JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 675 Reduction of Magnet Size in Direct Zeeman Atomic Absorption Spectrometry Roger Stephens Department of Chemistry Dalhousie University Halifax Nova Scotia Canada B3H 4J3 A field-on-source design is described which gives Zeeman background correction while allowing a significant reduction of magnet size to be achieved. The design is inherently suited to observation of the longitudinal Zeeman effect. The system was tested with copper using polarization modulation to selectively detect the difference in absorption between the c+ and c- components of the longitudinal multiplet emitted by the source. Effective background correction is achieved including tolerance towards anisotropy at the optical boundary of the atomizer. In contrast to the transverse field-on-source configuration the present system did not produce an interfering off-set signal as a result of self-absorption. Keywords Zeeman correction; atomic absorption spectrometry; permanent magnet The use of Zeeman background correction has become com- monplace for measurements in atomic absorption spectrometry (AAS).The theory of the technique and the various types of instrumental design that are used for its implementation have been described in Current information on the practice and development of the method can be found for instance in the series of on-going reviews (Atomic Spectrometry Updates) which appear in the Journal of Analytical Atomic Spectrometry. Zeeman corrected spectrometers are generally designed around an atomization device usually a furnace which is contained within the magnetic field. The emission line from a conventional hollow cathode lamp then interacts with the Zeeman absorption multiplet at the atomizer to generate total and background absorption signals for subsequent electronic processing.A disadvantage which is inherent to this arrange- ment is that the magnet must be large enough to produce a suitably strong field over a gap of sufficient size to contain the atomizer. The need for such a magnet raises the cost of an instrument and reduces its flexibility in terms of the ease of exchange or modification of the atomizer. The present work was carried out in order to investigate the possibility of reducing the demand on magnet size. Theory The size of the magnet needed for Zeeman correction depends upon the volume over which the field must act.In the present apparatus this volume was reduced by use of the field-on- source configuration shown in Fig. 1. In the arrangement shown the planar cathode geometry supports a disc-shaped glow discharge the disc being parallel and immediately adjac- ent to the pole face of the magnet. The depth of the discharge which controls the maximum distance of emitting atoms from the magnet is governed by fill pressure. Visual observation through the lamp wall indicated that the depth never exceeded about 1 mm under the conditions used in this work. The width of the discharge which is set by the internal diameter of the glass envelope was about 2mm across. This degree of con- finement proved sufficient to give adequate Zeeman splitting using only a small bar magnet to generate the field.The geometry shown in Fig. 1 produces approximately paral- lel electric and magnetic fields within the source a condition which allows a stable d.c. discharge to be obtained. The fields are also essentially parallel to the optical axis over the discharge region so that a longitudinal Zeeman effect (no n component; circularly polarized 0 components) is observed. In order to use this configuration in conjunction with the steady field of a bar magnet the optical system shown in Fig. 2 was adopted. In this system an oscillating retarder the photoelastic modulator (PEM) is used to give selective modulation of the intensity difference between the o+ and 6- components emitted by the Vacuum A H 1 cm A r M Fig.1 Schematic diagram of the lamp A anode; C cathode; GD glow discharge; M magnet; and W window PE M L L L Fig.2 Schematic diagram of the optical system F flame-burner assembly; L lens; M monochromator; P polarizer; PEM photoelastic modulator; and S source source. Such an intensity difference is produced as a result of atomic absorption within the atomizer (see Fig. 3) but not by a wide-band background absorption. It should be noted that a structured interference i.e. any background absorption that changes significantly over the wavelength interval between o+ and 0- will be detected similarly and will lead to an error signal which is proportional to the difference in absorption between o+ and 6-. The optical train in Fig.2 is described quantitatively by the Jones calculus.6 The Jones matrix J of the system is written as J= P x R - I ( 0)PEM R( O)FT (1) where P PEM P are the Jones matrices of polarizer P of the PEM and of the flame atomizer 0 is the angle between the transmission axis of P and the stress axis of the PEM,676 2 700 0 0 c - 3 600 500 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 - - - h Emission Fig.3 Schematic diagram of line profiles showing a pressure shift between emission and absorption maxima and the resulting increase in absorption for one Zeeman component and decrease for the other as the magnetic field is applied. (The emission profiles shown here are equivalent to the o+ 0- envelopes in the case of an anomalous mu1 tiplet) and R ( 0 ) is the corresponding rotation matrix T is the transformation matrix between circular and linear basis vectors and connects the circularly polarized emission lines to the linear coordinate system of the other optical components.Separate Jones vectors must be written for o+ and o- because of the frequency shift between them. Hence eqn. (1) can be simplified to J=P,R-'(@)PEM TF (2) Further details on the form and use of the Jones matrices have been given el~ewhere.