Anal. Chem. 1980, 52, 1059-1064
carried out in dilute acid, but some recent studies with t h e I C P have been with solutions prepared in 10% hydrochloric acid (19). This might mask any natural AIR interferences. Work by Boumans a n d DeBoer (20) with dual ultrasonic nebulizers has shown effects which seem t o fit into the pattern of AIR interferences which we have observed with pneumatic nebulizers. By using two separate nebulizers, these workers were able to distinguish clearly between aerosol and plasma processes. Enhancements for zinc a t low concentrations, in binary mixtures containing potassium a t high concentrations ( u p t o 4000 Fg mL-') were found, and these were attributed t o processes occurring in the desolvation system. However, t h e zinc enhancement might well be explained as an AIR effect, a n d this would also account for t h e similarity in t h e interferences noted for ammonium, potassium, and cesium ions, which cannot otherwise readily be explained. Although these effects fit t h e pattern of AIR interferences, direct measurement of ionic redistribution in ultrasonically generated aersols will be necessary in order t o unequivocally establish its role in causing t h e observed enhancements.
CONCLUSIONS Aerosol ionic redistribution has been shown t o have t h e potential to cause significant interferences in analytical atomic spectrometry when pneumatic nebulizers are used for sample introduction. Its predicted influence in flame AAS is relatively small, d u e t o the nature of the operation of AAS nebulizers and the likely dominance of volatilization and ionization interferences in many practical analyses. In ICP spectrometry, on the other hand, AIR interferences may cause significant measurement errors. A particular sample matrix might not be anticipated on the basis of previous knowledge t o give rise t o interference problems due to volatilization or ionization effects in t h e plasma. However, an unexpected AIR interference may be active. Furthermore, changes in operating conditions for t h e nebulizer, or indeed a change in nebulizer, may cause t h e magnitude of the interference t o alter drastically. consequently, in solutions containing high concentrations of matrix elements, matching of sample and standard matrices is essential if highest accuracy
1059
is to be achieved. Nevertheless, it should be emphasized t h a t t h e magnitude of AIR interferences is very much a function of nebulizer design a n d operation. While significant interferences can be present when running a n ICP system within the normal range of operating conditions, it seems likely that the anticipation of a possible problem, combined with careful optimization of conditions t o minimize interference, should allow t h e full I C P analytical potential t o be realized. It is hoped t h a t a fuller parametric investigation of AIR interferences currently underway will lead t o guidelines for such optimization.
LITERATURE CITED (1) Aitken, J. Trans. Roy. Soc. Edinburgh 1881, 30, 337. (2) Boyce, S. G. Science 1951, 173, 620. (3) Judson, C. M.; Lerew, A. A.; Dixon, J. K.; Salley, D. ,J. J . Phys. Chem. 1953, 57,916. (4) Bloch, M. R.; Kaplan, D.; Kertes, V.; Schnerb, J. Nature(London)1986, 209, 396. (5) Bloch, R. M.; Luecke, W. J . Geophys. Res. 1972, 77,5100. (6) MacIntyre, F. Tellus 1970, 22,451. (7) Wilkniss. P. E.; Bressan, D. J. J . Geophys. Res. 1971, 76,736. (8) Wilkniss, P. E.; Bressan, D. J. J . Geophys. Res. 1972, 77,5307. (9) Glass, S. J.; Matteson. M. J. Tellus 1973, 25, 272. (10) Sugawara, K. Oceanogr. Mar. B i d . Ann. Rev. 1984, 3 , 59. (11) Barker, D. R.; Zeitlin, H. J . Geophys. Res. 1972, 77,5076. (12) Woodcock, A. H.; Duce, R. A.; Meyers, J. L. J . Atmos. Sci. 1971, 28, 1252. ~~~
(13) Blanchard, D. C. "Progress in Oceanography", M. Sears, Ed.; Pergamon Press: New York. 1963: DD 71-202. (14) Blanchard, D. C. "From Raindrops to Volcanoes", Doubleday and Co.: Garden City, N.J.. 1967; pp 46-127. (15) MacIntyre, F. J Geophys. Res. 1972, 77,5211. (16) Cresser. M. S.; Browner, R. F. Spectrochim. Acta. Part 6 1980, 35, 73. (17) Cresser. M. S.; Browner, R. F. Appl. Spectrosc. 1980, in press. (18) Novak, J W.; Browner, R. F. Anal. Chem. 1980, 52,576. (19) Larson, G. F.; Fassel. V. A.; Scott, R. H.; Kniselev. R. N. Anal Chem. 1975, 47,238. (20) Boumans, P. W. J. M.; De Boer, F. J Spectrochim. Acta, Par7 6 1976, 37. 365 (21) Lane, R. W. Ind. Eng. Chem. 1951, 43, 1312. (22) MacBain, J.; Swain, R. Proc. R . Soc. London, Part A 1938, 254,608. (23) Loeb, L. B. "Static Electrification"; Springer: Berlin. 1958; pp 61-80.
