822 Aqzalyst, September, 1979, Vol. 104, $9. 822-830 Correction System for Spectroscopic Determination of Trace Amounts of Cadmium Using the Atomic Faraday Effect with Electrothermal Atomisation K. Kitagawa, T. Koyama and (the late) T. Takeuchi Department of Synthetic Chemistry, Faculty of Engineering, Nagoya University, FuYo-cho, Chikusa-ku, Nagoya, Japan The effect of background absorption of radiation on the signal obtained in Faraday-eff ect atomic spectroscopy has been overcome by comparing the energies transmitted when the optical polarisers are in crossed and parallel configurations. Two systems were developed: one was a static system in which a Glan prism was used to divide the optical beam into two beams of orthogonally polarised radiation that were detected by two photomultipliers ; the other system used a rotating prism to rotate the plane of polarisation of the optical beam and a single photomultiplier with time-sharing electronics to separate the required signals.Cadmium in a starch matrix was deter- mined using the spectral line at 228.8 nm and losses of incident radiation of up to 99% were corrected successfully. Keywords : A tomic spectroscopy ; electrothevmal atomisation ; Faraday eflect ; cadmium determination ; background covrectio:n In a previous paper1 we described, as a novel technique of atomic spectroscopy, an applica- tion of the Faraday effect to the determination of trace amounts of cadmium with an electrothermal atomiser. It was shown that the technique gave a detection limit for cadmium comparable to that of atomic-absorption spectroscopy and wa-, insensitive to the background scattering frequently encountered in atomic-fluorescence spectroscopy.There was, however, a problem in the direct atomisation of biological samples where background absorption caused a reduction in the transmitted energy and led to negative interference. This problem was briefly stated in the previous paper and the principle of a method for correction was suggested. In this study, correction systems based on this principle were constructed and their performances evaluated. Three types of information were obtained either by time or spatial separation of the signals. These data consist of signals arising from the optical rotation due to the Faraday effect, the background absorption and the black-body radiation emitted from the electrothermal atomiser.Theoretical As discussed previously,l the intensity, I J , transmitted through the Faraday configura- tion when the number of atoms in the light beam is small can be expressed as follows: . . * * (1) I1 cc (NLB)21, J’p’(k)F(k,H)2dk . . . . s where N L is the number of atoms in the light beam, I? the transition probability, IB the source intensity, p ’ ( k ) the profile function of the source radiation, F ( k , H ) the Faraday function, s the band pass of the monochroma.tor and k the angular frequency. If any back- ground species are present in the light beam, equation (1) must be modified by multiplying it by a damping term, e-abLb: IL cc (NLB) 21se--a2Jp’( k ) F (k,H) 2dk . . .. * * (2) s where Mb is the apparent absorption coefficient for the background component and L b the length of its cloud in the light beam.If the atomic absorption is small enough, the intensity, 111, transmitted when the plane of polarisation of the polariser is parallel to that of the analyser can be expressed as follows:KITAGAWA, KOYAMA AND TAKEUCHI 823 .. * . (3) 111 oc I,e-"bLbJp'(k)dk . . .. .. S As the integral term is unity, If we take a ratio of Il to 111, defined as T,, then T , cc (NLB)2Sp'(k)F(k,H)2dk . . . . .. ' * (5) S Therefore, under the condition of a constant strength magnetic field and a definite line profile of the source radiation, the corrected term T , becomes and hence 2/Fc K N L B .. .. . . . . * * (7) Thus, the square root of the corrected term is proportional only to the number of atoms in the light beam and is independent of the presence of background components and variation in the source intensity.Experimental TO embody the principle stated above in a rapidly responding system, two kinds of optics were tentatively constructed and compared. One of them, illustrated in Fig. 1, incorporates a Glan-type prism with an escaping window on the lateral face (Karl Lambrecht Co. Ltd.) as the analyser and two photomultiplier tubes behind the exit slit of the monochromator. The other optical elements, the electromagnet, the electrothermal atomiser and the polariser are the same as those used in the previous experiment.l The analyser passes the ordinary ray along the optical axis and the extraordinary ray at an angle of about 24" to the axis in the ultraviolet region.The polariser and the analyser were set in such a fashion that the ordinary ray coincided with I , and the extraordinary ray with 111. The separated radiations were detected by separate photomultiplier tubes (Hamamatsu TV Co. Ltd., R928). The source radiation was chopped at a frequency of 285 Hz by a mechanical chopper (Varian- Techtron Co. Ltd.) located between the source and the polariser. The resulting photo- currents were fed to the electronic system, the circuit diagram of which is shown in Fig. 2. Each of the photocurrents was converted into