摘要:

324 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 Future Issues will Include- Applications of Spark Source Mass Spec- trometry in the Analysis of Semiconduc- tor Materials: A Review-Jozef Verlin- den, Renaat Gijbles and Freddy Adams Determination of Iron by Atomic Absorp- tion Spectrometry after Synergistic Extraction of the Iron(III)-5,5’-methyl- enedisalicylohydroxamate-TBP Complex -F. Captain, D. Gazquez, M. Sanchez and L. F. Capitan-Vallvey Determination of Tin by Non-dispersive Atomic Fluorescence Spectrometry Coupled with the Hydride Generation Technique-Taketoshi Nakahara and Tamotsu Wasa Direct Microcomputer Controlled Deter- mination of Zinc in Human Serum by Flow Injection Atomic Absorption Spec- trometry-K. W. Simonsen, B. Nielsen, A. Jensen and J.R. Andersen A Kapid Screening Method for the Deter- mination of Platinum and Palladium in Geological Materials by Batch Ion- exchange Chromatography and Graphite Furnace Atomic Absorption Spec- trometry-Charles H. Branch and Dawn Hutchison Determination of Chromium, Vanadium and Barium in Silicate Rocks by Energy Dispersive X-ray Fluorescence Analysis Using a Cobalt Anode X-ray Tube. Part 1 : Optimisation of Excitation Condi- tions- Philip J. Potts, Peter C. Webb and John S. Watson A Study of the Formation of Atoms and Dry Aerosols Above a Graphite Rod Sample Introduction Used for Inductively Coupled Plasma Atomic Emission Spec- trometry-John R. Dean and Richard D. Snook Determination of Trace Amounts of Nickel and Cobalt in Standard Silicate Rocks by Graphite Furnace Atomic Absorption Spectrometry after Fusion with a Lithium Carbonate - Boric Acid Mixture-R.Kuroda, T. Nakanao, Y. Miura and K. Oguma Surface Properties of Graphite Tube Coatings: Limiting Factor in the Determi- nation of Refractory Elements by Elec- trothermal Atomisation Atomic Absorp- tion Spectrometry-M. Hoenig, F. Dehairs and A.-M. De Kersabiec Analyses of Solid Samples by Graphite Furnace Atomic Absorption Using Zee- man Background Correction-Glen R. Carnrick, B. K. Lumas and W. B. Barnett Determination of Phosphorus by Graph- ite Furnace Atomic Absorption Spec- trometry. Part 1: Determination in the Absence of a Matrix Modifier-A. J. Curtius, G. Schlemmer and B. Welz Determination of Silicon in Gallium Arsenide by Electrothermal Atomisation Atomic Absorption Spectrometry Using the L’vov Platform-M. Taddia Atomic Spectrometry Update The Update in the December issue is- Minerals and Refractories-David Hickman, Joan M. Rooke and Michael Thompson

ISSN:0267-9477    

DOI:10.1039/JA9860100324

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 325 Spatial Emission Characteristics and Excitation Mechanisms in the Inductively Coupled Plasma A Review John Davies Trace Analysis Laboratory, Department of Chemistry, Imperial College of Science and Technology, London SW7 ZAY, UK Richard D. Snook Chelsea Instruments Ltd., 5 Epirus Road, London SW6 7UR, UK Summary of Contents 1 Introduction 2 Atomic and Ionic Emission Profiles 3 Nomenclature Systems 4 Analytical Implications 5 Temperature and Excitation Mechanisms 6 Conclusions 7 References Keywords: Inductive1 y coupled plasma; spatial emission characteristics; excitation mechanisms 1. Introduction Since the introduction of the inductively coupled plasma (ICP) as a useful source for atomic emission spectrometry by Greenfield et al.and Wendt and Fassel2 the ICP has emerged as an excellent source for the determination of trace elements in solution. The reason why the ICP has become such a valuable tool for spectrochemical analysis is perhaps best answered by a description of the ICP and a look at the desirable properties it possesses. The ICP is simply a volume of partially ionised gas (usually argon) inductively coupled to high radiofrequency radiation (5-50MHz). The plasma is contained in a torch usually consisting of three concentric quartz tubes through which coolant gas is supplied to prevent the plasma melting the torch, plasma gas (which in some, and probably most, torches is superfluous) is supplied to sustain the plasma and an injector gas flow supplied to facilitate sample introduction.High-frequency power is continuously applied via an induc- tion coil (fabricated from copper tubing to facilitate water cooling) in which the plasma torch is mounted and enables the plasma to become self-sustaining through a series of excitation mechanisms once the plasma has been initiated. The analytical utility of any excitation source is dependent on its detection limits, degree of freedom from inter-element interferences and its signal to background and signal to noise ratios. For the ICP, detection limits for many elements are often in the p.p.b. range, and calibration graphs obtained using the ICP are rectilinear over 5-6 orders of magnitude with respect to analyte concentration, permitting element determinations to be undertaken at a single dilution of the sample.3 As the plasma is sustained without the use of supporting electrodes the background is relatively