~ Let the retardance 6 of the PEM be written as 6 = 6 sin(ot) where 6 is the peak retardance of the device and 4 2 7 ~ is its modulation frequency. Let the atom density and path length in the atomizer be N and 1 respectively and the Beer's law absorption coefficients for o+ 6- be K + K -. Evaluation of eqn.(2) then gives the intensity I transmitted by the optical system as I / I = [exp( - K + N l ) + exp( - K - N1)]/2 -sin( 2 0 ) x J1( 26,)[ exp(- K +NI) -exp(- K -NI)] sin@) (3) where J1( ) is the Bessel function of order 1 and ZO=source intensity for o -. Eqn. (3) shows that the optical train selectively modulates the absorption difference between the o+ and 0- components. The frequency shift between the source and atomizer lines caused by pressure broadening ensures that atomic absorption in the atomizer will cause such a difference to occur as seen in Fig. 3. The figure also shows that use of this optical system causes some loss of sensitivity since the absorption difference between o+ and o- is always less than the value of the zero field absorbance for a given N .At the same time however the system is well adapted for use with a low field strength since it is only necessary to produce a CT displacement up to the maximum of the atomizer absorption profile. The field strength required to produce such a displacement depends upon the particular element and type of transition being considered. Experimental Copper was used to test the system. A copper foil cathode (2 x 8 x 0.04 cm thick) was sealed to a glass case with epoxy as in Fig. 1. The anodes were of 14 gauge chrome1 wire. Argon was used as the fill gas. The bar magnet from a magnetic stirrer was 2.8 cm in length x 0.9 cm in diameter weighed 15 g and produced a field of 2.2 kG at the pole faces. Aqueous test solutions from 1 to 100 pg ml-' of copper as copper sulfate were atomized in an air-acetylene flame using a standard Varian-Techtron burner assembly and a 10cm head.The copper line at 624.8 nm was isolated by a McKee-Pederson MP-1018A monochromator with a linear dispersion of 3 nm mm-' at a slit-width of 1 mm. Polarizer P was a 1 x 1 cm calcite prism (Melles-Griot). The modulator was a PEM 80 (Hinds International Portland OR USA) with a fused silica head and oscillating at 0=50 kHz. All other electronics were built in the laboratory. Results Lamp Performance The lamp ran at 5 mA 3 mm of Hg pressure for 24 h before use to condition the cathode. Figs. 4-6 show current versus voltage intensity versus current and emission spectra respectively for the lamp with the magnet in place. Behaviour appears normal with a low pressure current limit and a fairly broad pressure range over which satisfactory performance can be obtained.The optimum output intensity is comparable to that of a commercial source (Varian); however the line-to-background ratio is worse. Stability plots (Fig. 7) show that good magnetic stability is obtained and confirm that the presence of the magnet does not cause any great perturbation of the lamp output. The optimum 1% absorption sensitivity for the source was 800 gOOO 400 ' 1 I I I I 0 1 2 3 4 5 Current/mA Fig. 4 Lamp current versus voltage for fill pressures 0 0.25; 9 1; 0 3 ; and U 6 mm Hg I 0 100 80 - (0 C 0 a z 60 .- c - 40 20 0 Current/mA Fig. 5 Relative intensity versus lamp current for fill pressures a 0.25; 0 1; 0 3 ; and U 6 mm HgJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL.9 677 333 3,” 3 (b) 324.8 327.4 ~ 3 318 5% NaCl Start End -Time Fig. 7 Stability plots over 30 min lamp current 3 mA; fill pressure 3 mm Hg. Upper curve field on; lower curve magnet removed 0.4 pg m1-I; about 10% better when the source was replaced by the Varian lamp. System Performance Absorption signals are shown in Fig. 8. The direction (phase) of the 50 kHz Zeeman signals reversed when the orientation of the magnet was reversed to bring the opposite pole up against the cathode of the lamp an observation confirming that the signals indeed represent a difference in absorption between CT and u- as required by eqn. (3). Calibration curves (Fig. 9) showed little sensitivity to the operating conditions of the lamp. However the signal-to-noise ratio varied markedly with current and pressure.Variations were consistent with the results shown in Fig. 5 for a shot- noise limited response. The roll-over observed in Fig. 9 is probably exacerbated by the low magnetic field strength which was used. Fig. 10 shows the dependence of the signal on the separation between