Comparison of the Analytical Performance of Flame Atomic Magneto- optic Rotation Spectrometry in the Faraday Configuration With That of Flame Atomic Absorption Spectrometry Journal of Analytical Atomic Spectrometry AHMET T. INCE,* JOHN B. DAWSON AND RICHARD D. SNOOK Department of Instrumentation and Analytical Science (DIAS) University of Manchester Institute of Science & Technology (UMIST) P. 0. Box 88 Manchester U K M60 1 QD Atomic magneto-optic rotation spectrometry (AMORS) is an analytical technique for the determination of elements and has much in common with AAS. This paper compares the analytical performance of the two techniques for several elements using flame atomization. Using the same instrumentation for all measurements Mg and Ag were studied in depth by AAS and AMORS.The best AMORS results were obtained when the angle between the planes of polarization of the polarizing and analysing prisms was +45". Under these conditions the detection limits and linear ranges were slightly better than those obtained by AAS. Keywords Atomic magneto-optic rotation; atomic absorption; frame atomization Faraday configuration; magnesium and silver determination In 1924 Hanle' observed the rotation of the plane of polarized light when it passed through atoms located within a magnetic field. The theory of magneto-optic rotation was introduced by Corney et a1.' on a semi-classical basis. The determination of elements by magneto-optic rotation was first proposed by Church and Hadeish? in 1974; they determined mercury in air as did Stephens4 in 1978.Since that time magneto-optic rotation has been shown to be potentially useful as an analyt- ical technique for trace element analyses.'-12 AAS is a well established technique and has been successfully utilized since the 1 9 4 0 ~ ' ~ and commercialized since 1955.14 A general prob- lem in AAS is background absorption by molecular species and the sample matrix. Atomic magneto-optic rotation spec- trometry (AMORS) is a more recent technique for trace element analysis and has not yet been commercialized. However by the very nature of its physical measurement it is a technique which discriminates against background absorp- tion. AMORS depends on the rotation of the plane of polariz- ation of resonance radiation by birefringence and dichroism generated in a magnetized atomic vapour at the wavelength of the resonance line of the analyte element. The measurement of the rotation of the incident plane polarized light also leads to AMORS having a multi-element spectroscopic capability when a continuum light s o ~ r c e ~ ~ * ' ~ is employed.This arises because only rotated light at the particular wavelength under investigation is subsequently detected. In its simplest form the AMOR spectrometer consists of a magnetized atomic vapour situated between a pair of crossed polarizers a light source and a detector. The magnetic field can be applied either in the transverse (Voigt) or longitudinal (Faraday) configuration. If the atomic vapour is not magnetized ~~~ * Present address National Metrology Institute (UME) Scientific and Technical Research Council of Turkey P.0. Box 21 41470 Gebze-Kocaeli Turkey. or is absent light from the source is blocked by the crossed polarizer arrangement so that no light reaches the detector. This configuration is illustrated in Fig. 1 where only light leaking from the polarizers can be detected. This crossed polarizer system produces signals with a quadratic dependence on concentration; hence detection limits are poor and means of linearizing the calibration graph have been sought. THEORY Generalized theoretical treatments have been presented by Dawson and c~-workers~'-~~ and Kankare and StephenP for crossed and partially crossed polarizers in the Voigt and Faraday configurations. The conclusion of these treatments as applicable to a Faraday-configured instrument will be summa- rized here as a basis for examination of the experimental results. Polarizer-Analyser Axes Orthogonal ('Crossed') The intensity of the transmitted radiation I is given by the expression ctL2 K2 I,= ~ [( 1 - p)2 + 4psin28] 8 where P=aR/aL (dichroic component); ctL and aR are the attenuation coefficients of the left- and right-handed circularly polarized components of the electric vector (amplitude 6) of the incident linearly polarized radiation; intensity = I = 1/02 28 is the phase difference between the components induced by birefringence of the magnetized atomic vapour and is proportional to the number of atoms N in the optical path.When p-1 and 8 is small eqn. (1) reduces to i.e. IT cc N 2 (3) For low concentrations of analyte eqn.