clean, consisting of only excited Ar lines and weak molecular band emission for species such as OH, CN, NO and Nz.As well as these properties the ICP has a high sensitivity to ion lines, a good signal to background ratio and reduced inter-element interferences compared with other excitation sources. However, inter-element interferences, though they are small, do exist in the ICP, and the mechanisms involved in excitation of such a source are not well understood. Much conflicting evidence has arisen in the literature concerning the magnitude and reason for interferences as well as the excitation mechanisms involved. To elucidate the spatial dependence of emission and interference effects it is impor- tant to grasp an understanding of the possible excitation mechanisms that might occur in the ICP. The plasma is not homogeneous with respect to temperature because it is formed from a flowing stream of the supporting gas that sweeps plasma matter consisting of ions, electrons and neutral atoms away from the energy input region (within and near the induction coil).As a consequence of the inhomogeneity of temperature throughout the plasma, both radially and axially, the emission intensity of many atom and ion lines show marked spatial dependencies. These dependencies can be linked in some instances to the temperature directly but in other instances we have to study the role of excitation mechanisms other than those which are thermal in nature.Clearly, before we can make predictions about such excitation mechanisms we must have a reasonable model for tempera- ture variations in the plasma. The plasma is not in local thermal equilibrium (LTE); LTE exists in a system when all temperature dependent processes ( e . g . , the Boltzmann popu- lation distribution and the Maxwell velocity distribution function), can be described by a single temperature for the system. A simple way of interpreting non-LTE character is to say that the excitation temperature T,,,, , ionisation tempera- ture Tion, and gas temperature Tgas are not the same. A more rigorous definition for plasma matter is that for LTE to exist collisional de-excitation processes must predominate and this is likely to be true only for high electron densities.2. Atomic and Ionic Emission Profiles It was the study of the atomic and ionic spatial emission profiles that led to the renewed interest in excitation mechan- isms in the ICP in order to explain the emission profiles observed, and therefore it is perhaps appropriate to consider the spatial emission profiles in the ICP first before we look at proposed excitation mechanisms in detail. Edmonds and Horlick4 were two of the first to study the spatial emission profiles in an ICP and the effect of operating parameters upon them. They found that lines with high energy emitting states peaked higher in the plasma than correspond- ing lower energy lines. In later papers Horlick and co- workerss-9 measured the vertical spatial profiles of analyte emission for various atom and ion lines and divided the plasma5 into two regions by dividing the lines into two basic categories: (i) soft lines (lower energy atom lines) whose spatial emission behaviour depended on applied high- frequency power, aerosol flow, analyte excitation and ionisa-326 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 tion characteristics and (ii) hard lines (ion lines and the more energetic atom lines) whose spatial emission behaviour was relatively insensitive to the parameters affecting the soft lines. They observed that, under fixed ICP conditions, all hard lines had their peak emission at essentially the same position in the plasma discharge and was always higher in the discharge than for soft lines.Kawaguchi ef al. 10 measured the axial profiles of Tex,, , Tgas and analyte emission lines and their measurements led them to divide the ICP into two excitation regions governed by different excitation mechanisms along the axis of the plasma. In the first region, 0-8mm above the coil, low-energy lines were predominantly excited and their emission profiles seemed to be controlled primarily by dissociation rates. In the second region, 10-20mm above the load coil, high-energy lines predominated, while lines with intermediate excitation energies possessed peaks in both regions. They concluded that two different excitation mechanisms were involved in each excitation region. Volatilisation effects were only observed in the first excitational region.This was also borne out by Roederer et aZ.11 who found maximum central ion emission occurring higher in the plasma than atomic emission and the absence of volatility effects on ionic species. Furuta and Horlickl2 also observed that atomic emission was species dependent while ionic emission behaviour was species independent and peaked higher in the plasma than atom emission. In addition, they observed that intense ion emission relative to atom emission for several elements was spatially distributed along the boundary regions of the injector channel