488
Anal. Chem. 1980, 52, 488-492
Simultaneous Determination of Five Trace Metals in Seawater by Inductively Coupled Plasma Atomic Emission Spectrometry with Ultrasonic Nebulization S. S. Berman, J. W. McLaren,” and S. N. Willie Analytical Chemistry Section, Chemistry Division, National Research Council of Canada, Ottawa, Canada, K 1A OR9
i t 5 concentration is more t h a n five times the detection limit-in other words, in a region where a reliable analysis might be accomplished. Five of the ten metals (Fe, Mn, Zn, Cu, and Ni) lie a t or above the dashed line, but P b , Co, and Cd are still below detection limits while Cr and As are in the region of uncertainty between t h e two lines. I t should be pointed out here that a 20-fold preconcentration is about the maximum conveniently possible with the commonly-used A P D C / M I B K solvent extraction technique. Furthermore, recent evidence (12, 13) indicates t h a t some of the seawater concentrations used for Figure 1 may be unrealistically high owing to contamination during sampling. T h e need for a technique which affords a t least 100-fold preconcentration is thus all the more clear. A number of studies (14-16) have shown t h a t it is possible to decrease ICP-AES detection limits, a t least for simple dilute acid solutions, by an order of magnitude or more by t h e substitution of ultrasonic nebulization with aerosol desolvation for t h e conventional pneumatic nebulization. In fact, Fassel and co-workers ( I ) have suggested this as a possible route t o seawater analysis. Surprisingly few applications involving the use of an ultrasonic nebulizer (USN) developed by Olson e t al. ( 1 4 ) have appeared despite the commercial availability of a nebulizer which is virtually a replica of this design. No doubt some workers have been discouraged by t h e possibility of desolvation interferences as reported by Boumans e t al. ( I 7 ) and also by Boumans’ comments (18) that the upper limit for dissolved solids concentrations, a t least for his USN,is about 0.1% if matrix effects are to be limited t o *lo%, whereas the cross-flow and concentric pneumatic nebulizers generally work well a t least u p to 1% dissolved solids. A recent publication by Haas et al. (19)reported the use of ultrasonic nebulization with aerosol desolvation in the simultaneous determination of trace elements in urine by ICP-AES. The use of an internal standard was necessary because the varying dissolved solids concentrations in t h e samples caused variations in the rate of aerosol production. T h e authors also commented that they were unable to reproduce extremely impressive detection limits reported in an earlier publication (14) and t h a t this reduced their expected capability of multielement analysis of urine. Some improvements in the procedure enabling more elements t o be determined a t normal levels in urine are included in a recent technical report (20). It is important to stress that detection limits attainable with ultrasonic nebulization uithout desolvation are not significantly better than can be achieved with pneumatic nebulization, despite the fact t h a t the ultrasonic nebulizer is about ten times more efficient than the cross-flow or concentric glass nebulizers, in terms of aerosol production. I t has been suggested ( 1 4 ) that the much greater volume of water produced by the USK cools the plasma sufficiently a t normal operating power to seriously degrade i t s excitation characteristics. T h e use of higher power does not seem to compensate for this. Bearing in mind that t h e USN will introduce about ten times as much solid material into the plasma per unit time
Simultaneous determination of five trace metals in seawater by a combination of ion-exchange preconcentration and inductively coupled plasma-atomic emission spectrometry (ICP-AES) is described. Ultrasonic nebulization with aerosol desolvation is used to introduce the metal concentrates to the plasma. Results are presented for iron, manganese, copper, zinc, and nickel in a relatively unpolluted coastal seawater sample, and these results are compared to those obtained by two other independent methods. Accuracy is within 0.1 pg L-’ at the 1 pg L-’ level and standard deviations range from 0.1 to 0.3 pg L-’. Chromium, cadmium, cobalt, and lead levels in the concentrates remain below values at which reliable analyses can be performed. The performance of the unrasonic nebulizer is assessed with reference to that of a cross-flow pneumatic nebulizer.
