forms: (1) an apparent reduction in the signal intensity, and (2) a shortening of decay time T. The

strong laser irradiance employed in the experiment ensures saturation of the pumped transition. Consequently, the peak signal intensity is independent of de-excitation collisions (Measures, 1968). However, the presence of such collisions would short en the fluorescence decay time~. Table 7 reveals that the mean of the measured decay time constant for both 100 ppm and 1000 ppm of chromium are the same within experimental precision and correspond to the natural radiative lifetime of the monitored 7P2

°

state. Furthermore, mixing to the nearby 7p~o fine structure level has been observed to be less than 1%. These results inaicate that at the time of measurement collisional quenching and mixing were insignificant. A thorough discussion of the implication of these observations will be given in Chapter 7.

4.4.2.4 Spectral Interference

Spectral interference effects were absent in the current investigation for chromium.

Line-line interference may exist in some other elements such as detecting copper in iron where their spectral lines are only 0.026 nm apart (Cu 324.75 nm; Fe, 324.728 nm) or detecting magnesium in iron where thetwo spectral lines

are only 0.001 nm apart

(Mg,

285.212 nm; Fe, 285.213 nm). In these cases, alternative resonance lines should be chosen. An outstanding feature of the TABLASER is its ability to monitor two useful parameters simultaneously, i.e., amplitude of the induced fluorescence and its decay time~. As mentioned in the discussion on self~absorption and collisional de-excitation, the decay time

T monitored is the same as the natural radiative lifetime of the excited state. This. additional parameter can thus be employed to show whether line-line inter-ference poses a problem. By the proper choice of transition, it is quite unlikely to have a coincidence in frequency for two excited states of two different elements in the same sample .

•

Line-band interference can be detected by either detuning the dye laser or using a.J)~ank if it is available. If line-band interference is significant, fluorescent signals should be observed in both cases. In the current investi-gation, fluorescent signals of this kind have not been observed. The 45° slope of the calibration curve clearly indicates the absence of line-line or line-band interference.

Intense plasma emission is invariably associated with the ablation event and could limit the sensitivity of the system as it does in the case of the emission laser microprobe. A maj or advantage of the TABLASER lies in i ts ability to eliminate this source of interference by a judicious choice of temporal delay and spatial probing location. For the present configuration adopted for the investigation, i.e., 1.1 ~s delay, an observation window positioned at 0.55 cm in front of the target surface and a 2 nm band pass, no appreciable p~ emission has been observed.

4.4.2.5 other Constraints of the TABLASER

Operation of the.TABLASER at background pressures greater than 10-2 torr may perhaps impose a problem. With reference to Fig. l3b, a significant plasma emission has been observed. This line-background interference will severely limit the sensitivity of this technique. The operating background pressure should not exceed 10-3 torr. For a higher sensitivity measurement, however,

the operating pressure may have to be even lower to avoid contamination from the environment (Zief and Speight, 1972).

In addition to the above pressure constraint, the duration of the ablation laser pulse plays a crucial role in the analytical capability of the TABLASER. When the Q-switched ruby laser pulse (30 ns FWMH) was replaced by a relatively long pulse (400 ns FWHM) fram a flash-lamp-pumped dye laser, two deleterious effects were observed: (1) significant plasma background emis sion , and (2) the early arrival of the particulate debris introducing strong Mie-scattering. The combined effects do not allowan interference-free space-time window to be found and prematurely limit the detection limit of this technique.

The intensity of the ruby laser currently used was in excess of 109 W cm- 2 .

and was primarily dictated by the plasma production threshold. In fact, an ablation event was not observed if the ruby laser. intensity was reduced by a factor of three or more. This plasma production threshold has been discussed at length by Ready (1973) .