of the plasma. This led them to suggest that atom excitation is a result of electron collisions and that ion species excitation is a result of collisions with energetic argon species and is therefore spatially somewhat limited.This is in agreement with the observations of Edmonds and Horlick4 already discussed. Eckert and Danielsson13 derived expressions for the rela- tive intensity distributions of analyte atom and ion lines for the limiting cases of full and vanishing ionisation. They observed that for a given temperature and profile the shape of the spatial emission profile was partly determined by the excita- tion energy and partly by the ionisation energy of the line. The profiles showed a gradual transition from double off-axis maxima to a single-peak profile with increasing observation height . Caughlin and Blades14 found a similar transition when they measured spatially resolved plots of electron density, n,. They compared ion - atom intensity ratios (Zi/la)exp, with n, plots and observed that the maximum (Zi/Za)exp, in the central channel did not correspond to the region of maximum n,.The relationship between n, and (Zi/Za)exp. was also found to change with position in the plasma indicating that n, is not the only factor in determining Zi/Za. Caughlin and Blades also observed that (Zi/la)exp, values were less than (Zi/la)L~E at all spatial positions and powers (the reverse of Boumans and DeBoer’s findings15) suggesting that the ground state was over-populated with respect to LTE indicating that the aersosol channel could be characterised as an ionising plasma. The contradiction of the results of Caughlin and Bladed4 and Boumans and DeBoer15 may be a consequence of the different operating parameters used. Certainly in our laboratory we have found that ion to atom ratios are critically dependent on injector flow-rate.In summary then, low energy atomic emission (soft) lines are observed to peak lower in the plasma than the more energetic atomic and ionic emission (hard) lines. The spatial profiles of soft lines are dependent on the operating para- meters ( e . g . , increasing the r.f. power shifts the spatial profile towards the induction coil) while hard lines are relatively insensitive to changes in the operating parameters. This spatial dependence of atomic emission and independence of ionic emission indicates that the excitation mechanisms in the ICP are not just the result of thermal excitation and electron collisional mechanisms. Moreover the difference in peak heights and dependency of soft line emission on the operating parameters makes comparison of different plasma systems more difficult.In the next section we review how different workers have suggested possible reference points for such comparisons. 3. Nomenclature Systems Koirtyohann and co-workersl”l* suggested a nomenclature system for the plasma discharge using an internal reference point as an aid for comparing plasma conditions. All previous work that had studied the spatial effects of matrix inter- ferences had used the top of the load coil as a reference point to specify relative positions in the plasma and this makes comparison of different pieces of work much more difficult because of differences in operating parameters and torch design. Koirtyohann et al. divided the central axial region into zones, namely, pre-heating zone (PHZ) , initial radiation zone (IRZ) and normal analytical zone (NAZ).The IRZ was observed by aspiration of (i) a solution containing Na and observing the intense atom emission or (ii) a solution containing 1000 pg ml-1 of Y and observing the intense red atom emission. They found that enhancements observed for both atomic and ionic lines in the region of high atomic emission with the addition of matrix elements (lower in the plasma, IRZ) became much less severe or disappeared in the region of high ionic emission (higher in the plasma, NAZ). Moreover, the zone where the interferences were observed could be shifted higher or lower in the plasma by varying the central (injector) flow or radiofrequency power.The proxim- ity of the IRZ (where inter-element effects are severe) and the NAZ (where such effects are much less severe) led them to suggest to observe the IRZ and ensure that the region viewed is at least 5 mm above it. Motooka et al. l9 suggested the use of the apex of the central vortex of the plasma as a reference point, which is clearly distinguishable in the centre of the plasma. The advantage over the reference point suggested by Koirtyohann is that no aspiration of a solution is necessary to observe it, Both methods however depend on visual observation as opposed to experimental determination. Anderson et ~1.20 however, have suggested the use of the intersection of normalised atom and ion emission intensity profiles as a possible internal reference point, which is experimentally determined and reproducible.They measured several different pairs, in all instances the peak ion emission being greater than the peak atom emission, and proposed