Atomization of lead from whole blood using novel tungsten filaments in electrothermal atomic absorption spectrometry

Abstract

Several novel W-filament designs are investigated and compared as atomization devices for blood Pb measurements by atomic absorption spectrometry. Two designs (short and long wire filaments) are based on rigidly-wound W-filaments from Osram, while a third (etched filament) is based on lithographic imaging and photoetching technology. The filaments are positioned in an aluminum atomization cell that is mounted in the optical path of a compact AAS instrument. This instrument is designed specifically for blood lead testing using W-filament technology with a Pb hollow cathode lamp and incorporates a self-reversal background correction system. Whole blood is diluted 1 + 4 with a phosphate modifier, Triton X-100 and nitric acid, and 15 µL are deposited on the filament using a micropipette. A simple heating program dries, pyrolyzes and atomizes the sample. Integrated absorbance measurements are used for improved precision and detection limits. Matrix-matched Pb standards are necessary for blood lead calibration. Of the three filaments studied, the long wire filament proved most suitable, with a characteristic mass (m₀) of 200 pg, a method detection limit (3σ) of1–2 µg dL⁻¹ (0.05–0.10 µmol L⁻¹), and a lifetime of the order of 60–70 firings. Method accuracy is verified using blood lead reference materials from the National Institute of Standards and Technology (SRM 955a,b), the Centers for Disease Control (BLLRS) and the New York State Department of Health (RMs 028, 040, and 043). For the long wire filament, blood lead results below 40 µg dL⁻¹ were mostly within ±1 µg dL⁻¹ of certified values, and within ±10% above 40 µg dL⁻¹; within-run precision was almost always better than ±10%. Additional validation is reported using proficiency test materials and human blood specimens. All blood lead results were within the acceptable limits established by regulatory authorities in the US.

---

Introduction

Childhood lead poisoning is a major public health problem in many countries, and blood lead is the assay most widely used to assess lead exposure. In 1991, the US Centers for Disease Control and Prevention (CDC)¹ reduced the blood lead concentration deemed as harmful to children from 25 to 10 µg dL⁻¹ (1.21 to 0.48 µmol L⁻¹). It is estimated that as many as 890 000 children aged between 1–5 years in the United States have blood lead levels of 10 µg dL⁻¹ or greater.² As a consequence, many states have implemented universal screening programs for identifying children at risk of lead poisoning, and the screening is usually accomplished using blood lead measurements.

Currently, the principal technology for blood lead measurement is electrothermal atomization atomic absorption spectrometry (ETAAS) using a graphite furnace (GFAAS) with platform atomization. Its high sensitivity, accuracy and low detection limits makes it suitable for screening purposes and clinical diagnosis of lead poisoning. However, its relatively large size, high cost and electrical power requirements limits its use to centralized clinical laboratories. In the US, such laboratories are regulated under the Clinical Laboratories Improvement Amendment of 1988 (CLIA '88)³ and must participate successfully in a blood lead proficiency testing (PT) program approved by the US Department of Health and Human Services.

Since 1993, the CDC has encouraged the development of new, innovative technologies for blood lead measurements suitable for use in point-of-care testing. The desirable characteristics of such technologies are: precision of <±1 µg dL⁻¹ to support a detection limit of 1–2 µg dL⁻¹; accuracy performance within CLIA '88 requirements (±4 µg dL⁻¹ for blood lead levels of less than 40 µg dL⁻¹; ±10% for blood lead levels of above 40 µg dL⁻¹); minimum specimen volume of around 200 µL to enable capillary blood lead measurements; and portable, low cost per test, easy-to-operate, and analysis time of less than 5 min per specimen.

Electrochemical technologies such as anodic stripping voltammetry (ASV) are well established for blood lead measurements.⁴ Recent improvements in electrochemical methods have led to new approaches for blood lead measurement based on disposable microarray electrodes^5,6 and screen-printed sensors.^7,8 Portable, hand-held instrumentation for blood lead testing based on ASV with disposable microarray electrodes are now being marketed commercially. However, electrochemical methods have the disadvantage of requiring Pb to be decomplexed from proteins and other macromolecules for reaction at the working electrode.

Others have proposed using a capacitively coupled microwave plasma (CCMP) coupled with atomic emission spectrometry (AES) for blood lead determinations.^9,10 In one approach, a 2 µL aliquot of whole blood is directly deposited on a tungsten filament, which becomes the plasma-supporting electrode. A microwave-induced plasma subsequently dries, pyrolyzes and atomizes the sample leading to Pb atomic emission signals.