RECEIVED for review September 10,1979. Accepted February 29, 1980. This material is based upon work supported by the National Science Foundation under Grant No. C H E 77-07618.
Zeeman Effect in Flame Atomic Absorption Spectrometry K. G. Brodie" and
P. R. Liddell
Varian Techtron Ry. Limited, P.O. Box 222, Springvale, 3 777, Victoria, Australia
An electromagnet is used to apply a modulated transverse magnetic field to a 6 t m length burner. A fixed linear polarizer is used to eliminate the T component. Sixty-six elements are examined in the appropriate acetylene fuelled flame, using either nitrous oxide or air as the supporting gas. Zeeman sensitivity ratios range from 3 6 % to 94%. The accuracy of background correction for non-atomic absorption in the flame Is superior to conventional background correction systems. Detection limits for some elements are better due to the reduction in flame noise. However, the reflex nature of the analytical curve reduces the linear dynamic range and introduces the possibility of ambiguous results.
There has recently been considerable interest in the use of 0003-2700/80/0352-1059$01 .OO/O
the Zeeman effect for background correction in atomic absorption spectrometry. Since background absorption is a more serious problem with graphite furnace atomizers t h a n with flames, most of the work has been with furnaces ( I ) . However, there have been several studies using flames and a variety of configurations have been utilized. The magnetic field has been applied to the flame (2-6) or the light source (5, 7-9) and has been applied parallel ( 4 ) or perpendicular ( 2 . 3, 5-9) t o t h e light path. We have discussed the merits of the various approaches to Zeeman atomic absorption spectrometry (ZAAS) with graphite furnace atomizers ( I ) and selected the modulated longitudinal field on t h e furnace as the preferred approach. Application of the field to the furnace meant t h a t background correction was carried out a t exactly the same wavelength as the atomic absorption, special lamps were not required, a n d self-abC 1980 American Chemical Society
1060
ANALYTICAL CHEMISTRY, VOL. 52, NO. 7, JUNE 1980
1 I
I
/
'I-
I
Flgure 1. Schematic diagram showing the position of the magnet and burner relative to the light path
/
/ /
sorption fluctuations in the lamp could be corrected. Use of a modulated rather than a fixed field meant that better sensitivity was obtained for elements exhibiting the anomalous Zeeman effect, and use of a longitudinal rather than a transverse field meant that there was no need for a polarizer with its associated loss of light. A longitudinal field has been applied to the flame from a 1-cm Meker burner ( 4 ) b u t it is not practical t o use a longitudinal field with the 5-10 cm slot burners usually used in flame AAS. Therefore the configuration which we have chosen t o investigate is the modulated transverse magnetic field applied to t h e flame. If a fixed linear polarizer is used to eliminate the A component, this approach is equivalent to the modulated longitudinal field approach, except t h a t the base-line noise will be slightly higher due to the light loss a t the polarizer.