voltage and amplified through an ax. amplifier. The coupling capacitor between the photomultiplier tube and the amplifier D . Fig. 1. Schematic diagram of static system for correction of energy loss.A, EDL and microwave cavity; B, chopper; C, lens and polariser; D, electro- magnet and electrothermal atomiser ; E, monochromator; F, analyser (Glan prism with escape window) ; and G, photomultiplier tube (H'I'V K928).824 KITAGAWA et al. : CORRECTION SYSTEM FOR SPECTROSCOPIC Analyst, Vol. 104 I- - 1 I’ . -. & l i i . Off-set Operation :.vj, V ~ ~ i T $ - ~ ; R e c o r d e r __o J s R1 R2 Fig. 2. Circuit diagram of electronic system for the static system. removes the d.c. voltage resulting from the black-body radiation emitted by the electro- thermal atomiser. The operational amplifier employed was a bi-FET IC, LF356 (Intersil Co. Ltd.), the high input impedance of which allowed it to be the load of the photomultiplier tube.Because of the large difference in energy between I , and I,!, the gain factor for the former was set higher (100) than that for the latter (10). Following a buffer amplifier, each of the a.c. signals was fed through a band-pass filter tuned to a frequency of 285 Hz with a Q-factor of 10 and rectified by an ideal diode circuit. The resulting d.c. signals were operated upon by a programmable multifunction analogue module (Analog Devices Co. Ltd., Model 433 J) according to the equation where m = R2/(Rr + R2). Let V , GC I,, V , cc 111, m = 4 and Vy = 0-9 V for adjusting the over-all gain factor. Then, VOUt cx: d I ~ ~ , , and Vout oc NL. Therefore, if the strength of magnetic field and the profile of the absorption line are defined, the output voltage varies proportionally only with the number of atoms in the light beam.However, a problem arose C B A I U Fig. 3. Schematic diagram of time-sharing system for correction of energy loss. A, EDL and microwave cavity; B, modulation system; C, electromagnet and electro- thermal atomiser; D, analyser and lens ; E, monochromator; and F, photomultiplier tube (HTV R928). in that the monochromator partially depolarised the incident radiation. current even if the atomic vapour was absent. was inserted between the rectifier and the function module. voltage is modified as in equation (9) : This caused a bias To offset this current, an additional circuit Consequently, the output {September, 1979 DETERMINATION OF Cd USING THE ATOMIC FARADAY EFFECT 825 where k,Iil is the additional term for cancelling the voltage resulting from the depolarisation. When atoms are absent, the numerator is pre-set to zero.The alternative system is illustrated in Fig. 3, the details of its mechanical modulator are shown in Fig. 4 and the circuit diagrams of the gating and signal processing electronic circuits in Fig. 5 . The mechanical modulator consists of a disc rotor for chopping the source radiation and a holder for the polariser. The holder is rotated by a synchronous motor at 3600 rev min-l through a belt drive and the disc linked to it by gears. The correlation in time sharing is shown for the detected radiations veysus the rotational angle of the plane of polarisation of the polariser in Fig. 6. At positions a and c the rotor blinds the detector from the source radiation and only black-body radiation from the electrothermal atomiser falls on the detector.Synchronously, an analogue switch Sb is closed and the voltage corresponding to the black-body radiation is sampled and stored in the capacitor Cb. At the instant b, the source radiation is allowed to pass through an aperture in the rotor and reach the polariser. As the polariser and the analyser are in the crossed configuration at this moment, the detected energy is the magnetically rotated radiation, I,, plus the black-body radiation, Ib. The following operational amplifier subtracts the voltage Ib from this voltage, giving the output voltage as the net energy of the magneto-rotation. Concurrently, a switch SI operates and the voltage is sampled and held in the capacitor C,.At moment d, the rotor allows the source radiation to reach the polariser and the plane of polarisation of the polariser is now parallel to that of the analyser, so that the resulting voltage is the reference voltage, I , , , plus the voltage owing to the background radiation, I b . After subtracting Ib, the net voltage Ill is sampled and held in the capacitor CII by closing and opening the analogue switch S,,. The resulting signals I , and I , , are fed to the same function module as that used in the preceding system. The sequence of operations is repeated during the remaining half cycle. Fig. 4. Schematic diagram of modulator in the time- sharing system. E, EDL; L, small lamp; D, rotatory disc; PH, phototransistor; G, gear system; P, polariser; S, stripes for synchronisation; and M, motor.In the preliminary experiment, the triggering pulse for the synchronous operation of the analogue switches was generated by detecting radiation from a small lamp after passing through holes bored on the rotor. However, the stability of the pulses was not satisfactory, partially attributable to fluctuations in the gear system. To improve the stability, the radiation from the small lamp was detected by a