The use of tungsten filaments as electrothermal atomization devices in AAS began as early as 1972 with the pioneering work of Williams and Piepmeier,¹¹ who employed a wound filament purged with Ar gas. Berndt and Schaldach¹² later proposed using inexpensive commercially available tungsten coils (150 W, 15 V) enclosed in a glass cell and mounted on the optical bench of a commercial AAS instrument in place of a graphite furnace or burner head. This technology was later transferred to other laboratories for further application development.¹³ Several groups have described the use of the tungsten filament atomizer for trace element determinations in biological materials,¹³ ground water and tap water.¹⁴ Other applications include barium in water,¹⁵ Al in hemodialysis fluid¹⁶ and cobalt in animal feces.¹⁷

The application of tungsten filament atomization to blood lead measurements has been independently proposed by several groups.^18,19 In our early work, lead was atomized directly from diluted blood using a 150 W tungsten filament operated with a current controlled 12 V power supply. Further refinements and additional details were later published²⁰ simultaneously with a report from another group²¹ who described using W-filament AAS to measure Pb in blood samples following protein precipitation to overcome matrix interferences. Other workers have also explored W-filament AAS for blood lead measurements, with some proposing pretreatment of blood with either microwave digestion^22,23 or preconcentration and extraction into methyl isobutyl ketone (MIBK) to overcome matrix interferences.²⁴

In this paper, we report advances in the measurement of blood lead using tungsten filaments of various designs mounted on a compact AAS instrument designed specifically for blood lead measurements using W-filament atomization, and operated with a self-reversal background correction scheme.

Experimental

Tungsten filaments

Three different W-filament designs were investigated as atomization platforms for AAS measurements. We obtained unmounted 150 W Osram HLX 64633 filaments (short wire filaments) directly from Osram Gmbh (München, Germany). The short wire filaments are mass produced for use as photographic projector halogen lamps, and consist of a 0.28 mm diameter tungsten wire wound into a flat coil of 4.8 mm in length. These coils can hold up to 20 µL of liquid sample deposited on the surface, the liquid occupying the space between the flattened layers of the coil (Fig. 1). The filaments were spot welded in-house to nickel posts mounted on a stainless-steel backplate, which was then installed in an atomization cell, machined from a 6.6 × 6.6 × 5.6 cm block of aluminum (Fig. 2). The cell was purged with argon gas containing 6% H₂.

Fig. 1 Schematic drawings of the Osram HLX short wire filament, the P-290 etched filament and the Osram H-1431 long wire filament. All dimensions are in millimeters.

Fig. 2 Exploded view (not to scale) of the atomization cell used in these studies.

A second filament (etched filament) was manufactured from a 0.18 mm thick tungsten sheet (Rembar, Inc., Dobbs Ferry, NY, USA) using photoetching technology performed by Buckbee-Mears, Inc. (St. Paul, MN, USA). The design is first developed into a mask that is used to transfer the coil images to the tungsten sheet using photo lithography. The filaments are then separated from the tungsten sheet using a chemical etching process. We explored two styles of etched filaments before settling on the final design (P-290) that was 22 mm in length, of which 5.6 mm comprised the middle segment that could accommodate up to 20 µL of sample (Fig. 1). The etched filament incorporated a novel three-segment arrangement, designed to improve the isothermality of the middle section of the filament on which the sample is deposited by isolating it from the cooler end segments. Initially, etched filaments were installed by Scientific Instrument Services (Ringoes, NJ, USA) on nickel posts mounted on a stainless-steel backplate. Later, we spot welded them in the laboratory, but with less precision than attainable through SIS, such that each new filament required minor optimization adjustments with respect to the light beam. The backplate was then installed in the atomization cell (Fig. 2).

A third filament (H-1431 long wire filament) was obtained unmounted from Osram, Inc., (Waldoboro, ME, USA). It was similar in design to the short wire filament, but was longer by a factor of about three (13 mm) and so could easily accommodate up to 50 µL of sample (Fig. 1). These long wire filaments were spot welded in-house to nickel posts and installed in the atomization cell as described above.

AAS instrumentation

We used a Model PBA-10 AAS instrument built by Exeter Analytical, Inc.† (North Chelmsford, MA, USA) to investigate filament atomization. This spectrometer is a compact unit (88 × 37 × 20 cm) designed specifically for blood lead measurements based on W-filament atomization AAS technology. It is equipped with a hollow cathode lamp, operated in self-reversal mode for simultaneous background correction, with a small photomultiplier tube mounted at the exit slit of a small monochromator with an effective band pass of 0.7 nm (Fig. 3). A grating (Carl Zeiss, Oberkochen, Germany) is used to isolate the 283.3 nm resonance line for Pb.

Fig. 3 Optical layout of the PBA-10 instrument; dashed lines indicate the optical path.

For self-reversal background correction, the hollow cathode lamp is pulsed at 100 Hz such that uncorrected AA measurements are taken during a 1.0 ms low current (9 mA) pulse, followed by a 0.1 ms high current (167 mA) pulse during which the background absorption (BKG) signals are read. The lamp then remains off for 8.9 ms. The complete timing sequence is given in Fig. 4. Data acquisition is operated at 100 Hz to synchronize with the pulsed hollow cathode lamp. Integrated absorbance measurements are used throughout these studies.

Fig. 4 Timing scheme of the pulsed hollow cathode lamp operating in self-reversal mode (not to scale). During each 10 ms cycle, uncorrected AA measurements are taken at low current pulse stage and BKG measurements taken at high current pulse stage.

