Analyst, June, 1979, Vol. 104, pp. 491-504 491 Determination of Cadmium in Blood and Urine by Flame Atomic-fluorescence Spectrometry R. G. Michel, M. L. Hall and J. M. Ottaway Department of Pure and Applied Chemistry, University of Strathclyde, Cathedral Street, Glasgow, G1 1XL and G. S. Fell Department of Clinical Biochemistry, Royal Infirmary, Glasgow, G4 OSF The development is described of an atomic-fluorescence method for the determination of cadmium in blood and urine. The method involves only the direct aspiration of acidified urine or diluted and acidified blood into the flame. Calibration is achieved simply by using acidified aqueous standards and by the application of a pre-determined correction factor to account for changes in the uptake rate. The sensitivity, accuracy and precision are comparable to those given by most techniques that are currently in use for the determination of cadmium in biological materials. The simplicity of the method permits rapid analyses of large numbers of samples (more than 25 samples per hour) and is particularly useful for the survey of populations of people exposed to cadmium.The instrumentation used employs a two- source background correction system. This is essential for maximum accuracy and allows automatic correction for the scatter, which is the primary cause of inaccuracies in the atomic-fluorescence spectrometric determination of cadmium. Keywords : Flame atomic-fluorescence spectrometry ; cadmium determination ; blood analysis ; urine analysis The most commonly used method for the determination of cadmium in biological samples is currently flame atomic-absorption spectrometry.Other methods that have been used or that are under development include spectrophotometric methods involving extraction with diphenylthiocarbazone (dithizone),lS2 neutron a~tivation,~,~ atomic-absorption spectrometry using the Delves cup5s6 or tantalum boat7s8 for blood cadmium, atomic-absorption spectro- metry with electrothermal atomiser~~-~~ and electrochemical methods.12-14 Pulido et al.15 have used a long path length absorption cell to increase the sensitivity of the determination of cadmium in urine and serum by direct aspiration into a flame. Smith et aZ.16 used an electrically heated ceramic tube through which the sample, after dithizone extraction of blood, was aspirated into a hydrogen flame.The relative merits of some of these techniques have been discussed by Friberg et aZ.,17 Pierce et aZ.l8 and O’Laughlin et al.19 Most methods have the required sensitivity and specificity , and inter-laboratory s t u d i e ~ l ~ - ~ ~ are beginning to give indications of the accuracy and precision of these methods. Almost all of the reported procedures employ some form of chemical pre-treatment of the sample in order to remove interferences from the biological matrix and to concentrate the cadmium in the final solution for analysis. Examples in the literature include neutron- activation analysis4 and atomic-absorption s p e ~ t r o m e t r y , ~ ~ ~ ~ 7 - 1 ~ ~ ~ ~ - ~ ~ where it is necessary to employ acid digestion or dry ashing, often followed by chelation and extraction steps.An alternative is to use ion-exchange techniques20s25 to separate cadmium. Electrochemical methodsl2,l4 also usually involve destruction of the biological matrix by wet digestion or low-temperature ashing before analysis by the various versions of anodic-stripping voltam- metry. Preliminary results from this l a b ~ r a t o r y ~ ~ - ~ ~ have shown that flame atomic-fluorescence spectrometry has potential as a simple and rapid method for determinations of cadmium in blood and urine. This speed and simplicity are attributable to the high sensitivity of the atomic-fluorescence technique and the minimum sample pre-treatment required. In this paper we report the use of more sophisticated atomic-fluorescence instrumentation.Cadmium in urine is determined by direct aspiration into a separated air - acetylene flame and cadmium in blood by 1 + 4 dilution and then direct aspiration. Our previous r e s u l t s ~ ~ ~ ~ demonstrated492 MICHEL et al. : DETERMINATION OF CADMIUM IN BLOOD AND Analyst, VOZ. 104 less than satisfactory accuracy for the atomic-fluorescence determination of cadmium in urine. However, the improved instrumentation includes a two-source background correction facility, which ensures high accuracy when determining low (pg 1-l) levels of cadmium, AI f r Mechanical- chopper J I Experiment a1 P.M. tube Instrumentation The choice of components for atomic-fluorescence instrumentation has been reviewed recently by Winefordner31 and there are also a number of other review papers available329 that discuss atomic-fluorescence spectrometry.A schematic diagram of the instrumentation used here is shown in Fig. 1 and a list of components in Table I. Cadmium atomic fluorescence was excited using a microwave-excited electrodeless discharge lamp (EDL) that was the subject of two previous p~blications.3~~~~ I I Double Separated monochromator Fig. 1. Instrumentation. Choice of jlame Cadmium atomic-fluorescence signals have been shown36 to be greater in hydrogen-based flames than in acetylene-based flames. However, we found that a number of disadvantages arose from our use of hydrogen flames. Salt deposits on the burner head appeared much more rapidly when using the air - hydrogen flame rather than an air - acetylene flame.This caused unacceptable instability while aspirating urine and blood samples with their high solids contents. Further, despite the low background and slightly larger cadmium fluorescence signals in the hydrogen flame, the sensitivities obtained with acetylene and hydrogen flames were comparable when aspirating biological samples.37 This was