Polarization rejection of scattered incident laser light in resonance fluorescence flame atomic detection

oss4-8547j84$3.00+ .oo Pergamon pressLtd.

Specmxhimic~ &a, Vol. 39B,No. 4 pp. 807-812, 1984. &&cd in&cat Britain.

Polarization rejection of scattered incident laser light in resonanee fluorescence flame atomic detection* R. P.

FRVEHOLZ

and J. A.

CELBWACHS

The Aerospace Corporation, Chemistry and Physics Laboratory, P. 0. Box 92957, Los Angeles, CA 90009, U.S.A. (Received 22 September 1983; in revised form 9 December 1983) Abstract--In the detection of trace atomic species by resonance fluorescence, scattering of the incident light may cause a significant interference. Fluorescence detection of flame atomized sodium in the presence of intense scattering resulting from the combustion of al~inium in an air-acetylene flame was investigated. A linearly polarized cw dye laser was used as the excitation source. Due to the high polarization of the scattered light, insertion of a polarizer in the detection optics allowed effective rejection of this interference. Improvements in the signal-tonoise and ba&round ratios of 460 and 200, respectively were observed upon intr~uction of the polarizer. Further improvement was observed employing harmonic saturated spectroscopy. I r INTRODUCTION IN ATOMIC resonance fluorescence spectroscopy, light scattering has been found to be a particularly significant interference [I-S]. This effect is most severe in flames in which nonvolatil~ed components scatter the incident radiation. This scattered light results in reduced signal-to-noise ratios degrading detection limits. Additionally, the background signal resulting from this light may be misinterpreted leading to e~oneously high analyte concentrations. This interference while most often observed using flame atomization may also be present when a graphite furnace is employed [7]. In resonance fluorescence, scattering is particularly annoying because it occurs at the same wavelength as the fluorescence signal. Consequently it cannot be rejected by placing a dispersive element in the detection system. Fluorescence detection has been demonstrated to be a powerful method for trace elemental analysis [I, g-121. BOLSHOV et al.[S] speculate that a major limitation to the ~alyti~l sensitivity of laser-excited atomic fluorescence in flames is the scattering of the incident laser light by nonvolatilized components. While scattering of incident radiation will occur whether laser or conventional excitation is employed, it is demonstrate in this paper that properties inherent to laser radiation allow effective rejection of this interference. Explicitly, we consider scattering as it interferes with the fluorescence detection of flame atomized sodium, excited at the D1 line (589.59 nm) by a cw dye laser. Aluminium is known to be a particularly intense scatterer and is believed to convert to a refractory oxide (Al,O,) upon combustion [3,7]. For this reason, it was chosen to generate the scattering interference in the present experiments. Sodium was chosen as a prototypical atom because it can be conveniently excited by our cw dye laser. With other commercial systems, tunable cw laser

*This work was supported by Office of Basic Energy Sciences, Department of Energy. [l] V. SYCHRA,V. SVOBODAand I. RUEESKA,Atomic Fluorescence Spectroscopy, p. 176. Van Nostrand-Reinhold, New York (1975). [2] N. OMENEXTO, L. P. HART and J. D. WINEFURD~ER, Appl. Spectrosc. 26,612 (1972). [3] P. 0. LARKINSand J. B. WILLIS, Spectrochim. Acta 29B, 319 (1974). [4] J. P. S. HAARSMA, J. VLOGTMANand J. AGTERDENBOS,Spectrochim. Acta 31B, 129 (1976). [5] K. J. D~OLAN and L. E. SM~THE, ~~ctr~~~. Acta 32B, 115 (1977). [6] D. A. GOFF and E. S. YEUNG, Artal. Chem. SO, 625 (1978). [7] K. J. D~OLAN and L. E. SMYTHE,Spectrochim. Acto 34B, 187 (1979). [8] M. A. BOLSHOY, A. V. ZYBIN and I. I. SMIRENKINA,Spectrochim. Acta J6B, 1143 (1981). [9] G. F. KIRKBRIGHTand M. SARGENT, Atomic Absorption and Fluorescence Spectroscopy. Academic Press, London (1974). [lo] J. A. GELBWACHS,C. F. KLEIN and J. E. WESSEL, Appi. Phys. Lett. 30,489 (1977). [ll] J. A. GELBWACHS,C. F. KLEIN and I. E. WESSEL, IEEE J. Quantum Electron. QE-14, 121 (1978). [12] S. J. WEEKS, H. HARAGUCHIand J. D. WINEFORDNER,And. Chem. 50,360 (1978). 807