A compact power-controlled internal 12 V dc supply was initially used to drive the etched filaments. Later it became necessary to use two different power supplies to drive the two wire filaments. We used a current-controlled 12 V dc external power supply to drive the short wire filaments through the drying, pyrolysis, atomization and cleaning stages. The long wire filaments were driven through drying, pyrolysis and atomization using the internal 12 V dc power supply, however, a 28 V dc external power supply was required to clean the filament at 75 W. A compact power controller was used to regulate the power delivered to the filament via a series of software setpoints. The setpoints are arbitrarily related to actual power, which is calculated empirically. Setpoint–power relationships were established for the three filament–power supply arrangements, and optimized power settings are given in Table 1. For the measurements in this paper, the instrument was interfaced to a personal computer, running software to control the heating program and data acquisition.

Table 1 Optimized heating programs for different W-filaments

Programstep	Power/W			Hold time/s	Gas flow rate/L min^−1
	HLX(Short)	P-290(Etched)	H-1431(Long)
Dry	0.8	0.8	0.8	200	1.0
Pyrolysis	2.6	2.8	2.6	30	1.0
Cool down	0	0	0	10	1.0
Atomization	38	40	45	3	1.0
Clean	65	73	75	60	1.0

---

For validation purposes, we used a Perkin-Elmer Model 4100ZL AAS instrument equipped with a transversely heated graphite furnace and a longitudinal Zeeman background correction system, and a well established method for blood lead.²⁵

Reagents and sample preparation

We used a well established modifier for both W-filament and furnace AAS blood lead work, consisting of 0.2% (m/v) NH₄H₂PO₄, 0.5% (v/v) Triton X-100 and 0.2% (v/v) HN₃.²⁵ Calibration Pb standards were prepared by serial dilution of a NIST-traceable 1000 mg L⁻¹ stock solution (Perkin Elmer, Inc., Norwalk, CT, USA). Aqueous Pb standards were prepared by diluting intermediate stock solutions (1 + 4) with a modifier. Matrix-matched Pb standards were prepared by diluting 100 µL of the intermediate Pb stock solution with 100 µL of "base blood" and 300 µL of modifier. Base blood is defined as that obtained from an undosed goat,²⁶ with a blood lead level determined to be below 1 µg dL⁻¹ by the established GFAAS method.²⁵

All reference materials, quality control and human blood specimens were diluted 1 + 4 with a modifier. The volume of diluted sample deposited on the filament was generally 15 µL, equivalent to 3 µL of whole blood except where indicated. A manual double-action micropipette (Finnpipette, Helsinki, Finland) was used to deposit samples on the filament. An alignment jig mounted on top of the atomization cell was necessary to guide the micropipette tip, to ensure reproducible sample deposition onto the coil (Fig. 5). The alignment jig, which was machined from an aluminum block, was designed with an inverted internal taper to minimize sample cross contamination, and a thread to provide fine adjustment of the tip to coil distance. This device was particularly important for depositing samples on the long wire filament.

Fig. 5 Schematic drawing of the alignment jig.

Certified reference materials and human blood samples

A variety of certified reference materials (CRMs) for blood lead were obtained for validation studies. Standard Reference Material (SRM) 955a and 955b were purchased from the National Institute of Standards and Technology (NIST), Gaithersburg, MD, USA. Additional reference materials were also available from the New York State (NYS) Department of Health (RMs #028, 040, 043) and from the CDC Blood Lead Laboratory Reference System (BLLRS # 1299, 2398, 1198). Blood lead proficiency test samples were available from the New York State PT program and from the Wisconsin State Laboratory of Hygiene (WSLH). Twenty-three human blood specimens were obtained for comparison purposes from our routine blood lead laboratory at the Wadsworth Center.

Results and discussion

Optimization studies

Chemical modifiers. In our previous work with the short filaments,²⁰ we successfully transferred our prior experience with the ammonium dihydrogen phosphate modifier from a graphite furnace to a W-filament atomizer for the blood lead analysis. In this work, we briefly examined other modifiers including Pd alone, and Pd in combination with Mg. We found that a Pd or Pd–Mg modifier distorted the etched coil after only 20–30 firings. Both relatively small (0.072 µg Pd + 0.15 µg Mg) and large (15 µg Pd + 24.6 µg Mg) amounts of Pd–Mg modifier had a noticeable pitting effect on the coil surface after only a few firings. We abandoned Pd–Mg and concluded that phosphate was a more suitable modifier for direct blood lead measurements yielding better sensitivity and lower detection limits. Our observations appear inconsistent with those of Bruhn et al.,²³ who reported that a Pd modifier provides better performance than phosphate modifiers for blood lead determinations by W-filament AAS. However, we note that Bruhn et al.²³ used blood that had been microwave-digested with HNO₃ and H₂O₂ before injection on the coil and so their samples were essentially free of the organic matrix. In our approach, we prefer to deposit whole blood that has been previously diluted with modifier directly onto the filament surface and proceed with the analysis. From our experience, we have found that whole blood diluted 1 + 4 with 0.2% (m/v) NH₄H₂PO₄, 0.5% (v/v) Triton X-100 and 0.2% (v/v) HNO₃ phosphate modifier works well enough to obtain clinically useful data. Studies are currently underway in our laboratory comparing the performance of a permanent Rh modifier in a graphite atomizer with its application in W-filament AAS.