a result of the greater scatter of excitation-source radiation by unvapourised sample particles in the hydrogen flame. This erroneous signal could be background corrected. However, noise levels associated with the scatter signal degraded the detection limits more in hydrogen than in acetylene flames.37 The flame background noise at 228.8 nm, the cadmium analytical wavelength, was five times smaller in the separated air - acetylene flame than in the same unseparated flame. This was a result of a reduction in the total flame background, upon separation, by a factor of 25-30.Separated flames have been discussed in more detail by various authors.3839 The nitrogen-separated air - acetylene flame supported on a circular capillary burner was therefore used throughout this work. It was noted that the reduction in noise upon flame separation was equivalent to the square root of the reduction in flame background. This indicated that the noise on the flame background was shot noise at this wavelength.June, 1979 URINE BY FLAME ATOMIC-FLUORESCENCE SPECTROMETRY TABLE I INSTRUMENTATION AND OPERATING CONDITIONS 493 Component Double monochromator .. Manufacturer Spex Industries, Metuchen, N.J., USA Operating conditions Spectral band pass 0.6 nm, grating 1200 grooves mm-l, blaze 300 nm, wavelength 228.8 nm Model No. 1672 Photomultiplier . . .. .. 9789 QB EMI, Electron Tube Division, Hayes, Middx. Photomultiplier housing . . PRI 400RF Products for Research, Danvers, Mass., USA EM1 Photomultiplier power supply Photon counting system . . Synchronous sampler . . .. PM 28A 6C1 5C21 High voltage, 1100 V Count time 1 or 10 s. Ortec Brookdeal, Bracknell, Berks. Synchronous sampler used for demodulation Pre-mix flame system . . .. Perkin-Elmer Nitrogen-separated air - acetylene flame operated under stoicheiometric conditions. Signals observed 30 mm above burner head Capillary burner and flame separator .. .. .. Chopper .. .. .. .. Broida Q-wave cavity . . ..Laboratory constructed Laboratory constructed Electromedical Supplies, Wan tage Modulation frequency 300 Hz 210L Rlk I11 Microwave generator . . .. (EDL) . . .. .. .. facility . . .. .. .. Electrodeless discharge lamp EDL temperature control Electromedical Supplies Laboratory constructed References 34 and 35 Laboratory constructed. (Fan and heater elec- tronics-surplus components) EM1 - Varian Ltd., Middx. Temperature of heater con- trolled by 0-240-V Variac v1x-300uv High-pressure xenon arc . . See text Xenon arc housing . . .. Xenon arc power supply . . Lenses . . .. .. .. R300-2 P5300-1 EM1 - Varian EM1 - Varian Thermal Syndicate Ltd., Tyne and Wear Ealing Beck Ltd., Watford Focal length 50 mm, diameter 50 mm Mirror . . .. .. .. Focal length 75 mm, diameter 75 mm Optical benches and optical mounts .. .. .. Mechanite Ealing Beck Choice of monochromator The monochromator chosen for the instrument was an f/4 double monochromator (Table I ) . A double monochromator was used in order to prevent stray light, resulting from wavelengths other than the analytical wavelength, from reaching the exit slit. By reducing the stray light the associated noise was reduced and the signal to noise ratio of the measure- ment improved. The magnitude of this improvement was a factor of three in the detection494 Analyst, VOl. 104 limit for cadmium in urine at 228.8 nm. At wavelengths other than 228.8 nm the magni- tude of the improvement in noise level varied. Stray light from the flame when aspirating biological samples is a result of the sum of the flame background emission and the emission from the biological matrix itself.This emission is both atomic, i.e., line, and molecular, i.e., broad band in character. The stray light problem in flame atomic-fluorescence spectro- metry is discussed in detail el~ewhere.~’ MICHEL et al.: DETERMINATION OF CADMIUM I N BLOOD AND Two-source background correction The background correction instrumentation that was used is shown in Fig. 2 and was based on that described by Rains et aL40 The mechanical chopper modulated the signal at a frequency of 300 Hz. The principle of operation of this atomic-fluorescence scatter correction system is identical with that discussed by Rains et al. and is also similar to the two- source systems used in atomic absorption.*l At omic-fluorescence signals were excited using a cadmium microwave-excited electrodeless discharge lamp (EDL) and the scatter was simulated by using a high-pressure xenon arc.The EDL and xenon arc scatter signals were balanced using a 2% aluminium solution in a similar fashion to that described by Rains et al. However, the intensity of the xenon arc was varied by using a series of metal gauze discs of different densities placed directly in front of the arc lamp. An alternative could be a series of neutral density filters. Each gauze disc was placed around the circum- ference of a wheel that could be rotated in front of the lamp. Fine control of the lamp intensity was obtained by using the power control on the xenon arc power supply. This method allows variation of intensity without changing the size of the image of the xenon arc in the flame.To match the EDL image and the xenon arc image at the flame the optical system was aligned to ensure that the centres OE each image were coincident and that both images were taller than the 10-mm slit height of the monochromator, wider than the flame and in the same horizontal plane as the slit. This alignment is simple to carry out and the different geometry of the two sources is not a problem if the above conditions are fulfilled. This procedure ensures detection of both scatter signals over the same region and depth of the flame and hence will give an accurate scatter correction. Electrodeless Front view \ Plane mirrors discharge lamp Mechanical chopper ,\~:qpl Xenon arc \ ’ ? 