Advances in atomic spectroscopy, volume 3

ADVANCES IN ATOMIC SPECTROSCOPY

V o l u m e 3 �9 1997

This Page Intentionally Left BlankThis Page Intentionally Left Blank

ADVANCES IN ATOMIC SPECTROSCOPY

Editor: JOSEPH SNEDDON Department of Chemistry McNeese State University Lake Charles, Louisiana

V O L U M E 3 �9 1997

Greenwich, Connecticut

~~~~ JAi PRESS INC. London, England

Copyright �9 1997 by JAI PRESS INC. 55 Old Post Road, No. 2 Greenwich, Connecticut 06836

JAI PRESS LTD. 38 Tavistock Street Covent Garden London WC2E 7PB England

All rights reserved. No part of this publication may be reproduced, stored on a retrieval system, or transmitted in any form, or by any means, electronic, mechanical, photocopying, filming, recording, or otherwise, without prior permission in writing from the publisher.

ISBN: 0-7623-0072-8

155N: 1068-5561

Manufactured in the United States of America

CONTENTS

LIST OF CONTRIBUTORS

PREFACE Joseph Sneddon

PLASMA SOURCE MASS SPECTROSCOPY Andrew S. Fisher and Les Ebdon

MULTIELEMENT GRAPHITE FURNACE AND FLAME ATOMIC ABSORPTION SPECTROMETRY

Joseph Sneddon and Kimberly S. Farah

DIRECT CURRENT ARCS AND PLASMA JETS Rudi Avni and Isaac B. Brenner

DIRECT AND NEAR REAL-TIME DETERMINATION OF METALS IN AIR BY IMPACTION-GRAPHITE FURNACE ATOMIC ABSORPTION SPECTROMETRY

Joseph Sneddon

INDEX

vii

ix

33

63

203

225

This Page Intentionally Left Blank

LIST OF CONTRIBUTORS

Rudi Avni

Isaac B. Brenner

Les Ebdon

Kimberly S. Farah

Andrew S. Fisher

Joseph Sneddon

Nuclear Research Center-Negev Beer-Sheva, Israel

Geochemistry Division Geological Survey of Israel Jerusalem, Israel

Department of Environmental Sciences University of Plymouth Plymouth, England

Department of Science Lasell College Newton, Massachusetts

Department of Environmental Sciences University of Plymouth Plymouth, England

Department of Chemistry McNeese State University Lake Charles, Louisiana

vii


PREFACE

Volume 3 of Advances in Atomic Spectroscopy continues to present cutting edge reviews and articles in atomic spectroscopy as did the previous two volumes in this series.

Chapter 1 of this volume is devoted to plasma source mass spectroscopy, in particular inductively coupled plasma mass spectrometry. This was proposed in the early 1980s and has been commercially available since the mid-1980s. It has been suggested that it will be the dominant force for trace and ultatrace metal determination in the coming years. This chapter describes the basic theory, instrumentation, sample introduction techniques, and selected applications.

Chapter 2 covers simultaneous multielement atomic absorption spectrometry, mostly with graphite furnace atomization but, where appropriate, with flame atomization. Atomic absorption spectrometry has been around since the early to mid- 1950s and is a well-established and accepted technique for trace and ultratrace determination of elements. However, it is primarily regarded as a single-element technique. The need to perform simultaneous multielement analyses became a need and a desire in the early 1970s (the resurgence of atomic emission spectrometry with the inductively coupled plasma at this time). Atomic absorption spectrometry was slow to respond to the challenge of multielement analyses, with most work from the early 1970s to the late 1980s using laboratory-constructed or modified systems. However, since the late 1980s through to the present time, simultaneous multielement atomic absorption spectrometry has attracted interest. This chapter

x PREFACE

describes instrumentation and applications of simultaneous multielement atomic absorption spectrometry.

Chapter 3 describes the direct current arc and plasma jets. Direct current arc and plasma spectrometry has been around for a number of years but is still an integral and indispensable method for determining metals in solids and liquids in many laboratories. This chapter describes the arc and plasma, and the physical and chemical interferences of the sample and its trace elemental constituents in the direct current discharge and their correlation with spectral line intensities of each trace element. The authors describe their experiences in the determination of trace elements in refractory-type samples such as uranium, thorium, and plutonium oxides, rare earth oxides, rock phosphates, silicate rocks, aluminum and titanium oxides, and molybdenum and tungsten oxides.

Chapter 4 describes basic principles, design, instrumentation, evaluation, characterization, and selected applications of the use of a single-stage impactor combined with a graphite furnace for the direct collection of metals in air and subsequent determination by atomic spectroscopic methods, primarily atomic absorption spectrometry. The advantage of this type of system is the ability to determine low concentrations of metals, in the ng/m 3 range, within a few minutes.

Joseph Sneddon Editor

PLASMA SOURCE MASS SPECTROSCOPY

Andrew S. Fisher and Les Ebdon

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

I. Introduction and Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . 2

A. Interferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

B. Isotope Ratio Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

II. Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

A. Sample Introduction Systems . . . . . . . . . . . . . . . . . . . . . . . . 8 B. The Plasma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

C. The Interface Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

D. The Ion Lenses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

E. Mass Analyzer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

F. Electron Multiplier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 G. Vacuum Pumps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

III. Sample Introduction Techniques . . . . . . . . . . . . . . . . . . . . . . . . . 16

A. Laser Ablation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

B. Electrothermal Vaporization . . . . . . . . . . . . . . . . . . . . . . . . . 18

C. Slurry Nebulization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

D. Flow Injection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 E. Chromatography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

F. Hydride Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Advances in Atomic Spectroscopy Volume 3, pages 1-31 Copyright �9 1997 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0072-8

2 ANDREW S. FISHER and LES EBDON

IV. Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 V. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

ABSTRACT

The use, the instrumentation, and some of the applications of inductively coupled plasma mass spectrometry (ICP-MS) is described. A brief description of some of the basic theory is followed by an outline of the different components and their functions and a more substantial description of the different methods of sample introduction and their inherent advantages and disadvantages. The use of survey analysis and isotope dilution analysis has also been described. A description of numerous applications involving different methods of increasing sensitivity or reducing interference effects has also been included. Conclusions and a prediction of possible future trends have also been made.

!. I N T R O D U C T I O N A N D BASIC THEORY

The concept of plasma source mass spectrometry was first proposed by Gray and in collaboration with Fassel's research group the inductively coupled plasma was identified as the most suitable source. Inductively coupled plasma mass spectrometry (ICP-MS) was first demonstrated in Fassel's laboratory in the mid- 1970s (Houk et al., 1980; Date and Gray, 1981). It is a marriage between two already successful techniques (ICP and MS). The main problem encountered during this coupling was the interface between the two. The ICP part is at atmospheric pressure, whereas the MS is under high vacuum. This interface region has been developed and improved over several years and this will be discussed in a later section (Section II.C).

There are several readable summaries of ICP-MS and ICP in general. Books by Boumans (1987) and by Montaser and Golightly (1992) give excellent accounts of the theory behind ICP, and a book by Date and Gray (1989) summarizes the early developments of ICP-MS. This latter reference also contains a very large number of applications. Other publications that are of use include the books by Thompson and Walsh (1989), the handbook of ICP-MS edited by Jarvis, Gray, and Houk (1992) and the Royal Society of Chemistry monograph by Evans et al. (1995). In addition to this, there are journal articles that explain in simple terms the use of such instruments (Ebdon and Evans 1988).

There are several types of instruments available commercially. The standard instrument has a quadrupole-based mass spectrometer and costs in the region of $150,000 to $200,000. These are low resolution spectrometers (approximately 0.5 daltons), but are sufficient for the vast majority of applications. For some applications, a more highly resolving spectrometer is required. Such instruments with magnetic sector mass spectrometers are available but at much higher cost (approxi-

Plasma Source Mass Spectroscopy 3

mately $400,000). Much of the most recently introduced instrumentation is either of the high resolution type or reduced size, i.e. benchtop versions of quadrupole instrumentation.

An ICP-MS instrument may be used to obtain concentration data for a large number of analytes (approximately 70) in a very short period of time. It has several advantages over other trace element techniques, and may be used in several different ways. For the survey (semi-quantitative) method of analysis, the concentration of approximately 70 analytes may be determined simultaneously to within a factor of 3 of the true concentration. Although this method is not particularly accurate, it does serve to identify contaminants in a previously unknown sample. Once identified, these contaminants may be determined more accurately using fully quantitative software with calibration standards in the normal way. In addition, most instruments have single-ion monitoring and time-resolved analysis facilities. These are software packages that monitor the signal at one isotope for single-ion monitoring or at several isotopes quasi simultaneously for time-resolved analysis. They are most useful with transient signals such as those obtained with laser ablation, flow injection, electrothermal vaporization, or when chromatography is being coupled with ICP-MS, although there are other applications. Another advantage of ICP-MS is ultratrace sensitivity. Modern ICP-MS instruments have limits of detection superior for most analytes than even electrothermal atomic absorption spectrometry. As well as having ultratrace detectability (limits of detection < 0.1 ng m1-1 for many analytes), it is also multielement and has a large linear working range (manufacturers claim 8-10 orders of magnitude). Another advantage is that it can supply isotopic information. This can be especially useful for analytes such as lead whose isotopic ratio varies according to geological origin. The technique also has some disadvantages, the most obvious of these being the high purchasing and operational costs of the instruments.

The very basic principles of an ICP-MS instrument are that the sample enters a plasma and is ionized. The ions are then taken from atmospheric pressure through nickel cones into the interface region which is under partial vacuum. From this region, the ions are directed using ion lenses under a stronger vacuum to a mass spectrometer which is under still stronger vacuum. The mass spectrometer sorts the ions into the mass-to-charge ratio and the ions are then detected by an electron multiplier. A more detailed description of the method of work of each of the components will be described later in Section II.

For the majority of applications, the sample is introduced to the plasma via the sample introduction system consisting of a peristaltic pump, a nebulizer, a spray chamber, and a torch. In the plasma the sample is rapidly desolvated, atomized, and ionized. The extent of ionization depends on the ionization energy of the respective analytes. The ionization energy for argon (the most commonly used plasma gas) is 15.76 eV. Any analyte with an ionization energy below this value will be at least partially ionized and hence available to be detected. For most metallic analytes the ionization energy lies in the range 5-10 eV. For analytes with a low ionization

4 ANDREW S. FISHER anti LES EBDON

energy (e.g., caesium), the ionization is close to 100%; however arsenic which has a much higher ionization energy may only be 30-40% ionized. The extent of ionization obviously has some effect on the sensitivity of the analyte. Other prospective analytes such as the halogens (ionization energies 11.8-17.4 eV) are either very insensitive or completely undetectable. The extent to which the various species within the plasma are ionized may conveniently be described by the Saha equation shown below; although it must be stated that this equation assumes the plasma is in local thermal equilibrium, which most plasmas are not,

NONe _ (2xmekT) 3/2 2zij e -eHkr

Naj h 3 Z aj

where: N U = ion concentration of species j; Naj = atom concentration of species j; N e = free electron concentration; m e = electron mass; k = Boltzman's constant; k = Planck's constant; zij = partition function of ions of species j; Zaj = partition function of atoms of species j; ej = ionization energy of species j; and T = ionization temperature.

A. Interferences

Although, when first produced, these instruments were regarded as being interference-free, this has proved not to be the case. A number of interferences and types of interference exist. There are basically four types of interference encountered using ICP-MS. These are termed isobaric, polyatomic, doubly charged, and interferences observed arising from signal drift or decreased/increased nebulization efficiency. There have been several reviews on the different types of interference and methods of overcoming them (Vanhoe et al., 1994a; Evans and Giglio, 1993; Sah, 1995). Examples of each of these types of interference and how they are overcome are given below.

Isobaric Interferences

These are interferences that occur when more than one analyte has the same nominal mass. Examples of this include 4~ on 4~ and l13In on l l3Cd. Such interferences may be overcome using alternative isotopes, such as 44Ca, but this is a far less abundant isotope and therefore the sensitivity is limited. Another method of overcoming this problem is to use a high resolution spectrometer. As explained earlier, this is an expensive solution.

Polyatomic Interferences

This is by far the most troublesome type of interference. There are a large number of polyatomic ions that interfere with numerous analytes. The majority of interferences occur below mass 80, although some oxide based ones do exist at higher mass, e.g. for the rare earth elements. Some of the more common interferences are


shown in Table 1. The interfering ions tend to come as part of the matrix, part of the solvent, or are entrained from the air. There are several possible ways to overcome the interferences. The use of an alternative isotope is the obvious one, but some analytes (e.g. arsenic) are monoisotopic. Judicious tuning of the ion lenses and optimization of torch position, forward powers, and injector (nebulizer) gas flow lead to a decrease in the interferences, but do not always eliminate them completely. The use of alternative gases bled into the nebulizer flow has had great success in removing interferences arising from chloride (Branch et al., 199 la; Hill et al., 1992a), oxides (Hill et al., 1992b; Ebdon et al., 1993a) sulfates and phosphates (Ebdon et al., 1994b). Some authors have also bled nitrogen into the coolant gas flow to overcome interferences (Lam, 1990).

Desolvation of the sample will also decrease interference effects derived from the solvent, e.g. oxides, nitrogen-containing interferants, and hydrides. Cryogenic desolvation has been used by Alves et al. (1992) with great effect. By cooling the sample to -80 ~ oxide ratios for many analytes were decreased to 0.02-0.05% while ArO § ArC1 +, and C10 § were also decreased. Desolvation has been achieved in a variety of other ways. The most common method is to use a cooling jacket on the spray chamber, but other methods include the use of Peltier coolers (Hartley et al., 1993), gas permeable membranes (Branch et al., 1991b; Botto and Zhu, 1994; Tao and Miyazaki, 1995) and a mixture of all of these. At least one instrument manufacturer produces a nebulizer that specializes in desolvating organic solvents.

Table 1. A List of Some of the More Common Polyatomic Ion Interferences M/Z Element Abundance (%) Interfering Ions

47 Ti 7.72 PO t 48 Ti 73.48 32SO§ POH § 51 V 99.76 35C10 +, 34SOH+ 52 Cr 83.76 4~ 35C1OH +

53 Cr 9.55 37C10 § 54 Fe/Cr 5.92/2.38 4~ +, 37C1OH+ 55 Mn 100 4~ 56 Fe 91.52 4~ + 63 Cu 69.09 4~ § 64 Zn 48.89 HPO~, 32SO~, 32S~, 63Cull +

65 Cu 30.91 H32SO~ 69 Ga 60.16 37C10~ 75 As 100 4~ 76 Se 9.12 36Ar4~ + 77 Se 7.58 4~ 78 Se 23.61 4~ + 79 Br 50.54 38Ar4~ + 80 Se 49.96 4~176 +

6 ANDREW S. FISHER and I.ES EBDON

Membrane separators also have the effect of removing interferences. Tao and Miyazald found that the levels of ArO § and C10 § were decreased by one and two orders of magnitude, respectively, while the CeO+:Ce § ratio was decreased to 1.1 x 10 -3. The Ba2+:Ba + ratio was also decreased to a larger extent than that found using a cooled spray chamber. Desolvation devices made inhouse have also been used. Tittes et al. (1994) have described such a device and found that interferences arising from chloride (e.g., C10 § C1OH § ArC1 § CI~, and C10~) were largely overcome. The detection limit for V in 0.4 M HC1 was found to be improved by a factor of 25.

Doubly Charged Interferences

This occurs only when an element which has a very low second ionization energy is present in the sample. The most problematic element for this is barium. This element produces Ba 2+ ions which, when in high concentration, can interfere when determinations of gallium are being made at m/z 69. This can be circumvented by determining Ga at its less abundant isotope at rn/z 71. Alternatively, if barium is the analyte, erroneously low results can be obtained if the double ionization is not prevented. This can be achieved by optimization of power etc.

Signal Drift

The signal may drift over a long-term period if the temperature of the room changes. This is the reason that instruments should always be placed in a temperature-controlled room. Short-term signal drift occurs when the dissolved/suspended solids content of the sample changes. This would lead to a change in the nebulization efficiency and hence to the amount of sample reaching the plasma. Signal drift can be accounted for by the addition of an internal standard to all samples and standards. The prime requisites for an internal standard are that it should not be present naturally in any measurable amount in the sample, it should preferably have an ionization energy close to the analyte(s) of interest, and should be reasonably close in mass to prevent any mass discrimination effects. Some of the more common internal standards include Co, In, and Rh. The use of an internal standard for normal analyses is now routine, but care must be taken if chromatography is being coupled with the ICP-MS since an internal standard is not always readily available. Simi- larly, for slurry introduction, an internal standard will not necessarily correct for the transport efficiency of the slurry particles to the plasma if these particles are too large. An added disadvantage of samples containing high dissolved solids or an organic solvent is that gradual blocking of the nebulizer, torch, or cones may occur. This will inevitably lead to a decrease in signal.

Mass Discrimination Effects

In addition to the types of interference listed above, mass discrimination effects may also occur. This happens especially when there is a high concentration of an easily ionized heavy element. The most badly affected analytes are the lighter


elements that have a high ionization potential. Careful optimization of the plasma operating conditions may overcome these problems to a large extent. It must be noted that the effect is dependent on the absolute amount of the heavy element rather than the molar ratio to the analyte. Therefore, if sensitivity permits, the effects can also be partially overcome by diluting the sample.

B. Isotope Ratio Analysis

Applications of isotope ratio measurements fall into one of two classes. Unspiked analyses are used primarily in the geological sciences for the determination of the ages of samples, measuring sedimentation rates and determining paleo- and mag- matic-temperatures, or in the nuclear industry where applications include the monitoring of isotopic composition during the production of enriched uranium and monitoring for environmental contamination. Spiked or isotope dilution analysis is used to determine the concentration of an analyte by adding a known concentration to the sample. In ICP-MS, stable isotopes are normally used. In a way it is the ideal form of internal standardization because any matrix-induced interference will affect all the isotopes equally. For this type of analysis it is necessary to know the natural abundance of each of the isotopes in the sample to ensure that the spike is as different from it as practical. The advantage of this technique is that once the sample has been spiked, it is not necessary to analyze it quantitatively. Another advantage is that it can yield results with exceptionally good precision (typically < 1%). This method also compensates for any losses of analyte during the sample preparation procedures. If the amount of spike isotope added is known and by measuring the isotope ratio in the spiked sample, the analyte concentration can be calculated using the formula below (Thompson and Walsh, 1989),

C ~ M s K ( A s - BsR )

W(BR - A)

where: C is the analyte concentration (~tg g-l); M~ is the amount of isotope spike (~tg); A is the natural abundance of the reference isotope; B is the natural abundance of the spike isotope; A~ is the abundance of the reference isotope in the spike; B~ is the abundance of the spike isotope in the spike; K is the ratio of natural and spike atomic weights; W is the sample weight (g); and R is the measured reference/spike isotope ratio after spike addition.

This technique has found applications in several different fields, including geological, environmental and health sciences, and the nuclear industry. One of the problems associated with the technique is that it relies on an analyte having several isotopes that are stable. When isotopic abundance can vary because of radioactive decay, as is the case with lead and uranium, then problems may occur unless a "double spike" technique is used, whereby two accurately mixed spikes are added


to a portion of the sample and then the spiked portion and the unspiked portion are analyzed separately and the data is normalized to the ratio of the spikes.

There have been numerous applications of isotope dilution published. A few of them are outlined below. A study of the systematic and random errors for the determination of fission products and actinides has been made by Garcia'Alonso (1995). Systematic errors arising from mass discrimination effects, detector non- linearity, and isobaric interferences were corrected.

Blood lead has been determined by Paschal et al. (1995). These authors spiked blood with NIST 983 radiogenic lead isotopic standard enriched with 2~ to 92.15%. After acid digestion of the blood in a microwave oven, the isotope ratio of 2~176 was measured. The authors claimed that exceptionally accurate results were obtained and up to ten whole blood samples could be analyzed in a day. Another application of isotope dilution has been demonstrated by Enzweiler et al. (1995). These authors determined Ir, Pd, Pt, and Ru in sodium peroxide fusions of geological materials. Analysis of several certified reference materials (WGB-1, TDB-1, UMT-1, WPR-1, WMG-1, and SARM-7) yielded results in excellent agreement with certified values.

Elemental speciation with liquid chromatography ID-ICP-MS has been addressed by Heumann et al. (1994). The isotope spike was applied in one of two ways. When the chromatographic system was well-characterized and the species were well-defined, a spike was added to the sample before the separation stage. If the species were not well-defined, a continuous on-line introduction of a spiked solution was used. For the species specific spikes, the method was applied to the determination of iodide and iodate in mineral water. The range found was 0.5-20 ng ml -l, and the precision of the analysis was 2%.

!1. INSTRUMENTATION

The instrumentation may be conveniently split into several sections, starting from the sample introduction systems and ending with the electron multiplier.

A. Sample Introduction Systems

There are a number of different ways that samples can be introduced to an ICP-MS instrument. The most common method is by aspiration of aqueous-based samples, but other methods such as laser ablation (LA), electrothermal vaporization (ETV), flow injection (FI), hydride generation (HG), chromatography, and slurry nebulization also exist. These sample introduction techniques will be discussed in more detail in Section III.

The sample introduction system for simple aqueous samples consists of a peristaltic pump, a nebulizer, a spray chamber, and a torch. In general the nebulizer produces a fine mist of droplets that enter the spray chamber which acts as a


dampener for pump noise and droplet size filter. The smallest droplets (typically 1-2% of the total solution) reach the torch and enter the plasma.

Nebulizers

There are several different types of nebulizers available commercially. These include: concentric glass pneumatic nebulizers such as the Meinhard; V-Groove nebulizers such as the De Galan, Babington, and Ebdon (Ebdon and Cave, 1982); direct injection devices; and ultrasonic nebulizers (Bear and Fassel, 1986). Various frit (Layman and Lichte, 1982) and grid (Brotherton et al., 1987) style nebulizers have also been developed.

The type of nebulizer used will depend on the application. The Meinhard nebulizer is effective for aqueous applications, but may become blockecl if the solution has a high dissolved/suspended solids content. The Meinhard operates with low noise levels since the solution need not necessarily be pumped to it by a peristaltic pump. The frit and grid style nebulizers have greater nebulization efficiency (and hence a higher percentage of the sample reaches the plasma leading to reduced wastage of sample), but are also prone to blocking if samples with high solids are aspirated. The V-Groove nebulizers are far more robust. They are often made of a polymer, thus are less fragile than the glass Meinhard. For example, the Ebdon nebulizer is made of corrosion-resistant Kel-F and has proved to be virtually unblockable, therefore it is used when the sample is corrosive or if it contains a high dissolved/suspended solids. The disadvantage with V-Groove nebulizers is that they rely on a peristaltic pump to transport the sample. This means that they are more liable to pump noise, but less prone to variations in uptake rate due to sample viscosity.

Direct insertion (or direct injection) devices lead to far higher sensitivity because close to 100% of the sample reaches the torch (i.e. there is no spray chamber). Since there is no spray chamber, the dead volume is very much decreased and the time of analysis is decreased because there does not have to be a wash-out period between samples. This has advantages for chromatographic applications, but the sample flow rate has to be low (typically < 100 ~tl min -1) to prevent plasma extinction. In addition, the DIN can be easily blocked if a sample or a chromatographic mobile phase has a high dissolved solids content. There have been several papers published using this type of nebulizer. A DIN has been evaluated in a paper published by Zoorob et al. (1995). Other authors to have used DIN's include Shum et al. (1992), Shum and Houk (1993), and Powell et al. (1995). It should be remembered that one of the roles of a spray chamber is to dampen noise and fluctuations in sample uptake, therefore nebulizers with no, or reduced, volume spray chambers may result in noisier signals.

Other types of nebulizers such as the thermospray (Vanhoe et al., 1994a) have also been developed. The advantages of this device are that it improves the sensitivity by an order of magnitude and improves the M+:MO + and M+:M 2§ ratios by a factor of 2.5 when compared with a pneumatic nebulizer used in conjunction


with a spray chamber. Vanhoe et al. have also produced other papers on this topic (Vanhoe et al., 1994b and Vanhoe et al., 1995). Other authors have also reported the use of a thermospray device (Arpino, 1992; Koropchak and Veber, 1992).

The hydraulic high pressure nebulizer (HHPN) has been described (Jakubowski et al., 1992). The same research group has also used this type of nebulizer for speciation of chromium (Jakubowski et al., 1994). Once combined with an effective method of desolvation, the HHPN increases the sensitivity for most analytes when compared with a conventional pneumatic nebulizer. The results for the speciation yielded detection limits of down to 1 ng ml -~ for the different chromium species.

Although ultrasonic nebulizers were developed in the 1920s, they have been used by several authors to improve the detection limits in atomic spectrometry (Woller et al., 1995). Detection limits for many analytes are improved by factors of 5 to 50 when compared with a nondesolvated pneumatic nebulizer, or 3 to 8 times when desolvation is used. Desolvation is normally used for ultrasonic nebulization, otherwise the increased solvent loading of the plasma leads to plasma cooling and possibly extinction. The increased solvent loading of the plasma would arise because of the smaller droplet size generated by this type of nebulizer (Tarr et al., 1992). Some commercial USN's consist of a temperature-controlled heated cell followed by a water-cooled condenser to obtain dry aerosol particles. Ultrasonic nebulizers are also prone to blockage by samples with high dissolved solids and may suffer from troublesome memory effects. Yang and Jiang (1995b) have also reported on the use of a USN recently.

High efficiency nebulizers have also been developed (Sang-Ho Nam et al., 1994). This nebulizer operates at a very low solution uptake rate (10-100 ~tl min -1) and its analytical performance in terms of detection limits (ng I-l), precision (0.7-4%), and the M+:MO § and doubly charged species ratios compared favorably with a conventional nebulizer.

An excellent review containing 209 references on the theory, mechanism of operation, and operating characteristics of pneumatic nebulizers was produced by Sharp (1988a). In a more recent review, the noise characteristics produced by the aerosols of different ICP nebulizers has been published (Luan et al., 1992).

Spray Chambers The basic function of the spray chamber is to ensure that only the smallest

droplets reach the plasma. A review (140 references) of the fundamental processes occurring within spray chambers has been produced by Sharp (1988b). This review also covers references that compare nebulizer and spray chamber types. Droplets in excess of 5-8 ~tm are effectively removed from the system in the spray chamber and drawn or are pumped to waste. There are several types of spray chambers that may be used, but the large majority of applications use a double-pass Scott-style spray chamber. These are frequently made of glass but variants exist for when corrosive materials such as hydrofluoric acid are being analyzed. These variants are usually made from PTFE. The drawback with a double-pass spray chamber is that


it has a high internal surface area and a large volume. For chromatographic applications this can lead to peak broadening. A single pass spray chamber has less volume and a smaller internal surface area, so is often used for coupling high performance liquid chromatography (HPLC) or flow injection (FI) with ICP-MS. A large number of custom-made spray chambers have also been produced. Wu and Hieftje have developed a cyclone style spray chamber that has a substantially higher transport efficiency than many other spray chambers.

A water jacket is incorporated into most commercial spray chambers. This is to cool the aerosol and hence lower the solvent loading of the plasma. This can be important when the solvent is water since large amounts of oxygen entering the plasma can lead to deleterious effects and interferences. Decreasing the solvent loading can be even more important when organic solvents are being used, because many solvents cause quenching of the plasma and increased interferences. A detailed study on the effects of organic solvents on plasmas has been made by Boorn and Browner (1982). Although this work was performed on an emission spectrometer, the overall conclusions hold true for MS instruments.

Torches

There are many types of torches available, but most are basically similar in design. The Fassel-based torch (18 mm in diameter) is used far more than the larger Greenfield type. The internal diameter of the injector (typically 2 mm) may differ depending on the sample type. For high solids, a wider bore may be necessary (3 mm), whereas for organic solvents a narrow bore (1 mm) will reduce solvent loading in the plasma. Demountable torches have the advantage that a variety of injector bore sizes may be used in an attempt to optimize the system. Low-flow torches in which the gas flow may only be half that of conventional torches have also found some use (Evans and Ebdon, 1991).

Other Methods of Sample Introduction Many manufacturers produce "bolt-on" devices that may be bought as accesso-

ries and are easily interchangeable. Such devices exist for LA, ETV, FI, and HG, although many laboratories use equipment that they already possess. Certainly for FI and HG applications it is common for simple manifolds to be prepared in-house rather than opting for the more expensive manufacturers products. For LA and ETV applications, modification to already existing devices is common, but it is more usual to invest in the manufacturers product for software compatibility reasons.

B. The Plasma

Torch boxes and RF generators differ between manufacturers. Some manufacturers use a standard 27.12-MHz generator, whereas others make use of the extra stability and tolerance of 40.68-MHz generators. Virtually all generators are solid state, but older instrumentation made use of the Henry generator. Most generators


supply power up to 2000 W, but for most applications a power of 1200-1500 W is sufficient. Exceptions include the low power plasma work developed by Evans and Caruso (1993) where powers of less than 100 W may be used. This will be discussed in more detail later.

The tolerance of generators to organic solvents depends on the design of the instrumentation. Most modem ICP-MS instruments can withstand very high per- centages of solvents such as methanol, ethanol, and acetonitrile without plasma extinction, although for many, a bleed of oxygen into the nebulizer gas flow is necessary to prevent the buildup of carbon on the cones, which will lead to excessive signal drift and ultimately to complete blockage.

The basic principles of the ICP have been given in several texts (e.g., Boumans, 1987). Basically, a flow of argon gas is seeded with free electrons via a Tesla coil. This produces a potential that overcomes the dielectric resistance of the gas. The load coil produces a fluctuating magnetic and electric field which sustains the plasma. These fields couple energy into the plasma by accelerating free electrons into a region within the load coil. These electrons then transfer energy to other plasma species by collision. This produces further breakdown and an avalanche effect is produced. The argon then continues collisional energy exchange and a fireball (plasma) is produced. The temperature reached in the plasma ranges between 10,000 K at the hottest part to 6000 K in the sampling part.

C. The Interface Region

The plasma acts as an ion source, i.e. the constituents of the sample are dried, atomized, and then ionized. The ionized analytes then pass from the atmospheric pressure plasma to the interface region (also called the expansion chamber). A complete diagram of the plasma and interface region is shown in Figure 1. A comprehensive description of ion sampling in plasma mass spectrometry has been given by Douglas and French (1988). Douglas has also given an account in a book

Figure 1. A diagram of the plasma and interface region.


edited by Montaser and Golightly (1992). The sampling and skimmer cones are usually made of nickel because it is inexpensive, relatively easy to machine, and is durable, but any material with high conductivity will suffice. Other materials that have been used include aluminum, copper, and platinum. Platinum tipped cones are still used if the sample is likely to be corrosive to the nickel ones, e.g. if it contains large amounts of phosphate or sulfate.

Behind the sampling cone a rotary pump produces a partial vacuum. The ionized sample in the form of a gas passes through the aperture in the sampling cone and on entering this region of lower pressure it accelerates until it exceeds the speed of sound. The temperature also drops dramatically. Under these conditions, the kinetic energy of the sample is converted into a directed flow along this axis. In effect, a free jet is formed that is bounded by a shock wave known as "barrel shock". Barrel shock helps prevent the gas jet from mixing with any surrounding gas and hence helps prevent the formation of molecular species. A second shock wave exists across this axis. This is formed when the expansion is halted by the background gas pressure. This second shock wave is called the Mach disc. TheMach disc's position is dependent on the diameter of the aperture in the sampling cone and on the pressure. Typically, the Mach disc is approximately 10 mm behind the aperture. Behind the Mach disc the ion beam becomes subsonic again and may mix with any surrounding gas. To prevent this as much as possible a second nickel cone (the skimmer) is placed at a distance of just over 6 mm away from the aperture on the sampling cone. This then allows the gas jet to pass through to the next stage of the spectrometer. The condition of the aperture on the skimmer cone is also vital. If it is misshaped then further shock waves will be caused and this will attenuate the transport of the gas jet through the orifice. Once passed the skimmer, the gas jet becomes random and requires focusing onto the detector by a set of electrostatic ion lenses.

Cleaning of the cones has been shown to assist in the prevention of the formation of some polyatomic interferences. Secondary discharge has also been shown to cause interference effects that have a severely detrimental effect on the performance of the instrumentation. Secondary discharge is caused by an excessively high plasma potential causing discharge to occur between the plasma and the sampling orifice. A bad discharge causes crackling of the plasma and is characterized by bright white emission from the gas flowing into the orifice. The overall result is that the sampling orifice becomes ablated (i.e. it becomes much larger), leading to decreased vacuum in the expansion region which therefore leads to increased interference effects such as more doubly charged ions and more polyatomic ions. There have been several methods used to overcome this problem. Among these are careful optimization of the sampling depth (i.e. from what part of the plasma the ions are sampled) using a low nebulizer gas flow (<11 min -l) and reducing the solvent loading of the plasma.

The account given above is a very simplified version of the events going on in the interface region. Intense research into the positioning and dimensions of the


cones is continuing. Since the inception of the technique the sensitivity of the analysis has been improved over 10-fold by refining the engineering of the interface region. Much fuller accounts of molecular beams and the sampling of them and of the way in which polyatomic interferences may be formed in this region are given in the literature (Campargue, 1984; Olivares and Houk, 1985; Douglas and French, 1988, Vaughan and Horlick, 1990). Another readable account is given in a monograph written by Evans et al. (1995).

D. The Ion Lenses

The function of the ion lenses is to focus as many ions as possible from the cloud formed behind the skimmer cone through a differential pumping aperture into an axial beam of circular cross section at the entrance to the quadrupole mass analyzer. The ion lenses are basically metal rings with electric potentials applied to them. They are housed in the intermediate region of the instrument. There are several different systems in use, but all have a photon stop present on the axis to prevent photons from the plasma reaching the detector and adding to the background signal. The diameter of the photon stop is another factor that determines the geometry of the ion lenses. The geometry of the lens system used by one manufacturer is shown in Figure 2. The negative voltage on the extraction lens attracts the positive ions in the sample cloud and accelerates them towards the lens stack. Negative ions are repelled and the neutral atoms pass to waste in the vacuum pump. The ion beam is then collected on the "collector electrode" prior to being focused through the differential pumping aperture by lens 1 and 2. Lens 3 and 4 then refocus the ion beam into the entrance aperture of the quadrupole mass analyzer.

It must be noted that the region between the skimmer cone and the collector has a very high "space charge" and a high background pressure of neutral species. This makes the calculation of ion trajectories difficult. The optimization of the ion trajectories obviously enables greater sensitivity to be achieved, therefore instru-

slide valve

,

'L4"~ L3"~ ~ ~ ~ ~ ' ~ ~ -r" . "~"skimmer cone

L1 extiactibri I Figure 2. A diagram of the interface region and the ion lens arrangement.


ment manufacturers have performed intense research into this area of the design. The effects of space charges etc. have been described in a series of papers by Tanner (Tanner, 1992; Tanner et al., 1994).

E. Mass Analyzer

The function of the quadrupole is to produce an electric field that selectively allows a stable trajectory for ions that have a narrow mass-to-charge ratio. Although it does not have as good a resolution as a magnetic sector instrument, it does have an acceptable sensitivity/resolution trade-off. Typically, it is comprised of four electrically conducting rods (12 mm diameter x 230 mm long) made of molybdenum that are arranged to produce an oscillating electric field between them. As the rods are arranged in a square, the electric field they produce between them is a good approximation of the ideal hyperbolic quadrupole field. Before the main mass analyzer there is a series of pre-rods that are only about 20-25 mm long. These rods are used to improve the transmission of the lighter ions and to prevent contamination of the main analyzer. Similarly at the end of the main analyzer, a set of rods are present to improve the extraction of the ions. The entire assembly is under high vacuum to ensure that there is no residual gas that can disrupt the ion trajectories by scattering, and hence causing decreased sensitivity. The basic principles of a quadrupole mass analyzer are beyond the scope of this text, but accounts of how they work are given elsewhere (Dawson, 1986). It is sufficient to say that the ions of the selected mass units "corkscrew" their way through the filter while the ions of unrequired mass units are deflected off this trajectory and are lost. Quadrupole analyzers have the advantages of being capable of scanning mass units very rapidly (up to 3000 mass units per second), and they are substantially cheaper and can withstand higher operating pressures when compared with magnetic sector spectrometers.

As discussed earlier, high resolution spectrometers that utilize a magnetic sector instead of a quadrupole mass selective filter also exist. These spectrometers are more highly resolving than the quadrupole-based instruments (they have a resolving power of about 5000). It must be noted though that sensitivity and resolution are not always compatible. Although the magnetic sector instruments are far more resolving, they are frequently unable to compete with the detection limits obtainable using quarupole-based instrumentation. There are however some applications that require high resolution to obtain accurate results, including the semiconductor industry, some geological applications, the nuclear industry, and laboratories that analyze certified reference materials.

F. Electron Multiplier

The ions transmitted from the quadrupole mass analyzer are detected by an electron multiplier. Again, the detailed description of how electron multipliers work is not appropriate in this text, but basically when a positive ion strikes the funnel


of the multiplier one or more secondary electrons are ejected from the surface and are accelerated down the tube. During the passage down the tube they collide with the walls dislodging further electrons and thus an avalanche effect quickly builds up. The curved nature of the tube prevents ionic feedback, i.e. the traveling of electrons to the beginning of the funnel to restart another avalanche effect. At the bottom of the tube the cloud of electrons leaves the base of the channel and is attracted by a collector electrode. The signal is therefore measured as an electrical pulse.

Recently, another type of detector has been developed. The use of this "active film multiplier" detector has become fairly widespread in the U.S. and it is envisaged that its use will spread further. It is a new type of discrete dynode multiplier that is coated with a material that is reputedly both resistant to chemical attack and which is stable in air. Other advantages this detector possesses include an increased total surface area (1100 mm 2 compared with 160 mm 2 in conventional electron multipliers), which should enhance its lifetime substantially and offer a large linear dynamic range.

The Faraday cup is also used by some manufacturers. This device is especially useful when high concentrations of analytes are being determined.

G. Vacuum Pumps

As mentioned earlier, the mass spectrometer from the interface region to the detector is under vacuum. The interface region is evacuated to typically 2 mbar by a single-stage rotary pump. Often two rotary pumps in series are used i.e. the output from one is the input for the second. This improves the vacuum further. The intermediate (the ion lenses) and analyzer (quadrupole and detector)regions are under much stronger vacuum. In older style instruments, diffusion pumps were used but more recently the large majority of ICP-MS instruments use turbomolecular pumps. This has several advantages. After cleaning the lens stack the turbo pumps may evacuate the inner chambers within only a few minutes, whereas the diffusion pumps would take several hours. This has the obvious advantage of decreasing the downtime of the instrument. Although cleaning the lens stack is normally a 3 to 6 monthly task, if samples containing organic solvents are in regular use. the procedure may have to be repeated every 2 to 3 weeks.

!!!. SAMPLE INTRODUCTION TECHNIQUES

As well as the conventional aspiration of aqueous samples, there is an array of other methods of introducing samples to the instrument. These include the direct analysis of solids by laser ablation, ETV or slurry nebulization; or the use of chromatography, hydride generation, and flow injection. The relative merits of each of these will be discussed in this section. An annual review of the advances of methods of sample introduction to ICP-MS instruments is provided in the Atomic Spectrometry Update


produced in each October issue of the Journal of Analytical Atomic Spectrometry. This review also describes the fundamental studies, theory, and, to some extent, the applications of ICP-MS.

A. Laser Ablation

Most manufacturers produce a laser accessory that simply bolts on to the front end of their instrument. A laser (often a Nd:YAG laser) is focused onto (or close to) the surface of a solid sample causing it to vaporize. The sample vapor is then swept by a flow of carder gas to the plasma for ionization and detection. The sample does not have to be conducting, so it has advantages over arc/spark technology. Laser ablation (LA) has enjoyed great popularity with geologists since it enables them to analyze rocks without resorting to traditional methods of sample destruction such as fusion, or acid digestion using hydrofluoric acid. Other advantages it has include the ability to analyze small samples and depth profiling (i.e. the ability to determine elemental constituents at different distances from the surface of the sample). The production of a dry vapor of the sample also leads to decreased interference effects that normally arise from the solvent (i.e. the production of oxides, hydrides etc.). There are, however, problems associated with the technique. As the laser only vaporizes very small areas of the sample, spurious results will be obtained if the sample is not homogeneous. Another very important disadvantage is that of standardization. The calibration standards have to be very closely matrix-matched with the samples and this can be very problematic. The use of certified reference materials as standards is common for LA-ICP-MS. Because of the difficulty of standardization, LA is used most frequently for qualitative analysis of samples.

As described earlier, the technique has had many applications in the geological field. A review of the applications for geological exploration has been published by Hall (1992). In a very comprehensive review (1019 references) the interaction of laser radiation with solid materials and its significance to analytical spectrometry has been discussed (Darke and Tyson, 1993). This review gives a simple explanation of the ablation process, an introduction to laser microprobes, and a large number of applications. Because this is such a comprehensive review, only a few of the more recent applications will be discussed here so that a feel of what is possible is obtained.

Laser ablation has been used to determine numerous analytes in several different geological samples. These include the rare earth elements in geological materials (Jarvis and Williams, 1993); chalcophile elements in rocks, soils, and sediments (Guo and Lichte, 1995); sulfide minerals (Watling et al., 1995); rocks (Lichte, 1995); garnet (Fedorowich et al., 1995); alkaline and rare earth elements in marine ferromanganese deposits (DeCarlo and Pruszkowski, 1995): and weathered marble (Ulens et al., 1994). As well as geological applications, LA has been applied to numerous other samples. These include the analysis of steel (Yasuhara et al., 1992), biological tissue (Wang et al., 1994), tree rings (Hoffmann et al., 1994), glass (Stix


et al., 1995), and teeth (Outfidge and Evans, 1995). Other applications that have been performed include the analysis of individual fluid inclusions by Shepherd and Chenery (1995), the analysis of powders (Raith and Hutton, 1994), and the laser vaporization of small volumes of solutions (Prabhu et al., 1993). Laser ablation has been used to obtain semiquantitative results in a paper by Cromwell and Arrows- mith (1995). A paper that reports the direct determination of metals in silver and gold without matrix-matched standards has been published by Kogan et al. (1994). This is a novel application, because as described earlier, the standards used for calibration normally have to be matrix-matched.

Some papers have also been published on hardware developments. Cousin et al. (1995) have developed an autofocus system that enables reproducible focusing of the laser.

Laser ablation using a microprobe has been used in several applications (Chenery and Cook, 1993; Fryer et al., 1995; Gunther et al., 1995). A microprobe has the advantage of permitting ablation from a much smaller site (10% of the size required by conventional laser ablation). This greatly facilitates the analysis of small individual grains, crystals, and powders. In another application a laser microprobe was used to find the three-dimensional distribution of precious metals in sulfide minerals.

Isotope ratio measurements of copper have been made by Allen et al. (1995). Precision was found to be 0.85%. This paper also utilized a novel twin quadrupole mass spectrometer. Having two spectrometers enabled the authors to analyze two masses simultaneously.

B. Electrothermal Vaporization

Electrothermal vaporization can be used to analyze liquid or solid samples. Some manufacturers produce an accessory, but many laboratories simply choose to modify existing hardware (Lamoureux et al., 1994; Wang et al., 1994). A review of ETV into ICP-MS has been produced by Carey and Caruso (1992). The overall benefits offered by this technique have been summarized in a paper produced by Beres et al. (1994). This paper also discussed the importance of vaporizer design and illustrated its conclusions with several applications. For the majority of ETV applications the sample is dispensed on a rod or platform made of graphite or other suitable material (tungsten or some other refractory metal), dried, ashed, and then vaporized. The vapor is then swept to the plasma by a flow of carrier gas. This technique has the advantage of separating the analytes from the matrix, which is removed during the ash phase. This obviously decreases the prospective interferences arising from phosphates, halides, sulfates etc. Since the sample is dry, the determinations are also free from interferences arising from solvents (e.g., oxides and hydrides). A study of the interferences in ETV-ICP-MS has been made by Shibata et al. (1993). These authors concentrated on the formation of oxides. The attenuation of oxide formation in ETV-ICP-MS has also been reported by Clemons


et al. (1995). These authors used a graphite torch injector to decrease the amount of oxide formation. Many applications have been produced recently. Vaporization from a metal platform has been achieved by Marawi et al. (1995); the vaporization of boron has been assisted by the addition of mannitol as a matrix modifier (Wei et al., 1995); rare earth elements, uranium, and thorium have been determined (Gre- goire et al., 1995); and As, Cd, Pb, and Zn have been determined in soils and sediments (Zaray and Kantor, 1995).

The analysis of solids in this way leads to greatly enhanced sensitivity as no sample preparation or dilution is required. An alternative approach is to place the sample on the end of a graphite rod and then insert the rod into the base of the plasma. This technique is not so common, although there has been a move recently towards the "electrically heated wire loop in torch vaporization" (Karanassios et al., 1995). Similarly, an assessment of direct solid sample analysis by graphite pellet ETV-ICP-MS has been made by Ren et al. (1995).

As well as being an analytical tool, ETV-ICP-MS has been used to elucidate vaporization mechanisms in electrothermal atomic absorption spectrometry. The mechanism of chloride interference has been discussed by Byme et al. (1993) and the mechanism of vaporization of uranium in a graphite tube has also been reported (Goltz et al., 1995). In the former paper, the use of ICP-MS allowed direct observation of the signals for manganese along with the matrix components during the ash (pyrolysis) and vaporization steps. The loss of Mn was reported to be due to vaporization of manganese chloride.

Exotic combinations of techniques have also been published recently. An ETV method of introducing slurries to an ICP-MS instrument has been reported by Gregoire et al. (1994). This technique also used a patented ultrasonic slurry agitator to ensure homogeneity of the sample. Electrothermal vaporization has also been used for isotope dilution analysis (Bowins and McNutt, 1994). This procedure enabled the determination of lead in blood to below the ng m1-1 level.

Although the technique of ETV-ICP-MS has proved very successful and sensitive, there are a few drawbacks. These include the very slow sample throughput, and the possibility of some refractory analytes not being vaporized quantitatively from the graphite rod. This has led to the use of matrix modifiers as in electrothermal atomic absorption.

C. Slurry Nebulization

Slurry nebulization has had wide acceptance as a means of sample introduction into plasma emission spectrometry, and its use has spread into plasma mass spectrometry. Slurry nebulization requires no extra instrumentation, although the use of a high solids nebulizer is obligatory. Slurry nebulization has several advantages. It eliminates lengthy sample preparation procedures, decreases the number of sample preparation steps, decreases the use of hazardous acids, and may decrease the amount of contamination. A major advantage is that slurries may be analyzed


using aqueous solutions as calibrants. A comparative study of aerosol sample introduction for solutions and slurries in atomic spectrometry has been made by Ebdon et al. (1989). There are several methods of making slurries, but basically, sample is powdered and then ground in the presence of an aqueous dispersant using either a ball mill or the "bottle and bead" method. The dispersant used will depend on the sample properties, but Aerosol OT and Triton X-100 are frequently used for carboniferous samples, whereas sodium pyrophosphate or sodium hexametaphos- phate are used for samples that are more inorganic in nature. The most important parameter for slurry nebulization is the particle size. Particles larger than 2 ktm are not transported to the plasma as efficiently as water droplets, hence low recoveries will be obtained if calibration is against aqueous solutions. The particle size may be decreased by lengthening the grinding time. The use of an internal standard may not compensate for the poor transport efficiency since the standard is normally in the aqueous phase.

There have been several applications of slurry nebulization ICP-MS. These include geological samples (Jarvis, 1992; Halicz et al., 1993; Totland et al., 1993) and environmental samples, e.g. sediments (Zaray and Kantor, 1995). A wide range of analytes may be determined in this way. Zaray and Kantor determined As, Cd, Pb, and Zn, whereas Totland et al. determined platinum group metals and gold. For some analyses, slurries have been analyzed using ETV-ICP-MS (Gregoire et al., 1994; Fonseca and Miller-Ihli, 1995; Zaray and Kantor, 1995).

Some studies have concentrated on the theoretical aspects of slurry introduction. A paper by Goodall et al. (1993) gave a good theoretical basis for slurry nebulization into plasmas. Fonseca and Miller-Ihli (1995) concentrated on transport studies of slurries in ETV-ICP-MS and Hartley et al. (1993) found that desolvation improved the transport and atomization efficiencies of slurries.

D. Flow Injection

Flow injection has been used by numerous workers to obtain a variety of effects. It has proved to be a very popular and successful method of sample introduction because it is so easily coupled with ICP-MS and is so versatile. The first authors to couple flow injection with ICP-MS were Dean et al. (1988). Basic flow injection has been used to introduce samples with a very high solids content (Stroh et al., 1992; Richner, 1993), or to introduce organic solvents (Hill et al., 1992c,d). The use of flow injection for these samples prevents a continual loading of the plasma with high solids, thereby preventing torch/nebulizer/cone blockage; or, in the case of the organic solvents, prevents continually high-reflected powers. Richner compared the limits of detection for samples with increasing solids loading. He found that LOD's at the ng g-1 level were obtained when samples containing 3% m/m Ni were analyzed. This was attributed to the very low dilution factor used. Precision was 2% on 13 replicate 200-~d injections. Organic solvents have been introduced into ICP-MS by flow injection to maintain plasma stability (Hill et al., 1992c).


Trimethylgallium and methyllithium were dissolved in diethyl ether and aliquots of this solution (10-25 ~tl) were introduced to a 2% nitric acid stream. Limits of detection in the ng ml-I range were obtained for several analytes including A1, Cu, In, Pb, and Zn. Detection limits were improved further by the use of a desolvation membrane. The authors concluded that this approach improved detection limits, increased sample throughput, decreased memory time, avoided sample pretreat- ment, and therefore improved the safety aspects. Flow injection is also useful when only a limited amount of sample is available.

If microcolumns of exchange media are used, analytes may be preconcentrated and the matrix can be eliminated and hence interferences removed. There have been a large number of applications papers that have used this approach. Examples include the determination of trace analytes in seawater (Orians and Boyle, 1993; Bloxham et al., 1994), concentrated brines (Ebdon et al., 1993a), biological samples (Ebdon et al., 1993b; Ebdon et al., 1994; Huang et al., 1995), steels (Coedo and Dorado, 1994; Coedo and Lopez, 1994), iron (Coedo et al., 1995), waters (Dadfar- mia and McLeod, 1994; Fairman and Sanz-Medel, 1995; Gomez and McLeod, 1995), high purity zinc (Sayama et al., 1995), soils (Hollenbach et al., 1994), and geological materials (Eaton et al., 1992). Many of these papers use ion exchange resins to retain the analytes while the bulk matrix is eluted to waste. This obviously has the advantage that when the analytes are eluted to detection, the large majority of potentially interfering species have already been removed. Large preconcentration factors are also obtainable using this methodology. If 5 ml of sample is passed through the column and then the analytes eluted to detection in 250 ~tl of eluent, a theoretical preconcentration factor of 20 is obtained. Often the preconcentration factor will be substantially higher than this; e.g. Gomez and McLeod achieved a factor of 160 when they determined gold in natural waters. Many exchange or adsorption media have been used. Examples include chelating resins (especially for transition metals, Cd, Pb, and Zn), acidic alumina (for As, Se, V, and Cr), ion exchange resins and sulfydryl cotton (for Hg and Au). Work has also been performed in which flow injection of samples into a gaseous carrier (air) was achieved (Beauchemin, 1993).

Some workers have again been coupling several well known techniques together to improve the overall detection limit or precision. Lu et al. (1993) have coupled flow injection with isotope dilution ICP-MS to determine Cd, Cu, and Pb in biological and environmental samples; Colodner et al. (1993) have used the same technique to determine Re and Pt in waters and Ir in sediments. Stroh and Vollkopf (1993) have described a method of flow injection-hydride generation ICP-MS for the determination of As, Hg, and Sb in water and seawater. Their results were validated by the analysis of several certified reference materials. Quijano et al. (1995) have used a similar approach to determine Se in water and serum. Using this method, the authors claimed that the sensitivity was improved by two orders of magnitude when compared with pneumatic nebulization (35 ng 1 -l compared with 3 I.tg 1-1).


Some manufacturers produce accessories for achieving flow injection/matrix elimination/preconcentration, but basically all that is required is an injection valve, a peristaltic pump, a glass tube, and some fittings to make the column. The vast majority of laboratories make their own manifolds since it is substantially cheaper and is often more robust.

E. Chromatography

While ICP-MS gives information on total elemental concentration, it is often the form of the element, so-called "speciation", which is important. To obtain speciation information it is necessary to couple a separation technique such as HPLC or GC with ICP-MS. A review of coupling chromatography with plasma spectrometric detection has been made by Hill et al. (1993), while a review concentrating on coupling chromatography with ICP-MS has been produced by Seubert (1994). There have been numerous papers appearing in the literature describing the coupling of high performance liquid chromatography (HPLC) with ICP-MS. Chroma- tography is frequently used for speciation studies and there have been numerous papers published speciating several different analytes in many different sample types. Examples include gold from anti-arthritis drugs (Zhao et al., 1992), aluminum in tea infusions by size exclusion chromatography (Owen et al., 1992); platinum from anticancer drugs (Zhao et al., 1993); arsenic in water (Thomas and Sniatecki, 1995), in urine (Larsen et al., 1993a), in fish (Branch et al., 1994), and in chicken (Dean et al., 1994); organotin (Kumar et al., 1993 and Rivas et al., 1995); and cadmium in pig kidney (Dean et al., 1987; Crews et al., 1989). Coupling HPLC with ICP-MS is in theory very simple. Normally a PTFE tube from the end of the column to the nebulizer is sufficient, but occasionally, when the mobile phase contains organic solvents, it may be necessary to desolvate the nebular in some way. There have been several methods published in the literature on this subject. A summary of methods available for desolvation is given in Section I.

Some authors are again becoming more adventurous and in attempts to improve the sensitivity or the precision of their analyses they have coupled together several techniques. An example of this is the use of chromatography to separate Sb III and Sb v, followed by hydride generation into ICP-MS detection (Smichowski et al., 1995). In this paper the detection limits for a 100 ~tl sample were 0.04 and 0.008 ng of Sb lII and Sb v, respectively, which was an improvement of over an order of magnitude compared with HPLC-ICP-MS. In a similar approach, lead has been speciated by HPLC-HG-ICP-MS (Yang and Jiang, 1995a). Again, the limits of detection for the HPLC-HG-ICP-MS were comparable to or better than conventional pneumatic nebulization. The limits of detection were reported to be 0.6-6 ng 1-1 depending on the species. Lead has also been speciated using HPLC - isotope dilution -ICP-MS (Brown et al., 1994).

Gas chromatography has been coupled with ICP-MS on far fewer occasions. This is partially because of the difficulty in coupling the two together. A heated transfer


line from the end of the chromatograph to the torch is required, but ensuring that there are no cool points where analytes can condense often proves to be very problematic. Another problem is ensuring that the transfer line reaches as far into the torch as possible while ensuring that it does not act as an aerial for RF power which, if directed back to the chromatograph, could lead to severe instrumental failure and possible personal danger to the operator. That said, a few workers have succeeded in making the coupling (Kim et al., 1992a,b; Peters and Beauchemin, 1993; Pretorius et al., 1993; Ebdon et al., 1994b). All of these papers describe the transfer line or the interface between the chromatograph and the ICP torch. The majority of papers in this field analyze organometallic compounds such as al- kylleads in fuel (Kim et al., 1992b) and metalloporphyrins (Pretorius et al., 1993; Ebdon et al., 1994b). A novel interface that can double for both GC and normal sample nebulization has been described by Peters and Beauchemin. This interface reportedly yielded detection limits that were favorable compared with those obtained from a conventional nebulizer and spray chamber. Other workers that have attempted to couple GC with ICP-MS include Prange and Jantzen (1995) and Hintelmann et al. (1995). This latter paper described the determination of mercury and organomercury species in sediments.

Supercritical fluid chromatography is one of the latest types of chromatography to be developed, and this too has been coupled successfully with ICP-MS. Appli- cations of this have been produced by Vela and Caruso (tin compounds)(1992) and Blake et al. (various organometallic compounds)(1995). An overview of the method has also been published (Carey and Caruso, 1992). The basic instrumentation including the transfer line is similar to that for gas chromatography.

A few papers coupling capillary electrophoresis with ICP-MS have also been published. One of the drawbacks with coupling these two techniques is the different flow rates of the sample. Capillary electrophoresis uses a flow rate at the ~tl min -1 range or below, while ICP-MS typically uses 1 ml min -l. Lu et al. (1995) have developed a special interface to overcome this problem. A speciation application has also been produced by Liu et al. (1995).

F. Hydride Generation

Hydride generation (HG) is a technique that is used for several of the metalloid elements. Arsenic and selenium are particularly prone to interference from chloride ions. The use of hydride generation is an extremely useful way of separating the analyte from the matrix. This means that even for difficult samples such as seawater, As and Se may be determined. The other advantage of HG is that the transport efficiency of the analyte to the ion source is far higher than for aqueous nebulization (close to 100%). This inevitably leads to a substantial increase in sensitivity. Hydride generation has a particular advantage for arsenic determinations in that by separating As from the matrix the troublesome interference from ArC1 § at mass 75 may be avoided. At very low levels of As, chloride in the aerosol carried over from


a conventional U-tube gas-liquid separator may be problematic. The use of a membrane separator eliminates this very fine aerosol and has therefore particular advantage in ICP-MS (Branch et al., 1991).

Other analytes have also been determined using this technique. Besides As and Se, Ge, Pb, Sn, and Te may be determined by their hydrides. Other analytes such as Hg may be determined by reducing any mercury present in a sample to the elemental form using stannous chloride and then sweeping the Hg to the plasma in a flow of argon. Cadmium has also been determined by ethylating it into a volatile vapor.

