Capacitively coupled microwave plasma atomic emission

Capacitively coupled microwave plasma atomic emission spectrometer for the determination of lead in whole blood. Michael W. Wensing, Benjamin W. Smith...
0 downloads 0 Views 484KB Size
Anal. Chem. 1994,66,531-535

Capacitively Coupled Microwave Plasma Atomic Emission Spectrometer for the Determination of Lead in Whole Blood Mlchael W. Wenslng, Benjamln W. Smlth, and James D. Wlnefordner' Department of Chemistrv, University Of Florida, Gainesville, Florida 3261 1

The determination of lead in whole blood by atomic emission spectrometry using a capacitively coupled microwave plasma and a tungsten filament electrode is presented. When the plasma-supportingelectrode is also used as the sample holder, transfer of the sample to the plasma is 100%. Microwaves are used to dry the sample and, at higher powers, ignite a helium plasma which results in the atomization and excitation of Pb. Using this methodology, a detection limit of 3 pg of Pb was L samples. The precision was 9%. obtained using ~ - Maqueous Whole blood samples were subjected to a drying stage similar to that of the aqueous samples. Following this drying stage, a low-power (30 W) helium plasma was ignited and used to ash the blood sample. Higher power plasmas (>150 W) were used to atomize and excite the Pb. Recovery of Pb from the blood samples was 88%,when compared to aqueous standards. It has become apparent that lead is toxic to the human body at lower concentrations than previously thought. As a result, in 1991 the Centers for Disease Control (CDC) lowered the level of concern for blood lead to 10 pg/dL.' The most popular method previously used to screen blood lead, an indirect measurement involving the fluorescence of zinc protoporphyrin, cannot determine lead below 20 pg/dL.2 Therefore, it is important to develop new methods or methodologies which are sensitive enough to determine lead at these levels, yet affordable enough to be used as a screening device. There are several methods which are sufficiently sensitive to determine lead at these levels in blood: anodic stripping voltammetry (ASV), potentiometric stripping analysis (PSA), and graphite furnace atomic absorption spectrometry (GFAAS). However, all of these methods require dilution of the blood sample, and the addition of chemical reagents, which increases the likelihood of sample contamination. ASV and PSA require the addition of acids to lyse the red blood cells where lead is bound, a buffer solution, and a surfactant which helps stabilize the s ~ l u t i o n .Accurate ~ and precise results for the rapid determination of lead in whole blood by GFAAS require dilution of the blood sample with Triton X-100 (a surfactant) and the use of a matrix modifier containing phosphate such as ammonium phosphate, diammonium hydrogen phosphate, or phosphoric acida4 The Triton X-100 US.Centers for DiseaseControl (CDC) PreuentingLeadPosioningin Young Children. A Statement by the Center for Disease Control in October, 1991. U S . Department of Health and Human Services/hblic Health Service/ Centers for Disease Control: Atlanta, GA, 1991. (2) Noble, D. Anal. Chem. 1993,65, 267A. (3) Roda, S. M.; Greenland, R. D.; Bornschein, R. L.; Hammond, P. B. Clin. Chem. 1988, 34, 563. (4) Benzo, Z.; Fraik, R Carrion, N.; Loreto, D. J . Anal. At. Spectrom. 1989,4, 397. (1)