(3) predicts a quadratic response to changes in analyte atom numbers. Such a response curve may be linearized by taking the square root of the transmitted intensity as the analyte-dependent function. It should be noted that the attenuation coefficients ctL and ctR are also a function of N . When N is small aL and d R approach unity. Polarizer-Analyser Axes Offset (Uncrossed) by a Small Angle The general expression for the intensity of radiation transmitted by a system consisting of a magnetized atomic vapour between Journal of Analytical Atomic Spectrometry October 1996 Vol. 1 1 (967-972) 967Magnet solenoids Monochromator Fig. 1 Block diagram showing the AMOR spectrometer arrangement. HCL hollow cathode lamp; PSU power supply unit; L1 L2 L3 L4 lenses F flame; PMT photomultiplier tube polarizers uncrossed by a small angle A is (4) CtL2 Vo2 I T = - [( 1 - p)2 + 48 sin2(A + 8)] 8 - [(I -p)2+4p(sin2 8 cos2 A+cos2 o sin2 A (5) 8 1 2 1 + - sin 28 sin 2A) When 8=0 the transmitted intensity is a L 2 L IT = - [( 1 -p)2 + 4psin2 A] 8 and represents the background against which the analytical signal is measured. The analytical signal is however linearized and enhanced when 8 is close to 0 and the term 1/2 (sin 28 sin 2A) of eqn.( 5 ) becomes dominant. In these circum- stances if the offset angle is set negative i.e. against the direction of optical rotation the initial gradient of the cali- bration graph will be negative. The sensitivity is increased and background signal cancelled if the difference signal between the positive and negative offset angles (+A) is used as the analytical signal I D .From eqn. ( 5 ) ID = IT( + A) -IT( - A) Thus (sin 2A sin 28) M L ~ R I ~ I D = - 2 When N is small and aL and aR x 1 IDKN (7) Polarizer-Analyser Axes at 45" This condition is a special case of eqn. ( 5 ) with A = +45" and can be used to calculate the optical rotation by the magnetized atomic vapour from the measured transmitted intensities 1.45" and 1-45" of the two orientations. Substitution of sin45" = 1/$; cos45" = 1/$; sin( - 45") = - 1/$ and cos(- 45") = 1/$ in eqn. (5) gives I+450= 8 ( 1 + p 2 + 2p sin 28) (9) and (1 +D2-2p sin 28) - 8 aR210 Combining eqns. (9) and (10) leads to (11) e= -sin-' 1 (a) 1+p2 ( + 4 5 " 4 - 4 5 " ) K N 2 +45" + I - 45" Eqn. (1 1) predicts a linear relationship between the number of atoms in the optical path and a simple function of the measured transmitted intensities modified by a dichroic term ( 1 + p2)/2p.Linear Range In order to facilitate comparison between the analytical per- formance of the various modes of operation of the apparatus the upper limit of linearity ( U ) will be defined as the concen- tration at which the gradient of the calibration graph is half that at the origin or in the event of a concave curve twice that concentration. The relative linear range (R) will be expressed as the ratio of the upper limit of linearity down to five times the detection limit D i.e. (R = U/5D). The detection limit is calculated as the analyte concentration corresponding to twice the standard deviation of a determination at near zero analyte concentration.The relative linear range is an indication of the analyte concentration ratio over which analy- ses can be effected with an RSD of 10% or better. EXPERIMENTAL Apparatus Atomic magneto-optic rotation generated by several analyte elements was studied using atomization in an air-C2H2 flame with the magnetic field in the Faraday configuration. The apparatus used is shown schematically in Fig. 1 and its specifi- cation summarized in Table 1. A particular feature of the apparatus was the use of a Rochon prism as the analysing polarizer. This prism was chosen because it generates two orthogonally polarized beams