The only difficulty with this procedure is that the different species of an analyte may form a hydride with different efficiency. Examples include Se TM which is hydride forming and Se vI which is not, and inorganic arsenic (As nI and As v) which form hydrides (albeit at different rates) and arsenobetaine (the main arsenic compound in many marine samples) which does not form a hydride (Dean et al., 1990). There are procedures to overcome this problem, e.g. reducing the selenium and arsenic species to Se TM and As Iu, respectively, using a reagent such as hydrochloric acid or L-cysteine. This however does not work for some species, e.g. arsenobetaine and selenomethionine, which are extremely stable. To destroy these compounds, either very harsh acidic conditions such as perchloric acid are required, which can on occasions be hazardous for the analyst; or photolysis may be used. There have been numerous papers that have been published recently that describe the use of HG-ICP-MS. Antimony, arsenic, and selenium have been determined in waters (Haraldsson et al., 1992); As, Hg, and Sb have been determined in seawater (Stroh and Vollkopf, 1993); and total and leachable As and Se have been determined in soils (Anderson et al., 1994). Other solid materials have been analyzed. Arsenic and selenium have been determined in marine CRM's with a detection limit of 0.3 and 2.5 pg ml -l respectively (Tao et al., 1993).

Other researchers have coupled different types of chromatography with HG-ICP- MS to obtain speciation data. However, one of the major problems with this technique is obtaining full recovery of all the species present without altering the speciation. This obviously prohibits the use of conventional acid digestion of solid samples. Arsenic speciation has been achieved as reported in papers by Le et al. (1994) and Rubio et al. (1993). This latter paper is interesting because it involved on-line photooxidation to convert the nonreducible species into ones that can form hydrides. Arsenic has also been determined in seawater by HPLC-HG-ICP-MS (Hwang and Jiang, 1994) and water (Thomas and Sniatecki, 1995). This work utilized an interesting in-situ nebulizer hydride generator. Other speciation studies that have been performed include the separation of Sb Iu and Sb v (Smichowski et al., 1995), lead compounds (Yang and Jiang, 1995a), mercury compounds in biota (Brown et al., 1995), and organotin compounds in mussels (Rivaro et al., 1995). Other types of chromatography that have been used include micellar liquid chromatography for the speciation of tin (Inoue et al., 1995) and vesicle-mediated HPLC for the speciation of Hg (Aizpun et al., 1995) and for As (Liu et al., 1993). The


vesicle-mediated HPLC methods have used a Cl8-reversed phase column and a mobile phase consisting of didodecyldimethylammonium bromide vesicles in a phosphate buffer containing 0.5% methanol. Under these conditions, the four toxic forms of As were separated in 10 minutes with sub ng ml -l detection limits. As an application, As was speciated in urine and tap water.

Several papers have reported on the coupling of flow injection with HG-ICP-MS (Dean et al., 1990). In one study Se was determined in water and serum (Quijano et al., 1995) and in another the effects of organic solvents on the hydride generation of selenium was evaluated (Olivas et al., 1995). In this latter paper, it was found that the presence of the organic solvents enhanced the Se signal. Under optimum conditions (sodium borohydride 0.2%, pH 1, methanol load 6%) an LOD of 1 pg for a 200 ~t 1 sample was obtained.

Isotope measurements of antimony have been made in seawater by Kumamaru et al. (1994): A method was developed to overcome the well-known chloride interference on As determinations (Story et al., 1992). Other more exotic method- ologies include the preconcentration of As, Bi, and Te hydrides in a palladium- coated graphite tube, followed by ETV into the ICP-MS (Marawi et al., 1994). This technique has the advantage of separating the excess hydrogen produced during reduction, enabling a more stable plasma to be obtained. The linearity was limited to the sub ng ml -~ range, but the procedure was validated by the analysis of CRM's. The LOD for As was 0.002 ng m1-1.

IV. APPLICATIONS

There have been a very large number of applications involving determination by ICP-MS over the last 10 years. Many of these applications have already been detailed in the individual sections describing the different types of sample introduction. Extremely useful sources of information include the Atomic Spectrometry Updates (ASU) published in the Journal of Analytical Atomic Spectrometry. Fundamental papers involving the theory, hardware, modifications, etc. of ICP-MS are published annually in the October issue of the journal. Environmental applications including water, geological materials, air/airborne particulates, soils, etc. appear in the February issue; clinical, biological materials, food, and beverages appear in the April issue, and industrial analysis incorporating metals, chemicals, and advanced materials appear in the December issue. These updates give information on papers published in all major journals during the previous year. The majority of the ASU's tabulate the data so that the samples being analyzed, the sample preparation procedures, the techniques used, the analytes determined, etc. may be seen at a glance.

Several reviews have also been published in recent years. Gray, (1994) has discussed the current status of ICP-MS and considered the major problems. Other reviews concerning clinical applications have also been published. These include ones by Barnes (1993)(138 references) who summarized the advances in the


application of ICP-MS to human nutrition and toxicology; Vanhoe (1993) who reviewed the capabilities of ICP-MS for trace element analysis in body fluids and tissues; and McKay (1993) who reviewed (with 17 references) the use of ICP-MS for the study of pharmacokinetics of platinum drugs. Carey (1993) has discussed the relative merits of conventional sample nebulization, ETV, HG, GC, LC, and SFC (120 references). Uden (1995) has discussed the relative merits of HPLC- and GC-ICP-MS. In a similar paper, Hill et al. (1993) also reviewed the coupling of chromatography with ICP-MS. In another invaluable review, McLaren (1993) provided an ICP-MS applications bibliography update. This review was designed to include all of the most important ICP-MS applications papers. A review of speciation in aquatic and biological environments has been produced by Lespes et al. (1992). Reviews have also been produced for solid sample analysis by ICP-MS (Baumann, 1992; Voellkopf et al., 1992; Darke and Tyson, 1994). Darke and Tyson concentrated on slurry introduction, ETV, and laser ablation.

Recently, the use of mixed gas or alternative gas plasmas has increased substantially. Sheppard and Caruso (1994) have produced a review concerning the use of mixed gas plasmas to overcome polyatomic ion interferences (64 references). The use of helium as an alternative gas has been used successfully to overcome the interferences arising from argon, e.g. ArO § ArC1 § and ArAr § Similarly, mixed gas plasmas that use ethene (Ebdon et al., 1994c), nitrogen (Hill et al., 1992a), hydrogen (Ebdon et al., 1993a), and methane (Hill et al., 1992b) to reduce polyatomic ion interferences have been reported. The use of mixed gas plasmasmcertainly nitrogen to remove chloride interferencesmis now routine in many laboratories.

The use of the low-powered plasma developed by Evans and Caruso will probably gain popularity as it is a halfway house between conventional ICP-MS and other types of mass spectrometry (Evans and Caruso, 1993; Castillano et al., 1994). Depending on the operating conditions, it may be used to break down molecules completely in a similar manner to ICP-MS, but it may also be used to obtain a fragmentation pattern in a similar manner to typical mass spectrometry techniques (Evans et al., 1994). This will therefore enable a more unequivocal characterization of a sample. Although some modification to the hardware of the interface region of the instrument is required, this can be machined and may easily be interchangeable with normal hardware.

V. C O N C L U S I O N S

Inductively coupled plasma mass spectrometry has now come of age. A review by Horlick (1994) containing 36 references compared the current status of ICP-MS in the context of Shakespeare's seven ages of man, and there has been an overview by Gray (1994). Both concluded that the method had reached an encouraging degree of maturity. Many of the problems associated with ICP-MS have been overcome, such as the removal of interferences by either matrix separation using flow injection, hydride generation, or ETV; or by the use of mixed gases.


As the quest for more knowledge on the speciation of analytes progresses, it would seem obvious that the number of papers that couple chromatography with ICP-MS will increase. As more authors require greater sensitivity with fewer interferences, the coupling of other techniques such as ETV, HG, LA, etc. will increase. Similarly, as a better understanding of space charge effects, ion optics, and the events in the expansion chamber is obtained, instrumentation with better sensitivity will be developed. As work continues into the use of low-powered plasmas, it is envisaged that this too will become more popular.

High resolution spectrometers are becoming more widespread as the need for interference-free determinations increases. It is therefore envisaged that their use will continue to increase in the future.

REFERENCES Aizpun, B., Fernandez, M.L., Blanco, E., Sanz-Medel, A. J. Anal. At. Spectrom. 1995, 9, 1279. Allen, L.A., Pang, H.M., Warren, A.R., Houk, R.S.J. Anal At. Spectrom. 1995, 10,267. Alves, L.C., Wiederin, D.R., Houk, R.S. AnaL Chem. 1992, 64, 1164. Anderson, S.T.G., Robert, R.V.D., Farrer, H.N.J. Anal. At. Spectrom. 1994, 9, 1107. Arpino, P. Mass Spectrometry Reviews 1992, 11, 3. Barnes, R.M. Anal. Chim. Acta 1993, 283, 115. Baumann, H. Fresenius J. Anal. Chem. 1992, 342, 907. Bear, B.R., Fassel, V.A. Spectrochim. Acta 1986, 41B, 1039. Beauchemin, D. Analyst 1993, 118, 815. Beres, S., Thomas, R., Denoyer, E., Bruckner, P. Spectroscopy 1994, 9, 20. Blake, E., Raynor, M.W., Cornell, D. J. High Res. Chromatog. 1995, 18, 33. Bloxham, M.J., Hill, S.J., Worsfold, P.J.J. Anal. At. Spectrom. 1994, 9, 935. Boom, A.W., Browner, R.E Anal. Chem. 1982, 54, 1402. Botto, R.I., Zhu, J.J.J. Anal. At. Spectrom. 1994, 9, 905. Boumans, P.W.J.M., Ed. Inductively Coupled Plasma Emission Spectrometry Part 1: Methodology,

Instrumentation and Performance; John Wiley and Sons: New York, 1987. Bowins, R.J., McNutt, R.H.J. Anal. At. Spectrom. 1994, 9, 1233. Branch, S., Ebdon, L., Ford, M.J., Foulkes, M., O'Neill, P. J. Anal. At. Spectrom. 1991, 6, 151. Branch, S., Corns, W.T., Ebdon, L., Hill, S., O'Neill, P. J. Anal. At. Spectrom. 1991, 6, 155. Branch, S., Ebdon, L., O'Neill, P. J. Anal. At. Spectrom. 1994, 9, 33. Brotherton, T., Barnes, B., Vela, N., Caruso, J. J. Anal. At. Spectrom. 1987, 2, 389. Brown, A.A., Ebdon, L., Hill, S.J. Anal. Chim. Acta 1994, 286, 391. Brown, R., Gray, D.J., Tye, D. Water, Air and Soil Pollution 1995, 80, 1237. Byrne, J.P., Lamoureux, M.M., Chakrabarti, C.L., Ly, T., Gregoire, D.C.J. AnaL At. Spectrom. 1993, 8,

599. Campargue, R. J. Phys. Chem. 1984, 88, 4466. Carey J. Chromatogr. Sci. 1993, 31,330. Carey, J.M., Caruso, J.A. Trends in Anal. Chem. 1992, 11,287. Castillano, T.M., Giglio, J.J., Evans, E.H., Caruso, J.A.J. Anal. At. Spectrom. 1994, 9, 1335. Chenery, S., Cook, J.M.J. Anal. At. Spectrom. 1993, 8, 299. Chenery, S., Cook, J.M., Styles, M., Cameron, E.M. Chemical Geology 1995, 124, 55. Clemons, P.S., Minnich, M.G., Houk, R.S. Anal. Chem. 1995, 67, 1929. Coedo, A.G., Dorado, M.T.J. Anal. At. Spectrom. 1994, 9, 1111. Coedo, A.G., Lopez, T.D. ISIJ International 1994, 34, 997.


Coedo, A.G., Lopez, T.D., Algucil, E Anal Chim. Acta 1995, 315, 331. Colodner, D.C., Boyle, E.A., Edmond, J.M. AnaL Chem. 1993, 65, 1419. Cousin, H., Weber, A., Magyar, B., Abell, I., Gunther, D. Spectrochim. Acta 1995, 50b, 63. Crews, H.M., Dean, J.R., Ebdon, L., Massey, R.C. Analyst 1989, 114, 895. Cromwell, E.E, Arrowsmith, P. Anal. Chem. 1995, 67, 131. Dadfarmia, S., McLeod, C.W. Appl. Spectrosc. 1994, 48, 1331. Darke, S.A., Tyson, J.E J. Anal. At. Spectrom. 1993, 8, 145. Darke, S.A., Tyson, J.E Microchem. J. 1994, 50, 310. Date, A.R., Gray, A.L. Analyst 1981, 106, 1255. Date, A.R., Gray, A.L., Eds. Applications of Inductively Coupled Plasma Mass Spectrometry Blackie

and Son: London, 1989. Dawson, P.H. Mass Spectrometry Reviews 1986, 5, 1. Dean, J.R., Ebdon, L., Crews, H.M., Massey, R.C.J. AnaL At. Spectrom. 1988, 3, 349. Dean, J.R., Ebdon, L., Foulkes, M.E., Crews, H.M., Massey, R.C.J. Anal. At. Spectrom. 1994, 9, 615. Dean, J.R., Munro, S., Ebdon, L., Crews, H.M., Massey, R.C.J. Anal. At. Spectrom. 1987, 2, 607. Dean, J.R., Parry, H.G.M., Masey, R.C., Ebdon, L. ICP Information Newsletter 1990, 15, 569. DeCarlo, E.H., Pruszkowski, C. At. Spectrosc. 1995, 16, 65. Douglas, D.J., French, J.B.J. Anal. At. Spectrom. 1988, 3, 743. Eaton, A.N., Hutton, R.C., Holland, J.G. Chemical Geology 1992, 95, 63. Ebdon, L., Cave, M.R. Analyst 1982, 107, 172. Ebdon, L., Evans, E.H. European Spectroscopy News 1988, 79, 9. Ebdon, L., Evans, E.H., Pretorius, W.G., Rowland, S.J.J. Anal. At. Spectrom. 1994, 9, 939. Ebdon, L., Fisher, A., Handley, H., Jones, P. J. Anal. At. Spectrom. 1993, 8, 979. Ebdon, L., Fisher, A.S., Worsfold, P.J.J. Anal. At. Spectrom. 1994, 9, 611. Ebdon, L., Fisher, A.S., Worsfold, P.J., Crews, H., Baxter, M. J. AnaL At. Spectrom. 1993, 8, 691. Ebdon, L., Ford, M.J., GoodaU, P., Hill, S.J. Microchem. J. 1993, 48, 846. Ebdon, L., Ford, M.J., Hutton, R.C., Hill, S.J. Appl. Spectrosc. 1994, 48, 507. Ebdon, L., Foulkes, M.E., Hill, S. Microchem. J. 1989, 40, 30. Enzweiler, J., Potts, P.J., Jarvis, K.E. Analyst 1995, 120, 1391. Evans, E.H., Caruso, J.A.J. Anal. At. Spectrom. 1993, 8, 427. Evans, E.H., Giglio, J.J.J. Anal. At. Spectrom. 1993, 8, 1. Evans, E.H., Giglio, J.J., Castillano, T.M., Caruso, J.A. Inductively Coupled and Microwave Induced

Plasma Sources For Mass Spectrometry; RSC Analytical Spectroscopy Monographs; Royal Society of Chemistry; Cambridge, 1995.

Evans, E.H., Pretorius, W., Ebdon, L., Rowland, S. Anal. Chem. 1994, 66, 3400. Fairman, B., Sanz-Medel, A. J. Anal. At. Spectrom. 1995, 10, 281. Fedorowich, J.S., Jain, J.C., Kerrich, R. Canadian Mineralogist 1995, 33, 469. Fonseca, R.W., Miller-Ihli, N.J. Appl. Spectrosc. 1995, 49, 1403. Fryer, B.J., Jackson, S.E., Longerich, H.P. Canadian Mineralogist 1995, 33, 303. Garcia-Alonso, J.I. Anal. Chim. Acta 1995, 312, 57. Goltz, D.M., Gregoire, D.C., Byrne, J.P., Chakrabarti, C.L. Spectrochim. Acta 1995, 50b, 803. Gomez, M.M.G., Mcleod, C.W.J. Anal. At. Spectrom. 1995, 10, 89. Goodall, P., Foulkes, M.E., Ebdon, L. Spectrochim. Acta 1993, 48b, 1563. Gray, A. Anal. Proc. 1994, 31, 371. Gray, A.L. Spectrochim. Acta 1986, 41b, 151. Gregoire, D.C., Ansdell, K.M., Goltz, D.M., Chakrabarti, C.L. Chem. Geol. 1995, 124, 91. Gregoire, D.C., Miller-Ihli, N.J., Sturgeon, R.E.J. Anal. At. Spectrom. 1994, 9, 605. Gunther, D., Longerich, H.P., Forsythe, L., Jackson, S.E. Am. Lab. 1995, 27, 24. Guo, X., Lichte, EE. Analyst 1995, 120, 2707. Halicz, L., Brenner, I.B., Yoffe, D. J. Anal. At. Spectrom. 1993, 8, 475. Hall, G.E.M.J. Geochem. Explor. 1992, 44, 201.


Haraldsson, C., Pollak, M., Ohman, P. J. Anal. At. Spectrom. 1992, 7, 1183. Hartley, J.H.D., Hill, S.J., Ebdon, L. Spectrochim. Acta 1993, 48b, 1421. Heumann, K.G., Rottmann, L., Vogl, J. J. Anal. At. Spectrom. 1994, 9, 1351. Hill, S.J., Bloxham, M.J., Worsfold, EJ. J. Anal. At. Spectrom. 1993, 8, 499. Hill, S.J., Ford, M.J., Ebdon, L. J. Anal. At. Specrtrom. 1992, 7, 719. Hill, S.J., Ford, M.J., Ebdon, L. J. Anal. At. Spectrom. 1992, 7, 1157. Hill, S.J., Hartley, J., Ebdon, L. J. Anal. At. Spectrom. 1992, 7, 895. Hill, S.J., Hartley, J., Ebdon, L. J. Anal. At. Spectrom. 1992, 7, 23. Hintelmann, H., Evans, R.D., Villeneuve, J.Y.J. Anal. At. Spectrom. 1995, 10, 619. Hoffmann, E., Ludke, C., Scholze, H., Stephanowitz, H. Fresenius J. Anal. Chem. 1994, 350, 253. Hollenbach, M., Grohs, J., Mamich, S., Kroft, M., Denoyer, E.R.J. Anal. At. Spectrom. 1994, 9, 927. Horlick, G. J. Anal. At. Spectrom. 1994, 9, 593. Houk, R.S., Fassel, V.A., Flesch, G.D., Svec, H.J., Gray, A.L., Taylor, C.E.Anal. Chem. 1980, 52, 2283. Huang, M.E, Jiang, S.J., Hwang, C.J.J. Anal. At. Spectrom. 1995, 10, 31. Hwang, C.J., Jiang, S.J., Anal. Chim. Acta 1994, 289, 205. Inoue, Y., Kawabata, K., Suzuki, Y. J. Anal. At. Spectrom. 1995, 10, 363. Jakubowski, N., Feldman, I., Stuewer, D., Berndt, H. Spectrochim. Acta 1992, 47b, 119. Jakubowski, N., Jepkens, B., Stuewer, D., Berndt, H. J. Anal. At. Spectrom. 1994, 9, 193. Jarvis, K.E. Chemical Geology 1992, 95, 73. Jarvis, K.E., Gray, A.L., Houk, R.S. Handbook oflnductively Coupled Plasma-Mass Spectrometry 1992. Jarvis, K.E., Williams, J.G. Chemical Geology 1993, 106, 251. Karanassios, V., Drouin, E, Reynolds, G.G. Spectrochim. Acta 1995, 50b, 415. Kim, A., Hill, S.J., Ebdon, L., Rowland, S. J. High Res. Chromatogr. 1992, 15, 665. Kim, A.W., Foulkes, M.E., Ebdon, L., Hill, S.J., Patience, R.L., Barwise, A.G., Rowland, S.J.J. Anal.

At. Spectrom. 1992, 7, 1147. Kogan, V.V., Hinds, M.W., Ramendik, G.I. Spectrochim. Acta 1994, 49b, 333. Koropchak, J.A., Veber, M. Critical Reviews in Anal. Chem. 1992, 23, 113. Kumamaru, T., Yamamoto, M., Nakata, E, Tsubota, H., Nishikida, K.Anal. Sci. 1994, 10, 651. Kumar, U.T., Dorsey, J.G., Elder, R.C.J. Chromatogr. 1993, 654, 261. Lain, J.W., McLaren, J.W.J. Anal. At. Spectrom. 1990, 5, 419. Lamoureux, M.M., Gregoire, D.C., Chakrabarti, C.L., Goltz, D.M. Anal. Chem. 1994, 66, 3217. Larsen, E.H., Pritzl, G., Hansen, S.H.J. Anal. At. Spectrom. 1993, 8, 557. Larsen, E.H., Pritzl, G., Hansen, S.H.J. Anal. At. Spectrom. 1993, 8, 1075. Layman, L.R., Lichte, EE. Anal. Chem. 1982, 54, 638. Le, X.C., Cullen, W.R., Reimer, K.J. Talanta 1994, 41,495. Lespes, G., Potingautier, M., Astruc, A. Environmental Technology 1992, 13, 207. Lichte, EE. Anal. Chem. 1995, 67, 2479. Liu, Y., Lopez Avila, V., Zhu, J.J., Wiederin, D.R., Beckert, W.E Anal. Chem. 1995, 67, 2020. Liu, Y.M., Sanchez, M.L.E, Gonzalez, E.B., Sanz-Medel, A. J. Anal. At. Spectrom. 1993, 8, 815. Lu, G.H., Bird, S.M., Barnes, R.M. Anal. Chem. 1995, 67, 2949. Lu, EL., Huang, K.S., Jiang, S.J. Anal. Chim. Acta 1993, 284, 181. Luan, S., Pang, H.M., Shum, S.C.K., Houk, R.S.J. Anal. At. Spectrom. 1992, 7, 799. Marawi, I., Olson, L.K., Wang, J.S., Caruso, J.A.J. Anal. At. Spectrom. 1995, 10, 7. Marawi, I., Wang, J.S., Caruso, J.A. Anal. Chim. Acta 1994, 291, 127. McKay, K. Cancer Surv. 1993, 17, 407. McLaren, J.W. At. Spectrosc. 1993, 14, 191. Montaser, A., Golightly, D.W., Eds. Inductively Coupled Plasmas in Analytical Atomic Spectrometry,

2rid Ed.; UCH Publishers: New York, 1992. Olivares, J.A., Houk, R.S. Anal. Chem. 1985, 57, 2674. Olivas, R.M., Quetel, C.R., Donard, O.EX. J. Anal. At. Spectrom. 1995, 10, 865. Orians, K.J., Boyle, E.A. Anal. Chim. Acta 1993, 282, 63.


Outridge, P.M., Evans, R.D.J. Anal. At. Spectrom. 1995, 10, 595. Owen, L.M.W., Crews, H.M., Massey, R.C. Chemical Speciation and Availabili~. 1992, 4, 89. Paschal, D.C., Caldwell, K.L., Ting, B.G.J. Anal. At. Spectrom. 1995, 10, 367. Peters, G.R., Beauchemin, D. Anal. Chem. 1993, 65, 97. Powell, M.J., Boomer, D.W., Wiederin, D.R. Anal. Chem. 1995, 67, 2474. Prabhu, R.K., Vijayalakshmi, S., Mahalingham, T.R., Viswanathan, K.S., Mathews, C.K.J. Anal. At.

Spectrom. 1993, 8, 565. Prange, A., Jantzen, E. J. Anal. At. Spectrom. 1995, 10, 105. Pretorius, W.G., Ebdon, L., Rowland, S.J.J. Chromatogr. 1993, 646, 369. Quijano, M.A., Guttierez, A.M., Conde, M.C.E, Camara, C. J. Anal. At. Spectrom. 1995, 10, 871. Raith, A., Hutton, R.C. Fresenius J. Anal. Chem. 1994, 350, 242. Raith, A., Hutton, R.C., Abell, I., Crighton, J. J. Anal At. Spectrom. 1995, 10, 591. Ren, J.M., Rattray, R., Salin, E.D., Gregoire, D.C.J. Anal. At. Spectrom. 1995, 10, 1027. Richner, P. J. Anal. At. Spectrom. 1993, 8, 927. Rivaro, P., Zaratin, L., Frache, R., Mazzucotelli, A. Analyst 1995, 120, 1937. Rivas, C., Ebdon, L., Hill, S.J. Quimica Analitica 1995, 14, 142. Rubio, R., Padro, A., Alberti, J., Rauret, G. Anal. Chim. Acta 1993, 283, 160. Sah, R.N. Appl. Spectrosc. Reviews 1995, 30, 35. Sang-Ho Nam, Jong-Soo Lim, Montaser, A. J. AnaL At. Spectrom. 1994, 9, 1357. Sayama, Y., Hayashibe, Y., Fukuda, M. Fresenius J. AnaL Chem. 1995, 353, 162. Seubert, A. Fresenius J. Anal. Chem. 1994, 350, 210. Sharp, B.L.J. AnaL At. Spectrom. 1988, 3, 613. Sharp, B.L.J. Anal. At. Spectrom. 1988, 3, 939. Sheppard, B.S. Caruso, J.A.J. AnaL At. Spectrom. 1994, 9, 145. Shepherd, T.J., Chenery, S.R. Geochim. et Cosmochim. Acta 1995, 59, 3997. Shibata, N., Fudugawa, N., Kubota, M. Spectrochim. Acta 1993, 48b, 1127. Shum, S.C.K., Pang, H.M., Houk, R.S. AnaL Chem. 1992, 64, 2444. Shum, S.C.K., Houk, R.S. Anal. Chem. 1993, 65, 2972. Smichowski, P., Madrid, Y., Guntinas, M.B.D., Camara, C. J. Anal. At. Spectrom. 1995, 10, 815. Stix, J., Gauthier, G., Ludden, J.N. Canadian Mineralogist 1995, 33, 435. Story, W.C., Caruso, J.A., Heitkemper, D.T., Perkins, L. J. Chromatogr. Sci. 1992, 30, 427. Stroh, A., Vollkopf, U. J. Anal. At. Spectrom. 1993, 8, 35. Stroh, A., Vollkopf, U., Denoyer, E.R.J. Anal. At. Spectrom. 1992, 7, 1201. Tanner, S.D. Spectrochim. Acta 1992, 47b, 809. Tanner, S.D., Douglas, D.J., French, J.B. Appl. Spectrosc. 1994, 48, 1373. Tao, H., Lam, J.W.H., McLaren, J.W.J. Anal. At. Spectrom. 1993, 8, 1067. Tao, H., Miyazaki, A. J. Anal. At. Spectrom. 1995, I0, I. Tan', M.A., Zhu, G., Browner, R.E J. Anal. At. Spectrom. 1992, 7, 813. Thomas, P., Sniatecki, K. J. Anal. At. Spectrom. 1995, I0, 615. Thompson, M., Walsh, J.N. Handbook of Inductively Coupled Plasma Spectrometry, 2nd Ed.; Blackie

and Sons: London, 1989. Tittes, W., Jakubowski, N., Stuewer, D., Tolg, G. J. Anal. At. Spectrom. 1994, 9, 1015. Totland, M., Jarvis, I., Jarvis, K.E. Chemical Geology 1993, 104, 175. Uden, EC. J. Chrom. A. 1995, 703, 393. Ulens, K., Moens, L., Dams, R., Van Winckel, S., Van de Velde, L. J. Anal. At. Spectrom. 1994, 9, 1243. Vanhoe, H. Journal of Trace Elements and Electrolytes in Health and Disease 1993, 7, 131. Vanhoe, H., Goossens, J., Moens, L., Dams, R. J. Anal. At. Spectrom. 1994, 9, 177. vanhoe, H., Moens, L., Dams, R. J. Anal. At. Spectrom. 1994, 9, 815. Vanhoe, H., Saverwijns, S., Parent, M., Moens, L., Dams, R. J. Anal. At. Spectrom. 1995, 10, 575. Vaughan, M.A., Horlick, G. Spectrochim. Acta 1990, 45b, 1327. Vela, N.P., Caruso, J.A.J. Anal. At. Spectrom. 1992, 7, 971.


Voellkopf, U., Paul, M., Denoyer, E.R. Fresenius J. Anal. Chem. 1992, 342, 917. Wang, J.S., Carey, J.M., Caruso, J.A. Spectrochim. Acta 1994, 49b, 193. Wang, S., Brown, R., Gray, D.J. Appl. Spectrosc. 1994, 48, 1321. Watling, R.J., Herbert, H.K., Abell, I.D. Chemical Geology 1995, 124, 67. Wei, W.C., Chen, C.J., Yang, M.H.J. Anal. At. Spectrom. 1995, 10, 955. Woller, A., Mester, Z., Fodor, P. J. Anal At. Spectrom. 1995, 10, 609. Wu, M., Hieftje, G.M. Appl. Spectrosc. 1992, 46, 1912. Yang, H.J., Jiang, S.J.J. Anal At. Spectrom. 1995, 10, 963. Yang, K.L., Jiang, S.J. Anal. Chim. Acta 1995, 307, 109. Yasuhara, H., Okano, T., Matsumura, Y. Analyst 1992, 117, 395. Zaray, G., Kantor, T. Spectrochim. Acta 1995, 50b, 489. Zhao, Z., Jones, W.B., Tepperman, K., Dorsey, J.G., Elder, R.C.J. Pharm. Biomed. Sci. 1992, 10, 279. Zhao, Z., Tepperman, K., Dorsey, J.G., Elder R.C.J. Chromatogr. Biomed. AppL 1993, 615, 83. Zoorob, G., Tomlinson, M., Wang, J.S., Caruso, J.A.J. AnaL At. Spectrom. 1995, 10, 853.


MULTIELEMENT GRAPHITE FURNACE AND FLAME ATOMIC ABSORPTION SPECTROMETRY

Joseph Sneddon and Kimberly S. Farah

Io II.

III.

IV.

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

A. Multielement Line Sources . . . . . . . . . . . . . . . . . . . . . . . . . . 35

B. Continuum Source Multielement AAS . . . . . . . . . . . . . . . . . . . . 36

C. Multielement Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

D. Multielement Systems Using Lasers . . . . . . . . . . . . . . . . . . . . . 46

Flame Atomization Applications . . . . . . . . . . . . . . . . . . . . . . . . . 47

A. Multichannel Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

B. SIMACC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

C. Hitachi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

D. Thermo Jarrell Ash-Baird . . . . . . . . . . . . . . . . . . . . . . . . . . 49

Graphite Furnace Applications . . . . . . . . . . . . . . . . . . . . . . . . . . 49

A. FREMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

B. Hitachi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

C. Dual-Channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52


33

34 JOSEPH SNEDDON and KIMBERLY S. FARAH

D. SIMAAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 E. Time-Divided, Single-Channel . . . . . . . . . . . . . . . . . . . . . . . 54 F. Thermo Jarrell Ash-Baird . . . . . . . . . . . . . . . . . . . . . . . . . . 54

G. Fast Fourier Transform (FFT) . . . . . . . . . . . . . . . . . . . . . . . . 55 H. Perkin-Elmer SIMAA . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

V. Multielement System Summary . . . . . . . . . . . . . . . . . . . . . . . . . 58 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

ABSTRACT

In recent years there has been a renewed interest in the use and application of simultaneous multielement atomic absorption spectrometry (AAS). This chapter describes the historical development of simultaneous multielement AAS, tracing its use in laboratory-constructed or modified systems through to commercially available systems. Several approaches have been proposed and used and are described, including multielement line sources, continuum source, multichannel systems, and laser- based systems. Selected results of the applications of simultaneous AAS are presented, primarily on graphite furnace atomization, but where appropriate flame atomization.

I. I N T R O D U C T I O N

Atomic absorption spectrometry (AAS) is widely used for the determination of trace elements in numerous and complex matrices. It offers excellent and low detection limits, high precision, high sensitivity and selectivity, and acceptable accuracy. One drawback of AAS has been its lack of ability to simultaneously determine several elements (multielement determinations). If knowledge of several elements in a sample is desired, an analysis must be run for each sample separately. This leads to an increase in the time required to complete the analysis. It also requires that a separate sample be used for every analysis. This can prove to be difficult when the available sample is small in quantity or is difficult to obtain.

Since the late 1960s, researchers have used a variety of techniques in performing multielement AAS analysis. Both continuous and line sources have been used. Early problems with these systems included a decrease in sensitivity, lack of linearity, and spectral interferences. Not until the early 1980s did systems become available that eliminated some of these problems.

In the past few years several commercial systems have become available. Newer systems, some using a laser light source and solid state electronics for detection, are being studied. Commercial multielement AAS systems are available from Hitachi (Danbury, Connecticut) since 1988, Thermo Jarrell Ash-Baird (Franklin, Massachusetts) since 1990, Leeman Labs (Lowell, Massachusetts) since 1993, and

Graphite Furnace and Name AAS 35

Perkin-Elmer (Norwalk, Connecticut) since 1994. The authors of this chapter anticipate that most major manufacturers will have commercial simultaneous multielement AAS instrumentation available in the near future. The authors also anticipate that multielement systems will only be capable of determining a maximum of six elements in a simultaneous mode. This is primarily due to the degree of difficulty in engineering.

A recent review on multielement AAS in general is available from Farah et al. (1993) and graphite furnace AAS by Farah and Sneddon (1995). Hamly (1986) also reviewed multielement AAS with a continuum source. An interesting book by Butcher and Sneddon (1997) describes a practical guide to graphite furnace AAS. In this chapter, the development of simultaneous AAS will be described. The primary focus will be on simultaneous graphite furnace AAS and selected applications. The increased sensitivity of the graphite furnace over the flame (10-100 x improvement) makes simultaneous multielement graphite furnace AAS attractive to the analyst. However, major developments and selected applications which involved the flame will also be described.

II. I N S T R U M E N T A T I O N

A. Multielement Line Sources

Hollow cathode lamps (HCLs), which are the most widely used line sources, commonly require that one lamp be used for each element. The multielement HCL and multielement electrodeless discharge lamp (EDL) [primarily for atomic fluorescence spectrometry (AFS)] were proposed and widely used beginning in the mid-1960s. While still available, multielement lamps are not widely used. This is primarily due to the reduced sensitivity (necessitated by compromise operating conditions) of the element in the multielement lamp compared to the element in a single element lamp. Multielement HCLs are most suited to elements that have similar properties and are usually limited to between two and four elements, e.g., Ca-Mg, Co-Fe-Ni, Au-Pd-Pt-Rh, and so forth. The use of multielement HCLs was demonstrated by Zeeman and Brink (1968). The multielement lamp was made using a cathode body of 0.1-mm thick Pt with strips of Pd and Rh soldered with Au to the inside of the Pt cylinder. A neon atmosphere was used inside the lamp. The multielement HCL could simultaneously determine Pd, Pt, and Rh. Van Rensburg and Zeeman (1968) attempted to improve on the sensitivity of the multielement lamp by using a high-intensity lamp with selective modulation. Detection limits were not as low as those using the low-intensity lamps. Multielement line sources have been used in combination with multichannel systems since the early 1970s. Multichannel systems are discussed in more detail later in this chapter.

Fulton et al. (1970) studied the use of dual-element microwave EDLs as a multielement line source for both AFS and AAS. Their results indicated that the dual-element As and Sb lamp was almost as sensitive as a single-element EDL.


Interference studies for several elements showed that no extra interference was caused by the use of the dual-element lamp. The stability of the lamp and variations introduced during lamp preparation were noted.

B. Continuum Source Multielement AAS

The use of a continuum source as a light source was proposed and used to overcome the problems associated with multielement lamps. The continuum source can operate in both the ultraviolet and visible region of the electromagnetic spectrum. A disadvantage of the continuum source is that the spectral bandpass is much greater than that of the hollow cathode lamp line profile widths and the intensity can be unstable. This results in decreased sensitivity, higher (poorer) detection limits, and a reduced linear range.

Hamly et al. (1979) developed a multielement system based on a xenon arc (300 W) continuum source and an echelle polychromator modified for wavelength modification, which was named simultaneous multielement atomic absorption with a continuum source (SIMAAC). A schematic diagram of the system is shown in Figure 1. The system was compatible with either the flame or graphite furnace. The instrument was capable of analyzing up to 16 elements at one time using double- beam operation and background correction. Detection limits were found to be

_ _ .

9 ~ :::' )..

! ,a , , 1 | i

, ,

l , - i l I _ Figure 1. Block diagram of the SIMMAC system (Harnly et al., 1979). L1 and L2, lenses. QP, quartz plate. EG, echelle grating. P, prism. M1 and M2, mirrors.

Graphite Furnace and Flame AA5 37

comparable to those of a single-element systems for many elements. However, a major drawback was a lack of sensitivity for elements whose major resonance (most sensitive) wavelength was below 280 nm. For example, a comparison between inductively coupled plasma atomic emission spectroscopy (ICP-AES) and SI- MAAC for the determination of several elements in U.S. Geological Survey (USGS) reference manganese nodules was reported by Kane and Harnly (1982). The SIMAAC system lacked precision and accuracy for Co and Ni, and imprecision near the detection limits was noted.

Harnly and Kane (1983) studied the effect of air-acetylene flame parameters on the detection limits and analytical accuracy of 11 elements. The use of a multielement system requires a compromise in experimental conditions which produce various detection limits. Using one set of experimental conditions gave different set of detection limits than another set of experimental conditions. Harnly and Kane (1983) found that the best compromise for detection limits was observed at heights low in the flame and with richer flames. Compromise experimental conditions will be necessary in simultaneous AAS whether a flame or graphite furnace is used as the atomizer.

The effect of the acid matrix, the measurement mode (peak height or peak area), the atomizer surface (unpyrolized and pyrolized graphite), the atomization mode (from a wall or from the platform), and the atomization temperature on the simultaneous graphite furnace atomization of Co, Cr, Cu, Fe, Mn, Mo, Ni, V, and Zn were also examined by Harnly et al. (1983). Detection limits using simultaneous analysis were comparable to those of single-element systems provided the wavelength used was greater than 280 nm. Matrix effects in 5% HNO 3 were noted, as well as the requirement for a high atomization temperature when analyzing for Mo or V.

Carroll et al. (1985) have reported that the use of a graphite probe was more accurate than tube-wall atomization methods, but less precise than platform atomization methods when used in conjunction with the SIMAAC system.

Lundberg et al. (1988) used the SIMAAC system to compare an integrated contact cuvette furnace (spatially isothermal) and a two-step atomizer (spatially and temporally isothermal). They found that the spatially isothermal furnace was less affected by carryover contamination and was not as prone to interference effects. A drawback of the two-step furnace was its lack of capability for determining nonvolatile elements.

Masters et al. (1989) reported on the development of an echelle-spectrometer/ image dissector combination for continuum source atomic absorption (CSAA). The design failed to produce sufficient resolution to produce quantifiable results for multielement analysis. The authors then replaced the echelle grating with one that would produce the desired spectral characteristics. The modified instrument included a 20-groove/mm echelle grating blazed at 76 ~ However, their work showed that the total spectral coverage and ability of the instrument to analyze in the ultraviolet range for elements whose major resonance (most sensitive) wave-


length were limited. Removal of the magnetic shield from the image dissector significantly improved detector performance. However, when the entire spectrum was displayed simultaneously, the image detector was not capable of simultaneously accumulating information. A minimum of 0.5 s was required to obtain a reasonably certain intensity measurement. It was possible to perform sequential analyses for flame atomization provided a large enough sample size was present. Sensitivity was reduced by a factor of at least 2 in most cases. Due to the transient nature of the graphite furnace atomization signal, sequential analysis could not be performed using the CSAA system. The study indicated that the use of higher quality grating and integrating two-dimensional detectors would be required.

C. Multielement Systems

A major drawback of early multielement lamp AAS systems was the lack of effective multichannel measurement systems. Horlick and Codding (1975) proposed a computer-coupled photodiode array spectrometer. A multielement hollow cathode lamp was used as a source. The system used a 256-element array. The array simultaneously covered only about 130/~ (13 nm) of the continuous spectrum. A severe drawback of the system was in the choice of analytical lines since only a limited number of spectral lines were available for analysis. In addition, overlap of nonabsorbing lines with an absorbing line caused negative deviations from Beer's Law.

A silicon-target vidicon detector that responded in the ultraviolet range was developed by Jackson et al. (1974). This system used a half-silvered quartz mirror (beam splitter) and combined radiation from two hollow cathode lamps. A vidicon tube was mounted in place of the exit slit photomultiplier assembly. The system showed that most sensitivities obtained were comparable with those of a higher resolution, single-channel AAS. The use of a 50-ktm entrance slit, which resulted in a resolution of approximately 1 nm, limited the choice of analytical lines for multielement analyses due to overlap of nonresonance lines. Interferences and poor signal-to-noise (S/N) ratios compared to those of single-element systems were also encountered.

To overcome the limitations in the spectral range, a two-dimensional silicon target vidicon tube was coupled to an echelle grating spectrometer by Felkel and Pardue (1977). To simplify data reduction and increase resolution, a random-access mode was used to measure only the spectral lines of interest. Multielement experiments showed that wavelengths of elements below 325 nm failed to yield acceptable detection limits. This result may have been due to the low line intensities associated with the multielement HCL as compared to that of the silicon target vidicon system.

Lundberg and Johansson (1976) used a multislit spectrometer that permitted the simultaneous registering of several absorbance signals using a single photomultiplier tube (PMT). A schematic diagram of the system is shown in Figure 2. Detection limits and sensitivities were comparable to those of single-element

Gra 9hite Furnace and Flame AAS 39

M : , T

IF | | I I I I ,!

MINICOMPUT|II

Figure 2. Block diagram of a computer-controlled multislit flameless atomic absorption spectrometer (Lundberg and Johansson, 1976). S1, multielement hollow cathode lamp; $2, hydrogen lamp for background correction; B, beam splitter; 11, and L2 are lenses; F, graphite furnace; PS, lamp power supply; IF, interface; CRT, 2-channel oscilloscope; R, strip chart recorder; TTY, teletype; E, concave mirror; 1F, photomultiplier tube; M, motor; P, photon coupled interrupter module; C, chopper with moving slits; A, adjustable fixed slits.

systems. Drawbacks of the multislit system were the lack of availability of multielement HCLs and the narrow operating range of the system.

Nakamura and Kubota (1990) have reported on the use of a single-channel, time-divided, simultaneous multielement AAS system using an electrothermal atomizer. The system consisted of a multielement HCL, a single detection channel with one PMT, and a time-divided high-speed computer data acquisition system. Precision and calibration ranges of the system were comparable to those of single-element systems for A1, Cu, and Fe. However, sensitivity and precision at the lower end of the calibration range was poor. The researchers proposed several reasons for this observation: (1) the tungsten atomizer, in contrast to a more conventional graphite atomizer, (2) the nondispersive optics instead of a monochromator, and (3) the multielement cathode lamp.


To overcome the lack of stable emission which occurred when multielement HCLs were used, Salin and Ingle (1978a,b) combined four single element HCLs with beam splitters that were directed through a carbon rod atomizer and into a special monochromator with a separate slit for each element. A single PMT was used and absorbance signals determined for each element using a time multiplex approach. The time multiplex approach requires that the HCL for each element be pulsed on and off in a sequential manner at the same frequency, thus only one lamp was on at a time. Detection limits were poor compared to a single HCL lamp as the shot noise was increased. Another disadvantage of the system was the inability to carry out background absorption correction for each element since a hydrogen lamp was used as a continuum source for background correction.

To improve detection limits, Alder et al. (1976) proposed the use of one PMT for each HCL. The system is shown schematically in Figure 3. To overcome the standing background potential, which made measurement of the signals to the PMT systems difficult, pulses were differentiated and then integrated to reconstitute the original shape. This modification did not improve detection levels or precision due to high background noise and line overlap.

Gulloch (1988) described the use of optical quartz lenses and fibers instead of mirrors for collection of radiation from four hollow cathode lamps. The system is shown in Figure 4. The radiation was collected from the quartz lenses and focused into the ends of four, single-mode quartz fibers. A polychromator was employed in lieu of a monochromator after initial studies indicated poor detection limits. Results obtained with the polychromator were comparable to those obtained using conven-

Figure 3. Block diagram of multichannel AAS system (Alder et al., 1976).

Flame, Electro or Non Thermal Atomization

ee e

e ! e

Polychronmter

[ ~ l l l .i i

,1 Monochremtor

tl i i

i I i

De~ection Amplil'wr Rudeut . . . . . . .

I)etectkm Amplil'nr Readout

m

Lillhtsource Atomintk)n Unit Dlelmmbn and I)eteclion Unit

Figure 4. Simultaneous AAS (Golloch, 1988).


tional single-element AAS. Gulloch also reported that a prototype model was being developed by Spectr-Analytical Instruments, Kleve, Germany.

A frequency modulated simultaneous multielement AAS system (FremsAAS) with background correction has been developed by Lehnert et al. (1993). FremsAAS allows for the determination of three elements at one time using a deuterium continuum source for background correction. Light signals sent to four furacated optical fibers are sent through interference filters and then to a photodiode detector. The use of lock-in amplifiers allowed for selection of the signal from the individual HCLs. The FremsAAS system was tested, using flame atomization, on three reference materials. Results showed that the values obtained using FremsAAS were dependent on the type of matrix analyzed. Matrices with a high salt content produced results that had larger deviations from certified values.

Multichannel systems with improvements in the technology of data processing equipment, electronics, and optics have been developed and were conunercially available before being replaced by newer systems. Dual-channel systems have been marketed by Hitachi (model 170-50) and Nippon Jarrell-Ash (Kyoto, model AA- 8200). Kumamaru et al. (1989) used the Nippon Jarrel-Ash system for the simultaneous determination of Cu and Pb using a graphite furnace atomizer. Their results showed good agreement with certified samples or those values determined by ICP-AES for a variety of samples.

The use of a dual-channel Hitachi system has been reported by Sakurada et al. (1971). The model 170-50 Hitachi system used a high-speed data acquisition system to alternately collect the absorption signals. The authors validated the dual system for Ag and Cu flame AAS. The same bandpass of the monochrometer was used since the wavelength for copper (327.40 nm) and silver (328.07 nm) were similar. The use of a t-test showed that there was no statistically significant difference in the results obtained using the two methods.

In the mid- 1980s, a multielement system employing two multielement HCLs and a polychromator along with nine detectors to simultaneously determine nine elements in wear oils was developed through a joint effort between Perkin-Elmer Corporation and the U.S. Air Force (Niu et al., 1987). The system was designed to be rugged and portable with a total weight of only 100 lb. The system was not widely used and was never commercially developed and marketed.

Since the late 1980s, Hitachi has marketed a simultaneous multielement system, Model Z-9000 (Retzik and Bass, 1988). A schematic diagram of the system is shown in Figure 5. The system utilized Zeeman background correction and used four HCLs focused simultaneously through the atomizer. Four PMTs allowed for continuous monitoring of each element. The Z-9000 Hitachi system can analyze four elements in one furnace firing. The multielement system offers the following advantages over the single-element or dual element systems: high sample throughput; shorter analysis time; saving on consumables; expansion of dynamic working range; internal standards; and smaller required working sample (Retzik and Bass, 1988). Yasuda et al. (1989) used the Z-9000 to investigate the expansion of the

Graphite Furnace and Flame AAS 43

F .mm $R p o ~ l ~

~Jhi~ Fwn~,e

Exit .Slit

Beam p t ~

~ Group2 |sup I (S - e) (1.4) J4oUow C~ocJ, LJnn~

Figure S. Schematic diagram of the Hitachi Z-9000 system consisting of four HCLs and four photomultiplier tubes (Retzik and Bass, 1988).

dynamic range using peak height measurement and a new data processing method. The working range was extended between a factor of 1 to 2 orders of magnitude. Optimization of parameters for multielement determinations of Cd, Pb, and Zn indicated that higher heating rates and rapid atomization produced reliable results.

Thermo-Jarrell Ash-Baird (originally called the Smith-Hieftje 4000 and 8000 when it was introduced in the early 1990s and as of 1993 called AA-Scan 4 and AA-Scan 8) have developed a fully automated spectrometer which can run up to eight HCLs in an unattended run with the capability of measuring simultaneously four elements (Dulude, 1992). The system is shown schematically in Figure 6. This is accomplished through use of a synchronized, rapidly moving, galvanometer- driven grating and lamp selection mirror which allows for both multielement and multiwavelength analysis. The system can utilize either Smith-Hieftje or deuterium arc background correction systems. Software features allow for extended linear range analysis of samples to be conducted.

Farah and Sneddon (1993b) utilized a sequential simplex to optimize the operating conditions for a multielement measurement of Cu, Fe, Mn, and Zn by flame AAS. Factors that were varied included slit width, hollow cathode lamp currents of each element, air-to-fuel ratio, and height above the burner head. The simplex optimization was useful in attaining a region of optimum. Once the region of optimum had been attained, univariate searches can be used to verify that the simplex values were in the region of true optimum.


Hg lamp p h m mlllm~ ~, ~ gmt ln l

. ! , / r . . . . . d - - - / - _ . . . .

k " ~ l , , - I L . - = - ~ - ~ ' - " ' i ! I -

o.,,1

Figure 6. Optical diagram of Thermo Jarrell Ash-Baird AA-Scan 8 (Dulude, 1992) (note that only four HCLs are shown, another four are on top).

Berglund et al. (1993) constructed a simultaneous GFAAS system based on a Perkin-Elmer 4100ZL transversely heated graphite atomizer (THGA) connected to a computer-controlled power supply which also controlled the Zeeman magnet and hollow cathode lamps. Beams of up to four HCLs were split and imaged using toroidal mirrors prior to entering a 500-mm echelle spectrometer. One PMT was used for each of the four line sources. Both EDLs and HCLs can be used. The use of a longitudinal magnetic field at the atomizer proved to be the most effective means for background correction. As with commercial systems, the use of beam combining resulted in loss of intensity. The authors proposed that this may be reduced through the use of fiber optic cables.

Another drawback of simultaneous determination of real samples is the limited working range for multielement analysis. A three-field modulated ac Zeeman-effect method for background correction and extension of the working range was suggested by L'vov (1959). He also suggested that computerized linearization of calibration curves could also extend the working range. Harnly (1994) has recently evaluated computer modeling techniques for normal, linearized, and three-field Zeeman GFAAS. Using linearized Zeeman, the calibration range was extended by 1.5 orders of magnitude. The study did not attempt to implement the linearized Zeeman calibration curves in actual determinations of samples.

A simultaneous multielement AAS with an inverse polychromator and fast fourier transformation (FFT) was developed by Kitagawa and Shimazaki (1993). The system is shown schematically in Figure 7. Simultaneous determination of up to 10 elements was possible. A plane grating is used to merge beams from the individual hollow cathode lamps. A solar blind PMT was used followed by a beam splitter, which then directed half the light into a wide-range PMT. The atomic absorption signals were then discriminated through FFr. Detection limits were poor compared with those of single element systems due to relatively poor time resolution using the FFT processor. The authors proposed that a multielement lock-in


! 4

a Cd

F'II Co Ni

14

/ ~ 1 3

17 I '-"H 116

Figure 7. Schematic diagram of the simultaneous multielement AAS (Kitagawa and Shimazaki, 1993). 1, hollow cathode lamps; 2, lenses (f = 120 mm); 3, plane grating (2400 grooves mm-1 ); 4, lens (f = 80 mm); 5, the first focal point; 6, lens (f = 120 mm); 7, graphite tube atomizer; 8, lens (f = 120 mm); 9, photomultiplier tube (R166); 10, beam splitter; 11, photomultiplier tube (R3060); 12, absorption filter or monochromator (f= 100 mm); 13, graphite apertures (2 mm in diameter); 14, dc cu rrent-to-ac voltage converters with a high-pass filter (> ca. 800 Hz); 15, analog adder; 16, fast recorder; 17, printout.

amplifier may be used as an alternative to the FFT processor. The lack of background correction also caused poor results.

Tong and Chin (1994) coupled a photodiode array spectrometer to a graphite furnace for simultaneous multielement AAS. They used multielement HCLs with spectral lines dispersed through a spectrograph with three selectable gratings. Sensitivities for Ni, Co, and Fe were within the same order of magnitude as single-element determinations with a PMT.

A recent introduction by Perkin-Elmer is the SIMAA 6000 which is a true simultaneous multielement graphite furnace atomic absorption system using solid state detection and the Echelle polychromator. A schematic diagram of the system is shown in Figure 8. The system is capable of determining up to six elements simultaneously and uses the transverse heated graphite atomizer and longitudinal Zeeman-effect background correction.

Leeman Labs has recently introduced the Analyte 5 Simultaneous Graphite Furnace AA System. This is a dedicated system capable of simultaneously determining As, Pb, Se, Sb, and T1 in about 2 minutes.


STM' LS !

$TM ' . . . . " a P-E 21 I

ECH

P M T , P R A

PC AID

! ! I i i m m . , i ~ - - m |

FO

PC 2 1 0 0

Figure 8. Block diagram of Perkin-Elmer 6000 system (Berglund et ai., 1993) (this was a based on a modified P-E 2100 system). LS, element specific light sources; STM, semi-transparent mirrors; TM, toroid mirror; THGA, transversely heated graphite atomizer; P-E 2100, modified Perkin-Elmer model 2100 spectrometer; PC 2100, personal computer to control the P-E 2100; ECH, echelle polychromator; FO, fiber optic cables; PMT, photomuitiplier tubes; PRA, preamplifier for the PMT signals; PC A/D, personal computer with analog to digital conversion for data acquisition and processing.

D. Muitielement Systems Using Lasers

The semiconductor laser as a light source offers the possibility of multielement analysis, background correction, and internal standardization (Hergenroder and Niemax, 1989). Another advantage of the laser as a light source is that the dynamic range can be extended toward higher concentrations by tuning the laser into the wings of the absorption lines. A primary disadvantage of the laser to date has been the limited wavelength range (660-860 mm). With the use of frequency doubling, however, a diode laser radiation in nonlinear crystals can be extended to a spectral range of about 315 nm (Hergenroder and Niemax, 1989).

The use of a tunable semiconductor laser as a multielement light source for AAS has been reported by Hergenroder and Niemax (1989). The intensities of two lasers were modulated sinusoidally at different frequencies. A Fourier analyzer was used to discriminate the two absorption signals recorded by a single photodiode. The determination of more than two analyses increases the quality and cost of the Fourier analyzer required. However, to overcome this problem, Groll and Niemax


(1993) developed a "quasi-simultaneous," time-multiplex technique for an analysis. One laser is tuned to resonance for a short period of time while the other laser is off-resonance. The laser wavelengths are changed using a frequency driver. Multielement determinations using laser AAS indicated that the limiting factor was the noise of the flame and graphite-tube atomizer. Characteristic masses given by L'vov for conventional single-element AAS were typically 1 to 2 orders of magnitude larger. Groll and Niemax (1993) suggested that optimization of the operating conditions would be required to produce improvements in characteristic concentration of the analyte.

Ng et al. (1993) have used a multiple-mode diode laser as a multielement line source. A Zeeman background correction system was used. As with semiconductor lasers, the available wavelengths of the multimode laser were limited. However, the authors reported that only 11 wavelengths were available to use for analysis.

!11. FLAME ATOMIZATION APPLICATIONS

This section will present selected applications of the simultaneous multielement determination using flame atomization. Most of the initial work in simultaneous AAS used flame atomization and selected results are presented here which show the advantage of simultaneous AAS.

Farah and Sneddon (1993b), utilizing a commercially available system produced by Thermo-Jarrell Ash-Baird, showed that optimization of operating systems can significantly increase detection limits when utilizing a system in multielement mode. Kane (1988) performed an optimization study, characteristic concentration, and detection limit study for the optimization of a nine-element analysis using the SIMAAC system. The results showed that matrix interferences, some approaching 75% of the certified value, could be resolved by optimization of the system prior to analyses.

A. Multichannel Systems

Jackson et al. (1974) utilized a silicon-target vidicon detector for the simultaneous determination of trace wear metals in used lubricating oils. Twenty samples of used aircraft lubricating oil were obtained from engines with no worn components, thus ensuring that the metals to be analyzed were in relatively low concentrations (gg/g). Dilution with methyl isobutyl ketone (MIBK) was required prior to analysis due to the high viscosity of the samples. One limitation that was noted was that the detector was not suitable with a nitrous oxide-acetylene flame due to atomic emission interferences.

Aldous et al. (1975) also utilized a vidicon atomic absorption spectrometer for the simultaneous determination of seven trace metals in potable water. While detection limits were higher than those for the single-channel analysis, they were


satisfactory. Some metals, such as Cr and Pb, could not readily be determined due to either spectral interferences or inadequate sensitivity.

B. SIMACC

Kane and Hamly (1982) utilized the SIMAAC system for the multielement analysis of manganese nodules. Five manganese nodules, including two USGS reference nodules were analyzed for Co, Cu, Fe, K, Mg, Mn, Na, Ni and Zn. The nodules were digested utilizing two different digestion methods: 1) a high temperature digestion 2) a room-temperature digestion with nitric and hydrofluoric acids. Due to matrix interferences, poor recoveries, and/or poor precision several dilutions were made for each digestion. Three orders of magnitude were covered for each element, with four calibration standards per order of magnitude. Experimental results showed that the optimum individual recoveries were obtained when Co, K, Mn, Na, and Ni were determined simultaneously in concentrated digest. The simultaneous determination of Cu, Fe, Mg, Mn, and Zn in a dilute solution yielded optimum recovery values.

Optimization of air-acetylene flame parameters for the simultaneous determination of 11 elements (Ca, Co, Cr, Cu, Fe, K, Mg, Mn, Na, Ni, and Zn) was investigated by Harnly et al. (1983). It was noted that a compromise in detection limits and characteristic concentrations resulted when all the above elements were determined simultaneously. The determination of Mg required a relatively high observation height in the flame with a resulting decrease for the analytical recovery of the other elements. Recoveries for Ca, Fe, and Na were also poor. To overcome this problem, Harnly et al. (1983) utilized a 100-fold dilution of the sample digest in 5% La. This modification allowed for acceptable recoveries of Ca, Fe, Mg, and Na.