0003-2700/94/0366-053 1$04.50/0 0 1994 Amerlcan Chemical Society

lyses the red blood cells, while the phosphate-containing salts allow higher ashing temperatures, ensuring the complete removal of carbonaceous material. Using simple dilution procedures, i.e.,dilutionof the blood samplewithTriton X-100 and water, lower ashing temperatures are used, resulting in carbonaceous depositswhich build up inside the furnace. These deposits retain lead and are not completely removed during the atomization step or the cleaning step. This results in a continual decrease in instrumental sensitivity for subsequent ana lyse^.^ Using GFAAS, the end result is at least a 5-fold dilution of the sample.6 In our laboratory, we are developing a technique which requires no sample pretreatment for the analysis of lead in whole blood. This paper describes the optimization of a capacitively coupled microwave plasma (CMP) for the direct determination of lead in blood. Since the CMP is less well known to analytical atomic spectroscopists than the microwave induced plasma (MIP), it will be briefly compared to the MIP. These plasmas differ by the way the energy is transferred to the plasma. In an MIP, microwaves are transferred from the microwave generator usually via a coaxial cable to an external resonator cavity. The cavity supports a standing microwave and sustains a gas discharge by interaction of the discharge gas with the microwave magnetic field. The discharge gas is contained by a quartz tube.7 MIPShave been operated at a variety of powers (30-1600 W) and, in general, can handle greater sample loading at higher powers. MIPS operated below 120 W have been used primarily in the analysisof GC eluates, where plasma loading is minimal. At these low powers, pressure gradient supercritical fluid chromatography (SFC) was not successful as the analyte signal was depressed as the pressure of the SFC mobile phase was increased.8 For these reasons, 500-WMIPS have been developed and, due to their higher power, have been used sucessfullyin handling SFC eluent. Limited success has been achieved in solid sampling into an MIP, where 1-mg quantities of finely ground coal have been directly introduced into a 500-W MIP and analyzed for carbon. The precision was limited to 20%.9 CMPs require an electrode to sustain the plasma. Microwaves are transferred via a waveguide to the electrode, which is supported in a widened, central channel of an ICP torch. The electrode couples to the microwave field, and when the microwave power is strong enough, a plasma ignites at the tip of the electrode. Powers ranging from 100 to 2000 W are used' CMP plasmas are very robust and have been ( 5 ) Subramanian, K. S . Prog. Anal. Spectrosc. 1986, 9, 237. (6) Jacobson, B. E.; Lockitch, G.; Quigley, G. Clin.Chem. 1991, 37, 515. (7) Zander, A. T.; Hieftje, G. M. Appl. Spectrosc. 1980, 35, 357. (8) Wu, M.; Carnahan, J. W. J. A n d . At. Spectrom. 1992, 7 , 1249. (9) Gehlhausen, J. M.; Carnahan, J. W. Anal. Chem. 1991, 63, 2430.

Analytical Chemistry, Vol. 66,No. 4, February 15, 1994 531

successfully with aqueous aerosols, and such matrices as tomato leaf residue and coal fly ash residue.IO In addition, CMP atomic emission spectrometry (CMP-AES) is very sensitive. Ali and Winefordner" obtained CMP-AES detection limits in the 1-100-pgrangefor Ag, Ba, Cd, Cu, Ga, Ge, In, Li, Mg, and Zn by vaporizing samples directly off a small- (1-3 mm) diameter tungsten loop electrode. Discrete sample introduction is simplified in a CMP versus an MIP or ICP because microliter sample volumes can be deposited directly on the plasma-supporting electrode, ensuring complete transfer of the sample to the plasma with minimal dilution. At low microwave powers, the sample is dried by direct absorption of microwaves and also indirectly through resistive heating of the electrode. At higher microwave powers, a plasma ignites on the electrode, around thesample, atomizing and exciting the sample. The emission signal is detected as a transient signal. Ali and Winefordnerl0Jl used this approach with the tungsten filament loop and with a graphite cup electrode. This approach is advantageous over other sample introduction methods, including electrothermal vaporization, where the sample is diluted by the carrier gas and some of the sample is retained on the tubing walls leading from the vaporization device to the plasma.12 Direct insertion of the sample into an Ar-ICP plasma was first performed by Salin and Horlick,13who inserted previously dried samples carried on a graphite electrode directly into the ICP; they introduced the electrode axially through the injector tube of an ICP torch, which also ignited the ICP upon insertion. Others have introduced the graphite electrode directly into a continuously running ICP.l"-l7 Abdullah, Fuwa, and Haraguchi18 raised the sample-containing graphite electrode in stages, first using indirect heating from the ICP to ash the sample, before it was introduced into the continuously running ICP. For the CMP, we report the use of a plasma ashing step, where after first charring the blood with microwaves, a very low power (30 W) microwave plasma is ignited on the electrode and is allowed to ash the sample. A disadvantange of the CMP electrode is that it erodes with time and appears in the emission spectra, and so it is necessary to choose an electrode material which will not produce interfering spectral emission lines. The electrode must be conductive and have a higher melting point than the plasma temperature. Several materials have been successfully used as electrodes in our laboratory, including tungsten, tantalum, and graphite.19.20 This paper characterizes the CMP for the atomic emission spectrometric determination of Pb in blood using a tungsten filament electrode and the discrete sampling technique mentioned above. The optimization procedure is described for aqueous samples, and the technique is applied to blood ~~~