the use of the Cu I - Ba I1 intercept as a good reference point for inter-laboratory comparison. The validity of all these methods decreases the further away from the reference point one goes. It would seem more appropriate to correlate the spatial distribution of emission by measuring, for example, the spatial electron density. Raaij- makers and co-workers21>22 have shown that the state of an ICP can be adequately described by a measured electron density distribution, where measurement is independent of LTE. Thus it would seem that electron density measurements are a preferential method for internal reference points.Although the electron density distribution is still not the unique factor in determining spatial distributions it is useful for comparing local plasma conditions in different plasmas. In our studies23 we have shown that in two markedly different plasmas (tangential-flow and laminar-flow torches) the peak ion and atom emission intensities occur at different viewing heights in the two torches but at the same electron densities and background intensities. The latter is not surprising as the background is caused by the recombination continuum, the intensity of which must be proportional to electron density via Ar+ + e- s Ar + hvJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 327 4. Analytical Implications The main consequence of the different spatial dependencies of ionic and atomic emission with respect to operating para- meters is that compromise conditions have to be employed for simultaneous multi-element determinations; that is, consider- ation has to be taken of the position of all the peak heights of the elements to be determined and a compromise selected.This problem has been shown to be overcome by “end-on,” or, axial viewing of the plasma. Several workers24-29 have studied the use of axial viewing and have shown that there is no degradation of analytical utility with such a viewing configuration and its major advantage is indeed in its use for simultaneous determinations. Davies et ~ 1 . 2 8 have shown that by using the correct optics the spatial dependencies are not seen with axial viewing due to optical integration and summation of emission from all parts of the plasma.Such a viewing configuration, although of little use for diagnostic measurements where spatial resolution is necessary, is an extremely useful way of negating the variations in spatial dependence of soft and hard line emission and, in particular, is useful for multi-element analysis. 5. Temperatures and Excitation Mechanisms As already stated one of the features of the ICP is that it is not in LTE. (LTE is not necessarily desirable as it implies that all radiative processes are balanced by absorptive properties in the local environment of the plasma and hence little or no emission would be observed.) This became apparent soon after the introduction of the ICP as an emission source when workers began to characterise the discharge.Different values of temperatures were found depending on the spectroscopic technique used to determine the temperature. Moreoever, the ratio of the intensity of ion lines to atom lines of the same element were found to be much greater when compared with theoretical ratios determined assuming LTE. It was these observations coupled with the fact that the more energetic emission lines had their maximum emission intensities higher in the plasma relative to the less energetic lines that led workers to seek an explanation for these observations. Merrnet30J1 was the first to suggest that ionisation tempera- ture, Tion,, excitation temperature, T,,,, , and electron tem- perature, T,, were significantly different, and that LTE did not exist in the ICP.Since that publication30 many other workers have confirmed his suggestion5~1”~15~16~3~37 and we can regard the statement that “the argon ICP under normal operating conditions is not in LTE” as a fact, although it may approach it under certain conditions (see later). Walters et al.32 measured Tion,, T,,,, and Te in a 144-MHz ICP and found that in their plasma Tion, was lower than Text,, while T, was an order of magnitude larger than Text.. This latter difference in temperature they attributed to a longer mean free path of electrons resulting in higher kinetic energy: therefore the population of higher energy levels arises from energy transfer from the high kinetic energy of the electrons resulting in higher temperature.A significant contribution to our understanding of the ICP came when Mermet30 postulated Penning ionisation as an explanation for the high detection power attained with ion lines. (Penning ionisation is ionisation and excitation of analyte atoms through interaction with an excited metastable atom, in this instance argon, Arm.) The process may be written as A r m + X - + A r + X + + e - where the ions produced, X+, may be either in a ground state, or an excited state, and where The Penning ionisation process leads to departures of the ionic and atomic populations from