EXPERIMENTAL Spectrometer. A Varian AA-6 was used with the original IhI-6 indicator module replaced by a specially designed module. The indicator module provided a digital readout of the Zeeman absorbance. There were recorder outputs for both Zeeman and normal absorbance. Varian hollow cathode lamps were operated a t 200 Hz with a duty cycle of 20%. It was necessary to shield the photomultiplier tube from the adjacent magnetic field by constructing a soft iron housing in which the tube was enclosed. If this shielding was not provided, there was magnetic interference on the photomultiplier tube. A quartz pile-of-plates polarizer was inserted in the light path adjacent to the monochromator entrance slit. The polarizer consisted of six Suprasil plates of 0.5-mm thickness (Heraeus Quarzschmelze GmbH, West Germany), and the degree of polarization was about 80%. Transmission through the pile-of-plates polarizer was 40% at 193 nm and 50% or greater above 300 nm. Magnet. A C-shaped laminated core electromagnet was situated transversely about a 6-cm length titanium burner as shown schematically in Figure 1. The same burner was used for air and nitrous oxide supported acetylene flames. Two copper wire coils, each with 384 turns, were fitted about the pole pieces. The pole gap was 16 mm. To prevent overheating of the pole pieces and coils, water was passed through copper tubing which was cemented to the coils. The magnet power supply delivered a peak field strength of 0.71 T with a modulation frequency of 100 Hz. It was possible to vary continuously the field strength from about 0.15 T to its maximum. Field strength measurement has been previously described ( I ) . The waveform of the magnetic field was basically sinusoidal, modified to permit a measurement at zero field. The lamp pulses were synchronized t o coincide with peak and zero field. Reagents. All reagents were analytical grade. Aqueous solutions were used throughout, the water being once-distilled in a borosilicate glass vessel. For some elements, BDH Chemicals stock solutions for atomic spectroscopy (1 mg/mL) were diluted to provide the required standards. For other elements, solutions were prepared in the appropriate acid(s). Blank solutions were also prepared.
I
~
L p -
21
32
03 '.'ASPYET
C.
C F8E.J
35
08
'?I
-e5 " I
Figure 2. Relationship between R, and magnetic field for copper
RESULTS A N D DISCUSSION Zeeman Sensitivity Ratio. The theory of ZAAS has been covered ( 1 , 10). T h e Zeeman absorbance, AZ, is defined by AZ = log I H / Z where IHis the measured light intensity with the magnetic field on and I is the intensity with the field off. T h e normal absorbance, AN, is given by where Io is the primary light source intensity. T h e Zeeman sensitivity ratio, RZ, is defined by
RZ = AZ/AN Using an applied magnetic field of 0.67 T peak, RZhas been determined for all elements commonly analyzed by AA and these values are presented in Table I. The magnet was usually run a t 0.67 T rather than the maximum field of 0.71 T t o prevent overheating. The conditions used for the analysis were the standard conditions recommended for the Varian AA-6. T h e values of RZ were determined from solutions giving a normal absorbance between 0.1 and 0.4. T h e highest values of RZ found were 94% for S m and 93% for T e and Nd. T h e lowest values found were 39% for In and 3670 for Li. As the magnetic field increases, the splitting of the components increases. In general, this will cause RZ t o increase, and most elements show a monotonic increase in RZ with increasing field. This contrasts with the fixed transverse field configuration, where splitting of the K components results in poorer sensitivity a t higher fields ( 7 , 8). However, even with a modulated field, RZ may not increase monotonically with field if the atomic line exhibits hyperfine splitting or if several isotopes are present. In these cases, RZwill increase initially as the field increases and the u components of the absorption line move away from the emission line. As the field increases further, some of the Zeeman shifted hyperfine components of the absorption line may coincide with different components of the unshifted emission line, causing a decrease in RZ. This effect can be seen in the plot of RZagainst field for Cu a t 324.7 nm, shown in Figure 2. Similar curves were obtained with In and Hg and the same effect has been observed with Cu (11, 12) and Hg (23) in a fixed magnetic field. RZ is dependent on lamp current in a manner which reflects the changed shape of the emitted resonance line (due t o self-absorption broadening) as the lamp current is increased. Figure 3 shows the results for Zn a t 213.9 nm. T h e narrower half-width of the emitted resonance line with lower lamp current results in a higher value of RZ,making it desirable to use a low lamp current. However, this must be weighed against the higher base-line noise a t the lower current. T h e
ANALYTICAL CHEMISTRY, VOL. 5 2 , NO. 7, JUNE 1980
1061
Zeeman Sensitivity Ratios ( R z )
Table I.