2 I I wheel Measurement ; : ; \\\ \* Fig.2. Chopper for background correction. Electronic components A photon counting system was used to monitor signals from The photon counter included a phase-sensitive detector driven the photomultiplier tube. by a photodiode-derivedJune, 19 79 URINE BY FLAME ATOMIC-FLUORESCENCE SPECTROMETRY 495 reference signal from the mechanical chopper. To reduce the cost of an atomic-fluorescence instrument it is feasible to use a conventional analogue lock-in amplifier instead of the photon counting system without any significant loss in performance or essential facilities. The background count (photomultiplier dark count plus stray light) of the instrument with the flame and sources off was typically 8 counts s-l whereas flame background was typically 130 counts s-l with a 0.5-nm spectral band pass at 228.8 nm.Scatter signals were of the same order of magnitude as the flame background (130-230 counts s-l). The noise at the detection limit was assumed to be the quadratic sum of the noise on the flame background and the noise on the scatter signal. This was measured by aspirating a blood, urine or aqueous solution and observing the total signal in the background channel of the phase-sensitive detector. When the continuum source is operating and the system balanced to correct for scatter this total background represents the sum of the flame and scatter backgrounds. The square root of this number was the noise figure used for calculating detection limits. This method of defining and measuring the detection limit is essentially the same as that used by Johnson et ~ 1 .~ ~ and the same assumption was made that white (shot) noise was dominant. This assumption was verified by measurements of the variation of noise with (a) flame background, (b) slit width, (c) analytical signal and ( d ) count time. The results confirmed that the limiting noise followed the square root relationship with each variable. Optical components optical benches with conventional mounting components. chopper had facilities for vertical and horizontal translations. that shown in Fig. 1, placed around the lens in front of the monochromator. mirror provides the normal double pass of source radiation through the flame. A count time of 1 s was used for routine analyses. All lenses, the mirror, mechanical chopper and light sources were mounted on triangular The burner and mechanical The only light baffle was The spherical Reagents content.Glassware was acidiwashed and rinsed with de-ionised water before use. Cadmium stock solution A stock solution of 2000 pg ml-l of cadmium in 0.04 M hydrochloric acid was prepared by dissolving a known amount of spectrographically pure cadmium in 10 ml of cadmium-free 11 M hydrochloric acid. Standard solutions of cadmium over the range 0.001-0.1 pg ml-1 were prepared daily in 0 . 0 4 ~ hydrochloric acid. Solutions of metals for the interference study were prepared from the AnalaR-grade chlorides. The 2% aluminium solution for balancing the scatter correction system was prepared from spectrographically pure aluminium wire dissolved in AnalaR-grade hydrochloric acid.BZood Standard disposable syringes and needles were used to take blood by venepuncture (5 ml). The blood was kept in plastic sample tubes containing anticoagulant (potassium EDTA or lithium heparin). This collection procedure was shown to be free from cadmium contamina- tion on this and on previous o c c a ~ i o n s . ~ ~ ~ ~ ~ No significant changes in cadmium concentration occurred when venous blood collected in this way was stored for more than 1 month either at 4 to 10 "C or at -10 to -20 "C. Development work on calibration was carried out using pooled blood from outdated blood bank samples. Urine Twenty-four hour urine samples were collected in plastic bottles, containing thymol as a bacteriostatic agent.Concentrated (1 1 M) hydrochloric acid was added dropwise to 25-ml portions to give a final concentration of 0.04 M hydrochloric acid. This adjustment required about 3 drops, or 0.3 ml of 11 M hydrochloric acid. The acidified portions showed no deterioration in cadmium content when stored in their 25-ml plastic specimen tubes at 4-10 "C for up to 4 weeks. All reagents were of the highest purity available and each batch was checked for cadmium High-purity de-ionised water30 was used for the preparation of all solutions.496 Blood sample preparation Blood (2ml) was taken from the thawed samples after ensuring thorough mixing and added to a clean 10-ml centrifuge tube. The sample was diluted to 10 ml with 2 ml of 0.2 M hydrochloric acid, 2 ml of 2.5% Triton X.(to ensure complete haemolysis) and 4 ml of de-ionised water. The diluted blood was then centrifuged (30s; 3000revmin-1) to remove cellular debris and aspirated directly into the flame. The 10-ml sample was sufficient for duplicate analyses. MICHEL et Ul. : DETERMINATION OF CADMIUM I N BLOOD AND Analyst, V d . 