808

R. P. FRUEHOLZ and J. A. GELBWACHS

radiation is available from approximately 420 to 900 nm. Additionally, recent advances in ring dye laser technology have provided useful cw output power in the 3OOnm region by internal second harmonic generation and frequency mixing. This wide range of wavelengths should allow application of the present results to most atoms of analytical interest. Due to the highly polarized nature of the incident laser light and the strict polarization requirements of Mie and Rayleigh scattering, introduction of a polar~ing element in the fluorescence detection optics is shown to effectively reject the scattering. 2.

DESCRIPTIONOF RAYLEIGH AND MIE ~CA~ER~NG POLARIZATION REQUIREMENTS

The quantitative theory of light scattering from small particles is complicated [13-161. Rayleigh scattering occurs for particles much smaller than the wavelength of the incident radiation while Mie scattering ocours for particIes whose dimensions are of the same order or larger than the wavelength. The rigorous theory of Mie scattering treats scattering from spheres. For particles of irregular shapes, as might occur in a flame, some deviation from the Mie results is expected. The degree of this deviation depends, in a complex manner, on the size and shape of the scattering particles [I?--201. Since the present ex~rimental results are consistent with predictions based on the simple theory of scattering from spherical particles, certain results of the Mie analysis are presented. For incident light that is plane polarized and for detection in the planes parallel or ~~ndicular to that plane, both Rayleigh and Mie scattering satisfy simple and rigorous polarization requirements [13-161. Consider the incident light propagating along the z axis with its polarization vector along the x axis. In the xz plane both Rayleigh scattering and Mie scattering are linearly polarized with polarization vectors laying in the xz plane perpendicular to the propagation direction of the scattered light. As an example, assume the detector views along the x axis. The scattered light’s polarization vector lies along the z axis. In the yz plane, both Rayleigh and Mie scattering are linearly polarized in the x direction. It is clear that when the detector is in either the xz or yz planes, insertion of an approp~ately oriented linear polarizer in the detection optics should significantly reduce the scattered light detected. Since resonance fluorescence is not strongly polarized [2,3], the observed fluorescence to scattering ratio should be increased. The use of pol~i~tion to reject scattering has been recognized by other authors [2,3,21]. Previous experimental studies have used unpolarized sources which require polarization of the exciting radiation prior to the sample. OMENETTO et al. [2] used polarization to minimize scattering in the analysis of several different metals. They found increases in the signal-tobackground ratios from 3 to 16 when compared to unpolarized studies. LARKINS and WILLIS [33 performed similar experiments yielding factors of improvement in signal-to-background ratios from 2 to 7 upon insertion of polarizers. DAILY [21] investigated theoretically the rejection of scattering by polarization. His work considered a laser excitation source which was linearly polarized, thus eliminating the need for the first polarizer. These calculations indicate that introduction of a linear polarizer in the detection optics should reduce the detected scattered light by approximately a factor of 1000. This improvement factor is specified primarily by the degree of the laser polar~tion and the rejection ratio of the detection polarizer. A factor not considered by Daily is that typically the incident laser beam is focused to some extent by insertion of a lens. With the convergence of light ray paths associated with the focusing, the polarization vectors of all the light rays will no longer be

[13] [14] [lS] [16] [ 173 [18] [19] [20] [21]