element
lamp current, mA
wavelength, nm
3 5
328.1 309.3 193.7 242.8 249.8 553.6 234.9 223.1 422.7 228.8 240.7 357.9 852.1 324.8 327.4 421.2 400.8 459.1 248.3 294.4 368.4 265.2 307.3 253.7 410.4 303.9 208.9 766.5 550.1 670.8 336.0 285.2 279.5 313.3 589.0
3 As Au BU Ba Ben Bi Ca Cd
co Cr
cs cu
DY Ern Eun Fe Gaa Gd' Gea Hf a
Hg
Ho a
In Ir K Lan Li Lua Mg Mn M0 Na
n
4 15 10
R z ,R
element
80 75 78 72 53
NbQ Nda Ni osa Pb
81
20 8
3 3 5 5 20 3 3 15 10 10
5 4 25 5 10
48 62 84 77 78 91 48 44 70 89
Pd Pr Pt Rb Re ' Rh Ru
88 88 88
Si
76 88 87 92 67
3
15 5 20
85
5 20
91 82 36 88 58
39 88
D
10
3 5 10 5
Sb
sc Se Sm Sn Sr Ta a Tba
Te Ti TI Tm
u=
80
92 88
wavelength, nm
lamp current, mA
R Z , 7%
334.9 492.5 232.0 290.9 217.0 283.3 244.8 495.1 265.9 780.0 346.0 343.5 349.9 217.6 391.2 196.0 251.6 429.7 235.5 286.3 460.7 271.5 432.7 214.3 364.3 276.8 371.8 358.5 318.5
20 10
255.1
20
75 88 93 77 69 80 92 75 81
410.2 398.8 213.9 3 60.1
10
88
5
91 60 86
80
93 75 87 58 78 78
5
20 El El
5 8 10
85
70 86 64 87 85 90 89 74
15 20 5 10
10 10 10
15 10 5 5 10 20
85
94 67 77 80
15 8
20 20 15
20 10
5
20
Determined in a nitrous oxide-acetylene flame. All other elements determined in air-acetvlene.
1
--.
i i
.r
-
L
~
3'
3, "IAGNET
,
0"
C FIEL:
,
,
0=
~
31
C"
le5 5
Figure 3. Relationship between R, and magnetic field for zinc at different lamp currents. (0)3 rnA, ( 0 )6 m A
negative value of RZ for Zn under conditions of low applied magnetic field and high lamp current is due to the shape of the self-absorbed resonance line from the hollow cathode lamp. As t h e absorption line splits, it exhibits a greater overlap of the emission line, resulting in higher absorption with the field on than the field off, and therefore a negative value of AZ. The effect of self-absorption in t h e lamp has also been observed
with Mg in the fixed field configuration (11). Some preliminary results with Ca and Cr indicate that RZ can vary slightly as flame composition is varied. I t is known that the atom population profile in the flame changes as the flame stoichiometry is changed (14). I t is postulated, therefore, t h a t atoms a t different heights in t h e flame are subject t o different magnetic field strengths, resulting in slight differences in R Z as flame stoichiometry is varied. Analytical Curves. Analytical curves for both the normal and Zeeman absorbance have been established for the following elements: Ag, Al, As, Bi, Ca, Cd, Co, Cr, Cu, Fe, Hg, In, Mg, Mn, Ni, P b , Sb, Se, S m , S n , a n d Zn. The magnetic field strength was 0.67 T in each case and t h e principal analytical lines, as listed in Table I, were used. Alternative analytical lines were also examined for some elements: Cu 327.4 nm, P b 283.3 nm, and S n 286.3 nm. The theory ( I , 10) predicts that if RZ is less than 1007~and there is a component of unabsorbable light, then the curves should reflex a t high concentrations. An example of this is seen in Figure 4 which shows both t h e normal absorbance and t h e Zeeman absorbance a t difference field strengths for Mn a t 279.5 nm. Since two different concentrations can give t h e same Zeeman absorbance, there is the possibility of an ambiguous result. Figure 4 also illustrates t h a t RZ increases and the dynamic range increases as the field strength becomes greater. With 0,7 T, RZ for Mn is 8270 (for a 5 pg/mL solution). For all the elements examined in this study. except Cu and In, RZ is greater at 0.7 T than at any lower field strength. T h e analytical curves for Cu a t 324.7 nm a t two field strengths of 0.35 T and 0.7 T are shown in Figure 5 demonstrating a greater ~
