Optical enhancements and applications of rapid atomic emission spectrometry acousto-optic deflector background correction

Abstract

A glass acousto-optic deflector (AOD) is mounted in a 0.5 m spectrometer for atomic emission background correction. The system allows sequential and rapid observation of adjacent wavelengths for "on" and "off" atomic line background correction. Particular attention is directed to optical masking and data acquisition. While AOD components slightly degrade spectroscopic resolution, proper optical masking minimizes this problem and enhances AOD diffraction efficiency. Conversely, reduction of light throughput by masking forces a compromise in minimal useful aperture diameters. Representative atomic spectra with a neon hollow cathode lamp are presented. Utilizing appropriate AOD masking arrangements produced an increase in observed neon line widths to 0.14 nm from the 0.09 nm widths observed in the absence of the AOD. While the acousto-optic has the capability to alternate observed wavelengths in the 100 kHz range, the 3000 Hz upper frequency limit is a function of the current computer manipulation software. The best signal-to-noise ratios were obtained at higher sampling rates. Background correction was performed with a microwave-induced plasma atomic emission source. Spectral characterizations were obtained with the 425 nm scandium emission line, the sodium doublet near 589 nm and the chromium triplet in the 427 nm region. Calibration plots were obtained with solution nebulization and electrothermal vaporization. Time–detector response profiles are illustrated for electrothermally vaporized analytes.

---

Introduction

Atomic emission background correction techniques with high efficiency single unit detectors such as photomultiplier tubes have produced impressive results.^1–11 Historically, techniques have often included the movement of devices such as refractor plates, choppers, gratings, etc. However, these moving device-based approaches have inertia-based limitations. Array-type detectors offer an alternative background correction approach. However, for higher resolution, more complex echelle spectrometers are generally required with these systems. Other factors which limit the use of some background correction techniques include cost and overall complexity. The application of acousto-optic devices offers a solid-state alternative for background correction.

The theory behind acousto-optic deflectors (AOD) and modulators has been presented.^12–16 An initial introduction to AOD background correction for atomic emission spectrometry is detailed in ref. 17. Briefly, optical radiation traverses a transparent acousto-optic material. A piezo-electric transducer is utilized to direct a near-acoustic velocity, high-frequency compression wave through the material at an angle often nearly perpendicular to the direction of the optical radiation. The acoustic waves yield periodic variations in the index of refraction of the material. The optical radiation undergoes diffraction; the angle is dependent upon the acousto-optic driver frequency. Mounting the AOD in the light path of an atomic emission spectrometer allows AO diffraction of analyte line emission on and off the exit slit.^17,18 With this configuration, the potential exists to monitor atomic spectrometric background and analytical signals in a rapid fashion by varying the applied frequency.

Background correction based upon the use of an AOD has shown promising preliminary results.^17,18 In the initial publication on this topic,¹⁷ an AOD was mounted in a 0.75 m focal length polychromator and the potential for background scanning was characterized with the 486.1 nm atomic line of a hydrogen pen lamp. The background correction proof-of-concept was verified by slowly scanning the spectral region of the atom line and "stepping" from the line to an adjacent background wavelength.

This manuscript presents the characterization of a 0.5 m focal length spectrometer based acousto-optic background correction system. The system is demonstrated with a hollow cathode lamp and microwave induced plasma (MIP) atomic emission sources. Effective line-width and system efficiency are measured using neon line emission from a hollow cathode lamp. The system is demonstrated for the determination of sodium, scandium and chromium with solution nebulization and electrothermal vaporization. Background correction modulation rates of up to 3000 Hz are applied.