Several years later, Kane (1988) optimized flame parameters for the simultaneous determination of Ag, Bi, Cd, Co, Cu, Mn, Ni, Pb, and Zn utilizing a SIMAAC system. An open-beaker acid decomposition with a mixture of HF, HNO 3, and HC104 was used for dissolution of the samples. Matrix interferences were as high as 75% of the certificate value when using non-optimum measurement conditions. However, analytical results within 95-105% of the certified value were obtained when the air, acetylene, and observation height above the burner head were optimized.

Lewis et al. (1984a) analyzed blood serum for Ca, Cu, Fe, Mg, Na, K, and Zn. The extended range capability of the SIMAAC system was utilized for the analyses. This allowed for analyses to be performed utilizing only a single fivefold dilution. A comparison of the SIMAAC system with routine clinical methods such as colorimetric tests and flame photometry produced excellent results for all elements. The performance limits at the 95% confidence limit were calculated for the above elements and compared to the reference range (i.e., range of normal values in human serum) and were found to be acceptable for all the elements listed above.


C. Hitachi

Sakurada et al. (1991) utilized a Hitachi Model 170-50 with an air-acetylene flame for the simultaneous determination of silver and copper in silver brazing filler metals. Samples were first dissolved in nitric acid. For the analysis, the sample was irradiated alternately with the silver and copper HCLs. A high speed computerized data acquisition system was used to collect the absorption signals. Results were obtained using both peak area and peak heights. A comparison of the results indicated that the proposed system provided values which were in agreement with the reference values as reported by the manufacturer of the samples.

D. Thermo larrell Ash-Baird

Farah and Sneddon (1993a) utilized the Smith-Hieftje 8000 system for the eight inorganics included as part of the Environmental Protection Agency (EPA) Toxicity Characteristic Leaching Program (TCLP). Three different methods were required to analyze for Ag, As, Ba, Cd, Cr, Pb, and Se. Hg, the other metal normally analyzed for in the TCLP method, was not included since analysis is normally performed using cold-vapor techniques. After extraction of the metals from the solid wastes, each sample was spiked with known concentrations of contaminants. Three analyses, with four replicates, were performed on each sample. Ag, Cd, and Pb were analyzed using an air-acetylene flame with Smith-Hieftje background correction. Ba and Cr was analyzed simultaneously using an nitrous oxide/acetylene flame with no background correction. To reduce ionization of Ba in the flame, all samples were made so as to have a final concentration of 1000 gg/mL of K in solution. As and Se were analyzed simultaneously utilizing an ultraviolet background correction system produced excellent results and all elements had recoveries of between 85 and 102%.

Farah and Sneddon (1993c) compared background correction methods and dissolution techniques on National Institutes of Standards & Technology (NIST) estuarine sediment and bovine liver. Simultaneous analysis of Cu, Fe, Mn, and Zn, and Fe and Zn were performed on both of the above standards. The use of Smith-Hieftje background correction produced results which were close to accepted values than either no background correction or deuterium background correction for Cu, Fe, and Zn. Poor results were obtained when either the Smith- Hieftje or deuterium systems were used for the 372.0-nm line of Fe. This was due to a double-valued effect when using the Smith-Hieftje system and the lack of intensity of the deuterium continuum source in this region.

IV. GRAPHITE FURNACE APPLICATIONS

This section will present applications for simultaneous multielement determination using graphite furnace atomization. A summary of selected applications for graphite


Table 1. Selected Summary of Multielement Graphite Furnace AAS Applications Elements System Matrix Reference

Cd, Cr, Cu, Mn, Ni, Sb, FREMS Sewage sludge, estuarine Lehnert et al. (1993) Zn sediment, and

phosphate rock Cd, Mn, Pb, V Hitachi Waste waters Kitagawa and Shimazaki

(1993) Kitagawa and Shimazaki

(1993) Sen Gupta (1993a)

Ba, Ca, Mg, Sr Hitachi Soda industry

Sy, Er, Eu, Ho, Nd, Sc, Hitachi Sm, Tm, Y, Yb

Ag, Au, Ir, Pd, Pt, Rh, Ru Hitachi

Co, Cr, Fe, K, Mg, Mn, SIMAAC Na, Ni, Zn

Co, Cr, Cu, Fe, Mn, Mo, SIMAAC Ni, V, Zn

Ca, Cu, Fe, K, Mn, Na, Zn SIMAAC

Co, Cr, Cu, Fe, Mn, Mo, SIMAAC V, Zn Cu, Pb Dual-Channel

Cd, Zn Dual-Channel

A1, Cu, Fe

Au, Bi, Cd, Co, Cu, Fe, Mg, Mn, Ni, Pb

Ag, Co, Cr, Ca, Fe, Mn, Ni, Si

Pb/Se

As, Cd, Pb, Se

Cd, Pb

Cd, Cr, Mn, Pb

Cd, Cu, Pb

Time divided single channel

FFT and Inverse Polychromator

Multi-channel

Thermo Jarrell Ash-Baird



Thermo Jan'ell Ash-Baird

Perkin-Elmer prototype

Rocks and geological samples Silicate rock,

metallurgical samples Manganese nodules

Standard Reference acidified waters

Bovine liver, rice flour, and wheat flour

Serum

NBS orchard leaves, vehicle exhaust particulates, human organs, and water samples

NBS coal fly ash, NBS orchard leaves, and mineral feed

Aqueous solutions

Orange juice

Hair

Tap water

Saline solutions

Certified blood samples

Cigarette smoke

Seawater

Sen Gupta (1993b)

Kane and Hamly (1982)

Hamly and Kane (1983)

Carroll et al. (1985)

Lewis et al. (1984a)

Kumamaru et al. (1989)

Lawson et al. (1982)

Nakamura and Kubota (1990)

Kitagawa and Shimazaki (1993)

Alder et al. (1976)

Dulude and Sotera (1993)

Sneddon and Farrah (1994), Farrah (1993)

Sneddon and Deval (1995)

Lee et al. (1996a,b,c), and Sneddon et al. (1995)

Berglund et al. (1993)


furnace analysis is shown in Table 1. The following discussion is divided into different systems used for multielement AAS.

A. FREMS

The FREMS AAS with deuterium background correction was used to test for Cd, Cr, Cu, Mn, Ni, Pb, and Zn in sewage sludge (BCR/CRM 146), estuarine sediment (BCR/CRM 277), and phosphate rock (BCR/CRM 32). All three reference materials were digested according to DIN 38414-$7 prior to analysis (Lehnert et al., 1993). In a determination by the commercial Instrumentation Laboratory (now part of the Thermo Jarrell Ash-Baird Company) an AASpektrometer IL 951 was used for a reference measurement. Cd, Ni, and Zn were measured with deuterium arc background correction. The method of standard additions was used for all elements except zinc.

B. Hitachi

Early applications for the multielement system currently manufactured by Hita- chi were reported by Yasuda et al. (1989). The prototype system used four gratings and four detectors corresponding to the four HCLs. Background correction was done using the polarized Zeeman effect. The results of analysis for waste water from the electronic industry, the soda industry, and two beverages were reported. The use of a pyrometric control was shown to produce better reproducibility than a constant current heating method. The simultaneous analysis of Cd, Mn, Pb, and V in waste water gave approximately the same values as reported for conventional AAS. A saturated solution of sodium chloride from the soda industry was analyzed for Ba, Ca, Mg, and Sr. Concentrations of the elements varied by as much as 103 to 104 times. The authors reported that even with the large concentration differences and varied background absorbances, the analyses produced acceptable results.

Sen Gupta (1993a) used the Hitachi Z-9000 to determine a group of four rare earth elements simultaneously in rocks and geological standards in one firing. These included Er, Eu, Dy, Ho, Nd, Sc, Sm, Tm, Y, and Yb at the low pg/g to ng/g levels. After dissolution of the rocks and preconcentration by cation-exchange chromatography or coprecipitation with calcium oxalate and hydrated iron oxide, the method was developed. Intereferences were corrected by Zeeman background correction and the method applied to Standard Canadian Certified Reference Geological Standards. Results obtained compared satisfactorily with an independent study using ICP-AES and inductively coupled plasma-mass spectrometry (ICP-MS).

A further study by Sen Gupta (1993b) determined up to four noble elements simultaneously in silicate rocks and metallurgical samples using the same system. Dissolution was achieved using hydrofluoric acid and aqua regia, and preconcentration by ion exchange chromatography. The method was tested for Ag, Au, Ir, Pd, Pt, Rh, and Ru with three Canadian certified reference materials and then applied


to the determination of ng/g amounts of these elements in four new Canadian candidate reference materials.

C. Dual-Channel

Graphite furnace AAS is required for analysis of several inorganics in the EPA's contract lab program (CLP). The use of GFAAS is required for analysis of As, Pb, TI, and Se. Use of conventional single-channel systems can produce a backlog in the laboratory analysis.

The automated simultaneous determination of As/Se and Pb/TI using a two- channel AAS was reported by Dulude and Sotera (1988). The method of standard additions was used along with automated sample introduction. A chloride interference was overcome by adding 0.04% ammonium hydrogen phosphate as a chemical modifier. Magnesium resulted in an enhancement of the T1 signal. This was overcome by the addition of 500 parts per million (ppm Mg) (as magnesium nitrate) as a chemical modifier.

The determination of Pb and Se in tap water and secondary effluent water (discharge from a sewage treatment piano has been described by Atsuya et al. (1991). In many cases, the instrumental sensitivity for determination of Pb and Se using GFAAS is less than necessary for the levels present. To increase analytical sensitivity, a miniature cup technique was applied to coprecipitation samples obtained using a nickel/pyrrolidine dithiocarbamate complex. An atomization time of 14 s was chosen to eliminate the effect of Cu contamination in the samples. Since the absorbances of Pb and Se were shown to remain constant between 500 and 800 ~ an ashing temperature of 600 ~ was chosen. Using these compromise conditions, absolute sensitivities for Pb and Se were 0.2 ng and 0.5 ng, respectively.

Kumamaru et al. (1989) have used a dual-channel system for the simultaneous determination of Cu and Pb by GFAAS. The accuracy of their method was examined using orchard leaves (NBS standard reference material), vehicle exhaust particulates (National Institutes for Environmental Studies, Environmental Agency of Japan-certified reference material), human organs, and water samples. This determination required the use of a liquid-liquid extraction of the ion pair using tetradecyldimethylbenzylammonium (zephiramine). The simultaneous determination enabled the researchers to reduce the amount of sample size and reagents typically required for a conventional extraction. The compromise set of instrument operating conditions included a drying stage of 180 ~ s, ashing temperature of 1050 ~ s, and atomization of 2750 ~ s. Decreasing the argon flow rate increased the sensitivity for Pb. Results for the samples listed above were comparable with certified values in all cases.

Lawson et al. (1982) demonstrated the use of their dual-channel, single-grating AAS in a dual-element analysis of solids. Biological samples such as mineral feed and NBS orchard leaves produced acceptable results. Recoveries of Zn and Cd from NBS coal fly ash produced poor results for Zn. To attempt to increase recovery for


Zn, an atomization temperature of 2100 K was used. This also failed to produce acceptable results.

D. SIMAAC

The most diverse applications for simultaneous GFAAS have been studied by Harnly and co-workers using various configurations of the SIMAAC system. Harnly et al. (1983) showed that an extremely volatile element such as Zn could be simultaneously determined with an extremely nonvolatile element such as V. A residual peak for zinc continued to be present in the atomization profiles even after replacement of the graphite rod and end cones. Further studies indicated that for multielement analysis, it was crucial to measure peak area versus peak height for the SIMAAC system. Harnly et al. (1983) showed that there is considerable uncertainty in the peak height measurement. They also showed that the linearity could be increased by one-half order of magnitude when peak area was used for calibration instead of peak height. This occurred even when the system was not specifically optimized for each element.

Using SIMAAC, Harnly and Kane (1983) compared the performance results for the simultaneous determination of nine elements in the National Bureau of Stand- ards (NBS) (now called the National Institutes of Science & Technology) (NIST) acidified waters, Standard Reference Materials (SRM) 1643 and 1643a. Peak height and peak area measurements and atomization from a pyrolyzed tube and pyrolzed platform were all compared. The NBS certified values fell within the 95% confidence limits of the SIMAAC determinations with Zn being the exception. Low recoveries for Zn were found for both the pyrolyzed tube and platform using peak height measurement mode.

Lewis et al. (1984a,b,c) also used SIMAAC in the simultaneous multielement analysis of microliter quantities of serum. It was shown that peak height measurements were adversely affected by the organic and inorganic constituents in the serum. The peak height was substantially altered by matrix components that changed the atomization rate and/or the vaporization temperature. To reduce chloride interferences present in the serum, nitric acid and magnesium nitrate are commonly used. In the multielement analysis this was not possible due to the volatility of Zn. A combination of nitric acid and ammonium nitrate reduced the chloride interferences by about 60%. Peak area measurements were used with atomization from platform in a pyrolytically coated graphite tube. Serum specimens were diluted 1:21 with Triton X-100, nitric acid, and ammonium nitrate. Results showed that the mean SIMAAC values for the bovine serum were within one standard deviation of the assay values.

Using a graphite probe, Carroll et al. (1985) performed simultaneous determinations for Ca, Cu, Fe, K, Mn, Na, and Zn in bovine liver, rice flour, and wheat flour. Bovine liver had previously been analyzed using a platform atomization with the SIMAAC system. The accuracy and precision of the probe were not as good as


those using the platform. In comparison to tube-wall atomization, however, the accuracy of the probe was better, although detection limits were worse by a factor of 2 for the involatile elements. The authors suggested that this may have been due to chemical interferences. The probe does not give better or improved detection limits than those of platform atomization. It was also noted that the agreement of SIMAAC values and certified values varied for different elements in different samples. For example, the K concentration showed good agreement with the certified value in the rice flour; a much greater discrepancy was found for the concentration of the K in the wheat flour.

A comparison of the performance of three furnace systems with respect to multielement AAS, interference effects, and carryover contamination was conducted by Lundberg et al. (1988) using a SIMAAC system. Hamly and Kane (1983) conducted most of their work using a Perkin-Elmer HGA-500 furnace. This furnace had the disadvantage of being temporally nonisothermal although this problem can be reduced through use of a platform. The use of either an integrated contact (IC) cuvette or a two-step atomizer were shown to further increase performance of the SIMAAC system. The advantage of the two-step furnace is that a consecutive two-stage, constant-temperature atomization procedure can be used. For a simultaneous determination of A1, Cd, Cr, Mn, and Pb, the sensitivity for Cd and Pb were increased twofold according to the work done by Lundberg et al. (1988). A further advantage of using a spatially and temporally isothermal furnace is a reduction in both spectral and nonspectral interferences. Since multielement analysis requires that the ashing temperatures has to be adapted according to the most volatile element, it is common to have large amounts of matrix remaining after ashing. The two-step furnace and IC cuvette have been shown to produce a greater reduction of interferences in multielement analysis. This is due to a greater atomization efficiency.

E. Time-Divided, Single-Channel

The use of multielement HCLs for multielement graphite furnace AAS has not been successful. Nakamura and Kubota (1990) used a single detection channel with one PMT and a time-divided, high-speed computer acquisition system in conjunction with a multielement HCL for the determination of A1, Cu, and Fe. Although the precision and calibration ranges were comparable to a single-element system, the sensitivity was not satisfactory. This was due to a lack of linearity at the lower limits of the linear calibration ranges.

F. Thermo larrell Ash-Baird

Sneddon and Farah (1994) investigated the simultaneous determination of As, Cd, Pb, and Se in a saltwater (0.5 M) solution. Nickel nitrate and mixed palladium nitrate-magnesium nitrate chemical modifiers were compared for their effective- ness in the simultaneous analysis. The mixed palladium nitrate-magnesium nitrate


modifier was found to be preferable since higher pyrolysis temperatures could be employed. A pyrolitically coated, delayed atomization cuvette (DAC) with platform was used to overcome or compensate for the complex pyrolysis profiles resulting when an uncoated DAC was used. An initial study by Farah (1993) showed that the analysis of artificial seawater was difficult due to background overcorrection when the Smith-Hieftje background correction system was used.

Sneddon and Deval (1995) simultaneously determined Cd and Pb in blood reference standards using the Thermo Jarrell-Ash-Baird 8000 system (now called AA-Scan 8) with Smith-Hieftje background correction. The ashing profiles for Cd and Pb are given in Figures 9a and 9b, respectively (they are shown separate for clarity but data were obtained in the simultaneous mode). The use of a chemical modifier, specifically potassium hydrogen phosphate, was necessary to allow ashing and removal of the blood matrix without loss of Cd. This highlights one of the problems when a simultaneous determination using two or more elements is being proposed, namely the compromise in experimental conditions. Optimum ashing temperature for Cd is around 250 ~ whereas the optimum ashing temperature for Pb is around 550 ~ In the simultaneous mode, the ashing temperature cannot be used above 250 ~ without loss of Cd. In this work the use of chemical modifiers allowed the use of a higher ashing temperature (Deval and Sneddon, 1995). Detection limits in the simultaneous mode were comparable to the single mode and were 0.2 ~tg/L for Cd and 1.06 ~tg/L for Pb. The method was applied to certified blood reference materials with satisfactory results.

Lee et al. (1996a,b,c) and Sneddon et al. (1995) describe the combination of a single-stage impactor and a graphite furnace AAS system for the direct and near real-time multielement determination of Cd, Cr, Mn, and Pb in cigarette smoke. The absorption profile for these four elements in cigarette smoke is shown in Figure 10. This again highlights a potential problem in simultaneous graphite furnace AAS, namely that various atomization times and atomization temperatures will depend on the elements being determined. A compromise in atomization time (in this case around 6 s is needed for Cr whereas around 2 s is needed for Cd) and atomization temperature (around 2400 ~ for Cr and around 1400 ~ for Cd) must be used. In practice, the higher atomization temperature and longer atomization time is used in the simultaneous multielement graphite furnace AAS determination. The major advantage of the impaction system was that it allows for direct collection of particles in air on the graphite furnace followed by the determination stage. Results showed background levels of around low ng/m 3 in air could be increased 10-fold in cigarette smoke. Results were regarded as semiquantitative due to lack of suitable standards (Lee et al., 1996a).

G. Fast Fourier Transform (FFT)

Kitagawa and Shimazaki (1993) evaluated a simultaneous multielement AAS system with an inverse polychromator and FFT for the analysis of up to 10 elements

0.2

0.1

f

0.0

i 0.1 r

a

0 200 400 600 800 1000

Temperature, ~

0.2 . . . . . . . . . b

0.0 0 200 400 600 800 1000


Temperature, ~

Figure 9. Ashing or pyrolysis profile of simultaneous detection of cadmium (a) and lead (b). No modifier, magnesium nitrate, ammonium dihydrogen phosphate (Deval and Sneddon, 1995).


I I I I I i Hill

--I , '~ m

ca 2350 0.30

o

i j i i i I I l l

0 6 ATOMIZATION TIME, s

Figure 10. Absorption profiles of cadmium, chromium, lead, and manganese obtained from cigarette smoke collected on an impaction-graphite furnace system and subsequently determined by simultaneous multielement AAS (Lee et al., 1996a).

in orange juice. Palladium nitrate was used as a chemical modifier. Recoveries for an orange juice spiked with Au, Bi, Cd, Co, Cu, Fe, Mn, Ni, and Pb using the chemical modifier ranged from 77 to 120%. The authors noted that recoveries for Cd, Mn, and Ni were enhanced when palladium nitrate modifier was added. However, without the modifier, analyte losses for all elements were significant. This work again highlights a potential problem in simultaneous graphite furnace AAS, namely different calibration curves. This is illustrated in Figure 11 which shows the calibration curves, in absolute mass, for mixed solutions of the 10 elements under investigation in the absence of a chemical modifier and in the presence of 1 lag of Pd (as nitrate). Clearly different elements have different calibration curve ranges. It should also be noted that the lack of linearity in the calibration curves (typically 2 to 3 orders of magnitude above the detection limit compared to 5 to 7 orders for inductively coupled plasma-atomic emission spectrometry) can be considered a weakness of simultaneous AAS (both flame and graphite furnace).

H. Perkin-Elmer SIMAA

Simultaneous determination of Cd, Cu, and Pb in seawater samples following preconcentration and matrix separation in a flow injection system was demonstrated by Bergland et al. (1993). A spatially isolated, transversely heated graphite atomizer with integrated platform (THGA) and Zeeman background correction was used. A major drawback of multielement AAS is that all the elements to be determined must lie within the working range of the instrumentation.


1 0 . . . . . . . .~l..~ ~.-~Co 1o Mn Au ~o

~ / . t l Bi

1" , ( / C d ~ Bi 1 Cd

0.1 "g'" .

,~ , oo

: o2~ i lb~o:o~oo oi : .... 1o z ; o ~obo ,,.~ 10 ' ' - 10- ' ' N i

Mg . Mg Fe-

Z '~ 1 1 o Pb

4 Ol ~" ~' 0.1 , , , L 1 l L 0 |

o.o~ o.1 i lb zbo looo o,bl .z : ~o ~oo 1ooo Mass."ng M a s s ; n g

Figure 11. Calibration curves obtained by simultaneous measurements of 10 elements. Right, one l.tg Pd modifier added (as the nitrate); and left, no modifier added (Kitagawa and Shimazaki, 1993).

To overcome or compensate for this problem, Bergland et al. (1993) used a concentration elution step prior to the determination to select optimum concentration ranges for the analytes. This step also simplifies the matrix. Results showed that determined and certified values were in good agreement. Detection limits were about 10 times poorer than in the single-element mode.

V. M U L T I E L E M E N T S Y S T E M S U M M A R Y

Several different techniques have been used for multielement determination in atomic absorption spectrometry (AAS). A summary of multielement AAS systems is given in Table 2. The advantages and disadvantages of each system are summarized.

Multielement AAS is receiving increased interest as an alternative to traditional single-element AAS for determination of trace and ultratrace levels of metals. This is primarily due to the recent commercial availability to these systems.


Table 2. Advantages and Disadvantages of Multielement AAS Systems System Advantages Disadvantages

Multielement lamps (Fulton et al., 1970)

Continuum source (Harnly et al., 1979)

Multichannel systems (Kitagawa and Shimazaki, 1993)

Laser systems (Hergenroder and Niemax, 1989)

Several elements can be analyzed using only one lamp.

Low detection limits. Multielement detection of up to 20 elements. Accurately corrects for background absorption.

High sample throughput. Commercially available. Shorter analysis time. Expansion of dynamic working range.

Possible extension of dynamic range. Internal standardization can be used.

Spectral interferences. Poor S/N ratio. Limited availability of lamps.

Not commercially available. Susceptible to chemical interferences. Poor detection limits below 280 nm.

Costly. Susceptible to chemical interferences.

Limited wavelength range. Not commercially available.

Multielement systems offer an alternative to the more expensive simultaneous inductively coupled or direct current plasma spectrometers. In addition, multielement graphite furnace AAS offers better sensitivity and detection limits than either type of plasma.

Current commercially available systems use multichannel systems. The literature shows, however, that there is still much research required to make full use of the capabilities of these systems and to identify possible shortfalls. Several problems that need to be addressed in multielement graphite furnace AAS include the need for a universal modifier or a chemical modifier which suits a particular group of elements to be determined. The lack of linearity in AAS will limit the analyses which can be performed in a simultaneous mode. However, there appears to be a great deal of potential in the use of multielement systems since such systems will significantly reduce the amount of time required for analysis and will require a smaller sample for analysis.

ACKNOWLEDGMENTS

The authors gratefully acknowledge the support of Thermo Jarrell Ash-Baird Corporation, in particular Zach Moseley, Gerry R. Dulude, and John J. Sotero. This work was supported, in part, by Louisiana Education Quality Support Fund (LEQSF) Research Program for 1994-96-RD-A-21.

REFERENCES Alder, J.A., Alger, D., Samuel, A.J., West, T.S. Analytica Chimica Acta 1976, 87, 301. Aldous, K.M., Mitchell, D.G., Jackson, K.W. Anal. Chem. 1975, 47(7), 304. Atsuya, T., Itoh, K., Ariu, K. Pure & Appl. Chem. 1991, 63, 1221.


Bergland, M., Frech, W., Baxter, D.C., Radziuk, B. Spectrochim. Acta 1993, 48B, 1381. Butcher, D.J., Sneddon, J. Practical Guide to Graphite Furnace Atomic Absorption Spectrometry, John

Wiley & Sons: New York, 1997. Carroll, J., Miller-Ihli, N.J., Harnly, J.M., Littlejohn, D., Ottaway, J.M., O'Haver, T.C. Analyst (London)

1985, 110, 1153. Dulude, J.R. Spectrosc. 1992, 1, 3. Dulude, J.R., Sotera, J.J. Paper # 1133, 39th Pittsburgh Conference on Analytical Chemistry and Applied

Spectroscopy, 1988. Farah, K.S. Ph.D. Dissertation, University of Massachusetts-Lowell, 1993. Farah, K.S., Sneddon, J. Am. Environ. Lab. 1993a, 5(2), 1. Farah, K.S., Sneddon, J. Talanta 1993b, 40(6), 879. Farah, K.S., Sneddon, J. Anal. Lett. 1993e, 26(4), 709. Farah, K.S., Sneddon, J., Farah, B.D. Microchem. J. 1993, 48(3), 318. Farah, K.S., Sneddon, J. Appl. Spectrosc. Rev. 1995, 30(4), 351. Felkel, H.L., Pardue, H.L. Anal. Chem. 1977, 49, 1112. Fulton, A., Thompson, K.C., West, T.S. Anal. Chim. Acta 1970, 51,373. Golloch, A. Proceedings 41st Chemical Conference, London, 1988, 123. Groll, H., Niemax, K. Spectrochim. Acta 1993, 48B, 633. Hamly, J.M. Appl. Spectrosc. 1994, 48(9), 1156. Hamly, J.M., Kane, J.S., Miller-Ihli, N.J. Appl. Spectrosc. 1983, 36, 637. Harnly, J.M., Kane, J.S. Anal. Chem. 1983, 56, 48. Harnly, J.M., O'Haver, T.C., Golden, B., Wolf, W.R. Anal. Chem. 1979, 52, 2007. Harnly, J.M. Anal. Chem. 1986, 58, 933A. Hergenroder, R., Niemax, K. Trends Anal. Chem. 1989, 8, 333. Horlick, G., Codding, E.G. Appl. Spectrosc. 1975, 29, 167. Jackson, K.W., Aldous, K.M., Mitchell, D.G. Appl. Spectrosc. 1974, 28, 569. Kane, J.S.J. Anal. Atom. Spectrom. 1988, 3, 1039. Kane, J.S., Harnly, J.M. Anal. Chim. Acta 1982, 139, 297. Kitagawa, K., Shimazaki, Y. Anal. Sci. 1993, 9, 663. Kumamaru, T., Okumoto, Y., Hara, S., Matsuo, H., Kiboku, M. Anal. Chim. Acta 1989, 218, 173. Lawson, S.R., Nichols, J.A., Viswanadham, P., Woodriff, R. Appl. Spectrosc. 1982, 36, 375. Lee, Y.I., Smith, M.V., Indurthy, S., Sneddon, J. Spectrochim. Acta 1996a, 51B(1), Number 1,109. Lee, Y.I., Indurtha, S., Smith, M.V., Sneddon, J. Anal. Lett. 1996b, 29(14), 2515. Lee, Y.I., Smith, M.V., Indurtha, S., Sneddon, J. J. Anal. Atom. Spectrom. 1996c, 11, in press. Lehnert, R., Quick, L., Rump, T., Winter, F., Cammann, K. Fresenius'J. Anal. Chem. 1993, 346, 392. Lewis, S.A., O'Haver, T.C., Harnly, J.M. Anal. Chem. 1984a, 56, 1651. Lewis, S.A., O'Haver, T.C., Harnly, J.M. Anal. Chem. 1984b, 57, 1. Lewis, S.A., O'Haver, T.C., Hamly, J.M. Anal. Chem. 1984e, 57, 1066. Lundberg, E., Frech, W., Harnly, J.M.J. Anal. Atom. Spectrom. 1988, 3, 1115. Lundberg, E., Johansson, G. Anal. Chem. 1976, 48, 1922. L'vov, B.V. Spectrochim. Acta 1984, 39B, 159. A translation from Inzh. Foz. Zh. 1959, 2, 44. Masters, R., Hsiech, C., Pardue, H.L. Talanta 1989, 36, 133. Nakamura, S., Kubota, M. Analyst (London) 1990, 115, 283. Ng, K.C., Abdalla, H.C., Barber, T.E., Winefordner, J.D. Appl. Spectrosc. 1993, 44, 849. Niu, W., Haring, R., Newman, R. Am. Lab. 1987, 19(11), 40. Retzik, M., Bass, D. Am. Lab. 1988, 20, 70. Salin, E.D., Ingle, J.D. Appl. Spectrosc. 1978a, 32, 579. Salin, E.D., Ingle, J.D. Anal. Chem. 1978b, 50, 1745. Sakurada, O., Tanaka, S., Taga, M., Kazizaki, T. Analyst (London) 1991, 116, 31. Sen Gupta, J.G.J. Anal. Atom. Spectrom. 1993a, 8, 93. Sen Gupta, J.G. Talanta 1993b, 40(6), 791.

Graphite Furnace and Flame AA5 61

Sneddon, J., Deval, A. Microchem. J. 1995, 52(1), 23. Sneddon, J., Farah, K.S. Spectrosc. Lett. 1994, 27(2), 257. Sneddon, J., Smith, M.V., Indurthy, S., Lee, Y.I. Spectroscopy 1995, 10(1), 26. Tong, S.L., Chin, K.S. Spectrochim. Acta 1994, 49B, 459. Van Rensburg, H.C., Zeeman, EB. Anal. Chim. Acta 1968, 42, 173. Yasuda, K., Okumoto, T., Yonetani, A., Yamada, H., Ohishi, K. 5th Colloquium Atomspectrometry

Spurenanal, 1989, p. 133. Zeeman, EB., Brink, J.A. Analyst (London) 1968, 93, 388.


DIRECT CURRENT ARCS AND P LAS MA J ETS

Rudi Avni and Isaac B. Brenner

II.

III.

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 A. Description of the DC Discharge . . . . . . . . . . . . . . . . . . . . . . . 65 B. The Vertical DC Arc (Free Burning) . . . . . . . . . . . . . . . . . . . . . 67

73 C. DC Plasma Jet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Systematic Behavior of the Analytes in the DC Discharge . . . . . . . . . . . . 74 A. Electrode Effects (EEs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 B. Plasma Effects (PE) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 Techniques in Spectrochemical Analysis by DC Arc Plasma . . . . . . . . . . 157 A. The Cathode Layer Technique . . . . . . . . . . . . . . . . . . . . . . . 157 B. The Cathode Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 C. Buffers, Fluxes, and Internal Standards . . . . . . . . . . . . . . . . . . 159 D. Carrier Distillation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 E. Development of General Schemes for Multielement Analysis . . . . . . 163 F. Analysis of Uranium, Thorium, Zirconium, and Plutonium Oxides . . . . 164

G. Rare Earth (RE) Oxides . . . . . . . . . . . . . . . . . . . . . . . . . . 172


63

64 R. AVNI and I.B. BRENNER

H. Rock Phosphate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176 I. Multielement Analysis of Silicate Rocks . . . . . . . . . . . . . . . . . . 181 J. Aluminum and Titanium Oxides . . . . . . . . . . . . . . . . . . . . . . 186

K. Molybdenum and Tungsten Oxides . . . . . . . . . . . . . . . . . . . . . 193 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

ABSTRACT

This chapter describes the direct current (dc) arc and plasma jets. Direct current arc and plasma spectrometry has been around for a number of years but is still an integral and indispensable method for determining metals in solids and liquids in many laboratories. This chapter describes the arc and plasma, and the physical and chemical interferences of the sample and its trace elemental constituents in the direct current discharge and their correlation with spectral line intensities of each trace element. The authors describe their experiences in the determination of trace elements in refractory type samples such as uranium, thorium, and plutonium oxides, rare earth oxides, rock phosphates, silicate rocks, aluminum and titanium oxides, and molybdenum and tungsten oxides.

I. I N T R O D U C T I O N

This chapter deals with the spectrochemical analysis of trace elements in solid or

liquid samples by direct current (dc) discharges (arc and plasma jets). Its main purpose is to describe the physical and chemical transformations of the sample and

its trace elemental constituents in the dc discharge and their correlation with spectral

line intensities of each trace element. Solid and liquid samples in the dc discharge undergo processes such as evaporation, dissociation, atomization, ionization, new molecule formation, among others, which affect or contribute to spectral line intensity and its relation with the unknown concentration of the chemical element to be determined. Topics such as the electric discharge, the structure of the atomic

or molecular spectra, light dispersion and diffraction, and detection of the light

intensity will not be treated in this chapter. Phenomenologically, the topics that will

be treated can be described schematically by the "Mandelshtam scheme" (Man-

delshtam, 1962; de Galan, 1965) namely:

1 2 3 4 5

Cx---~ Qx-.--~ nt---~ nex--+ Jx-.+ Sx

A chemical element x in a solid substance introduced in a probe into a dc arc or in a liquid sample aspirated into a plasma jet, at a concentration C, is related to the

measured spectral signal S by the following links:

Direct Current Arcs and Plasma Jets 65

C x or part of it, is evaporated into the plasma. The number of particles leaving the sample and injected into the plasma per unit time is defined as the volatilization rate Qx s-~ of the element x. The evaporated particles entering the plasma are subject to physical and chemical processes such as atomization, ionization, diffusion, convection, viscosity, electric fields, etc., which determine their residence times in the discharge region. The rate by which particles leave one region in the plasma for another is defined as the transport parameter of the element x:

t I J n , c m 3 S - 1 x

The total concentration of free particles (molecules, atoms and ions) per unit plasma volume is denoted by n x cm -3.

3. In the plasma the free particles undergo processes like atomization, dissociation, excitation, and ionization. The total amount of excited states (molecules, atoms, and ions) of element x emitting light at different wavelengths is denoted by nex.

4. The amount of energy emitted in unit time per unit volume of the plasma, along the optical axis of the spectrometer (solid angle), resultant from a particular quantum transition is the emittance Jx, as determined by the concentration (nex)2 in the upper level of the energy diagram (concentration of excited particles in the energy level 2).

5. The emmitance detector measures the integral of Jx for a period of time denoted by S x.

The spectrochemical topics relating to the concentration C x and the measured S x of a spectral line of the chemical element x will be treated quantitatively in the following sections of this chapter.

A. Description of the DC Discharge

In spectrochemical analysis of materials using dc discharges, the electrode material differs from the substance to be analyzed--graphite or carbon electrodes are commonly used. The discharge between the electrodes takes place in a gaseous media, for example, air, Ar + 0 2, He + 0 2, among others. Both carbon and air, for example, together with their reaction products constitute the major plasma population. Huldt (1948), Smit (1951), Boumans (1966), Finkelenburg and Maecker (1956), and Roes (1962) have shown spectroscopically the occurrence of the following species and calculated their concentration: N 2, N, N § N~, 02, O, O § O~, NO § C 2, C, C § CN, and CO.

In the plasma an energy balance between the electrical energy introduced in the discharge and its thermal losses per unit time is established by the population. According to Ellenbaas (1934) and Heller (1935), this energy balance in a steady- state dc arc plasma operating at atmospheric pressure is"

66 R. AVN! and I.B. BRENNER

jE = -div(p gray T) (1)

Where j is the current density in A cm -2, E is the electric field strength in V cm -1, p is the thermal conductivity expressed as erg cm-is-ldeg -1, and T is the absolute temperature. This equation indicates the nature of the variables governing the dc discharge, namely:

1. Electric variables such as the density of electrons and positive ions, their mobility and degree of ionization, in addition to current density and electric field strengths.

2. Thermal variables such as temperature and thermal conductivity and ther- mally dependent variables such as convection and diffusion.

The material balance in a dc discharge can be established by four basic equations, namely:

1. The summation of partial pressures in the system. For a dc arc in air, the summation can be expressed in terms of particle density as follows,

)-"n = 7.34 x 10 21/T (2)

where n represents the densities of molecules, atoms, and ions in the plasma. For example, for a temperature of 6000 K, the overall density for carbon or graphite electrodes and air at one atmosphere is 1.2 x 1018 cm -3.

2. The material balance for carbon, nitrogen, and oxygen; for example, the balance for carbon is:

nc = nc + 2nc2 + nc N + nc ~ (3)

3. The material balance of oxygen and nitrogen as their composition in air, namely:

n N + 2nN2 + nCN + (4) =3.7

n o + 2no2 + nco + nNO

4. The Saha equilibrium constant for a dissociation or an ionization process. For example, for a dissociation process (Boumans, 1966),

AB---~A + B

1 0-5 0 J' K d = ~ = 1.88 x 1 t-=-iT==j ' where for the respective particles, n is the concentration cm -3, M the mass, Z is the partition function, T is the absolute temperature, and V d is the dissociation potential of the molecule AB in eV.


The addition of a new matrix (the sample to be analyzed spectrochemically) to the two plasma matrices formed from carbon and air will result in changes both in the energy and the material balances (Maessen, 1974; Avni, 1978).

B. The Vertical DC Arc (Free Burning)

Mechanical Structure

The basic difference between the arc and the plasma jet stems from their mode of burning; vertical arcs burn freely with free convection, while the dc plasma jet has a suppressed plume with forced convection owing to the flow of gas around the discharge which also determines its geometry.

The free-burning arc consists of several parts: the core or column (inner zone), the mantle, and the zones immediately adjacent to the electrodes. The boundaries between the arc core and the mantle cannot be defined distinctly since they merge gradually one into the other (Boumans, 1966; Avni, 1968a). Photographs of the vertical dc arc shows a cylindrical core and a poorly defined mobile mantle (Brinkman, 1937; Ahrens and Taylor, 1961; Avni, 1968a). The former is produced by the anode crater (lower electrode) and is influenced by the diameter of the cathode (the upper electrode). The arc core can be considered as the positively charged part of the column (Smit, 1951; Boumans, 1966; Finkelenburg and Maecker, 1956). Photographs of the arc core for different arc gases are shown in Figure 1. The photographs were obtained using a Bolex H16cine camera with 1/100 exposure time and 16-64 frames per second (Avni, 1968a). The shape of the arc core in the figure is somewhat conical with a larger radius near the anode. The truncated cone of the arc core has been more clearly defined by densitometric measurements performed on the negative film (Table 1). Tables 1 and 2 show the relationship between core geometry and position as a function of the distance between the electrodes (up to an arc gap of 10 mm).

Spatial (axial and radial) distributions of line intensity, temperature, electron density, electric field strength, among other plasma gradients are more reliable for defining the arc core in comparison to images obtained by a cine camera. Avni (Avni, 1968a,b), defined three distinct axial zones in the core of a dc graphite arc in atmospheric pressure at 13 amp based on the axial distribution of spectral line intensity, excitation temperature, and voltage. An axial zone was defined where constant values of relative line intensities (for different atoms and ions), temperature, and voltage were obtained. Radial integration of a core of approximately 2-mm radius revealed the presence of the following three axial zones:

1. The anode zone, extending from 0.1 to approximately 0.5 mm above the anode. This range depends on the gap between the electrodes, namely from 2 to 10 mm, respectively.

2. The central zone strongly dependent on the arc gap. For example, for a 4-mm arc gap--from 1.5 to 3.0 mm above the anode; for a 6-mm arc gap--from


Figure 1. Plasma core images (a) 2.0 mm arc gap; (b) 4.0 mm arc gap; (c) 6.0 mm arc gap; (d) 8.0 mm arc gap (from Avni, 1968a, Figure 3).


Table 1. Densitometric Measurement of the Inner Zone of a 13-A DC Arc a'b

Gap 2.0 mm Gap 4.0 mm Gap 6.0 mm Gap 8.0 mm

Trans- Trans- Trans- Trans- Core mittance Core mittance Core mittance Core mittance Width Axial Width Axial Width Axial Width Axial (mm) (%) (mm) (%) (mm) (%) (mm) (%)

Near the cathode 3.10 42.0 3.10 47.0 3.20 34.0 3.15 37.0

Center 3.15 50.0 3.25 57.0 3.70 51.0 3.70 53.0

Near the anode 3.20 48.0 3.75 55.0 3.85 44.0 4.00 41.0

Notes: aAfter Avni, 1968a.

bMeasurement of a negative film. The measurements are averages over 25 pictures of the arc inner zone. Film 3 x Kodak, Cine-camera, Bolex H 16 with exposure--1/100 and 16-64 frames per second.

1.5 to 4.5 mm above the anode; for a 10-mm gap--from 2 to 8 mm above

the anode tip. The cathode zone extending from 0.2 to 0.5 mm below the cathode tip.

Other regions of interest in the dc arc discharge are those immediately adjacent to the anode and the cathode. The electric field created by the potential difference between the graphite electrodes and the conductivity caused by the free electrons

in the arc are the main factors which control the electric current in the arc core. The difference in the concentration of free electrons (e) and positive ions (i) determines

the space charge. A space charge in the core is formed due to the concentration of singly charged positive ions (only few doubly charged ions are formed in the dc arc) which is practically equal to that of the liberated electrons (Smit, 1951).

According to the Poisson relationship (Chen, 1965),

A E = 4rc9 (n i - ne) 8

Table 2. Lateral Shifting (mm) of the Inner Zone of a 13-A DC Arc a'b

Gaps

4.0 mm 6.0 mm 8.0 mm 10.0 mm

Near the cathode 1.5 2.7 3.5 4.5

Center 0.0 0.5 0.5 1.0

Near the anode 1.5 2.2 3.5 4.0

Notes: aAfter Avni, 1968.

bMeasurement of a negative film. The results represent the maximum shifting on one side, left or right of the axis of symmetry of the electrodes.

70 R. AVNi and I.B. BRENNER

the large space charge in the immediate vicinity of the electrodes, results in high electric field strengths of the order of several hundred volts per centimeter (Ecker, 1961). These potential decreases in the vicinity of the electrodes (the so-called "anode fall" and "cathode fall") were studied by Ecker (1961), Weizel and Rompe (1949) and Finkelenburg and Maecker (1956). The voltage change in the arc as a function of the distance from the cathode surface is represented schematically in Figure 2. The potential drop in the arc column is given by the linear section.

In Figure 2, Vtc is the total voltage over the cathode region, V c is the voltage in the contracted region, V~ is the voltage in the "cathode fall" extending over a distance governed by Debye shielding (Chen, 1965). V c covers a broad region of the order of 1 mm (Boumans, 1966). Similarly, VtA , V A, and VSa describe the anode voltage distribution.

History of the Free Burning Arc in Spectrochemical Analysis

Since the early 1930s, dc arcs have been used for quantitative analysis of geological materials (Strock, 1970). As early as 1925, Gerlach (1925) introduced the concept of internal standardization (intensity ratio method) and in 1931, in their "Geochemistry of Scandium" Goldschmidt and Peters (1931) correlated spectral line intensity with volatilization rates. The same authors (1933) in the "Geochem- istry of Ge" utilized the "cathode layer" method (developed in 1931 (Mannkopff and Peters, 1931)). These workers introduced the concept of the arcing period

Colhode

St "

,, ,

Anode

7 e i

X

Figure 2. Definition of cathode and anode fall regions, cr-contraction region, sr- space charge region. Reproduced from Boumans, 1966 and Ecker, 1961.


(pre-exposure and exposure) and stressed the importance of the dimensions of the crater and the matrix composition. During this period quantitative investigations on the geochemistry of Sc (Goldschmidt and Hefter, 1933), Li (Strock, 1936a) Ba (von Engelhardt, 1936), and La (Bauer, 1935) continued, for which the approach of a "completely burned sample" (total consumption) was developed.

Up to 1937, in quantitative spectrochemistry using the free-burning carbon arc, trace element spectral line intensities were directly correlated to matrix composition (differences were compensated by matrix matching procedures), crater dimensions, volatilization behavior, and the arcing period. The analytical application of the "cathode layer" (sample located in the cathode which is usually the lower electrode) was pioneered by Mannkopff and Peters (1931) and developed by Strock (1936b) and later by Mitchell (Mitchell, 1945, 1964; Scott et al., 1971), as a general spectrochemical technique for the analysis of soils, plants, and related agricultural materials.

The next milestone in the development of quantitative spectrochemical analysis occurred during the "Manhattan Project." Scribner and Mullin (1946) developed the fractional distillation approach, i.e. carder distillation, for the determination of trace elements in nuclear fuels such as uranium oxide. Numerous spectrochemical procedures used for the determination of trace elements, published in the period 1970 to the late 1980s, were based upon the principles elaborated during the 1930-1946 period. A review of quantitative spectrochemical procedures used for the analysis of geological, environmental, nuclear, and metallurgical materials shows that the methods can be classified into the following groups.

Chemical and Physical Separation of Trace Elements from their Matrices Prior to Spectrochemical Analysis. This approach is performed in order to reduce matrix effects in the electrode and in the plasma and to concentrate those trace elements having poor detection limits. Among the chemical separation methods, the most commonly utilized are precipitation, solvent extraction, and ion exchange (Avni, 1978; Hartenstein et al., 1985; McLaren et al., 1985; Horvath et al., 1991). Physical separation based on selective volatilization of several impurities by induction heating (Ahrens and Taylor, 1961), by the double arc, and the double arc furnace in air and in controlled atmospheres and in vacuum (Yukanovic et al., 1974, 1977; Nickel et al., 1984) have been described for the analysis of geological (Ahrens and Taylor, 1961) and nuclear materials (Zaidel, 1963; Nickel, 1984). However, chemical and physical separations are tedious compared to the convenience of direct spectrochemical analysis. The probability of contamination and loss of material due to multiple manipulations in a separation method is high.

Buffers or Fluxes-Matrix Modifiers. A statement that matrix effects have been overcome usually implies that the spectrochemical parameters for the standards and the samples are equal, thus line intensity is influenced to the same extent. There are several methods for compensating sample-standard compositional dif-


ferences. Geological materials, for example, are notoriously variable with respect to their chemical and physical composition. In order to overcome this problem, numerous standard reference materials have been prepared (Flanagan, 1969; Sine et al., 1969; Roelandts, 1990; Govindaraju, 1994) enabling calibration with materials which are similar to the samples. Nevertheless, ideal similarity between standards and samples is rarely achieved. As a result, matrix effects can be minimized by adding various chemical compounds to the samples and standards. These additives, called spectrographic buffers or modifiers, can compensate for differences in volatilization rates from the anode cup and for differences in the behavior of the particles and the atom cloud in the plasma. Buffers, such as lithium metaborate, persulfates, and others have been added in high concentration from 50 to 500% (Rossi, 1960; Ahrens and Taylor, 1961; Joensuu and Suhr, 1962; Boumans and Maessen, 1970; Govindaraju, 1970; Maessen, 1974). Additions of high concentrations can be regarded as matrix conversion or modification.

Carrier Distillation. Scribner and Mullin (1946) were first to employ carriers in analytical spectrochemistry. They developed a method for the determination of trace elements in uranium oxide. Two percent of Ga203 was added to the U308. Carriers and modifiers are widely employed for the determination of trace elements in refractory materials (Avni, 1968), also using electrothermal vaporization atomic absorption spectrometry (Matousek, 1981; Styris and Redfield, 1993), ICP-AES (Kirkbright and Snook, 1979; McLeod et al., 1992; Nickel and Zadkorska, 1995), and MS (Gregoire, 1988; Moens, 1995), and direct sample insertion inductively coupled plasma AES and MS (Brenner et al., 1987; Karanossios, 1989, 1990). The addition of these compounds to a matrix in a concentration range of 1-10% is considered to play the following roles:

1. Depression of the plasma temperature owing to the lower ionization potential of the carrier element (Avni and Chaput, 1961; Raikhbaum and Molych, 1961; Siemenova and Levchenko, 1962), and as a result reduction of the abundance of atom and ion spectral lines of the matrix (U, Th, Zr, Pu, and rare earths).

2. Selective transport of the trace elements from the matrix into the plasma as a result of the higher vapor pressure of the carrier (Atwell et al., 1958; Vainstein and Belayev, 1959; Vainstein, 1961).

3. Increase of the residence time of the trace element particles in the discharge zone (Goldfarb and Ilina, 1961; Samsonova, 1962; Vukanovic, 1964). Vain- stein (1959, 1961) employed radioactive tracers to study the spatial distribution of the elements in the plasma, and showed that the spatial distribution of most of the elements in the discharge zone was similar to the carder element. Samsonova (1962) demonstrated that carriers do not influence line intensity of the trace elements as a result of evaporation mechanisms of the matrix. Daniel (1960), classified trace elements in a Ga203-U30 8 matrix according to their relative ease of volatilization.


The large variety of carders and modifiers used in spectrochemical analysis can be explained by their limited suitability for various trace elements and matrices. The so-called common trace elements can be determined with satisfactory accuracy and sensitivity (Pepper, 1967); however the determination of numerous refractory impurities in uranium, thorium, and zirconium oxide matrices is less satisfactory and other procedures are employed (Avni, 1969).

C. DC Plasma Jet

In their extensive survey of electrical arcs and thermal plasmas, Finkel~nburg and Maecker (1956) showed that the maximum plasma temperature in the free-burning carbon arc amounted to 10,000 K and that an increase in current resulted in an increase of the arc diameter without raising its temperature. Gerdien and Lotz (1935) restricted the arc plasma near the electrodes, thus forcing the temperature to increase. Lorenz (1951) demonstrated that the axial temperature in constricted plasmas amounted to 35,000 K and that the temperature increased with increasing current. Maecker ( 1951) proceeded to restrict the entire plasma column by placing it in a water channel, and measured temperatures of about 55,000 K along the arc axis using a current of 1450 amp. He pointed out that such an arc could serve as a suitable emission source for quantitative spectroscopic measurements of transition probabilities, electron-ion collisions, and other physical properties of ionized plasmas. Jurgens (1952) investigated a low current (50 amp) Maecker-type water- stabilized arc and observed that the axial temperature was 12,730 + 200 K. He also showed that the plasma was in local thermodynamic equilibrium (LTE).

Maecker (1956a,b) was probably the first investigator to propose a wall-stabilized high current arc with stabilization being achieved by nitrogen, argon, and other gases flowing along the electrodes. Richter (1958) employed Maecker's wall- stabilized arc with six sections and developed the cascade arc. The arc was formed between two center copper plates with carbon dioxide being introduced into the axial zone between bored carbon electrodes. This wall-stabilized arc was used at the Institute for Experimental Physics of Kiel (Hey, 1959) as an intense stable light source for the quantitative spectroscopic measurement of transition probabilities of various excited atoms and ions, recombination processes, continuum spectra, diffusion processes in arcs, and other physical properties of ionized plasmas (Richter, 1961).

These developments in the 1950s, aimed at producing more intense and more stable light sources, were the foundations on which dc plasma jets were developed as a source for atomic emission spectrochemical analysis. As early as 1959, Margoshes and Scribner (1959) and independently Korolev and Vainstein (1959) used a plasma jet source for spectrochemical analysis. Their plasmas were generated within a cylindrical chamber by a dc current discharge between a narrow orifice (as the anode) and an annular graphite cathode. Argon or helium, which transported the sample aerosol, was directed tangential into the chamber and released through


the annular cathode. Owen (1961) attributed the instabilities of such a plasma jet to the gas stream. He stabilized the jet by introducing a second cathode (a horizontal rod) which intercepted the plasma at a point downstream from the graphite ring cathode (Owen, 1961).

Several review articles on the application of plasma jets have been published such as Olsen (1963) and Greenfield et al. (1975). References to applications using the dc plasma were published by Schramel (1988) and by Ebdon and co-workers (Ebdon and Sparkes, 1985; Zander and Miller, 1985; Sparkes and Ebdon, 1986; Ebdon et al., 1989; Armstrong and Ebdon, 1990). Kleinmann and Svoboda (1975), Butler et al. (1974), and Keirs and Vickers (1977) described various configurations for producing plasma jets as well as their application in spectrochemical analysis. The use of "plasmatrons" (the Russian term for plasma jets) in spectrochemical analysis was summarized by Zilbershtein (1977) and Zheenbaer et al. (1976). In several studies Keliher and co-workers demonstrated that dc plasma jets are less robust sources and are less susceptible to matrix variations (Keliher and Boyko, 1982; Fox, 1985; Keliher, 1991). In the following sections a systematic study will be made of analyte behaviors in free burning dc arcs and in plasma jets.

!1. SYSTEMATIC BEHAVIOR OF THE ANALYTES IN THE DC DISCHARGE

According to Mandelshtam's scheme (1962), the sample to be analyzed (in solid or in liquid form), reacts with the electrode material and volatilizes. Free particles are formed, transported by the plasma from one zone to the other, and undergo atomization and/or ionization. Finally spectral intensifies are measured and recorded. In order to comprehend the behavior of the analytes, the following topics will be treated in this section:

1. Behavior of the analytes in the electrode cups; namely "electrode effects" (EE).

2. Behavior of the analytes in the plasma; namely "plasma effects" (PE). 3. The influence of the analyte and the matrix on plasma variables; namely

"third matrix effects" (TME).

This approach can be exemplified by the behavior of uranium oxide (U308) and trace elements in a free-burning vertical arc. Samples of U308 mixed with graphite in a ratio of 1:5 were loaded into the crater of a graphite anode. After 100 s of arcing, the composition of the residual material was as follows U30 8 = 8%; UO 2 = 60%; UC = 18%, and UC 2 = 14% (Rautschke et al., 1975a). In other words from a matrix containing 20% U30 8, only 8% remained after 100 s of arcing. New compounds were formed resulting in a change in the evaporation rate of the trace elements (Avni, 1978) accompanied by a change in the temperature gradient in the anode crater (Nickel, 1968). If the composition of this new matrix, resulting from EE,


differs for samples and standards, then the spectrochemical determination of trace elements is no longer reliable.

In the plasma, the free particles of U, C, and any combination between them, together with those of the trace elements, are transported so that the uranium particles diffuse from the arc column (near the anode) toward the margins, while the majority of trace elements concentrate in the "cathode region" (Avni and Boukobza, 1969a,b). Owing to PEs, a "spatial separation" occurs between the matrix element (U) and the trace elements. PEs play an important role in the development of direct spectrochemical methods for the determination of trace elements in U30 8 and other complex types of materials (Avni, 1978). If PEs on the analyte-free particles are constant, the particle density in the plasma will be the same for both samples and standards (Avni and Boukobza, 1969b).

A. Electrode Effects (EEs)

The conventional quantitative determination of trace elements in a given matrix requires that samples are matrix-matched to the calibration standards. In the absence of materials that are similar in composition to the samples, synthetic standards are prepared by adding trace amounts of various oxides, carbonates, chlorides, and other chemical compounds to the matrix elements (major concentrations) which can take the form of metal powders, oxides, silicates and others. Thus the physical mixture usually differs from the samples which consist mainly of alloys, minerals, rocks, ceramics, etc. Moreover, samples undergo various chemical, mineralogical, or metallurgical transformations during their formation, and as a result, the trace elements are chemically and/or structurally bound to the bulk matrix. On the other hand, in synthetic standards, the trace elements are usually not chemically bound to the major constituents. Because of this fundamental difference between samples and standards, various additives (graphite, buffers, carriers, etc.) are introduced in order to reduce the possible differences in EEs. EEs have been attributed mainly to thermochemical reactions (Nickel, 1963, 1968; Rautschke, 1967, 1968; Rautschke et al., 1975a,b) between (1) matrix elements and the electrode material (usually carbon or graphite), (2) the trace elements (in the matrix) and the electrode material, and (3) the trace elements and the matrix.

Thermochemical Reactions between Matrix and Carbon or Graphite

Thermochemical reactions between the matrix and the anode material are related to thermal conditions in the electrode, i.e., the longitudinal temperature distribution. The physical differences between carbon and graphite will not be treated here and were studied by Euler (1956), Russman (1958), and Mellichamp and Grove (1978). A knowledge of the longitudinal temperature distribution is essential for understanding the chemical processes in the electrode cavity.

The energy input is derived mainly from the electrical energy dissipated in the anode fall, from energy released by electrons, the heat of combustion of the


electrode material (carbon), and the energy transferred from the plasma column to the anode either by radiation or by conduction. The energy dissipated in the anode fall (voltage drop x current strength) contributes by far the greatest portion of the total power input at the anode; the last two sources are considered to be negligible (Euler, 1956). Energy is consumed mainly by three processes--radiation, convection, and thermal conduction in the electrode cup.

The temperature of the carbon anode spot was measured to be 4000 K by Euler (Euler, 1956). At about 10 amp, the temperature decreased to 3700 K across a narrow layer approximately 0.05 mm thick. The temperature further decreased to about 3400 K at approximately 0.5 mm beneath the anode surface. Leuchs (1950) estimated the temperature of a carbon anode by observing the thermochemical behavior of various compounds and studying diffusion and volatilization phenomena in the anode cavity. This method was also used by other investigators (Nickel, 1965; Rautschke, 1965; Decker and Eve, 1968; Brill, 1969). However, the temperature in the anode cavity depended on the substance introduced. For example, Sb yielded 360 K, Cu - 1083 K, Mo - 2620 K, WC - 2857 K, W - 3390 K, NbC - 3500 K, and TaC - 3877 K. Leuchs (1950), used a carbon anode 3 mm in diameter and an arc 1.3 mm in length at 5.5-8 amp, and estimated the following longitudinal temperatures above the anode: 3800 K at 0 mm, 2800 K at 1 mm, and 1900 K at 5 mm. It should also be mentioned that the temperature gradient in the graphite electrode was considered to be smaller than that in a carbon electrode (Mellichamp and Grove, 1978). The above temperature gradients in the anode can now be used to evaluate thermochemical reactions between the matrix elements and graphite or carbon electrodes in a dc arc burning in air at atmospheric pressure.

