Catalytic processes in graphite furnaces for electrothermal atomic absorption spectrometry

SPECTROCHIMICA ACTA PART B

ELSEVIER

Spectrochimica Acta Part B 51 (1996) 1573-1589

Review

Catalytic processes in graphite furnaces for electrothermal atomic absorption spectrometry Anatoly B. Volynsky1 Sektion Analytik und HOchstreinigung, Universitiit Ulm, D-89069 Ulm, Germany

Received 1 January 1996; accepted 30 April 1996

Abstract The chemical processes in graphite furnaces used for electrothermal atomic absorption spectrometry (ETAAS) are discussed from the viewpoint of catalysis. The main attention is given to consideration of the processes which occur in the presence of platinum metal compounds and transition metal carbides. It is shown that both these groups of chemical modifiers accelerate either the processes of reduction of inorganic compounds or the thermal destruction of organoelement compounds and volatile hydrides in the graphite fumace. It is assumed that the similarities in the action of platinum metal compounds and transition metal carbides are based on the resemblance of their catalytic properties. The approach proposed may be used both for the improvement of already known chemical modifiers and for the creation of new ones. In general, adoption of the methods and concepts widely applied in catalysis for fundamental investigations of processes in graphite furnaces appears to be rather useful. Keywords: Catalysts; Chemical modifier; Electrothermal atomic absorption spectrometry; Graphite furnace

1. Introduction Catalysis refers to the phenomenon in which a relatively small amount of a foreign material, called a catalyst, augments the rate of a chemical reaction without itself being consumed (see, for example, Ref. [1]). The simplest example of a catalytic chemical reaction (with A and B as reagents and C as catalyst) is

A+C---* AC

(1)

AC + B ~ AB + C

(2)

A+BCAB

(3)

i On leave from ChemistryDepartment, Moscow State University, 119899 Moscow, Russia.

Here, Eqs. (1) and (2) describe the real processes taking place in the presence of the catalyst. Eq. (3), obtained by summing the first two equations, corresponds to the general process. As a result of fast intermediate reactions, catalysts allow chemical equilibrium to be achieved by pathways having significantly lower activation energies than those for the uncatalyzed process. The application of catalysts to the improvement of the selectivity of processes that are extremely important, for instance, in petrochemistry, does not appear to be of real significance for the reactions in graphite furnaces used for electrothermal atomic absorption spectrometry (ETAAS). Here it is more important to achieve a sharp increase in the reaction rate at a specific temperature caused by a catalyst. For example, fast decomposition of highly volatile analyte compounds in the pyrolysis stage significantly reduces

0584-8547/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved PI1 S0584-8547(96)01545-5

1574

A.B. Volynsky / Spectrochimica Acta Part B 51 (1996) 1573-1589

analyte losses. Catalytic acceleration of the rates of processes during the atomization stage may allow the atomization to be carried out at lower temperatures. Catalysis may be either homogeneous or heterogeneous in nature. In the former case, the catalyst is in the same phase as the reactants, for instance, in solution or as a gas. However, heterogeneous catalysis (where either reactants and catalyst are solids, or a solid substance catalyzes a gas-phase reaction, etc.) appears to be significantly more important for the processes taking place in ETAAS. According to the typical classification [2], the most widespread types of heterogeneous catalysts are (1) metals (Fe, Ni, Pd, Pt, Ag); (2) semiconducting oxides and sulphides (NiO, ZnO, MnO2, Cr203, Bi203MOO3, WS2); (3) insulating oxides (A1203, SiO2, MgO); and (4) acids (H3PO4,H2804, SIO2-A1203). It is rather interesting that this list of the most effective catalysts contains Pd, MgO and NiO, which are also the active forms of the most popular chemical modifiers in ETAAS. Except for such specific species as Bi203-MoO3 and WS 2, all other catalysts listed are typical for ETAAS systems occurring as matrix components and/or as analyte compounds. To make the list complete, active carbon should also be included. As the product of the thermal decomposition of many organic compounds and of the thermal/chemical destruction of graphite, active carbon possesses some catalytic properties as well [3]. Highly effective catalytic processes are widespread in nature (e.g. in biochemistry where enzymes are biocatalysts) and in industry. Certainly, they are also common among the reactions taking place at every stage of the thermal programme in graphite furnaces for ETAAS. The most important for industry, and therefore the most intensively studied, catalytic processes (hydrogenation, dehydrogenation, isomerization, alkylation, etc.) sometimes occur in the graphite furnace, but they have little influence on ETAAS results. Investigations of the systems that are of primary importance for graphite atomizers (e.g. the catalytic reduction of element oxides with graphite) have been rather scarce. That is why one can only assume the catalytic nature of some of the processes taking place in the graphite furnace. The most important peculiarity of the catalytic process in ETAAS appears to be the "reversed" ratio of the catalyst to the reactants [4]. According to Eq. (1), a

catalyst is one of the reagents taking part in the reaction. Though its concentration is the same at the beginning and at the end of the process, the rate of a chemical reaction is proportional to the concentration of the catalyst in accordance with the rate law. As a general rule, the concentration of a catalyst is lower than that of the reactants. However, under typical ETAAS conditions, the quantity of chemical modifier is several orders of magnitude higher than that of the analyte. Generally, a specific interaction between modifier and matrix may significantly decrease the quantity of modifier that can interact with the analytes. However, if the quantity of modifier is rather large (or if its interaction with matrix components is weak), the rate of processes involving the analyte, catalyzed with a modifier, may be extremely high. Therefore, the investigation of catalytic processes in graphite furnaces should be a very important part of fundamental research in this domain. Catalytic mechanisms were invoked as explanations of the processes in graphite furnaces in 1977 [5-7]. After a rather long "induction" period, the ideas of catalysis became rather popular among specialists working in ETAAS. In this review an attempt has been made to analyze published data and to discuss directions of research with the most prospects.

2. Platinum group metals (PGMs) as chemical modifiers

The first proposal that the action of palladium modifier is partially based on its catalytic properties was made in 1988 [8]. Subsequently, Rettberg and Beach [9] noted a similarity between the reduction of palladium modifiers on the surface of the graphite tube using hydrogen or graphite as reductants and the procedures used for the preparation of the supported catalysts, including Pd catalysts. Further research has shown that the role of catalytic processes in the overall action of PGM modifiers is very important. 2.1. Reduction o f the analyte oxides

Platinum group metals are well-known catalysts for the reduction of metal oxides [10,11]. For instance,

A.B. Volynsky / Spectrochimica Acta Part B 51 (1996) 1573-1589

the reduction of V205 by methane starts at 500°C in the absence of a catalyst and at 270°C in the presence of platinum [12]. If catalyzed by palladium, the reduction of V205 by ethylene and o-xylene starts at 250°C and 320°C, respectively [13]. When carbon monoxide is used as a reducing agent, without a catalyst, the reduction of V205 starts at 575°C; in the presence of palladium it starts at 25°C [10]. Analogous data have been obtained for the reduction of Fe203 and Fe304 [14,15], MoO3 [16], NIO[11,17], CuO [10], PbO2, Pb304, SnO2 and Re207 [15] by hydrogen. According to Frety et al. [15], the higher the oxidation state of the metal in its oxide, the sharper the catalytic effect of platinum on the reduction. Information concerning catalytic reduction by carbon is significantly more scarce. For the reduction of ZrO2 catalyzed by platinum, considerable formation of Ph_xZrx alloy was observed at temperatures higher than 600°C [18]. According to II'chenko [10], a limiting step in the reduction of an oxide by molecular hydrogen is electron transfer from the adsorbed molecules of the reducing agent to the solid oxide. The excellent catalytic activity of palladium and platinum in these reactions is based, firstly, on their ability to adsorb hydrogen rapidly and to dissolve it even at low temperatures and, secondly, their having the lowest overpotential for the cathodic liberation of hydrogen [19]. This second property of the PGMs may also be important for facilitating electron transfer from reducing agents other than hydrogen (e.g. CO, CH4, C2H4, o-xylene and carbon) to the metal oxides. The proposal that the reduction of analyte oxides by the graphite of the atomizer, catalyzed by elemental palladium, precedes the formation of solid solutions between analytes and palladium in the graphite furnace, was published in 1989 [20]. For the verification of this hypothesis, the reduction of lead and gallium oxides by graphite powder in the presence of palladium and nickel chlorides was studied. Fourier transform-infrared (FT-IR) spectroscopy was used for the simultaneous monitoring of the gaseous reaction products (CO and CO2) during heating of the reaction mixture in the temperature range 100-1000°C [4]. As the sensitivity of FF-IR spectroscopy is relatively low in comparison with that of ETAAS, the experiments were carried out using milligram quantities of the reactants. It has been shown that palladium chloride sharply decreases the temperature of reduction of

1575

the metal oxides by the graphite. Under the experimental conditions used, the initial temperature of the reduction of PbO is decreased from 430°C without a modifier to 340°C in the presence of PdC12; the corresponding data for Ga203 are 735°C and 360°C. Zhuang et al. [21] studied the behaviour of Zn and Cd in the presence of a mixture of palladium chloride and citric acid as a chemical modifier. On the basis of the shapes of the absorption signals observed, they also concluded that palladium catalyzes the reduction of analyte oxides by carbon. In addition to the catalytic reduction of lead oxide [22], the existence of analogous reactions at the pyrolysis stage was assumed for In, Sn, Ag, Mn, Co and TI oxides [23-27]. Palladium and nickel chlorides not only facilitate the reduction of oxides, but they may also change its mechanism. Without the catalyst, the probable products of Ga203 reduction with graphite are CO [4] and Ga20 [28-30] Ga203(s) + 2C(s) ~ Ga20(g) + 2CO(g)

