Instrumentation James M. Harnly Nutrient Composition Laboratory, BHNRC U.S. Department of Agriculture Bldg. 161, BARC-East Beltsville, Md. 20705
Multielement Atomic Absorption with a Continuum Source Historically, there has been continued interest in a multielement atomic absorption spectrometer (AAS). The development of AAS, inherently as a single-element technique, revolutionized the field of atomic spectrometry in the mid-1950s. The conceptually simple design, excellent detection limits, high sensitivity and specificity, and the low cost of flame AAS quickly made it the method of choice for trace metal determinations despite its single-element limitations. Today, carbon furnace atomization provides AAS with detection limits that, for most elements, are unsurpassed in the field of atomic spectroscopy. The combination of carbon furnace atomization with multielement detection is, potentially, a powerful analytical tool. The key to the success of AAS and the greatest hindrance to the development of multielement AAS is the light source. Conventional AAS uses a hollow-cathode lamp (HCL) that emits intensely and almost exclusively at the wavelength of the element to be analyzed. A different HCL is used for each element. Efforts to develop multielement HCLs or to combine the output of several HCLs using beam splitters have proven largely unsatisfactory. Continuum sources (specifically short-arc, high-pressure xenon arc lamps) have the advantage of provid-
ing energy across a broad spectrum (200-1500 nm). Unfortunately, the very nature of these sources, their broad-band emission and the instability of the arc, lead to poor sensitivity (or characteristic concentrations) and detection limits, respectively. Direct substitutions of continuum sources for HCLs, without modification of any of the other instrumental components, have been unsuccessful. In retrospect, the barriers presented by these undesirable characteristics of continuum sources have proven almost insurmountable. The only commercial use of continuum sources in the field of AAS has been as secondary sources for the correction of background interferences. In the past eight years, a simulta-
SIMAAC components 300-W Cermax xenon arc lamp (Model LX300UV) and power supply (Model PS300-1), ILC Technology, Sunnyvale, Calif. Echelle polychromator (Spectraspan III), Beckman Instrument Co. Galvanometer (Model G325D) and scanner controller (Model CCX-650), General Scanning Inc., Watertown, Mass. 11/34-VE Declab minicomputer with analog-to-digital and digital-to-analog converters, real-time programmable clock, digital I/O, floating-point processor, video terminal, and line printer, Digital Equipment Corp., Maynard, Mass.
This article not subject to U.S. Copyright Published 1986 American Chemical Society
neous multielement atomic absorption continuum source (SIMAAC) spectrometer has been developed that overcomes many of the problems associated with continuum sources by modifying the rest of the instrumental components. The success of SIMAAC can be attributed to the synergistic combination of a high-resolution échelle polychromator, wavelength modulation, and computerized, highspeed data acquisition (see box on this page and Figure 1). Each of these components is essential to the acquisition of useful analytical data from a continuum source. Ironically, the selection of components for SIMAAC, which was aimed at duplicating AAS characteristics in the multielement mode, has produced several capabilities that surpass conventional AAS. The high-resolution échelle polychromator provides characteristic concentrations, curve linearities, and elemental specificities comparable to line source AAS and provides multielement detection for up to 20 elements. Wavelength modulation minimizes the troublesome continuum source instability, provides detection limits comparable to line source AAS (above 280 nm), and accurately corrects for background absorption. High-speed data acquisition, in conjunction with wavelength modulation, permits extension of the calibration range to 5-6 orders of magnitude for each element. SIMAAC is compatible with either flame or carbon furnace atomization. Compared with commercial multielement spectrometers, specifically the inductively coupled plasma-atomic emission spectrometer (ICP-AES), SIMAAC offers superior detection limits using smaller sample volumes. The calibration ranges of the two spectrometers are comparable, but ICP-AES with a direct-reading polychromator can determine more ele-
ANALYTICAL CHEMISTRY, VOL. 58, NO. 8, JULY 1986 · 933 A
Continuum source
Figure 1. Block diagram of SIMAAC
ments simultaneously. In general, SIMAAC is more susceptible to chemical interferences, whereas ICP-AES is more susceptible to spectral interferences. SIMAAC was developed jointly by the Nutrient Composition Laboratory at the U.S. Department of Agriculture (USDA) and by T. C. O'Haver at the University of Maryland. For the past five years, SIMAAC has been used routinely at USDA for the analysis of agricultural, biological, and food materials. Resolution
A simple substitution of a continuum source for an HCL will not produce comparable analytical signals.