Reduction of a metal oxide sample placed into a cavity of carbon or graphite electrode can be expressed by the following reaction (Rautschke, 1967; Rautschke and Holdefleiss, 1968; Rautschke et al., 1975a):

k (6) MxOy + yC --~ X M + YCO

Thermodynamically the reaction rate, k, has the form,

AG log kg = log gco = 4.574 T R (7)

where AG is the free energy of formation and T R is the temperature at which the reduction process takes place. Assuming that gco is the total pressure of the system at atmospheric pressure at T R, then the reaction rate assumes the value log Yco = 0,

(8) n29s - rR s298 = o log Kg = 4.574 T R

where AH:98 is the enthalpy at 25 ~ C and AS298 is the entropy at 25 ~ C, from which,


TR __ z~./298/z~S298 (9)

Equations 8 and 9 also represent the reaction rate for the carbide reaction:

M + C --~ MC (10)

When metal carbides are formed in the electrode cavity AG is negative. For various temperatures and reactions, AG can be expressed in the following general form:

r~ rr2 AC~ AG =/'-/298 + ~ AC~, d T - T 5298 -- T j T d T (11)

T l T l

The dependence of AG on temperature for various reactions between SiO 2, TiO 2, WO 3, and MnO and carbon is illustrated in Figure 3. Metal carbide reaction products formed in the anode cavity after various arcing times, can be identified by quantitative X-ray diffraction analysis. This method was extensively used by Nickel (1968), Rautschke ( 1967, 1968), and Bril (1969), who classified the metal carbides formed as a result of thermochemical reactions between metal oxides and carbon in an anode during a dc discharge. Several examples of the reaction products are listed in Table 3.

Reaction rates of metal oxide reduction and those of carbide formation are large; i.e., the reduction to a lower oxide (or metal) state and carbide formation occurs rapidly during arcing. The example of the reactions of a mixture of 1:4.5 V205 and C is illustrated in Figure 4. For this mixture, the reaction rate for the reduction of V205 to V203 is K l = 1 s and for the VC formation from V203, K 2 = 0.23 s (Rautschke et al., 1975a). Rautschke et al. (1975a) also reported the reactions of WO 3 + C and U308 + C as a function of arcing time.

Thermochemical reactions leading to carbide formation can markedly alter the volatilization characteristics of matrix and trace elements. Changes in volatilization rates of several refractory oxides with and without fluorides from the crater of a graphite anode are listed in Table 4. The data show that the addition of fluoride inhibits carbide formation, resulting in higher volatilization rates.

Thermochemical Reactions between Trace s and the Electrode Material (Carbon, Graphite)

X-ray diffraction, which has been successfully applied to detect carbide formation between matrix elements and carbon or graphite, is inadequate for the detection of trace element carbides, owing to the poor detection limits of the technique. Although no direct evidence for the formation of trace carbides can be found in the literature, their occurrence has been identified indirectly. For example, Rautschke (1967, 1968; and Rautschke et al., 1975b) and Nickel (1963, 1968), observed high concentrations of carbides in electrodes when the graphite-to-element ratio was


<]

I 0 0

50

- 5 0

- I 0 0

\

"~e I00

" ' ~ ~ ( o ) so

. . . . . . . . . . . . . . . . . . . . . . . . . . o

', \o ~" \ <!

SiO z + C �9 SiO + C O ' �9 SiO z + 2C �9 Si + 2C0 . . . .

_ SiO 2 + 3C -- SiC + 2C0 SiO + 2C : SiC + C O .. . . . . . . . SiC + SiO �9 2Si + CO ---.a.--. SiO z + 2SIC: 3Si + 2 C 0 - - � 9 I I

I0O0 2 o o o

T, : K

- 5 0

- I 0 0

\ \

- \

" - ~ '\.

- ~ o "".~ \ �9

_ .~

TiO z + 3C : TiC + 2C 0 e , ~

2TiO z + C :T i zO 3 + C O . . . . . . . . Ti02 + 2C : Ti + 2C0 - - ' - -

- - Ti20 3 + C : 2TiO + C O

3TiOz + C : Ti30 5 + C O . . . . TiO z -f C �9 TiO +CO

~ 2.Ti30 s + C : 3Ti20~+ CO ,--- '�9 ,ooo 2000

T, ~

o" <]

IOC

50

- 5 0

- I 0 0

w + c : w c w 0 3 + c �9 w o z + c o - - ' - - 2w03 + 4 c ; w o ~ + w + 4 c o .......

_ \ w 0 3 + 3 C - W " + 3 C 0 - - o - -

\ W03 + 4C:WC + 3 C 0 - - - �9 2 W O 3 + T C , W C + W + 6 C O - e - k.\ ~176176

" . � 9

,~ (c)

_ <1

I 0 0 0 2 0 0 0 T , ~

I 0 0

50

- 5 0

- IOO

3MnO + 4C :Mn3C + 3CO ~ MnO + C :Mn + CO - - " - - . ~ ~ C - 3Mn 4- C --.o----

! . t I 0 0 0 2000

T, ~

Figure 3. Dependance of AG on reaction temperatures between SiO2 (a), TiO2 (b), WO3 (c), and MnO (d) with carbon (after Rautschke, 1967, Figures 2a-d).


Table 3. Examples of Reaction Products Formed by the Thermochemical Reactions between Metal Oxides and Carbon a

Probe SiC:C Reaction Products

TiO2/SiC 1:1:0 TiC, TiSi 2

1:2:0 TiC, TiSi 2, Ti5Si 3

1:3:0 TiC, TiSi 2, TisSi 3

1:4:0 TiC, TiSi 2, TisSi 3

TiO2/SiC/C 5:2:7 TiC

5:3:7 TiC, TisSi 3

5:4:7 TiC, TisSi 3

5:5:7 TiC, TisSi 3

1:1:1 TiC, TisSi 3

1:2:1 TiC, TiSi 2

1:3:1 TiC, TiSi 2

1:1:3 TiC, TiSi 2

1:1:5 TiC, TiSi 2

Ti/SiC 1:10:0 TiSi2, TisSi 3, TiC

TiO2/SiC 1:10:0 TiC, TiSi 2, TisSi 3

TiC/SiC 1:10:0 TiC, TisSi 3, TiSi 2

TisSi3/SiC 1:10:0 TisSi 3

TisSi3/C 1:0:10 TiC, TisSi 3, TiSi 2

ZrOz/SiC 1:1:0 ZrO 2, ZrSi2 1:2:0 ZrO2, ZrSi 2, ZrC

1:3:0 ZrOz, ZrSi 2, ZrC

1:4:0 ZrO z, ZrSi 2, ZrC

ZrO2/SiC/C 1:2:2 ZrO 2, ZrC, ZrSi 2

1:2:4 ZrO 2, ZrC, ZrSi 2

1:2:6 ZrO 2, ZrC, ZrSi 2

Zr/SiC 1:10:0 Zr, ZrC, ZrSi 2

ZrO2/SiC 1:10:0 ZrO 2, ZrC, ZrSi 2

ZrC/SiC 1:10:0 ZrC, ZrSi 2

ZrSi2/SiC 1:10:0 ZrSi 2, ZrC

ZrSi2/C 1:0:10 ZrC, ZrSi 2

V205/SiC 1:2:0 VsSi 3

1:4:0 VsSi 3

1:6:0 VsSi 3

1:8:0 VsSi 3

V2Os/SiC/C 1:2:1 VsSi 3, VC

1:3:1 VsSi 3, VC

1:4:1 VsSi3, VC

3:2:13 VC

3:3:13 VC

3:4:13 VC


Table 3. Continued Probe SiC:C Reaction Products

VC/SiC 1:10:0 VC

Ta205/SiC 5:3:0 Za205, TaC, TaSi 2 5:5:0 Ta20 5, TaC, ZaSi 2 5:7:0 TaC, Ta205, TaSi 2

Ta2Os/SiC/C 1:4:1 TaC, Ta205 1:4:3 TaC, Ta205 1:4:5 TaC, Ta205

Ta/SiC l:10:0 TaC, Ta

Ta2Os/SiC 1:10:0 TaC, Ta20 5 TaC/SiC 1:10:0 TaC TaSi2/SiC 1:10:0 TaC, TaSi 2, TasSi 3 TasSi3/SiC 1:10:0 TaC, TasSi 3, TaSi 2 TaSi2/C 1:0:10 TaC, TaSi 2 TasSi3/C 1:0:10 TaC, TasSi 3

Note: aAfter Rautschke, 1967.

I00

gO

80

70

60

50

40

30

20

I0

Mol %

/ O # % Y l % / C ' f

z%%%% %%%

% %

, ,, , . . . . . ! �9

$ IO

~ O ~

/ ~ A O ~

v~o s .c ,, 1 . 4 s

Ks. tsec "t K 2 ~O,2ase~ t

- V~O s

V20~ ----o--- VC

~ ~ ~ ~ ~ " ~ " ~ ~ ~ ~ , . ~ . m . . , . i f . . . .

15 20 t(sec)

Figure 4. Reaction rate for the reduction of V205 to VC in the presence of carbon (according to Rautschke et al., 1975, Figure 2).


Table 4. Volatilization Rates of U, Th, Zr, La, and Nd Matrices With and Without Addition of Fluoride (PTFE) a'b

Without F With F (4% w/w) Arcing Time (s)

U308 1.4 3.5 60

ThO 2 1.0 2.5 60

ZrO 2 1.5 3.7 60

La203 2.0 3.7 40

Nd203 2.5 4.0 40

Notes: aArc gap 6 ram; arc current 12 A. Qx x 10 -17 atoms s -I.

t~I'he relative standard deviation was 30% for 10 electrodes for each matrix.

high (C/element = 4/1). These results are listed in Table 3. Extrapolation for several trace elements indicates that the respective carbides can be formed with the exception of Co and Ni (Rautschke, 1967; Rautschke and Holdefleiss, 1969).

In general, elements can be classified on the basis of their ability (AG < 0) to form stable carbides at high temperatures (Figure 3). Such a classification was made by Rautschke (1967; Rautschke and Holdefleiss, 1969) and Nickel (1965) who showed that the carbides of B, Si, Ti, V, Zr, Nb, Mo, Hf, Ta, W, the majority of the rare earth elements, and the natural actinides are extremely stable at high temperature.

Formation of trace element carbides is also implied from the detection limits obtained for certain elements in a graphite matrix. Indeed, the detection limits of the stable carbide-forming elements listed in Table 5, are higher by at least one order of magnitude compared to other elements which do not form stable carbides at arcing periods not exceeding 60 s. For arcing periods of up to 180 s (i.e., total sample consumption in the electrode), the detection limits were improved as documented in Table 5, but not without a substantial increase in the background.

Thermochemical Reactions between Impurity and Matrix Elements

The DC Arc. The nature of the thermochemical reactions between trace and matrix elements have not yet been fully elucidated. Available techniques such as X-ray diffraction are inadequate to identify the high temperature end products, and modem techniques such as ESCA and SIMS have not yet been applied widely. Although the evidence for direct thermal reactions between trace elements and their matrix is obscure, indirect evidence emphasizes that volatilization rates of trace elements correlate well with the volatilization rates of their matrix (Avni, 1978), and that volatilization rates for the same trace elements may differ in various matrices.

Volatilization rates for several impurity elements in U308, SiO 2, and A1203 matrices are listed in Table 6. The anode cavity was filled to the same height with


Table 5. Detection Limits in a Graphite-Silicate Matrix a

Carbide-Forming Elements DC-OES Solid mg/kg

Ag 5

Ba 5

Be 1

Cd 20

Co l0

Cr 2

Cs 10

Cu 1

Mo 20

Nb 20

Ni 10

Pb 20

Rb 5

Sb 50

Sc 10

Sn 50

Sr 2 Ta 25

Th 25

U 30

V 5

W 20

Y 25

Zn 20

Zr 10

La 10

Ce 25

Pr 25

Nd 25

Note: aUnpublished data from Brenner, 1995.

the three matrices. The arc gap and current were constant during the arcing period which was the same (60 s) for the three matrices. The values listed in Table 6 were obtained by measuring the difference between the amount introduced into the electrode cavity prior to arcing and the residue after 60 s of arcing. The residual material in the electrode was analyzed in order to determine the content of the matrix element. Flame atomic absorption spectrometry was employed for the determination of the trace elements after fusion and dissolution. Several of the trace elements were preconcentrated and separated from the matrix using ion-exchange proce-


Table 6. Volatilization of Matrix and Trace Elements in a Free Burning Arc in Air a'b

Elements~Matrix U308 (mg/min) SiO 2 (mg/min) Al203 + AlF 3 (mg/min)

U 18,000 - -

Si - - 4800

A1 - - - - 12,000

Mn 120 30 140

Pb 160 30 160

Cr 120 30 120

Cu 120 30 150

Ga 120 40 150

Ge 200 50 130

K 180 40 130

Bi 120 50 180

Mo 70 18 130

V 80 18 120

Ti <0 .3 18 130

Ba 80 25 50

Sr 80 25 50

Notes: aData from Avni, 1978.

bConcentration of each trace element - 500 mg/kg (25 mg each in 50 mg U308; 15 mg each in 30 mg SiO 2 and A1203 + 10% AIF 3) in the anode cup; arc gap 6 mm; current 12 A; arcing period 60 s.

dures. The data show that in general, most impurities volatilized together with their matrices. Relative to the Si matrix, the high volatilization rates of U and A1 are accompanied by high rates of volatilization of their impurities. The results also indicate that Ti, Mo, V, Ba, and Sr form more stable compounds.

Daniel (1960) demonstrated that the vaporization of G a 2 0 3 - U 3 0 8 mixtures occurred at approximately 1650 ~ C, which is significantly lower than the melting points of Ga203 or U30 8. Daniel classified the trace elements according to their relative ease of volatilization as follows:

1. Trace elements readily affected by the volatilization of the carrier, such as B,

Cd, and Sn. 2. Elements which form compounds with the carrier at the eutectic, followed

by volatilization; for example A1, Fe, and Mn. 3. Elements whose compounds are involatile at the anode temperature. They

form stable compounds with matrices such as Ca and Mg. 4. Elements having volatilization rates of the same order of magnitude as that

of the mixture (carrier and matrix) and are not affected by the distillation of the carrier; for example, Zr, Th, and the rare earths.


The Plasma Jet. Chemical reactions and physical interferences between trace elements and the matrix have been reported in dc plasma jets. Serin and Ashton (1964) described the effect of acids on the relative intensities of trace element spectral lines. They found that the addition of 5 N HC1, HNO 3, and H2SO 4 suppressed the spectral line intensities ofMn, Fe, Cr, Mg, V, Ni, and Cd in aqueous solution. Sulfuric acid was most influential in this respect. Rippetoe et al. (1975) investigated the suppression of Ca emission by H3PO 4 in an Ar plasma jet. Golightly and Harris (1975) analyzed geological materials in an Ar plasma jet and found that the spectral line intensities of the trace elements were dependent on the bulk chemical composition of the rock; these matrix effects were reduced by buffering with K and Li. Denton et al. (1975) studied a capillary arc Ar plasma jet and reported that the line intensity due to 1 mg L -t Cd was depressed by the addition of sodium silicate, HC1, and H2SO4, among others. Zilberstein (1977) stated that in plasma jets the analyte spectral line intensities depended greatly on the anion composition of the solution.

These trace element intensity depressions are relative to those obtained in a pure aqueous solution. These effects together with reported enhancements can be attributed to chemical reconstitution processes in the spray chamber taking place between the particles in the droplets prior to injection into the dc plasma as demonstrated by Borowiec et al. (1980) and Skogerboe and Butcher (1985), or by changes in solution viscosity as described by Greenfield (1976) and Dahlquist and Knoll (1978). On injection into the plasma, the droplets are heated, the pressure increases, desolvation occurs, and stable compounds such as chlorides, nitrates, sulfates, phosphates, and silicates can form in the plasma. In the plasma, decomposition depends on plasma temperature and the height where total decomposition takes place. In pure water, chemical reactions occurring between the trace elements and the water droplets are minimum. Consequently more atoms per unit volume of plasma are released and excited, resulting in higher spectral line intensities for the trace elements.

This mechanism in the sprayed droplet was investigated and explained by Alkamade and Voorhuis (1958a,b) who employed two liquid introduction systems coupled in parallel to the same C2H2-air flame. A Ca solution was introduced into the flame through one system while H3PO 4 was aspirated in the other. It was found that Ca line emission was not depressed by PO43. This result is in strong contrast with the findings of many investigators that the depression of Ca by PO 4 is a phenomenon which can occur in the plasma (Larson et al., 1978). The reader is referred to Alkemade and Hermann (1979) for further discussion of the reactions, between the plasma and the aerosol droplets (Gustavsson, 1992; Fister and Olesik, 1991; Olesik and Fister, 1991).

Volatilization Rate (Qx) The number of particles of a chemical element, x, leaving the electrode crater and

introduced into the plasma per unit time is defined as the volatilization rate of the


element (Qx s-I)" For a solution, the volatilization rate of an element will be the number of particles per unit time entering the plasma jet. Methods for evaluation of Qx will be described only for solid samples and not for solutions. To the best of our knowledge, methodology for the determination of Qx (rate of atomization) in solutions has not been reported. Three experimental techniques for determining Qx will be discussed: the chemical (ach), wire (ax), and total consumption ("sample

transport efficiency") techniques.

Chemical Determination of the Volatilization Rate (Q~xh). The chemica l method has been described in detail (Avni and Goldbart, 1973a). The electrode charge was weighed prior to arcing and the residue in the graphite cup chemically and spectrochemically analyzed. Lateral losses of particles due to spattering were taken into account by installing a silica dome (similar to that used in the Stallwood jet) surrounding the arc gap and by placing an aluminum foil below the anode as shown in Figure 5. Both the dome and the aluminum foil were immersed in a suitable acid and the solution analyzed in order to determine the lateral losses of the matrix elements. A minimum period of arcing of 60 s was required in order to detect lateral losses with reasonable accuracy. The amount of matrix deposited on the dome was found to be negligible compared to that found on the foil. Results for

U30 s are shown in Figure 6.

The "Wire Method" (Q~ . In this procedure (Avni and Goldbart, 1973a), aluminum wires traversed the plasma in the vicinity of the anode with uniform velocity. In this way the volatilization rate of the matrix element can be evaluated by determining the amount of material collected on the wires. Deposits on the wire were analyzed by neutron activation (for U and La matrices) and by emission

spectrochemical analysis (for Th, Zr, and Si). The deposit on the wire can be related to the volatilization rate (QxW). For the

matrix x, the particle flux from the anode cup into the plasma can be approximated

if diffusion is ignored, by,

ntx Vx = Qx/Ac (12)

where nix (cm -3) is the total particle concentration in the plasma; V x (cm S -1) is the linear velocity of the particles in the anode region, and A C (cm 2) is the cross section of the anode crater from which particles enter the plasma. When a homogeneous flux of particles, ntx and v x, introduced into a plasma of cylindrical symmetry is considered, the number of particles N w colliding with the wire in a single traverse

is given in Eq. 13,

N w = f(ntx Vx) A(t) dt (13)


Figure 5. Instrumental configuration for the measurement of lateral losses during the arcing period (Q~:h) (reproduced from Avni, 1978).

Direct Current Arcs and Plasma Jets

E 0

p.

W 0

6 -

o o

4 - 0

U ar

r

0 q . 0 N 2

~ , u o 0

r Qw �9

�9 Z ~ O

I I I 40

Figure 6. Volatilization rates of several matrio chemical method, Q w - wire method. �9 U30[ (according to Avni, 1978).

where A( t ) (cm 2) is the area of the wire expose The uniform velocity of the wire in the x axis d exposed to the plasma is 1 (cm). Substitution

71

NCw=I(ntxV x) R

c

for which d w (cm) is the diameter of the wire Integration of Eq. 14 leads to:

Nw = ntx Vx-~-I

By substituting ntx from Eq. 12 and assumin~ N~ and the absolute number of particles N w

obtained from Eq. 15"

Q w = 2 N w 1

c1.7r,(,

For the derivation of Eq. 16, the followir velocity v x in the anode vicinity is equal for att proportionality constant, c~, which is relatec particles, is constant for different wire diamet

87

I I , I ,L 8 0 1 2 0 "

T i m e , s e e

.~s as a function of arcing time. Qch - ;, �9 ThO2, �9 ZrO2, �9 La203, * SiO2

:1 to the plasma for a given time t (s). irection is v w (cm s -~) and the length 9f A( t ) in Eq. 13 gives:

dw. dx (14) 2 v w

and Re is the radius of the cylinder.

r

-WA c (15) w

a proportionality thctor (c~) between ~n the wire, the volatilization rate is

W

(16) W

Lg assumptions w~ire made: (1) the )ms, ions, and molecules; and (2) the

to the cohesion coefficient of the ~rs and velocities.


Figure 7. Carriage and wire configuration for the "wire" method (a), and schematic illustration of the wires for the Q~ measurement (b) (reproduced from Avni, 1978, with permission).

The radial distribution of Qx w can be determined by cutting the wire into several equal segments and the material collected on each segment analyzed. An Abel inversion procedure, similar to that used for measuring radial distributions of spectral line intensities was applied. Calculations were made in terms of particle flux in ntx(r ) v x (cm -2 s-l), which is related to the volatilization rate QW(r) according to Eq. 17:

QW(r)dr = 2 x r nt(r ) vxdr (17)

The wire arrangement in the carriage for the measurement of Qw is shown in Figures 7a and 7b; the wire displacement is schematically represented in the latter. Four wires located in the same wire carriage penetrated the plasma sequentially at the same height (0.2 mm) above the anode. The distances between the four wires were adjusted so that the second wire entered the plasma when the first left it. The velocity of the carriage through the plasma was determined by impact of steel springs. The carriage was released through the arc gap at various times during the arcing period.


Figure Z. (Continued)

The wires were subsequently treated in two ways: (1) wires were totally analyzed to compare QW with Qxh; (2) 2-mm wire segments were used to determine the radial distribution of the particle flux.

In previous publications (Avni and Goldbart, 1973a; Avni, 1978) the "wire" method was shown to be reliable and that QW represents the absolute particle concentration of the matrix element in the anode region derived from the anode crater per unit time. Q~x h and QW values for refractory matrices such as U308, La203, ThO 2, ZrO 2, and SiO 2, as function of arcing time, are represented in Figure 6. Between 20 to 40 s, Qx ~ is almost equal to Q~x h. At a later stage this comparison is no longer valid owing to electrode erosion. For arcing periods less then 20 s only Qx ~ is valid because the determination error in Q~x h was large.

The radial distribution of the particle flux in the anode region for U308 is illustrated in Figure 8. In the anode region (0.2 mm), the radial diffusion of the particles beyond the arc axis varied by 2 orders of magnitude. For example, for U308 and similar matrices, only 5% of the particles entering the plasma moved vertically along the axis toward the cathode, whereas 95% diffused radially when they enter the dc arc (Figure 8). In spectrochemical analysis, the arc core is the most important part of the plasma because it is the brightest. For the determination of QW(r) the wires were cut into 2 mm portions. This enabled the measurement of radial diffusion at 2 mm intervals around the arc axis. The data in Table 7 show that

90 R. AVNi and i.B. BRENNER

0 i l l

?E

0 ~= 100 = ~

r

M

~ - �9

e .u

L .

,p 10 _ �9 �9

o 9~

o

i i I 1 I 1 I I i 1 1 1 0 5 10

1 I i W

D i s t a n c e f r o m a r c a x i s , m m

Figure 8. Radial distribution of uranium particle flux in the anode region. The AI wires are indicated by the following symbols: *, o, v, B, first to the fourth, respectively (according to Avni, 1978).

approximately 30% of the particle flux diffused outward from the arc core (radius 2 mm). These data pertain to the free particles of the matrix.

The relationship between the volatilization rates of trace elements and the refractory matrices was quantitatively evaluated for U308, A1203, and SiO 2. A1, Ba, B i, Cr, Cu, Ga, Ge, In, K, Li, Mn, Mo, Ni, Pb, Sr, Ti, and V were introduced into each ultrapure matrix as oxides to give a 500 mg&g concentration for each element. Each matrix was arced in triplicate for 10, 20, 40, 60 s. After each period of arcing the residual material in the graphite cup was analyzed chemically and spectrochemically for matrix and trace element losses (Avni and Goldbart, 1973a). Differences in contents prior to and after arcing divided by the arcing period is the volatilization rate for the matrix and for each trace element. Results are listed in Tables 8 and 9.

Due to electrode erosion, the material in the anode crater was continuously exposed to the plasma, resulting in the contiguous variation of Qx. Consequently, data in Table 6 were integrated over the given arcing period. Volatilization rates are also presented as matrix-to-trace element ratios. In general, the constant matrix/trace element ratios for various arcing periods indicate that trace element volatilization rates and behaviors were similar to the matrix. However, there are two exceptions: (1) Qx values for A1, Ba, Mo, Sr, and V followed that of the matrix


Table 7. Radial Particle Flux of U308, ThO2, and ZrO 2 on 2-mm Aluminum Wire Portions, in the Anode Region

Radial Distance Particle Flux [nt(r)v j x 10 -15 cm 2 s -l ] Relative Standard from Arc Axis Deviation c (mm) U308 d ThO2 d ZrO2 d (%)

0 - 2 12.0 12.5 11.5 25

2 - 4 3.5 4.0 3.0 25

4 - 6 0.7 0.8 0.5 25


bArc gap 8 mm; current 12 A; wire velocity 110 cm s -~.

CCalculated from four wires.

~ crater charge: 50 mg U308; 60 mg ThO2; 35 mg ZrO 2. The amount of the material in the anode cups was equal for the three matrices.

Table 8. Volatilization Rates, Qj, of Uranium and Trace Elements in a U308 Matrix a'b

Arcing Time (s)

10 20 40 60

Q; ktg s-I Qj/QTr Qj ktg s -1 Qj/QTr Q: ktg s -1 Qj/QTr Qjd ~tg s -I Qj/QTr

U 200 - - 150 - - 200 - - 300

Mn 1.3 150 1.0 150 1.4 143 1.9 157

Pb 1.5 133 1.3 115 1.4 143 2.6 115

Ni 0.3 666 0.2 750 0.5 400 0.05

Cr 1.3 150 1.0 150 1.2 167 2.0 150

Cu 1.5 133 1.2 125 1.3 150 2.0 150

Ga 1.5 133 1.0 150 1.3 150 2.0 150

In 1.2 167 1.0 150 1.5 133 2.0 150

Ge 1.6 125 1.3 115 1.5 133 3.5 85

K 1.5 133 1.3 115 1.5 133 3.0 100

Bi 1.3 150 1.0 150 1.4 143 2.0 150

AI <0 .05 - - <0 .05 - - 0.8 250 0.9 335

Mo < 0.05 - - < 0.05 - - 1.0 200 1.2 250

V < 0.05 - - 0.70 215 1.0 200 1.3 230

Ti < 0.05 - - < 0.05 - - < 0.05 - - < 0.05 - -

Ba < 0.05 - - < 0.05 - - 1.0 200 1.3 230

Sr < 0.05 - - < 0.05 - - 0.8 250 1.3 230


bThe concentration of each trace element was 500 mg/kg (25 lag in 50 mg matrix in the anode cavity); arc gap 6 mm; current 12 A.

CRelative standard deviation for 4 electrodes: 40%.

dRelative standard deviation from 4 electrodes: 25%.


Table 9. Volatilization Rates, Qi of AI or Si and Their Trace Elements in AI203 and SiO2 Matrices a'b

Arcing Time (s)

40 60

Al203 + 10% AlF 3 Al20 3 + 10% AIF, SiO 2 Matrix Matrix Si02 Matrix Matrix

Qf ~tg s -1 Osi/Qrr Qf ~tg s -l Qat/Qrr Qf ~tg s -t Osi/aTr Q~ p.g s -| QAI/QTr

Si 100 D __ D 80 - - - -

AI ~ ~ 100 ~ - - D 200

Mn 0.7 143 1.3 77 0.5 160 2.4 83

Pb 0.7 143 1.5 66 0.5 160 2.7 74

Ni 0.5 200 1.0 100 0.5 160 2.2 91

Cr 0.8 125 1.2 83 0.5 160 2.0 100

Cu 0.8 125 1.3 77 0.5 160 2.5 80

Ga 0.7 143 1.3 77 0.7 115 2.5 80

In 0.8 125 1.3 77 0.5 160 2.4 83

Ge 1.0 100 1.0 100 0.8 100 2.2 91

K 1.0 100 1.0 100 0.7 115 2.2 91

Bi 1.0 100 1.4 71 0.8 100 3.0 67

Mo 0.4 250 1.0 100 0.3 266 2.2 91

V 0.5 200 1.0 100 0.3 266 2.0 100

Ti 0.5 200 1.0 100 0.3 266 2.2 91

Ba 0.4 250 0.5 200 0.4 200 0.8 250

Sr 0.4 250 0.5 200 0.4 200 0.8 250

Notes: aReproduced from Avni, 1978. bThe concentration of each trace element in the anode cavities was 500 mg/kg (15 ktg in 30 mg matrix); arc gap 6 ram; current 10 A.

CRelative standard deviation from 4 electrodes: 30%; Qj, ~tg s -I" Qrr for trace elements.

only after 20 s. (2) the Qx value of Ti differed from that of the matrix even after 60 s of arcing.

This exceptional behavior is ascribed to the formation of new matrix-trace element compounds (Table 6). The new A1, Ba, Mo, Sr, and V compounds volatilized only after 20 s, and Ti after 60 s (the temperature along the electrode increased with increasing arcing time). The behavior of other refractory trace elements--for example, rare earths--was similar to that of titanium carbides (Rautschke, 1967; Rautschke and Holdefleiss, 1969). Addition of 2-4% UF 4 to a U308 matrix produced an increase in the volatilization rate (Avni, 1978) without affecting the temperature of the plasma or the longitudinal temperature of the electrode (Avni and Goldbart, 1973a; Avni, 1978). These data indicate that the presence of fluoride


in the matrix prevented carbide formation, promoting formation of other more volatile compounds (fluorides), and resulted in the simultaneous volatilization of matrix and trace elements.

Sample Transport Efficiency. Boumans and Maessen (1969a,b, 1970) and Maessen (1974) defined the transport efficiency (E) of elements into the arc column,

t

E = I " I Q dt (18) n

0

in which t is the arcing period, n the number of atoms of the given element in the electrode, and Q the volatilization rate (s -1) of free particles from the electrode into the discharge zone.

For the determination of E, Q can be calculated according to,

~g = Q/n or Q =mg (19)

for which n is the particle concentration (cm -3) of a chemical element in the plasma and ~g (cm 3 s -1) is the flow rate, better known as the transport parameter (Roes, 1962; de Galan, 1965; Kirkbright and Snook, 1979). Equation 19 was applied by Boumans and Maessen on the assumption that the radial particle distribution in the center of the arc gap is uniform. This assumption cannot be made in view of the distinct radial gradient of n for various elements in buffered matrices as reported by Decker and McFadden (1975). In a more general case of a nonbuffered matrix, the radial distribution of n is even more pronounced (Avni, 1978). Boumans (1966) defined Qx more accurately, i.e., ~g(z.r) = Qx/n(z,r) ' in which axial and radial gradient were not omitted. Owing to the existence of these radial gradients, Q must be determined in close proximity to the anode cavity.

In order to solve Eq. 19, Boumans and Maessen (1969a,b, 1970) calculated n from the absolute intensity of an atom (I-o0 or from the absolute intensity of an ion ((x). ((x is the degree of ionization of the atom). By applying Eq. 20,

2rr Z exp(Vq/kT) n =--~f gq Aqp hvqp I (20)

in which f = 1/(1 - or) for an atom line, and f = 1/a for an ion line, R is the radius of a uniform cylindrical source, Vq is the excitation energy, Aqp is the transition probability, vqp the frequency of the spectral line, gq the statistical weight of the upper level, Z the partition function, k the Boltzmann constant, h Planck's constant, T the absolute temperature, and I the absolute intensity (erg s -1 cm -2 s -1) of a spectral line.

Boumans and Maessen (1969a,b, 1970) neglected molecules and their dissociation products because many of the elements they investigated do not form stable oxides. However, this is an approximation since many of the trace elements were


present in the synthetic samples in the form of oxides, and as such were injected into the plasma where they decomposed. Thus the degree of dissociation 13, should be taken into account. Moreover, Pearce and Gaydon (1976) identified band spectra for several of the elements investigated; for example, B iO, PbO, SbO, MnO, GdO, YO, etc.

To evaluate Q, ~ values have to be calculated (Eq. 19). This was achieved by experimental correlation of ~ with the degree of ionization, ot (de Galan, 1965).

~r = 300.10 a (2~)

Equation 21 requires elaboration before it can be applied to solve Eq. 19. Equation 21 was obtained by de Galan (1965) from Eq. 19 which in logarithmic form and rearrangement is,

log ~t = log N / t - log ff (22)

for which N is the number of element particles in the graphite cup and t is the time required for complete evaporation of the sample. Thus N / t = Q defines the average evaporation rate of an element, de Galan (1965) succeeded in evaluating xr by applying Eq. 22. Using an arcing period of t = 300 s, given N (the element concentration in the sample), he calculated log n by measuring line intensities of an atom or an ion line. The values of log ~, which de Galan obtained, varied from 2.5 (for elements with ionization potentials from 7.5 to 9 eV) to 3.5 cm3s -l (for elements with ionization potentials from 6 to 7.5 eV), and they were independent of matrix elements (A1203/C = 1:3, LiF/C = 1:4 and KF/C = 1:1). According to de Galan (1965): "Because the transport parameter, log ~, varies between 2.5 and 3.4, and because the relative amounts of elements in the vapor correspond exactly to the sample composition, and as long as the degree of ionization (a) of the elements are alike, we may put log ~ = 2.5 + a".

On the basis of the log ~ value, and taking into account the transit time determined by Russian investigators, Boumans (1966) corrected log ~ to log ~ = 2.0 + Gt, i.e.,

= 300.10 a as in Eq. 21. It has been shown that Eq. 21 can be obtained logically from Eq. 19, given volatilization rates Q. How Boumans and Maessen (1969a,b, 1970) applied Eq. 21 to calculate Q using Eq. 19 is not clear. Nevertheless, they used the proportionality,

W: 10 = (23)

for determining relative ionization in the efficiency expression. Numerical efficiency values were determined (Boumans and Maessen, 1969a,b,

1970) with the aid of racking plate spectra; temperatures and electron pressures were calculated for the successive periods, and the values of n, ~, and Q computed for the elements tested, for each period employing Eqs. 20, 21, and 19, respectively.

Efficiency was determined by approximating the integral in Eq. 18,


i--m

1 8 = -~ E QiAti

i=1

(24)

for which i is the arcing period, A t i is the exposure time, and m is the number of periods.

Several factors related to the spectral sensitivity of the optics and the photoplates, transition probabilities, and the coefficient 300 in Eq. 21 were eliminated so that Boumans and Maessen (1969a,b, 1970) were able to apply relative efficiency which was defined as,

8 ere I - (25)

8ref

where e is defined by Eq. 24, and Cre f is the absolute efficiency for the same element using conditions described in the investigations. The data obtained for the rates of evaporation of various elements in fused samples [SiO 2 + 10% (w/w) CaO + 50% lithium metaborate and trace elements Sb, Mn, Pb, Sn, Ga, W, Bi, Be, Y, Pd, Cd, Er, Zn, Gd, Ag, Zr, and Co] are shown in Figure 9.

B. Plasma Effects (PE)

On evaporation from the anode cup, particles enter the plasma and undergo dissociation, atomization, excitation, ionization, recombination, de-excitation, and molecule formation, according to their spatial position in the plasma. For example, recombination and molecular formation processes can occur in the peripheral portion of the arc column, whereas dissociation, excitation, and ionization occur in the central regions. Mass transfer in a dc arc plasma at atmospheric pressure is influenced by convection (temperature gradients) and electric field strength (space charge gradients). The influence of radiation on mass transport at atmospheric pressure is negligible (Boumans, 1956; Roes, 1962). Prior to describing these phenomena in detail, the axial line intensity distribution will be used to illustrate some of the processes taking place in the dc arc plasma.

Axial Distribution of Line Intensity

The method of measuring the axial spectral line intensity distributions for trace and matrix elements between the anode and the cathode in adc arc plasma was described in a previous publication (Avni, 1968b). The dc arc gap was focused on an 8-aperture diaphragm mounted on the entrance slit of Rank Hilger 3-m Ebert spectrograph (Figure 10). For a given spectral line, the relative intensity at a given position in the arc gap was normalized to the relative intensity of the same spectral line in the anode region of the plasma. This procedure minimized systematic and random variability in line-intensity measurements. Thus intensity ratios obtained

Figure 9. Evaporation patterns for the various test elements derived from racking-plate spectra taken when a fused sample was arced under different conditions. �9 Intensity of spectral line corrected for background; [] rate of evaporation of element (i.e., number of free particles that enter the discharge region per second). Block ordinates represent exposure times of separate racking-plate spectra. Block abscissae represent intensities or evaporation rates. The 2-s racking interval has not been drawn to scale The normalization underlying the diagrams conceals the differences in scale for the three evaporation modes. This information is therefore brought out separately by scale marks at the bottom of the diagram. The length of each mark links up with the factor that the corresponding intensities (or rates of entry) must be multiplied with to obtain comparable scales for the three evaporation modes. (a) Slow evaporation mode; (b) medium evaporation mode; (c) fast evaporation mode (after Boumans and Maessen, 1969b).


represent ratios of particle concentration and temperatures at two levels (z) in the plasma. These ratios can be explained in the following way:

Let 1 and 2 be two different axial positions in the plasma, observed by apertures 1 and 2 in the diaphragm in Figure 10 a and b. At these levels (z), equal volumes of emission were transmitted to the spectrograph. At these positions, the absolute line intensity J for the same atom transition is described by,

m t X ~ m 2 2

~!l" 215 ,500 . . . . . - r

14 $ a

d J " S

W . S II al

b Figure 10. Optical conf igurat ion for the Ebert 3.4. m spectrograph. (a) Electrodes focused on the entrance slit; a - electrodes. Focal length of lenses: L1 = 1 O0 mm, L2 = 200 mm; L3 = 150 mm, L4 = 450 mm; m - mirror, ml and m2 - mirrors (on different planes for radial measurements), d - diaphragm, s - entrance slit. (b) Cathode region focused on the entrance slit; s - slit, c - cathode projection, w - wedge.


hv 1 glA1 Jl - 4x Z hi(1 - t~)l exp[-E 1/kT]

(26)

hv2 g2A2 n2(1 - ~ ) 2 exp[-Ez/kT] (27) z

for which h is Planck's constant, n the line frequency, g is the statistical weight of the upper term, A the transition probability, Z the partition function, n the particle concentration (atoms + ions), ~t the degree of ionization, k the Boltzmann constant, and T the absolute temperature. Therefore, the line intensity ratio can be expressed as;

J l _ nl(1 -Gt)l exp T2 - T1 (28)

"]2 n2(1 - ~)2 T1 T2

Equations 26 and 27 assume local thermodynamic equilibrium (LTE) at each level, and that only atoms and ions populate the dc plasma. The presence of molecules are neglected. Axial distributions of normalized relative line intensities of several trace and matrix elements in U308, ThO 2, ZrO 2, rare earth oxides, SiO 2, and Ca3(PO4) 2 matrices are shown in Figures 11-14. From these figures the following conclusions can be derived for trace element behavior:

1. Three distinct plasma regions were observed--anode, central, and cathode. 2. The axial spectral line intensity distributions of the trace elements in the

cathode region exceeded their intensities in the anode region. The maximum line intensities for matrix elements were located in the anode region and were minimum in the cathode region.

3. In the central region the normalized line intensities of the trace elements always exceeded that of the matrix element.

4. The normalized line-intensity distributions of the trace elements were not similar to those of the matrix elements, i.e., in the plasma, trace elements were "separated" from the matrix elements.

The volatilization rates of matrix and trace elements are listed in Tables 6, 8, and 9. In general, the volatilization behavior of the trace elements is similar to that of the matrix element. A comparison of these results, with those shown in Figures 11-14, emphasizes the conclusions concerning "spatial separation" between trace and matrix elements in the plasma. In order to evaluate "spatial separation" processes taking place in the dc arc, plasma temperature, degree of ionization, and free particle concentration, expressed in Eq. 28, were measured.

Voltage and Electric Fields In practice, the voltage-current relation in a dc arc does not obey Ohm's Law

(Smit, 1951; Boumans, 1956; Finkelenburg and Maecker, 1956). For a given


/ /

.C

Ih

1 l II

J

AmMi, D l s i a m : ,

,1 . 1 , L

Figure 11. Axial distribution of relative line intensity normalized to the anode region. U308, ThO2 and ZrO2 matrices; 6.0 mm arc gap, 13 A. [-] Line intensity for "common" trace elements in U308 ( ~ ) , ThO2 ( - - . - - ) , and ZrO2 ( . . . . ), REE traces in U308 (without markers ...... ) Matrix element line intensities" �9 U, [] Th, �9 Zr (reproduced from Avni, 1978).

distance between the electrodes, an increase in the current is accompanied by a voltage decrease. Smit (1951), Finkelburg and Maecker (1956), Boumans (1966) and Ecker (1961) described the voltage-current relations in a dc plasma.

The electric current, i, in the arc is determined by two factors: (1) the electric field created by the voltage difference between the electrodes; and (2) the conductivity caused by free electrons and ions due to the thermal excitation and ionization. The difference in free electron and ion concentrations determines the space charge. With only a few doubly charged ions in the arc column, the concentration of ions is practically equal to that of the liberated electrons; i.e., the space charge in the arc column is low with respect to the total charge of the free electrons (Smit, 1951; Ecker, 1961). The resultant space charge in the arc creates a field that promotes discharge. The electric field in the arc consists of the following components

(neglecting microfield strength): 1. The longitudinal field, frequently termed the potential gradient, E z.

In the carbon arc column this field is assumed to be homogeneous since the column is almost cylindrical and the relation between voltage and arc length at constant current is practically linear (Eqs. 18-20). E z depends on the arc gap


8.0

e.OJ &

.11 0 G

- ~ 4.0 l h

e

- 2 0

e N .,m

l "0.8 e

2:

OA - --A 2 4 6

Anode Cathode Distance m m

Figure 12. Axial distribution of relative line intensities normalized to the anode region. REE oxide matrices. 6.0 mm arc gap, 13 A. -D- Normalized intensity for "common" trace elements; __ .m normalized intensity for As, Hg, P, Zn and Cd; �9 normalized intensity of La, Y and Sm; I normalized intensity for Ce, Dy, Ey, Gd, Ho, Nd, and Yb (reproduced from Avni, 1978).

(increases with increasing gap length (Smit, 1951; Avni, 1968a) and on the ionization potential of the gases (Smit, 1951; Boumans, 1966).

2. The radial field, E r, is primarily caused by the small positive space charge within the arc core.

Electrons and ions are separated in this field despite their tendency to mix by diffusion. In an arc plasma, E r never exceeds E z (Smit, 1951; Finkelenburg and Maecker, 1956). According to the theory of gas discharge in tubes (Smit, 1951), E r can be expressed as,

k T L (29) E r = - ~ d n e / d r

e n e

where n e is the electron density (era-3), r the distance from the core axis, and L the length of the arc gap. E r is assumed to be maximum at the margins of the arc core because in this location d n e / d r is maximum (Mandelshtam, 1962).

E z a n d E r can be emulated theoretically. For cylindrical symmetry, the solution of the time-dependent Laplace equation, vZv = 0 (Avni, 1978), indicates that radial and axial electric fields can exist in a given arc gap. The experimental method for measuring the longitudinal electric field strength (Ez ) has been described elsewhere

102 R. AVNI and i.B. BRENNER

6.0

e40 "u o E

_- 2O

" u Q . 0.8

o , n

o E �9 . 06 o

Z

0.4

/t /

_

! I I I, I ! ~ 2.0 4.0 6.0

Anode Cathode Distance mm

Figure 13. Axial distribution of relative line intensities normalized to the anode region - SiO2 matrix; 6.0 mm arc gap, 13 A. D Normalized intensity of "common" trace elements; o normalized intensity of Si spectral lines (reproduced from Avni, 1978).

(Boumans, 1966; de Galan, 1965). The disadvantage of the "V~ + V c + E z x L" method (Boumans, 1966; Finkelenburg and Maecker, 1956; Ecker, 1961) (where V a is the voltage on the anode, V C on the cathode, and L is the arc gap) is the variability of the arc gap L during measurement. E z is assumed to be constant for a given arc gap and varies only by varying L. However, this assumption contradicts the theory underlying the Laplace equation (Avni, 1978). Methods for the measurement of E r in a dc arc plasma have not been described in the literature' We have measured E z a n d E r by the so-called "wire" method described in Section II.A, which entails passing a probe through the plasma (Avni, 1968a, 1978). Axial and radial voltages in the plasma, measured by wire intrusion, showed that the voltage values are not influenced by the material (element or alloy), by the diameter (cross section) of the wire, or by the velocity of wire passage through the plasma. The values obtained in the vicinity of the anode and the cathode with this "wire method" were identical to those obtained employing the " V a + V c + E z x L " method.

Voltages measured with an oscilloscope employing the "wire" method are illustrated in Figure 15. The variation of axial and radial electric field strengths when U30 8, ThO 2 and ZrO 2 were introduced in a graphite-air plasma, for a 6-mm arc gap, is indicated in Figures 16 and 17. The change in E z when trace elements are added to these matrices is illustrated in Figure 18. These figures show that the addition of the matrix to a graphite-air plasma resulted in an increase in both E z


I/I ANODE 14.0~

6.0_

5.0_

o / / /-

/ /

oj /

4.C_ /,, / �9 I o II ! t I lE

3.0_ ,, / . 2 " I / ~ l

2.0_ _.~.-.~.-.-~o ,~ ~. I o o

O.

Anode 1 2 3 4 Cathode Distance m m

Figure 14. Axial distribution of relative line intensities normalized to the anode region of natural and synthetic phosphates; 4.0 mm arc gap, 10 A. 0 Mg II, A Mn II, [] Ba II, ~ normalized trace element atom line intensities in a synthetic Ca(PO3)2 matrix, - ..... Normalized trace element intensities in natural rock phosphate.

Figure 15. Oscillographs for the measurement of voltage by the "wire" method. (Left) anode region; (right) cathode region. Abscissa - 5 m s/cm. Ordinate- 10 v/cm (reproduced from Avni, 1978).


o Cn O

. m

40

30

20

[ I

I

f ,F/ / i ]

J i I w S " I 'l ~'

2 ' 4 6 "8 Cafhode Anode Anode Anode

Distance mm

Figure 16. Axial distribution of voltage in a graphite-air dc arc plasma obtained with the "wire" method. � 9 graphite plasma; � 9 U308 with various arc gaps (reproduced from Avni, 1978).

E

.c 6

~ 4

: ' 3 I,P

>

-u o i 1 ,, I I ! I 1 1 - w

2 4 6 8 Anode C a t h o d e Dis lance m m

Figure 17. Radial distributions of electric field strength in plasmas containing U308, ThO2, and ZrO2 employing the "wire method". Radius is 1-7 mm. Arc gaps 4 mm ]~; 6 mm I; 8 mm~ (reproduced from Avni, 1978).


25 ~ ~ z 4 - . . ; / , ; ; Y

. .%=s. :2=,%.-"- / . /

/

13~" L ~ ~ ' ' ~ 6 " 1 2 3 4 5

C a t h o d e Anode D i s t a n c e m m

Figure 18. Axial distribution of voltage as a function of trace element concentration, using the "wire" method, o, [], zx - 100, 500, and 1000 mg/kg "common" trace elements, respectively, in a U308 matrix and e, N, A, respectively, in ThO2. - - - 500 mg/kg common trace elements in graphite (reproduced from Avni, 1978).

and E r values (Figures 16 and 17), while the addition of trace elements and the third

matrix resulted in a decrease of the E z and E r values of the plasma (Figure 18). Several conclusions can be made on the basis of the correlation between electric field strength (E), current density (j), and electrical conductivity (~):

E = j / ~ (30)

1. The addition of a third matrix to a graphite-air plasma resulted in an increase in the electric field and consequently an increase in the current density (j) to electrical conductivity (a) ratio. For the entire arc column, j becomes the current strength, i, which is normally constant in the dc arc (Boumans, 1966).

For a constant i value, the addition of the third matrix caused an increase in E and a decrease in ~.

2. The addition of trace elements in the third matrix, to a graphite-air plasma resulted in a decrease of E z and the i / a ratio. In other words the amount of ions and electrons in the plasma derived from the trace elements exceeded

those produced by the matrices.

Temperature (T) Plasma temperatures exceeding 4000 K can be measured using atom or ion

spectral line intensities. For elaboration of the principles underlying the procedure,


the reader is referred to Smit ( 1951 ), Boumans (1966), Lochte-Holtgreven (1955), Orenstein and Brinkman (1934), Pearce (1961), and others (Alkemade, 1956; Finkelenburg, 1956; Roes, 1962; Hollander, 1964; Gaydon and Wolfhard, 1970). In this section we shall deal with the temperature and its gradient in the plasma with and without the third matrix. The optical configuration used for axial and radial temperature measurements employing line intensities is illustrated in Figure 10.

For the determination of the radial temperature distribution, an Abel inversion was used. The line-intensity distribution obtained from a photographic image represents intensities integrated along the line of observation and thus cannot be used for evaluating radial distribution. The solution of the Abel inversion integral takes into account the radial distribution. This integral is represented by the equation,

L I'(x)dx r (31) J (x 2 r2) '/2 J(r) = Tr.

where J is the emission per cm 3 per step, l(x) the measured intensity at a distance x from the center of the circle measured at fight angles to the optic axis (Boumans, 1966; de Galan, 1965), l ' ( x ) is the derivative of l(x) with respect to x, and r is the radius of the dc plasma. Equation 31 has a unique solution but since l(x) is obtained as a set of numerical data, a numerical method was applied (Decker and McFadden, 1975). Empirical relationships involving the Abel transform were summarized by Bracewell (1956), Boumans (1956), Pearce (1958), Nester and Olsen (1960) and Lochte-Holtgreven (1955).

Decker and McFadden (1975) compared the radial distributions obtained for Zn (307.6 nm) using the methods described by Pearce (1958) and by Nester and Olsen (1960). This comparison is given in Figure 19. Kulagin et al. (1972) also compared the method of Nester and Olsen with that of Pearce and showed that the former is more accurate in the vicinity of the plasma axis.

The atom lines of Zn (307.2, 307.6, and 328.2 nm), Cu (510.5 and 521.8 nm), and Mn (262.404 and 279.984 nm) can be used as the thermometric species. However, it should be mentioned that Mn lines are restricted to temperatures up to 6000 K (Boumans, 1966, 1968, 1971). Line pairs have been used widely for the so-called "two-line ratio" method for measuring excitation temperatures in free- burning arcs. The "one-line" method is commonly employed to measure temperatures in plasma jets. For example, Krantz (1964, 1968) and Baronnet (1971) measured temperatures employing N~ band spectra and atom lines of Ba, Ca, Co, Cu, Fe, and Na, and W. Elliot (1971) used the "one-line" method with Ti, V, Cd, and Fe atom lines. Merchant and Veillon (1974) used Ar lines and Rippetoe et al. (1975) used atom lines of Fe and Ar spectral lines. Skogerboe et al. (1976) employed Na line self-reversal as well as Ar lines.


I .Z ~ .

\ . \

I ' 0 . " \ /

0 . 8

.+-_ c

o -,, 0 . 6

0 I 2 3 �9 4

r, n'rn

Figure 19. Comparison of the radial intensity distribution obtained by solving the Abel inversion integral by the Nestor and Olsen and Pearce methods. - - . . . . measured, ~ Nestor and Olsen method, - . . . . Pearce method (according to Decker and McFadden, 1975).

A comparison of the temperatures measured for different types of plasma jets is given in Table 10. The temperatures were measured in the analytical region above the orifice (Butler et al., 1974). The temperature profiles in the Krantz plasma jet, the commercial Spectra Jet II, and in "two-jet" plasmatrons are illustrated in Figure 20. A nearly linear temperature decay with distance beyond the orifice was observed. When using the "one-line" method, curvature of the function may be caused by departure of the Boltzmann distribution, by self-absorption of several spectral lines, and/or by spatial inhomogeneity of the light source. These factors require careful consideration.

In the free-burning arc in which the third matrix was introduced together with trace elements, significant temperature differences were not observed up to a concentration of 500 mg&g for each trace element (B, Si, Ag, Mn, Pb, In, Ni, Mo, Ca, Sr, Ba, Bi, Co, Ge, Li, Na, Ti, and V). Axial temperature distributions measured for various plasma matrices, using 6- and 10-mm arc gaps are shown in Figures


Table 10. Temperatures of Various Plasma Jets

Plasma Jet Type Gas flow

Temp. K Current A Power kW L/min Refs.

Kranz laminar

Kranz turbulent

Plasmatrons

Spectra Jet II

Spectra Jet I

Two-Jet plasmatrons

Bent arcs

Seeded arcs

Rotating arcs

Double-flow

5400 15 4-8

5000 15 4-8

11000 15

2300 a

5600 b 7.5 0.3

6140 0.2

13000 100

5700 9

3900 11 0.8

5000 11 0.33

6400 10

5.7

5.7

>> 4

3

5

>> 50

3

12.5

7

6

114

114

71a

l17a

115

71b

116

91

116

97

Notes: aUsing Na spectral line reversal method. bUsing Ar lines.

Figure 20. Temperature profiles of a Kranz-type plasma jet. a - Laminar; b- turbulent (compiled from Figure 10 in Keirs and Vickers, 1977, and from Figure 3 in Zheenbaer, 1976).


21-26. Radial temperature distributions in plasmas with and without a refractory matrix for the same arc gaps are illustrated in Figures 27-31.

Axial Temperature Distribution. 21-26, demonstrate the following:

The axial distributions shown in Figures

1. Three distinct arc regions (anode, central, and cathode), independent of the presence of matrix particles, occur in the plasma.

2. The maximum temperature of the graphite-air and the de Galan plasmas (de Galan, 1965) were observed in the anode region. On addition of refractory matrices, these maxima deviated to the cathode region.

3. Beyond the anode region, higher temperatures were observed only in the presence of third matrix particles.

In a dc arc plasma containing refractory matrix particles, the cathode region is characterized by maximum temperatures. This is in accordance with the maximum line intensities of the trace elements in this region (Figures 11 to 14). However, matrix element line intensities in this high temperature region are minimum!

750O

7OOO

6 5 0 0

T~ I"

I I ! I ! ] - 1 3 5

Anode Cathode

l : ~ , ta~ �9 m m

Figure 21. Axial temperature distributions in dc arc plasmas containing ThO2 (A) and ZrO2 (o) matrices; 6 mm gap (reproduced from Avni, 1978).


T~ 10 2

7 0

60

/ j i L / j /

"1" I l ,I I I . I 2 4 6 Distance mm

Anode Cathode

Figure 22. Axial temperature distribution in plasmas containing La203, Sm203, and Y203 (II), and other REE oxide matrices (1); gap 6 mm (reproduced from Avni, 1978).

~ T "K

e4oo~ \\

62o \ \ \ \ \ . //, ~

6 0 0

5800~t 1 " ~ ~ 1 ! I 1 Anode 2 4

Distance m m C'o, e

Figure 23. Axial temperature distributions in a graphite-air (,) plasma; and plasmas containing AI203 (e) and MgO (A) matrices; 6 mm gap (reproduced from Avni, 1978).

6400

T ~

. . % .

_ / /

/ /

t

I ~ ~ 11 J

6000 ~ - " " " "

- ~ , ~ .

~ - . & . . . . . . . _&

5600 L ! I , i J J ~ 2D 4.0 6.0 A n o d e C a t h o d e

D i s t a n c e m m

Figure 24. Axial temperature distribution in a dc arc plasma containing SiO2 (o) SiO2 + 5% each of Ca, Mg, AI, and Fe (a); SiO2 + 5% Na + 5% K (A); SiO2 + 12% (Na + K) + 500% graphite (u); SiO2 + 12% (Na + K) + 300% C (*); arc gap 6 mm (reproduced from Avni et al., 19XX).

T O K

6 4 0 0 -

6 2 0 0 '

6000 o

:1" 1 I I , i _ A n o d e 1 2 3 4 C a t h o d e

D i s t a n c e . m m

Figure 25. Axial temperature distribution in dc arc plasmas containing Ca3(PO4)2 (&); natural rock phosphate (E3); rock phosphate and synthetic Ca3(PO4)2 + 20% graphite (0); arc gap 4 mm (reproduced from Avni, 1978).