the cathode and the pole face of the magnet. The sharp fall-off illustrates the need to hold the plasma as near to the magnet as possible (however looking at Fig. 3 it can be seen that this would not be the case if the magnet were Fig. 8 Absorption signals observed after successive aspiration of NaCl (5%) and of increasing concentration of Cu2+ Wavelengthhm Fig. 6 Spectra from 318 to 333 nm. Lamp current = 5 mA for both (a) a Varian lamp; and (b) the present source 10 0 20 40 60 80 100 Concentration/pg ml-’ Fig. 9 Calibration curves for fill pressures of 0 0.25; 0 3; and 0 5 mm Hg and lamp currents of 0.8 2 and 5 mA respectively.Fourier series used for curve fitting 100 [i 0 5 10 15 Cathode-pole face distance/mm Fig. 10 Dependence of signal on magnet position stronger allowing both u+ and c- components to be displaced beyond the absorption profile). Background correction for wide-band absorption within the optical path appeared to be satisfactory. No spurious signals were seen under a knife-edge test or as a result of absorption by smoke particles or upon aspiration of a 5% solution of678 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 NaCI. No change in baseline was observed when the modulator was switched on indicating that no background signal occurred as a result of self-absorption within the lamp.This observation contrasts with the behaviour of the transverse field configuration.* The lack of a background signal here due to self-absorption is the result of the symmetry between the emission and self-absorption line frequencies; i.e. because there is no longer any pressure difference between the emitting and absorbing species no frequency shift exists between the zero- field emission and absorption lines in Fig. 3. Hence the differential absorbance between K + and K - remains zero even after the magnetic field is applied. In contrast again to the transverse field arrangement no spectral interference at the edge of the flame was seen.The reasons for this finding are discussed further in the Appendix. Conclusions A Zeeman-corrected spectrometer built around a small bar magnet was found to be functional. The longitudinal-field configuration reduces the spurious signals due to self- absorption and scatter from thermal boundaries which occur with the analogous transverse-field arrangement. No error signals were detected as a result of wide-band absorption. However the occurrence of spectral interferences due to line background absorption must be anticipated. The severity or otherwise of such interferences relative to their effect on a conventional instrument is expected to depend on the nature of the Zeeman splitting shown by the particular interferent. The need to use a specially designed source which is essential to the operation of the present system is a significant disadvan- tage. However it should be noted that fabrication of the source is not complex and that the whole assembly including the magnet is small enough to fit quite easily inside the envelope of a conventional hollow cathode lamp.Appendix Signal Harmonic Generation Through Atomizer Boundary Effects When a Zeeman corrected instrument uses a transverse field applied to the source then the atomizer itself can produce a spectroscopic interference. The effect arises because the thermal gradients around the atomizer cause anisotropic scattering to occur. As a result the n and CJ components of the transverse multiplet are not scattered equally. The consequent imbalance between n and (T intensities creates a spurious signal unless a suitable modulation procedure is adopted.This situation does not occur with the present configuration for the following reason. The Jones matrix X of the thermal boundary can be written in the general form where x1-x4 are complex. Using the same symbols as in the main text the Jones matrix J of the optical train shown in Fig. 2 becomes J = P xR - (n/4)MR( n/4)XT 0 exp(-i6) A cos(6) - iB sin(6) iA* cos(6) + B* sin(c’i) =2a( 0 where a = ( 1 +i),’2J2 A =xl + i x and B=x3+ix4. The output intensity is given by I = { A A * [ l + c o s ( ~ ~ ) ] + B B * [ ~ - c o s ( ~ ~ ) ] } (la+ +Z,-)/2 +(AB*- A*B) sin(26)(1,+ -Z0-)/2 Thus the output contains the even harmonics at 0 2w 4 0 ... However the first harmonic which carries the atomic absorp- tion signal is reduced by the term (la -la-). This term is zero in the absence of any wavelength-dependent absorption over the multiplet or if any absorption which does occur is symmetrical about the zero field line. References 1 Hadeishi T. and McLaughlin R. D. Science 1971 174 404. 2 Grassman E. Dawson J. B. and Ellis D. J. Analyst 1977 102 804. 3 de Loos-Vollebregt M. T. C. and de Galan L. Prog. Anal. Specrrosc. 1985 8 47. 4 Slavin W. and Carnrick G. R. At. Spectrosc. 1985 6 157. 5 de Loos-Vollebregt M. T. C. de Galan L. and van Uffelen J . W . M . Spectrochim. Acta Part B 1988 43 1147. 6 Kliger D. S. Lewis J. W. and Randall C. E. Polarized Light in Optics and Spectroscopy Academic Press 1990 ch. 4. 7 Kankare J. J. and Stephens R. Spectrochim. Acta Part B 1980 35 849. 8 Stephens R. Talunta 1978 25 435. Puper 3106236H Receiued October 19 1993 Accepted Junuary 26 1994