T h e application of inductively coupled plasma-atomic emission spectrometry (ICP-AES) t o the analysis of natural fresh waters is now well established. Two detailed papers have appeared ( 1 , 2 ) and numerous others have been presented a t meetings, a n d in technical reports (3, 4). T h e extension of this development to seawater analysis has been slow, however, for two major reasons. T h e first is t h a t the concentrations of almost all the trace metals of interest are 1 Fg L-’or less, below detection limits attainable with conventional pneumatic nebulization. T h e second is t h a t the seawater matrix, with some 3.5% dissolved solids, is not compatible with most of the sample introduction systems used with ICPs. Thus direct multielement trace analysis of seawater by ICP-AES is impractical, a t least with pneumatic nebulization. In view of this, a number of alternative strategies can be considered: (1) preconcentration a n d removal of the metals of interest from t h e seawater matrix prior t o I C P analysis; (2) the use of ultrasonic nebulization with aerosol desolvation; a n d (3) a combination of t h e above two strategies. Preconcentration techniques such as solvent extraction, ion-exchange, and carrier precipitation have been extensively studied with reference to seawater analysis and reviewed recently ( 5 ) . T h e most common method of seawater analysis for trace metals in use a t present involves t h e combination of one of these techniques with graphite furnace atomic absorption spectrometry (GFAAS). I t is only very recently that reliable methods of direct analysis for some of these metals by GFAAS have been proposed (6-9). Figure 1indicates the feasibility of multielement analysis of seawater by ICP-AES for 10 metals of interest, assuming t h a t a 20-fold preconcentration is accomplished. T h e metal levels in the concentrate (based on a recent tabulation (10) of typical seawater concentrations) are plotted against ICP detection limits attainable with pneumatic nebulization (11). If the point for an element lies above t h e solid line, its concentration is above the detection limit; similarily, if the point lies above the dashed line. 0003-2700/80/0352-0488$01 0010
C
1980 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 52, NO. 3, MARCH 1980
489
GENEXITOR
c
!ogL+-h+-
100
I C F DETECTION LIMIT,p gL-'
Figure 1. Plot of metal concentrations in a seawater "ex?ract" (assuming
a 20-fold preconcentration) against ICP detection limits with pneumatic nebulization
as d o the pneumatic nebulizers, the feasibility of direct analysis of seawater by this route seems doubtful. Seawater at 35 parts per thousand salinity contains about 400 pg mL-' Ca, 1300 pg mL-' Mg a n d 11000 pg mL-' Na. Even if t h e torch did not salt u p in a short time, accurate analyses would be made very difficult by spectroscopic interferences due t o very high levels of Ca and Mg. Perhaps the best available approach to seawater analysis by ICP-AES is a combination of a preconcentration technique with t h e use of ultrasonic nebulization and aerosol desolvation. As shown in Figure 1, even a 20-fold preconcentration is insufficient for five of ten metals of interest when conventional pneumatic nebulization is used. In this paper we describe an attempt to determine nine trace metals (Fe, M n , Cu, Zn, Ki, Cr, Co, Cd, and P b ) in seawater by a combination of ion-exchange preconcentration a n d ICP-AES using ultrasonic nebulization with aerosol desolvation. Results a r e presented for t h e first five of these in a relatively unpolluted coastal seawater sample. T h e performance of t h e ultrasonic nebulizer is assessed with reference t o that of a cross-flow pneumatic nebulizer.