(4)

In the presence of NiC12, and especially PdCI2, the main gaseous product of the reaction is carbon dioxide [4]. The pronounced increase in the maximum pyrolysis temperature for gallium [31] suggests that, in the presence of metal modifiers, the second product of the reduction of Ga203 is metallic gallium (b.p., 2403°C) 2Ga203(s ) + 3C(s) ~ 4Ga(1) + 3CO2(g)

(5)

This proposal correlates well with the conclusion drawn for the catalytic reduction by hydrogen, that if the reaction is carried out at a sufficiently high temperature, the metallic catalyst facilitates a greater degree of reduction of the metal as its oxide (certainly, within the limits allowed by thermodynamics) [10]. Qiao et al. [32] assumed that the stabilization of thallium in the presence of prereduced palladium is caused by the process of underpotential deposition. They referred to the work of Mallat et al. [33], devoted to the investigation of the poisoning of the palladium catalyst in the system "palladiumcompound of Ge, Hg, Pb, etc.-hydrogen". Mallat et al. [33] demonstrated that ions of the poisoning elements are reduced to the free elements by hydrogen on the surface of the palladium acting as a catalyst. Thus, the stabilization of analyte compounds by

1576

A.B. Volynsky / Spectrochimica Acta Part B 51 (1996) 1573-1589

underpotential deposition is simply another term for their catalytic reduction [4]. The succession of chemical reactions in the presence of modifiers on the base of PGMs [4,21,34,35] can be written as [36]: MC1k + nC* ---* M + C,,CIk

(6)

AmOn + g(bulk) -"* AmOn(ad)'M(bulk)

(7)

AC1 m + M(bulk) "--*AClm(ad ).M(bulk)

(8)

AmOn(ad ) + n/2 C ~ mA +n/2 CO 2

(9)

AmOn(ad ) + nC _.M mA +nCO

(10)

AClm(ad) __.MA + m / 2 Cl2

(11)

kA + nM ---*AkM,,

(12)

where M represents the modifier, A is the analyte and C* represents the active sites on the graphite surface. PGM modifiers are presented here as their chlorides because this situation is the most widespread [37]. If modem graphite furnaces are used, one can confidently assume that the products of the interaction between analytes and PGM modifiers decompose completely in the atomization stage. The effectiveness of platinum metal compounds as chemical modifiers will then be determined by processes (6)-(12) [38]. In particular, this scheme clearly indicates some reasons for the extreme effectiveness of pre-reduced palladium as a chemical modifier [37,39,40]. As a rule, the palladium surface is covered with a layer of adsorbate [41,42]. In comparison with other palladium modifiers, a surface of pre-reduced palladium provides the best accessibility for chemisorption. Variations in the catalytic properties of elemental PGMs may lead to the reduction of analyte compounds at different temperatures. The higher the temperature of formation of elemental analyte, the larger the analyte losses in the form of its volatile compounds. Though the catalytic properties of platinum metals relating to the real processes in graphite furnaces have not yet been systematically studied, for the reduction of metal oxides by hydrogen

the effectiveness of palladium is always higher than that of platinum [14,16]. For instance, the reduction of MoO3 by hydrogen starts at 185°C and 260°C in the presence of PdCI2 and PtCI4, respectively [16]. In 1989, Rettberg and Beach [9] noted that the size and surface structure of the palladium particles may influence their catalytic properties relative to the processes in the graphite furnace. For the reduction of Ni(II) ions in aqueous solutions, clusters of palladium, platinum, rhodium and iridium having a diameter of less than 1 nm are all catalytically active [43]. However, an increase in their size to 2 nm leads to the complete disappearance of catalytic properties for platinum, rhodium and iridium. Palladium retains its catalytic activity even in the form of a solid film. In the work of Del Angel et al. [44], supported rhodium and palladium catalysts were studied in some reactions with participation of C-H, C-C, C-C1 and C-O bonds. The general trend was a nondependence of the reaction rates upon the size of the palladium particles, whereas the opposite was true for rhodium. In principle, the changes in the size of the modifier particles, caused by the sample matrix and/or temperature programme of the atomizer, may influence the catalytic activity and, consequently, the modification efficiency of the PGMs other than palladium. However, the influence of particle size on the catalytic properties of supported metals is very complex in character [45,46]. Systematic investigations are necessary in order to evaluate the importance of this factor for catalytic processes in the graphite furnace. For the PGM-catalyzed reduction of metal oxides by hydrogen, II'chenko [10] postulated a so-called "saturation effect". At fairly low concentrations of platinoid, the rate of the limiting step (and, consequently, the rate of the overall process) increased proportionally with the content of catalyst. However, at a particular concentration (for the reduction of vanadium pentoxide it was 0.10-0.25% Pt), the rate of the former reaction reached such a value that a second reaction became the slowest step in the process, e.g. diffusion in the solid phase. The possible change in the rate limiting step with increase in the catalyst concentration should be kept in mind when the activation energies of the corresponding processes in graphite furnaces are determined.

A.B. Volynsky/ SpectrochimicaActa PartB 51 (1996)1573-1589 2.2. Reduction (or dissociation) of other analyte compounds Sakurada et al. [47] studied the behaviour of lead and palladium chlorides on the surface of a graphite tube, using X-ray photoelectron spectroscopy. According to their data, in the absence of the modifier, PbCI2 remains unchanged during heating up to 400°C and evaporates from the graphite tube at higher temperatures. By the addition of PdCI2, Pd(II) and Pb(II) were found to be reduced to the metallic states already at 200°C. Analogous processes may be responsible for the increase in sensitivity observed when palladium modifiers were used for the determination of analytes in the form of their volatile halides. For instance, palladium did not influence the absorption signals of bismuth as Bi(NO3)3 in dilute nitric acid [48]. However, for bismuth iodide (b.p., about 500°C) in methyl isobutyl ketone, 1 /xg of palladium increased the bismuth absorbance by a factor of about 13. Tserovsky et al. [49] postulated that palladium and platinum modifiers catalyze the reduction of diethyldithiocarbamates of volatile elements in the graphite furnace to the free elements.

2.3. Decomposition of volatile hydrides In 1989, Doidge et al. [50] used palladium, platinum and nickel to enhance the trapping efficiency of germanium hydride in the graphite tube. The application of palladium as a trapping substrate, which was the best of the modifiers studied, permitted a decrease in the characteristic masses from 5.5-fold for bismulth and tellurium to 95-fold for germanium. Sturgeon et al. [51] reported that, in the presence of pre-reduced palladium, a deposition temperature of 200°C allowed the effective trapping of volatile AS, Se, Sn, Bi and Sb hydrides in the graphite tube. This temperature is significantly lower than those established for the deposition of the analyte hydrides on the pure graphite surface (510°C, 680°C and 300°C for AS, Sb and Bi, respectively) [52]. Analogous results were obtained for other PGMs studied (platinum, ruthenium and rhodium). The authors assumed that the hydrides underwent catalytic dissociation on the PGM surface, rather than thermal decomposition. Rapid, dissociative chemisorption of arsenic hydride onto

1577

palladium film was reported to occur even at -80°C [53]. Similar results were obtained for the chemisorption of selenium hydride on platinum and tungsten films [54]. A comparison of the effect of various metals on in situ preconcentration of the volatile hydrides in a graphite furnace was also presented in Refs. [51,55-57]. As a rule, PGMs are the most effective modifiers and palladium is the optimal. However, some distinctions were also observed. For instance, using palladium, nickel and platinum in the graphite tube enhanced the sensitivity for germanium by factors of 95, 57 and 32, respectively [50]. For the determination of arsenic, the effectiveness of mercury, copper, iron and nickel was very low, and for the other metals studied it decreased in the order Pd > Ru > Ce > Ag > Pt [56]. Palladium, platinum and iridium similarly influenced the determination of arsenic, antimony and selenium, but the sensitivities in the presence of palladium were the best [55]. Sturgeon et al. [51] could not use compounds of gold, nickel and copper for the trapping of arsenic hydride even after their pretreatment in the graphite tube at 1200°C for reduction to the elemental metals. For As, Se, Sn, Bi and Sb hydrides, the typical order of the modifier effectiveness was Pd > Pt > Rh > Ru [51]. AI-Daher and Saleh [53] observed rather good decomposition of arsenic hydride on the surfaces of iron, nickel, palladium and tungsten. Moreover, according to their data, the uptake of AsH3 on oxidized films of these metals was more extensive than on clean films. Significant discrepancies between data from Refs. [51] and [53] may be caused by the differences in the reaction conditions applied. At present, the catalytic theory of the trapping of volatile hydrides in the graphite tube when metal modifiers are used is generally accepted [22,56-60]. Sturgeon et al. [51] assumed that the superiority of platinoids over other metals is based on the ability of PGMs to participate in a hydrogen abstraction reaction. The excellent performance of palladium modifiers may be the result of an enhanced affinity of this metal for molecular hydrogen [61].