111


CATHODE 10

~ ala_

0 5000 5500 sooo s5oo ANODE T(OK)

Figure 26. Axial temperature distribution �9 AI203/C 1:3; [] LiF/C 1:4; �9 KF/C 1:1 (after de Galan, 1965).

Radial Temperature Distribution. The radial temperatures distributions illustrated in Figures 27-31 show that:

1. In comparison to a pure graphite-air plasma (Figure 30), the addition of LiF, KF (Figure 27), and CaF 2 (Figure 28), resulted in a reduction of the temperature, whereas the presence of refractory matrices such as A1203, U308, ZrO 2, and ThO 2, caused increases in the temperature.

2. In plasmas containing refractory matrices (Figures 29-31), the radial temperature gradient was similar in all plasma regions.

3. In a graphite-air plasma the radial decrease was insignificant--up to 2 mm from the axis, as opposed to the abrupt decrease in this region when a third refractory matrix was added.

In order to comprehend this radial temperature distribution in these plasmas, the energy balance equation (Eq. 1) should be recalled. This equation can be used to calculate the radial temperature gradient, dT/dr. In an ideal arc where the symmetry axis of the electrodes coincides with that of the plasma, Eq. 1 for cylindrical (z, r) symmetry takes the following form:


Pel(a t rn) o A 1 a

I . L i

2 K

Io -3

&

3 - ~ r(rnm)

! I I I 0 2 4

T( ) 6500

6 000

5500

5000

4 5 0 0

o A I ~ L i

0

~ r(mm) t

i ! ! i 0 2 4

b

Figure 27. Radial distribution of electron concentration (a) and the temperature (b) in a dc arc. The arrow is the standard error, o AI203/C 1:3. [] - LiF/C 1:4; �9 KF/C 1 "1 (after de Galan, 1965).

8 0 0 C

E 60oo

E

30OW ~ ) o W

- t - h w l w l I I ! 0 I 2 3 O I 2 3

m m

5OOW

I I I I_ 0 I 2 3

Figure 28. Radial arc temperature distributions obtained from samples buffered with 50% CaF2, at different arc powers (after Decker and McFadden, 1975).


7oooT_T ~ K

6000

F _ _1 i, i _ i 0 - - 10 2D

I" f i b r i l

]A ,,v

Figure 29. Radial temperature distribution in the anode region of dc plasmas containing U308 (e 4 mm, �9 6 mm, and & 8 mm gaps); ZrO2 (- - - 6 mm gap); ThO2 (-. - .- 6 mm gap) (reproduced from Avni, 1978).

7 0 0 '~ T'ol~ K

5 0 0 0

! I , 1 1 I . L 1 2 "

I" I~1 Im

Figure 30. Radial temperature distribution in the central plasma region. U308 (e 4 ram, m6 mm, and �9 8 mm gaps); Zr02 ( . . . . 6 mm gap); Th02 ( ~ . ~ ) ; and graphite-air (~- 6 mm gap) (reproduced from Avni, 1978).


I~K

~ \

7

0 1.0 2D r ' I N I N

Figure 31. Radial temperature distribution in the cathode region of a dc arc plasma. U308 (o 4 mm, �9 6 mm, and �9 8 mm gaps); ZrO2 ( . . . . 6 mm gap); ThO2 (-.-.-); 6 mm gap (reproduced from Avni, 1978).

18 j x E = ~ (rp grad T)

r c3r

(32)

Integrating and using dT /dr = 0 as a boundary condition along the z axis, Eq. 32 gives,

dT fr (33) - rp -d--r-r = J J E r

&

0

indicating that the electric energy dissipation in a cylinder of unit height (dz = 1) is equal to the heat dissipated from this cylinder per second. For a set of concentric cylinders of radius r = 0 to r = R and unit height, it has been shown (Avni and Klein, 1973a) that for mean values ofj( j ) and E(E), Eq. 33 with,

R

I f j=~:R2 j r d r , 0

takes the form:

1 (34) m m

j E = Rp d T / d R rcR 2

Using Eq. 30 and substituting j, Eq. 34 can be written for the radial temperature gradient as:


Sdrr =E2 "R p ~r = ~ (35)

in which S = 1/x R 2 The calculation of the mean electrical conductivity, accounting for the mean

electric field strength E, and electron density, was described in a previous publication (Avni and Klein, 1973a). The thermal conductivity for N 2 was determined by Roes (1962) and Avni and Klein (1973a). With the values of o, E, and 9, the temperature gradient, d T / d r was measured at a given height in the plasma according to the Elenbaas-Heller equation (Eq. 35). The radial temperature gradient, and the thermal and electrical conductivity in the central region of the plasma with and without the third matrix are listed in Table 11. As shown in these tables, the calculated d T / d r (Eq. 35) and the measured T / R ("two-line method") values essentially coincide. Moreover, the values of o, 9, and d T / d r for plasmas with and without the third matrix show that:

1. The Elenbaas-Heller equation (Eq. 35) requires only steady-state conditions in the plasma, d T / d r values were equal to those of T / R in the same region. Consequently, the excitation temperatures ("two-line") represented the temperature in the given region, i.e. LTE exists in dc arcs with and without the refractory matrix (see Section II.B).

2. The mean electrical conductivity, s, for a graphite-air plasma had approximately one-half the value of a plasma containing a U308 matrix plus trace elements.

Electron Density (n~) The electron density (ne) in a dc plasma is usually calculated employing the "Saha

method" (Huldt, 1948; Alkemade, 1956; Mandelshtam, 1962; Roes, 1962: Bou- mans, 1971). Since this method assumes LTE in the plasma, an ion-atom spectral line intensity ratio of a given chemical element (trace or matrix) can be used for the electron density calculation. As for the temperature measurement, established transition probabilities and partition function values are prerequisite. An additional requirement is that self-absorption of the spectral lines is absent. Magnesium, manganese, and chromium spectral lines (Finkelenburg and Maecker, 1956; de Galan, 1965; Avni, 1968) were used mainly for the n e determination. The electron density was obtained from line-intensity measurements using Saha's constant and Eqs. 36-38,

[(2KTme) 3/2] 2Z i �9 ~ exp[-V l/kT] (36) K l = neni/n a = h3 Za

where indices i, a, and e refer to ion, atom, and electron, respectively, KI is Saha's constant for the ionization process, ni /n a is the relative ion and atom densities, k is

Table 11. Spatial Distribution of Temperature, Electron Density, and Electrical Conductivity (c) in Graphite-Air and U Plasmas a

Plasma Region

Measured Radial Electric Measured Measured Measured Calculated

Distance Strength b T Zn Lines T/R x 10 -3 n e x 10 -16 pC (R, cm) (E, V cm -!) (deg K) (deg K cm -I) (Saha, cm -3) (Watt cm -I deg -I)

Calculated Electron

Mobility Value Calculated Calculated Calculated x 10 -4 ~ oh m -! fie • 10-16 dT/dr x 10 -3

(cm 2 v -I s -j ) cm -] cm -3 deg K cm -I

Center

Error

Without U308 Matrix

0-0.05 6200 124 0.2 0.022

0.05-0.1 20 6000 120 0.08 0.017

0.1-0.15 5700 114 0.03 0.011

+/-20% +/-200 +/-25% +/ - 15%

With U308 Matrix

Anode 0-0.05 6400 128 0.4 0.026

0.05-0.1 90 6100 122 0.25 0.02

0.1-0.15 5700 114 0.16 0.011

Center 0-0.05 6700 134 I 0.038

0.05-0.1 20 6500 138 0.5 0.031

0.1-0.15 6100 130 0.4 0.024

Cathode 0-0.05 7200 144 4 0.05

0.05-0.1 10 6900 138 1.6 0.045

0.1-0.05 6500 ! 30 1.2 0.03 I

Error +/-20% +/-200 +/-25 % +/- 15 %

2.85 34 0.7 97

2.75 !1 0.15 121

2.5 4.1 0.07 116

+/-15% +/ - 15% +/-20% +/- 20%

3 3 0.07 139

2.8 0.7 0.02 134

2.5 0.2 0.005 120

3.6 80 1.5 127

3 22 0.5 133

2.9 10 0.3 134

4 460 7.2 138

3.7 130 2.2 136

3 50 1 129

+/-15% +/ - 15% +/- 20% +/-20%

Notes: aCalculated from the energy balance Eq. 26; arc gap 6 mm, current 10 A. = E 2" b E 2 E~ + ~ Measured using the "wire method" (Avni and Klein, 1973). The values are for a cylinder of R = 2 mm.

':Calculated values after (Roes, 1962) for N 2.

'JCalculated from the energy balance equation, using T/R values (Eq. 35).


the Boltzmann constant, h the Planck constant, m e the electron mass, T the absolute temperature, Z a and Z i the partition functions of the atom and ion, respectively, and V i is the ionization potential of the element involved (eV); The hi~ha ratio is obtained from the spectral line intensity ratio Ji/Ja as given by the following relation,

( v g A / Z ) a (37) ni /n a = JI /'12 exp[E i - E a /kT]

(vgA /Z ) i

in which v is the line frequency, g the statistical weight, A the transition probability, and E the excitation potential. With the values of K~ and n / n a the electron density can be calculated by:

ne = Ki(na/ni ) (38)

The optical arrangement for axial and radial line-intensity measurements was similar to that used for the temperature measurement (see Section II.B). The experimental data used for the temperature measurement were also used for the calculation of n e. The following spectral line ratios were employed: Mn II 257.6/ Mn 1 279.4; Mg II 279.5/Mg I 285.2; Cr II 283.5/Cr I 302.1; and U II 434.7/U I 424.6. The matrix in the anode crater contained 200 mg/kg of each trace element. The axial distribution of electron density for 6-mm arc plasmas with and without various refractory matrices is illustrated in Figures 32-36.

A x i a l Dis t r ibu t ion o f he. The following features of distribution of flee-electron density are depicted in Figures 32-36:

1. Distinct anode, central, and cathode regions are formed in plasmas containing refractory matrix particles. These regions are poorly defined in a graphite- air plasma and in plasmas containing Na, K, or Li (Figures 35 and 36).

2. In a graphite-air plasma, n e values were maximum in the anode region. M a x i m u m B e values occurred in the cathode region when a refractory matrix was added. When Na, K, or Li matrices were added to the plasma, maximum n e values were still observed in the anode region (Figures 35 and 36).

3. With the exception of MgO, electron density values of plasmas containing a third matrix, exceeded those of graphite-air plasmas, i.e., neiTM ) > 1l e (where T M refers to the third matrix). The fact thatne(rM ) > he, and that the maximum value was located in the cathode region requires elaboration.

According to Saha's equation, higher electron densities imply higher n a / n i ratios and/or high values for the ionization constant K i (Eq. 36). The dependence of log K i with temperature is shown in Figure 37. For uranium, an increase of 1000 K from 6000 to 7000 K, resulted in an increase of K i values by 1 order of magnitude. Based on plasma neutrality, ne(TM ) > n e cannot be explained by an increase in the n~/n i ratio, implying that a lower value of n i requires that n i ' ~ n e. Consequently if n e

1=;30 C

o �9 " 5.0 ";'E

tul

- , - - - - - - : - z . . - % x ,

"=1

w

~ 0.6[- 1 1 l I I , Jl

2.0 40 6.0 Cathode Anode

Distance Figure 32. Axial distribution of electron density (ne) in plasmas containing U308 (0), ZrO2 (A), and ThO2 (m). Mean electron density values (ne) for plasmas containing U308 (,); Arc gap 6 mm (reproduced from Avni, 1978).

8 'f .L.I | I

I 1 I ~1 ~ ~ 1 ~ . I I I �9 ', i - . - ~ , ,

, ,~ w I I \ I

, ,r I 1

[ o

i u J

I J | I , I . I 2.0 4.0 6.0

Cathode Anode Distance mm

Figure 33. Axial electron density (ne) distribution in plasmas containing REE oxides. La203, Y203 and Sm203 ( .... ); other REE oxides (--); arc gap 6 mm (reproduced from Avni, 1978).

119


t,l

16 10 "*

=,,, 1015 _ =D IIQ r @

"U

= 5 L

@ 0

1014 Anode

J

m \

\ \ J

f \ / \ /

A A

T

l l L [ I ,I -, 2 4 6 " Cafhode D i 'stance m m

Figure 34. Axial electron density (ne) distribution in a graphite-air plasma ( . . . . ); with A I203 (e); M g O (zx); arc gap 6 mm (reproduced from Avni, 1978).

increases, n i should increase as well. Thus the correlation of n e and K i values was due to the temperature increase of plasmas containing a third matrix.

The maximum electron density values observed in the cathode region of a plasma containing a third matrix as indicated by n e = n i, implies that the maximum ion concentration also occurred in this region. Table 12 lists the normalized intensities (Jcathode[Janode) for several trace element atom and ion spectral lines. The normalized intensities of ion lines were approximately twofold higher than their respective atom lines in the cathode region; thus n i in the cathode region was higher by a factor of 2 (Eq. 27).

Radial Distribution of ne. If the measured values of T/R ("two-lines method") are introduced into the energy balance equation instead of d T / d r , Eq. 35 can be used for the determination of the mean electrical conductivity, a, as described in a previous publication (Avni and Klein, 1973a). From the s values the mean electron


('1 'E ~t

T -i-

.~10 m a -... - .~

"0 \ \ / ~) %% /

10-1- I ! ! t !

2 D 4 0 6.0 Anode Cathode

D i s t a n c e mm

Figure 35. Axial electron density (ne) distribution in plasmas containing SiO2 (o); SiO2 + 5% each of Ca, Mg, AI, and Fe (A); SiO2 + 12% (Na + K) + 500% graphite (mn); SiO2 + 12% (Na + K) + 300% graphite (.); arc gap 6 mm (reproduced from Avni et al., 1972).

(3 A THOD E 10 , , , , ,

m m r ~

�9 A A I ~

A ~ ~

A A A l b

! ! i i

b -

I i |

o q o k -dl~ c' 0 ~ 0 "

. ~ s , . \ o " ~ .

i 1 I 15

10 5 ne(Cm -3)

i , i i I , , �9

5 1 016 A NOD E

Figure 36. Axial electron density distribution in a dc arc. (o) AI203/C 1:3; (n) LiF/C 1:4, (A) KF/C 1:1 (de Galan, 1965).


18

14

- n i n e ,;_log K

_

/ ,,.-,,.--

/ / / . /

Y

5000 7000 ~r TOK

F i g u r e 3 7 . Saha's constant for ionization (log Ki) as a function of temperature of a dc arc plasma containing U308 (e), Z r 0 2 (A), and ThO2 (I) (reproduced from Avni, 1978).

density can be calculated from Eq. 40. Table 11 lists the radial n e values in the cathode, center, and anode regions with and without the third matrix. Figure 38 illustrates the radial distribution of both electron densities, i.e. one derived from atom and ion spectral lines, the second derived from the energy balance equation (Eqs. 35 and 39, respectively). The values of electron pressures, determined by Dekker and McFadden (1975), are reproduced in Figure 39. The radial distributions of electron density in Figure 36 and Table 11 show the following features:

Good agreement was obtained between ne and ne values in the cathode central regions (Figure 36), while agreement was relatively poor in the anode region. The equality between n e (the Saha equation which requires LTE) and n e (the energy balance which requires steady-state conditions) indicates that LTE exists in the cathode and in the central regions.


1 0 . --

5.0

tO

O.5

0.1

0.05

" 10 ~a ne ; ne c m "3,

-u\

\ rk I~ ' r

- ' \ j ! I & \

-'~"7 ~ ! " . Lii" i

, u"

\ _

1 ! I I 1.0 2.0

R m m

Figure 38. Radial distribution of electron densities ne and ne for a 6 mm gap in the cathode (C), central (B), and anode (A) regions, ne measurements ( - - - ) for plasmas containing La203 (A), U308 (e), ZrO2 (*), and ThO2 (o). ne calculations (horizontal bars) for plasmas with La203 ( _ . m ) and ZrO2 (--) (reproduced from Avni, 1978).

2. The electron density decreased by approximately 1 order of magnitude over a radial distance of 2 mm from the arc axis. The rate of electron density decrease was similar in the anode, central, and cathode regions. The radial gradient decreased with increasing electrical power in the dc arc (Figure 39) (Decker and McFadden, 1975).


4"0

c =E 2~ to d-- o Z Q 0 $ _

L l I , I I _ I

_ _ _

0 I 2 3 0 I 2 3 r , rnrn

o o~-0,o

I I I ~ I _

0 I 2 3

Figure 39. Radial electron pressure distributions obtained from samples buffered with 50% CaF2 at different arc powers (according to Dekker and McFadden, 1975).

In the graphite-air plasma, ne(r ) values were lower than the value in the presence of the refractory matrix by a factor of 3 (Table 12), i.e. the mean electrical conductivity (s) was higher in plasmas containing a third matrix.

Local Thermodynamic Equilibrium (L TE) When a plasma is in complete thermodynamic equilibrium (LTE), the neutral and

charged particles and photons are in equilibrium, both mutually and with the

Table 12. Normalized Spectral Line Intensities (Jcathode/Janode) for Several Ion and Atom Spectral Lines of Trace Elements in U3Oa and

SiO2 Matrices a'b

Jcathode/Janode

Spectral line (nm) U308 Si02

Fe II 259.9 6 3

V II 311.0 4 0 .9

Cd II 228.8 4.8 2.8

Cr II 283.5 5.8 3.5

Mg II 280.2 6.9 3.3

Mn II 257.6 8.4 5

Fe 1 248.8 3.5 0.83

V 1 313.8 0.95 0.9

Cd 1 326.1 2.4 1.2

Cr 1 302.1 2.2 1.2

Mg 1 285.2 3.3 0.8

Mn 1 279.4 3.7 2

Notes: aReproduced from Avni, 1978. bArc gap 6 mm, current 12 A; 100 mg/kg concentration of each trace element.


surrounding environment. LTE is characterized by the following conditions (Bou- mans, 1966; Finkelenburg and Maecker, 1956; Lochte-Holtgreven, 1968):

1. The velocity distribution of all free particles in all energy levels satisfies Maxwell's distribution law.

2. For each particle the relative population of excited energy levels conforms to Boltzmann's distribution law.

3. The ionization of all free particles (molecules, radicals, and atoms) satisfies the Saha-Eggert equation.

4. The radiation density is consistent with Planck's law. Due to the absence of barriers having temperatures equal to that of the arc gas, equilibrium between light quanta and material particles is not attained, i.e. Planck's law is not satisfied for all parts of the discharge. Nevertheless for small regions in the dc arc core, Planck's law can be satisfied and a local LTE exists.

P l a s m a T e m p e r a t u r e s . The first three constraints (1-3 above) imply that there are five of the following different plasma temperatures (Lochte-Holtgreven, 1968):

1. Electron temperature (Te), determined by the kinetic energy of the electrons. 2. Gas temperature (Tg), defined by the kinetic energy of the neutral atoms

and/or molecules. 3. Excitation temperature (Tex), determined by the population of various energy

levels. 4. Ionization temperature (Ti), controlled by ionization equilibria. 5. Dissociation temperature (Td), governing molecular or radical dissociation

equilibria.

The radial temperature gradient in a dc arc inhibits thermal equilibrium; however, if the change in temperature along one mean free path is small in comparison to the mean temperature value in that region, then the influence of the temperature gradient on the equilibrium conditions is negligible. Since the mean free path in a dc discharge at atmospheric pressure is small, the temperature gradient will not significantly influence thermal equilibria.

One of the criteria for LTE is when all the different temperatures are equal, i.e., T e = Tg = Tex = T i = T d. Generally in adc discharge the various temperatures depend on the electric field strength (E), the gas pressure (p), and the p / E ratio (Smit, 1951; Boumans, 1966; Finkelenburg and Maecker, 1956). At reduced pressures and in the presence of strong fields, T e > Tg, LTE will not be obtained, whereas at atmospheric pressures and attenuated E, T e .~ Tg, conditions for LTE are fulfilled. An additional criterion for LTE was developed by Griem (1964), namely that at electron densities n e > 1016 cm -3, and LTE is reached, while at lower values, n e < 1016 cm -3, and the plasma is not in LTE.


Boumans (1966) reviewed the experimental evidence (see also references cited by Boumans (1966, chapter 5)) and concluded that LTE exists in the dc arc plasma burning freely in air (except in the cathode region). Avni and Klein (1973a) reported measured values of AT = T e - Tex, T z, Tex, a n d T v (vibrational temperatures using CN bands). For a graphite-air plasma containing third matrices such as U308, ThO 2, La203, Cr203, and BaCO 3, LTE was reached within the experimental error in the three regions of the arc column, namely the anode, center, and the cathode. Accurate temperatures were determined in the central region of a free-burning arc in air (Avni and Klein, 1973).

Olsen (1963) discussed the experimental evidence for LTE in the central part of a stabilized free-burning dc arc and showed that it is suitable for measuring transition probabilities of excited spectral lines. Experimental electron densities of a free-burning atmospheric dc arc indicates that it is in LTE (Avni and Klein, 1973b, Figures 32-39). The electric current heats the arc gas resulting in the high temperature which controls the dissociation, excitation, and ionization of atoms and molecules (Boumans, 1966).

Before discussing the status regarding LTE in a dc plasma jet it is worthwhile extending the discussion of the Elenbaas-Heller energy balance equation. Vukanovic et al. (1971) and Pavlovic et al. (197 la,b) studied the Elenbaas-Heller equation (Eq. 32). They investigated the free-burning arc in N 2 (Vukanovic et al., 1971), Ar, and water vapor (Pavlovic et al., 197 l a) and evaluated the effect of lithium (Pavlovic et al., 1971b). This equation was solved in order to obtain the radial temperature distribution in the center of the arc discharge. This assumed an a p r i o r i knowledge of the radial distribution of electron density n e and appropriate values of the thermal conductivity (r). Equation 32 can also be used to determine the radial distribution of ne, given the radial temperature distribution as was shown for a free burning dc arc plasma containing uranium oxide (Avni and Klein, 1973a).

The Elenbaas-Heller equation (Eq. 35) for radial temperature distribution can be rewritten using Eq. 30,

a E 2 = - 1 / r d/dr(rp dt/dr)

where ~ is the electrical conductivity in ohm -~ cm -~, E is the electric field strength in V cm -1, r is the radial coordinate in cm, and p is the thermal conductivity in erg s -1 cm -l deg -l. As stated previously, in order to solve Eq. 39 for the radial temperature distribution, a and p must be calculated. According to Vukanovic et al. (1971), the electrical conductivity, a, is given by,

= e n e ~ e = e 2 ~ e n e / m e V e (39)

where ~te = e ~ , e / m e V e is the electron mobility in cm2V-ls -l, e is the electron charge in C, ~e is the mean free path in cm, m e is the electron mass, and v c the average random velocity of the electron.

Equation 39 is based on the assumption that 99% of the electric current in the plasma is carried by electrons. Vukanovic et al. (1971) calculated the electrical

Direct Current Arcs and Plasma Jets 12 7

1 4 -

2 f

I O - E u

tm

b 6

o

, , , �9 I 1 1 , , , 1 1__ 1 ..... 1 _ L

2oo0 - 4 0 o 0 6 0 0 0 eooo ~o, ooo T,eK

Figure 40. Electrical conductivity of a nitrogen plasma as a function of temperature (according to Vukanovic et al., 1971, Figure 12, and Pavlovic et al., 1971 a, Figure 5).

conductivity as a function of temperature (Figure 40; Vukanov et al., 1971; Pavlovic et al., 1971 a). This was based on the fact that the thermal velocity of the electrons at arc temperature [v e = (3KTe/me) 1/'2] exceeds the drift velocity (Ved = L e x E) at an electric field strength in the arc (0 < E < 500 V cm -~) burning in air, N 2, or in Ar at atmospheric pressure. The calculation of electron density, required to solve Eq. 39 for o, is described in Section ll.B.

The thermal conductivity is the summation of the classical thermal conductivity (Pn), which includes free particle motions, the thermal conductivity due to thermal diffusion (Or), and the thermal conductivity due to reaction energy transfer (9r)" The reaction energy includes dissociation of molecules into atoms and ionization of atoms. Because py > p,, > Pt in a dc arc at atmospheric pressure [Roes (1962), Finkelenburg and Maecker (1956), and Boumans (1966)], Pn and Pt Can be neglected to a first approximation (Vukanovic et al., 1971). As a result, thermal conductivity 9r can be expressed as,

P~= Cpfl D (40)

where Cp,. is the specific heat at constant pressure due to dissociation and ionization, d is the density of the plasma, and D the coefficient of diffusion:

C p r = ( S H l a T ) p (41)

At the high temperature prevailing in the dc arc, the thermal conductivity of the electron (Pe) cannot be neglected, and the following equation (Drawin, 1966; Vukanovic et al., 1971) was employed to derive Pe,

Pe = 2/3ne~'evek(1 + cL) (42)

7

U U

5 T E

~" 4 U

where k is Boltzmann's constant and a is the degree of ionization. The values of P = Pr + Pe at various temperatures, depicted in Figure 41, have been calculated by Vukanovic et al. (1971) and Pavlovic et al. (1971 a). For the mathematical solution of E in the Elenbaas-Heller equation the reader is referred to Maecker (1959a, b). Experimental radial temperature distributions agree adequately with the calculated distributions (Vukanovic et al., 1971; Avni and Klein, 1973b), indicating that the central portion of the dc arc discharge is in LTE. Excitation temperatures were measured by the "two-line" method, and the gas temperature was determined by solving the Elenbaas-Heller equation. The solution of the Elenbaas-Heller equation (Eq. 35) is not strictly necessary for a dc arc burning at atmospheric pressures (Pearce, 1958; Boumans, 1968; Kulagin et al., 1972). The approximate procedures used here (either given gradrn e and predicting gradrT, or vice versa) are powerful tools for understanding and predicting:

o

t 4 1

0 . B

3

L . . . . L I I 2OO0 4OO0 6OOO 80OO

T , ~


Figure 41. Thermal conductivity of nitrogen as a function of temperature (according to Vukanovic et al., 1971, Figure 11, and Pavlovic et al., 1971 a, Figure 6).


1. The nature of the processes and parameters governing the dc arc plasma at atmospheric pressure.

2. Dissociation and ionization reactions which determine transport properties in the plasma.

3. The distribution of temperature and electron densities (Section II.B)--variables directly related to spectral intensity which is of prime concern for the analytical spectrochemist.

Another important, well-known feature of Eq. 35 for dc plasmas is that since it is an energy balance it requires only an equality in steady state (not LTE) between energy influx (electrical) and energy loss (which is only thermal at atmospheric pressure) per unit time per unit volume.

Plasma Jets. The dc argon plasma jet will now be discussed taking into account the features of the Elenbaas-Heller energy balance equation. French physicists have performed extensive investigations on LTE in de plasma jets at low and at atmospheric pressures (Borasseau et al., 1970; Cabannes et al., 1970; Czemichowski et al., 1970; Cabannes and Chapelle, 1971; Ranson and Emard, 1973; Drawin, 1974). Models for analyte emission behaviors were evaluated in detail by Miller et al. (1984).

Bourasseau et al. (1970) investigated whether LTE exists in an Ar plasma jet between pressures of 6 to 700 torr, at various gas flows, and at a power of 4.5 kW (400 A current). Spectral measurements were made 0.5 cm above the anode. Excitation temperatures (Tex) were measured using the "two-line" method with Ar (415.859 and 518.775 nm) and by the measurement of the absolute line intensity of Ar (415.859 nm). In the latter technique the temperatures used for calculating atom and ion populations of Ar in the plasma were unknown. The third temperature measurement was obtained from the Saha equation (Tio n) with known electron density values (ne). Again the plasma temperature used for calculating the Ar atom population was unknown. The electron density was obtained from Stark broadening of Ar (415.859 nm) (n~t), from the continuous spectrum in the visible region (n~) and from the Stark broadening of Hf) obtained by introducing 1% H 2 into the Ar plasma jet. Bourasseau (1970) reported approximately equal temperature values for Tex ,~ T 1 = 13,000 K at 700 torr, Tex = 10,000 K, and T i = 9000 K at 5.8 torr. Electron densities were also found to be approximately equal to nes t z nec z 1017 cm -3 at 700 torr and nes t ~, nec -~ neHf) = 4.4 x 1014 cm -3 at 5.8 tort. On the basis of these results it can be concluded that LTE exists in an argon jet plasma jet between 6 and 700 torr or between electron densities of 4.4 x 1014 _< n e < 1017 cm -3. The criteria for LTE determined by Griem (1964) imposes values n e > 1016 cm -3 and not n e _> 1014 cm -3.

Bourasseau and co-workers checked the validity of the LTE criteria at lower pressures, i.e. at n e < 1015 cm -3 and observed departures from LTE, probably due to total absorption of the spectral lines (415.8 and 518.7 nm) used for the temperature and electron density measurements. Cabannes et al. (1970) measured tempera-


tures using the Ar neutral species (Tg) in a Pitot tube (dynamic pressure) employing a Mach-Zender interferometer (refraction index). They concluded that LTE occurred at a height of 0.5 cm above the anode (the hollow anode was the counter electrode of the dc arc), since T ~ Tex z Tg ~ 11,000 K for a plasma jet operating at 5.3 kW, 315 A, and 6 L min -l (Ar flow). At a height of 4 cm in the plasma jet, LTE was not attained, i.e. Tex > T at 10,000 K and 4500 K, respectively. These authors (Cabannes et al., 1970) attributed nonexistence of LTE to the diffusion of air into this zone and, to avoid this, the plasma jet was enclosed in a chamber (upper part open) and argon was introduced laterally with respect to the jet axis. LTE was still not attained under those conditions.

Czemichowski et al. (1970) extended this study and performed measurements at a height of 4 cm. Cabannes et al. (1970) measured Ar line intensities of a wide range of energy levels (starting from the 4p levels just above the ArI metastable state) and correlated them with the energy levels. The slope of the linear "one-line" method indicated that Tex was 4800 K, which is approximately equal to T,, measured using a Mach-Zender interferometer (Cabannes et al., 1970). Even though Tex ~ T,,, LTE was not obtained at 4800 K (Cabannes and Chapelle, 1971; Ranson and Emard, 1973; Drawin, 1974). However, LTE does not exist since temperatures measured using ArI (down to the metastable state) resulted in a temperature of 10,000 K. Thus the plasma jet behaves like a tank of metastable Ar atoms in LTE with upper excited levels which are not in equilibrium with ground levels (Czernichowski et al., 1970; Cabannes and Chapelle, 1971; Ranson and Emard, 1973; Drawin, 1974). In other words, Ar and Ne dc plasma jets (when observed several centimeters above the orifice) are recombination plasmas in which levels are populated from above (cascading) and not from the ground state as dictated by the Boltzmann distribution. In conclusion, the plasma region adjacent to the jet orifice (at a height (I) < 1.0 cm) is in LTE, whereas at a height of several centimeters, LTE does not exist.

The radial distribution of electron density and electric field strength can be predicted in the analytical regions of the jet (above the exit orifice where LTE does not exist) with the use of the Elenbaas-Heller energy balance equation and the measured radial distribution of the gas temperature. The results of this calculation are listed in Table 13 where average literature values of Tex and T r were used for calculating the average electron density [n e and electric field strength (E)]. As shown in Table 13, the calculated n e values are in good agreement with the electron densities reported in the literature for an argon plasma jet. It should be stressed that the best procedure for applying the Elenbaas-Heller equation (Eq. 35) is to use the measured radial distributions of electron density (Stark broadening, continuum or Saha) to determine the radial distribution of gas temperature (Tg).

At a height of 4 cm and higher above the plasma jet orifice, the use of the energy balance equation is restricted (Table 13) because the dc electrical energy input T E 2

is very low due to the deterioration of the dc plasma discharge. The large discrepancy between the calculated and other n e values (Czemichowski et al., 1970) is indicated in Table 13.


Table 13. Comparison of the Calculated Average n e and E Values from the Elenbaas-Heller Equation (Eq. 40) with Literature Values

of T, c~, and p for Argon Plasma Jets Height Above Orifice T (K)

(crn) Measured

(~a e Le c kte d Measured n e n e E t

Calculated Calculated (x lt) ~ mho pb Measured Calculated Measured Calculated (x 10 -5 m) (m2v-ls -I) m -I) (Wm -I g -I) (x 1015 cm 3) (cm -2) (Vcm -I)

1-2 8500 g 1.42 4.0 2.0 0.332 3.2 1.2

1-2 6000 h 1.0 3.37 0.7 0.168 1.3 -1.0 j 1.2

1-2 5500 i 0.92 3.24 0.5 0.155 0.96 1.4

1-2 5000 h 0.83 3.06 0.3 0.144 0.6 1.6

3 6600 j 1.09 3.47 0.85 0.186 1.5 1.6 j 1.2

3 7000 g 1.16 3.62 1.05 0.204 1.8 1.2

4 4500 k 0.75 2.92 0.15 0.131 0.32 2

4 48001 0.8 2.92 0.23 0.139 0.48 2.61

Notes: aFrom Figure 6 in Boumans and de Galan, 1966. bBoulos (written communication). cEe = l/nQ.

dlae = e~ e / mve eDerived from Eq. 38.

tDerived from Eq. 35. gCzernichowstzi et al. (1970). hSkogerboe and Butcher (! 985). iKranz (1964).

JGaydon and Wolfhard (1970). kAlkemade and Herrman (1979).

ICabannes (1974).

Free Particle Concentration

This section deals with two "types" of particle concentration, namely nj and ntj, defined as follows,

n j = Z n a j + Z n i j ( 4 4 )

and:

= (45) ntj nJ + Z n m j

The concentration nj represents a summation over the concentration of atoms (nai) and ions (nij) of the element j. The total particle concentration, ntj, takes into account also the summation of the molecular concentration (nmi).


The free particle concentration (nj) in a plasma was evaluated from absolute line intensity measurements (Boumans, 1966; de Galan, 1965; Borasseau et al., 1970) (see Eqs. 26 and 27). In this method we assumed that atoms and ions are only present in the plasma; i.e., the element j does not form stable molecules. Presence of molecules of elementj was neglected by convenience in the calculations, owing to two factors: (1) inadequate data on the molecular spectra ofj (Pearce and Gaydon, 1976); and (2) calculation of ntj is complicated even if the molecular spectrum ofj is known (Boumans, 1966).

In dealing with trace elements, the nj calculation is legitimate since their concentration in the plasma is small. However, one should proceed with caution since trace elements can form stable molecules with oxygen, carbon, or nitrogen (see Section II.A) and with refractory matrices. For a refractory matrix element, and a stable trace element molecule, ntj was calculated instead of nj.

Suitable methods for the evaluation of n o are not available. For the present purpose, the estimation of n o was made without identifying the type of matrix and the trace element molecules present in the plasma. Whether the molecules are carbides, oxides, or nitrides is of no significance provided they are computed as a whole.

The "wire method" described in Section II.A was used for measuring molecule, ion, and atom concentrations (see also Figure 42 for the configuration of the carriage). The spatial distribution of particle flux ntj (r,z)vj of the matrix and the trace elements were then computed. The deposits on the wires were analyzed using flame AAS (after immersion in nitric acid), neutron activation analysis (La and U), and dc arc OES (Avni and Goldbart, 1973a).

The results obtained correspond to the particle flux ntj (z, r)vj of the matrix (Avni and Goldbart, 1973a). The radial distribution of the particle flux is reported in Figure 43 for various heights (z). The total particle concentrations were calculated from the flux, and (data derived from Figure 43 in next section) was fitted with exponential or Bessel functions. Thus an experimental solution of the "'dnt/dt" model (Boumans, 1966; de Galan, 1965) was obtained. This model requires the evaluation of the axial particle velocity vj as will be shown subsequently.

The dn/dt Model. The dn/dt model as described by Boumans (1966) considered the flow of particles through an infinitesimal element of volume and can be expressed in the form of cylindrical coordinates (r, z) for a steady state:

-d-/-=D + az2j (46)

where D and v i are diffusion coefficients and the axial particle velocity, respectively. Solution of this partial differential equation for a 10-mm arc gap with a core radius (r) of approximately 5 mm was given by Ginsel (1933) and Bavinck (1965). Ginsel


j ~

Figure 42. Schematic wire configuration for measuring the total particle concentration (nt) in plasmas. (1-4) Anode to cathode regions (reproduced from Avni, 1978).

(1933) considered the supply of particles to the plasma from a point source. At appropriate boundary conditions his solution has an exponential form:

nt(z,r) = Oj v j ( 4 7 ) 2rtD4z2 + 2,- exp[- ~ (4z 2 + r 2 - z)

Bavinck (1965) considered a disc-shaped source for particle supply to the plasma. At appropriate boundary conditions (Boumans, 1966) his solution was a first-order Bessel function,

n,(zR, r /R=

oo a 2Qj J(n "~) r

reaD Z B,~,n[j~(~nj + j~(~,,)] Jo(~'n-R ) exP(RB) n-I

(48)

B' = qW 2 + Ln 2 + w; B = qW 2 + Ln z - I/(/; W - up 2D

in which n is an infinite series of positive numbers satisfying the boundary conditions; J0 and J1 are zero and first-order Bessel functions, R is the radius of the


arc core, and a is the radius of the anode crater. With R and Qj known, the functions nt(z, r) can be calculated from Eq. 48 using appropriate values of a, D and vj.

In a previous publication (Avni and Goldbart, 1973b) we demonstrated that the results illustrated in Figure 43 can be fitted to Eq. 48. In this case the ratio v/2D was obtained directly from the slope (Avni and Goldbart, 1973b) of the exponential plots of particle flux nt(z, r)vj in Figures 43-45 for U and La, respectively.

Particle Velocity. Particle velocity values of vj/2D can be obtained from the slopes of the lines shown in Figures 43-45 using Eq. 48. The axial particle velocity vj was obtained from this parameter after the diffusion coefficient D was calculated. The following equation (Avni and Goldbart, 1973b; de Galan, 1965) was used for D,

D = constant Tl"Vs/'~M*dc (49)

11

'E

"= 10

I r

o

I

o �9 �9 �9

I I i

1 " 5 10 "

l)istance trom arc axis,ram

Figure 43. Radial distribution of U (U308) total particle flux at various heights (z) in the plasma as derived from the 2 mm wire. Gap 8 mm, carriage velocity 110 cm s -1 . Distance above anode z = 0.2 mm (*), 2.2 mm (o), 4 mm (u), 5.9 mm (&), 7.8 mm (o) (reproduced from Avni, 1978).


in which M* is the reduced mass, d c the diameter of the diffusing particle, and T the absolute temperature in the given region of the dc arc plasma. The diffusion coefficient was calculated using Eq. 49 for different temperatures of the various regions in the plasma (1 < r < 3 mm and z up to 8 mm). A plot of the experimental values of the axial velocity of U and La particles is given in Figure 46. With exception of the anode region, vj was constant over the arc gap within the experimental error.

Total Particle Concentration of the Third Matrix. In a previous paper (Avni and Ooldbart, 1973a), the proportionality factor between the particles deposited on

lC

5

E U

cO

"0 o ~

A

e. I l l

v o

I :

N 1

~L.

_o U

mm I,.

I 0 mm ,o

1 3 5 7 9 1 1 1 3

Distance from arc axis r , m m

Figure 44. Radial distribution of uranium (U308 + 4% UF4) total particle flux at different heights (z) in the plasma, as derived from 2 mm wire portions. 8 mm gap; carriage velocity 110 cm s -1. Distance above anode" z = 0.2 mm (*), 2.2 mm (o), 4 mm (,,), 5.9 mm (A), and 7.8 mm (e) (reproduced from Avni, 1978).

ut

% 3 1 : u g.

"-;-~oo x" I

0 U

N tO

1 5 10

Distance from arc axis, mm

Figure 45. Radial distance of La total particle flux at different heights (z) in the plasma, derived from a 2 mm wire segments. 8 mm gap, carriage velocity 110 cm s -1 . La203 (A) and (B) cathode and anode regions respectively; La203 + 4% LaF3 (+ and �9 cathode and anode regions, respectively.

E l , l

o m

E l m

o I , ,

Anmle

7 0

6 0

5 0 0 ~\,,\ ~

4 0 0 ~ ~ ~ " ~ ~ ' ~ . 4

Ci

2 4 e ' s c=;'h~~ Distance z m nn

Figure 46. Axial distribution of particle velocity (Vj). (A) U308, (1) U308 + 10% UF4, A La203, [] La203 + 10% LaF3 (reproduced from Avni, 1978).

136


the wire and the particle flux in the plasma was shown to be approximately unity within the experimental error. In other words, absolute values of the particle flux in the plasma can be measured using the "wire" method. Using vj values, experimental total particle concentrations were obtained from the particle flux. The axial distribution of nt: is listed in Table 14, together with particle concentrations, ny, calculated from the absolute intensities of U (428.9 and 431.0 nm) and La (279.1 and 272.5 nm) lines. According to this table the concentration of U and La molecules in the arc plasma (R = 2 mm) was approximately 20 times that of their atom and ion concentrations. The high concentration of molecules and the relatively low concentration of atoms and ions of the refractory third matrix can explain the relatively high temperature of the plasma in spite of the relatively low first-ionization potentials of these elements.

The radial distribution of n t o f the third matrix element in the anode, central and cathode regions of the plasma are given in Table 15. The high radial diffusion of U and La particles in the anode region decreases toward the cathode. Compared to the n t values in the anode region, a lower axial particle concentration was observed in the cathode region. Beating in mind that the width of the plasma in the cathode region exceeds that in the anode region (see Figure 1 and Table 1), then the decrease toward the cathode can be explained by axial movement of particles to the outer part of the arc core. Therefore, for radii larger than 2 mm, n t w a s almost constant at various heights in the arc gap (Avni and Klein, 1973b) as shown in Table 16.

Table 14. Axial Distribution of Matrix Element Particle Concentrations (nt and n i) in U308, ThO2, and ZrO2 Matrices a'b

Total Particle Concentration (cm -3)

n t x 10 - l l c nj • 10 - l l d n t - nj /nj

(mm) U Th Zr U Th Zr U Th Zr

Anode

0.2 270.0 280.0 220.0 11.0 13.0 6.0 23.5 20.5 35.7

1.8 110.0 140.0 100.0 6.0 9.0 4.0 17.3 14.6 24.0

3.5 90.0 100.0 60.0 5.0 7.0 2.0 17.0 13.4 24.0

6.0 110.0 80.0 30.0 4.0 5.0 1.5 26.5 15.0 19.0

7.8 60.0 60.0 20.0 3.0 4.5 1.0 19.0 12.2 19.0

Cathode

Notes: aFrom Avni, 1978.

bArc gap 8 mm, core radius 2 mm, exposure 35 s.

CWire method. Mean value of 4 wire segments; RSD = 25%.

dMean value of 4 spectra (arc focused on entrance slit); RSD = 20%. Spectral lines used: U 1424.6 and II 431.0 nm; Th 1401.2 and II 402.5 nm; Zr 1423.6 and II 423.1 nm.


Table 15. Radial Distribution of Total Particle Concentration (n t) of U (U3Os) and La (La203)

n t x 10 -11 cm -3 Relative Standard

Plasma Region r (ram) U308 Matrix La203 Matrix Deviation (%)b

Anode 0 -2 270.0 30.000 25

2 -4 35.0 - - 25

4 -6 5.0 - - 25

Center 0 -2 100.0 3500 25

2 -4 45.0 - - 35

4 -6 9.0 - - 35

Cathode 0 -2 60.0 27.000 25

2 -4 30.0 - - 35

4 -6 7.0 - - 35

Notes: aArc gap 8 mm, current 8 A, exposure 35 s. bObtained from the passage of 4 carriages; i.e. 16 wire segments in the anode region, 8 wires in the center, and 4 wires in the cathode region.

Particle Concentration of Trace Elements

Axial Distribution. Trace element particle concentrations njr r, were calculated only in the arc core (R >> 2 mm) as a function of height in the plasma. This axial distribution of njr r was calculated on the assumption that stable molecules do not form in the plasma. The axial distribution of njr r for trace elements in U308 and La203 matrices as well as a comparison with the nj of U and La is illustrated in

Table 16, Total Par t ic le C o n c e n t r a t i o n Ratio N o r m a l i z e d to t he Va lue

in the A n o d e Region a

Plasma Region

nt /n t A node

Radial Distance (mm) U Th Zr

Anode 0 -2

2 -4

4 - 6

Center 0 -2

2 -4

4 -6

Cathode 0 -2

2 -4

4 -6

Note: aReproduced from Avni, 1978.

1.0 1.0 1.0

1.0 1.0 1.0

1.0 1.0 1.0

0.37 0.32 0.32

1.30 0.80 1.0

1.80 0.90 1.25

0.22 0.21 0.14

0.86 0.80 0.67

1.40 1.00 1.25


Figures 47 and 48. When the concentration of the trace elements is 100 mg/kg, njr r values are smaller by 2 orders of magnitude than that of the matrix element, nj, in the central and cathode regions of the plasma. In the anode region, njr r is approximately 3 orders of magnitude smaller than that of the matrix, nj, indicating that radial diffusion of trace elements is smaller than that of the matrix element. The accumulation of njr ~ in the cathode region is in sharp contrast to the values observed in the anode region.

A comparison of n i t r (Figures 47 and 48) with n t values for matrix elements in the anode region (see Table 14) showed that n t > njT r by approximately 104. This is in a good agreement with the 100 mg/kg contents in the solid matrices. In other words, the ratio of matrix particles to trace element particles in the solid phase was similar to that in the anode region in the plasma, indicating that molecular species of Cr, Mn, Fe, and Mg can be disregarded in the calculation of their particle concentration in the plasma.

R a d i a l D i s t r i b u t i o n . de Galan (1965) measured the radial distribution of particle concentration in the anode and the cathode regions of a de arc at atmospheric

M 'E 20

I I o

~" 16 @ 0

U L

�9 12 e u

o u

�9 8

L. a

",ado \

%

4 ~ ~ / / "---...

1 3 5 Anade Cat~ode

Distance m m

Figure 47. Axial distribution of several trace element particle concentrations (nj) in U308 for Cr 302.1 and 283.8 nm (m), Mn (279.4 and 257.6 nm (o), Fe 278.8 and 259.8 nm (A), and Mg 279.0 and 278.2 nm (,).( .... ) nj of U atoms and ions (reproduced from Avni, 1978).


E

k-

O m qm o "12- b,

I : O Idl

= o . . l O Z O 81- ' k - - - -

�9 _ .o

h 4 - ~ . e - - - - -

\

\ \

1 3 S A n o d e C a t h o d e

l ) i s t a n c e m m

Figure 4& Axial distribution of several trace element particle concentrations (nj) in La203. Cr (n), Mn (o), Fe (A), and Mg (.). ( .... ) nj of La atoms and ions (reproduced from Avni, 1978).

pressure and an electrode gap of 100 mm. Table 17 shows the radial distribution of log n and log njr r in a KF--graphite matrix ( 1" 1). In this table several features of the spatial distribution are indicated:

1. The ratio of nj ln jr r in the plasma is similar to the value in the solid phase in the anode crater (de Galan, 1965, Table 3.2).

2. Major differences in particle concentrations between the anode and cathode region were not observed, owing to the buffeting effect of K in the plasma, the decrease in arc temperature, and its homogeneity in the anode-cathode regions (see Figure 26). The njr r values correlate well with the nj (K) distribution.

3. Small radial gradients were observed for both nj and njr r up to 2.4 mm away from the arc axis. Radially the njr r values correlated well withj distributions.

Owing to the effect of K in the plasma, the anode regions were evenly populated. njr r values decreased by about 1 order of magnitude for radii larger than 2.4 mm, namely up to 4 mm.

\ \ \

3 S

Cath

ode

l)istance

m m

several trace element particle concentrations (nj) in

(.). ( .... ) nj of La atoms and ions (reproduced

mm

. Table 17 show

s the radial distribution of m

atrix ( 1" 1). In this table several features of the

plasma is sim

ilar to the value in the solid phase in 1965, T

able 3.2). concentrations betw

een the anode and cathode ow

ing to the buffeting effect of K in the plasm

a, tem

perature, and its homogeneity in the anode-cathode

njr r values correlate well w

ith the nj(K) distri-

observed for both nj and njr r up to 2.4 mm

away

njr r values correlated well w

ithj distributions.

plasma, the anode regions w

ere evenly populated. order of m

agnitude for radii larger than 2.4 mm

,

Table 77. Radial Distributions of Particle Concentrations in a DC Arc Plasma Containing Ka

Values of log n near the lower, supporting electrode (the anode); unit ~ r n - ~

r ( m ) T(KJ k x n , K In TI Ga A1 Pb M g Sn B Sb Be Zn

0.2 5590 15.24 1.2 5350 15.24 2.4 4960 15.07 3.6 4790 -

A log n (r = 0): 5 logn ( r = O ) - logn ( r = 4 mm):

0.2 5660 15.51 1.2 5330 15.26 2.4 5130 15.10 3.6 4620 14.83

log n:

anode and cathode region:

-

difference in log n between

14.61 15.10 15.08 -

14.9

14.88 14.87 14.74

14.83 +o. 1

~ ~~~

11.51 12.15 11.82 11.88 12.04 12.73 12.03 11.78 12.29 12.05 11.93 12.00 12.92 12.51 11.39 12.05 11.55 11.74 11.76 12.48 12.32 10.87 11.29 11.63 11.08 1 1 . 1 1 - 11.83

11.6 12.2 11.9 11.9 12.0 12.8 12.4 0.7 0.8 0.3 0.8 0.8 0.6 0.6

Values of log n near the upper electrode (the cathode); unit cm-'

11.30 11.96 11.69 11.53 11.23 12.16 12.23 11.70 12.18 11.83 11.79 11.70 12.33 12.17 11.55 12.10 11.55 11.54 11.56 12.16 12.01 11.64 12.20 11.72 11.75 11.73 12.54 11.95

11.51 12.10 11.70 11.69 11.55 12.40 12.09 +0.1 +O.l +0.2 +0.2 +0.4 N.4 +0.3

13.50 14.48 14.21 13.44 14.34 14.17 12.85 14.06 14.06 11.99 13.25 -

13.5 14.5 14.2 0.5 1 .o 0.4

12.92 13.49 13.46 12.97 13.93 13.80 12.53 13.72 13.50 12.4 13.94 -

12.94 13.78 13.59 +0.5 +0.7 N.7

14.90 14.79 14.49 13.9 1

14.8 0.9

14.24 14.39 14.18 14.41

14.26 +0.6

Notes: "From de Galan. 1965, Table 4.8. bRadial decrease of log n; value of refers to region near the arc axis.


Decker and McFadden (1975) investigated the radial distribution of njT r in LiE KF, CaF 2, and BaF 2 mixtures with graphite (1:1). Some of their radial distribution measurements of njT r are reproduced in Figure 49.

Figure 49 shows that when a CaF 2 buffered arc is observed off axis, the njT r values increased at a radius of about 1 mm in an 8-mm arc gap burning in air. Approxi- mately similar radial distributions of n j r r were reported for the other matrices (Decker and McFadden, 1975).

Zn 3076 Sn 3262 Mo 3170

0.4 0.4 0.4

Height above anode

7ram

I'0 I'0 I '0 ' V ~

lu

I:: 0-4 0"4 0"4

0

I '0 I-0 1.0

0.4 0.4 0.4

5 mm

3 mm

1.0 I-0 1.0

0.4 I 0-4 - 0.4 l ~l_ o i 2 s o i 2 3 o ~ 2 3

r, mm

I m m

Figure 49. Relative atomic particle distribution in a dc arc at various distances from the anode. Sample was buffered with CaF2 (from Decker and McFadden, 1975, Figure 6).


Transport Phenomena Equation 19, the transport parameter of free particles in a dc arc plasma, can be

rewritten as follows (Avni and Goldbart, 1973b),

Wj(z,r) = Qntj(z" r) = Vj S(z ) (50)

in which z and r are the axial and the radial coordinates of cylindrical symmetry, Qj is the volatilization rate in s -1 of a chemical element j, ntj the total particle concentration (Eqs. 44 and 45) in cm -3, vj the axial velocity of the free particles in the plasma in cm x sec -1, and S is the cross section at any height (z) in the plasma.

Several theoretical models (Boumans, 1965; de Galan, 1965; Vukanovic et al., 1971) for calculating vj have been used: (1) the "velocity model" based on axial migration of particles; (2) the "dn/dt" model based on simultaneous effects of axial and radial migration of particles as described in Section II.B.

The experimental methods for determining Vj or Wj are based on spectral line intensity measurements namely:

1. Measurement of the transit time (Malyck and Sard, 1964) of particles in the plasma; the velocity vj, in the arc plasma is derived transit time over a given distance (Ilina and Goldfarb, 1962; Zilbershtein, 1977).

2. Calculation of particle concentration from the absolute values of line intensifies (Eqs. 26 and 27). The concentrations thus obtained represent only atoms and ions (nj). The value of Vj was calculated from nj using mean values of the volatilization rate, Qj, determined in the same experiment. Both methods, can be employed only when the test elements do not form stable molecules. This was demonstrated by de Galan (1965) using the following empirical correlation,

log Wj = log a j - log nj = 2.5 + txj (51)

in which czj is the degree of ionization of the atom and nj is the average particle concentration of elementj. Thus Wj depends on the degree of ionization only and not on the degree of dissociation of molecular species.

Axial and radial transport of the matrix particles were calculated using Eq. 51. Qj and n t were measured by the "wire" method (Sections II.A and II.B). The radial distribution of Vj in the anode, central, and cathode regions of the plasma are given in Table 18. The outward increase of Wj (arc core radius > 2 mm) indicated that radial transport occurs in the arc mantle.

The axial distribution of the transport parameter in the arc core for a radius of R = 2 mm for both matrix and trace elements is illustrated in Figures 50 and 51. The "wire" method was also employed for the determination of the trace element volatilization rates (see Tables 8 and 9). Our calculated value of vj is compared with


Table 18. Radial Distribution of the Transport Parameter (yj) for U and La Oxide Matrices a'b

~lj = a j / n t c m 3 3 -1 c

Plasma Region r (mm) U308 Matrix La203 Matrix

Anode 0-2 48.0 70.0 2-4 370.0

Center 0-2 130.0 600.0 2-4 290.0

Cathode 0-2 220.0 78.0 2-4 420.0

Notes: aReproduced from Avni, 1978.

bArc gap 8 ram; current 10 A.

r value of 4 carriages. RSD was 25% in the anode region and 30% in the central and cathode regions.

6(

30(: 7 u 25(] e i

M I E

u 210

h

�9 1 9 0 qlm e E E 15o

q= I,,, o 110 Q, m r Q L

~- 7 0

3 0

/ /

/

f /

\ \

\ \ \ \ \ \ \

1 3 ,5 7

Anode C a t bode D i s t a n c e m m

Figure 50. Axial distribution of the transport parameter (~Fj). U308 carriage velocity 110 cm s -1 (A); U308 carriage veloci ty- 250 cm s -1 (e); U308 + UF4 - 110 cm s -1 (m); (---) Cr, Fe, Mn, and Mg trace elements; arc gap 8 mm (reproduced from Avni, 1978).


'E q,I

- 500

o

! I i

2 4 6 C o l b o d e

Dislonce z m m Anode

Figure 51. Axial distribution of the transport parameter (~j). (o) La203; (+) La203 + LaF3 at 110 cm s -1 carriage velocities; arc gap 8 mm (reproduced from Avni, 1978).

those given by de Galan (1965) in Table 19; de Galan's values appear to be too high for both matrix and trace elements owing to the following reasons:

1. In comparison to the matrix (e.g. A1203:C), the volatilization rates for trace elements used by de Galan (1965) is high.

2. The average nj of the matrix element (A1) computed by de Galan is too low; i.e. molecules were not considered. Low values of nj and high values of Qj result in high values for vj, by 1 order of magnitude.

Todorovic et al. (1975) measured the transported velocity of particles introduced in a dc arc plasma by the "bullet" method (Vukanovic et al., 1975); i.e. liquid droplets containing the element of interest were introduced laterally into the plasma. The axial mass transport of metal volatilized from the "bullet" was influenced by diffusion, convection, and electric fields. The velocities of Na and Li are reported in Table 20; these were obtained by the oscilloscope method and with a high speed camera (Ilina and Goldfarb, 1962). The authors defined and calculated the convection velocity of Na (v c = 122-198 cm s-l), and the velocity due to the electric field (v e = 140-175 cm s-l).

Temporal Variation of Plasma Variables

The transport of the matrix elements, i.e. elements in high concentration, transported to a graphite (or carbon)-air plasma, as has been shown in this chapter, is


Table 19. Comparison of the Transport Parameters (yj) Calculated by the Wire Method with Those Quoted by de Galan (1965) for U3Oa, ThO2, ZrO2, and AI203

Matrices in the Central Region of the Plasma a Log ~j "Wire" Method b Log vTj "de Galan" Method c

U308 ThO 2 ZrO 2 Al203/C = 1/3 U308 ThO 2 ZrO 2 AI203/C = 1/3

2.5 2.3 2.7 2.9 3.8 3.6 3.9 4.1 6.41 d

Notes: aReproduced from Avni, 1978. bQj and nj values for a 0-2 mm radius. CQj and nj values for r = 0-2 mm.