Pyrolysis and atomization. We used a NYS reference material (Lot # 040) with a blood Pb concentration of 42.8 µg dL⁻¹, certified through interlaboratory validation studies,²⁶ for pyrolysis and atomization investigations. Pyro-lysis and atomization parameters were optimized for the three different filament designs (Table 1). Fig. 6 shows a typical pyrolysis and atomization study for the long wire filament. These curves are similar to those observed in furnace AAS except that, unlike furnace work, the calculated power is plotted rather than temperature, and on a log scale. In this work, we do not attempt to estimate the coil temperature from theory as others have done^14,27 because of the uncertainty involved, and because calculated power is a more reliable variable.

Fig. 6 Pyrolysis and atomization curves of the long wire filament using a NYS blood lead reference material (42.8 µg dL⁻¹). Bars indicate the range of the measurements (n = 3).

Coil position. We varied the coil position vertically with respect to the hollow cathode lamp beam by raising the filament into the optical path and observed changes in the sensitivity for Pb atomized from aqueous standards and blood. For blood matrices, the optimized coil position was found to block approximately 20% of the light beam. We chose not to block more than 20% of the light beam to avoid degradation of the detection limit. Although this effect was the same for all three coil designs, serious differences in sensitivity were observed between blood and aqueous matrices for the same coil position. This was a major factor that prevented system calibration based on aqueous Pb standards, and thus matrix-matched calibration became necessary.

Purge gas flow rate. We examined the effect of purge gas flow rate on sensitivity using an etched filament. Sensitivity remained largely stable as the purge gas flow rate was increased from 0.5 to 1.5 L min⁻¹ (Fig. 7). However, above 1.5 L min⁻¹, sensitivity decreased markedly. Based on these data, a purge gas flow rate of 1.0 L min⁻¹ was used throughout our studies.

Fig. 7 The effect of purge gas flow rate on sensitivity using an etched filament. A NYS blood lead reference material (42.8 µg dL⁻¹) was used for this study. Bars indicate the range of the measurements (n = 3).

In contrast, our previous experience with peak height measurements using a commercial AAS optical bench²⁰ showed improving sensitivity when flow rate was varied over roughly the same range. Similarly, Luccas et al¹⁶ reported that the aluminum peak height sensitivity almost doubled as the flow rate was increased from 0.4 to 1.4 L min⁻¹. Use of integrated absorbance measurements here is almost certainly responsible for avoiding the susceptibility to purge gas flow rate observed with peak height.

Sample loading. We investigated the maximum amount of blood that can be deposited on the three filaments without generating carbonaceous residues during the analysis. Such residues worsen precision and, once established, continue to accumulate, rapidly degrading coil lifetime. The amount of blood deposited depends on the injection volume and dilution factor. A 15 µL aliquot of 1 + 4 diluted blood, i.e., 3 µL of whole blood, was found to be optimum for all three filaments. Although the long wire filament can accommodate as much as 20 µL of 1 + 4 diluted blood, i.e., 4 µL of whole blood without build-up, positioning the sample at the center of the filament becomes much more critical, such that small amounts of the matrix can easily remain unpyrolyzed prior to atomization at the cooler ends of the filament. This overwhelms the self-reversal background correction system. Using the optimum 15 µL of 1 + 4 diluted blood requires 60 s of cleaning at 75 W for the etched and the long wire filaments, or 65 W for the short wire filament.

Atomization profiles. We compared the atomization of lead from blood using the various filament designs. A NYS blood lead reference material containing 42.8 µg dL⁻¹ Pb was used for this purpose. Typical atomization profiles are shown in Fig. 8. The background signal for the blood matrix on the HLX short wire filament [Fig. 8(a)] appears as two distinct components and its intensity is more than double that of the atomic signal, whereas for the P-290 etched filament [Fig. 8(b)] and the H-1431 long wire filaments [Fig. 8(c)], the background signal follows the atomic absorption signal with roughly the same intensity. The reason for the differences in background signals may be due to incomplete pyrolysis of the blood trapped at the cooler ends of the short wire filament. Interestingly, the integrated absorbance was only 0.0142 s (m₀ = 396 pg) on the short wire filament, compared to 0.0216 s (m₀ = 260 pg) for the etched filament and 0.0220 s (m₀ = 256 pg) for the long wire filament. The reason(s) for the poorer sensitivity of the short wire filament are not clear. In Fig. 8(c), the earlier appearance of Pb in the long wire atomization profile may simply be due to the higher atomization power used (45 W).

Fig. 8 Typical atomization profiles for the (a) HLX short wire filament, (b) P-290 etched filament and (c) H-1431 long wire filament using a NYS blood lead reference material (42.8 µg dL⁻¹).