of radiation with a small angle of separation from an incident radiation beam. The prism was mounted on the optical axis such that the two beams passed through the spectrometer entrance slit one above the other.On emerging from the spectrometer the beams were directed to two separate photomultipliers. By means of this arrange- ment in addition to single channel measurement of IT for 968 Journal of Analytical Atomic Spectrometry October 1996 VoE. 1 1Table 1 Specification of instrumentation Monochromator RLD*/A O mm-' Grating spacing/lines mm- ' Entrance slit-width/mm Exit slit-width/mm Polarizer Anal yser Electromagnet maximum strength/T Atomizer Beam area on flame/cm2 Pathlength/mm HCL current/mA PMT voltage/V Nebulizer uptake/ml min Coil current/A Rank Hilger monospek. 8 1200 0.5 0.5 Glan-air prism Rochon prism 0.7 30 Air-acetylene flame 0.4 20 8 lo00 8 * RLD is the reciprocal linear dispersion.various configurations of the polarizers simultaneous channel measurement of I + 450 and I - 450 could be made. The polarizing component was a Glan-air prism which was attached to a rotatable mount so that the angle of polarization of the incident radiation relative to the polarization axis of the fixed analyser could be varied. Four principal orientations of the polarizer were used crossed polarizers A =O; small angle offset A= & 6 O ; large angle offset A = +45" and for atomic absorption measurements A = 90". Test solutions were pre- pared from analytical-reagent grade chemicals and doubly distilled water. Preliminary Experiments Several experiments were conducted with a view to selecting the two elements with the strongest analytical signal for further investigation.The crossed polarizer configuration was used and the response of the AMOR spectrometer to Ag Ca Cu Mg Ni and Pb was determined. The atomic transitions and wavelengths of the analytical lines are given in Table 2; the splitting components were found from the All of these analytes gave an AMORS signal but those of Ca and Ni were poor and so were not studied further. For Ca this lack of intensity was attributed to non-optimum flame con- ditions for the element. The dependence of the signal on the magnetic field strength was examined for Ag Cu Mg and Pb. The results are shown in Fig. 2. As the magnetic field strength increases the rotating power of the atomic vapour is initially enhanced as the components of the Zeeman-split analyte spectral line move apart and their optical interaction with the emission line from the light source increases.However at high magnetic fields the components may move beyond the profile of the emission line and thus the AMORS signal will begin to decrease. All the response curves in Fig. 2 show a maximum. Multiple maxima arise from the fine structure and splitting patterns of the spectral line. As Ag and Mg appeared to be the most sensitive elements they were chosen for further study using magnetic field strengths corresponding to the maximum Table 2 Atomic transitions for analytical lines used in this work & 6 component splitting Element Term symbols l/nm pm T-' (refs. 20-22) Ag+ :sl/2 -'2p3/2 328.1 k 5.0 Ca so JPI 422.6 k 8.2 cu* 2s1,2 d2P3,2 324.7 +_ 5.0 Mg 'so +'PI 285.2 & 3.8 Ni* 'F4 +3Gj 232.0 + 3.0 Pb 3P0 +3P1 283.3 5.6 * Zeeman splitting component has more than one sigma component on each side of the analytical wavelength.1.4 1.2 ' 1 E 0 T3 3 C .- - 8 0.0 - a K 0 a 0.6 U 0 .- 0.4 0.2 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Applied magnetic field strengthn Fig. 2 Optimization of magnetic field strength in crossed polarizer configuration for A Mg (40 ppm); B Ag (80 pprn); C Cu (100 ppm); and D Pb (1000 ppm) responses of the two elements (for Ag 0.27T and for Mg 0.55 T). RESULTS AND DISCUSSION AMORS for Ag and Mg With Crossed Polarizers In the conventional AMOR configuration i.e. with crossed polarizers the calibration graphs obtained for Mg and Ag and are shown in Figs. 3A and 4A respectively. As predicted by 3 T I 2.5 0 10 20 30 40 50 60 Concentration (ppm) Fig.3 AMORS calibration graphs for Mg.A Crossed polarizer; B square root of curve A amplitude; C small positive offset polarizer (+6"); D small negative offset