~

~~

(IO) Ali, A. H.; Ng, K. C.; Winefordner, J. D. J . Anal. At. Spectrom. 1991,6,21I . (11) Ali, A. H.; Wincfordner, J. D. Anal. Chim. Acta 1992, 264, 327. (12) Alvarado, J.; Cavalli, P.; Omenctto, N.; Rossi, G.Anal. Lett. 1989, 22, 2975. (13) Salin, E. D.; Horlick, G . Anal. Chem. 1979. 51, 2284. (14) Sommer, D.; Ohls, K. Fresenius 2.Anal. Chem. 1980, 304, 97.

( I S ) Kirkbright, G. F.; Walton, S.J A ~ I y s 1982, t 107, 276. (16) Kirkbright, G. F.; Zhang, L.-X. Analysf 1982, 107, 617. (17) Zhang, L.-X.; Kirkbright, G.F.; Cope, M. J.; Walton, J. M. Appl. Specfrosc. 1983, 37, 250. (18) Abdullah, M.; Fuwa, K.; Haraguchi, H. Specfrochim.Acta 1984,398,1129. (19) Masamba, W. R. L.;Smith, B. W.; Winefordner, J. D. Appl. Spectrosc. 1992, 46, 1741. (20)Patel, B. M.; Heithmar, E.; Winefordner, J. D. Anal. Chem. 1987,59,2374

532

Analytical Chemistry, Vol. 66,No. 4, February 15, 1994

voltage glass chlmney

Flgurs 1. Capacltively coupled microwave plasma setup. Table 1. CMP Equlpmenl

item

dc high-voltage power supply; max output 1.2 kW, voltage regulation 15% magnetron aluminum waveguide two concentric tube quartz torches 0.5 m spectrometer; 1200 grooves/", 300-nm blaze wavelength intensified photodiode array

model

manufacturer

803-330

Hipotronics, Brewster, NY

OM75A

Samsung lab constructed lab constructed

1870

Spex, Edison, NJ

TN-6122A

Tracor Northern, Middleton, WI

analysis. Helium is used as the plasma gas due to its excellent excitation efficiency and low background characteristics.

EXPERI MENTAL SECTION Apparatus. The experimental apparatus is shown in Figure 1; the instrumental components are described in Table 1. The setup is typical of CMP-AES systems, with the exception that a high-voltage switch is used to gate power to the magnetron and water cooling of the electrode is unnecessary. The electrode is constructed from 0.25-mm-diameter high-purity (99.98%) tungsten wire (Aesar, Ward Hill, MA). The electrode loop is formed by tying a knot 3 mm in diameter in the wire (samples deposited on the loop are held by adhesion) and then bending the ends of the wire so that they are perpendicular to the loop. The ends extend 63 mm downward from the loop. This length was found to be optimum for efficient coupling to the waveguide. The electrode is held in the 1-mm-diameter central channel of a modified ICP torch by compression of the wire against the central channel walls. The bottom of the central channel is sealed. Procedure. Aqueous samples of 5-pL volume are deposited on the electrode with a pipet and dried at 75 W for 90 s. After the drying step, the high-voltage switch is opened, and the voltage is increased. The helium plasma gas is set to 10L/min, and when the switch is closed, the plasma autoignites. Closing the switch also triggers the detector, which integrates signals over 0.1-speriods for a total of 4 s. A total of 40 time-resolved spectra are obtained. The results are stored and analyzed with a computer. For all determinations, the monochromator was operated with a slit width of 120 pm. For blood samples, 2-1L aliquots are deposited on the electrode. The sample is dried first at 75 W for 30 s. The plasma gas is then ignited at low power (30 W) and allowed to ash the sample for 3 min. The switch is opened, and the

250

1

1

0

404.0

404.5

405.0

405.5

406.0

406.5

407.0

407.5

408.0

1.0

Wovelength (nm)

O

Flguro 2. Aqueous emission spectrum near 405.8 nm.