aAccording to de Galan (1965).

directly influenced both by electric and thermal plasma variables; consequently the spectral intensifies of the trace elements. Because of time-dependent variations in spectrochemical analysis, the temporal variations of plasma variables will now be addressed.

Both precision and accuracy will be improved if the plasma variables can be kept constant during the arcing period. In this context, the volatilization and the rates of volatilization of particles play an important role in transport processes in the plasma. Variations of volatilization rates of refractory matrices such as U308, ThO 2, ZrO 2, La203, and SiO 2 are shown in Figure 6. For a period of 40 s, i.e. from 20 to 60 s, their volatilization rates were constant. One of the factors affecting the volatilization rate is the consumption of the anode cup walls during arcing, thus exposing the third matrix to higher temperatures and enhancing its volatilization rate. Mel- lichamp (1978) measured the consumption rate of electrodes as a function of arcing time as shown in Figure 52 and with current supplied to the anode as shown in Figure 53. The consumption rate expressed in mg/s indicates a 20% loss in carbon

Table 20. Axial Velocities and Mean Axial Velocities (in cm s -1) of Li and Na in a DC Arc Plasma a

Li Na

V 1 (to cathode) 930

V2 (to anode) 690 V1 680 V2 -80 b V3 210 V4 390

630

280

Notes: aData from Todorovic et al., 1975.

bThe negative sign denotes upward movement.

3 5 0 ' I ' ' I ' ' I I ' I ' i~-

3 0 0

E 25O "O E

c 2 0 0 8 i -

~

", 150 "13 O L

U . , 1 0 0 W

C a r b o n A n o d e f

f f

_ ~ ~ - G r a p h i t e A n o d e J

,,p _ ~ ( ~ ~ ~ " O ~

~ ~ ' ~ ' / / G r a p h i t e C a t h o ~ e z : ~ - - ~ . , ~ ~ r

5 0 - I ~ / ~ / " ~ ~ . . . . . i

, I , , I I , I ,,, I I 0 3 0 6 0 9 0 1 2 0 1 5 0 1 8 0

T i m e (sec)

Figure 52. Relation between rates of consumption of carbon and graphite electrodes in a 10 A dc arc (Mellichamp, 1978, Figure 3.5).

2 0 0

1 7 5

--- 1 5 0 et ot E ~, 1 2 5 O .J

.c 1 0 0 01

= 75 13 c

~ 50

A m p e r e s (Open) C a r b o n 5 10 15 2 0

, . , I 1 I v, , I v . G r a p h i t e ~ - f / ,

//

C a r b o n / - /d,, ' /

0.12" Diameter / /- A n o d e s / / / /

. ~ , p ~ Graphite

~ ,~ o~ ~" "~4~

25

J I I I .... ! , I I I 5 10 15 2 0 25

A m p e r e s ( S h o r t )

Figure 53. Weight loss of carbon and graphite from anode cups as a function of arc current (Mellichamp, 1978, Figure 3.6).

147


or graphite during 100 s of arcing (Figure 52). A smaller loss was observed by increasing the current up to 15-20 A (Figure 53).

Szabo (1974), in a review of Hungarian studies, attributed electrode consumption to oxidation (anode) and reduction (cathode) phenomena of the electrode surface during arcing in various gas atmospheres. Mellichamp (1978) also demonstrated the influence of electrode consumption on the volatilization rates of trace element impurities from silicon, measuring their line intensities by the moving-plate method as shown in Figure 54. An alternative procedure for expressing the change of Qj with arcing time is shown in Figure 55 (Mellichamp and Grove, 1978). Mellichamp expressed Qj by the voltage drop measured in the arc during volatilization.

1.2

1.0

0 .Sq :b e.,

~. o.s O

0.4

0.2

Graphite -, Fe29290A Crater

I ) 'kk

, , , . . . . . :<L 10 20 30 40 50 60 70 80 90

1.6~k" 1.4~ " ~ . ~ . Ti 2956.1A ,~ "'"-o-.... Carbon 1"2 l "k AI Fe 2929.0A'~ Crater

/ ""~r,6o.4~,~,, " ~ / ' , , \ / ,, , "o r '-~-'~--.~..~ ~, I~ 0.8 ; Si 2452.1A ~ , ;

J ~

~

0.2 b

~ ~o 3'0 ,;) s'o 60 r do g'o Time in Seconds

Figure 54. Trace element volatilization from silicon using graphite (a) and carbon (b) anode cups (Mellichamp, 1978, Figures 3.10a and b).


Boumans and Maessen (1969a,b, 1970) employed stabilized mixed gas Ar and 02 dc arc plasmas for investigating the change in Qj (rate of evaporation (Boumans and Maessen, 1969b)) with arcing time for several elements in fused and nonfused synthetic silicate samples buffered with LiE The moving-plate method was used for measuring time-dependent changes in line intensities. Their results are shown in Figure 9 for the fused samples. Qj values were identified for different groups of elements characterized by specific time-dependent rates of volatilization. In comparison to a slow evaporation mode (5% 02 in Ar), Qj changed markedly with time

z o _> Q

o

.J 0

120 I00 80 60 4 0 20 0

S E C O N D S

Figure 55. Rates of volatilization of In from anode cups varying in cross-sectional area (ram 2) (a) 1.16, (b) 1.77, (c) 3.52, (d) 4.94, (e) 7.04 (Mellichamp, 1978, Figure 3.11).


Ca) 0 ..=.

- " *"*~. YL ~ / "- i

0,~

r ,e,- O ~

i,.,. ~ 0

~ - " X _o _ "- O ~ o ~ Y I , I I �9 0 - O / I > , ~

0 - �9 ~ ~ I = \ e- _ I

e ~

o r - e ~ o ~ , , , 4 P . J

-I , I ,, I I, I J I

6 2 0 0 ;3 7 q~. f o 6 0 0 0 ~ "~

" ~ o o + ~ - 4D �9 �9 . . . . .

E 5 6 0 0 ~ 1 7 6 1 7 6 I "~ I I I , I I

~ - 3 -

0

" - o ~ O , , . I . . o ~ O ~ O ~ % % Q .

" E 0 , ~ - %

~ %

g. _ --I I i I__ I I I , ,

L

I 1

lcJ e "

r - o ~ , " D

C

0

U 0

, , a

0 . i

- / r ' o %

I. I I .. I I .... I., I_ I 0 0 2 0 0 : 3 0 0

Time, see

(a). S low evaporat ion mode (5 ~ oxygen) .

Figure 56. Effect of arcing time on the volatilization rate, plasma variables, and line intensities. Fused samples (according to Boumans and Maessen, 1969b, Figures 6 a-c).


(b) .o. "*"~'*~ YL

I

c: I o

, o . - ._~ o e..

>,,. �9 = o- \...l I

C

._c

o . J _ \

-I - i 1, I - t - - . - -. " i

6 2 0 0 - /

y 6 0 0 0 - " ~.., ssoo ~ . . ~ - - - L ~ , ' ~ E 56oo ~= l I I

@ I h W

o oD o

_1

-5

- 4

-.../'X

. ~ . ~ I j

I ! I

| , | .

j o i n w - -

/

! ,1

o

c

e

u o

,ID

o

-v" \ e'e~e

:

m

i l 1 I I 0 I 0 0 2 0 0

T i m e , s ~

(c)

Z~Yx.~ \

\ .

% ! _ _ 1

t /

I L

m

I f

Time, sec

(b). Medium evaporation mode (10 oxygen)

(c). Fast evaporation mode (25 ~ oxygen).

Figure 56. (Cont inued)


o ..... o

tn I : /D _= Ob _o o

ul c 4D

oD o

- I

(o)

o

0 �9 �9

-I I l

I/-.. \ .>"-- 3 6 z o o

Q. E 5600

I ~ I I I I I I

= - 3 ira in

C E I o o.. u

"~ - 4 �9

o _ I I

>. QfJ C ..e

._a

=3 g u o

. o

I I I I I , I I 0 0 2 0 0 3 0 0

T i m e , s e c

(s). Slow evaporation mode (5 oxygen).

Figure 57. Effect of arcing time on the volatilization rate, plasma variables, and line intensities. Unfused samples (according to Boumans and Maessen, 1969b, Figures 7 a-c).


0

0

1-

.c

_o

0

e~

.c_

0 _1

(b)

- e

- Yu

- e

- e ~ ~ Q

! ! !

3 6200 v 6000 I- ~ "

r~" 58ooi--. " - - ~ e s6oo t-

I~" ",~ ! l . . . l

i / )

- \ ~ - \ / 0 = '6 u

0 . . I I I I

_>,

r ,D-

" 0

u s

.J o

\.,--\ I ,

I I 1 1 IO0 200 Time, sec

('c)

..tip

,l

1 . ! .

i , . L L . I _ J

I//'-." f" 1 l _ 1

\%o J

I t , I 1 0 I00

T ime, sec

(b). Medium evaporation mode (10 oxygen).

(e). Fast evaporation mode (25~ oxygen).

Figure 57. (Continued)


when a rapid volatilization (25% 0 2 in Ar stream) was employed for both fused and nonfused samples. Generally, higher values of Q were obtained for fused samples in comparison to the nonfused samples (Boumans and Maessen, 1969b).

The effect of arcing time on the volatilization rate of both matrix and trace elements is reflected in plasma variables such as temperature and electron density and as a consequenceRline intensities. Boumans and Maessen (1969b) demonstrated this relationship for fused and nonfused silica samples (Figures 56 and 57, respectively). These figures also show that Tex, n e, line and background intensities behaved in a similar way to Q for the different modes of volatilization (slow, medium, and fast) for fused and nonfused samples.

Decker (1973a) showed that Tex and n e behaved differently with arcing time for samples buffered to various extent. Fluorinating agents such as 50% LiE NaF, and

* F 6 0 0 0 j,,e- ---e ,,,,.it.,,

L,, Q

O . s 0

U L

< / ~ I I ! I ! , 5 0 0 0 I 2 3

Time, min

2 0 - I E -

z

! _

o

0 L,,

_o O - - W

1 I t I t I I l I 2 3 4

Time, min

Figure 58. Arc temperature and electron pressure variations during the arcing of samples containing high LiF and LiCI contents. 50% LiF ( �9 ) 50% LiCI (---.---) (Decker, 1973, Figure 5).


BaF 2 were considered to be good modifiers, while the effect of 50% LiC1 and 30% CaF was poor. The difference between efficient and ineffective additives was expressed by the dependency of Tex and ne with arcing time. A good buffer resulted in constant Tex and n e values over a period of up to 240 s, whereas they varied with arcing time if the buffer was inferior (Figure 58). Decker and Eve (1970) investigated the anode spot obtained in a free-burning dc arc in air. The addition of buffers controlled the movement of the anode spot. According to these authors, ideal buffers are alkali and alkaline earth fluorides, preferably of elements low in the periodic table.

Although we have demonstrated that the volatilization of trace elements closely resembled the volatilization pattern of the matrix, as described in Section II.A, the addition of buffers and carriers can modify this process; namely trace elements can be volatilized selectively depending on the volatilization rate of the modifiers-buffers (de Galan, 1965; Boumans and Maessen, 1969b) or carriers (Daniel, 1960; Pepper, 1967). In the spectrochemical determination of trace elements in refractory matrices, using a free-burning dc arc, Qj and other plasma variables were constant with arcing over a period of up to 60 s (see Figure 6), while the addition of graphite (de Galan, 1965; Avni, 1978), buffers (Boumans and Maessen, 1969b, 1970; Decker, 1973a) and carriers (Atwell et al., 1958; Vainstein and Belayev, 1959; Daniel, 1960; Avni and Chaput, 1961; Goldfarb and Ilina, 1961; Raikhbaum and Molych, 1961; Vainstein, 1961; Samsonova, 1962; Siemenova and Levchenko, 1962; Vucanovic, 1964; Pepper, 1967) resulted in constant Qj values for longer periods of arcing time.

Almost complete volatilization and atomization of refractory materials (A1, A1203, W, WO 3, WC, Ta, Ta205, TaC, Mo, ZrO 2, ZrC, SiC, SiO 2, ThC, ThO 2, Nb, NbC, and HfC) was described by Gordon (1976) in a AgC1 buffered argon arc (Gordon and Chapman, 1970). Microvolumes of liquid or suspended samples weighing 25-45 mg were loaded into graphite cups containing 4 mg of AgC1 (Gordon and Chapman, 1970). An argon-supported dc arc was employed for a cycle of 15 s in which the dc current was changed from 10 A (at the beginning) to 30 A at the end of the arcing cycle. Gordon's data indicated that refractory materials were vaporized completely in the arc column with the exception of NbC, W, and TaC (Gordon, 1976).

In dc plasma jets, the role of time on the plasma variables have not been reported. The variation of background emission for 120 min at 368.8 nm in the two-electrode "Spectrametrics" plasma jet was measured and compared with the three-electrode plasma configuration (Poirier, 1979). A low and constant background intensity was observed in the three-electrode system, while in the two-electrode system the background intensity increased markedly after 80 min of operation. This was probably due to instrumental factors more than fundamental processes in the plasma.

1 56 R. AVNI and I.B. BRENNER

Summary of the Spstematic Behavior of the Analyte

The volatilization rate of major and the minor constituents, their dissociation, free particle transport, atomization, ionization, and concentration per unit volume of the dc arc plasma were treated theoretically and experimentally in Section I on the basis of the Mandelshtam's scheme. The last link of the scheme namely, spectral line intensities, were only briefly elaborated. Zeidel et al. (1963), Kaiser (1947, 1964), Boumans (1966, 1971). Mika and Torok (1974), and Zilbershtein (1977) treated spectral line-intensity-related aspects of spectrochemical analysis. The following summarized topics have a major effect on the behavior of the solid phase analyte in the electrode and in the dc discharge.

1. Volatilization rate (Q). This is influenced significantly by the electric power input in the electrode, which also produces the thermal energy necessary for volatilizing the sample. During the arcing period, Q values change mainly due to erosion and corrosion of the electrode material. Thermochemical reactions between the major elements and graphite or carbon inhibit volatilization due to the formation of refractory carbides. The volatilization pattern of the trace elements is similar to that of the major constituents (matrix). High volatilization rates of the matrix translate into higher volatilization rates for the trace elements. Addition of graphite, suitable buffers, and carriers results in regular and constant volatilization rates over the arcing period.

2. Nonbuffered arcs: (a) Temperatures, mean electric conductivity, and electron density attain maximum values in the cathode region of the plasma within a radius of 2 mm from the "axis of symmetry" of the de discharge. These maxima are attributed only to the presence of the matrix in the plasma. (b) The radial distribution of trace element free particles is small. The maximum values of nr~ occur in the zone surrounding the arc axis (r >> 2 mm). The radial distribution of the matrix free particles from the column is high. (c) Owing to (b) above the highest trace element particle concentration is located in the cathode region (relative to the central and anode regions) where the n t of the matrix elements is smallest. The arc column in the cathode region can be depicted to be shielded by an outer zone containing a high concentration of n~ of the matrix particles and molecules such as N 2, CO, O 2, and NO (Avni, 1978). (d) An axial "separation" occurs in the plasma, where the anode and cathode is discriminated with respect to the matrix and trace elements. This "separation" effect is predominant in the cathode region.

3. Graphite as a buffer In the presence of graphite, the Qj values are constant over longer arcing periods in comparison to unmodified plasmas. The presence of graphite also modifies the reactions taking place in the electrode. The plasma variables are similar to those for unmodified plasmas, as summarized in paragraph 2 above.

4. PTFE or organic fluorides as buffers. Qj values increase for both matrix and the trace elements. The pattern of trace element volatilization is similar to that of


the matrix as stated in paragraph 1 above. Qj is constant over longer arcing periods, compared to the unbuffered matrix. The behavior of plasma variables is similar to the behavior summarized in paragraph 2.

5. Chlorides and fluorides of alkali or alkaline earth elements as buffers. (a) Qj values decrease due to a decrease in anode temperature. Qj is constant over long arcing periods (2-4 min). (b) Temperatures and electron densities are maximum in the center of the arc column (>> 2 mm). Tex decreases while the value of ne increases when the ionization potential of the buffer element is z 5 eV). The axial distributions of these parameters are small-large differences in T e and n~ values in the anode, center, and cathode regions were not observed. (c) The radial distribution of the matrix (buffer) particle concentration was similar to the particle concentration of the trace elements. The separation effect was not observed between matrix and trace elements. (d) The residence time of particles in the arc column is probably increased in the presence of alkali and alkaline earth metal buffers, in particular for those that are located in the lower parts of the periodic table (K, Rb, Sr, and Ba). (e) Various buffers have to be added to the same sample in order to obtain a multielement analysis.

6. Carriers. (a) The general trend is an increase in the Qj values for some trace elements. (b) The plasma variables behave in a similar way summarized in paragraph 5b and c. (c) The residence time of particles in the dc arc column increases. (d) As in item paragraph 5e, various carders need to be used for a multielement analysis.

Iii. TECHNIQUES IN SPECTROCHEMICAL ANALYSIS BY DC ARC PLASMA

A. The Cathode Layer Technique

The cathode layer can be defined as the region adjacent to the cathode when the cathode is the lower electrode containing the sample and the anode is the upper electrode. As stated in Section I of this chapter, Mannkopff and Peters (1931) as early as 1931 developed this procedure for spectrochemical analysis of rocks and minerals. They reported a 10-fold enhancement of the spectral line intensities of Zn, Cd, and Cu relative to the other regions of the dc arc. Elements of relatively low ionization potential exhibited cathode layer enrichment, whereas the intensity of elements of high ionization potential such as B and Hg were reduced. A large arc gap of 10 mm was more conducive to cathode layer enhancement. Ahrens and Taylor (1961, page 65) stated that the advantages of cathode layer excitation did not appear to be significant, and that the advantages had been overstated. Never- theless, Mitchell (1940, 1945, 1964), Scott and Mitchell (1943), Scott (1945a,b), and Scott et al. (1971) successfully applied this technique for several decades to the analysis of biological and geological materials.


When it was recognized that the alkali metals with low ionization potential (present in high contents in geological and environmental materials) degraded the cathode effect, Scott and Mitchell (1943) developed separation techniques to eliminate the alkali cathode effect. Decker (1973b) concluded that the cathode layer is a suitable region for analysis. The technique allowed the use of small samples and produced good detection limits. However, it was necessary to position the cathode accurately, and to maintain the observation zone and the gap of the plasma (0.1 mm) during the analysis period. Decker (1973b) also made a detailed study of the processes taking place in the cathode layer. He observed that the cutoff region of the cathode layer was located approximately 1 mm above the cathode when buffers of low ionization potential were not added. The variation of temperature and electron density was relatively small in this region. When a K2CO 3 buffer was added, the temperature and electron density decreased and the signal-to-background ratio (SBR) increased. An increase of the arc current to 12 A resulted in an increase of the temperature and the electron density, but a decrease in the SBR of the trace elements. These results indicated that the plasma in the cathode layer was not in LTE for the following reasons: (1) if the temperatures measured (using the "two-line" method with Zn as the thermometric species) do not represent the plasma temperature (i.e. LTE exists), then the populations of the excited states of Li, Zr, Ce, and Ba do not comply with Boltzman's distribution law; (2) temperatures represent only the excitation temperature of Zn and not the temperature of the cathode layer.

The axial and radial distribution of SBR is a function of the ionization potential, i.e. the higher the ionization potential, the lower the axial and radial regions of maximum SBR. This indicates that the factors causing cathode layer enrichment hinder axial and radial diffusion of atoms. The application of stationary inhomo- geneous magnetic fields and the use of mixed gas plasmas (Ar/O 2 and He) resulted in the deterioration of the detection limits.

B. The Cathode Region

A direct spectrochemical method for trace element determination in refractory matrices such as U, Th, Zr, Pu, REEs, A1, Ti, Mo, and W oxides, as well as phosphorites were developed by Avni and co-workers (1978, and references therein). In order to avoid confusion with the cathode layer technique, this method was called the "cathode region" method. The sample is loaded into the anode cup (the lower electrode) and only the region near the upper electrode (the cathode) is observed. Since it is a direct procedure, carders, buffers, and internal standards are not usually added. However, in the case of refractory material analysis, buffers and carriers are added to enhance the volatilization rates of the refractory sample from the anode cup. Fluorides, at a concentration up to 10% (w/w) and/or graphite are added for the determination of refractory trace elements such as the REEs. The observation zone of the atmospheric free-burning dc arc is between 0.2 to 1 mm


below the cathode tip with an arc gap not less than 4 mm. As described previously the cathode region method is based upon the following features of refractory matrix behavior in the dc arc plasma:

1. The temperature, electron density, and line intensity of the trace elements are maximum in the cathode region (Figures 11-14, 21-23, 32-34). The line intensity is 1 order of magnitude higher in this region than in the anode region.

2. In comparison to the anode region, the cathode region contains the highest particle concentration for the trace elements and the lowest for the matrix elements. This discrimination is indicated in Table 12.

3. This separation between trace and matrix elements is due to the strong radial diffusion of the matrix particles, whereas only a small radial diffusion was observed for trace constituents. The strong radial diffusion, n t, of the matrix element is clearly shown in Tables 14-16.

4. This separation is analytically beneficial since the spectra of the matrix (W, U, Th, REEs) which are normally very dense, are less intense at the cathode and spectral interferences are reduced. Detection limits for the common trace elements and the REEs in U308 are compared with those obtained by carder distillation (see Tables 26 and 27).

The disadvantage of the cathode region method is the need to accurately maintain the observation zone and the arc gap during the analysis--0.2-1 mm below the cathode.

C. Buffers, Fluxes, and Internal Standards

As mentioned previously in Sections I and II, matrix match procedures have been developed in order to compensate for differences in composition of the samples and the standards. Various buffers and fluxes are added in order to modify the vaporization and excitation behaviors of the trace elements and the matrix in the plasma and in the graphite cup. Thus working curves, established using synthetic standards, can be used to determine the composition of the unknowns. In order to apply buffers it is necessary to investigate the behavior of the analytes and the matrix elements in the standards and in the samples both in the graphite cup and in the plasma. It should be mentioned that in cases where the mineralogical and chemical compositions of the samples vary, the effect of buffers are limited and sample fusion may be necessary.

A buffer was defined as a substance, which when added to the samples and to the standards, regulates the volatilization rates of the electrode load and the arc plasma parameters, i.e. electrical and thermal variables. These modifiers promote higher and constant volatilization rates and stabilize the temperature and electron concentration gradients of the arc plasma. They are added in as much as 50% to cause a


suppression of the matrix effect. Among the buffers used in dc arc spectrochemical analysis are graphite, carbon, organic fluorides, Li2CO 3, LiE NaC1, KC1, NaF, KF, C a C O 3, B a C O 3, and others.

The enhancement effect of organic fluorides (PTFE) on the volatilization rates is shown in Table 4. The addition of 4% w/w PTFE enhanced the Qj values of several refractory matrices by a factor of 2. A similar enhancement factor was observed when 20% w/w of graphite was added to phosphorite samples and synthetic standards as shown in Table 21. The Qj values of silicates were also enhanced when 20% (w/w) graphite was added (Figure 59). The effect of 10% A1F 3 on the regulation of the volatilization rate of A1203 is shown in Table 9. The effect of the three- to fivefold addition of graphite to a silicate sample on the axial distribution of temperature and electron density is illustrated in Figures 24 and 25, respectively.

In addition to the thermal effect of buffers, the residence time of the particles in the discharge ig increased. The axial transport parameter, ~gj, for U308 and U308 + 4% UF 4 (w/w) particles is shown in Figure 50. A twofold decrease was observed in the axial values of ~gj on the addition of fluoride (UF4). The minimum value of ~gj for a given plasma cross section (perpendicular to the arc axis) is compatible with the minimum axial velocity of the particles (Eq. 5) and with the increase in the residence time of the particles in the arc column.

Because of the unique properties of the elements, a universal buffer for spectrochemical analysis has not been found and numerous buffers and additives are required to achieve the optimum analytical performance (LODs, minimum matrix effects, etc.) (Boumans, 1966; Ahrens and Taylor, 1961; Corliss, 1962; Maessen and Boumans, 1968; Boumans and Maessen, 1969b, 1970; Decker and Eve, 1970; Scott et al., 1971; Decker, 1973a; Maessen, 1974; Maessen et al., 1976; Avni, 1978).

D. Carrier Distillation

Carriers were used by Scribner and Mullin (1946) to selectively volatilize components from the electrode for the determination of trace elements in refractory matrices. Scribner and Mullin developed an analytical procedure for the determi-

Table 21. Volatilization Rate (Qfh)in Natural and Synthetic Phosphates a'b

Matrix Sample consumed, mg Qj mg/sec

Synthetic Ca(P03) 2 20 0.68 + 10% c

Phosphorite 22 0.73 + 10%

Ca(P03) 2 + 20% C 40 1.33 _+ 20%

Phosphorite + 20% C 46 1.55 _+ 20%

Notes: aData from Avni, 1978. t'I'he anode charge was 50 mg; arc gap 4 mm; arc time 30 s.

CRSDs were determined from 10 replicates.


i

10 |

�9 0 , 5 tPt

t i

0

0 m

g N

m

o g

m o >

f f

- /

I / /

" e ' \ ~ ~

\ e \

\ N

N N

\ e \

\ \ \

\

1 1 I l , i 2 0 6 0 1 0 0

% C in Si02

Figure 59. Volatilization rate Q~:h of silicates (e) SiO2 + 12% (Na + K), without graphite, (o) the same with graphite (from Avni et al., 1972).

nation of trace elements in uranium oxidem2% was used as the carder. In addition to Ga203, which was studied further by Pepper (1967) and Feldman (1966), the application of other carriers has been cited in the literature; e.g. AgC1 (Spitzer and Smith, 1952; Nelms and Vogel, 1966), NaF (Belegisanin, 1953), In203 (Cesarelli and Rossi, 1970), NHnF (Anonymous, 1964), CsF (Whitehead and Heady, 1962; Strzyzewska et al., 1963), and Ag20 (Janda et al., 1963). Various mixtures like AgC1 + AgF (Strzyzewska et al., 1966), AgC1 + PbF (Mykytiuk et al., 1966), Ga203 + SrF 2 (King and Neff, 1962), AgC1 + GeO 2 (Russell, 1968), LiCO 3 + PbF 2 + NaCI (Day et al., 1968), and others (Feldman, 1966; Pepper, 1967) have been used. Evidently, a wide range of carriers are being applied to achieve multielement analysis. These modifiers are widely employed in direct solids analysis using electrothermal vaporization devices (Gregoire, 1988; Karanossios and Horlick, 1990) and direct solids insertion (Brenner et al., 1987; Karanossios, 1989) into inductively coupled plasmas with AES and MS detection.

In addition to the discussion and conclusions in Section II.A and studies by Strzyewska and others (Strzyzewska, 1971a,b, 1972; Boniforti et al., 1972; Strzyzewska and Minczewski, 1972), their effect can be listed as follows:

Trace element volatilization rates are increased due to the higher vapor pressure of the carrier-thermochemical reactions.


Table 22. Mean Residence Time x (103 s) in the Presence of Carriers a

Element Ionization Potential Without Carrier Ag AgCl Ga203

Li 5.39 1 1.1 2.4 1.95

T1 6.11 1.9 2.1 4.6 3.8

Zn 9.39 2.7 - - m 1.6

Hg 10.43 6 - - m 0.9

Note: aAdapted from Lorentz, 1951.

2. They cause a reduction of the plasma temperature and an increase of the

electron density. As a result of plasma temperature depression, the dense

atom and ion spectra of the complex matrix is diminished. 3. They cause a reduction in the degree of ionization (~t) of any element with

low ionization potential (IP ~ 4 eV). 4. The residence time of atomized particles is increased in the arc discharge

(see Table 22) for elements with IP < 8 eV, and as a result intensities are

enhanced (Vukanovic, 1960). 5. The axial and radial distribution of excited particles in the plasma are more

uniform than without a carrier.

In general, the following analytical classification of trace element responses to

carriers can be proposed:

1. Trace elements that are readily affected by the volatilization of the cartier. These include the volatile elements such as lead, cadmium, and tin.

2. Elements that preferentially form a eutectic compound with the cartier followed by volatilization. For example, aluminum, iron, and manganese.

3. Elements whose compounds are involatile at arc temperatures or which form refractory compounds with the matrix. For example, calcium and magne-

sium. 4. Elements which have volatility dissimilar to those of the mixture (carrier-

matrix) and therefore are not affected by the distillation of the carder, e.g.

zirconium, thorium, and the rare earths.

Thus, in the absence of a carrier it can be concluded that the volatilization rate of a solid sample from a graphite cup is affected mainly by the refractory matrix and that the volatilization rates of numerous trace elements are governed by that of the refractory matrix.


E. Development of General Schemes for Multielement Analysis

Introduction In developing a general method for multielement analysis of complex materials,

the principles described in previous sections were applied. As a result of the spatial variations of temperature, mean electric conductivity, and electron density (in many cases cathode region maxima), accompanied by selective diffusion of the third matrix, intensities of the matrix elements and the traces are differentiated. For example, maximum trace element particle concentrations for refractory matrices are observed in the cathode region where matrix-free particle concentration is smallest. This axial "discrimination" between the anode and cathode parameters results in the "separation" between trace and matrix elements in particular in the cathode region. Thus the observation zone can be optimized using this "separation", and a multielement analysis with maximum LODs and minimum spectral interference can be made.

The methods of analysis of refractory matrices treated in this review differ in several aspects: Volatilization rates are unequal, chemical reactions of the third matrix with air and graphite differ one from the other, and additional refractory matrices can be formed during the atomization period. For example, in natural phosphates and silicates, the behaviors of the major elements such as Ca, Mg, A1, Fe and Ti, P, and Si could differ from those of the trace elements. In certain cases a particular major element may act as a matrix carrier for several trace elements, while another as a matrix carrier for other traces. Many refractory matrices such as A1203, MOO3, TiO 2, ThO 2, ZiO 2, U308, PuO 2, and the REE oxides can be conveniently classified together since all have the following characteristics: (1) their volatilization rates are similar; (2) their effects on plasma variables are similar.

Thus if (1) and (2) above are well defined, the use of internal standards may not be necessary. In cases where a small bias occurs, matrix factors can be applied, i.e. trace element standards in any matrix calibration standard can be used for determining trace elements in the other matrices.

Sample Preparation Metal samples were converted into oxides. After homogenization employing a

"Wig-L-Bug" (Spex Industries, Edison, NJ) the samples were weighed into the anode crater (identical crater dimensions for all matrices) and pressed with a venting tool.

Plasma Operating Conditions During arcing (arcing periods for each matrix may differ) only the cathode region

is projected through a diaphragm onto the entrance slit of the spectrograph-spectrometer. The electrode gap is focused onto the diaphragm and onto the collimator. Both the current and the arc gap are kept constant during arcing. Synthetic standards


Table 23. Instrumentation

Spectrograph: Gratings:

DC arc source: Arc and spark stand: Densitometers:

Data acquisition:

Computer: Transient measurements: Photographic emulsions: Optical configuration: Electrodes:

Ebert 3.4 m Hilger and Watts (Rank Hilger, U.K.) 600 grooves/mm blazed for 5.2 ~ (Thermo Jarrell Ash)

1800 grooves/mm blazed for 18.6 ~ (Thermo Jarrel Ash)

3-25 A (Hilger and Watts) Spex Industries, Edison, NJ GII (Zeiss Jena), L459 (Rank Hilger), and scanning system

(Thermo Jarrell Ash) Analog-to-digital converter with a Kennedy tape recorder

(Nuclear Center, Negev, Israel) Control Data 3600 Oscilloscope, type 551 (Textronix): for voltage measurement SA-1; SA-3; N-1 (Kodak); R-50 (Ilford)

See Figure 10 Ultra purity graphite (Ultra Carbon)

are treated in an identical manner. The spectra obtained from the cathode region on the photographic plates are measured densitometrically for trace element content or the intensities can be collected by PMTs or C1Ds. The instrumentation employed in the present work using photographic detection is listed in Table 23.

Calibration Standards

Three different sets of multielement calibration standards were prepared containing about 50 trace elements [Ag, A1, B, Ba, Bi, Ca, Cd, Co, Cr, Cu, Cs, Fe, Ge, Ga, In, K, Li, Mg, Mn, Mo, Na, Nb, Ni, Pb, Rb, Sb, Sc, Si, Sn, Sr, Ti, T1, V, and Zn ("common" trace elements), and REEs] in each refractory matrix. Standard reference materials were also employed for the evaluation of the precision and accuracy where available.

The common trace elements were mixed with each matrix to produce a graded set of standards containing trace elements in a concentration range of 0.1 to 100 mg/kg each. Li, Na, K, Rb, and Cs were mixed as above in a concentration range of 0.1 to 100 mg/kg each. Hg and P were mixed separately with each matrix at the same concentration range. Standards consisting of A1, Ba, Ca, Mo, Nb, Sr, Ti, and REEs were prepared in each matrix. In several matrices, 2-10% of the matrix in the fluoride form was added. The range of contents was the same as that of the common elements.

F. Analysis of Uranium, Thorium, Zirconium, and Plutonium Oxides

Behavior of the Matrices and Theoretical Considerations

In accordance with the fundamental studies described in this paper, the following parameters were measured.

Direct Current Arcs and Plasma Jets "165

Volatilization Rates. Volatilization rates of U308, ThO2, ZrO 2 La203, and Nd203 are listed in Table 4. The addition of fluorinating agents (2-4%) in the form of UF 4, ThF 4, ZrF 4, or "PTFE Spray" resulted in an increase of the volatilization rates and detection limits for refractory trace elements such as the REEs. The results in Table 4 show that U308, ThO2, and ZrO 2 behave in a similar manner.

The Axial Temperature Distributions. The axial temperature distributions, given in Figures 21 and 60 show that the matrices are similar within the limits of the experimental error (+200 K). (Addition of fluoride to the refractory matrix did not significantly affect the temperature.)

Axial Distributions of Total Particle Concentrations. The above matrices are similar with respect to the axial distributions of total particle concentrations (see Table 14). (Particle concentrations were insensitive to fluoride additives.) The transport parameter (~j) of the free particles of the matrices are also similar (see Section II.B).

Axial Distribution of the Electron Density. The axial distributions of the electron density of the matrices (see Figure 32) are similar and they were unaffected by fluorination.

750(~_ T'K

700

650O

i ,.. J

6 0 0 0 ~ x "o'-~ l I - '

v 2 4 6 Cathode Distance m m

Anode

Figure 60. Axial distribution of temperature in a graphite-air dc arc plasma (,). With U308 (A) 6 mm gap, (o) 8 mm gap and with PuO2 (n) matrices (from Avni, 1978).


Normalized Axial Intensity Distributions. The normalized axial intensities (I/IAnode) (Figures 11-14, 61) of the trace elements in the refractory matrices are similar.

Inf luence o f Graphite and Fluorination. We have found that REEs cannot be determined using the routine employed for the determination of the common trace elements due to their refractory nature. The addition of 20% graphite and 4% fluoride enhanced REE sensitivities due to a significant increase of the volatilization rates of the matrix and the REEs and their free particle concentration in the plasma, while the axial distribution of normalized line intensities remained un- changed (Figures 11-14, Avni and Chaput, 1961). On the basis of these observa- tions, the cathode region was selected for multielement REE determination. The similarity of the matrices allowed the application of matrix factors i.e. the use of the standards in U308 for the analysis of ThO 2, ZrO 2, and PuO 2 matrices.

9 Q

7

6

5 41 "D 0 �9 <

t 3 0 ,qm C

3 . "6 2 E t,, 0

2[

I I I / )

/ i / I

- T / I .1. i /

T ~ i t I f I I ~ " 3"

"r ," ; T . , .~1

10 2 3 4 Anode Dislance m m Cafhocle

Figure 61. Axial distribution of relative line intensity normalized to the anode region, PuO2 matrix; 6 mm gap, 13 A. [~ Common elements, TREEs (reproduced from Avni, 1978).


Analytical Procedure

Sample Preparation. Samples of uranium, thorium, zirconium, and plutonium were transformed respectively in U308, ZhO2, ZrO2, and PuO 2 by roasting in air at 900 ~ Plutonium was handled in a glove box (Avni et al., 1970).

Calibration. Details of the operating conditions are listed in Table 24. Expo- sures were made at 35-40 s without a prebum. A seven-step (1:2) rotating sector was employed. A 1800 grooves/mm grating was used for the REEs, while a 600-grooves/mm grating was employed for the common trace elements. Comput- erized self-calibration were employed using Seidel and Kaiser conversion procedures (Kaiser, 1947, 1964; Frenkel et al., 1975). Slopes ofthe working curves [d(log I)/d(log C)] and the coefficients of correlation are given in Table 25 for the U308 matrix. The coefficients of correlation show that reliable working curves were obtained without the use of intemal standards.

Analytical Results

Detection Limits. A list of the spectral lines used together with their detection limits are listed in Tables 26 and 27. Detection limits for the common elements varied from 0.1 to 5 mg/kg. "Cathode region" LODs are also compared with those obtained by "carrier distillation" for the U oxide matrix. The comparison indicates that the LODs of the common trace elements are similar for both routines. However, in the "cathode region", the REE LODs (Table 27) are significantly better than those obtained using a conventional "cartier" method.

Matrix Factors (MFs) in the Cathode Region: Theoretical Considerations. Due to LTE in the cathode region, theoretical and experimental matrix factors were calculated. Line intensity ratios for the same spectral line in two different matrices, M 1 and M 2, were calculated using Eqs. 26 and 27,

Table 24. Analytical Operating Conditions Upper electrode (cathode): Lower electrode (anode):

Electrode gap: Diaphragm: Observation zone: Current: Exposure: Slit: Calibration standards:

Graphite, 3.17 mm diameter, flat Graphite, 6.35 mm diameter Crater, 3.97 mm diameter, depth up to 5.60 mm 4.0-8.0 mm 1.0 mm 0.2-1.0 mm below the cathode (see Figure 10b) 10-13A 35-40 s without preburn 0.035 mm width and up to 5.0 mm height Trace element oxides (Specpure, Johnson-Matthey)


Table 25. Statistics for Calibration Curves for Analysis of U308, ThO2, and ZrO 2 in the Cathode Region a

Analytic Working Curves Samples $1-$10

Slope Value Coefficient of Number of Coefficient of Analytical Line d(log J) ld (log C) Correlation, R b Determinations Variation c

AI (+ F) 0.75 0.95 15

Ba (+ F) 0.65 0.92 16

B 0.80 0.98 17

Ca (+ F) 0.65 0.93 15

Cd 0.4 0.98 14

Cr 0.80 0.99 17

Cu 0.70 0.94 17

Fe 0.78 0.96 17

In 0.82 0.99 15

Mg 0.85 0.94 17

Mn 0.85 0.96 17

Mo (+ F) 0.70 0.98 15

Ni 0.65 0.98 17

Si 0.80 0.95 12

Sn 0.85 0.98 12

Ti (+ F) 0.72 0.98 17

V (+ F) 0.75 0.98 17

Zn 0.55 0.99 15

La (+ C + F) 0.72 0.95 20

Ce (+ C + F) 0.65 0.95 20

Yb (+ C + F) 0.80 0.98 20 Nd (+ C + F) 0.70 0.97 20

15-17

17-20

12-15

17-20

12-13

10-14

15-17

9-12

9-12

15-20

9-15

9-15

8-10

15-20

10-12

8-10 10-13

20-25

20-25

25-30

17-20 18-20

Notes: aAfter Avni, 1978.

N ~ log X log r - ( E log Y) (~-' log X) bg [N E (log X) 2 - log E (log X) 2] [N E (log y)2 _ ~ (log ],,)2]

CMinimum and maximum values.

�9 slope of Y = a + bX.

j j j + 5o40 v r(M/) r(M~)

where V is the excitation potential in eV of the spectral line, and nj the particle concentration of the trace element under consideration. The ratio nj(M1)/nj(M2) was calculated from relative line intensities of the trace element in the two matrices. Partition function ratios Z 2 ( M 2 ) / Z I ( M I ) were corrected for temperature values in


Table 26 . Limits of Detection for the Common Elements in U308, ThO2, ZrO2 and PuO2 in the Cathode Region a'b

Detection Limits (ppm)

Wavelength U308 ZrO 2 ThO 2 PuO 2 U308 Element (,~) U308 ZrO 2 ThO 2 PuO 2 + F + F + F + F C.D. c

Ag 3280 0.5 0.1 0.2 1.0 3382 0.5 0.1 0.2

A1 3082 . . . . 0.2 0.5 3092 . . . . 0.2 0.5

B 2496 0.1 0.1 0.1 0.1 Ba 4554 . . . . 0.5 1.0 Bi 3067 0.1 0.5 0.5 0.5 Ca 4226 . . . . 0.5 1.0 Cd 2288 0.1 0.2 0.2

3261 0.5 1.0 1.0 2.0 Co 2432 1.0 1.0 1.0 2.0 Cr 3021 1.0 0.5 0.2 1.0

4254 1.0 m - - Cu 3247 0.1 0.1 0.1 0.5 Fe 2843 1.0 1.0 1.0 m

3021 0.5 0.1 0.1 0.5 Ga 2943 0.5 0.3 0.1 1.0

2944 2.0 0.8 0.5 Ge 2651 0.2 0.2 0.2

3039 0.2 0.2 0.2 1.0 In 3039 0.5 0.2 0.2 0.5

3256 0.3 0.1 0.1 Li 3232 0.5 0.2 0.2 0.5 Mg 2795 0.2 0.1 0.1 1.0 Mn 2798 0.1 0.1 0.1 0.5 Mo 3132 . . . . 0.5 2.0

Na 5889 0.1 0.1 0.1 0.1 Nb 4058 . . . . 5.0 10.0 Ni 3050 0.5 0.5 0.5 1.0

3002 0.5 0.5 0.5 Pb 2801 1.0 0.5 0.5 1.0

2833 0.5 0.3 0.3 Sb 2598 2.0 1.0 0.5 2.0 Si 2516 0.5 0.5 0.5 2.0

2881 0.5 0.5 0.5 Sr 4607 . . . . 0.5 1.0 Ti 3349 . . . . 0.5 1.0

3234 . . . . 0.5 1.0 TI 2767 1.0 1.0 1.0 1.0 V 3182 . . . . 0.5 0.5 Zn 3282 5.0 3.0 5.0 2.0

3345 5.0 3.0 5.0 K 7664 0.05 0.1 0.1 2.0 Rb 7800 0.05 0.5 0.5 5.0 Cs 8251 0.5 0.5 0.5

0.5

0.1

0.5 1.0 1.0 0.5

1.0 5.0

1.0 2.0

2.0 2.0

10.0 20.0

1.0 5.0 1.0 1.0 1.0 m

<1.0

0.1 1.0 0.3 1.0 0.1

1.0 1.0

0.1 1.0

1.0

0.3

0.3 1.0 0.1 1.0 0.1

1.0 4.0 1.0

0.5

1.0 1.0

1.0 1.0

1.0 0.5

10.0

1.0 10.0

1.0

Notes: aData from Avni, 1978. b4% (w/w) F was added. CCarfier distillation.


Table 27. Limits of Detection for the REEs in U308, ThO2, ZrO2 and PuO2 in the Cathode Region a'b

Element

Detection Limits (ppm)

U308 ZrO 2 ThO 2 PuO 2 Wavelength + + + +

(t~) UF 4 + C ZrF 4 + C ThF 4 + C Teflon Spray + C

Ce

Dy

Er

Eu

Gd

Ho

La

Lu

Nd

Pr

Sm

Tb

T m

Yb

Y

Sc

Zr

Th

3716.37 10 10 20 m

4222.60 20 - - - -

4289.91 - - 15 - -

3385.03 5 10 ~ m

4000.45 5 10 10 5

3372.75 5

3692.64 10 10

4007.94 5 10 10 50

3907.11 5 5

3971.96 5 5 5

3930.50 ~ 5 5 5

3362.23 5 ~ ~

3422.47 20 10 20 10

3398.98 15 10 ~ 50

3456.00 15 20 20

3337.49 5 10 5 5

3995.75 - - 10 10

3949.11 10 ~ 10

2615.42 5 5 10 5

4012.25 15 20 50

4247.38 20 10 - -

4303.58 20 20 ~

4222.98 100 100 100

4408.84 100 100 100

4256.39 10 10

4280.79 10 10 10 50

4424.34 ~ 10 ~

3509.17 15 ~ ~ 100

3650.40 10 ~ - -

3702.85 ~ 20 20

3291.00 10 5 ~

3362.61 5 5 ~ 5

3462.20 10 5 5

3289.35 1 1 1 1

3694.20 5 1 ~

3987.99 ~ 1 ~

3633.12 ~ 2 5

4374.94 2 2 5 2

3353.73 1 1 2

3630.75 1 1 2 10

3438.23 1 ~ 5

4019.13 10 10 - -

Notes: aData from Avni, 1978. b4% (W/W) F was added.


two matrices, i.e. T(M2) and T(M1) in the cathode region. In this way the theoretical matrix factor (MF) is expressed by the calculated ratio J(MI)/J(M2) (Eq. 52).

Experimental MFs were obtained by measuring J(M1) and J(M2) densitometrically and dividing one by the other. Table 28 lists the MFs for U308-ZrO 2, U308-ThO 2, and U308-PuO2; for several trace elements. These MFs were then applied for trace element determinations in each matrix. However, the observation region for the determination of As, Hg, and P was different and these elements were determined separately in the central region of the dc arc plasma. No matrix factors were applied.

Analysis Using Matrix Factors (MFs). Ten samples of U308, ThO 2, and ZrO 2

(S 1-S 10) were analyzed for their trace element contents. Two of the oxides were prepared from ultrapure nitrates (S 1-$2), the others were prepared from the metals ($3-$7). $8-S 10 were received in the oxide form. A sample of certified PuO 2 was also analyzed. Each trace element was determined 15 to 20 times. Analytical data were obtained by applying the following MFs (Table 28):

�9 U 3 O g - Z r O 2 mixtures--(MF)l = 1.0 for the common trace elements and 1.7 for the REEs.

�9 UaOs-ThOEm(MF)2 - 1.5 for the common trace elements and 2.2 for REEs. �9 UaO8-PuOEmIndividual values (MF) 3 for each common trace element'.

Table 28. Calculated and Experimental Matrix Factors for the Analysis of U308, ThO2, ZrO2, and PuO2 in the Cathode Region a

Calculated Factor Experimental Factor c Impuri~ Element b ju/Jzr jU/JTh ju/JPu ju/Jzr jU/JTb ju/Jeu

Pb 1.0 1.52 0.86 1.10 1.70 0.77

Cr 1.0 1.45 1.49 1.05 1.55 1.85

Sn 1.0 1.42 1.04 1.08 1.35 0.94

Ge 1.0 1.37 1.33 0.88 1.53 1.19

Fe 1.0 1.37 1.23 0.92 1.50 1.11

Ni 1.0 1.35 1.40 1.15 1.50 1.55

Bi 1.0 1.34 1.10 1.08 1.55 0.99

Si 1.0 1.45 6.38 1.05 1.37 5.96

Ca 1.0 1.67 - - 0.85 1.60

Zn 1.0 1.97 2.40 1.25 1.42 3.15

Mn 1.0 1.42 1.26 0.95 1.39 1.19

Ba 1.0 1.21 1.0 1.10 1.30 0.94

Sr 1.0 1.21 1.0 0.91 1.35 0.79

Notes: aData from Avni, 1978. bFor wavelengths see Table 26. eRSDs varied from 2-25%.

1 72 R. AVNI and I.B. BRENNER

Calibration curves and percent RSDs are listed in Table 25. Using mean values for the various MF's without background correction, the percent RSDs were about 15% for the concentration range of 5-500 mg/kg in the three matrices, with the exception of PuO 2.

G. Rare Earth (RE) Oxides

Behavior of Matrices and Theoretical Considerations

Volatilization Rate. QRE values for the matrices are listed in Table 29 (measured using the "wire" and chemical methods). The data show that addition of 20% graphite and 4% fluoride increased the volatilization rate of the REE by a factor of 2.

Axial Distribution of Temperature. The distribution of the axial temperature gradients for the REEs indicate that two distinct groups exist: (1) Y203-Sm203- La203, and (2) the remainder of the REEs. The temperatures observed in both groups attained a maximum value in the cathode region (see Figure 22). The addition of 20% graphite and 4% fluoride reduced the differences between the two REE groups within the experimental error. However a small maximum axial temperature occurred between the central and cathode regions of the plasma.

Axial Distribution of Voltage. Figure 62 illustrates the axial distribution of voltage measured by the "wire" method. Two different axial electric field strengths, E z, were observed: 120 V cm -l in the anode region and 13 V cm -l in the central and

Table 29. Rates of Volatilization of the REEs with and without Fluoride and Graphite a'b ach x 1 O- 17 (atoms s- l)c

Matrix Mixture Rare Earth Alone

QW x 10 -17

(atoms s -l )d Rare Earth Alone

La203 ~ 1.9 2.0

La203 + C + F 8.0 4.5 3.7

Sm203 ~ 1.5

Sm20 3 + C + F 7.2 4.0 m

CeO2 - - 2.3

CeO 2 + C + F 7.0 4.3

Nd20 3 m 2.2 2.5

Nd20 3 + C + F 9.0 5.0 4.0

Notes: aData after Avni, 1978.

b4% F + 20% graphite.

CRSD = 30%.

dRSD = 25%.


cathode regions. E z values were independent of the REE matrix within the experimental error. Furthermore, E z values for all the REE matrices, with and without 20% graphite and 4% fluoride, were similar.

A x i a l D i s t r i b u t i o n o f E l e c t r o n D e n s i t y . Axial n e values are illustrated in Fig- ure 33. The electron densities of Y, Sm, and La oxides (without graphite and fluoride) were equal. The n e values for the remainder of the REE oxides were also similar. An electron density maximum occurred in the cathode region in both groups. When 20% graphite and 4% fluoride were added to both groups their n e

values were similar.

Axial Distribution of Total Particle Concentration. The n t (obtained using the "wire" method) and nj values are listed in Table 30 for the cathode region. The values for all the REE matrices were similar in each REE group without additives. Addition of 20% graphite and 4% fluoride resulted in a twofold increase of the particle concentrations and the values for both groups were similar.

Axial Distribution of Normalized Line Intensities. Normalized line intensities of the common and RE trace elements in the REE matrices are similar (see

50

40

e = 30 I=

2 0

m 0

>

1 5

j f ~ f

- ,1 ! 1

I // /

/ / /

/ / /

/ /

I 1 1 , 1 I 3 5 7

Cathode Anode Distance mm

Figure 62. Axial distribution of voltage in plasmas with REE oxides; 6 mm gap, 13 A (from Avni, 1978).


Table 30. REE Matrix Particle Concentration in the Cathode Region a

Matrix n I x 10 -13 b nj x 10 -13 c

(cm -3) (cm -3) (n t nj)/nj

La203 270 4.0 66.0

La203 + 20% C + 4% LaF 3 420 10.0 41.0

Sm203 - - 7.0

Sm203 + 20% C + 4% LaF 3 - - 12.0

Nd203 100 1.4 70.0

Nd203 + 20% C + 4% NdF 3 320 9.0 35.0

CeO 2 - - 0.7

CeO 2 + 20% + 4% NdF 3 w 8.0

Eu203 - - 0.9

Eu203 + 20% C + 4% NdF 3 - - 8.5


bWire method; mean for four wires, RSD = 30%.

CMean for four spectra, RSD = 20%.

Figure 12) regardless of the presence of graphite and fluoride, and the maximum line intensity for each trace element was observed in the cathode region of the plasma.


Sample Preparation. REE compounds or metals were transformed into their oxides by roasting at 900 ~ for 2 h. After homogenization, 30 or 50 mg of the oxide were introduced into the anode crater. The material was pressed into the cup using a venting tool, the levels being constant and equal for all REE matrices

(La203, Y203, Sm203, Nd203, Ce203, DY203, Eu203, Gd203, and S c 2 0 3 ) . The analysis was divided into two groups: (1) determination of common trace elements in REE matrices without additives; and (2) determination of REEs in REE matrices~20% graphite and 4% fluoride was added.

Analytical Results As in the case of U308, ThO 2 and ZrO 2 (Section III.F), common and REE trace

elements in REE matrices were determined separately. The addition of 20% graphite and 4% fluoride improved REE LODs in the REE matrices. The observa- tions described in Section III.G indicate that the cathode region was optimum for trace multielement determination.

Matrix Factors (MFs) in the Cathode Region. Equation 52 was used to calculate matrix factors. The experimental line intensity ratios (J( M) /J(M)) for the same spectral line of the same element in two different matrices were used to calculate


Table 31. Matrix Factors for Trace Elements in REE Oxides in the Cathode Region a

Group 1 Group 2 Trace

Element JLa/Jy JLa/Js m JNd/Jc e JNd/Jp r JNd/JE u JNd/Jy b

Ba 1.10 1.30 1.15 0.70 1.12 1.30

Cr 0.90 1.20 1.30 0.85 1.20 1.50

Ga 1.05 1.25 1.20 0.80 1.15 1.45

Fe 0.90 1.20 1.20 0.85 1.25 1.50

Ni 0.95 0.10 1.15 0.90 1.20 1.60

Bi 1.0 1.20 1.15 0.80 1.25 1.35

Si 1.05 1.25 1.20 0.95 1.20 1.55

Mn 0.90 1.20 1.20 0.8 1.15 1.50

Averages from all 0.9 1.20 1.20 0.8 1.20 1.50 trace elements +_20% +_15% +_25% +_15% +_20% _+17%

Note: aReproduced from Avni, 1978.

the matrix factor. Table 31 lists MFs for common trace elements in each matrix

relative to La203 and Nd203. Table 32 records the MFs for several REEs in REE matrices relative to La203. The average values given in Tables 31 and 32 were used to determine the trace element contents in each REE matrix. Thus La203- and NdEOa-based calibration standards were employed for the determination of the common trace elements in each REE oxide group. LaEOa-based standards were used for the determination of rare earth traces in all REE matrices.

Detection Limits. The spectral lines and detection limits of in the REE matrices are given in Table 33. Detection limits were determined as described in the previous section and amount to several mg/kg.

Calibration. The cathode region was used for the determination ofthecommon trace elements. A preexposure of 5 s followed by 35 s was employed. The analytical observation zone for the determination of the REEs was located 0.8 mm below the cathode tip. A preexposure of 25 s followed by an integration period of 35 s was used. The slopes d(log J)/d(log C), of the working curves within the concentration

range of 5-5000 mg/kg are given in Table 34 for La203 and Nd203 matrices. The coefficients of correlation indicate that the calibrations were reliable despite the fact that intemal standards were not applied.

Statistics Using Matrix Factors. The accuracy was determined by analyzing

samples obtained from the Institute of Enerhia Nuclear, Sao Paulo, Brasil (B) and an in-house material (S). Table 34 lists the percent RSDs (10-20%) for the


Table 32. Experimental Matrix Factors for Some Rare Earth Trace Elements in Rare Earth Matrices in the Cathode Region a'b

Rare Earth Matrix Factor Trace Ele- Wavelength ment (t{) JLa/Jy J L a / J N d J L a / J s m J L a / J E u JLa/Jce

T m 3425 1.4 1.3 1.4 1.0 1.0

3362 1.6 1.3 1.5 - -

E r 3312 0 .90 w 0 .82 J 0 .92

3372 - - 0.65 - - 0 .88

G d 3362 ~ ~ - - 1.1 0 .85

3 3 5 0 0 .73 - - 0 .82 ~

H o 3456 1.08 1.05 1.2 1.10 1.0

D y 3407 - - 1.09 - - 0 .92 0 .76

3385 1.02 w 0 .96 ~


t'6.0-mm arc gap at 13 A; trace concentration 500 ppm. Grating used: 1800 grooves/mm; two orders. 20% graphite and 4% fluorides were added to rare earth matrices. Relative standard deviation from four spectra, 20%.

determination of the common trace elements based on 10 determinations and applying the appropriate MFs.

H. Rock Phosphate

Rock phosphates differ from refractory oxide matrices described above in that they their bulk chemical composition consists mainly of Ca and P (Table 35).