polarizer (-6"); and E difference between curves C and D Journal of Analytical Atomic Spectrometry October 1996 Vol. 11 9690 10 20 30 40 50 oa Concentration (ppm) Fig. 4 AMORS calibration graphs for Ag. A Crossed polarizer; B square root of curve A amplitude; C small positive offset polarizer (+ 6"); D small negative offset polarizer (- 6"); and E difference between curves C and D various workers"*'* [eqn. (3)] the calibration graphs are non- linear and appear to be quadratic in nature; this can be demonstrated by taking the square root of the transmitted intensity and plotting this parameter against concentration to obtain a more linear calibration graph (Figs.3B and 4B). AMORS for Ag and Mg With Small Angle Offset Polarizers When the polarizers are uncrossed radiation is transmitted by the system and all analytical measurements are made against this background signal. With a small positive offset (A = + 6") increasing slightly concave response curves were observed (Figs. 3C and 4C). When the offset was negative (A= -6") the transmitted intensity initially decreased with increasing analyte concentration to produce minima in the response curve (Fig. 3D and 4D) at 45 pg ml-' Ag and 17 pg ml-' Mg. These minima arise because the plane of the radiation incident on the analyser is rotated by the AMOR effect through an orientation which is orthogonal to the transmission axis of the analyser.The initial gradients of the curves for both positive and negative offsets are almost linear. When the difference signal between positive and negative offsets is plotted (Figs. 3E and 4E) the effect of the background signal is eliminated and the curve begins at the origin. AMORS for Ag and Mg With 45" Offset Polarizers Another method of obtaining a linear calibration graph is to use the linear relationship between the angle of rotation of the linear plane polarized light and sample concentration c pre- dicted by eqn. (1 1). When dichroism is small p~ 1 and the dichroic term I( 1 + P2))/2p3 z 1. Hence d= - 1 s i n - l ( ; + 4 5 - ~ - 4 5 0 ) ~ N 2 + 45" + 1-450 In order to measure 8 the axes of the polarizer and analyser prisms were set at 45" relative to each other.The latter prism (Rochon) then generated the two signals and I-45.= from which 8 was calculated using eqn. (12). Calibration graphs obtained at several magnetic field strengths are shown in Figs. 5 and 6 for Mg and Ag respectively. The graphs are initially linear but then begin to curve and 'roll-over'. Theoretically [eqn. (ll)] the function reaches its maximum value corre- sponding to an optical rotation of 45" i.e. the axes of the polarized radiation and the analysing prism are parallel in one detector channel and orthogonal in the other. In practice the combined effect of the dichroic term [( 1 +P2)/2P] and stray light becomes significant and leads to maxima of less than 45".Although the general form of the response curves is as predicted by eqn. ( l l ) the change in shape with change in magnetic field strength demonstrates the crucial role of the interaction of the plane of the emission and absorption line profiles in determining the precise shape of the curves. As the magnetic field changes the dichroic effect and the proportion of non- interactive radiation (stray light) change as the overlap between the emission and absorption line profiles changes. Thus eqn. (12) should be modified to eqn. (13) to include stray light terms I An example of this situation is seen in Fig. 6C where at high concentrations of Ag there is little change in the apparent signal because the peak has been broadened by the complex splitting peaks of the lines and much of the optical rotatable light has been absorbed and stray light has become significant.This effect may also be attributed to the second 'peak' seen in Fig. 2. AAS Studies With the AMOR Spectrometer Atomic absorption curves were obtained using the AMOR apparatus with the polarizers in the 45" setting. The absorbance (A) of the atomic vapour was calculated as I + 45" + I - 45")o ( I + 45" + I - 45% A =log where the subscript '0' indicates measurements when a blank solution is aspirated and 'c' when analyte is present. Calibration