104 Urine sample preparation If acid had not been added at the collection stage it was added shortly before analysis. Otherwise no further sample preparation was necessary. The undiluted urine was then aspirated directly into the flame. Results and Discussion Cadmium Calibration Standards It was possible to use aqueous cadmium solutions to construct calibration graphs for the analysis of both blood and urine solutions prepared as described above. However, it was found necessary to acidify samples and standards with hydrochloric acid to a level above 0 .0 3 ~ . Fig. 3 shows that the commonly occurring enhancing effect of hydrochloric acid on trace-metal atomic absorption and fluorescence in flames was observed and that in aqueous solutions this effect became constant at acid concentrations greater than 0.01 M. However, when the same experiment was carried out in urine the enhancing effect did not stabilise until above 0.03 M hydrochloric acid. 450 250 I I I I I I , 0.01 0.03 0.05 Hydrochloric acid concentration/M Fig. 3. Effect of hydrochloric acid on cadmium fluorescence. A, Aqueous stan- dards, 4 pg 1-1 of cadmium; B, 4 pg 1-1 of cadmium in urine. The addition of hydrochloric acid was advantageous for the usual purposes of stabilising the concentration of metal ions in solution and for taking advantage of the enhancing effect of the acid.With aqueous calibration graphs no change in linear range was observed. The results of standard additions of cadmium to urine with and without the presence of 0.04 M hydrochloric acid show that without acid linearity extends to 100 pg l-l, whereas with acid linearity is the same as for aqueous standards and extends to 2000 pug 1-1 (Fig. 4). Further, the calibration graph with acid added was identical with the calibration graph prepared using aqueous standards 0.04 M in hydrochloric acid (Fig. 4), except that there remained a constant 4% depression of the cadmium signals in urine relative to aqueous solutions. This depression corresponded with a 4% reduction in the rate of uptake of urine into the spray chamber. A correction for this and a similar correction for blood is discussed below but this correction has been applied and incorporated into the results shown in Fig.4. The standard additions calibration graph obtained for the diluted blood samples prepared as described above was identical with graphs prepared using aqueous standards 0.04 M in hydrochloric acid (Fig. 4) except that there remained the constant (19%) depression of the cadmium signals in blood This enhancement also has an important effect on calibration graphs.Jzcne, 1979 URINE BY FLAME ATOMIC-FLUORESCENCE SPECTROMETRY 497 relative to aqueous solutions. This corresponded to a 19% reduction in rate of uptake caused by the high viscosity of diluted blood.The blood calibration graph was linear up to 2000 pg 1-1 of cadmium. 1 I I I I I I I I I 0 1 10 100 1000 Cadmium concentration/pg I-’ Fig. 4. Effect of hydrochloric acid on cadmium in urine calibra- A, Spiked urine, 0.04 M acid (identical with aqueous calibra- tion. tion, 0.04 M acid) ; B, spiked urine, no acid. Correction for Changes in Uptake Rate The ratio of the rate of uptake of aqueous standards to the rate of uptake of urine or blood was used as the arithmetical correction factor to be applied to analytical measurements of cadmium atomic fluorescence. These ratios were defined as follows: Uptake rate of water (U,) Uptake rate of blood ( U b ) = Blood correction factor u , / u b was typically in the range 1.2-1.3 and Uw/Uu was in the range 1.03-1.07.Twenty consecutive measurements obtained each week for 6 weeks gave an average weekly ratio of 1.23 for blood (120 different samples of normal blood) and 1.04 for urine (120 different samples of normal urine) with a standard deviation in these ratios of 3% for blood and 2% for urine. This predictable and long-term precise performance of measurements of u w / u b and U,/Uu allowed the use of a correction factor rather than a more time-consuming standard addition to each sample or the addition of glycerol to the standards.43 It is possible that widely differing concentrations of haemoglobin in blood could change these rates by more than the above standard deviations indicate. The range of haemoglobin values over which the blood correction factor of 1.23 is valid was not determined.If massive deviations in haemoglobin values, e.g., for anaemic people, do change the rates of uptake significantly such changes could be accounted for by prior knowledge of particular samples and the application of specific rate measurements or the use of the standard additions technique. The procedure used to obtain and apply the correction factor was simple. The rates were obtained by using a 10-ml measuring cylinder and stop-watch to measure the time taken for the spray chamber to consume a fixed volume of liquid. A measurement of rates of uptake to obtain the correction factor was made in duplicate before each batch of cadmium analyses in blood and urine. The signals obtained throughout the batch were then multi- plied by the appropriate ratio to obtain the figures that could be used to deduce the cadmium concentrations from the aqueous calibration graphs.The correlation between the cali- bration graph obtained using aqueous standards 0.04 M in hydrochloric acid and the graph obtained by standard additions incorporating the above correction factors was excellent for498 MICHEL d.: DETERMINATION OF CADMIUM IN BLOOD AND Analyst, Vd. 104 both blood (correlation coefficient 1.02) and urine (correlation coefficient 0.99). To compute these correlation coefficients 11 concentrations were chosen to cover the full linear range of each calibration graph. Repeated measurements over a period of several months have demonstrated the continued reliability of this method of correction.Correction for Scatter of Source Radiation The procedure for scatter correction was aut.omatic once the scatter signals caused by each source had been equalised in the following manner. When the EDL was operating under optimum conditions of temperature and microwave a 2% solution of aluminium was aspirated into the separated air - acetylene flame. This solution gave a large signal caused by the scatter of EDL radiation off aluminium salt particles in the flame. The radiation from the xenon arc, irradiating the flame 180" out of phase with the EDL and scattered by the aluminium particles gave a similar large signal that was