Calibration and method validation

Calibration. We compared two approaches for calibrating various filament designs for blood lead measurements using (i) aqueous standards in the modifier and (ii) matrix-matched standards in the modifier. Fig. 9 compares aqueous (open circles) and matrix-matched (closed circles) calibration for the HLX short wire filament [Fig. 9(a)], P-290 etched filament [Fig. 9(b)] and H-1431 long wire filament [Fig. 9(c)]. For each filament studied, the sensitivity for aqueous Pb standards, expressed as the characteristic mass (m₀) is always poorer than for matrix-matched calibration standards. This loss of sensitivity may be due in part to a less than optimum coil position for aqueous standards. Examination of the r² statistic for the calibration curves indicates that matrix-matched standards provide a better fit to a simple linear regression model. The typical calibration range up to 120 µg L⁻¹, equivalent to a blood Pb concentration of 60 µg dL⁻¹, is within the limit of linearity for the EAI PBA-10 instrument and its self-reversal background correction system.

Fig. 9 Calibration curves obtained from matrix matched standards (•) and aqueous standards + modifier (o) using the (a) HLX short wire filament, (b) P-290 etched filament and (c) H-1431 long wire filament. Bars indicate the range of the three measurements. Characteristic masses (m₀) and r² statistics are labeled next to the calibration curves.

Analytical performance for various W-filaments. Typical figures of merit for various filament designs based on matrix matched calibration are given in Table 2. Sensitivity is defined as m₀ (pg) and method detection limits (3σ) are calculated from 10 replicate firings of a low concentration (6.7 µg dL⁻¹) blood lead reference material. Under a self-reversal background correction scheme, a loss of approximately 50% in sensitivity is expected. Indeed, the characteristic mass of the P-290 etched filament (200 pg) obtained from the EAI instrument is exactly twice that observed in previous studies (unpublished data) using a commercial AAS optical bench equipped with a D₂ continuum background correction system. A description of the commercial AAS arrangement (P-E 3110) adapted for W-filament atomization is given elsewhere.²⁰ Again, the short wire filament has poorer peak area sensitivity than either the etched or the long wire filament. Of the three filament designs investigated, the long wire filament consistently provided detection limits of the order of 1–2 µg dL⁻¹, which is necessary to support current clinical requirements for blood lead screening purposes. The linear dynamic range of the long wire filament is determined to be up to 500 µg L⁻¹, equivalent to a blood Pb concentration of 250 µg dL⁻¹ (Fig. 10), well beyond the typical clinical range for blood lead.

Fig. 10 Linear dynamic range ( 0–500 µg L⁻¹) of the W-filament AAS method using the long wire filament, m₀ = 304 pg.

Table 2 Figures of merit for different W-filaments

Filament	PE 3110 AAS			EAI Prototype PBA-10
	Optimized	Method detection limit/		Optimized	Method detection limit/		Linear range/µg L^−1
	m 0/pg	µg dL^−1	µmol dL^−1	m 0/pg	µg dL^−1	µmol dL^−1
a Data based on peak height measurements (Ap), reported in ref. 20.
HLX (Short)	26^a	2–3^a	0.10–0.15^a	400	4–5	0.19–0.24	—
P-290 (Etched)	100	2–3	0.10–0.15	200	2–3	0.10–0.15	—
H-1431 (Long)	—	—	—	200	1–2	0.05–0.10	0–500

---

Analysis of reference materials, PT samples and human blood specimens. A variety of CRMs and PT samples were analyzed for method validation purposes. Table 3 shows the results of analyses of NIST SRM 955a Lead in Blood and New York State blood-based RMs for the three filaments mounted on the EAI PBA-10 instrument. Five replicate measurements were made for each material, with the mean ± SD shown along with certified values. The definition of uncertainty in certified values varies among the RM producers; for the NIST SRM it is defined as a 95% statistical tolerance interval and for the NYS RM it is the between-lab SD from interlaboratory validation studies.²⁶ For the CDC BLLRS materials, the target values are established from in-house validation studies using either ICP-IDMS²⁸ or GFAAS.²⁵ Although all measured values for these materials on all three filaments were within the acceptable limits for accuracy, established under CLIA '88 for this assay (±4 µg dL⁻¹ for blood lead levels below 40 µg dL⁻¹; ±10% for blood lead levels above 40 µg dL⁻¹), the data for the H-1431 long wire filament were generally better with respect to both accuracy and precision. For example, the average precision (RSD) for the long filament is 5% compared to 12% for the etched filament and 13% for the short filament. The average bias for the long filament was −0.5 µg dL⁻¹ compared to −1.0 µg dL⁻¹ for the short filament. However, because the data are based on a relatively small number of measurements (n = 7), the differences observed may not be statistically significant. Nonetheless, further validation data for the long filament were obtained by analyzing additional reference materials from the NIST and CDC. Indeed, the H-1431 blood lead results below 40 µg dL⁻¹ were mostly within ±1 µg dL⁻¹ of certified values, and within ±10% above 40 µg dL⁻¹; within-run precision was almost always better than ±10%.