D. SINCLAIR and V. K. LAMER,Chem.Rev. 4d, 245 (1949). H. C. VANDEHULST,Light Scatteringby Small Particles. Wiley, New York (1957). M. KERKER The Scattering of Light and Other Electrmwgnetic Radiation. AcademicPress,New York (1969). M. BORN and E. WOLF, Principles ofOptics,pp. 549, 630-661. PergamonPress, London (1970). G. N. PUSS, Appl. Optics 3,867 (1961). W. L. PARTAIN, J. R. WILLIAMS and J. D. CLEMENT,NASA Spec. Publ., NASA-SP236,243 (1971). R. G. PINNICK,D. E. CARROLL and D. J. HOFMANN,Appl. Optics IS, 384 (1976). I. KIRMACIand G. WARD, Appl. Optics 18, 3328 (1979). J. W. DAILY,Appl. Optics 17, 1610 (1978).

809

Scattering of the incident light

parallel [22]. This reduction of the degree of polarization of the laser beam could reduce the scattered light rejection effectiveness. 3. EXPERIMENTAL The experimental apparatus is shown in Fig. 1. A tunable cw dye laser (Spectra Physics # 375) was pumped by an argon ion laser (Coherent CW-12) operating at 5 W ah-lines. The dye laser, operating with rhodamine 6G dye, produced an output power of 0.3 W with a linewidth of IOGHz (1.2 x lo-’ run). The dye laser output was found to be horizontally polarized to a degree [16] better than 0.998 as determined by a Glan-Thompson polarizer. The laser radiation was passed through a X6 beam expander and then mechanically chopped at 400 Hz by a slotted wheel with 5-mm slots. The resulting light was approximately sinusoidally modulated with its power varying between 2 and 98 % of the power prior to chopping. A 250-mm focal length lens focused the beam into an air-acetylene flame. In the focal region, the laser n-radiance was approximately 1 kW cm- 2. At this &radiance, the resonance fluorescence was partially saturated, i.e. the fluorescence to laser power ratio became sublinear. Solutions to be studied were aspirated into the nebulizer and premix burner assembly (Perkin-Elmer 303-0352 and 303-0191, respectively). A three-slot burner head (Perkin-Elmer 303-0401) was used. The detector was placed in the plane defined by the direction of propagation of the laser beam and its horizontal polarization. In this ~o~guration, Rayleigh scattering is zero and the polarization vector of Mie scattered light lies in the horizontal plane. In order to test how effectively polarization can be used to reject scattered light in trace analysis, it is desirable to have a large ratio of scattered to fluorescent signal prior to the introduction of the polarizer. This can be accomplished by either reducing the fluorescent species concentration or increasing the scattered signal. In the present case, the species emitting resonance fluorescence is sodium which is a contaminant in the material chosen to yield the scattering interference, AlCl,. To reduce the sodium signal, the reagent grade AlCl, was further purified via vacuum sublimation. This resulted in the aspirated aqueous solution having a 0.5-ppm sodium concentration with a 1% AlCl, concentration. The scattered light intensity can be increased by using the fact that Mie scattering is “forward peaked” with the highest scattering in the direction of laser propagation. To make use of this result, the detector was placed at an angle of 12” from the laser’s propagation direction. If the sodium concentration could have been easily reduced to the low ppb range as is typical for trace analysis, this configuration would not have been needed. The direct laser light is readily rejected through the use of appropriate apertures. The laser induced fluorescence and scattered light from the focal region were imaged by a l%mm focal length lens onto a photom~tiplier tube (PMT) (EMR9558B) after passing through a linear polarizer (whose axis of polarization was vertical), two absorption filters (Schott KG1 and 2), and a 0.8nm band pass interference filter with a 30% maximum transmission at its central wavelength of 589.3 nm. The polarizer was a sheet of dichroic plastic (Edmund P-60637). The output of the PMT was analysed by a lock-in amplifier (fthaco 391A) synchronized to the chopping frequency. A 4.0-s output time constant was used while the laser frequency was scanned at 0.01 nm min- *.

BEAM EXPANDER

&.I ‘7”n

CHOPPER

Fig. 1. Schematic drawing of ex~r~en~l [22] L. W.