EXPERIMENTAL Seawater Sampling. Near-shore surface samples are filtered through 0.45-pm Millipore filters and acidified to p H 2 with nitric acid. All analyses to date indicate them to be representative of relatively unpolluted coastal Atlantic seawater. h'o detectable changes in the metal concentrations in any of these samples have occurred over several weeks of standing. Reagents. Acids used in this work were purified by sub-boiling distillation (21). High purity water is produced by passing distilled water through a deionizing system. All reagent and sample preparations were done in a class 100 clean air laboratory. Ion-Exchange Preconcentration. The ion-exchange procedure used has been described in detail in a recent publication by Sturgeon et al. (22)and is similar to that reported by Kingston e t al. (23). The volume of seawater taken for each preconcentration varied from 225 to loo0 mL. Calibration was accomplished by the method of standard additions; either 2 or 3 spikes were used, so that the total volume of seawater consumed in each analysis ranged from 900 to 3000 mL. T o accommodate the 1000-mL samples, reservoirs fashioned from 1-L polypropylene separatory funnels were fitted to the "separatory columns" described in ref. 22. The reservoir funnels were fitted with a microbore tubing air vent. A brief outline of the procedure will be given here. Filtered acidified seawater (225-1000 mL), which has been spiked if necessary, is treated with 1 N ammonium acetate buffer solution (4 mL/100 m L seawater) and the p H is adjusted to 5.4 with 2 M ammonium hydroxide and 0.5 M hydrochloric acid. The sample is passed through the Chelex-100 column (after initial shaking with the resin as described in ref. 22) at a rate of 1.5 to 2.0 mL/min. The resin is then washed with 40 mL of water followed by four 10-mL aliquots of the ammonium acetate buffer at pH
Figure 2. Schematic diagram of the ICP-echelle spectrometer
5.2. These latter washings remove t:he bulk of the Ca and Mg and, in our experience, part of the NIn and (to a lesser extent) Fe and Cu from the resin. The resin column is again washed with 5 mL of water and aspirated dry. The mace metals are then eluted by shaking the resin with 5 mL of 5 M nitric acid followed by 5 mL of water. Blanks are prepared by running the appropriate volume of the ammonium acetate buffer solution (40 mL + 4/100 X sample volume in mL) through the column, washing with water, and eluting with nitric acid and water as described for a seawater sample. At least two blanks are run as a part of each analysis. Ultrasonic Nebulizer/Desolvation Apparatus. Introduction of the trace metal concentrates to the ICP was accomplished by means of an ultrasonic nebulizer (USN)used in conjunction with a desolvation apparatus very similar to that described by Veillon and Margoshes ( 2 4 ) . The cylindrical heating chamber, 15 cm long and 5 cm in diameter was wrapped with a 4-ft length of 1-inch wide "Briskeat" heating tape (Briscoe Mfg., Columbus, Ohio) which is rated for 384 W a t 115 V. The only modification necessary to a standard 35-cm Friedrichs condenser was the installation of a side arm to allow the dry aerosol to exit to the plasma. The temperature of the heating chamber was controlled with a VARIAC set to 70 V. This setting was just sufficient to vaporize a dilute acid aerosol completely by the time it entered the condenser. Operating at higher temperature tended to cause "pulsing" in the plasma due to large variations in the back pressure. Very rapid evaporation of the water entering the heating chamber is undesirable, since any variation in the rate of aerosol introduction from the USN will cause very noticeable changes in the back pressure of the system. In effect, the heating chamber amplifies any variation in the rate of sample introduction t o it from the USN. ICP-Echelle Spectrometer. Details of the custom instrument and the standard operating conditions are given in Tables 1-111. An earlier report (25) described the optical coupling of the ICP source and echelle spectrometer. Since then, the instrument has been further developed so that it can now be used either as a monochromator or as a 20-channel direct reader. A schematic diagram of the configuration for simultaneous multielement analysis is shown in Figure 2. Currents from the 20 photomultiplier tubes (Hamamatsu R292) in the detector module are led to the high speed scanner where they are rapidly (but sequentially) switched to the programmable current amplifier. This device is remotely programmable to operate on any of 6 decade current ra:nges from lo-'' A full scale to 10-eA full scale. The output from the current amplifier, filtered by a simple RC circuit (RC = 2.35 ms) goes t o a 12-bit analogto-digital converter which is part of the DECLAB 11/03 system. As indicated by the schematic, the scanner and current amplifier are normally operated under computer control, although channel and gain selection can also be accomplished manually. The software package provided with the computer permits us to write all the necessary programs for spectrometer control and data acquisition and manipulation in FORTRAN; no assembler language programming is necessary. Although the routines would probably run faster if programmed entirely in assembler language, having
490
ANALYTICAL CHEMISTRY, VOL. 52, NO. 3, MARCH 1980
Table I. Experimental Facilities ICP
sample introduction systems
spectrometer
current amplification system
data acquisition system
Plasma-Therm Model PT-1 ICP source including: R F generator Type HFS-3000D (27.12 MHz, 2.5 k W ) . Automatic power control APCS-3. Automatic match i ng network hl h-30 0 0 D , ( A ) Pneumatic nebulization. Plasma-Therm Model T N - 1 cross-flow nebulizer with simple conical fog chamber, model SC-1. Gilson Minipuls I1 peristaltic pump. ( B ) Ultrasonic nebulization with aerosol desolvation. Plasma-Therm Model UNS-1 system. Veillon-Margoshes type desolvation apparatus (24). Spectrametrics SMI I11 echelle grating spectrometer with sequential and 20-channel mu1tiel ement cassettes. Keithley Model 703 100-channel scanner main frame with 2 model 7029 10-channel low voltage plug-in cards and model 7032 parallel BCD interface card. Keithley model 18 000-20 programmable linear current amp1 if ier. DECLAB 1 1 / 0 3 system consisting of: PDP 1 1 / 0 3 microcomputer with 24K MOS memory. RXOl dual drive floppy disc mass storage. D R V l 1 parallel line interface. ADV11-A 12-bit analog-todigital converter. K W V l 1 -A programmable real-time clock. D L V l 1 serial line interface. Texas Instruments Silent 700 terminal.