Table 3 Accuracy and precision of blood lead measurement by different W-filaments on EAI prototype

Reference materials	Certified value^a/µg dL^−1	HLX (Short)			P-290 (Etched)			H-1431 (Long)
		Value found^b/µg dL^−1	RSD(%)	Bias/µg dL^−1	Value found^b/µg dL^−1	RSD(%)	Bias/µg dL^−1	Value found^b/µg dL^−1	RSD(%)	Bias/µg dL^−1
a Definition of uncertainty in certified values varies among the RM producers, see text for details. b Uncertainty for measured values are ±SD (n = 5).
NIST SRM 955a-1	5.01 ± 0.09	3.9 ± 1.7	44	−1.1	4.7 ± 0.9	19	−0.3	4.0 ± 0.3	8	−1.0
NIST SRM 955a-2	13.53 ± 0.13	11.2 ± 0.6	5	−2.3	10.4 ± 3.1	30	−3.1	13.3 ± 0.7	5	−0.2
NIST SRM 955a-3	30.63 ± 0.32	27.1 ± 1.2	4	−3.5	30.9 ± 3.3	11	0.3	30.3 ± 0.5	2	−0.3
NIST SRM 955a-4	54.43 ± 0.38	52.3 ± 3.1	6	−2.1	56.5 ± 0.7	1	2.1	51.7 ± 1.1	2	−2.7
NYS RM Lot 028	6.7 ± 1.2	6.6 ± 0.7	11	−0.1	6.5 ± 1.1	17	−0.2	6.4 ± 0.7	11	−0.3
NYS RM Lot 043	17.7 ± 0.5	18.7 ± 3.5	19	0.9	18.5 ± 0.2	1	0.7	17.1 ± 1.2	7	−0.7
NYS RM Lot 040	42.8 ± 0.9	43.9 ± 0.8	2	1.1	43.5 ± 1.6	4	0.7	44.4 ± 0.2	0	1.6
NIST SRM 955b-1	4.04 ± 0.15	—	—	—	—	—	—	4.5 ± 0.6	12	0.5
NIST SRM 955b-2	10.30 ± 0.10	—	—	—	—	—	—	9.5 ± 0.3	3	−0.8
NIST SRM 955b-3	20.59 ± 0.21	—	—	—	—	—	—	19.8 ± 0.1	1	−0.8
NIST SRM 955b-4	39.36 ± 0.36	—	—	—	—	—	—	41.1 ± 1.2	3	1.8
CDC BLLRS 1198	45.67 ± 0.12	—	—	—	—	—	—	45.2 ± 1.3	13	−1.1
CDC BLLRS 2398	18.49 ± 0.04	—	—	—	—	—	—	15.5 ± 1.9	12	−3.0
CDC BLLRS 1299	5.58 ± 0.27	—	—	—	—	—	—	4.5 ± 0.6	3	−0.5

---

Blood lead PT samples from NYS and the WSLH were also analyzed with the long wire filament and are shown in Fig. 11 as a scatterplot [Fig. 11(a)] and a difference plot [Fig. 11(b)]. These PT samples were only used as reference materials; target values were obtained from the routine PT practice. Again all reported values were within PT program acceptability limits (dashed lines) as defined above under CLIA '88.

Fig. 11 Analysis of blood lead PT materials from New York State and from the WSLH using the PBA-10 instrument with the long wire filament: (a) a scatter plot of W-filament AAS values versus the target values; and (b) a difference plot of the W-filament AAS data versus the target values. Dashed lines indicate the PT program acceptability limits as defined under CLIA '88.

For clinical validation purposes, we compared data obtained from the long wire filament with the blood lead values from a well established furnace AAS method²⁵ for 23 human specimens. Results are shown in Fig. 12 as a scatterplot of W-filament AAS data versus furnace AAS. Agreement between the two techniques was within the ±4 µg dL⁻¹ limits as defined above and shown as dashed lines in Fig. 12.

Fig. 12 Scatterplot of 23 human blood lead values obtained from W-filament atomization using the long wire filament versus values from a well established GFAAS method.²⁵ Dashed lines show the US regulatory limits for accuracy under CLIA '88.

Coil lifetime. Coil lifetime varied considerably from as little as 30 firings to over 100 firings. Direct analysis of blood matrices is probably the single most important factor in coil lifetime because of the need to clean the filament at high temperatures to prevent carbonaceous build-up. One disadvantage to W-filament AAS is the tendency of some filaments to bend and warp after as little as 15 firings when analyzing blood matrices. This effect is most troublesome with the flat etched filament design, which is structurally the most fragile among the three designs tested. Warping for the etched filament was observed to begin after as few as 15–20 firings and rapidly worsened thereafter. Consequently, the coil position in the optical beam changes over time, causing sensitivity drift, and thus the instrument requires frequent re-calibration. The warping effect is unpredictable and the etched filament lifetimes vary from 30 to over 100 firings. Several attempts were made to correct the warping effect including investigating various annealing procedures for mounted etched filaments and exploring alternative mounting arrangements that could accommodate a limited amount of expansion and movement during the heating process. None of these approaches satisfactorily resolved the coil warping effect.