2.4. Decomposition of organoelement compounds In 1981 Weibust et al. [62] reported that of nine metals studied, only Pd, Pt and Ag compounds

1578

A.B. Volynsky / Spectrochimica Acta Part B 51 (1996) 1573-1589

effectively stabilized tellurium when present in the organometallic form in the graphite furnace. The application of palladium nitrate as a chemical modifier for the analysis of organotin compounds permitted a significant decrease in the influence of their composition on the magnitude of tin absorption [20]. It was assumed that metallic palladium facilitated the decomposition of organotin compounds at 150-200°C. Further research supported the effectiveness of palladium modifiers for the determination of different organoelement compounds. Matsumoto and coworkers [63-65] proposed the use of organopalladium compounds that are soluble in organic media for this purpose. The best of six organopalladium compounds studied, PdC12(CH3CN)2, enhanced the sensitivity for tin by a factor of 100-1000, resulting in similar sensitivities for the different organotin compounds studied. Prereduced palladium and a mixture of PdC12 and Mg(NO3)2 improved the sensitivity for lead, as Pb(CEH5)4 [66], and for arsenic when present as different organoarsenic compounds [67], respectively. A drastic increase in the trapping efficiency was observed in the presence of palladium for Se(CH3) 2 (b.p., 54-55°C), Se(C2Hs)2 (b.p., 108°C) and Se2(CH3)2 (b.p., 153°C) [68,69]. Although heterogeneous catalysis of the thermal decomposition of organoelement compounds has not been studied specifically, it is known that about 10 g of tetraphenyltin are completely decomposed at 150°C within 17 h in the presence of metallic nickel [70]. The chemisorption of acetone, methanol, formaldehyde, dimethyl ether and acetaldehyde on polycrystalline palladium is accompanied by their decomposition at 25°C and the formation of chemisorbed carbon monoxide [71]. The chemisorption of organic molecules on palladium surfaces occurs primarily through the lone-pair orbitals associated with the oxygen atoms. Evidently, an analogous process may lead to chemisorption and subsequent decomposition of oxygen-containing organoelement compounds. Transition (and especially platinum group) metals are effective catalysts for hydrocarbon cracking [72]. The breakage of C - C and C - H bonds during the pyrolysis of organic compounds occurs primarily on defects on the catalyst surface [73]. As the strength of M e - C (where Me is metal or metalloid) bonds is

always lower than that of the C - C bond [74], their breakage by PGMs is quite possible. The behaviour of organoelement compounds in the presence of PGMs is strongly influenced by the nature of both substances and by the reaction conditions. For instance, the thermal decomposition of volatile Cr(III) chelate with trifluoroacetylacetone [75] and nickel carbonyl [76] is evidently not catalyzed by palladium. According to Laborda et al. [77], reduced palladium (formed either by thermal pretreatment or by use of chemical reagents) is about 50% less effective for the determination of selenium, as trimethylselenonium chloride, in comparison with nickel or palladium nitrates. Finally, Nixon et al. [78] obtained the best results for the determination of arsenic in the form of different organoarsenic compounds with a mixture of Pd(II) with potassium persulphate as chemical modifier. Potassium persulphate definitely promoted the oxidation of volatile organoarsenic compounds in situ. On the other hand, it also prevented the reduction of palladium compounds to metallic palladium in the graphite furnace. These data clearly show that detailed investigations are necessary for the optimization of the catalytic decomposition of organoelement compounds in graphite furnaces for ETAAS. 2.5. Oxidation of graphite and carbon monoxide

The oxidation of graphite and carbon monoxide, including catalytic reactions, by oxygen and water has been thoroughly studied [79-82]. However, the influence of these reactions on the processes taking place in graphite tubes for ETAAS is understood significantly worse. On the basis of mass-spectrometric data, Styris et al. [83] assumed that palladium catalyzes the oxidation of graphite (or carbon monoxide) by water vapour at about 130°C C(CO) + H20 ~ CO(CO2) -I-H 2

(13)

The low-temperature decomposition of water causes the disappearance of selenium hydroxides in the gas phase of the atomizer in the presence of palladium modifiers. The palladium-catalyzed oxidation of graphite results in the formation of channels in its structure [79,81]. According to Styris et al. [83], selenium may be trapped in these channels at temperatures up

A.B. Volynsky/ SpectrochimicaActa PartB 51 (1996) 1573-1589

to 880-1280°C. An analogous proposal was made by Majidi and Robertson [84] based on their data, which was obtained by Rutherford backscattering spectroscopy. Another probable consequence of an accelerated interaction between graphite and oxygen in the graphite tube in the presence of PGM modifiers is the decrease in the free oxygen content in the gas phase of the atomizer. Peile et al. [85] observed spectral interferences during the determination of selenium in the presence of high quantities of iron, probably caused by gaseous iron monoxide. The interferences completely disappeared when pre-reduced platinum was used as a chemical modifier along with a mixture of argon with hydrogen and carbon monoxide as a sheath gas. This effect was ascribed to the acceleration of the reaction

1579

carbide, but also on the procedure used for the modification [89]. Chemical and physical vapour deposition result in a solid layer of carbide formed on the surface of the graphite tube [90-92]. This completely prevents contact of the graphite with the analytes. However, the application of these modification methods requires rather expensive equipment. Therefore, the use of methods based on the treatment of graphite tubes with aqueous solutions of W, Zr, Ta, etc. salts is significantly more popular. As a rule, such procedures do not lead to the formation of a solid carbide layer on the graphite tube surface [89]. It is likely that the carbide formed acts as a low-volatility modifier that is removed from the atomizer very slowly [93]. The processes occurring in carbide-modified graphite tubes are discussed below. 3.1. Some chemical properties of refractory carbides

2CO + 02 ~ 2CO2

(14)

catalyzed by platinum. A decreased free oxygen content of the gas phase of the atomizer promoted better dissociation of iron monoxide. However, the effect observed might also be caused by catalytic acceleration of the process 2H 2 + 02 ~ 2H20

(15)

3. Graphite tubes modified with refractory carbides The modification of graphite tubes with refractory carbides was proposed in 1973 by Kuzovlev et al. [86]. Although the number of papers concerned with carbide-modified graphite tubes [87] exceeds 400, the mechanisms of the processes occurring in them are not well understood. Some of the processes in such tubes appear to be catalytic in nature. According to the classification presented in Ref. [88], metal-like carbides (TaC, WC, ZrC, etc.), covalent carbides of silicon and boron, and salt-like covalent-metallic lanthanum carbides may be applied to the modification of graphite tubes [89]. Further, the term refractory carbides refers to metal-like carbides, as they are used most frequently for this purpose. Properties of carbide-modified graphite tubes depend not only on the properties of the respective

Carbides of Zr, Ta, Nb, W, etc. possess both excellent thermal stability and mechanical properties. On this basis, it is assumed that their chemical inertness is also extremely high. However, this may not be true. Under typical ETAAS conditions, Sturgeon et al. [94] observed that oxidation of the pyrolytic graphite coating starts at about 1000°C. As a rule, the refractory carbides are oxidized by oxygen at lower temperatures [95]. According to Cizek et al. [96], graphite powder starts to interact with pure oxygen above 780°C. Under the same conditions, ZrC, VC, NbC, WC, TaC and HfC react with oxygen at 510-530°C, 550-570°C, 590-610°C, 650-680°C, 720-740°C and 750-770°C, respectively. Changes in the experimental conditions may influence this order, but the general tendency is maintained [95]. At relatively high temperatures, the refractory carbides may act as reductants with respect to the element oxides [97,98]. 3.2. Catalytic properties of the refractory carbides

Levy and Boudart [99] showed that the addition of carbon to non-noble metals (such as molybdenum and tungsten) confers to them some of the catalytic properties typical of platinum and palladium. The reason for this is the resemblance of the electronic structure of transition metal carbides to that of PGMs. Yang and Wong [100] reported that all refractory carbides

1580

A.B. Volynsky / Spectrochimica Acta Part B 51 (1996) 1573-1589

studied by them (WC, TaC and Mo2C) catalyzed the oxidation of single crystals of natural graphite by oxygen (20% in argon; p = 1 atm) at 680°C. According to their data, transition metal carbides act as dissociation centres for molecular oxygen generating highly active atomic oxygen. However, the refractory carbides may also promote the catalytic oxidation of graphite via an electron transfer mechanism, as discussed earlier in relation to processes in the presence of PGM modifiers. The third possible mechanism of the catalytic oxidation of graphite in the presence of transition metal carbides involves oxygen transfer. McKee [80] observed catalytic activity in the oxidation of graphite by oxygen only for compounds capable of participating in oxidation-reduction cycles in the temperature range used. For instance, both metallic vanadium and vanadium pentoxide were catalytically active at about 550°C. However, elements having oxides stable at 500-700°C (silicon, aluminium, tungsten and tantalum) were not active in this temperature range. Yang and Wong [100] noticed that different mechanisms of catalysis may be dominating for the gasification of graphite by various gases, and for the same gas at different temperatures. Studies of the catalytic activity of transition metal carbides in other oxidation-reduction processes are less frequent. For instance, it is known that Zr, Nb, Ta, Mo and W carbides promote the reduction of nickel and iron oxides by hydrogen [11]. According to published data, the majority of the reactions discussed above take place in graphite furnaces for ETAAS. 3.3. Reduction of analyte compounds