An optical explanation for the difference in the signals is presented in Figure 2. The HCL is a line source presenting "monochromatic" light to the absorbing species (Figure 2a). Here "monochromatic" is defined as photons arising from a single electronic transition. The finite, but small, line width is determined by the broadening processes associated with the source. The low characteristic concentrations (concentration necessary for an absorbance of 0.0044) and the calibration curve linearities (2V2 orders of magnitude of concentration) that are typical of line source AAS are the result of the narrow width of the HCL emission line. HCLs also provide low noise levels
and thus excellent detection limits. A monochromator or a band-pass filter is necessary to isolate the spectral region of interest, but the band-pass of the isolating device does not influence the monochromatic nature of the absorption measurement for an HCL. Upon substituting a continuum source for an HCL, the spectral region over which absorption is measured is determined by the spectral band-pass of the monochromator (Figure 2b). Mediumresolution monochromators usually used for AAS have spectral band-passes several orders of magnitude wider than HCL line widths. Thus, Strip with a continuum source, the chart absorption measurement will occur over a much broader wavelength region and will no longer be monochromatic. Because the conditions of Beer's law are not met, the characteristic concentrations and calibration curve linearities for a continuum source with these monochromators are significantly poorer than for line source AAS. The absorption measurement with a continuum source can be rendered monochromatic by use of a high-resolution spectrometer (Figure 2c). The narrow spectral band-pass of the spectrometer is substituted for the narrow HCL source. Fabry-Perot interferometers have excellent resolving power but are fundamentally unusable for routine use. Echelle monochromators have resolution comparable to the absorption profile width and are much more convenient to operate. In addition, échelle spectrometers have also
Figure 2. A comparison of instrumental, absorption, and light source profiles (a) Absorption with a hollow-cathode lamp, (b) absorption with a continuum source and a medium-resolution monochromator, (c) absorption with a continuum source and a high-resolution monochromator, and (d) absorption with a continuum source, wavelength modulation, and an échelle monochromator with 50-μπι slit widths
934 A · ANALYTICAL CHEMISTRY, VOL. 58, NO. 8, JULY 1986
Figure 3. Absorption profiles for wave length modulation with a continuum source (a) No background absorption and (b) 50 % back ground absorption
been designed as polychromators; i.e., they have multiple exit slits with dedi cated detectors. Unfortunately, con tinuum sources are considerably less stable than HCLs, and the high-reso lution spectrometers restrict the transmitted intensity. Thus, regard less of the resolution of the monochromator, the detection limits with a con tinuum source are significantly worse than those obtained with an HCL. Wavelength modulation
Snelleman demonstrated that both the instability of the continuum source and broad-band background absorption can be overcome by sweep ing back and forth, or modulating, across the analytical wavelength (Fig ure 2d) and detecting the resulting signal with a frequency-sensitive de tector (lock-in amplifier). Wavelength modulation is easily implemented by mounting a quartz refractor plate on a galvanometer immediately behind the entrance slit (Figure 1). As the quartz plate is rotated first in one direction and then the other (±12.5° maxi mum), the wavelength is shifted back and forth across the exit slit. The ana lytical signal produced by a single sweep across the absorption profile is presented pictorially by either trace a or trace b in Figure 3. Atomic absorption appears as an in verted peak or trough in the continu um. Whereas line source AAS is onedimensional, wavelength modulationcontinuum source AAS gives two-di mensional data—intensity vs. wave length. From this two-dimensional in tensity data, absorbances are comput ed in a slightly unconventional manner. For conventional AAS, the
reference intensity, In, is determined from light passed around the atomizer (double-beam instruments) or from To values measured prior to the sample measurement. With wavelength mod ulation and a continuum source, In is redefined as the average of the inten sities measured to either side of the analytical line (much like Zeeman AAS with the magnet around the HCL). The transmitted or sample in tensity, /, is defined conventionally as the intensity at the profile center. It can be seen that with this definition the reference and sample beam are perfectly superimposed. They differ only in wavelength and, due to the modulation, a slight time lapse be tween measurements. Thus, wave length modulation with a continuum source is double-beam in function al though not by strict physical design. Absorbance is a proportional proc ess; i.e., as given by Beer's law, the light absorbed is equal to the product of the absorptivity and the analytical concentration and is independent of the source intensity. Thus, a decrease in the source intensity produces a sim ilar decrease in both the reference and sample intensity, as shown in Figure 3. Profile b results when the light inten sity for profile a is attenuated by 50%. Because absorption is proportional to the source intensity, the ratio of / υ to I remains constant, as does the absor bance (log loll). The independence of the computed absorbance from the source intensity produces the inherent correction for the continuum source instability and background absorption that Snelle man described. Because the source fluctuation or flicker noise of continu um sources occurs at relatively low frequencies ( characteristic concentration, and
detection limit are computed and printed in the final report. These val ues allow the instrumental perfor mance to be monitored for each ex periment. In addition, each sample is categorized as to its analytical S/N ra tio. The sample signal may be quanti tative (>15σβ), semiquantitative (>3(7β but