Behavior and Theoretical Considerations

In developing a multielement procedure we evaluated the influence of the major elements on volatilization rates and plasma parameters. The variable content of the major constituents could significantly effect the plasma parameters, and consequently influence trace element behavior. The axial distributions of temperature and electron density are illustrated in Figures 25 and 63, respectively. These plasma variables for calcium metaphosphate were similar to those observed for natural phosphates when 20% graphite was added to both matrices. The presence of graphite in both matrices resulted in similar volatilization rates (see Table 21) and plasma variables (Figures 14, 25, and 63). An increase in the volatilization rate of the matrix resulted in an increase of the trace element particle concentration in the plasma (Table 21). Figure 14 also shows the normalized axial intensity distributions of Ca, P, Na, and trace elements. It can be observed that Ca behaves as a third matrix element. [The axial intensity distributions of all the trace elements are similar to that of calcium (Figure 14).] It can be assumed that the Ca molecule particles


Table 33. Detection Limits for the Analysis of REEs in the Cathode Region a'b

Element and Rare Earth Matrices Wavelength c

(/~) La203 CeO 2 Y b 2 0 3 Nd203 Sm203 G d 2 0 3 Sc203 Y203

Yb 3289 1 1

Tm 3425 2 2 2

Tm 3362 w ~ D

Er 3312 ~ 10

Er 3372 5 m 5

Gd 3422 ~ ~

Gd 3350 5 20 5

La 3337 ~ 10 5

Dy 3407 - - 25

Dy 3393 10 ~ 5

Ho 3456 5 10 5

Nd 4012 5 5 5

Eu 3971 2 ~ 2

Eu 3930 - - 10 m

Sm 3634 3 5 3

Ag 3280 0.2 0.2 0.2

A1 3082 2 2 2

B 2496 0.5 0.5 0.5

Ba 4554 20 20 20

Bi 3067 1 1 1

Cd 2288 1 1 1

Co 2432 1 1 1

Cr 2835 1 1 1

Cu 3247 2 2 2

Fe 2843 3 3 3

Go 2943 1 1 1

Ge 2651 1 1 1

In 3256 0.5 0.5 0.5

Mg 2779 1 1 1

Mn 2798 0.5 0.5 0.5

Mo 3132 5 5 5

Ni 3050 5 5 5

Pb 2833 1 1 1

Si 2576 2 2 2

Sr 4607 20 20 20

Ti 3234 10 10 10

V 3184 10 10 10

1 1 2 2 2

2 m __ 5 - -

2 1 ~ 2

10 m 10 10

5 - - 5 - -

10 10 2 10

10 . . . .

5 10 10 2 10

10 10 10 10

10 . . . .

5 5 5 10 10

5 5 10 10

2 2 ~ 2 10

~ 5 ~

5 ~ 5 2 5

0.2 0.2 0.2 0.2 0.2

2 2 2 2 2

0.5 0.5 0.5 0.5 0.5

20 20 20 20 20

1 1 1 1 1

1 1 1 1 1

1 1 1 1 1

1 1 1 1 1

2 2 2 2 2

3 3 3 3 3

1 1 1 1 1

1 1 1 1 1

0.5 0.5 0.5 0.5 0.5

1 1 1 1 1

0.5 0.5 0.5 0.5 0.5

5 5 5 5 5

5 5 5 5 5

1 1 1 1 1

2 2 2 2 2

20 20 20 20 20

10 10 10 10 10

10 10 10 10 ' 10

Notes: aReproduced from Avni, 1978. bData in mg/kg.

CFor REE determinations, 1800 gr/mm grating in two orders was used. 4% fluoride was added.


Table 34. Calibration Statistics for Analysis of REE Matrices Using Matrix Factors in the Cathode Region a

Working Curves b Number of Determinations

Coefficient of Coefficient of Element Slope Value Correlation, R c B Sample S Sample Variation (%)

AI 0.81 0.88 m 10 15

B 0.83 0.96 6 6 10 Ba 0.70 0.92 10 10 20

Co 0.74 0.92 10 w 12

Cu 0.79 0.90 10 4 10

Cr 0.69 0.95 8 8 15

Fe 0.76 0.96 8 8 10

Mg 0.82 0.99 5 5 10 Mn 0.80 0.99 10 w 8

Mo 0.68 0.92 10 4 15 Ni 0.72 0.98 12 - - 15 Pb 0.83 0.98 5 10 10 Si 0.70 0.94 m 15 16

Sr 0.68 0.95 10 - - 20 Ti 0.73 0.90 5 5 20

V 0.69 0.96 5 5 20

Notes: aReproduced from Avni, 1978. bConcentration range: 5-500 mg/kg. CSee Table 25 for equation.

supplying atoms and ions in the plasma, consist of CaO and/or C a 2 P 2 0 7 since both are relatively stable compounds. The volatilization rates of Ca(OH) 2, CaCO 3, and Ca(PO3) 2 (metaphosphate) are compared with that of the natural rock phosphate. Q~ values are measured by weight loss after an arcing period of 30 s (Table 2 I), The data indicate that the volatilization rate of calcium metaphosphate is similar to that of the natural samples, i.e. Ca(PO3) 2 can be used as a calibration matrix. Thus the cathode region of the plasma was the optimum region of observation for quantitative trace element determinations. With the exception of As, Hg, and B,

Table 35. Average Major and Minor Element Composition of Israeli Phosphorites

Constituent Oron Site (%) Arad Site (%)

P205 14-35 22-33

CaO 35-55 45-53

F 2 -4 2.5-3.5

SO 4 2-4 3-6.5

NaC1 1-2 1-2


F~

Q 4~ c !

- I ~ v l C 0

C 0

U

�9 2 u l id

I I., I ! A n o d e 1 2 3 4 C a t h o d e

D i s t a n c e m m

Figure 63. Axial distribution of electron density (ne) in plasmas containing phosphate. (A) synthetic Ca3(PO3)2, (11) phosphorite, (e) synthetic and natural phosphate + 20% graphite; 4 mm gap, 10 A.

maximum sensitivities were observed in the cathode region (0.5-0.8 mm below the cathode) (Figure 14).


Standard and Sample Preparation. Four ml of pure H3PO 4 were added to 30 g of CaCO 3 (Johnson Matthey Specpure) (Rautschke, 1968). After drying at 110 ~ for 2 h, calcium dihydrogen phosphate formed which was then converted to Ca(PO3)2 by heating at 400 ~ for 2 h. A mixture of 20% graphite and 80% calcium metaphosphate was homogenized and two sets of calibration standards were prepared: (1) rare earths, Sc, Y, Th, U, and Zr in concentrations varying from 1-500 mg/kg; and (2) the common trace elements (Table 36) varying 1-500 mg/kg for each element.

A phosphorite sample weighing 0.5 g (grain size < 20 mm) was treated with 0.5 ml H3PO 4. The sample was dried at 110 ~ and then at 400 ~ and mixed with 20% graphite. The anode crater charge was 50 mg for the standards and the samples. The operating conditions were similar to those described by Avni and Boukobza (1969).

Calibration. Working curves were prepared as described previously (page 164) with the exception of a 30 s integration period and without a prebum. The slopes of the working curves within the concentration range can be classified according to the behavior of the trace elements. Slope values and percent RSDs are given in Table 36.


Table 36, Calibration Statistics for Trace Element Determinations in Phosphates in the Cathode Region a

Working Curves 100 Samples

Element b

Slope Value Number of

With Ls.b Without l.S.b R Determinations Coefficient of Variation (%)

A1

Ba

Bi

Cd

Co

Cr

Cu

Fe

Ga

In

Mg

Mn

Mo

Ni

Pb

Sc

Sr

Ti

V

Y

Ce

La

Yb

Nd

Sm

Eu

Gd

Ho

Lu

U

0.74 0.75 0.95 5 15

0.53 0.52 0.97 10 20

0.90 0.89 0.98 5 10

0.87 0.88 0.99 5 10

0.42 0.42 0.98 5 15

0.75 0.73 0.96 15 10

0.84 0.84 0.98 15 9

0.43 0.42 0.98 15 12

0.85 0.86 0.95 5 10

0.75 0.75 0.88 5 13

0.89 0.88 0.96 10 18

0.80 0.82 0.95 10 8

0.70 0.70 0.95 10 12

0.69 0.68 0.94 10 10

0.74 0.75 0.98 20 8

0.75 0.75 0.92 10 10

0.53 0.53 0.95 10 20

0.73 0.72 0.97 10 15

0.73 0.71 0.92 15 15

0.80 0.82 0.90 20 15

0.50 0.51 0.96 5 12

0.55 0.54 0.98 15 13

0.49 0.48 9.95 5 18

0.43 0.45 0.95 5 17

0.41 0.40 0.92 5 17

0.59 0.57 0.93 5 12

0.59 0.59 0.95 5 15

0.76 0.75 0.88 5 20

0.60 0.59 0.95 5 18

0.50 0.51 0.96 20 20

Notes: aReproduced from Avni, 1978. bl.S.: With internal standard.


Analytical Results: "Cathode Region"

Detection Limits. The spectral lines used for the determination of the trace elements together with their detection limits are given in Table 37.

Accuracy and Precision. A comparison between calibrations obtained with and without Pd as an internal standard indicated that satisfactory results can be obtained without the use of this internal standard (see Table 36). Approximately 300 natural samples were analyzed using the "cathode region" procedure. Several in-house phosphorite samples with varying P205 content were analyzed 10 times in order to assess analytical precision which is given as percent RSD in Table 37. For trace element concentrations exceeding 500 mg/kg, samples were diluted with a mixture of Ca(PO3)2:graphite and analyzed as described previously.

The results in Table 37 show that the "cathode region" is an accurate and reliable method for quantitative trace elements analysis of rock phosphates. Analytical data for standard reference materials (NIST 120 a and b, BCR 32; Gladney et al., 1987; Govindaraju, 1994) in Table 38, indicate that this region can be employed for the accurate determination of trace and minor elements in phosphate rocks.

I. Multielement Analysis of Silicate Rocks

The bulk composition (SiO 2, Ca, A1, Mg, Fe, Ti, alkalis, etc.) of silicate rocks and minerals varies widely. Matrix match calibrations can be conveniently made using standard reference materials (Gladney et al., 1987; Govindaraju, 1994) that are representative of the large compositional variations found in nature. Alterna- tively, synthetic standards can be used and various matrix modifiers ("buffers" or "fluxes") such as lithium metaborate, persulphates, and others (Ahrens and Taylor, 1961) can be added to regulate the volatilization rates of the matrix and the trace elements. Internal standards (Ahrens and Taylor, 1961) can also be added to compensate for these differences. Trace element volatilization has been divided into several groups and internal standards selected accordingly.

The application of the matrix match approach to multielement analysis of silicates depends on the availability of a wide range of SRMs. Table 39 lists the compositions of several silicate rocks indicating that the major components vary widely. Therefore, in order to establish a general method for the spectrochemical analysis of silicates and related materials, we determined the volatilization rates and axial distribution of plasma variables with the aim of localizing a region in the plasma where the influence of the major components on the trace elements is minimum, thus facilitating the development of a general quantitative spectrochemical method. The discussion in the forthcoming sections are based on work by Avni et al. (1972) and Brenner and co-workers (1975, 1976, 1987).


Element

Ag

A1

Ba Bi

Cd Co Cr Cs

Cu

Fe

Ga Ge

In K

Li Mg Mn Mo

Na

Ni Pb Rb Sb

Sc

Sn Sr

Ti T1

V Y

Zn Be Ce Dy Eu Er

Table 37. Detection Limits in Phosphorites a

Wavelength Detection Limits (,~) (ppm)

3280.68 0.1

3082.16 0.5 3092.71

4554.03 2 3067.72 0.5

2288.02 0.5 3453.50 2 3021.56 1 8251.10 2

3247.54 0.1

3273.96

3020.49 3020.64 1 3021.07 2943.64 0.1 3039.06 0.2

3039.36 0.1 7664.91 0.5 6707.84 0.1 2779.83 1 2794.82 1 3132.59 0.5

3302.32 2 3050.82 1 2802.00 0.2 7800.23 0.5 2598.05 1

3353.73 0.5 3372.15

2839.99 1 4607.33 2

3234.52 2 2767.87 0.2

3181.41 0.5 3216.69 1

3242.28

3302.59 10 2348.61 0.5 3716.37 5 4045.99 5 3907.10 5 3007.97 1

(continued)


Table 37, Continued

Element Wavelength ( ,~) Detection Limits (ppm)

Gd 3422.47 1 Ho 3456.00 5

La 3337.49 2 Lu 2615.42 2 Nd 4303.58 10 Pr 4408.84 100

Sm 3621.23 5 Tb 3703.92 10 Th 4019.13 50 Tm 3462.20 1

U 3890.36 25

Zr 3391.98 0.5

3438.23 12

As b 2780.22 10

2860.44

B b 2496.78 1

2497.73

Hg b 2536.52 10

Notes: aFrom Avni, 1978. bin the central region.

Table 38, Compar ison of "Cathode Region" (CR) and R e c o m m e n d e d (RV) Trace

Element Contents in Standard Reference Phosphates a

NIST 120 a NIST 120 b BCR-32

CR RV b CR RV b

A1 4800 5000 5600 2680 2900

Ba 68 61 Cd 10 12 25 18 Ce 122 115 Cr 50 60 245 260

Cu 10 Fe 7500 6990 7000 7700 Mg 1300 1600 1550 1680 2500 2400 Mn 180 150 265 248 18 23 Ni 12 17 40 32

Ph 20 24 Sr 685 705 Ti 680 720 850 900 22 19 V 120 170

Notes: aAll data in mg/kg. bRecommended values according to Gladney and co-workers (1987), Govindaraju (1994), and NIST SRM reports.


Matrix Behavior and Theoretical Considerations

Volatilization Rate (Q~h). An important factor which was considered when examining the volatilization of solid silicate material into a plasma is sample vitrification, i.e. mineralogical changes as a result of thermochemical reactions. Formation of glass, a refractory material, was found to hinder volatilization of the sample. Glass formation was eliminated by adding modifiers (e.g. graphite) tothe sample prior to arcing.

Figure 59 shows Q~n values in mg s -l of SiO 2 from a graphite anode cup into the dc arc gap. For pure SiO 2, glass formation in the crater hindered volatilization 0.06 mgs -l. Although Q~h increased to 0.35 mg s -l when 12% (Na + K) was added to the SiO 2 sample, glass formation persisted. This addition resulted in the increase of the values o.f Q~h possibly due to the fusion effect. Samples containing 33% graphite and 12% alkalis were totally consumed after a 100 s arcing period. The volatilization rate of the synthetic SiO 2 standards was identical to that of the natural silicate only after 3 parts graphite were added to both matrices. Thus by eliminating glass formation, the refractory silicate matrix was converted into one which was less refractory.

Plasma Variables. The axial distribution of the normalized line intensity is shown in Figure 64 for the synthetic silicate matrix. As in the case of the other refractory matrices investigated, the maximum line intensities for the trace elements, and minimum intensities for the matrix elements, occur in the cathode region (see Figure 13). Addition of major components such as Ca, Mg, Al, and Fe to the SiO 2 matrix did not significantly alter the normalized line distribution as shown in Figure 64. However, the addition of 5% (Na + K) to the SiO2-based standards suppressed the normalized line intensities in the cathode region for all elements. Only the central region was not influenced by the presence of alkalis as illustrated in Figure 64. The axial distributions of temperature (Figure 24) and electron density

Table 39. Major and Minor Element Contents in Silicate Rocks a'b

G-2 Granite BCR- 1 Basalt PCC- 1 Peridotite

SiO 2 69.14 54.11 41.71

A120 3 15.39 13.64 0.68

TFe20 3 2.66 13.41 8.25

MgO 0.75 3.48 43.43

CaO 1.96 6.95 0.52

Na20 4.08 3.27 0.03

K20 4.48 1.69 0.007

Notes: aAfter Govindaraju (1994). bAll data in w%.


0 "U 0 C

' < 6 -

~ _ ,ira

IA

@

�9 4 - C

�9 ~ -

O N

E L 0 -

Z

Anode Calhode

/ /

/ i /

~ " ~ ' ~ T ! ! , ! l ! " 2r 4.0 6.0

Distance mm

Figure 64. Axial distribution of relative line intensities normalized to the anode region for silicates. ( . . . . ) Major and common trace elements in SiO2 + 5% each Ca, Mg, AI, Fe; ( . . . . . ) Na and common trace elements in SiO2 + 5% (Na + K); ( ~ ) Si in SiO2 + 5% Ca, Mg, AI and Fe (data from Avni et al., 1972).

(see Figure 35) confirm that the refractory SiO 2 matrix was converted when graphite was added in a 3"1 proportion. The data presented in Figures 24, 35, 59, and 64 can be summarized as follows:

1. The central region (the 2-mm central area of a 6-mm gap) of the plasma is the most suitable region for performing multielement determinations of trace elements in silicate rocks and minerals.

2. Dilution of SiO2-based synthetic standards containing 8% Na and K and the natural silicates with 3 to 5 parts graphite resulted in a similarity of parameters of these two matrices.


Calibration. An exposure of 50 s without preburn was used. Emulsions were calibrated using a computerized self-calibration method using Seidel densities (Frankel et al., 1975). Slope values of the analytical working curves and their coefficients of correlation are given in Table 40.

Analytical Results: "Central Region"

Detection Limits. The spectral lines used for the trace and minor element determinations are given in Table 40. The limits were similar to those obtained for phosphates.


Table 40. Calibration Data for the Trace Element Determinations in Silicate Rocks with and without Pd as the Internal Standard

Slope Value a Coefficient o f Correlation, R b Working Range

Analytical Line (ppm) With I.S. c Without LS. c With LS. r Without I.S. c

Cr 2843 5-5000 0.4645 0.4709 0.95 0.95

Cr 4254 5-5000 0.5971 0.5952 0.97 0.97

Mn 2933 10-5000 0.6681 0.6816 0.98 0.98

Mn 4034 50-5000 0.8307 0.8442 0.98 0.99

V 3184 5-1000 0.6209 0.6170 0.98 0.98

V 4379 100-1000 0.7733 0.7486 0.96 0.96

Ti 3242 0.01-3.5% TiO 2 0.5775 0.5815 0.99 0.98

Ti 3990 0.05-3.5% TiO2 0.7793 0.5168 0,99 0.88

Cu 3274 5-1000 0.7344 0.6929 0.98 0.96

Ni 3414 2-3000 0.5869 0.5855 0.98 0.98

Co 3453 2-1000 0.8336 0.8115 0.99 0.98

Pb 2833 5-1000 0.6160 0.6012 0.94 0.94

Sr 4077 5-2000 0.6120 0.6078 0.98 0.98

Sr 4607 20-1500 0.5840 0.7204 0.83 0.95

Ba 4554 5-3000 0.7771 0.7711 0.99 0.99

Notes: aLog Y = b log x + log a.

bSee Table 25 for equation. Cpd as internal standard (I.S.).

Role of the Internal Standard. Only minor improvement in analytical accuracy was obtained with the use of Pd as the internal standard as shown in Table 40. Analysis without internal standards was achieved by matching the plasma and volatilization parameters of synthetic SiO 2 and the silicate samples as discussed previously.

Tables 41 and 42 list the analytical data for a large variety of standard reference silicate materials. The results obtained were based on 5 to 50 determinations. The precision expressed as the percent RSD varied as a function of concentration and the overall accuracy was about 10-15% for all the common trace elements determined.

I. Aluminum and Titanium Oxides

Trace elements in A1 metal and alloys can be determined directly in the solid using spark OES (Methods for Emission Spectrochemical Analysis, Philadelphia, 1963, 149, 199.), X-ray fluorescence, and spark ablation coupled to an inductively coupled plasma (Aziz et al., 1974; Brenner et al., 1995). The refractory oxides of these metals have been analyzed for their trace element contents by dc arc techniques by adding "buffers" and "carriers" (Robert and Lloyd, 1962; Balfour et al., 1966). Alumina powders have been analyzed by slurry nebulization inductively

Table 41. Comparison of Data for Silicate Standard Reference Materialsafb

Sample Cr M n v %KO2 cu Ni co Pb Sr Ba Ga Zr B Cd

AGV-1 15 13

BCR-I 12 co v 16

DTS-1 4300 4200

G-2 26 37

GSP-I 15 13

PCC-1 3000 3090

G- 1 18 22

w- 1 125 120

SCO-I 65

785 728

1400 1350 1060 963 280 270 330 326 860 889 195 270

1250 1300 420

115 121 385 384

12 19 34 37 53 52 35 31 12 16

248 240 117

1.14 1.08 2.22 2.23 0.0094 0.02 0.5 1

0.53 0.71 0.69 0.0085 0.02 0.27 0.26 1 .o 1.07 0.75

68 64 19 22 9 8

12 1 1 38 35 11 10 13 13

125 110 30

14 18 14 15

2400 2330

7.5 6

13 11

2370 2430

4 2

70 78 25

17 40 13 35 37 20 36 18

140 8 132 8

4 32 5 29 7 60 8 52

126 8 112 6

2 50 2.4 49

43 50 8 11 28

675 1300 657 1410 340 770 345 790

1 6 6

519 1900 463 1950 255 1300 247 1360

1 5 0.3 I

240 950 250 1200 190 188 180 180 224 750

20 250 18 227 21 215 22 185

22 260 20 316 24 528 18 544

18 220 18 210 18 115 16 100 14 133 65

(continued)

Table 41. (Continued)

Sample Cr Mn v %7i02 cu Ni co Pb Sr Bu Gu Zr B Cd

SGR-I 34 BR 410

420 VSN 900

820

Fe-mica 80 90

Mg-mica 96 80

SY- 1 50 A 52

20 SY-3 14

b n.d. BCS-267 130

50

170

g SY-2 9

BCS-269 150

297 125 1400 250 1600 240 800 825 700 650

2660 140 2710 135 2000 100 1900 82 3200 88 3100 88 2500 52 2250 2700 55 2500 30 1 200 18 890 1 230 176 230 180

0.27 70 2.67 69 2.62 72 1.12 850 1 .o 800

2.65 4.3 2.55 4 2.15 3.5 1.67 4 0.46 25 0.48 22

0.14 0.13 13 0.13 18 0.20 57 0.10 64 1.31 68 1.47 64

28 235 270 890 786

22 35 85

105 37 37

14 n.d.‘ 26 14.5 79 88

13 53 50

700 734

18 20 24 20 25 18

4.5 n.d.c

7 I

23 23

41 8 8

1100 930

450 495 75 64

105 1 20

316 1 200 1300 1260 820

7 6

26 25

250 286 290 270 250 300

109 120

322 1070 1050 1500 960

150 140

4200 4700 330 282 470 430 370 410

520 660

12 52 30 20 275 26 240

384 710 250 lo00 372

120 885 95 34 30 24 3080 20 2900 29 33 19 43

137 140 206 230

Notest aRecommended data are according to Govindiraju (1994) and others ’All data in mgkg. ‘n.d., not detected.


Table 42. Coefficients of Variation for Trace Element Determination in Silicate Rocks

Sample Cr Mn V TiO 2 Cu Ni Co Pb Sr Ba Ga Zr B

AGV-1 C a 15 9 12 10 12 20 16 15 13 10 14 13

N 40 40 40 40 35 30 30 20 35 35 20 20

BCR-1 C 30 12 5 7 12 20 9 18 10 10 12 14

N 24 28 20 28 26 18 22 14 48 48 26 25

DTS-1 C 9 9 20 16 13 6

N 20 26 3 10 15 25 26 4

G-2 C 20 12 13 10 15 20 30 13 9 11 10 18

N 18 28 30 40 18 18 12 16 28 20 13 20

GSP-1 C 18 8 10 9 9 20 30 15 14 8 13 12

N 18 40 40 26 26 22 10 25 40 30 21 13

PCC-1 C 13 11 15 10 15 10 6

N 22 28 10 5 19 14 18 3

G-1 C 20 12 7 12 15 30 30 14 10 10 12 10

N 9 15 15 12 10 5 6 15 16 12 15 5

W-1 C 14 10 8 6 10 11 10 12 15 12 11

N 24 28 16 25 30 37 37 28 20 17 27

SCO-1 C 14 10 7 6 12 12 24 7 9 8 15.5 10.5 11

N 10 34 26 18 26 26 12 12 12 11 12 11 9

SGR-1 C 16 4 10 10 8 8 12 8 6.5 4.6 20 16 10

N 12 12 22 21 20 22 12 12 12 12 12 8 11

BR-1 C 10.5 8 8 8 12 15 15 15 10 12 10 6

N 50 50 40 50 50 50 50 50 50 50 40 40

VS-N C 12 12 15 5 15 11 11 15 10 10 5 6.5

N 25 25 30 25 25 25 25 25 25 25 25 20

Fe-mica C 8 12 15 11 20 12 15 18 11.0 15 13.5

N 14 14 11 10 9 6 6 7 22 6 4

Mg-mica C 12 11 14 10 13 15 12 9 11 12

N 6 12 11 11 6 6 6 15 15 6

SY-1 C 12 11 8 10 10 18 10 8 12 14 12 16

N 25 40 40 48 35 30 30 30 30 30 25 8

SY-2 C 20 10 13 10 15 12 12 15

N 6 9 7 9 7 ~6 6 4

SY-3 C 14 17 10 16 20 12 10 10 10 10

N 5 16 10 10 15 5 5 5 6 5 5

BCS-267 C 9.5 20 10 11 20 20 8

N 6 6 6 6 6 6 6

BCS-269 C 18 15 16 15 16 15 13 15 8.5 10

N 16 9 8 11 6 7 6 9 10 8

Note: aC = coefficient of variation, N = number of determinations.


coupled plasma AES (Raeymaekers et al., 1988). The direct "cathode-region" was applied for the analysis of A1-Ti alloys and related materials in order to overcome the need to separate Ti and employ tedious dissolution procedures for solution analysis of alumina.

Matrix Behavior and Theoretical Considerations

The development of the "cathode-region" method for the determination of the common trace elements was based on the study of the volatilization rates of the matrices and their mutual influence on dc arc variables.

Volatilization Rate. The volatilization rates for A1203 and TiO 2 matrices were measured chemically (~jjh). The value for the former was erratic and varied within a factor of 3 for the same arcing period. This variation was reduced by heating the charged graphite cups on a hot plate for several minutes before arcing and adding 10% of fluoride. (Al-fluoride and Teflon spray were added respectively to AI203 and TiO 2 samples.) Table 43 shows the improvement of the volatilization rate when fluoride was added, ~jjh increasing by a factor of 2 when 10% fluoride was added.

Based on the results reported in Table 9, it can be noted that the volatilization rates of the trace elements were similar to that of the refractory matrix. A 10-fold increase in the volatilization rate of the matrix enhanced the volatilization rates of the trace elements.

Plasma Variables. Figure 23 shows the axial distribution of temperature (Zn and Cu spectral lines were employed as the thermometric species) in a graphite-air plasma containing A1203 particles. In comparison toanode and central regions, the temperature in the cathode region was maximum. Figure 34 shows the axial distribution of electron density (Mg and ion and atom spectral lines were used) of the plasma with the refractory matrices. Again, maximum n e values were observed in the cathode region. Figure 65 shows the normalized intensities (I/Ianod e) of trace

Table 43. Volatilization Rate of AI203 and TiO 2 with and without Fluoride a'b

Matr ix

Volatilization Rate c Q~jh (mg s - l)

A1203 0.04-0.08

AI203 + 10% A1F 3 0.029-0.10

T~O2 0.035-0.06

TiO 2 + 10% Teflon 0.095-0.1 ,

Notes: aAfter Avni, 1978. b6 mm gap, 13 A, arcing time 40 s. ORange of values, n = 10.


Figure 65. Axial distribution of relative line intensities normalized to the anode region for AI203 and TiO2 with 10% fluoride; 6 mm gap, 13 A. ([-]) Common trace elements in TiO2; �9 Ti; (I) common trace elements in AI203; �9 AI (data after Avni, 1978).

and matrix elements. The maximum intensities were observed in the cathode region for the majority of the trace elements. Based on the similar temperature and electron density pattems for A1203 and TiO 2 matrices in the cathode region, and the normalized intensity distributions (Figure 65), this region was selected for mul- titrace element determinations. However, for B, Ba, Ca, and Sr the maximum line intensities were obtained in the anode region and they were determined in that region (approximately 1 mm above the anode).


Sample Preparation. A1 metal and alloy samples were dissolved in purified HC1 and subsequently transformed to A1203 by roasting at 900 ~ for 2 h. Alumina samples were analyzed without further treatment. Titanium metal or alloys, obtained in chips form, were transformed into TiO 2 by roasting in air at 1100 ~ for 2h.


Table 44. Calibration Data for the Determination of Trace Elements in AI203 and TiO2 Using the Cathode Region a

Working Curves

Detection Limits (ppm) Slope Value Al203 and TiO 2

Coefficient of Correlation, Number of Coefficient

Element b Al203 TiO 2 Al203 TiO 2 R c Determinations of Variation

Ag 0.5 0.5 0.80 0.78 0.95 10 9

A1 - - 1 - - 0.75 0.96 10 15

B 0.5 0.5 0.83 0.82 0.96 10 10

Ba 25 25 0.70 0.65 0.92 10 20

Bi 0.5 0.5 0.72 0.80 0.92 10 10

Cd 0.5 0.5 0.80 0.85 0.98 10 12

Co 0.5 1.0 0.69 0.70 0.95 10 15

Cr 2 5 0.70 0.70 0.94 10 15

Cu 0.5 0.1 0.78 0.75 0.92 10 15

Fe 0.5 2 0.76 0.75 0.96 10 12

Ga 0.1 0.1 0.80 0.75 0.98 6 8

In 0.1 0.1 0.80 0.75 0.98 8 9

Mg 1 2 0.80 0.75 0.95 10 13

Mn 0.5 2 0.72 0.70 0.95 10 12

Mo 1 3 0.65 0.60 0.95 10 15

Nb 50 50 0.52 0.55 0.90 10 20

Ni 3 3 0.58 0.60 0.96 10 15

Pb 1 1 0.80 0.70 0.98 10 10

Sb 5 5 0.50 0.52 0.96 10 10

Si 5 0.5 0.68 0.65 0.98 10 12

Sn 0.5 1 0.72 0.70 0.98 10 8

V 5 5 0.69 0.70 0.96 10 15

Zn 10 10 0.65 0.65 0.95 10 20

Notes: a6 mm gap, 13 A, 40 s exposure. bSee Tables 26 and 37 for wavelengths. CSee Table 25 for equation.

Analytical Results

Detection Limits and Working Curves. Table 44 lists the detection limits of the trace elements in each matrix together with the slope value of the calibration curves. For the common trace elements in AI203 (prepared from synthetic Johnson Matthey Specpure oxides), the cathode region was exposed for 4 s, and for TiO 2 for 35 s. It is evident from the correlation coefficients listed in Table 44 that the calibrations were linear.


Precision and Accuracy. The percent RSDs, based on 10 determinations for each trace element (listed in Table 44) varied from 10 to 20% for contents up to 500 mg/kg.

K. Molybdenum and Tungsten Oxides

The trace element contents in molybdenum and tungsten matrices have been determined directly using dc arc sources (Peterson and Chaney, 1961; Hubbard and Green, 1966; Spano and Green, 1966). MoO 3 and WO 3 matrices have been analyzed by chemical preconcentration and matrix elimination prior to arcing. Tungsten metal has also been analyzed by ICP-AES using a high resolution sequential spectrometer.

Behavior of Matrices

Volatilization Rate (Q~h). The volatilization rates for the metal oxides, measured chemically, produced erratic data due to sublimation of molybdenum and tungsten oxides, and the spread of ~jjh values varied by a factor of 5. The addition of 10% fluoride to the Mo oxides did not result in an improvement of the repeatability of Q~jh. However, the addition of graphite, resulted in a significant improvement, and constant volatilization rates were obtained for 35 s arcing. The minimum concentration of graphite necessary for constant volatilization rates of MoO 3 and WO 3 was 66 and 50%, respectively (Table 45). This table indicates that various amounts of graphite resulted in a decrease in the volatilization rate due to the formation of refractory carbides as discussed in previous sections.

Normalized Line Intensity. The axial distribution of line intensities of the common trace elements was measured only for the following matrix-graphite mixtures: MoO3:C = 1:2 and WO3:C = 1:1. Figure 66 shows the axial distribution of the normalized line intensities in the plasma. Maximum values for almost all common trace elements were obtained in the cathode region. Furthermore, the

Table 45. Volatilization Rates of M o O 3 and W O 3 with Varying Graphite Content a

Ratio WO 3/Graphite Ratio MoO 3/Graphite

1:0.5 1:1 b 1:2 b 1:1 1:2 b 1:3 b

Volatilization rate, 0.70-0.30 c 0.22 0.15 0.80-0.35 c 0.22 Q~h (mg S -1)

0.13

Notes: a6 mm gap, 13 A, 35 s. bn= 10.

CRange, n = 10.


9

o

�9 5

ti

z 1

I I I I l 2 4

I 8

C a t h o d e jlt~no de

D i s l a n c e m m

Figure 66. Axial distribution of relative line intensities of common trace elements, normalized to the anode region for MoO3 and WO3 with graphite.

intensity of the Mo and W spectral lines were low using a 6-mm arc gap and an exposure time of 35 s. This was attributed to the following processes:

1. Molecules of the Mo and W matrices were not volatilized nor atomized. 2. The Mo and W oxides reacted with graphite to form high melting point

carbides (MoC 2692 ~ WC 2870 ~ as described by Rautschke (1967) and Nickel (1968), and as a result the volatilization rates were very low.


Sample Preparation. Mo and W metal in chip and powder form were transformed into their respective oxides by roasting in air at 1100 ~ for 2 h.

Analytical Results

Detection Limits and Working Curves. Table 46 lists the detection limits of the trace elements in both matrices, the concentration range and the slopes of the calibration curves. For common trace elements, the cathode region was observed.


Table 46. Detection Limits and Calibration Data for the Determination of Trace Elements in MoO3 and WO3 a

Element b

Ag

A1

B

Bi

Ca

Cd

Co

Cr

Cu

Fe

Ga

In

Mg

Mn

Ni

Pb

Sb

Si

Sn

Sr

Ti

Zn

Working Curves

Detection Limits (ppm) Slope Value MoO 3 and WO 3

Coefficient of Number of MoO 3 WO 3 MoO 3 W03 Correlation, R c Determinations

1 1 0.88 0.85 0.99 5

2 2 0.75 0.70 0.95 10

3 1 0.83 0.82 0.98 10

1 1 0.72 0.80 0.98 10

10 10 0.60 0.65 0.90 10

0.5 1 0.80 0.82 0.97 10

10 10 0.69 0.70 0.95 10

3 3 0.65 0.68 0.95 10

2 2 0.70 0.70 0.96 10

2 3 0.68 0.70 0.95 10

1 2 0.72 0.75 0.98 5

1 1 0.80 0.82 0.97 5

1 1 0.85 0.85 0.97 7

1 1 0.72 0.75 0.95 7

3 3 0.65 0.70 0.92 12

1 2 0.80 0.75 0.99 10

5 5 0.53 0.55 0.89 10

5 2 0.68 0.65 0.90 10

2 2 0.75 0.80 0.96 10

25 20 0.60 0.62 0.90 10

15 15 0.60 0.55 0.95 10

10 10 0.65 0.65 0.95 10

Coefficient of Variation

10

15

10

10

25

15

15

15

15

17

12

10

15

12

17

10

15

20

10

25

18

25

Notes: a6 mm gap, 13 A, 35 s exposure. bSee Tables 26 and 37 for wavelengths. CSee Table 25 for equation.

Precision. Table 46 also shows the percent RSDs for the trace element determinations based on 10 determination of each element in each sample. The percent RSDs varied from 10 to 20% for the concentration range of 10-1000 mg/kg.

REFERENCES

Ahrens, L.H., Taylor, S.R. Spectrochemical Analysis; 2nd ed." Pergamon Press: London, 1961.

Alkemade, C.T.J., Voorhuis, M.H. Spectrochim. Acta 1958, 12, 394.

Alkemade, C.T.J., Voorhuis, M.H.Z. Anal Chem. 1958, 163, 91.


Alkemade, C.T.J. Thesis, State University Utrecht, Utrecht, The Netherlands, 1956. Alkemade, C.T.J., Herrmann, R. Fundamentals of Analytical Flame Spectroscopy; Hilger: Bristol, 1979. Anonymous, U.K. Atomic Energy Authority Production Group PG Report 464 (s), 1964, p. 85. Armstrong, J.A., Ebdon, L. ICP Info. Newslett. 1990, 15, 439. Atwell, G., Pepper, C.E., Stukenbroecker, G.L. USAEC TID 7568-Part I. 1958, 287. Avni, R. Spectrochim. Acta 1968a, 23B, 597. Avni, R. Spectrochim. Acta 1968b, 23B, 619. Avni, R. In Applied Atomic Spectroscopy; Grove, E.L., Ed.; Plenum Press: New York, 1978, Vol. 1, Chap.

4, pp. 149-150. Avni, R., Chaput, M., Re, C.E.A., 1908 (1961). (French) Translated by ET. Birks, UKAEA Trans 875,

1961. Avni, R. Spectrochim. Acta 1969, 24B, 133. Avni, R., Boukobza, A. Appl. Spectrosc. 1969a, 23, 483. Avni, R., Boukobza, A. Spectrochim. Acta 1969b, 24B, 515. Avni, R., Goldbart, Z. Spectrochim. Acta 1973a, 28B, 189. Avni, R., Goldbart, Z. Spectrochim. Acta 1973b, 28B, 241. Avni, R., Klein, ES. Spectrochim. Acta 1973a, 28B, 319. Avni, R., Klein, ES. Spectrochim. Acta 1973b, 28B, 331. Avni, R., Boukobza, A., Daniel, B. Appl. Spectrosc. 1970, 24, 406. Avni, R., Harel, A., Brenner, I.B. Appl. Spectrosc. 1972, 26, 641. Aziz, A., Broekaert, J.A.C., Laqua, K., Leis, E Spectrochim. Acta 1974, 39B, 1091. Balfour, B.E., Jukes, D., Thornton, K. Appl. Spectrosc. 1966, 20, 168. Baronnet, J.M., Fuachais, P. J. Physique. 1971, 32, C56-58. Bauer, H. Anorg, Z Algem. Chem. 1935, 221,209. Bavinck, H. Math. Centre Amsterdam, Amsterdam, The Netherlands Ref TW 98, 1965. Belegisanin, N. Rec. Travaus Inst. Rech. Struct. Matiere (Belgrade), 1953, 2, 27, 83. Boniforti, R., Cioncia, A., di Giorgio, G. Spectrochim. Acta 1972, 27B, 309. Borasseau, D., Cabannes, E, Chapelle, J. Astron. and Astrophys. 1970, 9, 339. Borowiec, J.A., Boom, A.W., Dillard, A.H., Cresser, M.S., Browner, R.E, Matteson, M.J. AnaL Chem.

1980, 52, 1054. Boumans, EW.J.M. Theory. of Spectrochemical Excitation; Hilger and Watts: London, 1956. Boumans, EW.J.M., Maessen, EJ.M.J. Appl. Spect~osc. 1970, 24, 241. Boumans, EW.J.M., Maessen, EJ.M.J. Spectrochim. Acta 1969a, 24B, 585. Boumans, EW.J.M., Maessen, EJ.M.J. Spectrochim. Acta 1969b, 24B, 611. Boumans, EW.J.M. Proc. 14th C.S.L (Ottawa," 1967) Adam Hilger: London, 1968, p. 23. Boumans, EW.J.M. Analytical Emission Spectroscopy; Grove, E.L., Ed.; Marcel Dekker: New York,

Chap. 6. Boumans, EW.J.M., de Galan, L. Anal. Chem. 1966, 38, 674. Bracewell, R.N. Austral. J. Phys. 1956, 9, 198. Brenner, I.B., Eldad, H., Argov, L., Harel, A., Assous, M. Appl. Spectrosc. 1975, 29, 82, 55. Brenner, I.B., Gleit, L., Harel, A. Appl. Spectrosc. 1976, 30, 56. Brenner, I.B., Lorber, A., Goldbart, Z. Spectrochim. Acta 1987, 42B, 219. Brenner, I.B., Eldad, H., Harel, A., Avni, R. Microchem. J. 1987, 36, 62. Brenner, I.B., Zander, A., Kim, S., Holloway, C. Spectrochim. Acta 1995, 50B, 562. Brill, J. Z. Anal. Chem. 1969, 246, 176. Brinkman, H. Thesis, State University Utrecht (in Dutch), Utrecht, The Netherlands, 1937. Butler, L.R.E, Human, H.G.C., Scott, R.H. Handb. Spectrosc. 1974, 1, 816. Cabannes, E, Chapelle, J., Czernichowski, A., Decrosisette, M., Zamarlik, J. Rev. Int. Hautes Temp. et

Refract. 1970, 7, 7. Cabannes, E, Chapelle, J. J. Phys. 1971, 32, C5b-39.


Cecarelli, C., Rossi, G. Euratoru Rep., 4541 E 1970, 84. Chen, F.E Electric Probes. In Plasma Diagnostic Techniques; Huddlestone, R.H., Leonard, S.L., Eds.;

Academic Press: New York, 1965. Corliss, C.H.J. Res. Nat. Burl Stand. 1962, 66A. Czernichowski, A., Chapelle, J., Cabannes, E C.R. Acad. Sci. Paris 1970, 270, 54. Dahlquist, R.L., Knoll, J.W. Appl. Spectrosc. 1978, 32, 1. Daniel, J.L. USAEC, HW-64299, 1960. Day, G.T., Serin, P.A., Heykoop, K. Anal. Chem. 1968, 40, 805, 93. Decker, R.J., Eve, D.J. Appl. Spectrosc. 1968, 22, 13. Decker, R.J., McFadden, P.A. Spectrochim. Acta 1975, 30B, 1. Decker, R.J. Spectrochim. Acta 1973a, 29B, 51. Decker, R.J., Eve, D.J. Spectrochim. Acta 1970, 25B, 411. Decker, R.J. Spectrochim. Acta 1973b, 28B, 339; de Galan, L. Ph.D. Thesis, Univ. of Amsterdam, Amsterdam, The Netherlands, 1965. Denton, H., Sharp. B.L., West, T.S. Talanta 1975, 22, 379. Dahlquist, R.L., Knoll, J.W. Appl. Spectrosc. 1978, 32, 1. Drawin, H.D. Collision and Transport Cross-Sections. Ref CEA-FG-383 1966. Drawin, H.W., Emard, F. Pure Appl. Chem. 1974, 39, 381. Ebdon, L., Sparkes, S. ICP Info. Newslett. 1985, 10, 797. Ebdon, L., Armstrong, J.A., Fairman, B. ICP Info. Newslett. 1989, 14, 459. Ecker, G. Ergeb. Exakt Naturew. 1961, 33, 1-104. Elenbaas, W. Physica. 1934, 1,673. Elliot, W.G. Am. Lab. 1971, 3, 45. Euler, J. Ann. Physik. 1956, 18(6), 345. Feigenson, M.D., Cam N.J. Chem. Geol. 1986, 51, 19. Feldman, C. U.S. Atomic Energy Comm. ORNL-TM-1950, 1966, p. 80. Finkelenburg, W., Maecker, H. Handbuch der Physik; Springer-Verlag: Berlin, 1956, Vol. 22, p. 254. Fister III, J.C., Olesik, J.W. Spectrochim. Acta 1991, 46B, 869. Flanagan, EJ. Geochim. Cosochim Acta 1969, 28, 447. Fox, R.L. Spectrochim. Acta 1985, 40B, 287. Frenkel, Y., Gil, D., Brenner, I.B. Comput. and Geosc. 1975, 1, 27. Gaydon, A.G., Wolfhard, H.G. Flames.; 3rd ed.; Chapman and Hall: London, 1970. Gerdien, H., Lotz, A. Wiss. VeroffSiemens. 1935, 14, 25. Gerlach, W. Zeit. Ansrg. Allgen. Chem. 1925, 142, 383. Ginsel, L.A. Thesis, State Univ. Utrecht, Utrecht, The Netherlands, 1933. Gladney, S.E., O'Malley, B.T., Roelandts, I., Gills, T.E. NBS Spec. Bull. 1987, 260-111,196. Goldfarb, B.H., Ilina, E.V. Opt. Spectrosc. 1961, 11,243. Goldschmidt, V.M., Peters, CI. Nach. Acad, Wiss. Gottingen Math-Physik. Fachgr. 1931, (iii) and (iv),

257-279. Goldschmidt, V.M., Hefter, D. Nach. Acad. Wiss. Gottingen Math-Physik. Fachgr. 1933, 245-252. Goldschmidt, V.M., Peters, CI. Nach. Acad. Wiss. Gottingen Math-Physik. Fachgr. 1933, 141-166. Golightly, D.W., Harris, J.L. Appl. Spectrosc. 1975, 29, 233. Gordon, W.A., Chapman, G.B. Spectrochim. Acta 1970, 25B, 123. Gordon, W.A. Appl. Spectrosc. 1970, 30, 329. Govindaraju, K. Geostands. Newslett. 1994, 18, 1-58. Govindaraju, K. C.N.R.S. Rep. 1970, 923, 133. Greenfield, S., McGeachin, H. McD., Smith, P.B. Talanta. 1975, 22, 1. Greenfield, S. Anal. Chim. Acta 1976, 84, 67. Gregoire, D.C.J. Anal Atom. Spectrom. 1988, 3, 309. Griem, H.R. Plasma Spectroscopy; McGraw Hill: New York, 1964.


Gustavsson, A.G.T. In Inductively Coupled Plasmas in Atomic Spectromet~, 2nd ed.; Montaser, A., Golightly, D.W., Eds.; VCH Publishers: New York, 1992, Chap. 15, pp. 679-720.

Hartenstein, S.D., Ruzicka, J., Christian, G.D. Anal. Chem. 1985, 57, 21-25. Heller, G. Physica. 1935, 6, 389. Hey, E Z Physik. 1959, 157, 79. Hollander, T. Thesis, State Univ. Utrecht, Utrecht, The Netherlands, 1964. Horvath, Zs., Lasztity, A., Barnes, R.M. In Contributions in Geoanalysis Using Plasma Spectrochem-

istr3,; Brenner, I.B., Ed.; Spectrochim. Acta Revs. 1991, 14, 79-94. Hubbard, G.L., Green, T.E. Anal. Chem. 1966, 38, 428, 159. Huldt, S.L. Thesis, Univ. of Uppsala, Uppsala, The Netherlands, 1948. Ilina, E.V., Goldfarb, V.M. Bull. Acad. Sci., USSR, Phys. 1962, 26, 943. Janda, J., Shausberger, J., Scroll, E. Mikrochim. Acta 1963, 122, 88. Joensuu, D.I., Suhr, N.H. Appl. Spectrosc. 1962, 16, 101. Jurgens, G. Z Physik. 1952, 134, 21. Kaiser, H. Spectrochim. Acta 1947, 3, 40. Kaiser, H. Optik. 1964, 21,309. Karanossios, V., Horlick, G. Spectrochim. Acta 1989, 44B, 1345. Karanossios, V., Horlick, G. Spectrochim. Acta 1990, 45B, 85. Keirs, C.D., Vickers, T.J. Appl. Spectrosc. 1977, 31,273. Keliher, EN. In Contributions in Geoanalysis Using Plasma Spectrochemistry; Brenner, I.B., Ed.;

Spectrochim. Acta Revs. 1991, 14, 24. King, H.G., Neff, C.M. AppL Spectrosc. 1962, 17, 51, 91. Kirkbright, G.E, Snook, R.D. AnaL Chem. 1979, 51, 1938. Kleinman, I., Svoboda, V. Chem. List3'. 1975, 69, 833. Korolev, V.V., Vainstein, E.E.Z. Anal. Khim. 1959, 14, 658. Kranz, E. Proc. 14th C.S.I.; Adam Hilger: London, 1968, p. 697. See also Emissionspektroscopie; Akad.

Verlag: Berlin, 1964, p. 160. Kulagin, I.D., Sorokin, L.M., Dubrovskaya, E.A. Opt. Spectros. 1972, 32, 459. Larson, G.E, Fassel, V.A., Scott, R.H., Kniseley, R.N. AnaL Chem. 1978, 47, 238. Leuches, O. Spectrochim. Acta 1950, 4, 237. Lochte-Holtgreven, W. Temperature, Its Measurement and Control of Science and Indust~. ; Reinhold:

New York, 1955. Lochte-Holtgreven, W. Plasma Diagnostics; North Holland: Amsterdam, 1968, p. 135. Lorentz, R.W.Z. Physik 1951, 129, 327. Maecker, H. Z Physik 1951, 129, 108. Maecker, H. Z Physik 1956a, 112, 32. Maecker, H. Z Physik 1956b, 112, 457. Maecker, H. Z Physik 1959a, 157, 1. Maecker, H. Z. Physik 1959b, 158, 392. Maessen, EJ.M.J. Thesis, Univ. of Amsterdam (in Dutch), Amsterdam, The Netherlands, 1974. Maessen, EJ.M.J., Boumans, P.W.J.M. Spectrochim. Acta 1968, 23B, 739. Maessen, EJ.M.J., Elgersma, J.W., Boumans, EW.J.M. Spectrochim. Acta 1976, 3lB. Malyck, V.D., Sard, M.A. Opt. Spectros. 1964, 16, 303. Mandelshtam, S.L. BulL Acad. Sci., USSR 1962, 26, 850. Mankopff, R., Petters, CI. Z. Physik. 1931, 70, 444. Marchant, P., Veillon, C. AnaL Chim. Acta 1974, 70, 17. Margoshes, M., Scribner, B:E Spectrochim. Acta 1959, 12, 138. Matousek, J.P. Prog. AnaL Atom. Spectrosc. 1981, 4, 247. McLaren, J.W., Mykytiuk, A.P., Willie, S.N., Berman, S.S. Anal. Chem. 1985, 57, 2907-2911.


McLeod, C.M., Routh, M.W., Tikkanen, M.W. In Inductively Coupled Plasmas in Analytical Atomic Spectrometr3,, 2nd ed.; Montaser, A., Golightly, D.W., Eds.; VCH Publishers: New York, 1992, Chap. 16, pp. 721-780.

Mellichamp, J.W., Grove, E.L. In Applied Atomic Spectroscopy; Plenum Press: New York, 1978, Chap. 3.

Mika, J., Torok, T. Analytical Emission Spectroscopy; Crane and Russak & Co.: New York, 1974, Chap. 5.

Miller, M.H., Eastwood, D., Hendrick, M.S. Spectrochim. Acta 1984, 39B, 13. Mitchell, R.L.J. Soc. Chem. India 1940, 59, 210. Mitchell, R.L. Commonwealth Bur. Sci. Tech. Communications No. 44; Macauley Inst. (UK) for Soil

Res; Farnham Royal Bucks: Aberdeen, 1964. Mitchell, R.L. Commonwealth Bur. Sci. Tech. Communications No. 44A; Macauley Inst. (UK)for Soil

Res; Farnham Royal Bucks: Aberdeen, 1964. Mitchell, R.L.J. Soc. Chem. Ind. 1940, 59, 210. Moens, L., Verrept, P., Boonen, S., Vanhaecke, F., Dams, R. Spectrochim. Acta 1995, 50B, 463. Mykytiuk, A., Russel, D.S., Berman, S.S. Talanta. 1966, 13, 175, 90. Nelms, J.R., Vogel, R.S.U.S. At. Energy Comm. MCW 1495. 1966, p. 82. Nestor, D.H., Olsen, H.N. Soc. Ind. Appl. Math Review 1960, 2, 200. Nickel, H., Arens, C., Mazurkiewicz, M. Spectrochim. Acta 1984, 39B, 1431. Nickel, H. Zadkorska, Z. Spectrochim. Acta 1995, 50B, 527. Nickel, H. Spectrochim. Acta 1968, 23B, 323. Nickel, H. Z. Anal. Chem. 1963, 198, 55. Nickel, H. Spectrochim. Acta 1965, 21,2031. Olesik, J.W., Fister III, J.C. Spectrochim. Acta 1991, 46B, 851. Olsen, H.N.J. Quant. S. Rad. Trans. 1963, 3, 305. Orenstein, L.S., Brinkman, H. Physica. 1934, 1,797. Owen, L.E. Appl. Spectrosc. 1961, 15, 150. Pavlovic, B., Vukanovic, V., Ikonomov, N. Spectrochim. Acta 1971a, 26B, 115. Pavlovic, B., Ikonomov, N., Vukanovic, V., Todorovic, M. Spectrochim. Acta 1971b, 26B, 123. Pearce, R.W.B., Gaydon, A.G. The Identification of Molecular Spectra. Pearce, W.J. Optical Measurements of High Temperature; Dickerman, P.J., Ed.; Univ. of Chicago Press,

1961. Pearce, W.J. Conference on Extremely High Temperatures; J. Wiley and Sons: New York, 1958, p. 123. Pepper, C.E. USAEC, NLCO 999, 1967. Peterson, M.J., Chaney, C.L.U.S. Bureau Mines Rep. Invest. 5903. 1961, p. 161. Poirier, C., Ph.D. Thesis, Univ. of Massachusetts, Amherst, Massachusetts, USA, 1979, unpublished

data. Raeymaekers, B., Graule, T., Broekaert, J.A.C., Adams, E, Tschopel, P. Spectrochim. Acta 1988, 43B,

923. Raikhbaum, D., Molych, V.J. Opt. Spectrosc. 1961, 10, 524. Ranson, P., Emard, EZ. Naturforsch. 1973, 28a, 1289. Rautschke, R., Amelung, G., Nada, N., Boumans, P.W.J.M., Maessen, EJ.M.J. Spectrochim. Acta 1975a,

30B, 397. Rautschke, R. Proc. XIV C.S.I.; Debrecen: Hungary, 1967, pp. 487-494. Rautschke, R. Spectrochim. Acta 1968, 23B, 55. Rautschke, R., Dowe, Ch., Isensee, G., Herbeck, J. Z. Anorg. Allg. Chem. 1975b, 413, 127. Rautschke, R. Z. Chem. 1965 5, 146. Rautschke, R. Spectrochim. Acta 1967, 23B, 55. Rautschke, R., Holdefleiss, H. Spectrochim. Acta 1969, 24B, 125. Richter, J. Z. Physik. 1958, 151, 114.


Richter, J. Z. Astrophys. 1961, 51, 177. Rippetoe, W.E., Johnson, E.R., Vickers, T.J. Anal. Chem. 1975, 47, 436. Robert, W.J., Lloyd, C. U.S. Bur. Mines Rep. Invest. 6105, 1962, p. 156. Roelandts, I. Spectrochim. Acta 1990, 45B, 815. Roes, R.L. Thesis, State Univ. Utrecht (in Dutch), Utrecht, The Netherlands, 1962. Rossi, G. Spectrochim. Acta 1960, 16, 25. Russell, G.M.S. Africa NIM (MINTEK) Rep NIM 450. 1968, 40, 805. Russman, H.H. Methodische Untersuchungen on Spektralkohlen; Ringsdorff Publication, Bonn-Bad,

Godesburg, Germany, 1958. Samsonova, Z.N. Opt. Spectrosc. 1961, 12, 466. Schramel, P. Spectrochim. Acta 1988, 43B, 881. Scott, R.O., Mitchell, R.L.J. Soc. Chem. Ind. 1943, 62, 4. Scott, R.O.J. Soc., Chem. Ind. 1945, 64, 189. Scott, R.O., Mitchell. R.L.J. Soc. Chem. Ind. 1945, 65, 291. Scott, R.O., Mitchell, R.L., Purves, D., Voss, R.C. Consultative Committee for Dev. Spectr. Work, Bulletin

2; Macauley Inst. for Soil Res.: Aberdeen, Scotland, 1971. Scribner, B.E, Mullin, H.R.J. Res. Natl. Bur. Std. 1946, 37, 379. Serin, P.A., Ashton, K.H. Appl. Spectros. 1964, 18, 166. Siemenova, O.P., Levchenko, M.A. Opt. Spectrosc. 1962, 13, 347. Sine, M.M., Taylor, W.D., Webber, G.R., Lewis, C.L. Geochim. Cosochim. Acta 1969, 33, 121. Skogerboe, R.K., Butcher, G.B. Spectrochim. Acta 1985, 40B, 1631. Skogerboe, R.K., Urosa, T., Coleman, G.N. Appl. Spectrosc. 1976, 30, 500. Smit, J.A. Thesis, State Univ. Utrecht (in Dutch), Utrecht, The Netherlands, 1951. Spano, E.E, Green, T.E. Anal. Chem. 1966,38, 1341, 158. Sparkes, S., Ebdon, L. ICP Info. Newslett. 1986, 12, 1. Spitzer, E.J., Smith, D.D. Appl. Spectrosc. 1952, 6, 9, 82. Strock, L.W. Nach. Acad. Wiss. Gottingen Math-Physik. Fachgr. 1936a, (iv), 171-204. Strock, L.W. Spectrum Analysis with the Carbon Arc Cathode Laver; Hilger: London, 1936b. Strock, L.W. Appl. Spectrsoc. Selected Reprints and Review Articles, 19711, 1, 1. Strzyzewska, B., Radwan, Z., Minczewski, R. Polish Rep. PAN/IBJ 5222/VIII. 1963, 86. Strzyzewska, B., Radwan, Z., Minczewski, J. Appl. Spectrosc. 1966, 20, 236. Strzyzewska, B. Spectrochim. Acta 1971a, 26B, 51. Strzyzewska, B. Spectrochim. Acta 1971b, 26B, 59. Strzyzewska, B. Spectrochim. Acta 1972, 27B, 227. Strzyzewska, B., Minczewski, R. Spectrochim. Acta 1972, 27B, 227. Styris, D.L., Redfield, D.A. Spectrochim. Acta Revs. 1993, 15, 71-123. Szabo, Z.L. Spectrochim. Acta 1974, 29B, 231 and references therein. Todorovic, P.S., Vukanovic, V.M., Simic, M.M., Peric, M.N. Spectrochim. Acta 1975, 31B, 103. Vainstein, E.E., Belayev, Y. Int. J. Appl. Rad. Isotopes 1959, 4, 179. Vainstein, E.E. Proc. IX C.S.1. Paris, G.A.M.S., Ed., 1961, I, 105. Von Engelhardt, W. Chem. Erde. 1936, 10, 187. Vukanovic, D., Simic, M.M., Vukanovic, V., Nickel, H., Mazurkiewicz, M. Spectn~chim. Acta 1977,

32B, 305. Vukanovic, D., Simic, M.M., Vukanovic, D.D. Spectrochim. Acta 1984, 39B, 33. Vukanovic, V. Emission Spektroscopie, Tagung der Physik; DDR Acad. Verlag: Berlin, 1964. Vukanovic, V., Ikonomov, N., Pavlovic, B. Spectrochim. Acta 1971, 26B, 95. Vukanovic, V.M., Todorovic, P.S., Mihail, T.A. Spectrochim. Acta 1975, 30B, 31. Vukanovic, V.M. Proc. VIII Coll. Spectrosc. Intern; Verlag H.R. Sauerlander, Aarau, 1960, 62. Weizel, W., Rompe, R. Theorie Elektrischer Lichtobogen und Funken; Barth: Verlag, Leipzig, 1949. Whitehead, A.B., Heady, H.H. US Bur. Mines Rep. Invest. 6091. 1962, 89.