4.4.3 Evidence Indicating Adverse Physical Interference Effects

4.4.3.1 Differential Vaporization Spatial and Temporal Interference Effect The separation of the response curves for the different sample materials is clear evidence of some remaining physical matrix effects. This can be attributed to two mechanisms: the mean streaming velocity of the atoms and

the amount of material vaporized. These characteristics depend to some extent upon the bulk physical properties of the target medium. Fortunately, the

differential spread in velocity between the particulates and the atams is large enough so that the former effect might be eliminated by expanding the volume of excitation to increase the sensitivity (see Chapter 5). It mayalso be possible to allow for differences in the amount of material vaporized by either using an internal s tandard (McDonald, 1977) or standards closely resembling the matrix of the sample to be examined.

5. SIGNAL OPrIMIZATION FOR THE TABLASER AND ITS RANGE OF APPLICATION 5.1 Detection Limit, Precision and Limitation of the Present Facility

A log/log plot of the signal intensity versus the known chromium concen-tration for each of the doped samples is presented in Fig. 13a. The 45° slope of each of these lines is a clear indication of the linearity of response. The error bars reveal the standard deviation over four to six shots for each sample. These observations were recorded, 1.1 ~s af ter the ablation even~ from a small volume situated at 0.55 cm along the ruby laser axis in front of the target

sample. In the case of NBS steel, the detection limit is rather poor due to the poor choice of delay time (see Fig. 15). The detection limit of our

current arrangement is around 1 ppm in the case of flour. As we shall indicate later in this chapter, we feel that an improved design could drive the detection limit to the sub-ppb range.

There are several factors that influence the signal intensity on a shot-to-shot basis. (1) Piepmeier et al (1971) report~d that the amount of material sampled is afunction of the ruby laser energy. The ruby laser output of our

system was, unfortunately, somewhat non-reproducible. The variation in its energy could be as high as

±

30% over a given experimental run. (2) There

was also a jitter of ± 30% in the delay time in firing the thyratron of the N2 laser. This led to a spread in the time of probing the vapor at a given location, which in turn created a spread in the amplitude of resonance fluore-scence since the density in the vapor cloud is a function of time due to its rapid expansion. (3) There may also have been some degree of inhamogeneity

in the distribution of the doped element. Inhomogeneity of the NBS steel samples has been discussed at length by Yakowitz et al (1965) and Nachaelis (1964). This would also contribute to variation in the observed signal. (4) Variation of power density of the dye laser may have contributed to the spread in the observed signal amplitude. However, this should only give rise to a second order effect since the power density of the dye laser was high enough to saturate the atomic transition (Appendix 3). (5) The non-reproducible surface condition of the sample surface may alter the power density of the ablation laser seen by the target

surface at the interaction zone (Figs. l4a, l4b).

The sensitivity of the current experimental facility is limited mainly by (1) the extremely small fraction of the atomized volume excited by the dye

laser, (2) the relatively slow optical system (f/4.5) which collects only .

0.33% of the resonance fluorescence emitted from the excitation region, (3) the poor optical design that allows some laser scattered radiation from the chamber

walls to enter the field of view.

5.2 Possible Imwrovement on Detection Limit

5.2.1 Signal Improvement by Increasing the Probing Volume

For the present measurement, the observed probed volume is less than 3.5 x 10-4 cm3. This volume is defined by the diameter of the probe laser beam (at less than 0.03 cm) (Appendix 4) and the projection of the entrance

slit height of the monochromator into the probed region (0.5 cm). To evaluate the temporal and spatial density distribution of chromium atom, the induced fluorescent signal as a function of delay time is plotted for various spatial locations (Figs. 14 and 16). The fluorescent intensity profiles suggest the fOllowing five points:

(1) Ground state chromium atoms are found not only along the ablation laser axis but also at considerable lateral distances away from this axis without a substantial decrease in density (Fig. 15).

(2) These neutral atoms are streaming into the low vacuum region at very high speed (Fig. 17).

(3) The superposition of the velocity profiles for the chromium at two locations (Fig. 17) indicates that the bulk of the neutral chromium is concentrated in an expanding hemispherical shell.

(4) The streaming velocity of the analyte (Chromium) is different for different matrix materials.

(5) The relative insensitivity of the laser induced fluorescent signal to the angular position of the probing volume relative to the ablation

laser axis suggests that the angular distribution is of the form proposed by Inoue (1971).