the LTE values. The evidence for this is that experimental ion to atom intensity ratios are greater than the intensity ratios predicted from LTE .15 The model of Boumans and DeBoerl5 also involved the role of Arm proposing that Arm acts as an ionisant (easily ionisable element, EIE) as well as an ioniser (Penning ionisation), and therefore increases the electron number density in the plasma. (Arm levels, 3P0 and 3Pz, have a relatively low ionisation potential of about 4eV.) This is of course self evident when one considers ionisation as an equilibrium process.They also pointed out that departure from LTE was favourable resulting in high sensitivity of ionic lines and a relatively low ionisation interference, both interrelated with the overpopulation of Arm. Such an over- population is a result of the mixing of cold carrier gas and hot plasma gas and is responsible for the suprathermal population of ionic levels (Penning ionisation) and the high suprathermal electron number density.Thus the ion lines are more strongly excited than expected by sample excitation and ionisation by interaction with Arm. (Overpopulation of Arm occurs because of their extended lifetimes compared with the lifetime of excited states38; the Arm do not spontaneously return to the ground state through a radiative transition but remain in the metastable state and the excitation energy is then transferred by the Penning ionisation process.) Edmonds and Horlick4 pointed out that a Penning ionisa- tion mechanism would result in (i) ion line emission not occurring until Arm diffuse into the central analyte channel, relatively high in the plasma discharge, (ii) spatial emission patterns of ionic species generated by Penning ionisation would be species independent and (iii) a significant decrease in ion line emission as nebuliser flow-rate is increased.Thus they were inclined to suggest that non-thermal mechanisms such as Penning ionisation would dominate analyte emission charac- teristics higher in the channel while lower in the plasma emission characteristics of the analyte channel would be more thermal in nature. This is a reasonable, but simple, model developed from simple experimental observations that can be confirmed by performing spatially resolved emission measure- ments of the ICP. The required practical support for the existance of metastable levels in the upper region of the plasma was given by Uchida et a1.39 who found that Arm clearly exist even at a distance of 8 mm from the plasma centre and at a height of 7.5 mm, and the density of Arm at the outer positions was higher than at the centre of the plasma at a height of 3.5 mm.In complete contrast to these theories Alder et ~ 1 . ~ 5 suggested that analyte excitation and ionisation is primarily dominated by electrons with radiative de-excitation influenc- ing the analyte population and giving rise to non-thermal effects. They measured T,,,, using various Fe I lines and Tion. for several species. They found that for the various Fe I lines the values of T,,,, obtained was dependent on the Fe lines chosen: the values of T,,,, increased with the upper energy of the transition chosen indicating an increase in the relative overpopulation of an energy level with increase in its excitation energy. They attributed the discrepancies to popu- lation through radiative decay processes and proposed the major energy exchange mechanism to be electron collisional in nature.They felt that the high T, observed in the ICP decreases the significance of Penning ionisation and the non-thermal level populations observed results from the electron - atom collision frequency being too low. Aeschbach40 suggested that the two main causes of non- LTE behaviour are due to (i) the kinetic energy of the electrons owing to the radiofrequency power input of the field being higher than the kinetic energy of the gas particles and (ii) ambipolar diffusion of electrons as a consequence of extremely high gradients in electron density and temperature328 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 distribution. Ambipolar diffusion arises because electrons move faster than ions, resulting in a charge separation which produces an electric field that exerts its force in such a way as to increase the drift velocity of the ion species and retard the drift velocity of the electron. Thus it seemed apparent that non-thermal as well as thermal excitation mechanisms exist in the ICP and as a consequence two theories arose for analyte excitation processes: ( i ) direct electron collisional excitation (thermal) and (ii) excitation by Arm (non-thermal). Boumans41 considered a recombining plasma model put forward by Fujimoto42-45 and suggested a mechanism (given below) that could explain the results of Alder et al.35 and act as a link between atom and ion line emission where excited Ar atoms, Ar*, not just Arm, are involved in Penning ionisation reactions and where ion - electron recombination plays a vital role in creating excited level populations.The