By contrast, the rigidly-wound Osram wire coils (both the short and long designs) showed better resistance to warping, perhaps because they are better able to tolerate the stresses produced under intense heating. Warping develops later in the coil lifetime compared to the etched filament. Typically Osram wire coil warping begins to form after 30–40 firings, but is not as severe as with the etched filament. Between the two Osram coils, the long wire filament is more flexible and has a lifetime of 60–70 firings compared to 40–50 firings for the short wire filament. As the long wire coil ages, the atomization peak height decreases while peak width increases, but the integrated absorbance remains virtually unchanged (Fig. 13). As with furnace AAS,²⁹ this is another good reason to use integrated absorbance measurements rather than peak height in W-filament AAS.

Fig. 13 Comparison of an atomization profile from a NYS blood lead reference material (42.8 µg dL⁻¹) on (a) a new long wire filament and (b) the same filament after 40 firings.

W-filament AAS blood lead methods

As stated by Welz,²⁹ integrated absorbances (peak area) are preferable to peak height to improve precision, accuracy and detection limits, and this is routine practice in furnace AAS. In previous reports,^{12–17,20,21,23} most workers including ourselves mounted a W-filament atomizer on the optical bench of a commercial AAS instrument equipped with a data acquisition system operating at the local line frequency (50 or 60 Hz). For the fast signals from W-filament atomization, e.g., full width at half-maximum of about 0.1 s, such a system is too slow to integrate peak area reliably, and peak height measurements are used in these situations.

In the EAI PBA-10 instrument, we operated the data acquisition system at 100 Hz to synchronize signal measurements with the pulsing frequency of the hollow cathode lamp. Faster data acquisition, while possible, might be incompatible with the limitations for pulsing the hollow cathode lamp to achieve self reversal correction. At 100 Hz, reliable integrated absorbance measurements are feasible. Peak area measurements also improve precision and thus the method detection limit for blood lead. For example, on the long wire filament the precision for peak height measurements at a blood Pb concentration of 6.7 µg dL⁻¹ is only 13.8% but, using integrated absorbance measurements, this improves by almost a half to 8.4%. The estimated 3σ method detection limit based on peak height is approximately 3 µg dL⁻¹ (n = 10) while, for integrated absorbance measurements, it is roughly 2 µg dL⁻¹.

Most previous workers^{12–17,21,23} used continuum background correction systems installed in commercial AAS instruments. In fact, our earlier report²⁰ also relied on a D₂ arc background correction system. But continuum correction is not really practical for a portable instrument because it requires an extra lamp and creates additional cost. One group^22,24 has used a "near-line" background correction method for a portable blood lead instrument based on W-filament atomization AAS. Near-line background correction was first proposed 40 years ago³⁰ for use with flame atomization AAS, and its application to Pb in biological fluids was described more than 30 years ago.³¹ This approach assumes that background absorption at both the resonance and non-resonance lines is exactly the same. However, for electrothermal atomization of complex matrices such as blood, where a structured background often appears, this is rarely true and can lead to serious errors. In the near-line method proposed by Sanford et al.,²² microwave-assisted digestion with concentrated nitric acid is required for blood lead measurements, and, in a later near-line method described by Salido et al.,²⁴ it is necessary to separate Pb from blood matrix by chelation and extraction into MIBK solvent, prior to deposition on the coil. Other workers have proposed methods that require sample pretreatment such as protein precipitation²¹ and microwave digestion²³ for blood lead measurement using W-filament AAS.

The self-reversal background correction system is simple, effective and requires no additional instrumental components. Self reversal is based on spectral line broadening that occurs within a hollow cathode lamp that is operated with a very high current, e.g., 100–300 mA.³² Atomic absorption measured during the high current pulse is greatly reduced but background absorption remains constant. Correction is accomplished by pulsing the hollow cathode lamp and taking the difference in absorbance measurements at low and high currents. Under a self-reversal correction scheme, sensitivity for Pb is reduced by as much as 50–60% and this is an important limitation. Thus, improvements in precision are critical for reaching the required detection limits.

Conclusion

Three different W-filament designs were investigated for the atomization of lead from blood matrices. A compact AAS instrument was assembled and used for atomization studies and method development. Background correction was accomplished using a self-reversal pulsing scheme. A long wire filament proved acceptable for routine blood lead measurements. Whole blood is simply diluted 1 + 4 with the same phosphate/Triton X-100 modifier used in conventional furnace AAS methods and 15 µL are deposited directly onto the long wire filament. Blood lead analyses are based on integrated absorbance measurements and calibration against matrix-matched blood standards. Detection limits of 1–2 µg dL⁻¹ are possible with precision of around 8% RSD at about 7 µg dL⁻¹. The method has been validated with a variety of blood lead reference materials and proficiency test samples, and verified against an established furnace AAS method using human blood specimens. Performance satisfies US regulatory requirements with respect to accuracy and precision, and meets public health requirements for screening children exposed to lead. Future investigations are planned that will include field studies using production instruments based on the instrument design described here.