Zaidel, A.N., Kaliteevski, N.J., Lipis, L.W., Chaika, M.R U.SIA.E. C Tr 5745 Chem. 1963. Zander, A.T., Miller, M.H. Spectrochim. Acta 1985, 40B, 1023. Zheenbaer, Z., Engelsht, V.S., Urmanbetov, K. FACSS Meeting: Philadelphia, 1976. Zilbershtein, Kh.I. Spectrochemical Analysis of Pure Substances; Translated by J.H. Dixon; Adam

Hilger: Bristol, UK, 1977.


DIRECT AND NEAR REAL-TIME DETERMINATION OF METALS IN AIR BY IMPACTION-GRAPHITE FURNACE ATOMIC ABSORPTION SPECTROMETRY

Joseph Sneddon

I~ II.

A~ B. C.

D.

E.

F.

A b s t r a c t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204

I n t r o d u c t i o n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204

I m p a c t i o n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205

Bas i c P r inc ip l e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205

Th e o re t i c a l C o n s i d e r a t i o n s . . . . . . . . . . . . . . . . . . . . . . . . . 206

I n s t r u m e n t a t i o n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208

E x p e r i m e n t a l Resu l t s . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212

A p p l i c a t i o n s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218

Leg i s l a t i ve L e v e l s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

A c k n o w l e d g m e n t s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

R e f e r e n c e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224


203

204 JOSEPH SNEDDON

ABSTRACT

The basic and theoretical principles, design, development, characterization, and evaluation of a impaction-graphite furnace system which is capable of collecting particles in air in a few minutes for subsequent determination of metal concentrations primarily by atomic absorption spectrometry is described. These unique systems are capable of giving a direct, in-situ, and near real-time (few minutes) determination of low concentrations of metals (ng/m 3 range) in air. A description of the systems, a discussion on the factors which effect the collection efficiency, and their application to determining selected metals in the laboratory air and second hand cigarette smoke is presented.

i. I N T R O D U C T I O N

The detection and quantitative determination of metals or metallic compounds in aerosols (a solid or liquid particle in a gaseous medium, e.g. air) suffers from lack of promptness. Typically, a collection stage on a filter (cellulose or glass fiber) system of at least one-half hour to several days is used, followed by sample preparation (digestion or dissolution) and subsequent determination by a variety of analytical techniques such as atomic spectroscopy. A long sampling period is needed to obtain a measurable amount of metal when concentrations are low. While this method is routinely used and provides valuable information, it is tedious, time-consuming, and can lead to losses (and reduced accuracy) due to storage, transportation, and contamination. A review of this standard method using atomic spectroscopic methods is available (Sneddon, 1983) as well the use of graphite furnace atomic absorption spectrometry (GFAAS) in determining metals in air (Noller et al., 1982). Furthermore, the standard methods are in retrospective and do not include the possibility that a short-duration, high-concentration exposure occurs which, when averaged over the sampling time period, may not appear to be dangerous.

Clearly there could be a need for a system which is capable of providing quantitative determination of metals in air on a real-time or near real-time basis. Atomic spectroscopic methods (atomic emission, atomic absorption, and atomic fluorescence spectrometry) are excellent methods for trace metal determination due to several factors: low sensitivity; no, minimal, or easily corrected interferences; high specificity; good (low) precision; acceptable accuracy; and low cost per sample after initial cost of the instrumentation. However, they are primarily regarded (and give the best performance) as solution- or solid-sampling techniques. They have been applied to determine metals in air after collection and digestion or dissolution, but have not been widely applied to direct air analysis for metal determination. Several atomic spectroscopic techniques have shown potential for directly determining metals in air including electrostatic precipitation-graphite furnace AAS (Torsi et al., 1982; Sneddon, 1989b; Sneddon, 1991; Torsi et al., 1996) and

I-GFAAS Determination of Metals in Air 205

laser-induced breakdown (emission) spectrometry (LIBS) (Cremers et al., 1983; Essien et al., 1988a; Essien et al., 1988b; Radziemski, 1994).

A further promising technique is that of combining a single-stage impactor to a graphite furnace for the direct collection of particles in air for subsequent determination of metals using atomic absorption spectrometry in near real-time (a few minutes). The technique has been referred to as impaction-graphite furnace atomic absorption spectrometry (I-GFAAS). This chapter will present and discuss this technique as it is applied to the direct and near real-time determination of metals in air (Sneddon, 1986, 1988, 1990a; Sneddon et al., 1995).

i l . I M P A C T I O N

Impaction techniques have been widely and extensively used in industrial hygiene for the collection and sizing of aerosols (Marples, 1970; Fuchs, 1978). The size of particles in an aerosol are considered important from a health point of view. In general, particles in the 1 to 15 micron size are carded more efficiently into the lungs. However, the specific metal and accumulation in the body may also play a role in characterizing the toxicity of a metal and subsequent health hazard.

A. Basic Principles

An impactor is an instrument in which an aerosol issuing from a narrow jet impinges on an impaction plate or surface. Aerosol particles are deposited on this surface because of their inertia. One of the first instruments was described in 1945 (May, 1945). It consisted of four jets and four sampling plates. The jets were progressively smaller, so that as the speed of the aerosol increased, the finer or smaller particles were collected or impacted on the impaction surface and removed from the aerosol. This aerosol continued to the next set of smaller jets and impaction plates; the latter were usually the same material and size. The final step would be a size grading of the aerosol.

The 1945 instrument was the forerunner for commercial systems including the Anderson Impactor (Anderson Impactors, Inc., Atlanta, Georgia). These systems were inexpensive and relatively straightforward to use. Several new and more sophisticated cascade-type systems are available including 10-stage piezoelectric systems which feature in-situ electronic weighing and give complete mass information (concentration and size distribution) in a few minutes (Californian Meas- urements, Sierra Madre, California). In these cases, the information on particle sizes was considered the major focus as outlined earlier. The use of impaction combined with a graphite furnace uses the principle outlined above' However, in this case, a single-stage impactor consisting of a single jet and graphite furnace (impaction plate) are used. The basic principle is shown in Figure 1.

The aerosol or air sample is drawn by vacuum at a known flow rate or speed through the single-stage impactor jet. Particles in the aerosol are deposited on the

206 JOSEPH SNEDDON

Figure 1. Schematic cross-sectional view of the impaction-graphite furnace system (from Sneddon, 1984).

impaction (graphite furnace) surface. This single-stage impaction-graphite furnace (I-GF) system will separate particles in an aerosol into two sizes: particles larger than a certain aerodynamic size are removed from the aerosol and deposited (impacted) on the graphite furnace and smaller particles will pass through the I-GF system. The particles removed from the aerosol and collected on the graphite tube can then be added to a graphite furnace atomization unit for subsequent quantifi- cation by (primarily) GFAAS.

B. Theoretical Considerations

The movement of an aerosol in a single-stage I-GF system is extremely complex: the motion relies on many factors including flow rate, jet diameter, and the distance from the jet exit to the graphite furnace. These factors will be critical in determining the particle size collected on the I-GE An ideal system would collect and remove all particle sizes from an aerosol. This is termed collection efficiency, and is defined as the mass of metal collected on the graphite furnace after entering the system divided by the mass of the metal initially entering the system, expressed as a percentage. Ideally this would be 100%. However, single-stage I-GFs do not have a 100% collection efficiency; instead they are rated by a cut-off particle size, ds0, which is defined as the particle size at which at least 50% of a certain particle size is collected.

If particle motion in an aerosol is governed by Stokes Law (Hinds, 1982), then,

Stk = p(dp)2 U Cc (1) 9 p D


where U is the average jet exit flow velocity (in cm/s), P is the density of the particle (1 g/cm3), D is the jet diameter (in mm), and p is the viscosity of air (1,81 x 10 -4 g/cm at 20 ~ dp is the particle diameter, Cc is the Cunningham correction factor (~1), and Stk is the Stokes number. Assuming (Stk) 1/2 = 0.475 (Hinds, 1982), substituting ds0 for dp, and rearranging Eq. 1"

1/2

=19S_~,D) (2) 0v i I

Using this equation and the constants, a plot of particle size collected, ds0, versus average flow rate and jet or nozzle diameter is shown in Figure 2. This figure reveals information on the theoretical particle size collected using the geometry of the I-GF system. At low flow rates of around 0.1 L/min and ajet diameter of 1.0 mm, particles of less than 6 ~m would not be collected. At higher flow rate, e.g. 5 L/min, particles of less than 2 ~tm would not be collected, and at 15 L/min, the size of particles collected is less than 0.5 ~m. In general, by increasing the flow rate and decreasing the jet diameter will decrease the particle size collected. For example, at a low flow rate of 0.1 L/min, the particle size collected would drop to 2 ~m (compared to 6 ~tm at 1.0 mm jet diameter). Increasing the jet diameter to 1.5 mm would increase the particle size collected. At 0.1 L/min, the particle size collected would be approximately 10 ~m. Increasing the jet diameter beyond the 1.5 mm diameter would greatly increase the particle size diameter collected, particularly at low flow

35

30 -4 O 25 i - ra In 20

3 ~ 5 0

3 0 @

-'.- rL .Ol ju . .

"" ,v4~ (L~/nl

Figure 2. Theoretical study of particle size collected, ds0, versus flow rate and jet or nozzle diameter (from Lee et al., 1996a).

208 JOSEPH SNEDDON

rates. However, a jet diameter of much greater than 1.5 mm is impractical because it would be difficult to introduce the aerosol or air sample into the graphite furnace. Furthermore, as described earlier, particle sizes in the range of 1-15 ~tm are considered to be the most damaging to health because of their efficiency in entering and remaining in the body.

Finally, it is worth noting that these theoretical calculations provide information on the particle size collected but do not take into account the possibility of "bounce-off" errors (Hinds, 1982). The "bounce-off" is where a particle will literally bounce off the surface although it should theoretically be collected on the surface. However, they do provide information which allows the construction of a system which has a high probability of efficiently collecting particles (1-15 ~tm) potentially dangerous to health.

C. Instrumentation

Impaction Systems Using the theoretical studies outlined in section II.B for design purposes, several

systems have been constructed, characterized, and evaluated. An early system is shown in Figure 3 (Sneddon, 1984, 1985). This system was designed and constructed as a complete unit in which the impactor was connected directly to the graphite furnace atomizer. Later systems separated the collection stage from the analysis stage.

The jet was made of tantulum to prevent melting at the relatively high temperature (--3000 K) obtained by the graphite furnace at the analysis stage. The inside

A

E

I r

Figure 3. Schematic diagram of impaction system connected directly to a commercial electrothermal atomizer. (A) Electrothermal atomizer (graphite tube); (B) aluminum face plate with viewing window and outlets (2); (C) impactor tube with 1.00 mm tantalum jet; (D) connecting faceplate; (E) sampling faceplate; (F) connecting rubber stopper with glass jet; (a) atmosphere sampled; (b) determination performed; (c) standard aerosol introduced (from Sneddon, 1985).


diameter of the jet was 1.00 mm and outside diameter was 2.00 mm. The jet was pressed into the impactor tube, which was made from easily machineable aluminum of 30.0 mm outside diameter. The inside diameter was gradually tapered from 20 mm to 1.00 mm over a length of 110.0 mm. The gradual decrease in diameter prevented any buildup of par-ticles on sharp edges. A 40.0 mm length of the impactor was threaded and was matched with the specially constructed faceplate which replaced the faceplate of the commercial graphite furnace atomizer. This was to allow the jet-to-impactor distance to be varied. The faceplate was made of 10.0 mm thick aluminum and had a quartz viewing window for visual alignment of the jet to the graphite tube. Two cylinder outlets of inside diameter 5.00 mm and height of 15.00 mm were pressed into the faceplate. These two outlets were connected via a tee piece to a flowmeter and then to a vacuum pump, with vacuum tubing used in all the connections. This system allowed a maximum flow rate of 15 L/min. All contact surfaces were sealed with a rubber seal and checked periodically to ensure no leaks. A connecting faceplate (D) and sampling faceplate (E) were connected to the impactor. In position (a) the air was sampled, in position (b) the determination was performed, and in position (c) a standard aerosol could be introduced. A rubber stopper with a glass jet (F) was connected to position (c) and to the standardization unit. This standardization unit was an aerosol deposition system as described by Tapia et al. (1984). The analysis (sampling and determination) could be performed by manually moving the faceplate, although clearly there was potential for auto- mation.

A second system was constructed by Liang et al, (1990) and is shown schematically in Figure 4. In this case the sampling unit was separate from the graphite furnace system. It consisted of the same impactor tube previously described (Sneddon, 1985) and a plastic chamber with the graphite furnace inserted into this chamber. Air was drawn by vacuum through the ends of the chamber. After collection at a known flow rate and sampling time, the system was disconnected and the graphite tube removed and placed on the graphite furnace atomization unit for determination. This system was used for GFAAS and laser-excited atomic fluorescence spectrometry (AFS).

A third system was designed and constructed and shown schematically in Figures 5a and b (Lee et al., 1996a). It was constructed from a nylon block. The impactor tube was also constructed from nylon and was 10 cm in length by 22 mm in diameter. The impactor tube decreased from an inside diameter of 20 mm to the jet exit diameter over the 10 cm length. Several jet diameters were available (0.5, 1.0, and 1.5 mm). The outside of the impactor tube was threaded to allow a complete fit to the rest of the system. This had two advantages: (1) an airtight fit, and (2) the distance between the jet exit diameter and graphite (impaction) surface could be carefully controlled and varied if required. Three nearly, identical impactor tubes were constructed, the only difference being that the jets were of 0.5, 1.0, and 1.5 mm inside diameter. When the 1.0- and 1.5-mm jets were used, the graphite tube entry port had to be slightly enlarged. This new system had a barrel-type fitting

210 JOSEPH SNEDDON

a

GRAPHITE TUBE TO PUMP

" : . : . " . ~ ,'r~;

IMPACTOR TUBE WITH 1.0ram TANTALUM JET

TO PUMP

b RUBBER O-RING

GRAPHITE TUBE

MAIN CHAMBER

GRAPHITE TUBE HOLDER

r HOLDER

GRAPHITE TUBE

JET WITH lmm DIAME TER NOZZLE

Figure 4. Three views of the impaction chamber: (a) general view; (b) impaction device mounted into the graphite furnace; (c) close up of the nozzle inside the graphite tube (from Liang et al., 1990).

which allowed up to four separate graphite tubes to be installed [See Figure 5b, part (d)]. An experiment could be performed using one graphite tube by rotating the system to a new graphite tube with the same impactor. This could be repeated a third and fourth time, which was considered to be particularly useful when studying precision from aqueous solutions introduced through the impactor.

Detection Systems

The graphite furnaces used as the collection or impaction surface are, for the most part, from commercial GFAAS systems. The exception was a laboratory-modified


Figure 5. (A) Impaction-graphite furnace system. (B) Schematic diagraph (a) top view; (b) left-hand side view; (c) right-hand side view; (d) exploded view of the inside (from Lee et al., 1996a).

atomization graphite furnace unit used for laser-excited atomic fluorescence spectrometry (LEAFS) (Liang et al., 1990). Initial work was performed using AAS systems which were capable of only single metal determination (Sneddon, 1984, 1985). Recently, this has been extended to simultaneous multimetal AAS systems (Sneddon et al., 1995; Lee et al., 1996a).

212 JOSEPH SNEDDON

D. Experimental Results

Effect of Geometry

Initial work and studies on the I-GF system found that its geometry played an important part in the collection efficiency. As mentioned earlier, an ideal collection efficiency would be 100%, but this is not possible with a single-stage impactor. A collection efficiency of 100% would require a system capable of collecting all particle sizes, i.e. a multistage impactor system. Collection efficiency is dependent on a number of factors related to the system including geometry (jet exit-to-impaction surface distance, jet diameter, and impaction surface) as well as flow rate and particle size. A rigorous study was undertaken by Lee et al. (1996b) to study the effect of these five parameters on collection efficiency. Ideally, four parameters would be kept constant and the fifth would be varied, and so on. However, in practice it was found that it was more realistic and practical to vary some factors while investigating other factors; e.g. the jet diameter would be varied while investigating the flow rate, etc. The optimum signal and hence the optimum collection efficiency was obtained using the peak area absorbance signal when a known concentration of metal in an aerosol was introduced into the system.

Effect of Flow Rate. The effect of flow rate for five aqueous chromium solutions (10, 20, 30, 40, and 50 ktg/mL) and a blank introduced through the impactor system is shown for jet diameter of 1.00 mm (Figure 6a) and 1.50 mm (Figure 6b). A linear calibration was found for all four flow rates (1.0, 5.0, 10.0, and 15.0 L/min) and the two jet diameters investigated. However, at low concentrations and a jet diameter of 1.00 mm the precision is poor, typically > 10%. Using a jet diameter of 0.50 mm (not shown), the poor precision became more pronounced (>15%) even at higher concentrations of chromium. It was noted that there appeared to be very little difference in absorbance signals at high flow rates (10 and 15 L/min) when using a jet diameter of 1.50 mm (see Figure 6b). This suggests that with higher flow rates (in combination with a large jet diameter) the collection efficiency is reduced.

Effect of Jet Diameter. Three jet diameters were investigated: 0.50 mm, 1.00 mm, and 1.50 mm. The results were discussed above.

Effect of ]et Exit-to-lmpactor Distance. The effect ofjet exit-to-impactor distance is shown in Figure 7 for several flow rates (1, 5, and l0 L/min) using l0 I.tg/L of lead aerosol and a jet diameter of 1.50 mm. At a low flow rate of 1 L/min the optimum distance is around 3 mm. It is relatively constant after 3 mm, but is reduced to zero at smaller distances. At higher flow rates (5 and 10 L/min), an optimum occurs around 3 mm but decreases after and before this optimum. If the jet diameter was 1.00 mm (not shown) the optimum still occurred at 3 mm with a decrease below and above this distance. These results suggest that at high flow rates there could be

I-GFAA5 Determination of Metals in Air 213

0.0 �9 i ' - i ' �9 i �9 ' 1 i 0 0 10 20 30 40 Concentration of chromium,/~g/L

l 0.2

0.1

0.18

o o ,

-0.02 0 10 20 30 40 50

Concentration of chromium,/~g~

Figure 6. The effect of flow rate for five aqueous solutions of chromium through a jet impactor size of (a) 1.00 mm and (b) 1.50 mm (from Lee et al., 1996b).

a "bounce-off" effect at distances greater than 3 mm. At low flow rates, the signal (collection efficiency) decreases. This suggests that the particles are not collected on the surface and pass through the system.

Effect of Impactor (Graphite) Surface Material. Different impaction surfaces could effect the collection efficiency due to the possibility of the "bounce-off" effect. A preliminary investigation involving the use of an uncoated and pyrocoated graphite fumace showed that the analytical signal obtained with a 10 lag/L solution of lead to be around 2-5% higher when using the uncoated graphite fumace compared to the pyrocoated graphite fumace. This could be predicted due to the more open structure of the uncoated graphite furnace compared to the less open graphite structure of a pyrocoated surface. However, it should be noted that the particle sizes of the aerosol generated by this nebulizer would be in the range of

214 JOSEPH SNEDDON

Figure 7. The effect of jet exit to impactor for several flow rates (1,5, and 10 L/min) with a jet diameter of 1.50 mm (from Lee et al., 1996b).

around 6 microns to submicron sizes and "bounce-off" errors are more critical with larger particle sizes. It is possible that the impaction surface material would be more critical if the system collected large particle sizes. However, the geometry of the system has been developed to collect particle sizes of less than 15 microns.

Effect of Particle Size. A preliminary study of the particles collected through the impaction system focused on the size of the collection area on the impaction surface. Initial studies of particle size involved introducing 10 lag/L of lead through the nebulizer for 1 min at a flow rate of 5 L/min with impaction on carbon planchets (12.7 mm diameter and 1.6 mm thick). The particle size of this aerosol was estimated to be from around 6 microns to submicron particle size (Browner and Boom, 1984). Particle sizes were studied using a scanning electron microscope (SEM). The planchets closely resemble a graphite fumace surface. The results for various jet diameter widths of 0.50, 1.00, and 1.50 mm were obtained and the impaction patterns were clearly visible. The results obtained could be predicted, i.e. the impacted area size increases with increasing jet diameter. Using a jet diameter of 1.00 mm, the flow rate was varied at 1, 5, and 10 L/min, and the same concentration of aerosol was impacted onto the planchets for 1 min and studied by SEM. The results showed that at 1 L/min the particles are low in number and scattered. Increasing the flow rate showed a significant increase in density.


Based on this work, Lee et al. (1996b) recently proposed that the optimum conditions for maximization of collection efficiency for particles of less than 6 microns is a combination of medium flow rates (around 5 L/min) and jet diameter of 1.0 mm, and a jet-to-impaction distance of 3 mm. The higher the flow rate, the more dense the particles are on the graphite furnace. The results obtained in this study confirmed an earlier study by Sneddon (1989b). In this case the work was performed using the early system which had the impactor connected directly to the graphite furnace atomization unit. This study also pinpointed the variables such as geometry, flow rate, and particle size which affect collection efficiency.

An earlier study on the specific effect of particle size on collection efficiency by Sneddon(1989a) used a box system shown in Figure 8. The system was made from 0.5 cm thick aluminum and measured 47 cm x 47 cm x 77 cm. The airtight system was divided into two separate chambers with one of the chambers containing a pressure relief valve, which was considered necessary for safety. A buildup of pressure could cause the system to explode. On top were two windows to allow viewing of the system. On the plate dividing the two chambers were two impaction systems. The first system, shown in an expanded view in Figure 8, consisted of the impactor (D) which was positioned opposite the entrance port of a graphite furnace. The distance from the jet exit to this surface could be varied. At the two sides of the graphite furnace was positioned a 0.8 ~tm pore size filter and filter holder which was connected via a tee-piece to a flowmeter and pump. The flowmeter and pump were outside the system. The second part had a second but identical impactor with the jet exit positioned at an identical filter and filter holder. This was connected to a flowmeter and pump (again outside the system). In both parts of the system were several filters and holders. A fluid bed (c) was positioned in one chamber. A mass of a few grams of known particle size was placed in the fluid bed and filtered air was introduced at two ends to create a dry aerosol of known particle size. This aerosol was pumped through the system and collected on the filter systems (Bland B 6) and impaction surfaces. The system was cleaned before the next series of experiments were performed.

The results were considered very preliminary and somewhat inconclusive (the aerosols with known particle size generated clumps of particles and not a uniform particle size as expected). A potential problem was that the flow patterns would change when filters are placed at the ends of the graphite furnace. However, they did show that large particles > 50 ~tm would not be collected on the system.

Standardization

To calculate the concentration of metal in the aerosol or air, a calibration curve is established using aqueous solutions standards as follows (Sneddon, 1983, 1985; Lee et al., 1996a),

Mm Cstd • Vstd (3) Cm=-~a = FrXt s

216 JOSEPH SNEDDON

/

Figure 8. Particle size grading box for impactor-electrothermal atomizer. (A) Pressure relief valve; (B) filter and filter holder (6); (C) fluid bed; (D) impactor (2); (E) electrothermal atomizer (from Sneddon, 1989).

where C m is the equivalent concentrations of metal in ng/m 3, M m is the mass of .

metal in aqueous standard in ng, V a is the volume of air sampled in m 3, Cst d is the

concentration of standard in ng/mL [parts per billion (ppb)], Vst d is the volume of

standard in mLi F r is theflow rate in m3/mL, and t s is the sampling time in minutes.

An example would be as follows: Cst d = 20 ppb (20 ng/mL); Vst d -- 20 ~tL (20 x 10 -3 mL); F r = 10 L/min (10 x 10 -3 m3/min); and t s = 5 min

20 n g / m L x 20 x 10 -3 mL = 8 n g / m 3 (4)

Cm = 10 x 10 -3 m3/min x 5 min

Thus a calibration curve of absorbance (usually peak area absorbance, but peak

height absorbance could also be used) versus concentration in ng/m 3 can be


established using various concentrations of aqueous standards. If the conditions change (i.e. flow rate, sampling time, concentration) or volume of standard change, then a new calibration curve has to be established. It should be noted that standardization is achieved by introduction through the impactor system in the same manner as the air samples.

Accuracy and Precision

Accuracy can be defined as "how close" to the "correct" answer. It is usually presented statistically and can be established using a number of complementary techniques including standard additions, internal standards, comparison of the results obtained using this method to a different method, and a standard sample containing a known concentration of the particular metal of interest, typically a National Institute of Science & Technology (NIST) Standard Reference Material (SRM) (Gaithersberg, Maryland).

Clearly, none of the above methods are available to assess the accuracy of the I-GF system. The NIST urban particulate Standard Reference Material (SRM), 1648, could be added as a powder or slurry although, at present, this has not been undertaken. An attempt to assess the accuracy of the I-GF system was undertaken by Sneddon (1989a, 1990c) by comparison to standard methods of sampling air by collection on a filter, followed by digestion of the filter, and analysis by flame AAS. The experiments involved the use of a filter (0.8 ~tm pore size) and pump which sampled the air at a known flow rate was positioned as close to the I-GF system as possible and simultaneously sampled air. In both systems air was sampled at 5 L/min for 5 min. A cap was placed on the end of the filter system during the approximate 2 min to analyze the air collected by the I-GF system. After approximately 210 min, 150 L of air had been collected using both systems. The total metal mass collected by the I-GF system was determined by adding the 30 separate analyses, and the total mass content collected by the conventional filter method was determined by analyzing the digested filter paper. Using three separate experiments, the mass collected on the I-GF system was found to be 59-69% compared to the conventional method. These results were quite reproducible. As stated previously, the design and geometry of this single-stage I-GF system would prevent the collection of large particle sizes, and therefore, it is not surprising that a direct 100% comparison was not obtained. The 30-40% difference was attributed to the fact that the conventional system will collect all particle sizes, including large particle sizes, whereas the I-GF system would collect particle sizes less than 15 ~tm in size.

A further factor which could affect accuracy is the fact the system is standardized using aqueous standards. In most cases it is not acceptable (accurate) practice to use aqueous standards to calculate the concentration of an air sample containing a complex matrix. However, for this type of work a factor of 2 would not be considered significant in industrial hygiene, i.e. the difference in (say) 4 ng/m 3 and 8 ng/m 3.

218 JOSEPH SNEDDON

Precision is defined as the "repetitive analysis" of the same sample. Several experiments (three for each method) were performed using the I-GF system with sample introduction by electrothermal vaporization (ETV), a pneumatic nebulization (PN) system, and air (Sneddon, 1990b). A precision of 2-3% was obtained using the ETV and PN, which is comparable to that obtained by direct sample introduction into a conventional GF-AAS. The air sample precision ranged from 7.6 to 9.9%, a reduction in precision by a factor of 4. This highlighted a potential problem with the air sampled, namely that air is not a homogeneous sample. Therefore, the precision obtained or long-term stability of the concentration of metal in air will not be constant or (necessarily be) expected. Liang et al. (1990) discuss this subject when they analyzed their results of six metals in laboratory air and in a clean room. Further discussion of this is presented in the applications part of this chapter.

Detection Limits, Characteristic Concentration, and Useful Working Range

The detection limits, characteristic concentration, and useful working range obtained using I-GFAAS are identical to that obtained from conventional GF-AAS. These properties for a selection of metals are presented in Table 1 in units of ng/m 3 using various wavelengths (Sneddon, 1986).

Practice of the I-GF System The advantage of the I-GF system is the ability to directly collect particles in air

for subsequent quantitation by atomic spectroscopy, most commonly atomic absorption spectrometry. The collection time will vary depending on the flow rate and concentration of the metal in air to be determined. The maximum flow rate in the impaction systems appears to be around 15 L/min. A collection stage of five min at the maximum flow rate will collect 75 L of air. In many instances this is sufficient to obtain a measurable amount. In some instances a much longer sampling period is required (see mercury in air by Sneddon, 1989). Early impaction surfaces were connected directly to the graphite furnace atomization unit of an AAS system but later systems were separate. Using the later systems involved dismantling the impaction system, adding to the graphite furnace (with practice) around 30(s), and the actual determination (assuming a predetermined set of experimental conditions of drying, ashing or pyrolysis, atomization, and clean around 2 min). This gave the results in about 5 min (assuming a collection stage of around a few minutes was adequate and the standardization had been achieved). This is why the I-GFAAS is referred to as near real-time. In practice, the system would be standardized prior to sampling and experimental conditions determined prior to analyses.

E. Applications

To date the I-GF system has not been widely applied to the determination of metals in air or aerosols. However, sufficient applications and investigations have


Table 1. Detection Limits, Characteristic Concentration, and Useful Working Range for Selected Metals by I-GFAAS a'b

Metal

Characteristic Wavelength Detection Limit c Concentration d Useful Range e

(nm) (ng/m 3) (ng/m 3) (ng/m 3)

Arsenic 193.7" 24.0

Barium 350.1 600

553.6* 10.0

Beryllium 234.9* 2.0

Bismuth 223.1" 20.0

227.7 300

Chromium 357.9* 12.0

425.4 60.0

520.8 1000

Cobalt 240.7* 20.0

391.0 1000

Iron 248.3* 1.0

372.0 10

392.0 200

Mercury 253.7* 100

Lithium 323.3 800

670.8* 2.0

Nickel 232.0* 10.0

341.5 50

362.5 1000

Osmium 290.9* 70

426.1 1000

Lead 217.0* 20.0

283.3 30.0

Selenium 196.0* 25.0

204.0 300

Silicon 251.6" 20.0

288.2 300

Silver 328.1 * 20.0

338.3 100

Tin 286.3* 11.0

Zinc 213.9" 0.5

376.0 300

3.0 40-400

1.0 20-150

0.3 4-40

3.0 5-40

100 500-800

6.0 20-80

20.0 100-300

8.0 30-100

0.2 5-90

20 40-140

30 300-600

20 200-400

0.7 3-30

1.3 20-200

10 70-200

18 100-300

5 40-200

5 40-200

5 50-300

5 30-200

5 30-300

20 150- 500

3.6 20-200

0.1 1-15

Notes: *Resonance (most sensitive) wavelength. aFrom Sneddon, 1986. bObtained using a volume of equivalent to 20 ~L, flow rate of 10 L/min and sampling time of 5 min. CConcentration giving a signal-to-noise ratio of 3. dConcentration which gives 1% absorption (0.0044 absorbance units). eRange of concentration for which the relative standard deviation (precision) is less than 5%.

220 JOSEPH SNEDDON

been performed to demonstrate the potential of the system. The following are three potential applications of the system; several other applications have been performed but are not presented in this section (Sneddon, 1983, 1986; Lee et al., 1996c).

Mercury in Laboratory Air

Sneddon (1987) determined mercury in air in two separate laboratories. Mercury is an extremely toxic metal and detection and determination at low levels in the atmosphere is required in clinical, environmental, and industrial hygiene studies. One laboratory contained polarography instrumentation and the other did not. In both cases the levels of mercury were extremely low with no detectable levels (less than the detection limit of 0.1 ng/m 3 obtained in these experiments with this system) in the laboratory not containing the polarography instrumentation, and in the ng/m 3 range in the laboratory which contained the polarography instrumentation. A considerable sampling period of several hours was needed in order to obtain a measurable amount. This was still below the legislative levels. The system was standardized using aqueous solutions and the author noted the difference in peak shape between the aqueous mercury and the air mercury samples. This could be a factor in accuracy.

Cadmium, Chromium, Lead, and Manganese in Cigarette Smoke Lee et al. (1996a) determined cadmium, chromium, lead, and manganese in

cigarette smoke using an impaction system and subsequent determination with simultaneous multimetal AAS. Multimetal AAS has not widely been used although there has been increased interest in this technique in the 1990s. A description of multimetal AAS is available from Farah et al. (1993), Deval, and Sneddon (1995), and Farah and Sneddon (1995). A typical AAS profile is shown in Figure 9 for the simultaneous measurement of these metals in cigarette smoke.

w 0 . 3 0 u z ~C

0 u')

0

Cd

Mn

2 3 5 0

0 6 ATOMIZATION TIME, s

Figure 9. Absorption profiles of cadmium, chromium, lead, and manganese obtained from cigarette smoke collected by the i-GF system and subsequently determined by multimetal atomic absorption spectrometry (from Lee et al., 1996).


Experiments were performed in the laboratory air prior to smoking, during smoking, and four hours after smoking. The results of this study (Table 2) show variation factors by as much as 3. For example, lead levels in background air are reported as 20 ng/m 3, whereas results from other experiments showed these concentrations to be from 8 to 25 ng/m 3. This variation is somewhat expected because the air sampled is not homogeneous. The results in Table 2 also show that with the introduction of cigarette smoke there is a significant increase in metal concentration by factors of almost 5 for lead to in excess of 15 for cadmium.

Copper, Iron, Lead, Manganese, Tin, and Thallium in Air of a Trace Metal Clean Room

Liang et al. (1990) collected particles in the air from a trace metal clean room using an impaction system and subsequently determined copper, iron, lead, manganese, tin, and thallium using AAS or laser-excited atomic fluorescence spectrometry (AFS). The impactor was placed on a laboratory bench, either in the normal laboratory or inside the clean area of a class 100 clean bench or hood in the clean room. The whole of the clean room did not have filtered air. Therefore, a third type of measurement was made inside the clean room, but not in the clean bench area. Measurements were made at, typically, 1.4 m above the floor.

The results (Table 3) show a difference in the concentrations of the selected metals in air for all three locations. The authors were satisfied that the air from the clean room contained less metals than normal laboratory air and that the concentrations of metals in the air from the clean bench were low. Also, the filters were filtering the air sufficiently to remove substantial amounts of metals in the particulate matter. The concentrations were in the ng/m 3 range with thallium at extremely low levels. The authors noted that tin levels were about the same in the clean bench or hood and in the clean room. A study of long-term stability was conducted 7 months later and a comparison of the results were made with each pair of metals using Student's t test at the 95% confidence limit. Only 5 out of the 17 pairs of metals compared could be considered different with iron changing by a factor of 3. The authors felt

Table 2. Concentrations of Cadmium, Chromium, Lead, and Manganese Determined in Air Using I-GFAAS a'b

Before Introduction of Cigarette 4 Hours After Cigarette Metal Cigarette Smoke (ng/m 3) Smoke (ng/m 3) Smoke (ng/m 3)

Cadmium 8 128 16

Chromium 4 23 4

Lead 20 88 17

Manganese 10 68 14

Notes: aFrom Lee et al., 1996. bAir sampled at 10 L/min for 10 s.

Table 3. Determination of Metals in Air by Impaction-GFAAS and Impaction-LEAFS a'b Concentration of Metals in Air

(ng/m 3)

Author's Laboratory Clean Room Clean Hood or Bench Detection Limit

bo I',o I,,o

Metal Nov. 1989 June 1989 Nov. 1989 June 1989 Nov. 1989 June 1989 Cu 1.89 + 0.13(6) 1.43 + 0.27(4) h'z 0.31 + 0.10(5) 0.41 • 0.05(4)* c c 0.02

Fe 6.65 • 0.75(6) 16.9 • 0.47(4) z 1.54 • 0.58(3) 4.13 • 0.61(4) z 0.049 + 0.029(4) 0.046 • 0.016(4)* 0.01

Mn 0.71 + 0.08(6) 0.70 + 0.12(3)* 0.16 • 0.07(4) 0.093 • 0.025(3)* c c 0.05

Pb 1.25 • 0.09(6) 1.24 • 0.03(4)* 0.19 • 0.02(4) 0.18 • 0.04(3) c c 0.01

0.32 + 0.05(4) e 0.60 _+0.18(4) e'z 0.072 • 0.0020(4) d 0.0063 • 0.0020(4) e'z 0.0001 e

Sn 0.21 • 0.04(6) 0.27 • 0.12(4)* 0.072 + 0.013(3) 0.038 • 0.015(3) z 0.072 • 0.001(4) 0.040 • 0.004(3) z 0.01

TI 0.0043 • 0.0079 + 0.0086 • 0.00099 • 0.000029 • 0.00043 • 0.00010 0.0009(4) d 0.0018(4) e* 0.00013(3) e 0.00017(3) e* 0.000014(3) e 0.000025(4) e*

mean RSD 13 18 26 22 34 34

Total mean RSD 22% (November, 1988 data) 23% (June, 1989 data) mean RSD of the pooled data: 23%

Notes: aFrom Liang et al., 1990.

bData is expressed as the mean + standard deviation, followed by the number of replicate measurements (n) in each set of data is given in parentheses: Measurements were by graphite furnace-AAS, except those indicated by d.

CBelow the detection limit.

dSampling time was 6 h, detection limit based on signal to noise ratio of 3.

eDetermined by laser excited atomic fluorescence spectrometry.

fSampling time of 12 h.

gStudent's t test of the pooled data indicated that, overall, the data of November, 1988 were no different from the data of June, 1988.

hStudent's t test of individual sets of data showed no significant differences in most cases of paired data (*), and a significant difference in five cases (z).

iMean relative standard deviation (by column).


that the particular contamination and ubiquitous nature of iron contributed to these results. They also concluded that given the different seasons, and potential changes caused by contamination, some variation would be expected. A study of short-term precision showed a variation of 22-26%, which is quite acceptable given the nature of the sampling. However, the authors were not entirely satisfied with the accuracy of the results.

F. Legislative Levels

The analysis of particles in air or "dust" has been of long concern to environmen- talists, industrial hygienists, and analytical chemists. The introduction of legislation in the United Kingdom in 1974, the Health and Safety at Work Act, conferred new and wide-ranging responsibilities on employers involving a legal obligation to be aware of hazards, particularly in processes within their industry. Similar legislation in the United States is controlled by the Occupational Safety and Health Agency (OSHA). In addition, the United States Environmental Protection Agency (EPA) maintains some control over the introduction of new and potentially harmful substances. For chemical substances in workroom air, the American Conference of Governmental Industrial Hygienists (ACGIH) publishes a list of Threshold Limit Values (TLVs) based on experiments on animals, human volunteers, reports of medical cases, and industrial experience. These can be in the form of a Threshold Limit Value-Short Term Exposure Limit (TLV-STEL) which defines worker exposure limits to a certain concentration or mass of metal over an 8-hour period, or a Threshold Limit Value-Time Weighted Average (TLV-TWA) which is a certain concentration or mass over (usually) a 15-minute period. These values vary for different metals and range from the low microgram to milligram per cubic meter levels. These values are continually assessed and in some case lowered as more knowledge is gained and newer instrumentation which can detect lower levels becomes available and accepted. It should be pointed out that this does not mean that a certain concentration or mass of metal is "safe" For example, there is no safe level of lead in the air which can enter the human body (lead has no known need or role in the human body). It does not take into consideration that many metals can concentrate and accumulate in the body. It is intended to provide minimal guidelines for safety, Documentation of the Threshold Limit Values for Substances in the Workroom Air, 1975.

ACKNOWLEDGMENTS

The author gratefully acknowledges the generous support of the Thermo Jarrell Ash-Baird Corporation, in particular, Gerald R. Dulude, Zach Moseley, and John J. Sotera. This work was supported by the Louisiana Education Quality Support Fund (LEQSF) Research Program for 1994-96-RD-A-21.

224 JOSEPH SNEDDON

REFERENCES Browner, R.F., Boom, A.W. Anal. Chem. 1984, 56, 786A-798A. Cremers, D.A., Radziemski, L.J., Loree, T.R., Hoffman, N.M. Anal Chem. 1983, 55, 1246-1251. Documentation of the Threshold Limit Values for Substances in the Workroom Air, American Conference

of Government Hygienists; Cincinnati, 3rd, edition, 1975. Deval, A., Sneddon, J. Microchem. J. 1995, 52(1), 96-100. Essien, M., Radziemski, L.J., Sneddon, J. Proceedings of International Conference on Laser's 87 STS

Press: McLean, Virginia, 1988a, pp. 908-912. Essien, M., Radziemski, L.J., Sneddon, J. J. Anal. Atomic Spectrom. 1988b, 3, 985-988. Farah, K.S., Farah, B.D., Sneddon, J. Microchem. J. 1993, 48(3), 318-325. Farah, K.S., Sneddon, J. Appl. Spectrosc. Rev. 1995, 30(4), 351-371. Fuchs, N.A. Fundamentals of Aerosols Science; Shaw, D.T., Ed.; John Wiley & Sons: New York, 1978,

Chap. 1. Health and Safety at Work Act, 1974, H.M.S.O., London. Hinds, W.C. Aerosol Technology; John Wiley & Sons: New York, 1982. Lee, Y.I., Smith, M.V., Indurthy, S., Deval, A., Sneddon, J. Spectrochim. Acta 1996a, 51B(1), 109-116. Lee, Y.I., Indurthy, S., Smith, M.V., Sneddon, J. Anal. Lett. 1996b, 29(14), 2515-2524. Lee, Y.I., Smith, M.V., Indurthy, S., Sneddon, J. J. Anal. Atomic Spectrom. 1996e, in preparation. Liang, Z.L., Wei, G.T., Irvin, R.L., Walton, A.P., Michel, R.G., Sneddon, J. Anal. Chem. 1990, 62(13),

1452-1457. Marples, V.A. Fundamental Study of Inertial Impactors, Ph.D. Thesis, University of Minnesota, 1970. May, K.R.J. Sci. Instrum. 1945, 22, 187-190. Noller, B.N., Bloom, H., Arnold, A.P. Progress In Atomic Spectroscopy 1982, 3, 81-189. Radziemski, L.J. Microchem. J. 1994, 50, 218-234. Sneddon, J. Talanta, 1983, 30, 631-648. Sneddon, J. Anal. Chem. 1984, 56, 1982-1986. Sneddon, J. Anal. Lett. 1985, 18(A10), 1261-1280. Sneddon, J. Am. Lab. 1986, 18(3), 43-50. Sneddon, J. Spectrosc. Lett. 1987, 20(6 & 7), 527-535. Sneddon, J. Trends in Anal. Chem. 1988, 7(6), 222-226. Sneddon, J. Anal. Lett. 1989a, 22(13,14), 2887-2893. Sneddon, J. Appl. Spectrosc. 1989b, 43(6), 1100-1102. Sneddon, J. In Sample Introduction in Atomic Spectroscopy; Sneddon, J., Ed.; Elsevier Publications;

Amsterdam, 1990a, pp. 329-352. Sneddon, J. Anal. Lett. 1990b, 23(6), 1107-1112. Sneddon, J. Appl. Spectrosc. 1990c, 44(9), 1562-1565. Sneddon, J. Anal. Chim. Acta 1991, 245(2), 203-206. Sneddon, J., Smith, M.V., Indurtha, S., Lee, Y.I. Spectroscopy 1995, 10(1), 26-30. Tapia, T.A., Combs, P.A., Sneddon, J. Anal. Lett. 1984, 17(A8), 2333-2347. Torsi, G., Desimoni, E., Palmisano, F., Sabbatani, L. Analyst 1982, 107, 96-101. Torsi, G., Reschiglian, P., Lippolis, M.T., Toschi, A. Microchem. J. 1996, 54(4), 437-445.

INDEX

Applications of ICP-MS, 25-26 Applications for multielement AAS

using flames, 47-49 Hitachi, 49 multichannel systems, 47-48 SIMACC, 48 Thermo Jarrell Ash-Baird, 49

Applications for multielement AAS using graphite furnaces, 49-58

dual-channel, 52-53 fast Fourier transform, 55-57 FREMS, 51 Hitachi, 51-52 Perkin-Elmer SIMAA, 57-58 SIMACC, 53-54 Thermo Jarrell Ash-Baird, 54-55 time-divided, single channel, 54

Basic principles of dc arcs and plasmas, 64-74

dc plasma jet, 73-74 description of dc discharge, 65-67 vertical dc arc (free burning), 67-

73 history, 70-71 mechanical structure, 67-70

Basic principles of ICP-MS, 3 Basic theory of ICP-MS, 2 Behavior of analytes in dc discharge,

74-105

electrode effects, 75-95 thermochemical reactions

between impurity and matrix elements, 81-83

thermochemical reactions between matrix and graphite, 75-77

thermochemical reactions between trace elements and graphite electrode 77-81

volatilization rate, 84-95 plasma effects, 95-105

axial distribution of line intensity, 95-99

axial electron density distribution, 118-120

axial temperature distribution, 109-112

electron density, 116-118 free particle concentration, 131-

134 jets, 129-131 local thermodynamic equilib-

rium, 124-125 particle concentration of trace

elements, 138-142 particle velocity, 134-135 radial distribution of electron

density, 120-124

225

226 INDEX

radial temperature distribution, 112-116

temperature, 105-109, 125-129 temporal variation of plasma

variables, 145-155 total particle concentration of the

third matrix, 135-138 transport phenomena, 143-145 voltage and electric fields, 99-

105 Impaction-graphite furnace AAS, 205

applications, 218-223 cadmium, chromium, lead, and

manganese, in cigarette smoke, 220-221

copper, iron, lead, manganese, tin, and thallium in air of a trace metal clean room, 221- 223

mercury in laboratory air, 220 basic principles, 205-206 experimental results, 212

accuracy and precision, 217-218 detection limits, characteristic

concentration, and useful working range, 218-219

effect of geometry, 212-215 practice of, 218 standardization, 215-217

instrumentation, 208-211 detections systems, 210-211 impactions systems, 208-210

legislative levels, 223 theoretical considerations, 206-208

Instrumentation for ICP-MS, 8-16 electron multiplier, 15 ion lenses, 14-15 interface region, 12-14 mass analyzer, 15 plasma, 11-12 sample introduction systems, 8-11

nebulizers, 9-10 spray chambers, 10-11

torches, 11 vacuum pumps, 16

Instrumentation for multielement AAS, 35-47

continuum source, 36-38 multielement line sources, 35-36 multielement systems, 38-46 multielement systems using lasers,

46-47 Interferences in ICP-MS, 4-7

doubly charged, 6 isobaric, 4 mass discrimination effects, 6 polyatomic, 4 signal drift, 6

Ionization energy of argon, 3 Isotopic ratio analyses, 7-8 Saha equation, 4 Sample introduction techniques in

ICP-MS, 16-25 electrothermal vaporization, 18-19 chromatography, 22-23 flow injection, 20-22 hydride generation, 23-25 laser ablation, 17-18 slurry nebulization, 19-20

Techniques in spectrochemical analyses by dc arc plasma 157-195

analysis of aluminum and titanium oxides, 186-193

analytical procedure, 191 analytical results, 192-193 matrix and theoretical considerations, 190-191

analysis of molybdenum and tungsten oxides, 193-195

analytical procedure, 194 analytical results, 194-195 matrix and theoretical considerations, 193-194

analysis of rare earth oxides, 172- 176

analytical procedure, 174

Index 227

analytical results, 174 behavior of matrices and theoret-

ical considerations, 172 analysis of rock phosphates, 176-

181 analytical procedure, 179-180 analytical results, 181 behavior of matrices and theoret-

ical considerations, 176-179 analysis of uranium, thorium, zirco-

nium, and plutonium oxides, 164-172

analytical procedure, 167 analytical results, 167-172 axial distribution of electron den-

sity, 165 axial distribution of temperature,

165 axial distribution of total particle

concentration, 165

behavior of the matrices and theoretical considerations, 164

influence of graphite and fluorination, 166

volatilization rates, 165 buffers, fluxes, and internal stan-

dards, 159-160 carrier distillation, 160-162 cathode layer, 157-158 cathode region, 158-159 development of general schemes for

multielement analysis 163-164 calibration standards, 164-165 plasma operating conditions,

163-164 sample preparation, 163

multielement analysis of silicate rocks, 181-186

analytical procedure, 185 analytical results, 185-186 matrix behavior and theoretical

considerations, 184-185

J A l

P R E S S

Advances in Atomic Spectroscopy

Edited by Joseph Sneddon, Department of Chemistry, McNeese State University

This series describes selected advances in the area of atomic spectroscopy. It is primarily intended for the reader who has a background in atomic spectroscopy; suitable to the novice and expert. Although a widely used and accepted method for metal and nonmetal analysis in a variety of complex samples, Ad- vances in Atomic Spectroscopy covers a wide range of materials. Each chapter will completely cover an area of atomic spectroscopy where rapid development has occurred.

Volume 1, 1992, 238 pp. $109.50 ISBN 1-55938-157-4

CONTENTS: Preface, Joseph Sneddon. Analyte Excitation Mechanisms in the Inductively Coupled Plasma, Kuang-Pang Li and J.D. Winefordner. Laser-Induced Ionization Spectrom- etry, Robert B. Green and Michael D. Seltzer. Sample Intro- duction in Atomic Spectroscopy, Joseph Sneddon. Background Correction Techniques in Atomic Absorption Spectrometry, G. Du/ude. Flow Injection Techniques for Atomic Spectrometry, Julian F. Tyson.

Volume 2, 1995, 297 pp. $109.50 ISBN 1-55938-701-7

CONTENTS: Preface. Effect of Molecular Orientation on Electron Transfer and Electron Impact Ionization, Phi/ R. Brooks and Peter W. Hat/and. Experimental Approaches to the Unimolecular Dissociation of Gaseous Cluster Ions, Terry B. McMahon. New Approaches to Ion Thermochemistry via Dissociation and Association, Robert C. Dunbar. Alkyl Cation- Dihydrogen Complexes, Silonium and Germonium Cations: Theoretical Considerations, Peter R. Schreiner, H. Fritz and Paul v. R. Sch/eyer. Symmetry Induced Kinetic Isotope Effects in Ion-Molecule Reactions, Greg /. Ge//ene. ion-Molecule Chemistry: The Roles of Intrinsic Structure, Solvation and Counterions, John E. Bartmess. Gas Phase Ion Chemistry under Conditions of Very High Pressure, W. Burk Knighton and Eric P. Grimsrud. Index

JAI PRESS INC. 55 Old Post Road No. 2 - P.O. Box 1678 Greenwich, Connecticut 06836-1678 Tel: (203) 661- 7602 Fax: (203) 661-0792

Advances in Gas Phase Ion Chemistry

Edited by Nigel Adams, and Lucia M. Babcock, Department of Chemistry, The University of Georgia

Volume 1, 1992, 329 pp. $109.50 ISBN 1-55938-331-3

CONTENTS: Preface, Nigel Adams and Lucia M. Babcock. Flow Tube Studies of Small Isomeric Ions, Murray J. McEwan. Anion Molecule Experiments: Reactive Intermediates and Mechanistic Organic Chemistry, Joseph J. Grabowski. Ther- mochemical Measurements by Guided Ion Beam Mass Spec- trometry, Peter W. Armentrout. Photoelectron Spectroscopy of Molecular Anions, Kent M. Ervin and W. Carl Lineberger. Ion Chemistry at Extremely Low Temperatures: A Free Jet Expansion Approach, Mark A. Smith and Michael Hawley. Theoretical Studies of Hypervalent Silicon Anions, Mark S. Gordon, Larry P. Davis and Larry W. Burggraf. Chemistry Ini- tiated by Atomic Silicon Ions, Diethard K. Bohme. Spectro- scopic Determination of the Products of Electron-Ion Recombination, Nigel G. Adams. Index.

Volume 2, 1996, 266 pp. $109.50 ISBN 1-55938-703-3

CONTENTS: Preface. Effect of Molecular Orientation on Electron Transfer and Electron Impact Ionization, Phil R. Brooks and Peter W. Harland. Experimental Approaches to the Unimolecular Dissociation of Gaseous Cluster Ions, Terry B. McMahon. New Approaches to Ion Thermochemistry via Dissociation and Association, Robert C. Dunbar. Alkyl Cation- Dihydrogen Complexes, Silonium and Germonium Cations: Theoretical Considerations, Peter R. Schreiner, H. Fritz and Paul v. R. Schleyer. Symmetry Induced Kinetic Isotope Effects in Ion-Molecule Reactions, Greg L Gellene. Ion-Molecule Chemistry: The Roles of Intrinsic Structure, Solvation and Counterions, John E. Bartmess. Gas Phase Ion Chemistry under Conditions of Very High Pressure, W. Burk Knighton and Eric P. Grimsrud. Index

JAI PRESS INC. 55 Old Post Road No. 2 - P.O. Box 1678 Greenwich, Connecticut 06836-1678 Tel: (203) 661- 7602 Fax: (203) 661-0792

.1 A 1

P R E, S S

J A l

P R E S S

Advances in Sonochemistry Edited by T i m o t h y J. Mason, Schoo l of Chemistry, Covent ry University, Eng land

Volume 1, 1990, 275 pp. $109.50 ISBN 1-55938-178-7

CONTENTS." Introduction to Series: An Editor's Foreword, Albert Padwa. Introduction, Timothy J. Mason. Historical Introduction to Sonochemistry, D. Bremner. The Nature of Sonochemical Reac- tions and Sonoluminescence, M.A. Marguli. Influence of Ultra- sound on Reactions with Metals, B. Pugin and A.T. Turner. Ultrasonically Promoted Carbonyl Addition Reactions, J.L. Luche. Effect of Ultrasonically Induced Cavitation On Corrosion, W.J. Tom- linson. The Effects of Ultrasbund on Surfaces and Solids, Kenneth S. Suslick and Stephen J. Doktycz. The Use of Ultrasound for the Controlled Degradation of Polymer Solutions, G. Price.

Volume 2, 1991, 322 pp. $109.50 ISBN 1-55938-267-8

CONTENTS: Sonochemical Initiation of Polymerization, P. Kruss. Free Radical Generation by Ultrasound in Acqueous Solutions of Volatile and Non-Volatile Solutes, P. Reisz. Ultrasonic Studies of Polymeric Solids and Solutions, R.A. Pethrick. The Action of Ultra- sound on Solidifying Metals, O.V. Abramov. Ultrasound in Chemi- cal Processing, N. Senapat. Ultrasonic Organic Synthesis Involving Non-Metal Solids, T. Ando. The Influence of Ultrasound on Oscillating Reaction, M.A. Margulis. The Manipulation of Parti- cles in an Acoustic Field, C.J. Schram.

Volume 3, 1993, 292 pp. $109.50 ISBN 1-55938-476-X

CONTENTS: Preface. Ultrasound and Colloid Science: The Early Years, David Gilling. Contributions to Various Aspects of Cavita- tion Chemistry, Arnheim Henglein. Sonochemistry: from Experi- ment to Theoretical Considerations, Jean-Louis Luche. Ultrasonic Agitation in Metal Finishing, Robert Walker. Ultrasonic Atomiza- tion, Andrew Morgan. Reations of Organosilicon Compounds in Acoustic Fields, 0.1. Zinovev and M.A. Margulis. Use of Ultra- sound in the Identification of Biological Molecules, John A. Evans. Initiation and Catalysis of Oxidation Proceses in an Acoustic Field, Evgen M. Mokry and Volodymyr L. Starchevsky. Index.

Volume 4, 1996, 295 ISBN 1-55938-793-9 Approx. $109.50

CONTENTS: Dosimetry for Power Ultrasound and Sonochemistry, J. Berlan (deceased) and T. J. Mason. Nuclear Magnetic Reso- nance Spectroscopy Combined with Ultrasound, J. Homer, Larysa Paniwyk and Stuart A. Palfreyman. Degassing, Filtration and Grain Refinement Processes of Light Alloys in an Acoustic Cavita- tion Field, Georgy Eskin. Sonochemistry in China, Yiyun Zhao, Jin- lei Yin and Ruo Feng. Uses of Ultrasound in Food Processing, T. J. Mason and L. Paniwnyk. Sonoelectrochemistry, S. J. Phull and D. J. Walton.