In 1975, Hocquellet and Labeyrie [101] reported that the sensitivity for tin by ETAAS was better if the graphite tube was modified with tantalum carbide. They assumed that TaC is a more effective reducing agent for Si, Ge, Sn and Pb oxides than graphite. Hodges [6] observed a sharp decrease in the background absorption when MoC-modified graphite tubes were used for the determination of lead in urine. It was assumed that the decreased background was caused by the reduction of alkaline metal phosphates to the respective phosphides catalyzed by molybdenum or its carbide. Later, it was established

that the rate of formation of free silicon atoms increased when the graphite furnace was modifed with niobium carbide [102]. For example, at 1200°C, the respective rate constant is 0.12 min -1 for the standard graphite furnace without a pyrolytic graphite coating, and 0.73 min -x for a modified graphite furnace. The activation energies for the reduction of sodium silicate by graphite and niobium carbide were estimated to be 134.0 kJ mo1-1 and 38.5 kJ mo1-1, respectively. The authors assumed that niobium carbide took part in the reduction of the silicon compounds, for example, by the incorporation of oxygen into its lattice [102]. The most detailed investigations concern the mechanism of the atomization of germanium in graphite furnaces modified with refractory carbides [60,103-105]. Modification of the graphite tubes with WC or ZrC enhanced the sensitivity for germanium up to eightfold and sixfold, respectively. In addition, in the modified graphite tubes, sharper absorption signals were observed for germanium even at lower atomization temperatures. The following processes during the atomization stage have been postulated (for the ZrC-modified graphite tube as an example) [60,103,105] 2ZrC(s) + GeOE(S) --* 2Zr(s) + Ge(s/l) + 2CO(g) (16) ZrC(s) + GeO(g) ---* Zr(s) + Ge(s/l) + CO(g)

(17)

Ge(s/l) ---* Ge(g)

(18)

Zr(s) + C(s) --* ZrC(s)

(19)

On the basis of kinetic investigations, analogous processes in graphite tubes modified with zirconium carbide were subsequently assumed for lead [22], indium [23] and tin [24]. Summarizing Eqs. (16)-(19), one obtains GeOE(S) + 2C(s) z_.~cGe(g) + 2CO(g)

(20)

GeO(g) + C(s) ~

(21)

Ge(g) + CO(g)

It is now evident that the role of the refractory carbides appears to be catalysis of the reduction of analyte oxides by the graphite of the atomizer. Eqs. (16)-(19) characterize the mechanism of catalysis as

A.B. Volynsky / SpectrochimicaActa Part B 51 (1996) 1573-1589

an oxygen transfer, but the data obtained are not sufficient to exclude other possible mechanisms discussed earlier. From a thermodynamic viewpoint, carbon seems to be a better reducing agent than Ta [106] or W [107] carbides. For example, the Gibbs free energy AG260ooc for the reaction 2C(s) + SiOz(s) --* 2CO(g) + Si(g)

(22)

is - 237 kJ mo1-1, and for the reaction 2TaC(s) + SiO2(s) ---, 2CO(g) ÷ Si(g) + 2Ta(s)

(23)

is - 33.2 kJ mo1-1 [106]. The kinetic nature of the enhanced reducing properties of the refractory carbides is the basis of their catalytic activity in the reactions under discussion. 3.4. Free oxygen content in the gas phase of carbide-modified graphite tubes

The molecular absorption of GeO(g) [103,104] and SnO(g) [108] is significantly lower in ZrC-modified graphite tubes than in standard tubes. Sometimes, such phenomena may be caused simply by better thermal dissociation of volatile analyte monoxides. For example, the appearance time for tin [109] and silicon [110] is greater in WC-modified graphite tubes than in standard tubes. Thus, in these cases, the effect of modification with tungsten carbide may be analogous to that produced by the graphite platform [111]. However, an increase in sensitivity for analytes that have relatively stable volatile monoxides (such as Pb, In, Sn, Ge, B, Si) in the carbide-modified graphite tubes is often accompanied by a decrease in the appearance time [20,22-24,109,110,112]. As a rule, such data are considered as arguments confirming the improved reduction of the analyte oxides in the carbide-modified graphite tubes (see above). However, the decrease in the concentration of the gaseous oxides may also be caused by a lower content of free oxygen in the graphite furnace. It is known that the sensitivity for elements having rather strong monoxides (such as Sn and Ge) depends strongly on the flow rate of the sheath gas through the graphite furnace during the atomization stage. For example, under "gas-stop" conditions, the sensitivity for tin is practically the same for the standard graphite tube without a pyrolytic graphite coating and

1581

for WC-modified ones, and more than twofold worse for the graphite tube with a pyrolytic graphite coating [113]. For "full flow" conditions the difference in sensitivities for tin exceeds two orders of magnitude for the graphite tubes studied. The characteristic masses for tin in standard graphite tubes with and without a pyrolytic graphite coating, and for WCmodified graphite tubes are 176 ng, 3.1 ng and 0.6 ng, respectively [113]. For the first two cases, the difference is no doubt caused by the lower reactivity of the pyrolytic graphite, resulting in a higher free oxygen content in the gas phase in the graphite furnace [114]. The same factor may be responsible for the higher sensitivity observed when WC-modified graphite tubes are used. Using the indirect "calibration graph displacement method" [114], Volynsky et al. [113] showed that the free oxygen content in WC-modified graphite tubes is lower than that in a standard tube. This phenomenon is probably caused by an acceleration of the interaction between the graphite of the atomizer and free oxygen for the graphite tubes modified with refractory carbides. Oxygen transfer was assumed to be the mechanism of catalysis (with WC as an example) WC ÷ 0 2 ---*W(WO3) -i-CO(CO2)

(24)

W(W03) + C ~ WC + (CO, C02)

(25)

C.4- 02 ~

(26)

C0(C02)

This decrease in the free oxygen content in carbidemodified graphite tubes in the atomization stage may enhance the sensitivity for lead, indium, tin, gallium, germanium and analogous elements as a result of better dissociation of their relatively stable volatile monoxides [114-116]. Certainly, the respective changes in the sensitivity are more pronounced when the sample is vaporized from the wall of the graphite tube and peak heights are used for the quantification of the analyte. However, even under STPF conditions, the decrease in the free oxygen content may enhance the degree of dissociation of monoxides having dissociation energies of about 650 kJ mol-1 or more (such as GeO, SiO and BO) [93]. The incomplete decomposition of refractory tin carbide during the atomization stage [117,118] is often used as the explanation for the increase in sensitivity

1582

A.B. Volynsky/ SpectrochimicaActa PartB 51 (1996)1573-1589

for tin in carbide-modified graphite tubes [109,112,119]. However, under normal conditions, tin does not form carbides at all [120,121]. The formation of such compounds was observed only under a pressure of about 50 kbar [122]. The modification of the graphite tubes may enhance the sensitivity for elements that actually form refractory carbides, such as silicon, yttrium and boron [110,123-125]. A typical explanation for this phenomenon is the prevention of the formation of refractory analyte carbides in the carbide-modified graphite tubes [30,120,123,125,126]. However, according to [127,128], the formation of solid BnC (m.p., 2350°C) does not influence the sensitivity for boron by ETAAS. The dissociation energies for monoxides of B and Si are larger than those of their monocarbides (DoB_.o=800 kJ mo1-1, DoB_.c=444 kJ mo1-1, Do si-o--783 kJ mo1-1, D o sic--428 kJ mo1-1) [129]. Certainly, these data cannot be used for the evaluation of the atomization mechanism without taking into account the actual concentrations of oxygen [114,130] and carbon [28,131] in the graphite tube during the atomization stage. However, they allow one to assume that the reason for the increased sensitivity for silicon and boron in the carbide-modified graphite tubes is better dissociation of stable gaseous monoxides of the analytes during the atomization stage (for example, SiO(g)---* Si(g) + O(g)[132]), caused by a lower free oxygen content in the gas phase of the atomizer [93]. Catalysis of the interaction between graphite and oxygen by the refractory carbides leads to larger weight losses for the carbide-modified graphite tubes as compared to the standard tubes [113]. The lowest weight losses were registered when carbon tetrachloride was added to the standard graphite tube before every atomization cycle. This effect was explained as a retardation of the oxidation of graphite by oxygen in the presence of chlorine compounds [801. The increased weight losses for the carbidemodified graphite tubes are accompanied by their prolonged lifetime [113]. This seeming contradiction is explained by the changed geometry of the carbide-modified graphite tube burn-out [93,113]. As a rule, standard graphite tubes fracture in their centre [113,133]; moreover, in the absence of strong acids in the sample solution the deterioration is observed to

occur mainly from the outer surface of the tube [93,113]. The evident reason for this is consistent with keeping the full-flow mode of the sheath gas outside the tube during the atomization stage. Free oxygen entrained in the sheath gas preferentially interacts with graphite in the centre of the tube which is at the highest temperature. For the carbidemodified graphite tubes, the catalytic effect causes a reduction of the initial temperature at which oxygen and graphite start to interact [80]. Therefore, oxygen is "removed" from the sheath gas as soon as it reaches the tube surface (for HGA-type atomizers, this occurs not far from the ends of the graphite tubes) [93,113]. The "pre-cleaned" sheath gas obtained destroys the centre of the carbide-modified graphite tube to a lesser extent than the starting gas. 3.5. Decomposition of organoelement compounds

In 1980, Vickrey et al. [134,135] investigated the determination of organotin compounds using ETAAS. For standard graphite tubes without a pyrolytic graphite coating, an almost thirtyfold difference in sensitivity for tin in the form of nine organotin compounds studied and tin tetrachloride was observed. However, when the graphite tubes were modified with zirconium carbide, this difference did not exceed 40%. Moreover, the average increase in sensitivity for tin was more than threefold [135]. Analogous results were also obtained for VC-, MoC-, TaC- and WCmodified graphite tubes [20,117,118,134,136]. For example, Bums et al. [136] successfully used ETAAS with TaC-modified graphite tubes for the direct analysis of 11 organotin compounds. Relatively low sensitivity was observed for only three compounds with Sn=O bonds. Modification of the graphite tubes with a pyrolytic graphite coating by zirconium carbide increased the sensitivity for lead, as (C2Hs)4Pb, about seventeenfold and significantly decreased the differences in sensitivity for lead for five organolead compounds studied [137]. The only explanation published for the effectiveness of the carbide-modified graphite tubes with respect to organotin and organolead compounds was their low-temperature decomposition on solid metal oxides [134,135]. Oxide species of Ta, Zr, etc. may be formed in the graphite furnace as a result of the interaction between the respective carbide and residual