mechanism can be represented by four equations as follows. Ar* + X + Ar + X+(*) + e- (i) Penning ionisation - excitation: (ii) Spontaneous emission (ionic level): X+(*) + X+ + hv (ion line) (iii) Recombination X+(*) + e- + X* (iv) Spontaneous emission (atomic level) X* + X + hv (atom line) Assuming Penning ionisation reactions create an over- population of ionic levels and ion - electron recombination subsequently populates atomic levels, then higher atomic levels may become more densely populated than the lower levels.While Mermet proposed a non-LTE mechanism for ionising atoms and populating excited states based on Arm and Boumans and DeBoer offered a model whereby Arm have a dual role of ioniser (Penning ionisation) and ionisant (EIE), neither model (nor Boumans recombining model) was able to explain the presence of excitation species above the ionisation potential of Ar (11.7 eV), i.e., Cd 11. In a later paper, however, Batal and Mermet46 discussed the roles of electrons and argon ions, molecules and metastables as precursors to excitation. They showed that although these species play a direct role the predominant process is charge transfer between ions or argon and the analyte, viz. A r + + X + X + + A r AT2+ + X -+ X+* + 2Ar They conclude that the charge transfer mechanism plays an important role in the excitation mechanisms in the energy range below 16 eV and suggest that the sensitive analytical lines will be located below this limit.In order to verify this hypothesis they carried out experiments for elements that have energy levels in the range 10-19 eV. For these lines, e.g., V I1 and Mg 11, Boltzmann plots were constructed. For lines up to 13 eV a temperature of 4500 K was obtained, whereas above this limit a higher temperature ( T < 10 000 K) was obtained showing that there are probably two different mechanisms. They postulate that there is an upper limit of 30eV for excitation and suggest that the charge transfer mechanism provides a possible explanation of the high sensitivity achieved in the ICP. Blades and HieftJe47 suggested that the populations of radiating states of atomic Ar (3P1, ‘PI) and metastable states of atomic Ar (3P2, 3P0) may be collisionally equilibrated.They proposed that radiating levels, e . g . , metastable levels, serve to store energy through radiation trapping. Noble gas transitions that couple to the ground state absorb strongly over a wide range of gas pressures48349 and a high probability therefore exists of radiation emitted by one atom being reabsorbed by surrounding gas atoms, which in turn leads to transfer of excitation from one to another. At atmospheric pressure this exchange of energy (emission from one atom which is then reabsorbed by another atom) can occur many times before such radiation escapes from the boundaries of the enclosed gas.This trapped radiation, which leads to longer apparent lifetimes for the radiating levels, effectively acts as a storage of excitational energy in the enclosing gas. In the model, radiating levels are created in much the same way as metastable levels but with rapid collisional equilibria between the two states occurring; because of radiation trapping little loss of excitational energy arises. Such trapping imprisons the energy within the plasma for sufficient time (1 ms) for processes such as Penning ionisation to take place. The model therefore incorporates much of that of the previous models of Mermet and Boumans and DeBoer, involving the presence of Arm but with the provision that radiating levels are included, i. e . , Penning ionisation can arise not only through collisions with excited Ar species but also through interception of photons imprisoned in the plasma volume.This radiation trapping mechanism coupled with the Boumans and DeBoer model of Arm acting as an ioniser and ionisant led Blades50 to postulate a mechanism of “assisted ionisation” based on a rapid exchange of excitational energy between higher temperature Ar gas in the annular region, where the photon flux is very high, and lower temperature Ar gas in the analyte channel. Thus the high concentration of ground-state atoms in the analyte channel at a relatively low temperature (3000-6000 K) can absorb photons from the annular region within and just above the load coil, which is at a higher temperature (10 000 K). Blades showed that the assisted ionisation of “resonance” levels of atomic Ar in the channel can be explained by interaction with trapped radiation originating from the annular region of the plasma to a degree characteristic of the temperatue annulus of 10 000 K.A significant difference between this theory and the theory of Boumans is that the population of Arm is now dictated by the trapped resonance radiation and not by the diffusion rate constant of Arm from the annular region to the injector channel. Furthermore, it is not necessary to invoke a mechanism that involves Penning ionisation. Blades further concluded that the region of the axial channel that should be affected most is the