Experimental

Spectrometer and AOD system

An acousto-optic deflector (Model ADM-40, IntraAction Corporation, Inc., Bellwood, IL, USA) was mounted in the light path of a 0.5 m Czerny–Turner scanning monochromator (Model 82-516, Jarrell Ash, Co., Waltham, MA, USA). The spectrometer utilized a 1200 groove per mm grating and a 30 µm exit slit. The reciprocal linear dispersion was 1.67 nm per mm and the calculated geometric spectral bandpass was 0.05 nm. A schematic showing the placement of the AOD can be seen in Fig. 1. The AOD was positioned 2.0 cm from the monochromator entrance slit and mounted so that it could be rotated using a precision gear drive mechanism. Light entered the monochromator, traversed the AOD, was separated by wavelength-based angular dispersion of the grating and was focused upon the focal plane. The dotted line represents the path of analyte line emission when the AOD frequency is adjusted so that the analyte signal is observed as first-order AOD diffracted line emission. By changing the driver frequency, the position of the radiation along the focal plane may be varied. The solid line represents the path of analyte line emission when the AOD frequency is adjusted to view the spectral background. The line emission wavelength of interest strikes the focal plane, but not the exit slit. The radiation striking the exit slit corresponds to that of a wavelength suitable for measurement of the spectroscopic background.

Fig. 1 Schematic of the 0.5 m scanning monochromator AOD background correction system.

The AOD utilized in this study is constructed of dense flint glass, with a manufacturer installed 2 × 20 mm optical aperture, a rise time of 170 ns with a 1 mm aperture and a manufacturer-specified wavelength range from 440 to 700 nm. The aperture was initially modified to allow a maximum window of 3.5 by 20 mm. Interchangeable circular optical masks with diameters of 0.8, 1.3, 1.7, 2.0, 2.5 and 3.1 mm were mounted on the AOD. The AOD driver (Model DE-40 VCO, IntraAction Corporation, Inc., Bellwood, IL, USA) has a center frequency of 40 MHz, a range of 27 to 54 MHz and a maximum sweep scan rate of greater than 20 kHz.

The 1P28A PMT (RCA) current output was converted to a voltage, amplified and filtered using an electronics system designed and constructed at Northern Illinois University. The amplifier output was directed to an SCB-68 connector and an AT-MIO-16XE-10 data acquisition board (National Instruments, Austin, TX, USA) installed in a 200 MHz Pentium Pro (Dell, Austin, TX, USA) computer. Software to acquire the PMT signal versus time and save it as an ASCII text file for further analysis was written using LabView (National Instruments) software.¹⁸

Radiation sources

For preliminary characterizations, a zinc hollow cathode lamp with neon fill gas was used as a radiation source. This lamp was powered by a hollow cathode lamp warm-up system (Perkin-Elmer, Corp., Norwalk, CT, USA).

The microwave-induced plasma (MIP) system consisted of a 500 W, 2.45 GHz microwave generator (Model 420B, MicroNow, Skokie, IL, USA) and a copper Beenakker TM₀₁₀ cavity similar to that described by Haas et al.¹⁹ and Urh et al.²⁰ The 12 L min⁻¹ helium plasma gas flow was directed through a tangential flow, Teflon insert-based quartz torch (9 mm id and 11 mm od) placed in the center of the cavity. The plasma was operated with a forward power of 350 to 450 W and minimal (≈0 to 10 W) reflected power. Samples were delivered to the central channel of the torch with a helium flow of 1.0 L min⁻¹. Solutions were introduced into the MIP using an ultrasonic nebulizer (Model U5000, Cetac, Omaha, NE, USA) with a liquid sample flow rate of 400 µl min⁻¹ delivered by a Rainin Rabbit peristaltic pump. Electrothermal vaporization was performed utilizing an apparatus similar to that described by Wu and Carnahan.²¹ Sample volumes of 10 µL were delivered to the carbon cup with a micropipet. A one-to-one image of the plasma was focused with a 5 cm diameter, 25 cm focal length borosilicate lens upon the entrance slit of the spectrometer.

Results and discussion

Optical masking

To examine the effects of the AOD in the monochromator optical path, neon emission lines at 585.2 and 588.2 nm were examined with and without the AOD in place. Fig. 2(A) shows the neon spectrum with no AOD in the monochromator. The 585.2 nm peak width measured at 50% of the peak maximum intensity is 0.09 nm.

Fig. 2 Observed spectral effects of the insertion of the AOD with a 0.35 × 2 mm optical mask: (A) neon emission spectrum with no AOD in the monochromator, (B) with the AOD off and the 0.35 × 2 mm aperture installed and (C) with the AOD on and the 0.35 × 2 mm aperture installed.