A.B. Volynsky / Spectrochimica Acta Part B 51 (1996) 1573-1589

oxygen in the sheath gas and/or traces of oxygen-containing solvents, such as water or methanol. It is known that organotin compounds may easily be decomposed by acids [138,139]. Vickrey et al. [134,135] assumed that transition metal oxides formed on the graphite furnace surface are identical to the solid "superacids" [140] and therefore may promote the decomposition of organoelement compounds. This proposal may not be correct. The preparation of a ZrO 2-SO 2- superacid is based on the treatment of Zr(OH)4 with 0.5 M H2804, drying and subsequent pyrolysis in air at 500°C for 3 h [140]. Certainly, this procedure differs sharply from the interaction of zirconium carbide with oxygen in the absence of sulphuric acid. Without special treatment, zirconium oxide possesses only weakly acidic and weakly basic properties [141]. It is more likely that the enhanced effectiveness of the carbide-modified graphite tubes is the result of PGM-like properties of some transition metal carbides [99]. If this assumption is true, the successful decomposition of organotin and organolead compounds is analogous to that observed in the presence of PGM modifiers (see above). Though the exact mechanisms of these processes are not yet known, some practical recommendations may be proposed. According to Reutov et al. [138], the rates of protolysis of (C6Hs)E(C2H5)3, where E is Si, Ge, Sn and Pb, by perchloric add, correlate as 1:36:3.5 x 105:2 x 108, respectively. In practice, carbide-modified graphite tubes are rather effective for the determination of organotin and organolead compounds that have relatively weak Me-C bonds [74]. Attempts to determine aluminium and germanium in a number of organoelement compounds using TaC-modified graphite tubes failed [142,143]. The volatilization of Cr(III) chelate with trifluoroacetylacetone was even better from the surface of WC-modifled graphite tubes than from standard tubes [144]. As discussed earlier, palladium modifier did not improve the decomposition of the same chelate in the graphite tube [75]. 3.6. Other catalytic p r o c e s s e s

There is little doubt that the effective trapping of lead and tin hydrides on the wall of a ZrC-modified graphite tube at 100°C and 400°C, respectively, is a result of catalytic decomposition [145,146]. In

1583

comparison to a palladium modifier possessing analogous properties, zirconium carbide is less volatile and permits simplification of the procedure of analysis. However, zirconium carbide is rather ineffective for the trapping of germanium hydride [60,147]. Graphite tubes modified with refractory carbides are significantly better than standard ones for the analysis of organic extracts [89]. This may be due to the catalytic decomposition of the volatile chelates of the analytes on active sites on the surface of the former tubes (for tin chelates, this explanation was proposed in Ref. [148]). The diminution of the negative influence of chlorine-containing organic solvents, such as CC14 and CHC13, on the sensitivity for nickel [149], tin [148] and other elements when carbide-modified graphite tubes were used may be caused by the fairly easy decomposition (or reduction) of the respective chlorides. Analogous effects were observed in the presence of palladium modifiers (see above). Sturgeon and Chakrabarti [5] assumed that carbideforming elements (zirconium, lanthanum, silicon, vanadium, boron, etc.) catalyze the graphitization of disodered carbon in the graphite tube. At first, disordered carbon dissolves into the refractory carbides at temperatures above 1700°C and then precipitates as non-reactive, small graphite crystallites. This hypothesis was based on the well-documented data on the catalytic graphitization of carbons [150,151] and therefore it was rather popular [123,152]. However, the reaction conditions used in the works devoted to the investigation of catalytic graphitization differ significantly from those existing in the graphite tubes for ETAAS. For example, in the work described in Ref. [151], mixtures of carbon with the metals were heated for 1 h at 2600°C and 10 min at 3000°C. According to scanning electron microscopy results, lanthanum compounds used as chemical modifiers in ETAAS do not facilitate catalytic graphitization in the graphite tubes [153]. Though these data cannot be applied to other refractory carbides that do not react with water, as lanthanum carbides do, experimental verification of catalytic graphitization in the graphite tubes for ETAAS is evidently absent.

4. Catalytic processes of a different nature

Among catalytic processes influencing the lifetime

1584

A.B. Volynsky / Spectrochimica Acta Part B 51 (1996) 1573-1589

of the atomizer, gasification of graphite by water catalyzed by lanthanum compounds is the most welldocumented. Runnels et al. [154] were evidently the first to note that during the pyrolysis and atomization stages, lanthanum compounds used as chemical modifiers were transformed into their respective carbides. However, as a result of the contact with the subsequently injected aqueous solution, lanthanum dicarbide decomposes, forming lanthanum hydroxide 4La(OH)3 + 11C ---*4LaC 2 + 6H20 + 3CO 2

(27)

LaC2 + H20 --'* La(OH)3 + C2H2 + C2H6

(28)

C + H20 ~ C2H2 + C2H6 + C O 2

(29)

Though these processes undoubtedly take place in graphite furnaces [155,156], no one has drawn attention to their relationship to catalysis. Its peculiarity lies in sharply divided (both in temperature and time) stages of gasification of graphite (Eq. (28)) and the regeneration of lanthanum carbide (Eq. (27)). The acceleration of the corrosion of the graphite tubes in the presence of lanthanum compounds has been well documented using scanning electron microscopy [153,157]. Such processes are especially undesirable for the determination of carbide-forming elements (such as molybdenum [158]) when lanthanum compounds are used as chemical modifiers together with the graphite tubes having a pyrolytic graphite coating. It is interesting that other elements forming hydrolyzable salt-like carbides (aluminium, beryllium, calcium, magnesium, strontium, barium and lanthanoids) [159] may also accelerate the corrosion of graphite tubes under the same mechanisms. Although the thermal stability of the carbides of alkaline-earth elements is lower than that of lanthanum carbide [160], they appear to dissociate incompletely during the atomization stage [29,30,161]. The carbides will then be decomposed by the next portion of the analyzed solution. Certainly, the rate of deterioration of the graphite tube caused by these processes depends greatly upon the nature and concentration of the respective element, material of the atomizer and temperature programme applied, etc. The investigation of the decomposition of salt-like covalent-metallic carbides in graphite tubes appears to be most important for

large quantities of calcium and magnesium, nitrates of which are widely used as chemical modifiers. An additional catalytic process occurring in graphite tubes, convincingly documented by scanning electron microscopy, is graphitization of glassy carbon by Fe, Mo and La compounds [157]. These processes are not discussed here because, at present, glassy carbon is practically not used as a material for electrothermal atomizers applied to AAS. Catalysed by gallium graphitization of the carbon residue formed as a result of the thermal decomposition of organic chemical modifiers was assumed to be a reason for the shift of the gallium absorption signals to higher temperatures [162]. However, subsequent experimental results did not support this assumption [163]. Szydlowski [7] used copper as a chemical modifier for the determination of selenium in the form of its chelate with 2,3-diaminonaphthalene. It was assumed that, in addition to the retention of selenium in the graphite tube during the pyrolysis stage, copper may catalyze the uniform breakdown of the coextracted chelating agent. However, the thermal decomposition of organic chemical modifiers is not catalyzed by gallium oxide [163]. Guevremont et al. [164] observed a sharp decrease in the atomization temperature for cadmium when the tetrasodium salt of ethylenediaminetetraacetic acid was used as a chemical modifier for the determination of cadmium in sea water. It was assumed that NaaEDTA acts either as a reducing agent or a catalyst to bring about the immediate atomization of cadmium at low temperatures. Dedina et al. [165] suggested the following mechanism for low temperature (1200-1600°C) atomization of selenium hydride: Sell 2 + H --* Sell + H 2

(30)

Sell + H --* Se+H 2

(31)

Hydrogen radicals that are necessary for its realization may be formed in the graphite furnace only in the presence of small quantities of oxygen that, consequently, acts as a catalyst for selenium hydride atomization. Possibly analogous atomization mechanisms occur for other semimetal hydrides as well [166]. During their investigation of the mechanism of the thermal decomposition of hydrated copper nitrate by

A.B. Volynsky / Spectrochimica Acta Part B 51 (1996) 1573-1589

mass spectrometry, L'vov and Novichikhin [167] observed the appearance of free oxygen from about 230°C. They assumed that its release was caused by the thermal decomposition of nitrogen dioxide, catalyzed by copper oxide 2NO2 c uo 2NO + 0

2

1585

graphite furnace during the atomization stage was used. However, direct experimental observations refuted the theory in general [171-173]. Therefore, the approach of L'vov and Savin is not analyzed here in detail.