phototransistor after reflection by stripes marked on a drum mounted on the axis of the polariser. Another pulse was taken to define the sequence of the operation of the analogue switches. As the photocurrent is significantly different in magnitude between the parallel and crossed configuration, the gain of the pre- amplifier is synchronously altered by switching the resistance in the feedback network.826 KITAGAWA et al.: CORRECTION SYSTEM: FOR SPECTROSCOPIC Analyst, Vol. 104 +5 v H -;a Pre-amplifier .v Blank hold Operation V, \/\I hold Subtract I -15 V U 48, B Q 741 21 Recorder Fig. 5. Circuit diagrams of (a) analogue and (b) digital electronic circuit for the time-sharing system. C, capacitor; P, switching pulse; and S, switch. Results andl Discussion The operating conditions were similar to those used in the previous work1 and were as follows: wavelength, Cd I, 228.8 nm; band-pass width of monochromator, 2.5 nm; drying temperature, 120 "C; atomisation temperature, 1800 "C; and strength of magnetic field, 4 kG. The two systems presented similar results from the standpoint of correction ability.However, in the first system, probably because of the depolarising property of the mono- chromator prism, a considerable leak of energy (about 2% of the incident radiation) was observed in the crossed configuration. This generated a bias photocurrent in the photo- multiplier tube intermediate in form between the continuous and pulse modes. The con- tinuous current was offset electrically according to equation (9) but the noise level on the base line was increased owing to the pulse current. The following results therefore relate mainly to the time-sharing system.September, 1979 Optimisation of Switching Pulses Fig. 6 shows a trace of the photocurrent (Vt) displayed on a cathode-ray oscilloscope (Iwasaki Tsushin Co. Ltd.) and synchronised with those of the switching pulses for the analogue gates of Sb, S, and St[.To offset the blank signal due to the black-body radiation, the switching pulses P b were set as close to P, and as possible; the time interval between the pulses of P b and P, or PI, was synchronised to be 1 ms. DETERMINATION OF Cd USING THE ATOMIC FARDAY EFFECT 827 a.l 0-l CI 0 - > p, v, n v b VI Pb 4 I , J-T-Tl c d 43 , I I I O0 90" 1 80° Rotational angle of the plane of polarisation of the polariser Fig. 6. Traces of photocurrent and pulses for analogue switching. V,, Photocurrent converted into voltage ; PL, switching pulse at crossed configuration ; VL, voltage sampled and held at crossed configuration ; PiI, switching pulse at parallel configuration; Vll, voltage sampled and held at parallel configuration ; Pb, switching pulse for sampling black-body radiation ; Vb, voltage of black-body radiation sampled and held; and P,, pulse for switching gain of pre- amplifier.Fig. 7 is the oscilloscope trace of the I , signal as a function of the width of the switching pulse. For greater widths, however, the noise level increased progressively. This increase may be attributed to the wider pulse sampling the transmitted radiation at moments when the polariser and analyser are incompletely crossed, and to some vibration in the mechanical system of modula- tion. Up to a 50-ps pulse width no variation in the energy of the stored signals was found when atomising 25 pg of cadmium at 1800 "C. Taking into account the slew rate Jf the operational amplifier, we determined the optimum pulse width to be 25 ps.Up to a width of 50 pus, no significant change in noise level was found. loops 1 5 0 ~ ~ 200~s Fig. 7. Oscilloscope trace of I, signal to show dependence of base-line noise on width of switching pulse with crossed polariser configuration. Effectiveness of Correction with Constant Optical Rotation that was stretched unidirectionally to produce optical activity. The corrective ability was estimated using 0.05 mm thick film of low-density polyethylene The film was mounted828 Analyst, vd. 104 between the pole pieces of the electromagnet and, by means of its optical activity, allowed a constant signal to be transmitted through the optical system. The microwave power supplied to the cadmium EDL was varied from 100 to 70 W.Fig. 8 shows traces of the reference signal, I,,, and of the computed signal, z/z, when cigarette smoke was introduced between the pole pieces. The peaks on the reference signal trace correspond to the loss of energy due to scattering by the smoke; these peaks, however, do not appear on the computed signal traces. KITAGAWA et d. : CORRECTION SYSTEM FOR SPECTROSCOPIC These results illustrate three important aspects for practical analysis : (1) the system does not respond to radiation scattered by non-atomic species; (2) the system corrects for the influence of changes in source intensity; (3) the system corrects for the reduction in incident and transmitted energy arising from background absorption. The noise on the corrected signal increased as the source intensity decreased.This effect is attributable to the decrease in the signal to noise ratio in the numerator and denominator of equation (9), due to the reduction in the radiation output of the lamp. 3 0 Time Fig. 8. Traces showing correction for energy loss Arrows and variation in intensity of