Utilizing the 3.5 × 20 mm aperture without a circular mask, the AOD was mounted in the spectrometer. With the AOD driver operated at a frequency of 33.7 MHz and a power of 4.0 W, the device was rotated until the AOD-diffracted first-order neon line intensities were maximized.

The spectrum in Fig. 2(B) was taken with the AOD driver off. Peak widths increased to approximately 2 nm. These peaks are characterized by "sloping shoulders" to the high-wavelength sides. The spectrum of Fig. 2(C), taken with the rf directed to the acousto-optic, shows AO diffraction of the 585.2 nm radiation at the monochromator-calibrated position of 584.5 nm. There is also AO diffraction superimposed on the zero-order emission "shoulder" at the calibrated position of 586.3 nm. Similar but less well defined features are seen for the 588.2 nm Ne line. The peak shoulders of Fig. 2(B) and (C) made observing the AO phenomena difficult. It is likely that these large 2 nm peak widths are due to reflections within the AOD.

To reduce potential light reflections and test whether this was a factor, the 2.5 mm diameter optical mask was placed on the aperture of the AOD. Fig. 3 illustrates postulated masking effects upon internal AO reflections. Fig. 3(a) shows a front view of the AOD with the 3.5 × 20 mm aperture. Light passing through the entrance slit and traversing the AOD with this large aperture is illustrated in Fig. 3(b). This cartoon illustrates that reflections from the sides within the AOD may give rise to redirection of radiation within the spectrometer. It is likely that this phenomenon yields the peak shoulders observed in Fig. 2(B) and (C). With the smaller circular aperture [Fig. 3(c)] the potential for reflected radiation is substantially decreased. These phenomena are illustrated in Fig. 3(d). Additionally, with the arrangement illustrated in Fig. 3(d), increased AOD diffraction efficiency should be observed because the optical radiation is more nearly collimated and a smaller acceptance angle is required than is the case without the mask. This increased efficiency may come with a sacrifice of light throughput.

Fig. 3 Postulated effects of smaller diameter masks: (a) the 0.35 × 2 mm aperture, (b) light traversing the AOD with the 0.35 × 2 mm aperture, (c) the 0.25 mm circular aperture and (d) light traversing the 0.25 mm aperture.

With the 2.5 mm diameter optical mask in place, neon spectra were obtained. For comparison purposes, Fig. 4(A) shows the neon spectrum with the AOD removed from the spectrometer. Fig. 4(B) shows the neon spectrum with the installed AOD and the driver off. Fig. 4(C) shows the spectrum with the driver on. With the AOD in place, the 584.2 nm peak widths increase slightly, with widths at half maximum of 0.13 to 0.14 nm compared to 0.09 nm with the AOD absent. With this exception, spectra are identical with the AOD in place, but turned off, and without the AOD in place. With the AOD on, first order diffraction of the 585.2 nm line can be seen at the 586.3 nm location and −1 order diffraction at the position of 584.5 nm. First order diffraction of the 588.2 nm line can be seen at the 589.2 nm position and −1 order diffraction can be seen at the 587.2 nm position.

Fig. 4 Observed spectral effects of the insertion of the AOD with a 2.5 mm circular aperture: (A) neon emission spectrum with no AOD in the monochromator, (B) with the AOD off and the 2.5 mm aperture installed and (C) with the AOD on and the 2.5 mm aperture installed.

Characterization was undertaken with masks of varying diameters. These studies examined trade-offs between the effects of maximizing light throughput versus minimizing reflection and light outside the angular acceptance aperture. With the AOD operated at 33.7 MHz and 4.0 W, spectra were acquired with the 0.8, 1.3, 1.7, 2.0, 2.5 and 3.1 mm diameter apertures. These data are shown in Figs. 5, 6 and 7.

Fig. 5 588.2 nm Ne peak widths as a function of aperture diameters. Solid rectangles, widths measured at one-half maximum intensity of AOD diffracted first order radiation. Crossed rectangles, widths measured at one-fifth maximum intensity of non-diffracted zero order radiation.

Fig. 6 588.2 nm Ne first order diffracted peak intensities as a function of aperture size.

Fig. 7 588.2 nm Ne first order diffraction efficiency as a function of aperture size.