0.5 I -0.5

voltage is increased. Closing the switch results in the vaporization, atomization, and excitation of lead. Detection is accomplished in the same way as in the case of aqueous samples.

RESULTS AND DISCUSSION Aqueous Background. The electrode and aqueous background around the 405.8-nm Pb line is shown in Figure 2. One can identify several tungsten lines, all adequately resolved from the Pb line. In addition, one can identify a very weak molecular peak near the Pb emission line. This peak is most likely due to the second positive system of nitrogen and can be corrected for by careful subtraction of the blank. The nitrogen originates from air that is entrained into the atmospheric pressure plasma. Radial, Axial, and Power Optimizations. It was important to determine spatial variation of the Pb signal/background ratio since the whole plasma was not imaged into the spectrometer. Two microliters of 1 ppm Pb was used. The Pb signal-to-background ratio was measured by integrating the Pb signal over time. For high concentrations of Pb (> 100 ppb), the full width at half-maximum (fwhm) of the transient Pb emission signal was determined to be 0.4 s. Therefore, all lead signals were integrated over this time period. The background was also added over this same period of time. The Pb signal-to-background ratio is optimum near the electrode and decays rapidly in both the axial and radial directions. The results are shown in Figure 3A,B. For the power optimization, 2 pL of 10 ppm Pb was used. The integrated Pb signal was measured over the fwhm of the transient Pb signal, which was different at different powers. The Pb signal profiles as a function of time are plotted in Figure 4 and show that the signal profile becomes sharper and more intense as the power is increased. However, above 172 W, there is no gain in the integrated signal-to-background ratio. At 172 W, the electrodecould sustain 100 firings, while at powers greater than this, the electrode lasted for only 5 firings. The filament loop slowly thinned at the edges of the loop until it broke. Therefore, a power of 172 W was used in further studies. The detector was triggered immediately after the high-voltage switch was closed, however, it took a few tenths of a second for the plasma to ignite. This produces the apparent delay preceding the appearance of the Pb signal. Aqueous Calibration Curve. A calibration curve was obtained for Pb trsing 5-pL aliquots and the optimum

I

I

0.0 0.5

I

I

I

I

1.0

1.5

2.0

2.5

1

I

I

3.0 3.5

Radial Distance from Electrode (mm)

0

P

B

0

-

-0.5

0.0

0.5

1.0

1.5

2.0 2.5

3.0

3.5

Distance above Electrode (mm) Flguro 5. Radial (A, top) and axial (B, bottom)emission profilesof Pb.

eE

5000

a

4000

e

3000

e e e 5

sP

C

2000 1000 0 -1 000 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 7.8 2.0 Tlme (8)

Figure 4. Temporal emission profilesof Pb as a function of microwave power.

atomization power of 172 W and viewing the center of the plasma less than 1 mm above the electrode. The Pb solutions were in the form of the nitrate salt and were made by serial dilution. The calibration curve was linear over three decades Analytical Chemistry, Vol. 66,No. 4, February 15, 1994

533

+

No Ashing

20 mA

15 mA

25 mA

Ashing Current (mA)

0

1

2

Figure 8. Relative amounts of KCI, PbCI2,and Pba(P01)2remaining as a function of ashing current.

Time (s)

1

Figure 5. Temporal emission profile of 5 ng of Pb.