source radiation. indicate the introduction of cigarette smoke. Correction in Electrothermal Atomisation Fig. 9 shows oscilloscope traces of the corrected response for pyrolysis of different amounts of starch a t the atomisation temperature of 1800 "C. As no ashing procedure was applied, a concentrated cloud of background species was generated in the electrothermal atomiser. In practice, the reference signal 11, was rapidly reduced as shown in Fig.9, but no background signal appeared on the dTc trace, except an increase in the noise level. Partly because of imperfection in the transient response of the electronic circuit for the static system, an errone- 0 0 Fig. 9. Oscilloscope traces of the corrected response for pyrolysis of different amounts of starch at the atomisation temperature of 1800 "C, showing the background scattering. 1, No starch; 2, 2 . 5 p g of starch; 3, 5 pg of starch; and 4, 10 pg of starch.September, 1979 829 ous signal equivalent to 5 pg of cadmium was generated by the correction system when the loss of the incident energy reached 99%. From this point of view, the “box-car” circuit employed for the time-sharing system gave better results than the combination of band-pass filter and rectifier employed in the two-detector system.Fig. 10 shows oscilloscope traces of the response of the system to 50 pg of cadmium in solutions to which starch was added in various concentrations to produce background absorption. After drying at 120 “C for 1 min, 5 p1 of the solution were atomised at 1800 “C without ashing. The traces of the reference signal, I,,, [Fig. 10 (c)] are a measure of the energy loss due to the resulting background component and the absorption by atomic cadmium. When starch was present in a concentration of 0.3%, only 5% of the incident radiation was transmitted by the smoke generated by the pyrolysis of starch. The base-line drift is assigned to the variation in the source intensity. The traces in Fig. 10 ( b ) are the uncorrected response corresponding to IL, and it is obvious that the resulting smoke reduces the transmitted intensity and causes the negative interference.In Fig. 10 (a) it can be seen that the use of the function .\/T, buffers the interference. DETERMINATION OF Cd USING THE ATOMIC FARADAY EFFECT Fig. 10. Correction of energy loss and variation in intensity of source radiation for electro- thermal atomisation. 1, 50 pg of cadmium; 2, 5 pg of starch added; 3, 10 pg of starch added; and 4, 15 p g of starch added. See text for explanation of (a), (b) and (4. Signal to Noise Ratio As the system involves a “box-car” circuit, the photocurrent that is sampled and held has a value corre- sponding to the moment for which the gate is open. Most of the noise included in the output voltage can be attributed to the pulses generated in the photomultiplier tube.When the radiation does not reach the photomultiplier tube, the pulses are assignable to thermal noise. When the number of photons is small, the photocurrent becomes a series of pulses. In this instance, the sample and hold circuit leads to significant variations in the peak response. In either instance, a short current pulse is held and converted into a longer voltage pulse. This leads to a disadvantage in that it is difficult to remove the noise by a circuit of low pass filter. The problem of the thermal noise can be reduced by cooling the photomultiplier tube. To resolve the problem of low light levels, it seems to be more promising to employ a modified photoelectron counting circuit of short cycle time.Another possibility lies in the enchancement of the source intensity for a short period. A possible There remains, however, a problem inherent in the time-sharing system.830 KITAGAWA, KOYAMA AND TAKEUCHI version of this is the use of the hollow-cathode lamp synchronously driven by a giant current pulse. An electronic system based on this technique is under investigation and will be dis- cussed elsewhere. On the other hand, although the system using two detectors did not give better results than the time-sharing system, an alternative configuration is possible that is potentially free from the bias current due to depolarisation. In this system, after separation by a Wollaston or Rochon prism located in front of the entrance slit of the monochromator, the ordinary and extraordinary rays can be dispersed and each detected by a photomultiplier tube. Conclusion The performance of the instrumental systems that we have developed for the determina- tion of cadmium using the Faraday effect has been shown to be sensitive and independent of changes in light-source intensity, scattering of radiation by smoke and absorption of radiation during atomisation of the sample matrix. The limitation of the system has been found to lie in the noise in the systems arising from low light levels and incomplete extinction when the planes of the polariser and analyser are crossed. This work was supported by a grant from the Ministry of Education of Japan. We are indebted to Dr. J. B. Dawson and Professor T. S. West for helpful discussions and advice. Reference 1. Kitagawa, K., Shigeyasu, T., and Takeuchi, T., Analyst, 1978, 103, 1021. Received January 3rd, 1979 Accepted February lst, 1979