graphs without and with an applied magnetic field are pre- sented in Figs. 7 and 8 respectively. The curves are initially linear but as stray light becomes a greater proportion of the measured intensity curvature begins to arise.Fig. 8 includes two AMOR calibration graphs to facilitate visual comparison of the linearities of AMORS and AAS. Until the 'roll-over' in the AMOR curves occurs the relative curvatures of all the curves are similar. The reduction in AAS sensitivity when the magnetic field is applied is relatively small (25% for Mg and 33% for Ag) and demonstrates that the optimum field strength for AMOR measurements is much less than that required to generate the maximum Zeeman AAS (ZAAS) signal. This difference is due to the fact that in AMORS the signal is generated by interaction of the overlapping emission and absorption line profiles whereas in ZAAS the maximum signal is generated by separating the emission and absorption line profiles as completely as possible in order that the intensity (field 0n):intensity (field off) ratio is as large as possible.970 Journal of Analytical Atomic Spectrometry October 1996 Vol. 1 10 20 40 60 80 100 120 140 (::,oncentration (ppm) Fig. 5 Calibration graphs for Mg with 45" offset polarizer cc' ?figuration. Magnetic field strengths used A 0.225; B 0.35; C 0.5; and D 0.65 T 35 1 C T 0 40 80 120 160 20C 240 280 320 360 400 440 480 520 560 600 Concentration (ppm) Fig. 6 Calibration graphs for Ag with 45" offset polarizer configuration. Magnetic field strengths used A 0.225; B 0.275; and C 0.37 T 0.35 0.30 0.25 a 2 0.20 2 0.15 a 0.10 0.05 0 10 20 30 40 50 t Concentration (ppm) Detection Limits Upper Limit of Linearity and Linear Range A summary of the experimental results is presented in Table 3.The best results for the AMORS system were obtained when the polarizer axes were offset by 45" and the worst when the axes were orthogonal. The latter is a consequence of the poor sensitivity of a quadratic response system at low concentrations and failure of the crossed polarizers to exclude non-interactive radiation completely. Although the upper limit of linearity was comparable to that with other configurations the relative linear range was poor. When the polarizer axes were offset from orthogonal by 6" the detection limits and relative linear range were improved 10-fold. However the greatest improve- ment was obtained when offset angles of 45" were employed. Under these conditions detection limits were further reduced and the upper limits of linearity extended.These improvements in performance are attributable to the increase in light trans- mitted by the system in this configuration which in turn leads to greater accuracy in measurements. As AAS is a well established technique for the determination of elements and in our situation can be carried out with the Fig. 7 Absorption calibration graph using AMOR apparai US. No magnetic field. A Mg; and B Ag same instrumentation as AMORS it provides an effective means for evaluating the merits of AMORS. The sensitivities 971 Journal of Analytical Atomic Spectrometry October 1996 Vol. 11linear ranges of the AAS measurements were worse (z 50%) than the best of those obtained by AMORS using 45" offset polarizer axes but were better than those obtained with all the other AMORS configurations.When AAS measurements were made with a magnetic field applied to the atomic vapour there was a reduction in sensitivity of approximately 50% and a doubling of the detection limits. The studies of Ag and Mg by both AMORS and AAS did not reveal any differences in their analytical behaviour which might lead to the preference of one technique over the other. CONCLUSION In the conventional crossed polarizer configuration AMORS 0" 50 100 150 2b 250 300 350 - o Concentration (ppm) Fig. 8 Comparison of atomic absorption calibration graphs with magnetic field and AMOR calibration graphs for Mg and Ag obtained Magnetic field strengths for Mg and Ag are 0.55 and 0.27 T respectively has an inherent quadratic dependence on analyte concen- with the 45" POlariZer Configuration.