~
~
~
1062
ANALYTICAL CHEMISTRY, VOL. 5 2 , NO. 7, JUNE 1980 2
i-
//
38t
." -0,
Figure 4. Analytical working curves for manganese at different magnetic fields: (0) A,; ( 0 )A , at 0.7 T, (A)A , at 0.5 T, (B) A , at 0.3 T
r r: -t17Ri-12\
PS
-
Figure 6. Analytical working curves for samarium: (0)A,, ( 0 )A ,
0 w
dT
c
-
m
0 2
A
1
e
1
c
3
1
c
1
3
1
4
Figure 7 . Recorder traces showing A , and A , for arsenic (A) Flame off, (B) flame on, (C) aspirating distilled water, (D) aspirating 50 pg/mL arsenic solution
COh,EtvT-ATC
LQ
-
Flgure 5. Analytical working curves for copper at different magnetic fields: (0)A,; ( 0 )A , at 0.35 T, (A)A , at 0 7 T
dynamic range a t the lower field strength. A similar relationship applies t o In. In general, whenever RZ is less than l o o % , reflex curvature will occur. T h e absorbance and concentration a t which this occurs depends on t h e particular element. When RZ a p proaches loo%, the calibration for A Z closely follows AN, even a t high concentrations, e.g., S m a t 429.7 nm with RZ of 9470, as shown in Figure 6. Background Correction. One of t h e major features of ZAAS is the potential for accurate background correction. The accuracy of correction has previously been determined for furnace atomizers ( I ) but not for a flame system. The accuracy of correction was measured for non-atomic absorption generated by the flame itself, by the aspiration of distilled water, a n d by a 1 % KC1 solution. T h e correction for flame absorption was measured a t 200 n m using a hydrogen hollow cathode lamp and an air-
acetylene flame. T h e fuel flow was adjusted t o give t h e maximum signal on t h e normal channel. This was 0.8 absorbance. T h e corresponding signal on t h e Zeeman channel was 0.0004 absorbance. This represents a background correction error ((AZ/Ah-)X 100) of 0.05%. For comparison, identical measurements were made on a Varian AA-6B fitted with a hydrogen lamp background correction system. T h e error in this case was 3 % . The correction for distilled water and the KCl solution was measured using an As hollow cathode lamp a t 193.7 nm. T h e distilled water produced a normal absorbance of 0.11 and the KC1 solution an additional absorbance of 0.04. T h e Zeeman channel gave no measurable absorbance with either sample. When the same measurements were performed on the AA-6B, there was an error of 370 in the correction for distilled water and 2 % for the KC1 solution. T h e excellent background correction of the Zeeman system is clearly demonstrated in Figure 7, which shows the analysis of 50 pg/mL As in an air acetylene flame. The flame and the aspiration of distilled water cause a base-line shift in Ah whereas AZ remains steady. The Zeeman absorbance produced by the As atoms is clearly recorded. Base-Line Noise. T h e major contributions to base-line noise in a single beam flame atomic absorption instrument have been found to be photon noise, lamp flicker noise, and flame transmission noise (15). T h e Zeeman system should have the following effect on these noise types.