(32) 5. Conclusion

Dabeka [34] observed double absorption peaks for lead when palladium was used as a chemical modifier. The high-temperature peak was especially pronounced when hydrogen was added to the sheath gas. The appearance of this high temperature peak was attributed to the formation of a refractory PbPd-C compound, catalyzed by hydrogen. Some attempts to use a catalytic approach as the explanation for the phenomena observed in graphite furnaces used for AAS were unsuccessful. In an early work, Frech and Cedergren [168] observed an extremely strong influence of sodium chloride on the ETAAS determination of silicon. As the rate of depression of the silicon signals excluded the possibility that stoichiometric chemical reactions (40% and 15%, respectively) occurred, they assumed the existence of some catalytic processes. However, this explanation cannot be true, because catalysts do not influence the equilibrium state. Evidently, the addition of sodium chloride led to the appearance of other processes which were not taken into account. Imai et al. [169] investigated the influence of organic matrices on the atomization mechanism of gallium. They reported that, in the absence of an organic matrix, the first stage of gallium atomization involves thermal dissociation of its oxide Ga203(s) ~ Ga20(g) + O2(g)

(33)

It was assumed that the active carbon, formed as a result of the thermal decomposition of the organic matrix, catalyzes this process. However, for the description of the catalytic reaction, an equation describing carbothermal reduction ((Eq. (4)) was used. Evidently, the authors of the above work confused a catalytic acceleration of the reacton rate with a change in the reaction type. L'vov and Savin [170] used an autocatalytic approach for the description of the carbothermal reduction processes leading to the formation of spikes of absorption signals when slow heating of the

It is evident from the data discussed that fundamental investigations of catalytic processes occurring in graphite furnaces used for AAS are absent, and systematic applied researches are rather rare. In spite of this, analysis of the results already published from the viewpoint of catalysis allows one to make some nonevident generalizations. Most importantly is the explanation of the very similar behaviour of platinum group metals and the refractory carbides as chemical modifiers as a consequence of the resemblance in their catalytic properties [99]. To date, these two very important classes of chemical modifiers have been studied practically independently. The investigations of the catalytic processes in graphite furnaces may be important both from theoretical and experimental points of view. For instance, both pre-reduced palladium and a "permanent" iridium modifier are typical examples of supported catalysts [1,2]. The characteristics of such catalysts depend strongly upon the catalyst-support interaction [44,46]. With respect to graphite furnaces used for AAS, the changes in this interaction may influence the modifier effectiveness when PGM compounds are applied together with WC- or ZrC-modifled graphite tubes [174]. The low efficiency of chemical modifiers possessing catalytic properties may sometimes be due to their poisoning by sample matrices. Wide application of the methodology and approaches developed for specific catalysis purposes relating to ETAAS investigations is also in prospect. To be sure, it is sufficient to compare the methods used for catalysis characterization [175] and for investigation of the processes in graphite furnaces [29,176]. In addition to their evident overlap, a number of other methods, which are common in catalysis, may provide valuable information concerning the form of the analytes, matrices, modifiers and the graphite of the atomizer at the different stages of the thermal programme. For example, the processes of chemisorption

1586

A.B. Volynsky / Spectrochimica Acta Part B 51 (1996) 1573-1589

and physical adsorption that play an important role in the stabilization of the analytes in the graphite furnace, especially at relatively low temperatures [35,51], are little studied. However, the methodology relating to such investigations is developed in detail concerning the form of the reagents on surfaces of catalysts of various types. Concluding the review, one may also notice that some achievements in ETAAS might be of theoretical interest to specialists in catalysis (see the title of Ref. [177]).

6. Note added to proof An alternative mechanism explaining periodic absorption spikes when large masses of analytes are slowly heated in the graphite tube, proposed by Ohlsson et al. [178,179], is also based on autocalysis like the original theory of L'vov and Savin [170]. However, in the former hypothesis, the decisive role belongs to the volatile oxides, not to the carbides of the analytes, as was assumed by L'vov and Savin. Although this new explanation is better supported by direct experimental observations [171], it also seems to contain some contradictions [180].

Acknowledgements The author thanks Professor V. Krivan for his support of this work and Dr. R.E. Sturgeon for editing the mannscript. The author also thanks Dr. D.C. Baxter and Prolessor W. Frech for their valuable suggestions. The Alexander von Humboldt Foundation (Bonn, Germany) is acknowledged for receipt of a research fellowship.

References [1] R.L. BurweU, Jr., Pure Appl. Chem., 46 (1976) 71. [2] G.C. Bond, Heterogeneous Catalysis: Principles and Applications, Clarendon Press, Oxford, 1974. [3] H.P. Boehm, in C. Monterra, A. Zecchina and G. Costa (Eds.), Structure and Reactivity of Surfaces, Proc. Europ. Conf., 1988, Elsevier, Amsterdam, 1989, p. 145. [4] A. Volynsky, S. Tikhomirov and A. Elagin, Analyst, 116 (1991) 145. [5] R.E. Sturgeon and C.L. Chakrabarti, Anal. Chem., 49 (1977) 90.

[6] D.J. Hodges, Analyst, 102 (1977) 66. [7] F.J. Szydlowski, At. Absorpt. Newslett., 16 (1977) 60. [8] A. Brzezinska-Paudyn and J.C. Van Loon, Fresenius' Z. Anal. Chem., 331 (1988) 707. [9] T.M. Rettberg and L.M. Beach, J. Anal, At. Spectrom., 4 (1989) 427. [10] N. I. II'chenko, Russian Chem. Rev., 41 (1972) 47. [11] H. Charcosset and B. Delmon, Ind. Claim.Belg., 38 (1973) 481. [12] V.Yu. Bychkov, M.Yu. Sinev, V.N. Korchak, E.L. Aptekar' and O.V. Krylov, Kinet. Katal., 27 (1986) 1190. [13] C. Blejean, P. Boutry and R. Montarnal, C.R. Acad. Sci., Ser. C, 270 (1970) 257. [14] K.M. Sancier and S.H. Inami, J. Catal., 11 (1968) 135. [15] R. Frety, C. Bolivar, H. Charcosset, L. Tournayan and Y. Trambouze, in P. Barret (Ed.), Reaction Kinetics in Heterogeneous Chemical Systems, Proc. 25th Int. Meet. Soc. Chim. Phys., 1974, Elsevier, Amsterdam, 1975, p. 681. [16] A.J. Hegediis, K. Sasvari and J. Neugebauer, Z. Anorg. Allg. Chem., 293 (1958) 56. [17] W. Verhoeven and B. Delmon, C.R. Acad. Sci., Ser. C, 262 (1966) 33. [18] R. Szymanski and H. Charcosset, J. Mol. Catal., 25 (1984) 337. [19] B.E. Conway and J.O'M. Bockris, J. Chem. Phys., 26 (1957) 532. [20] E.T. Kabeshova, A.B. Volynskii and A.N. Kashin, J. Anal. Chem. (USSR), 44 (1989) 1600. [21] Z-X. Zhuang, P.-Y. Yang, J. Luo, X.-R. Wang and B.-L. Huang, Can. J. Appl. Spectrosc., 36 (1991) 9. [22] X.-P. Yan and Z.-M. Ni, Spectrochim. Acta, Part B, 48 (1993) 1315. [23] X.-P. Yan, Z.-M. Ni, X.-T. Yang and G.-Q. Hong, Talanta, 40 (1993) 1839. [24] Z. Quan, Z.-M. Ni and X.-P. Yan, Can. J. Appl. Spectrosc., 39 (1994) 54. [25] Y.-Z. Liang and Z.-M. Ni, Spectrochim. Acta, Part B, 49 (1994) 229. [26] B. Gong, H. Li, T. Ochial, L. Zheng and K. Matsumoto, Anal. Sci., 9 (1993) 723. [27] P.-Y. Yang, Z.-M. Ni, Z.-X. Zhuang, F.-C. Xu and A.-B. Jiang, J. Anal. At. Spectrom., 7 (1992) 515. [28] T. McAllister, J. Anal. At. Spectrom., 5 (1990) 171. [29] D.L. Styris and D.A. Redfield, Spectrochim. Acta Rev., 15 (1993) 71. [30] T. Kantor, B. Radziuk and B. Welz, Spectrochim. Acta, Part B, 49 (1994) 875. [31] X-Q. Shan, Z-N. Yuan and Z-M. Ni, Anal. Chem., 57 (1985) 857. [32] H.-C. Qiao, T.M. Mahmood and K.W. Jackson, Spectrochim. Acta, Part B, 48 (1993) 1495. [33] T. Mallat, Z. Bodnar, S. Szabo and J. Petro, Appl. Catal., 69 (1991) 85. [34] R.W. Dabeka, Anal. Chem., 64 (1992) 2419. [35] H. Docekalova, B. Docekai, J. Komarek and I. Novotny, J. Anal. At. Spectrum., 6 (1991) 661. [36] C. Yoshimura and T. Huzino, Nippon Kagaku Kaishi, (1985) 1392.