subtracted by the phase-sensitive detector from the EDL scatter signal. The intensity of the xenon arc was adjusted until zero signal was obtained at the output of the photon counter. After this balance had been achieved the aspiration of the 2*& aluminium solution could be discontinued.The instrument would then correct automatically for scatter off blood and urine matrix particles in the flame. With the instrumentation described here the 2% aluminium solution scattered the EDL radiation to give a signal of, typically, 9000 counts s-l. Diluted blood and undiluted urine gave scatter signals in the range 130-230 counts s-l. The flame itself gave a scatter signal of about 25 counts s-l, i.e., after scatter correction the signal was zero to within the shot noise of the flame background while aspirating cie-ionised water. This observation verified that scatter off the flame gases and unvapourised water droplets does take place but its magnitude is small and equivalent to that of a cadmium fluorescence signal close to the detection limit.Two alternative solutions, sodium chloride and calcium hydrogen orthophosphate, were investigated to determine whether or not they gave the same results as the 2% aluminium solution. They gave similar results, i.e., the scatter correction for blood and urine turned out to be the same, as did the magnitude of the scakter signal while aspirating the concentrated balancing solution (approximately 9000 counts s-l) . The use of calcium orthophosphate was thought inadvisable in view of the possibility of PO molecular fluorescence at 228.8 nm.44 However, no PO fluorescence was observed, probably because the xenon arc was used at low power for background correction purposes.It was necessary to ensure that the EDL was stable before attempting to use the back- ground correction system because any instability caused an imbalance in the two source intensities. A 0.5-h warm-up time was found to be necessary in order to achieve sufficient stability and maximum fluorescence. Once this was obtained balance checks with 2% aluminium solutions demonstrated that the instrument was in balance and therefore sufficiently stable for 2-3 h operation. Rebalancing after this period took less than 1 min and was not therefore a serious problem. More precise temperature control is currently being developed to reduce the time required to stabilise the EDL and to maintain its output constant. This is because the stability of the fluorescence and scatter signals depends primarily on the operating temperature of the EDL.During operation the temperature has a tendency to increase slowly (approximately 2-3 "C h-l). This increase causes the total line width to increase owing to self-absorption and self-reversal and hence an increase in scatter and a simultaneous decrease in fluorescence are observed. A more precise temperature control system would prevent this temperature increase, stabilise both fluorescence and scatter and maintain the balance between the two sources. 'The magnitude of this effect is of the order of 6% in 2 h. No problems were encountered with the stability of the xenon arc. Interferences The effects of the inorganic ions Mg2+, Ca2+, Fez+, Na+ and K+ as chlorides and C1-, NO3-, PO4% and S042- as acids on the atomic fluorescence of cadmium were investigated for the two flames, air - hydrogen and nitrogen-separated air - acetylene.The ions were studied in the concentration range 10-5000 pg ml-l.June, 19 79 URINE BY FLAME ATOMIC-FLUORESCENCE SPECTROMETRY 499 Two methods were used to confirm whether or not an apparent enhancement interference was caused by scatter. Firstly, the signals from blank solutions, containing only the inter- fering ion and no cadmium, were subtracted from the observed signals from interference test solutions. Secondly, this result was checked by using the background correction facility, which automatically subtracts scatter and reveals the presence of other types of interference. The second method was the only one that could be used when looking at the interferences affecting cadmium in urine directly rather than cadmium in synthetic aqueous solutions.In the nitrogen-separated air - acetylene flame, without hydrochloric acid present, all ions enhanced the 4 pg 1-1 cadmium atomic-fluorescence signals. Hydrochloric acid itself gave the greatest enhancement of about lOOyo (Fig. 3). The remaining ions caused enhancements that gradually increased from low to high concentrations of interfering ion with an average maximum of 50% enhancement. For the ions NO3-, PO4-, Cl-, Na+ and K+ scatter of source radiation was negligible (less than 5% of the total signal) at all concentrations of interfering ion (e.g., Na+, Fig. 5). Slight scatter was evident at the 5000 pg ml-l levels of Fe2+ and SO,2- (e.g., Fez+, Fig.6). Scatter was serious for Ca2+ and Mg2+ at all concentra- tions and especially at high concentrations (e.g., Mg2+, Fig. 7). $ 1 I , I 0 I 0 100 1000 Concentration of sodium in aqueous soIution/pg m1-l Fig. 6 . Effect of sodium on cadmium fluorescence. A, 4 p g 1-1 of cadmium, 0.04 M hydrochloric acid; B, 4 pg 1-l of cadmium, no acid. In 0.04 M hydrochloric acid solution the cadmium fluorescence was enhanced by the acid as demonstrated in Figs. 3 and 4. None of the other ions investigated caused any change in this signal although Ca2+ and Mg2+ continued to cause large scatter signals (e.g., Mg2+, 0 10 100 1000 Concentration of iron in aqueous soIution/pg mI-’ Fig. 6. Effect of iron on cadmium fluorescence.0, .=, 4 pg 1-1 of cadmium in 0.04 M hydrochloric acid; 0, 0, 4 pg 1-1 cadmium, no acid; 0, =, no correction for scatter; 0, 0, scatter corrected.500 MICHEL et al.: DETERMINATION 0 1 7 CADMIUM IN BLOOD AND Analyst, VoZ. 104 I 0 10 100 1000 Concentration of magnesium in aqueous soIution/pg mI-’ Fig. 7. Effect of magnesium on cadmium fluorescence. 