As shown in Fig. 5, the 585.2 nm Ne line peak widths of first order diffracted and zero-order non-diffracted AOD radiation were monitored as a function of aperture diameter. The first-order AOD peak widths at half maximum have relatively constant values ranging from 0.15 to 0.20 nm. however, major differences in the widths of the zero-order peaks are observed. These differences were the most profound at the peak base. To fully describe this behavior in terms of the effects on the spectral baseline and background correction, the widths at 20% of the peak height were determined. These "fifth widths" increased from just less than 0.5 nm with the 0.8 mm mask to more than 1 nm with the 3.1 mm mask. For the 3.1 mm mask, the zero-order radiation overlapped the −1 order diffraction at the 584.3 nm location.

Other critical factors affected by aperture size include signal intensity and diffraction efficiency. A plot of first-order diffraction signal intensity versus mask diameter is shown in Fig. 6. Signal intensity is relatively constant for masks with diameters greater than 1.7 mm. There is a decrease in signal intensity for the smaller masks, indicating that these masks reduce the amount of radiation undergoing AOD diffraction. A plot of the diffraction efficiency versus mask diameter can be seen in Fig. 7. The diffraction efficiency represents the height of the diffracted peak divided by the height of the zero-order peak with the AOD "off." This plot shows a general increase in the diffraction efficiency with decreases in mask diameter. The diffraction efficiency increases due to the elimination of radiation at angles greater than the AOD acceptance angle. The range of efficiencies (25 to 60%) are more than acceptable for most atomic emission spectroscopy experiments.

For further studies presented in this manuscript, the 1.3 mm diameter mask was chosen. This diameter was chosen based upon the compromises made in order to minimize zero-order peak widths, while maximizing light throughput and AOD diffraction efficiencies.

Applied AOD power

Using the 1.3 mm aperture, power–frequency–intensity characterization of the 585.2 nm neon line was accomplished by obtaining a series of spectral scans. The driver frequency was varied in the range from 28 to 44 MHz in 2 MHz increments. The driver power was incremented in the range of 1.6 to 4.0 W in 0.8 W steps. First-order peak area was measured and plotted versus driver frequency in Fig. 8. The general shapes of these plots indicate maximum areas at 34 MHz. There is a general increase in the peak area with an increase in driver power. The maximum peak area is seen with the 4.0 W driver power. With the present device, this is the ideal operating condition for the AOD using the 585.2 nm neon emission line.

Fig. 8 588.2 nm Ne first order diffracted relative peak area as a function of driver power and frequency.

Plasma atomic emission AOD spectra

The MIP was utilized to examine AOD background correction with an atomic emission source. Separate solutions containing 100 ppm Na, 1000 ppm Sc and 1000 ppm Cr were nebulized into the MIP. For each analyte, the AOD was rotated for maximum first-order diffracted radiation intensity. With the 1.3 mm mask, spectra were recorded. Again, the AOD driver frequency was increased incrementally by 2 MHz from 28 to 44 MHz and the AOD driver power was increased in 0.8 W steps from 1.6 to 4.0 W. Spectra were recorded for each increment and the optimum conditions (4.0 W and 34 MHz) observed with neon were reconfirmed.

Intense sodium emission lines are seen at 588.9 nm and 589.5 nm with 100.0 ppm sodium. Fig. 9(A) and (B) show sodium spectra with the AOD off and on, respectively. With the AOD on, first-order diffracted 589.5 nm radiation is evident at the 590.7 nm monochromator position and –1 order diffraction is seen at 588.2 nm. For the 588.9 nm Na line, the spectral region of the scan shows only the +1 order diffraction at 590.1 nm. The AOD was rotated such that the diffraction was efficient for the 420 to 430 nm spectral region of chromium and scandium emission. Results are seen in Fig. 10 and 11. First and second order AOD diffraction are seen for the 424.6 nm scandium emission line at 425.8 and 427.0 nm, respectively. While some overlap of –1 order AOD diffraction is seen in the chromium spectrum, dominant first order diffraction is seen at wavelengths 0.8 nm greater than the 425.4, 427.4 and 429.5 nm emission lines.

Fig. 9 Spectrum of 100 ppm sodium nebulized into the MIP with the AOD (A) off and (B) on.