ANALYTICAL CHEMISTRY, VOL. 52, NO. 7 , JUNE 1980
1063
Table 11. Detection Limits, bg/mL element
As Ca Cr cu
cu
Figure 8. Relationship between peak-to-peak absorbance noise and wavelength with high I,: (0)A , flame on, ( A ) AN flame off, ( 0 )A, flame on; and with low I,: (0)A N flame on, (A)AN flame off, (M) A , flame on
(a) Photon Noise. T h e photon noise will be higher on A Z t h a n AN since AZ is calculated from the ratio of IHand I . In Figure 7, it can be seen t h a t t h e noise with the flame off is higher on A Z t h a n AN. (b) Lamp Flicker Noise. The Zeeman system automatically performs the same function as a conventional double-beam instrument in correcting for changes in lamp intensity. Therefore, any lamp fluctuations which occur a t a frequency lower t h a n t h e modulation frequency should be corrected. (c) Flame Transmission Noise. Since the Zeeman system corrects for flame absorption, any fluctuations in t h e transmission of light through the flame which occur a t a frequency lower t h a n the modulation frequency should be corrected. T h e relative magnitude of these three noise sources will determine whether t h e base-line noise is lower in AZ or AN. An experiment was performed using a hydrogen hollow cathode lamp in place of the normal hollow cathode lamp in order t o measure only non-atomic absorption. Both AN and AZ were then measured on a chart recorder with an airacetylene flame a n d distilled water aspirating, over the wavelength range from 195 t o 240 nm. In this region significant flame absorption occurs. T h e time constant on the recorder output was about 0.07 s. At each wavelength, the current t o the hydrogen lamp was set so t h a t the photomultiplier signal and supply voltage remained constant. This ensured t h a t t h e photon noise was constant over t h e wavelength range. Results obtained for the measurement of peak-to-peak noise in AN and AZ a t high and low source intensities are shown in Figure 8. T h e photomultiplier supply voltage was 225 and 350 V , respectively. T h e peak-to-peak noise level in Ar; in the absence of the flame is also shown. There was no measurable difference in the noise in A Z with t h e flame off. Figure 8 shows t h a t with a relatively high source intensity the peak-to-peak noise in A Z is lower than t h a t in AN a t wavelengths below about 220 nm. This is due to t h e dominance of flame noise in AN. With a relatively low source intensity, t h e peak-to-peak noise in AZ is lower than t h a t in AN only below about 200 nm. At higher wavelengths, AZwill always have a higher noise level, due to the dominance of photon noise in AN. Therefore, the capability for reduction in base-line noise exists with the Zeeman system. T h e extent of t h e noise reduction is finally limited by t h e source intensity. I t should be noted that the results in Figure 8 were obtained from chart recordings using a low time constant. With high time constants or long integration times, it would be expected t h a t t h e photon noise would show a greater reduction than the lamp flicker or flame transmission noise, since photon noise is “white” whereas lamp and flame noise have a strong
Fe K Li Mg Mn Ni Pb Sb Se Zn
wavelength, nm 328.1 309.3 193.7 422.7 357.9 324.8 327.4 248.3 776.5 670.8 285.2 279.5 232.0 217.0 217.6 196.0 213.9
Zeeman
normal
0.001 0.015
0.12
0.003 0.03 0.67
0.001.
0.002
0.004
0.008 0.003 0.004 0.005 0.002 0.002
0,003
0.007
0.006 0.001. 0.001.