0,. ., 4 pg 1-1 of cadmium in 0.04 M hydrochloric acid; 0, 0, 4 pg 1-l in cadmium, no acid; 0, a, no correction for scatter; 0, 0, scatter corrected. Fig. 7), SO,” and Fe2+ scatter was significant only at high concentrations of interferent (e.g., Fe2+, Fig. 6) and little scatter (less than 5:/0) was observed for the remaining ions (e.g., Na+, Fig. 5). It appears, therefore, that the presence of the metal chlorides and inorganic acids causes various enhancement interferences.Hydrochloric acid releases cadmium from the effects of the other ions and enables it to be determined in hydrochloric acid medium without interference. In urine the releasing effect is not completely effective until the hydrochloric acid concentration is above 0.03 M i B shown in Fig. 3. Moreover, it was found that orthophosphoric, sulphuric and nitric acid!; behave in a similar way. Urine acidified with one of these acids could probably be anal.ysed successfully using aqueous calibration standards containing the appropriate acid. This was not investigated further because hydrochloric acid provided the best analytical sensitivity, i.e., it caused the greatest enhance- ment of the cadmium signal in urine.The presence of 0.04 M hydrochloric acid in the diluted blood samples also appeared to have a releasing effect on cadmium in blood. This effect was demonstrated by the coincidence of the calibration graph (corrected for rate of uptake) obtained by standard additions to blood with the calibration graph obtained using aqueous standards which were 0 . 0 4 ~ in hydrochloric acid. Although the air - hydrogen flame was not used extensively, the ions K+, Na+, Ca2+, C1- and P0,3- were investigated in order to compare the behaviour of this flame with that of the acetylene flame. Similar results were obtained except that interferences in unacidified solutions were greater in magnitude with a maximum enhancement of 100%. Hydrochloric acid had a similar releasing effect to that found in the separated air - acetylene flame.Scatter was high for calcium solutions but little scatter (less than 5%) was observed for the remaining ions. Calcium scatter signals in the hydrogen flames were always three to five times greaterJune, 1979 URINE BY FLAME ATOMIC-FLUORESCENCE SPECTROMETRY 501 than the scatter signals in the separated air - acetylene flame; this was an example of the inadvisability of using the air - hydrogen flame for heavy matrices. A large scatter signal carries a noise component that degrades precision and detection limits. Results for Blood and Urine Analyses Table I1 summarises the results for the determination of cadmium in the blood and urine of a number of different groups of people. Blood and urine was taken from samples sub- mitted for analyses as part of health checks by Employment Medical Advisors.The samples that constituted the reference populations were obtained from hospital patients who were known to be unexposed to cadmium. The exposed groups of workers were coppersmiths engaged in brazing operations and men engaged in the shipbreaking industry. All these industrial workers showed increases in blood and urine cadmium relative to the reference population indicating evidence of recent exposure to cadmium. However, statistical comparison of results would require details as to the smoking habits of each of the populations studied as cigarette smoking can cause moderate elevations of blood cadmium. Samples analysed by standard additions gave agreement with samples analysed as described above, using aqueous standards.The average level of blood cadmium in the reference population, 3.1 pg 1-1 (27.6 nmoll-I), was similar to the control value of 35.6 nmol 1-1 reported by Cernik and Sayers9 and to other published values17 for non-exposed subjects. The mode (most frequently occurring value) of the reference population was lower than the average or mean value and indicated a non-normal distribution in the population. The average level of urine cadmium in the reference population, 0.5 pgl-1, was close to published valued7 and again the mode was lower than the mean. It was possible to measure the magnitude of the scatter signals associated with the atomic fluorescence signals by analysing blood and urine with and without use of the two-source background correction.For a batch of 54 blood samples from cadmium-exposed workers it was found that the average magnitude of the scatter signal was equivalent to 5.7 pg 1-1 of cadmium in the original undiluted blood with a standard deviation of 3.3 pg 1-1 (1.14 pg 1-1 and 0.66 pg l-l, as measured by the instrument, for diluted blood). Similarly, 25 samples of urine gave an average of 2.0 pg 1-1 and a standard deviation of 1.3 pg 1-l. To verify that the two-source system gave an accurate correction for scatter the alternative correction procedure described by Haarsma et aZ.45 was also applied. The Haarsma method is not rapid or automatic, but it is useful for verification purposes. It is based on the principle that the different dependence of EDL and fluorescence intensity on temperature permits a correction for scattering to be based on measurements at two EDL temperatures.Results using the Haarsma correction were in good agreement with the two-source method. TABLE I1 SUMMARY OF RESULTS OF BLOOD AND URINE ANALYSES Number in sample . . .. .. Standard deviation/pg 1-1 . . .. Rangelpg 1-1 . . .. .. .. Class interval of modelpg 1-1 Mean cadmium concentration/pg 1-1 .. - Reference population 100 3.1 1.5 1.1-6.4 1.4-1.7 Blood Copper- smiths* 20 10.4 4.8 3.3-20 6-6.9 - * Blood and urine samples were not from the same workers. 