A.B. Volynsky / Spectrochimica Acta Part B 51 (1996) 1573-1589 [37] A.B. Volynsky and V. Krivan, J. Anal. At. Spectrom., 11 (1996) 159. [38] A.B. Volynsky, V. Krivan and S.V. Tikhomirov, Spectrochim. Acta, Part B, 51 (1996) 1253. [39] Z. Grobenski, W. Erler and U. Voellkopf, At. Spectrosc., 6 (1985) 91. [40] B. Welz, G. Schlemmer and J.R. Mudakavi, Anal. Chem., 60 (1988) 2567. [41] G.A. Somorjai, Science, 201 (1978) 489. [42] Gmelin's Handbook of Inorganic Chemistry, Palladium, Syst. No. 65, Vol. B2, Springer, Berlin, 1989, p. 1. [43]J.F. Hamilton and R.C. Baetzold, Science, 205 (1979) 1213. [44] G.A. Del Angel, B. Coq, G. Ferrat and F. Figueras, Surf. Sci., 156 (1985) 943. [45] G.C. Bond, Surf. Sci., 156 (1985) 966. [46] M. Che and C.O. Bennett, The Influence of Particle Size on the Catalytic Properties of Supported Metals, in D.D. Eley, H. Pines and P.B. Weisz (Eds.), Advances in Catalysis, Vol. 36, Academic Press, San Diego, 1989, Chapter 2, p. 55. [47] O. Sakurada, H. Takahashi and M. Taga, Bunseki Kagaku, 38 (1989) 407. [48] L.-Z. Jin and Z.-M. Ni, Can. J. Spectrosc., 26 (1981) 219. [49] E. Tserovsky, S. Arpadjan and I. Karadjova, Spectrochim. Acta, Part B, 47 (1992) 959. [50] P.S. Doidge, B.T. Sturman and T.M Rettberg, J. Anal. At. Spectrom., 4 (1989) 251. [51] R.E. Sturgeon, S.N. Willie, G.I. Sproule, P.T. Robinson and S.S. Berman, Spectrochim. Acta, Part B, 44 (1989) 667. [52] R.E. Sturgeon, S.N. Willie, G.I. Sproule and S.S. Berman, J. Anal. At. Spectrom., 2 (1987) 719. [53] I.M. AI-Daher and J.M. Saleh, J. Phys. Chem., 76 (1972) 2851. [54] J.M. Saleh, Trans. Faraday Soc., 67 (1971) 1830. [55] L. Zhang, Z.-M. Ni and X-Q. Shan, Spectrochim. Acta, Part B, 44 (1989) 339. [56] M. Burguera and J.L. Burguera, J. Anal. At. Spectrom., 8 (1993) 229. [57] H.O. Haug and L. Yiping, Spectrochim. Acta, Part B, 50 (1995) 1311. [58] M.M. Chaundry, A.M. Ure, B.G. Cooksey and D. Littlejohn, Anal. Proc., 28 (1991) 44. [59] I.L. Shuttler, M. Feuerstein and G. Schlemmer, J. Anal. At. Spectrom., 7 (1992) 1299. [60] Z.-M. Ni and D.-Q. Zhang, Spectrochim. Acta, Part B, 50 (1995) 1779. [61] Gmelin's Handbuch der Anorganischen Chemie, Palladium, Syst. No. 65, Lieferung 2, Verlag Chemie, Weinheim, 1942, p. 133. [62] G. Weibust, F.J. Langmyhr and Y. Thomassen, Anal. Chim. Acta, 128 (1981) 23. [63] T. Katsura, F. Kato and K. Matsumoto, Anal. Sci., 6 (1990) 909. [64] T. Katsura, F. Kato and K. Matsumoto, Anal. Chim. Acta, 252 (1991) 77. [65] H. Li, B. Gong, T. Ochiai and K. Matsumoto, Anal. Sci., 9 (1993) 707.

1587

[66] R.E. Sturgeon, S.N. Willie and S.S. Berman, Anal. Chem., 61 (1989) 1867. [67] E.H. Larsen, J. Anal. At. Spectrom., 6 (1991) 375. [68] L. Zhang, Z.-M. Ni and X.-Q. Shan, Can. J. Appl. Spectrosc., 36 (1991) 47. [69] G.-B. Jiang, Z.-M. Ni, L. Zhang, A. Li, H.-B. Hart and X.-Q. Shan, J. Anal. At. Spectrom., 7 (1992) 447. [70] L. Spialter, G.R. Buell and C.W. Harris, J. Org. Chem., 30 (1965) 375. [71] H. Ltith, G.W. Rubloff and W.D. Grobman, Surf. Sci., 63 (1977) 325. [72] J.R. Anderson, Metal Catalysed Skeletal Reactions of Hydrocarbons, in D.D. Eley, H. Pines and P.B. Weisz (Eds.), Advances in Catalysis, Vol. 23, Academic Press, New York, 1973, Chapter 1, p. 1. [73] G.A. Somorjai and D.W. Blakely, Nature, 258 (1975) 580. [74] M.E. O'Neill and K. Wade, Structural and Bonding Relationships Among Main Group Organometallic Compounds, in G. Wilkinson, F.G.A. Stone and E.W. Abel (Eds.), Comprehensive Organometallic Chemistry, Vot. 1, Pergamon Press, Oxford, 1982, Chapter 1, p. 1. [75] Y. An, S.N. Willie and R.E. Sturgeon, Fresenius' J. Anal. Chem., 344 (1992) 64. [76] D. Erber and K. Cammann, Analyst, 120 (1995) 2699. [77] F. Laborda, J. Vinuales, J.M. Mir and J.R. Castillo, J. Anal. At. Spectrom., 8 (1993) 737. [78] D.E. Nixon, G.V. Mussmann, S.J. Eckdahl and T.P. Moyer, Clin. Chem., 37 (1991) 1575. [79] G.R. Hennig, Electron Microscopy and Reactivity Changes near Lattice Defects in Graphite, in P.L. Walker, Jr. (Ed.), Chemistry and Physics of Carbon, Vol. 2, Marcel Dekker, New York, 1966, Chapter 1, p. 1. [80] D.W. McKee, The Catalysed Gasification Reactions of Carbon, in P.L Walker, Jr. and P.A. Thrower (Eds.), Chemistry and Physics of Carbon, Vol. 16, Marcel Dekker, New York, 1981, Chapter 1, p. 1. [81] T. Baker, Chem. Ind., (1982) 698. [82] T. Engel and G. Ertl, Elementary Steps in the Catalytic Oxidation of Carbon Monoxide on Platinum Metals, in D.D. Eley, H. Pines and P.B. Weisz (Eds.), Advances in Catalysis, Vol. 28, Academic Press, New York, 1979, Chapter 1, p. 1. [83] D.L. Styris, L.J. Prell, D.A. Redfield, J.A. Holcombe, D.A. Bass and V. Majidi, Anal. Chem., 63 (1991) 508. [84] V. Majidi and J.D. Robertson, Spectrochim. Acta, Part B, 46 (1991) 1723. [85] R. Peile, R. Grey and R. Starek, J. Anal. At. Spectrom., 4 (1989) 407. [86] I.A. Kuzovlev, Yu.N. Kuznetsov and O.A. Sverdlina, Zavod. Lab., 39 (1973) 428. [87] A.B. Volynsky, Spectrochim. Acta, Part B, 50 (1995) 1417. [88] G.V. Samsonov, Ukr. Khim. Zh., 31 (1965) 1006. [89] A.B. Volynskii, J. Anal. Chem. (USSR), 42 (1987) 1223. [90] E. Norval, H.G.C. Human and L.R.P. Butler, Anal. Chem., 51 (1979) 2045. [91] E. Norval, Anal. Chim. Acta, 181 (1986) 169.

1588

A.B. Volynsky / Spectrochimica Acta Part B 51 (1996) 1573-1589

[92] H.M. Ortner, W. Birzer, B. Welz, G. Schlemmer, J.A. Curtius, W. Wegscheider and V. Sychra, Fresenius' Z. Anal. Chem., 323 (1986) 681. [93] A.B. Volynsky and E.M. Sedykh, J. Anal. At. Spectrom., 4 (1989) 71. [94] R.E. Sturgeon, K.W.M. Siu, G.J. Gardner and S.S. Berman, Anal. Chem., 58 (1986) 42. [95] R.F. Voitovich, Okislenie Karbidov i Nitridov (Oxidation of Carbides and Nitrides), Naukova Dumka, Kiev, 1981. [96] Z. Cizek, P. Borek, J. Fiala and B. Bogdain, Mikrochim. Acta, III (1990) 163. [97] F.A. Josien, Bull. Soc. Chim. Fr., (1956) 480. [98] W. Freundlich, F.A. Josien and A. Erb, Bull. Soc. Chim. Fr., (1960) 281. [99] R.B. Levy and M. Boudart, Science, 181 (1973) 547. [100] R.T. Yang and C. Wong, J. Catal., 85 (1984) 154. [101] P. Hocquellet and N. Labeyrie, Analusis, 3 (1975) 505. [102] G. Miiller-Vogt and W. Wendl, Anal. Chem., 53 (1981) 651. [103] Y.-Q. Gao and Z.-M. Ni, Acta Chim. Sin., 40 (1982) 1021 (in Chinese). [104] Z-M. Ni and X-Q. Shan, Spectrochim. Acta, Part B, 42 (1987) 937. [105] Y.-S. Zheng and D.-L. Zhang, Chin. J. Appl. Chem., 4(4) (1987) 21 (in Chinese). [106] D.J. Lythgoe, Analyst, 106 (1981) 743. [107] V.N. Muz'gin, V.B. Atnashev, A.A. Pupyshev and Yu.B. Atnashev, J. Anal. Chem. (USSR), 41 (1986) 1246. [108] W. Wendl and G. M~iller-Vogt,Spectrochim. Acta, Part B, 39 (1984) 237. [109] E. Iwamoto, H. Shimazu, K. Yokota and T. Kumamaru, J. Anal. At. Spectrom., 7 (1992) 421. [110] J.M.P. Parajon and A. Sanz-Medel, J. Anal. At. Spectrom., 9 (1994) 111. [111] B.V. L'vov, Spectrochim. Acta, Part B, 33 (1978) 153. [112] A.C. Sahayam and S. Gangadharan, Can. J. Appl. Spectrosc., 39 (1994) 61. [113] A.B. Volynsky, E.M. Sedykh, B.Ya. Spivakov and I. Havezov, Anal. Chim. Acta, 174 (1985) 173. [114] B.V. L'vov and G.N. Ryabchuk, Spectrochim. Acta, Part B, 37 (1982) 673. [115] R.E. Sturgeon and S.S. Berman, Anal. Chem., 57 (1985) 1268. [116] G.F.R. Gilchrist, C.L. Chakrabarti, J.T.F. Ashley and D.M. Hughes, Anal. Chem., 64 (1992) 1144. [117] M. Chamsaz and J.D. Winefordner, J. Anal. At. Spectrom., 3 (1988) 119. [118] C. Donaghy, M. Harriott and D.T. Bums, Anal. Proc., 26 (1989) 260. [119] E. lwamoto, H. Shimazu, K. Yokota and T. Kumamaru, Anal. Chim. Acta, 274 (1993) 231. [120] H.M. Ortner, H. Krabichler and W. Wegscheider, Beschichtung und Impr~igniemngvon Graphitrohren, in B. Welz (Ed.), Fortschritte in der Atomspektrometrisehen Spurenanalytik, Vol. 1, Verlag Chemie, Weinheim, 1984, p. 73. [121] Gmelin's Handbuch der Anorganischen Chemie, Zinn, Syst. No. 46, Teil C2, Springer Verlag, Berlin, 1975, p. 217.