Acknowledgements

This work was supported by Grant Number R08/CCR208632-04 from the Centers for Disease Control and Prevention. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the CDC. We thank the staff of the Wadsworth Center's Lead Poisoning Laboratory for providing the GFAAS analytical data on the routine human blood specimens. We are also grateful to Mr. E. Mills and Mr. W. Ahearn for their technical assistance with this project.

References

1. Centers for Disease Control, Preventing Lead Poisoning in Young children [Report], U.S. Department of Health and Human Services, Atlanta, GA, 1991..
2. Centers for Disease Control, Screening Young Children for Lead Poisoning: Guidance for State and Local Public Health Officials [Report], U.S. Department of Health and Human Services, Atlanta, GA, 1997..
3. US Department of Health and Human Services, Fed. Regist., 1992, 57, 7002.
4. R. M. Griffin, J. Occup. Med., 1986, 28, 615.
5. B. J. Feldman, J. D. Osterloh, B. H. Hata and A. D' Alessandro, Anal. Chem., 1994, 66, 1983.
6. B. J. Feldman, A. D'Alessandro, J. D. Osterloh and B. H. Hata, Clin. Chem., 1995, 41, 557.
7. J. Wang and B. Tian, Anal. Chem., 1993, 65, 1529.
8. C. Yarnitzky, J. Wang and B. M. Tian, Talanta, 2000, 51, 333.
9. M. W. Wensing, B. W. Smith and J. D. Winefordner, Anal. Chem., 1994, 66, 531.
10. A. D. Besteman, N. Lau, D. Y. Liu, B. W. Smith and J. D. Winefordner, J. Anal. At. Spectrom., 1996, 11, 479.
11. M. Williams and E. H. Piepmeier, Anal. Chem., 1972, 44, 1342.
12. H. Berndt and G. Schaldach, J. Anal. At. Spectrom., 1988, 3, 709.
13. M. F. Giné, F. J. Krug, V. A. Sass, B. F. Reis, J. A. Nobrega and H. Berndt, J. Anal. At. Spectrom., 1993, 8, 243.
14. C. G. Bruhn, F. E. Ambiado, H. J. Cid, R. Woerner, J. Tapia and R. Garcia, Anal. Chim. Acta, 1995, 306, 183.
15. M. M. Silva, R. B. Silva, F. J. Krug, J. A. Nobrega and H. Berndt, J. Anal. At. Spectrom., 1994, 9, 861.
16. P. O. Luccas, J. A. Nobrega, P. V. Oliveira and F. J. Krug, Talanta, 1999, 48, 695.
17. G. S. Lopes, A. R. A. Nogueira, P. V. Oliveira and J. A. Nobrega, Anal. Sci., 1999, 15, 165.
18. C. L. Sanford and B. T. Jones, Pittsburg Conference, 1994, (Abstract # 098)..
19. P. J. Parsons, K. M. Aldous, E. Mills, H. Qiao and W. Slavin, Pittsburg Conference, 1995, (Abstract # 1294)..
20. P. J. Parsons, H. C. Qiao, K. M. Aldous, E. Mills and W. Slavin, Spectrochim. Acta, Part B, 1995, 50, 1475.
21. F. J. Krug, M. M. Silva, P. V. Oliveira and J. A. Nobrega, Spectrochim. Acta, Part B, 1995, 50, 1469.
22. C. L. Sanford, S. E. Thomas and B. T. Jones, Appl. Spectrosc., 1996, 50, 174.
23. C. G. Bruhn, J. Y. Neira, G. D. Valenzuela and J. A. Nobrega, J. Anal. At. Spectrom., 1998, 13, 29.
24. A. Salido, C. L. Sanford and B. T. Jones, Spectrochim. Acta, Part B, 1999, 54, 1167.
25. P. J. Parsons and W. Slavin, Spectrochim. Acta, Part B, 1993, 48, 925.
26. P. J. Parsons, Environ. Res., 1992, 57(2), 149.
27. A. Salido and B. T. Jones, Talanta, 1999, 50, 649.
28. D. C. Paschal, K. L. Caldwell and B. G. Ting, J. Anal. At. Spectrom., 1995, 10, 367.
29. B. Welz, Spectrochim. Acta, Part B, 1992, 47, 1043.
30. J. B. Willis, in Methods of biochemical analysis, ed. D. Glick, Interscience, London, 1963, vol. 11, pp. 35.
31. R. J. Reynolds and K. M. Aldous, Atomic Absorption Spectroscopy, Griffin, London, 1970, p. 85..
32. S. B. Smith and G. M. Hieftje, Appl. Spectrosc., 1983, 37, 419.

---

Footnote

† Use of trade names is for informational purposes only and does not imply an endorsement by the New York State Department of Health.

---