1 Ship- breakers 20 7.3 1.9 4-11.8 6-6.9 Urine w Reference Copper- population smiths* 0.5 7.2 0.4 6.8 20 20 0.2-1.5 2-26.4 0.2-0.29 2-2.9 It was previously reported from this l a b o r a t ~ r y ~ , ~ ~ that the level of urine cadmium in a reference population was 7 & 2 pg 1-1, which was higher than recent published values and higher than some results obtained on the same urine samples during inter-laboratory com- parisons.The value reported here, 0.5 pg 1-1, is in closer agreement with published values and demonstrates the improvement brought about by using the atomic-fluorescence instru- ment described above. The figures for the magnitude of the scatter correction show that502 MICHEL et d. : DETERMINATION OF CADMIUM I N BLOOD AND Analyst, 'vd. 104 the increased accuracy can be attributed primarily to the use of the scatter correction facility with a contribution resulting from the use of the separated air - acetylene flame rather than an air - hydrogen flame.The air - hydrogen flame tends to give scatter signals a factor of two to three times greater than air - acetylene when aspirating blood and urine solutions. An at omic-absorption spectrometric met hod with electrothermal at omisation developed in this laboratoryll was compared with the flame atomic-fluorescence method. The results of the analysis of the urine of a group of 30 exposed workers using both methods gave a correlation coefficient between those two sets of results that showed good agreement (Table 111). Further comparisons with more diverse methods of cadmium analysis are required to characterise fully the accuracy of the atomic-fluorescence method relative to other techniques. TABLE I11 DETERMINATION OF CADMIUM IN URINE USING TWO TECHNIQUES Flame atomic-fluorescence spectrometry (AFS) and atomic-absorption spectrometry using an electrothermal atomiser'l (AAS) .Urine samples taken from industrial workers some: of whom had possible exposure to cadmium in the workplace. Samples 26-29 were from four of our students. Sample number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Cadmium contentlpg 1-1 - AFS AAS 0.3 0.5 0.5 0.5 0.4 0.5 0.2 0.2 0.6 0.6 0.4 0.6 3.1 2.6 1.8 1.4 0.8 0.3 1.6 1.0 11.0 10.0 4.6 4.4 15.0 14.0 2.1 2.2 2.6 2.5 Sample number 16 17 18 19 20 21 22 23 24 25 26 27 28 29 Cadmium contentlpg 1-1 -7 AFS AAS 7.9 8.8 1.5 1.3 1.9 1.5 0.6 0.2 8.4 8.5 0.3 0.7 0.2 0.4 3.6 3.8 3.6 3.2 0.3 0.4 0.3 0.2 0.2 0.2 0.3 0.2 0.3 0.9 .. . . 2.6 2.5 .. . . 3.6 3.6 .. .. 0.2-15 Mean cadmium concentration/ pg 1-l Standard deviationlhg 1-1 .. .. . . - Rangelpg 1-1 .. .. .. Correlation coefficient between the results of each technique . . .. .. .. .. .. 0.994 p <0.05 a t 95% confidence Regression equation .. .. .. . . A=mF + c (A = AAS, F = AFS) Gradient of regression equation . . .. . . 0.954 Intercept of regression equationlpg 1-1 . . . . 0.021 Some analytical figures of merit for typical analyses are shown in Table IV. The detection limit for blood is five times worse than that of urine because of the dilution required for blood. The noise on the scatter signal accounts for the difference in detection limit between aqueous standards and biological samples. Analyses of 15 aliquots each of a pooled blood and a pooled urine spiked with 20 pg 1-1 and 4pg1-1 of cadmium, respectively, were carried out to obtain the relative standard deviation within batch. The spiked, pooled blood and urine were then stored for about 3 months and an aliquot was analysed about once a week during routine analyses to obtain the between-batch figure (Table IV).The precisions and recoveries obtained were those normally expected of flame techniques. Conclusions The atomic-fluorescence method that has been described involves only direct aspiration of acidified urine or diluted and acidified blood into the flame. Calibration is achieved simply by using acidified aqueous standards and by the application of a pre-determined correction factor to account for changes in the uptake rate.June, 1979 URINE BY FLAME ATOMIC-FLUORESCENCE SPECTROMETRY 503 TABLE IV ANALYTICAL PERFORMANCE FOR THE DETERMINATION OF CADMIUM Precision as relative standard deviation (rsd)- Blood spiked with 20 pg 1-’ Urine spiked with 4 pg I-’ of Cd of Cd c A \ A Rsd, % Recovery, % Rsd, yo Recovery, %’ Within-batch (15 aliquots) ..2.8 101 2.0 101 Between-batch (9 aliquots) .. 9.5 103 10.0 102 Detection limitsipg l-l*- Aqueous Count timeis standards Blood Urine 1 0.2 1.4 0.3 10 0.07 0.5 0.1 * Spectral band pass was 0.5 nm. The sensitivity of the method is comparable to that of most techniques that are currently in use for the determination of cadmium in biological samples. The accuracy and precision are also satisfactory when compared with other techniques reported in the literature. Inter- laboratory studies need to be carried out to define the accuracy more completely. However, it is clear that background scatter correction is essential for accurate cadmium determinations when using flame atomic-fluorescence spectroscopy at the pg 1-1 level.This probably holds true for most elements, especially those with analytical lines in the ultraviolet region. The instrument described has a dynamic range for background correction that is more than adequate for all matrices. This is because of the use of the high-power second source. Moreover, the background correction will work throughout the visible and ultraviolet regions because the output of the xenon arc is maintained through them. The simplicity of the method permits rapid analyses of large numbers of samples (more than 25 samples h-l) and is particularly useful for the surveys of industrial workers currently being carried out in our laboratories.The instrumentation described has potential for