CASPERSON

and C. YEH, Appt. Optics 17, 1637 (1978).

”n

apparatus.

810

R. P. FRUEHOLZand J. A. GELBWACH~

4. Rnsu~rs ANDDISCUSSION 4.1. Ex~~ri~nt~i results In the present experiments, the dye laser wavelength was scanned slowly across the sodium D, resonance while the Ah& solution was aspirated into the flame. Figure 2a shows the detected signal with no linear polarizer present in the detection optics. The scattered signal is so intense that the sodium resonance fluorescence is undetectable. When aspiration stopped the scattered light dropped to a zero baseline value, The result of scanning the spectral region a second time with the linear polarizer present is shown in Fig. 2b. The sodium signal is now clearly visible on top of a small background. To quantify the improvement between Figs 2a and b, we use the standard signal-to-background (S/R) and signal-to-noise (S/N) ratios [23]. The signal magnitude in Fig. 2b is readily obtained. From this signal magnitude and the measured transmission of sodium D1 ftuorescence through the polarizer, it is possible to estimate the signal magnitude present in Fig. 2a. The scattering background magnitude is the difference between the average signal with and without aspiration of the AUS solution occurring and the laser tuned-off

6

4

2

0 -2

-2

-2

0 -4 --4 -4 -4 -4 0 F 56952 54 66 36 60 62 64 66 WAVELENGJH

inml

Fig. 2. Laser excited resonance fiuorescence of sodium from a game atomized aqueous soIution of 1% AQ and residual sodium (approx. 0.5 ppm). (a) Direct observation of fluorescenceand scattering, (b) detected signal with linear polarizer in detection optics, and (c) fluorescence signal from harmonic saturated spectroscopy in conjunction with linear polarizer in detection optics.

1233S. &VA and A. LONGONI,An introduction to signals, noise, and measurements, in AnolytiealLaser Spectroscopy, Ed. N.

OMENETI-O,

p.

418.

Wiley, New York (1979).

811

Scatteringof the incidentlight

resonance. Finally the noise amplitude is the r.m.s. deviation of the scattered light intensity from the background levels. For Figs 2a and b the magnitudes of the various quantities are summarized in Table 1. Insertion of the polarizer is seen to increase the S/N ratio by a factor of about 460 while increasing the S/B ratio by nearly 215. Clearly using polarization can be an extremely effective means of reducing the scattering interference. The signal observed in Fig. 2b still rests on a scattering background. A further reduction in background can be achieved by coupling harmonic saturated spectroscopy to polarization rejection. Harmonic saturated spectroscopy, which makes use of the nonlinear response of saturated fluorescence, has been found to effectively reject background [24]. A brief description of this method is now given. For low laser intensities, the fluorescence is not saturated and it responds linearly to modulation in the incident radiation. At high laser intensities, when partial saturation is present, fluorescence no longer responds linearly, distorting the incident modulation function. However, scattering will respond linearly even at high laser intensities. In the present experiments, the laser beam is sinusoidally modulated at a frequency f, by the mechanical chopper. The saturated fluorescence, due to its nonlinear response, contains signal at harmonics off0 with the most intensity at 2fo. Phase sensitive detection at 2f;t allows detection of fluorescence with a sibilant r~uction of background scattering. The results of polarization rejection coupled to harmonic saturated spectroscopy with detection at 2fo are shown in Fig. 2c, while pertinent data are tabulated in Table 1. The S/B ratio is improved over simple polarization rejection by a factor of 14 yielding a total improvement factor over the raw signal in Fig. 2a of 2950. At the relatively low 2fo signal levels, instrumental and flame noise not related to laser scattering are quite significant. It is possible to determine the magnitude of the noise due to sources other than laser scattering by studying the detected signal while no AlCl, solution is aspirated into the flame. Comparison of the noise level in this region to that when aspiration is occurring allows identification of noise due purely to laser scattering. This noise level is presented in Table 1 and is somewhat better than that obtained by detecting at&. The total improvement of the Sm ratio between detection at 2fo and the raw signal is approximately a factor of 500. It is seen that use of polarization along with harmonic saturated spectroscopy to reject scattered laser light can result in significant improvements in both S,% and Sm ratios. 4.2. Comparison of present reds to previous studies Polar~tion was much more effective in rejecting the’scattering interference in the present studies than in previous experiments [2,3]. It is of value to understand why this improvement was obtained. The possibility exists that the three-slot burner used in our work might be more efficient in producing a uniform and smaller droplet size distribution than the burners used in the prior studies. This could account for a closer agreement of experimental data with the predictions given by scattering theory. Certainly, though, a major factor is the use of a laser as the resonance fluorescence excitation source. Two characteristics of the laser radiation allow Table 1. Summaryof experimentalresults