2000 A

from thedetectionlimit (0.7 ppb) to 1 ppm. Themassdetection limit corresponded to 3 pg of Pb. The limit of detection was based on three standard deviations of the background obtained by averaging 10 diodes on either side of the Pb peak. The CMP-AES detection limit in aqueous samples is better than that necessary for the CDC (10 ppb) and rivals that of GFAAS (5.5 pg, 0.3 ppb) while being better than for ICP-AES (14 ppb).21 Figure 5 shows the background-corrected, timeresolved emission profile of 5 ng of Pb. The Pb emission signal decays to the baseline quickly as the majority of the signal is captured in 1 s. It actually takes 3 s for the Pb signal to return completely to the baseline. A plot of the log of the integrated Pb signal versus the log of the lead concentration had a slope of 0.99. The precision, as estimated by the relative standard deviation, was 15%. This precision was limited by the reproducibility of the power supply settings, which varied by 34%. A magnetron is a nonlinear device and has the same voltage-current characteristic as a diode. Therefore, in order to keep the power constant it is essential to control the current, not the voltage. However, our power supply was incapable of this. When the data set was limited so that the power varied only by 5%, the precision improved to 9%. Measurement of Pb in Blood. Volumes of 2 1 L were used to minimize the amount of sample that needed to be ashed. The CMP handles plasma loading better than the MIP; however, ashing is still necessary to minimize the amount of loading. In addition, a neighboring potassium peak completely obscured the lead peak unless some of the potassium was removed. It was determined that the best way to ash the sample was to use the plasma ashing step described above. In order to assess the effectiveness of the ashing step, aqueous samples containing either potassium chloride, lead chloride, or lead phosphate were dried for 90 s, ashed at various plasma currents for a period of 3 min, and then atomized at 172 W. In order to prepare the lead phosphate, 5 p L of 0.2% phosporic acid was added to 5 p L of lead chloride. (21) Varian Guide to ICP/AAS analytical values (PN 8 5 101009 00).

534

Analytical Chemistty, Vol. 66, No. 4, February 15, 1994

500

200

100 I5O

-50

t

,

i

'

1

'

'

I

'

I

'

I

'

I

"1"1 I

403

I

404

I

i

Pb

405

406

1

407

408

409

I

Pb

n A

I

1

Therefore, blood was ashed at an output power of 30 W. This ashing step resulted in the efficient removal of potassium. The spectrum of 330 ppb (mass/volume) Pb in blood is shown in Figure 7A,B. Figure 7A identifies the 405.8-nm region around Pb. Figure 7B shows an expanded view of the Pb peak. The percent recovery of Pb was 88% and was calculated by comparing the Pb in whole blood signal to aqueous standards with the same concentration of Pb. The precision in blood was 7% at 330 ppb. A detection limit of 6 ppb (12 pg) was obtained for Pb in whole blood. This is comparable to that achieved by GFAAS (9 pg).22

CONCLUSION The use of the tungsten filament electrode in combination with the CMP has been shown to be a sensitive way of determining lead in both blood and aqueous samples. Picogram quantities of lead can be detected. The plasma ashing step is a unique and useful way to ash the sample; however, further studies are necessary and are underway to ensure that this step is sufficiently reproducible in a variety of blood (22) Parsons, P. J; Slavin, W. Specfrochim. Acta 1993, 4 8 4 925.

samples. If this is a problem, matrix modifiers such as ammonium phosphatecan be used to make the lead lessvolatile and the potassium more volatile (Le., Mg(NO&), and the ashing time can also be decreased. In addition, other studies are underway using a current-regulated power supply to improve the precision of the method. To further reduce the cost of the device, shorter focal length monochromators and/ or interference filter-photomultiplier tube combinations will be evaluated. Finally, there is no part of this device that cannot be automated to a simple push-button device, which is necessary for routine screening of lead in blood by nonskilled operators.

ACKNOWLEDGMENT This work was supported by a grant from the National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, GA (Contract CCR408614-02). Received for revlew August 10, 1993. Accepted November 29, 1993."

* Abstract

published in Aduonce ACS Absfracfs, January 1 , 1994.

Analytical Chemistry, Vol. 66, No. 4, February 15, 1994

535