0.0003 0.003 0.015 0.02 0.015 0.10 0.001
0.0004
0.004 0.008 0.03
0.028 1.2 0.0015
Al w a s determined in a nitrous oxide-acetylene flame. All other elements were determined in air-acetylene. l / f component (16). Under these conditions, the Zeeman system would show a greater noise advantage. Detection Limits. The absorbance in the Zeeman channel is always lower than in the normal channel, as indicated by the Zeeman sensitivity ratio in Table I. Therefore, Zeeman detection limits will match t h e normal detection limits only if t h e noise in A Z is lower than the noise in AN. T h e magnitudes of the photon, lamp, and flame noise in t h e normal channel will determine whether this is possible. Table I1 shows some detection limits measured on both the Zeeman and normal channels using a 10-s integration time and lamp currents and spectral bandwidths recommended for t h e Varian AA-6. T h e detection limit is the concentration giving a reading equal to twice the standard deviation. T h e values in Table I1 show t h a t when flame noise is high (e.g., As, Se, Sb), the Zeeman channel yields a better detection limit t h a n t h e normal channel. T h e Zeeman detection limits are also better for those elements where the light source is intense a n d the photon noise is low compared with the lamp flicker noise (e.g., Ag, Al, Ca, Cr, K). Cu a n d Mg aIe also in this category but the low values of RZ cancel out the reduced noise. Elements which have a better deteciion limit in t h e normal channel are those where the lamp intensity is relatively low and photon noise is dominant (e.g., Fe, Ni). T h e major advantages of this system are t h e continuous correction for all non-atomic absorption as well as monitoring of source lamp intensity to provide effective double-beaming with a single beam optical system. The ability of the Zeeman system t o reduce flame absorption noise a t low wavelengths and consequently to improve detection limits has been demonstrated.
ACKNOWLEDGMENT We are greatly indebted to the late A. S. Pearl for introducing us to this subject and for his encouragement and advice. We also thank J. A. Chidzey for designing the processing electronics and the magnet.
LITERATURE CITED Liddell. P. R.; Brodie, K. G . Anal. Chem., in press. Parker. C.;Pearl, A. Australian Patent 474204, filed .January 5, 1971. Uchida, Y.: Hattori, S. Oyo BuLsuri1975, 4 4 , 852-7; Chem. Abstr. 1976, 8 4 , 11953~. Otruba, V.; Jambor, J.; Horak, J.; Sommer. L., Scr. Fac. Sci. Nat. Uni. Bmo. 1976, 6 , 1-16. Uchida, Y.; Hattori, S. Bunko Kenkyu 1977, 26, 266-71; Chem. Abstr. 1978, 88, 163211a. Otruba, V.; Jambor, J.; Komarek, J.; Horak, J.; Somrner, L. Anal. Chim. Acta 1976, 101, 367-74. Stephens, R.; Ryan D. E. Talanta 1975, 22, 659-62.
1064
Anal. Chem. 1980, 52, 1064-1067
(8) Koizumi, H.; Yasuda, K. Spectrochim. Acta, Part 81976, 3 1 , 237-55. (9) Stephens, R . Manta 1978, 25, 435-40. (10) de Loos-Vollebregt, M. T. C . ; de Gaian. L. Spectrochim. Acta, Part B 1978, 33, 495-512. (11) Grassam, E.; Dawson, J. B.; Ellis, D. J. Ana/yst(London) 1977, 102,
804-18. (12) Koizumi, H; Katayama, M. Phys. L e t t A 1977, 6 4 , 285-6. (13) Koizumi, H.; Katayama, M. Phys. L e t t A 1977, 63, 233-4.
(14) Rann, C . S.; Hambly, A. N. Anal. Chem. 1965, 37, 879-84. (15) Liddeil, P. R . Anal. Chem. 1976, 48, 1931-33. (16) Bower, N. W.; Ingle, J. D., Jr. Anal. Chem. 1977, 4 9 , 574-9,
RECEIVED for review January 23, 1980. Accepted March 13, 1980.
Laser Atomic Absorption Spectrometry for Histochemistry Kimiaki Sumino,
Ryoji Yamamoto, Fumikazu Hatayama, and Shoji Kitamura
Department of Public Health, Faculty of Medicine, Kobe University, Ikuta-ku, Kobe 650, Japan
Harumasa Itoh Nihon Denshi (JEOL) Co., Nakagami, Akishima, Tokyo 196, Japan
A flameless atomic absorption spectrometer with an optical microscope and laser oscillator for determination of metals in various materials in very small target areas has been developed. A microfield of the sample is irradiated by a narrow laser beam under microscopic survey and the element in this area is quickly atomized. The subject element in the vapor is detected by flameless atomic absorption spectrometry and measured by comparing with known amounts of the standard material. I n one application for histochemistry, localization of cadmium in human kidney cortex was quantitatively examined under optical microscopic observation by this instrument. After the method for making standard preparations was established, relatively higher amounts of cadmium were found in the proximal tubules in the cortex and the lesser amounts in distal tubules and the glomerulus.