Table 11. Wavelengths of Multielement Cassette (in a m ) Ca( 11) CO(I1) Cr(1 Cr(I1) CU(I 1 Cu( 11) Fe(I1)
328.07 317.93 228.80 211.13 237.86 123.44 267.72 324.76 213.60 239.94
Mg(II 1 Mn(1I) Ni(1) Ni(I1) Pb(1) Pb(I) Pb(I1) V( 111 W I ) Zn(I1)
280.27 267.61 341.48 231.60 283.31 368.35 220.33 292.40 213.86 202.55 -
Table 111. Standard Operating Conditions ICP
ul t r as on i c
nebulizer spectrometer
Power: 1.1 kW indicated incident power ( < 5 R reflected) Plasma Ar: 1 4 L m i n - ' Auxiliary plasma Ar: 1 L min-' Aerosol carrier Ar: 1 L min-' for ultrasonic nebulizer. 0 . 7 L min for pneumatic nebulizer Power: 30-15 W Sample introduction rate: 2.8 m L min-l Observation height: 1 2 m m above load coil
Entrance slit: 1 0 0 p m X 0.5 m m Exit slits: 50 p m Y 0.4 m m Photomultiplier tube voltages: 900 V
all the software in FORTRAN has allowed us to develop a viable system using a common high-level programming language, while also retaining complete flexibility to alter any aspect of our routines. Analytical Procedure. Any number of the 20 available wavelengths (Table 11) can be used in a particular analysis. At the beginning of the analytical sequence, the operator selects which channels are to be active. Before the measurement sequence for each sample or standard, an identification and desired number of repeats of the scanning cycle are entered. Each scanning cycle takes about 35 s and normally 3 repeats are run. The first part of the first scanning cycle for each sample is the automatic selection of the correct current amplifier range for each active channel. At the end of the desired number of scanning cycles, a table of photocurrent data is printed out, giving an overall mean photocurrent (with standard deviation) for each channel, plus a mean and standard deviation for each repeat. A t the same time, these data are transmitted to the mass storage floppy disk. Standards can be run (or re-run) a t any time, as no results are calculated until the end of the entire analysis. Then, a subsequent program searches the data file, assembles the calibration data, performs the linear regression analyses, and calculates the results for the samples. In the application to seawater analysis described here, it was necessary to use the method of standard additions, because quantitative recovery of the original and added metals in the samples could not be assumed. Because the method of standard additions provides no information about the blank, it was necessary to make a background subtraction and blank correction for each sample before performing the linear regression analysis to calculate the metal concentrations in the unspiked samples. Subtraction of the mean total intensities observed for the blanks a t each wavelength from those observed for the samples before performing the linear regression analyses provided background subtraction and blank correction in one step. This is not a t present part of our standard analytical program, as calibration by standard additions is the exception rather than the rule.