the determination of other elements, although success depends upon the availability of suitable excitation sources. For the determinations requiring greater sensitivity electrodeless discharge lamps are often suitable and efforts are being made in our laboratory to develop well controlled methods for the preparation of these lamps.34~~5 In addition, it can be seen, from Fig. 1, that if the EDL is not switched on the xenon arc is positioned suitably to excite atomic fluorescence. The xenon arc, used at maximum power, can simultaneously excite the fluorescence of a number of elements.42 With this instrument these fluorescence signals can be observed simply by changing the wavelength.The potential of using the xenon arc for biological sample analysis is now being explored in our laboratories. The authors acknowledge the support of the Scottish Home and Health Department for the purchase of the major items of equipment used in this project and for the award of a Postdoctoral Fellowship (to support R.G.M.). They also thank the HSE, Employment Medical Advisory Service, for a maintenance grant in support of M.L.H. Practical and equipment support of this work has also been given by Pye Unicam Ltd. and this is gratefully acknow- ledged. 1. 2. 3. 4. 5. 6. 7. 8. References Saltzman, B. E., Analyt. Chem., 1953, 25, 493. Fernando, Q., and Freizer, H., in Kolthoff, I. M., and Elving, P.J., Editors, “Treatise on Analytical Jervis, R. E., Tiefenbach, B., and Chattopadhyay, A., Can. J . Chem., 1974, 52, 3008. Lieberman, K. W., and Kramer, H. H., Analyt. Chem., 1970, 42, 266. Delves, H. T., Analyst, 1977, 102, 403. Cernik, A. A., Atom. Absorption Newsl., 1973, 12, 163. Hauser, T. R., Hinners, T. A., and Kent, J. L., Analyt. Chem., 1972, 44, 1819. Lewis, S. C., Forney, A. B., and Forney, R. B., J . Forens. Sci., 1976, 21, 1. Chemistry,” Part 111, Volume 3, Interscience, New York, 1963, pp. 171-229.504 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. MICHEL, HALL, OTTAWAY AND FELL Cernik, A. A., and Sayers, M. H. P., BY. J . Ind. Med., 1975, 32, 155.Kjellstrom, T., Lind, B., Linnman, L., and Nordberg, G., Envir. Res., 1974, 8, 92. Gardiner, P. H. E., Fell, G. S., and Ottaway, J. M., Talanta, in the press. Maienthal, E. J., J . Ass. 08. Analyt. Chem., 1972, 55, 1109. Ben-Bassat, A. H. I., Blindermann, J . M., Solomon, A., and Wakshal, E., Analyt. Chem., 1975, 47, Valenta, P., Rutzel, H., Nurnberg, H. W., and Stoeppler, M., 2. Analyt. Chem., 1977, 285, 25. Pulido, P., Fuwa, K., and Vallee, B. L., Analyt. Biochem., 1966, 14, 393. Smith, T. J., Temple, A. R., and Reading, J. C., Clin. Toxicol., 1976, 9, 75. Friberg, L., Piscator, M., Nordberg, G., and Kjellstrom, T., “Cadmium in the Environment,” Second Edition, CRC Press, Cleveland, Ohio, 1974. Pierce, J. O., O’Laughlin, J. W., and Hemphill, D. D., in “Edited Proceedings of the First Inter- national Cadmium Conference, San Francisco, 31st January-2nd February, 1977,” Metal Bulletin Ltd., London, p.88. O’Laughlin, J. W., Hemphill, D. D., and Pierce, J. O., “Analytical Methodology for Cadmium in Biological Matter-A Critical Review December 1976,” International Lead Zinc Research Organi- zation, Inc., New York. Lauwerys, R., Buchet, J . P., Roels, H., Berlin, A., and Smeets, J., Clin. Chem., 1975, 21, 551. Kjellstrom, T., Tsuchiya, K., Tompkins, E., Takabatake, E., Lind, B., and Linnman, L., “Pro- ceedings of an International Symposium on Environmental Health, Paris, 1974,” Commission of the European Communities, EUR 5360, 1974, 4, 2197. Berman, E., Atom. Absorption Newsl., 1967, 6, 57. Westerlund-Helmerson, U., Atom. Absorption h’ewsl., 1970, 9, 133. Evans, W. H., Read, J. I., and Lucas, B. E., Analyst, 1978, 103, 680. Vens, M. D., and Lauwerys, R., Arch. Mal. Prof. Mkd. Trav., 1972, 33, 97. Hough, D. C., and Ottaway, J. M., Proc. SOC. Analyt. Chem., 1974, 11, 223. Scott, R., Mills, E. A., Fell, G. S., Hussein, F. E. R., Yates, A. J., Paterson, P. J., McKirdy, A., Ottaway, J . M., Fitzgerald-Finch, 0. P., Lamont, A., and Roxburgh, S., Lancet, 1976, 396. Fell, G. S., Hough, D. C., Hussein, F. E. R., i%nd Ottaway, J. M., Proc. Analyt. Div. Chem. Soc., 1976, 13, 271. Fell, G. S., Ottaway, J. M., Hussein, F. E. R., Michel, R. G., and Hall, M. L., in Brown, S. S., Editor, “Clinical Chemistry and Chemical Toxicology of Metals,” Elsevier North Holland, Amsterdam, 1977, p. 367. Fell, G. S., Ottaway, J . M., and Hussein, F. E. R., BY. J . Ind. Med., 1977, 34, 106. Winefordner, J. D., J . Chem. Educ., 1978, 55, 72. Browner, R. F., Analyst, 1974, 99, 617. West, T. S., Analyst, 1974, 99, 886. Michel, R. G,, Coleman, J., and Winefordner, J. D., Spectrochim. Acta, 1978, 33B, 195. Michel, R. G., Ottaway, J. M., Sneddon, J., and Fell, G. S., Analyst, 1978, 103, 1204. Bratzel, M. P., Dagnall, R. M., and Winefordner, J. D., Analyt. Chem., 1969, 41, 713 and 1527. Michel, R. G., Hall, M. L., Rowland, S. A. K., Sneddon, J., Ottaway, J. M., and Fell, G. S., Analyst, Aldous, K. M., Browner, R. F., Dagnall, R. M., and West, T. S., Analyt. Chem., 1970, 42, 939. Kirkbright, G. F., and Sargent, M., “Atomic Absorption and Fluorescence Spectroscopy,” Academic Rains, T. C., Epstein, M. S., and Menis, O., Analyt. Chem., 1974, 46, 207. Zander, A. T., Int. Lab., Jan/Feb, 1977, 15. Johnson, D. J., Plankey, F. W., and Winefordner, J . D., Analyt. Chem., 1975, 47, 1739. Butrimovitz, G. P., and Purdy, W. C., Analytica Chim. Acta, 1977, 94, 63. Haraguchi, H., Fowler, W. K., Johnson, D. J., and Winefordner, J. D., Spectrochim. Acta, 1976, Haarsma, J . P. S., Vlogtman, J., and Agterdenbos, J., Spectrochim. Acta, 1976, 31B, 129. 534. 1979, 104, 505. Press, New York, London, 1974. 32A, 1539. Received September 2.5512, 1978 Accepted November Zlst, 1978