Experimentalconfiguration

Fluorescence Scattering r.m.s.scattering signal background noise (mV) (mv) (mV)

No polarizationrejection Simplepoiarizationrejection Harmonic saturated spectroscopy with polarizationrejection

1.6” 0.42

73 8.8x 1o-z

1.2 6.9x 1o-4

0.022t

3.4 x ;o-4

3.3 x 1o-5

S/B ratio

S/N ratio

0.022 4.7

600

65

1.3

670

*Signal not visible on spectrum but can be estimated from polarization rejection data and the known fluorescence transmission of the polarizer. t More sensitive second harmonic detection is possible if the experimental apparatus is fully optimized for this technique as was done in Ref. [24]. [24] R. P.

FRUEHOU

and

1. A. GELBWACHS, Appi.

0ptics

19,273s (1980).

812

R. P. FRUEHOLZ and J. A. GELBWACHS

polarization rejection of scattering to be effective. First, cw dye laser radiation is generally linearly polarized to a high degree. In the case of incoherent sources as were used by OMENETTOet al. [Z] and LARKINSand WILLIS [33, linear polarizers must be used to polarize the incident radiation. This immediately reduces the excitation intensity by a factor of two. Polarizers employed in these studies had crossed transmissions of 0.3 % at 535 nm. At this wavelength, both groups were detecting thallium fluorescence for which their highest S/B ratio improvements, factors of 16 [2] and 7 [3], were obtained. The factor of 0.3 % cross transmission implies that after the incident beam has passed through the polarizer, the degree of linear polarization [16] will be approximately 0.98. For comparison the degree of linear polarization of the dye laser used in the present studies was 0.998. The unpolarized portion of the incident light results in scattering that will not be effectively rejected by the insertion of a polarizer in the detection optics. Of course additional polarizers could be used prior to the excitation region, for example, OMENE~O et al. [Z] used two. The observed difference in scattering rejection efficiency between our laser studies and the previous studies cannot be explained entirely on the basis of polarization purity of the exciting radiation. Another factor which allows a higher rejection of scattering when using a laser excitation source is the well collimated nature of the beam. This collimation shows the use of high~number optics in the focusing of the laser beam into the sample. Thefnumber for the present experiment was approximately 50 yielding a maximum convergence or ray angle of 0.6”. In contrast, LARKINSand WILLIS [S] used optics with anfnumber of 2. This more sharply focused light source has a maximum ray angle of 14”. Once rays leave the plane, defined by the incident beam and the line from the focal point to the detector, the polarization requirements for rigorous scattering rejection are no longer met. It is this factor which most severely limits the effectiveness of polarization rejection when using a non-laser excitation soure. Taken together the two effects discussed explain why polari~tion rejection of scattering is so much more effective when a laser excitation source is employed rather than a conventional source. 5. CONCLUSIONS A signi~~nt reduction in the scattering interference to atomic resonance fluorescence detection can be obtained through the use of a linearly polarized cw dye laser for excitation and perpendicularly oriented polarizers in the detection optics. Use of a laser excitation source allows much greater scattering rejection than an incoherent source due to the inherently high polarization and spatial collimation of the laser beam. Harmonic saturated spectroscopy when used in conjunction with polarization discrimination further improves scattering rejection.