Fig. 10 Spectrum of 1000 ppm scandium nebulized into the MIP with the AOD (A) off and (B) on.

Fig. 11 Spectrum of 1000 ppm chromium nebulized into the MIP with the AOD (A) off and (B) on.

For all three sets of spectra, first order diffraction efficiencies were in the 25% range. The positive results obtained for scandium, chromium and sodium indicate that, for this particular glass, AOD is useful for background correction in the spectral range of at least 420 to 590 nm.

Wavelength modulation optimization

The system was used for background correction with the MIP radiation source to characterize PMT signal–frequency relationships. First-order diffracted radiation of the 588.9 nm sodium emission line was directed to the exit slit of the monochromator with the AOD operating at a driver frequency of 33.7 MHz and a driver power of 4.0 W. The "off-line" spectral background was observed with the driver frequency set at 43.7 MHz. A 0.5 V peak-to-peak square wave was directed to the AOD driver controller to modulate the driver frequency between these frequencies. The tested modulation rates were 2, 10, 100, 1000, 2000 and 3000 Hz for the analysis of a 0.25 ppm sodium solution.

General trends indicated that, as expected with a 1/f dominated source such as the MIP,¹⁸ higher modulation rates yielded lower background noise and higher signal-to-noise ratios. Software and hardware limitations restricted the maximum background correction sampling rates to 3000 Hz. It should be noted, however, that trends indicate that greater modulation rates should yield greater signal-to-noise ratios.

Solution nebulization calibration plots

Calibration plots were generated for sodium and scandium with 2000 Hz frequency modulation. Results for sodium indicated linear response for the calibration range from 0.05 to 0.5 ppm with a correlation coefficient (r²) of 0.9993. For scandium in the 1 to 50 ppm range, r² was 0.997. Both plots exhibit significant rollovers at higher concentrations. These rollovers can be attributed to self-absorption by the analytes.

Electrothermal vaporization background correction

Compared to pneumatic nebulization, using electrothermal vaporization (ETV) for analyte introduction often improves limits of detection by a factor of 5 to 100.^21–24 This characteristic is due to the selective elimination of water and sample concomitants prior to the introduction of vaporized analyte. However, ETV methods often introduce background fluctuations which degrade performance. When the sample is heated during vaporization, the carrier gas expands due to the increase in temperature. This leads to pressure fluctuations which result in background emission intensity changes. Background correction is needed to remove the effects of the background shifts.

For sodium, scandium and chromium, optimized ETV operating conditions were found for the drying, ashing and atomization steps. Following 60 to 120 s drying steps, samples were ashed for 5 s and atomized for 1.5 to 1.75 s. Data acquisition was initiated during the ashing step. Detector signal was monitored for 10 s with a 40 kHz data acquisition rate with a background correction modulation rate of 2 kHz. Sets of 400 background corrected data points (0.2 s) were averaged to produce single data points. The averaged background corrected, raw analytical and raw background signals were plotted.

Figs. 12 and 13 show plots of observed PMT signal versus time for 1000 ppm scandium and chromium during electrothermal vaporization. The upper lines are the raw analytical signals when the AOD is "on-line", the middle lines are the observed background signals when the AOD is "off-line", and the lower lines are the background-corrected signals. Significant background shifts during the vaporization steps are noticeably present. It is likely that these background shifts are due to a combination of the molecular carbon emission from the ETV and pressure-induced fluctuations in the plasma background intensity. In both cases, the background shift is equivalent to 10–20% of that of the analytical signal. Background correction allows these shifts to be eliminated.

Fig. 12 Electrothermal vaporization of 1000 ppm scandium: upper trace, background correction half-cycle monitoring line emission with an AOD frequency of 33.7 MHz; middle trace, background correction half-cycle monitoring spectral background with an AOD frequency of 43.7 MHz; lower trace, background corrected analytical signal, the difference between the above traces.

Fig. 13 Electrothermal vaporization of 1000 ppm chromium: upper trace, background correction half-cycle monitoring line emission with an AOD frequency of 33.7 MHz; middle trace, background correction half-cycle monitoring spectral background with an AOD frequency of 43.7 MHz; lower trace, background corrected analytical signal, the difference between the above traces.