[122] G.T. Dubovka, Fiz. Tekh. Vysokich Davlenii (Physics and Techniques of High Pressures), (13) (1983) 47. [123] H. S. Wahab and C. L. Chakrabarti, Spectrochim. Acta, Part B, 36 (1981) 463. [124] H.J. Gitelmal and F.R. Alderman, J. Anal. At. Spectrom., 5 (1990) 687. [125] Y.-M. Liu, B.-L. Gong, Y.-L. Xu, Z-H. Li and T.-Z. Lin, Anal. Chim. Acta, 292 (1994) 325. [126] M. Luguera, Y. Madrid and C. Camara, J. Anal. At. Spectrom., 6 (1991) 669. [127] G.A. Wiltshire, D.T. BoUand and D. Littlejohn, J. Anal. At. Spectrom., 9 (1994) 1255. [128] J.P. Byme, D.C. Gregoire, D.M. Goltz and C.L. Chakrabarti, Spectrochim. Acta, Part B, 49 (1994) 433. [129] K.S. Krasnov (Ed.), Molekulyamye Konstanty Neorganicheskich Soedinenii (Molecular Constants of Inorganic Compounds), Khimiya, Leningrad, 1977. [130] W. Frech, J.-A. Persson and A. Cedergren, Prog. Anal. At. Spectrosc., 3 (1980) 279. [131] P.S. Doidge and T. McAllister, J. Anal. At. Spectrom., 8 (1993) 403. [132] M. Taddia, Anal. Chim. Acta, 182 (1986) 231. [133] B. Welz, G. Schlemmer and H.M. Ortner, Spectrochim. Acta, Part B, 41 (1986) 567. [134] T.M. Vickrey, G.V. Harrison, G.J. Ramelow and J.C. Carver, Anal. Lett. Part A, 13 (1980) 781. [135] T.M. Vickrey, H.E. Howell, G.V. Harrison and G.J. Ramelow, Anal. Chem., 52 (1980) 1743. [136] D.T. Bums, D. Dadgar and M. Harriott, Analyst, 109 (1984) 1099. [137] T.M. Vickrey, G.V. Harrison and G.J. Ramelow, At. Spectrosc., 1 (1980) 116. [138] O.A. Reutov and I.P. Beletskaya, Reaction Mechanisms of Organometallic Compounds, North-Holland Publishing Co., Amsterdam, 1968, p. 223. [139] M. Sager, Mikrochim. Acta, II1 (1986) 129. [140] M. Hino, S. Kobayashi and K. Arata, J. Am. Chem. Soc., 101 (1979) 6439. [141] K. Tanabe, M. Misono, Y. Ono and H. Hattori, New Solid Acids and Bases, Kodansha, Tokio, Elsevier, Amsterdam, 1989, p. 47. [142] D.T. Bums and D. Dadgar, Analyst, 107 (1982) 452. [143] D.T. Bums, D. Dadgar, M. Harriott, K. McBride and W.J. Swindall, Analyst, 109 (1984) 1613. [144] S. Arpadjan and V. Krivan, Anal. Chem., 58 (1986) 2611. [145] X.-P. Yan and Z-M. Ni, J. Anal. At. Spectrom., 6 (1991) 483. [146] Z.-M. Ni, H.-B. Hang, A. Li, B. He and F.-Z. Xu, J. Anal. At. Spectrom., 6 (1991) 385. [147] Z.-M. Ni and B. He, J. Anal. At. Spectrom., 10 (1995) 747. [148] A.B. Volynsky, E.M. Sedykh, B.Ya. Spivakov and Yu.A. Zolotov, Anal. Chim. Acta, 177 (1985) 129. [149] J. Komarek and S. Gomiscek, Vestn. Slov. Kem. Drus., 30 (1983) 443. [150] D.B. Fischbach, The Kinetics and Mechanism of Graphitization, in P.L. Walker, Jr. (Ed.), Chemistry and Physics of Carbon, Vol. 7, Marcel Dekker, New York, 1971, Chapter 1, p. 1.

A.B. Volynsky / Spectrochimica Acta Part B 51 (1996) 1573-1589

[151] A. Oya and S, Otani, Carbon, 17 (1979) 131. [152] H. Fritzsche, W. Wegscheider, G. Knapp and H.M. Ortner, Talanta, 26 (1979) 219. [153] B. Welz, A.J. Curtius, G. Schlemmer, H.M. Ortner and W. Birzer, Spectrochim. Acta, Part B, 41 (1986) 1175. [154] J.H. Runnels, R. Merryfield and H.B. Fisher, Anal. Chem., 47 (1975) 1258. [155] V.J. Zatka, Anal. Chem., 50 (1978) 538. [156] T.M. Vickrey and M.S. Buren, Anal. Lett., Part A, 13 (1980) 1465. [157] B. Welz, G. Schlemmer, H.M. Ortner and W. Wegscheider, Prog. Anal. Spectrosc., 12 (1989) 111. [158] A.B. Volynskii and E.M. Sedykh, J. Anal. Chem. (USSR), 42 (1987) 844. [159] F.A. Cotton and G. Wilkinson, Advanced Inorganic Chemistry, 5th edn., John Wiley, New York, 1988. [160] Landolt-B6mstein, O. Madelung (Ed.), Group IV, Macroscopic and Technical Properties of Matter, New Series, Vol. 5, Phase Equilibria, Crystallographic and Thermodynamic Data of Binary Alloys, Subvol. b, Springer-Verlag, Berlin, 1991. [161] B. Neidhart and C. Lippmann, Fresenius' Z. Anal. Chem., 306 (1981) 259. [162] A.B. Volynskii, O.R. Valedinskaya and N.K. Kutseva, J. Anal. Chem. (Russia), 47 (1992) 361. [163] A.B. Volynsky, S.V. Tikhomirov, V.G. Senin and A.N. Kashin, Anal. Chim. Acta, 284 (1993) 367. [164] R. Guevremont, R.E. Sturgeon and S.S. Berman, Anal. Chim. Acta, 115 (1980) 163. [165] J. Dedina, W. Frech, E. Lundberg and A. Cedergren, J. Anal. At. Spectrom., 4 (1989) 143.

1589

[166] J. Dedina, Prog. Anal. Spectrosc., 11 (1988) 251. [167] B.V. L'vov and A.V. Novichikhin, Spectrochim. Acta, Part B, 50 (1995) 1459. [168] W. Frech and A. Cedergren, Anal. Chim. Acta, 113 (1980) 227. [169] S. lmai, T. lbe, T. Tanaka and Y. Hayashi, Anal. Sci., 10 (1994) 901. [170] B.V. L'vov and A.S. Savin, J. Anal. Chem. (USSR), 38 (1983) 1475. [171] D.A. Katskov, A.M. Shtepan, I.L. Grinshtein and A.A. Pupyshev, Spectrochim. Acta, Part B, 47 (1992) 1023. [172] A.Kh. Gilmutdinov, Yu.A. Zakharov, V.P. Ivanov, A.V. Voloshin and K. Dittrich, J. Anal. At. Spectrom., 7 (1992) 675. [173] E. Iwamoto, K.E.A. Ohlsson, D.C. Baxter and W. Frech, J. Anal. At. Spectrom., 7 (1992) 1063. [174] D.L. Tsalev, A. D'Ulivo, L. Lampugnani, M. Marko and R. Zamboni, J. Anal. At. Spectrom., 10 (1995) 1003. [175] J. Haber, J.H. Block and B. Delmon, Pure Appl. Chem., 67 (1995) 1257. [176] A.Kh. Gilmutdinov and Yu.A. Zakharov, Izv. Akad. Nauk SSSR, Ser. Phiz., 53 (1989) 1821. [177] D. Brennan, The Atomization of Diatomic Molecules by Metals, in D.D. Eley, H. Pines and P.B. Weisz (Eds.), Advances in Catalysis, Vol. 15, Academic Press, New York, 1964, Chapter 1, p. 1. [178] K.E.A. Ohlsson, E. lwamoto, W. Frech and A. Cedergren, Spectrochim. Acta, Part B, 47 (1992) 1341. [179] M.M. Lamoureux, C.L. Chakrabarti, J.C. Hutton, A.Kh. Gilmutdinov, Y.A. Zakharov and D.C. Gregoire, Spectrochim. Acta, Part B, 50 (1995) 1847. [180] B.V. L'vov, Spectrochim. Acta, Part B, 51 (1996) 533.