* degrees; + 5 absorbance. tration. This characteristic can be linearized by offsetting the polarizer prism from orthogonality by a small angle (x6"). Table 3 Detection limits and linear ranges in AMORS and AAS Configuration AMORS Field strength/T Crossed polarizers (linearized by taking square root of IT) Detection limit (ppm) Upper limit of linearity (ppm) Relative linear range (ppm) Offset polarizers A=++" Detection limit (ppm) Upper limit of linearity (ppm) Relative linear range (ppm) Detection limit (ppm) Upper limit of linearity (ppm) Relative linear range (ppm) A = -6" Difference [+6"-(-6")] Detection limit (ppm) Upper limit of linearity (ppm) Relative linear range (ppm) 45" Offset polarizers Detection limit (ppm) Upper limit of linearity (ppm) Relative linear range (ppm) AAS No field on Sensitivity (ppm per 0.0044 A ) Detection limit (ppm) Upper limit of linearity (ppm) Relative linear range (ppm) With field Sensitivity (ppm per 0.0044 A ) Detection limit (ppm) Upper limit of linearity (ppm) Relative linear range (ppm) Analyte 0.27 10 60 1.2 0.6 > 40 > 13.3 0.6 7.3 22 0.4 40 20 0.4 220 110 1.2 0.6 > 50 > 20 1.8 1.1 200 36 0.55 3 55 3.7 0.4 > 50 > 25 0.4 6.5 13 0.3 15.3 23 0.1 50 100 0.4 0.15 45 60 0.5 0.4 > 120 > 60 This modification improved the detection limit and relative linear range by an order of magnitude.The use of +45" offset angles produced the lowest detection limits with the greatest linear ranges. From our studies it appears that this arrangement is the most suitable for analytical applications of AMORS.The detection limits and relative linear ranges obtained were comparable to or better than those obtained by AAS using the same apparatus. AMORS however has the potential to discriminate against background absorption. The poor detec- tion limits resulting from the experimental configuration could be improved by increasing the optical path through the flame and more efficient utilization of the atomic vapour. Moreover as the magnetic field strength required to generate an AMORS signal is less than that for ZAAS an AMORS system can offer greater flexibility in the design of magnet/atomizer systems using either flames or furnaces. REFERENCES 1 Hanle W. Z . Phys. 1924 30 93. 2 Corney A. Kibble B. P.and Series G. N. Proc. R. SOC. London Ser. A 1966 293 70. 3 Church D. A. and Hadeishi T. Appl. Phys. Lett. 1974 24 185. 4 Stephens R. Anal. Chim. Acta 1978 98 291. 5 Davis L. A. Krupa R. J. and Winefordner J. D. Spectrochim. Acta Part B 1986 41 1167. 6 Hermann G. CRC Crit. Rev. Anal. Chem. 1988 19 323. 7 Davis L. A. and Winefordner J. D. Anal. Chem. 1987 59 309. 8 Debus H. Hanle W. Scharman A. and Wirz P. Spectrochim. Acta Part B 1981 36 1015. 9 Jolly G. S. and Stephens R. Spectrochim. Acta Part B 1984 39 335. 10 Monnig C. A. and Hieftje G. M. J. Anal. At. Spectrom. 1988 3 679. 11 Kitagawa K. Shigeyasu T. and Takeuchi T. Analyst 1978 103 1021. 12 Yamamoto M. and Murayama S. J. Opt. SOC. Am. 1979,69,781. 13 Woodson T. T. Rev. Sci. Instrum. 1939 10 308. 14 Walsh A. Spectrochim. Acta 1955 7 108. 15 Yamamoto M. Murayama S. Ho M. and Yasuda M. Spectrochim. Acta Part B 1980 35 43. 16 Kankare J. and Stephens R. Spectrochim. Acta Part B 1983 38 1301. 17 Kersey A. D. Dawson J. B. and Ellis D. J. Spectrochim. Acta Part B 1980 35 865. 18 Kersey A. D. and Dawson J. B. Anal. Proc. 1981 18 187. 19 Dawson J. B. King P. R. Duffield R. J. and Ellis D. J. J. Anal. and detection limits of AAS measurements made in this work are up to times than those achieved using tional equipment under optimum conditions. This reduced sensitivity is attributable to a short optical Path through the flame ( z 20 mm) inefficient flame geometry and non-optimum flame gas composition. However the lack of sensitivity does not invalidate compari- sons between AAS and AMORS as this limitation will affect both techniques similarly. The detection limits and relative At. Spectrom. 1989 4 245. 20 Wagenaar H. C. PhD Thesis Delft University 1976. 21 Mavrodineanu R. and Boiteaux H. Flame Spectroscopy Wiley 1965 New York p. 31. 22 Preli F. R. Jr. Dougherty J. P. and Michel R. G. Spectrochim. Acta Part B 1988 43 501. Paper 6/01 71 2F Received March 1 I 1996 Accepted June 7 1996 972 Journal of Analytical Atomic Spectrometry October 1996 Vol. 11