RESULTS A N D DISCUSSION P e r f o r m a n c e of the U l t r a s o n i c N e b u l i z e r . T h i s work provided an opportunity t o compare two sample introduction systems, pneumatic nebulization, and ultrasonic nebulization with aerosol desolvation. Detection limits were of primary interest, b u t it is also instructive t o compare the sensitivities, Le., t h e slopes of t h e calibration lines a t t h e various wavelengths. Before proceeding with a discussion of these results it is necessary to describe our method of calculating detection limits, as it provides a more conservative, but probably more useful value than d o other commonly used methods. In our method, the detection limit is defined as that concentration of analyte which gives a response equal t o three times the standard deviation of the blank or background value. This blank value is the ordinate intercept calculated from the linear regression analysis of t h e calibration data. Thus, not only the standard deviation of the background signal but also the standard error of the regression is taken into account. T h e least quantitatiuelg determinable concentration (relative standard deviation 1 2 5 % ) is equal to twice the detection limit when t h e latter is defined i n this way. In Table IV sensitivities and detection limits using the two sample introduction systems are reported. T h e sensitivities are simply t h e computed slopes of t h e calibration lines with precision expressed as t h e standard deviation. In each case a set of 6 standards in t h e range 0-200 kg L-' in 0.1 M H N 0 3 was used. It can be seen that ultrasonic nebulization with desolvation afforded a mean increase in sensitivity of about a factor of 9 (5-12) and a mean decrease in detection limits of about a factor of 4 (1-12) over pneumatic nebulization. These improvements are in accordance with the approximately tenfold greater efficiency of t h e USN. However, these d a t a d o not warrant a detailed discussion of relative efficiencies of the two nebulizing systems because not only were the carrier
ANALYTICAL CHEMISTRY, VOL. 52, NO. 3, MARCH 1980
Table IV. Comparison of Sample Introduction Systems ultrasonic nebulization with -_- desolvation detection limit, wavelength. nm hln(I1) 2 5 7 6 1 Cu(I1) 2 1 3 . 6 0 Zn(I1) 2 0 2 35 Co(I1) 2 3 7 . 8 6 Zn(1) 2 1 3 8 6 Pb(I1) 2 2 0 . 3 j Cd(1I) 2 1 1 . 1 3 Pb(1) 2 8 3 . 3 1 Fe(I1) 2 5 9 . 9 4 Ni(I1) 2 3 1 . 6 0 Cd(1) 2 2 8 . 8 0 Cr(I1) 2 6 7 . 7 2 CurI) 3 2 1 75 Pb(1) 3 6 8 3 5 V(I1) 2 9 2 . 4 0 Cr(1) 1 2 5 . 4 1 Ni(1) 3 4 1 1 8
slope, nA/ppmU
229.2 i 3.8 i 3.8 = 10.5 i 21.6 i 1.29 t 1.06 L 3.0 t 58.4 t 8.02 = 2.02 = 97.2 = 3121 3.3 = 160 = 36.0 = 24.6 =
0.8 0.1 0.1 0.1 0.1 0.03 0.03 0.2 0.4 0.06 0.02 0.7 1 0.2 2 0.9 0.6
L"$b 1
7 5 3 1
7 2
15 2 2 3 2 1 21 1 8
7
pneumatic nebulization detection limit, slope, nAippm
!J $b L-
25.5 i 0.2 0.36 L 0.03 0.31 = 0.03 1.1 I 0.1 2.21 = 0 . 0 2 0.11 L 0 . 0 1 0.37 L 0 . 0 1 0.42 I 0.02 7.5 I 0 . 1 0 . 7 9 1 0.04 0 . 1 7 = 0.01 9.6 = 0.2 65.3 :0.6 0.54 I 0 . 0 2 17.2 ! 0.3 1.8 i 0.2 4.1 :0.2
a Precision expressed as standard deviations. lated as defined in the text. -
491
______
--__
-
2 27 25 12 3 81 16 55 1 14 35
Table V. Effect of Nitric Acid Concentraticpn on Performance of Ultrasonic NebulizeriDesolvation Apparatus slope, n A / p p m 1.511 H N O ,
wavelength, nm
0 1 X1 H h O ,
l l n ( 1 I ) 2.57 61 Cu(I1) 2 1 3 6 0
2 0 0 .3 30 0 1 3 7 - 0 1 8 9 0 2 2 0 8 - 0 3 129 001 3 8 01 2 9 0 2 1954 0 3 75 0 2 84 - 2 278 6 131 2 289 0 6 2192 0 2
Zn(I1) 2 0 2 55
Co(I1) 237 8 6 Zn(1) 8 1 3 86 Pb(I1) 2 2 0 3