A calibration plot was generated using sodium as the analyte. With a calibration plot range from 0.01 to 0.25 ppm, r² was 0.9993. At concentrations higher than 1 ppm, significant calibration plot self absorption-induced rollover occurs. Comparing the calibration plot using ETV for analyte introduction to the one obtained using the USN for analyte introduction, the sensitivity is improved by a factor of five.

Conclusion

The AOD background correction system was successfully characterized using hollow cathode lamp and MIP radiation sources. The current system shows a relatively large spectral range, from at least 420 nm to 590 nm. Masking the AOD significantly improved the ability to detect the first-order diffraction, to remove scattered radiation, and to remove radiation outside the acceptance angle of the AOD. Diffraction efficiencies were in the range of 25 to 60%.

The characterization of the AOD background correction system with the MIP gave rise to the use of the system for trace-metal emission analysis. With AOD background correction, the analytical, background and background-corrected signals could be observed in a rapid sequential fashion. The AOD background correction system was successfully used to obtain calibration plots using solution nebulization and ETV analyte introduction.

References

1. I. G. McWilliams, J. Sci. Instrum., 1959, 36, 51.
2. W. Snelleman, T. C. Rains, K. W. Yee, H. D. Cook and O. Menis, Anal. Chem., 1070, 42, 394.
3. T. C. Rains and O. Menis, Anal. Lett., 1974, 7, 715.
4. M. S. Epstein and T. C. O'Haver, Spectrochim. Acta, Part B, 1975, 30B, 135.
5. M. S. Epstein, T. C. Rains and T. C. O'Haver, Appl. Spectrosc., 1976, 30, 324.
6. O. Rose, Jr., D. W. Mincey, A. M. Yacynch, W. R. Heineman and J. A. Caruso, Analyst, 1976, 101, 753.
7. H. Kawaguchi, M. Okada, T. Ito and A. Mizuike, Anal. Chim. Acta, 1977, 95, 145.
8. S. R. Koirtyohann, E. D. Glass, D. A. Yates, E. J. Hinderberger and F. E. Lichte, Anal. Chem., 1977, 49, 1121.
9. L. Lepine, M. Provencher, K. Thammavong, K. C. Tran and J. Hubert, Appl. Spectrosc., 1992, 46, 864.
10. R. G. Michel, J. Sneddon, J. K. Hunter, J. M. Ottaway and G. S. Fell, Analyst, 1981, 106, 288.
11. R. J. Sydor and G. M. Hieftje, Anal. Chem., 1976, 48, 535.
12. J. A. Deblaquiere, K. C. Harvey and A. K. Hellman, Am. J. Phys., 1991, 59, 443.
13. I. Abdulhalim, C. N. Pannell and D. N. Payne, Appl. Phys. Lett., 1993, 62, 3402.
14. E. Kruger and T. Kerseborn, Rev. Sci. Instrum., 1994, 65, 2482.
15. L. Zeng, H. Matsumoto and K. Kawachi, Opt. Eng., 1996, 35, 1662.
16. C. D. Tran, Anal. Chem., 1992, 64, 971A.
17. T. M. Spudich, B. A. Pelz and J. W. Carnahan, Appl. Spectrosc., 1997, 51, 765.
18. T. M. Spudich, Ph.D. Dissertation, Northern Illinois University, DeKalb, IL, USA, 1999..
19. D. L. Haas, J. W. Carnahan and J. A. Caruso, Appl. Spectrosc., 1983, 37, 82.
20. J. J. Urh, K. G. Michlewicz and J. W. Carnahan, Spectrochim. Acta, Part B, 1985, 40B, 493.
21. M. Wu and J. W. Carnahan, Appl. Spectrosc., 1990, 44, 673.
22. K. C. Ng and J. A. Caruso, Anal. Chem., 1983, 55, 2032.
23. U. Schafter and V. Krivan, Anal. Chem., 1998, 70, 482.
24. J. D. Ingle, Jr. and S. R. Crouch, Spectrochemical Analysis, Prentice Hall, New Jersey, 1988.

---

Footnote

† Present Address: School of Science, Penn State Erie, The Behrend College, Erie, PA 16563, USA.

---