A microanalytical method for t h e determination of trace metals in desired minute portions of samples, under microscopic survey, has been developed utilizing a flameless atomic absorption technique with a laser beam as an atomization device. The instrument is a modification of a commercial laser microprobe, which can vaporize a sample by laser energy and discharge of spark electrodes, and determine trace metals using emission spectrometry. X-ray fluorescence, microprobe, or ion beams will undoubtedly be used for histochemical analysis of heavy metals in t h e future, b u t no report has yet been found in this field, except the use of the laser microprobe mass analyzer (LAMMA). Microanalysis by LAMMA has been reported in freeze-dried sections of rat kidney (tubulus cells) a t 1 pm in diameter and 1 to 2 pm in thickness, but only light metals like N a , K, a n d Ca were detected ( I ) . Other reports on microanalysis have been found in the field of histochemistry by using the laser microprobe with emission spectrometer (2-4), but only one set of quantitative data was reported on t h e analysis of a heavy metal, cellular gold ( 4 ) . Emission spectrometry is available for the qualitative analysis of t h e components of samples. but its sensitivity, specificity, and quantitative accuracy are generally thought to be less than those of atomic absorption spectrometry. Furthermore, because atomic absorption spectrometry provides precise measurement, ease of use and simple pretreatment of samples, 0003-2700/80/0352-1064$01 OO/O
it has shown great usefulness in trace metal analysis in many fields, particularly in pollution analysis, especially since development of the flameless method. With atomic absorption spectrometry, the measurement of microsamples is possible by using a sampling cup or carbon tube, but direct search for and observation of desired microfields on a sample through a microscope has been completely impossible. A few attempts a t atomic absorption analysis using lasers have been made (5-7), b u t as reviewed by Hieftje e t al. (8), sensitivity remained mediocre, interferences were frequent, and precision was marginal. No microanalysis of a minute sample under direct survey by microscope has been reported. T h e new laser atomic absorption spectrometer with background correction system, therefore, has been developed t o take advantage of the good points of both the laser microprobe and t h e atomic absorption spectrometer, making new experimental techniques possible. T h e description of this apparatus and some basic data with regard to its use, sensitivity, and precision are described first. Thereafter, in one application, this instrument was used t o determine the cadmium within minute portions of t h e kidney cortex in Japanese autopsy cases.
EXPERIMENTAL Apparatus. The outline of the apparatus, manufactured by JEOL Co., in accordance with our instructions, is shown in block diagram in Figure 1. The laser is a neodymium glass laser of 6.6 mm rod diameter X 100 mm rod length, rearranged from JEOL laser microprobe JLM-200. The oscillation wavelength is 1.06 pm and the maximum power of the pulse is 2 J, with a water cooling system. The output of the laser can be selected at 0.5, 1.0, 1.5, and 2.0 J, and the diameter of the laser beam can also be controlled, at 0.5 to 7 mm by use of an iris diaphragm. The laser beam can be focused onto any selected areas in a sample by observation with a light microscope. A hollow cathode lamp for atomic absorption and D, lamp for the background correction are provided as light sources. Background Correctivn S y s t e m . The optical pathway constructed specially by JEOL for this system is shown in Figure 2. In the double-beam pathway adopted in this instrument, a dc lighting method was used rather than a pulse lighting method or chopper mirror, because it was considered that the simultaneous background correction was highly significant, in order to catch extremely high-speed phenomena caused by laser vaporization. Both beams become integrated into one by the partially coated mirror which is 5 cm x 7 cm with alternate slits of 0.5 cm, and pass through the atom cloud which is created by the laser shot. C 1980 American Chemical Society