boundary region between the channel and the plasma annulus as the photon flux at the centre is diminished due to absorption as it passes through the outer regions of the injector channel.Further support for this can be found in the work of Furuta and Horlick12 who demonstrated the importance of the plasma boundary regions, these being the regions where non-LTE behaviour is most marked. Although an elegant theory, the radiation trapping model does not stand up to examination because for argon these processes should result in a metastable population51 of ca. 1015 m-3 whereas measured concentrations suggest that the Arm concentration is 1018 m-3. This discrepancy becomes worse when we consider the lifetime of metastable states. For radiation trapping to be responsible for metastable excited emission at heights of 3 cm above the energy input region of the ICP (at a gas flow-rate of 30 m s-1) the lifetime of the metastable states would need to be a few milliseconds. As de Galan points outs1 the lifetime of metastables in the energy input region is more likely to be ca.10-8 s, which means that metastables produced by radiation trapping are unlikely to persist for more than a few microseconds, i.e., to a height of ca. 20 pm above the energy input region. Radiation trapping, therefore, cannot be considered as a mechanism for the apparent concentration of Arm above the load coil of the ICP. Indeed in later studies Mills and Hieftje52 calculated the extent and duration of radiation trapping and found that radiation trapping was only significantly responsible for sustaining the plasma within the load coil region.329 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 TO summarise the mechanisms discussed so far: Mermet, overpopulation of 3P0 and 3P2 Arm; Boumans and DeBoer, 3P0, 3P2 role of ioniser and ionisant; Boumans, recombining plasma model; and Blades and Hieftje, 3P0, 3P2 collisionally equilibrated with 3P1, ‘PI, radiation trapping. Lovett ,53 presenting a quantitative model of excitation in the ICP, considered the rates for all processes that involve excitation, de-excitation, ionisation or recombination of the analyte or ion by postulating a set of conditions likely to be found in a particular volume of plasma and attempted to calculate the response of the analyte to the conditions. The model revealed that even in a collisionally dominated plasma there are significant radiative processes that affect the population of the different levels, i.e.radiative decay, radiative recombination and radiative absorption. The model indicated that the significance of radiative decay was relatively small and was rate dependent giving rise to element - element variations. The extent to which radiative recombination is significant is depedent upon the ionisation potential of the element and gives rise to an increase in the ground-state atom population whereas radiative absorption affects the popula- tion of the levels by increasing the ion ground-state popula- tions and is again, dependent on the ionisation potential of the element. Thus Lovett’s calculations suggested that (i) for atom lines, overpopulation is more significant for lower energy lines than for higher energy lines and (ii) for ion lines, underpopula- tion is observed for excited levels but for the ground-state ion at LTE population would arise with coupling between the highly excited atomic lines and the ground-state ionic level.His model also predicted that Penning ionisation is a fairly insignificant process in the plasma. Raaijmakers et a1.21 put forward a theoretical description to explain quantitatively the density of the analyte and argon species in the ICP. The description was a radical deviation from previous models of excitation in this source, as outlined above. They postulated that although a deviation from LTE is apparent, the plasma can be described adequately by the measurement of just one parameter, e.g., the electron density, n,, or alternatively T,.They assumed that the density of argon ions in the plasma is equal to the density of electrons, which is self-evident if the plasma is taken to be macroscopically neutral. Not surprisingly they further state that T, can be found by: (a) assuming that the plasma is close enough to LTE and ( b ) measuring the electron density, n,. The reiteration of the preference to measure n, to get at Te is useful, however, as n, can be measured by methods such as the Hp Stark broadening meth0d,5~955 which is independent of LTE and is presumably why these workers have used it. An advance to our understanding of the plasma is hidden in their summary and in their reiteration of plasma physics. Their description centres around the concept of partial LTE (pLTE) where two parameters are needed to describe the plasma.The electron density is measured and then from this T, is estimated. In their second paper Schram et ~ 1 . 2 ~ postulate a “quasi-resonant asymmetrical charge transfer” mechanism for argon and analyte excitation. This mechanism leads to immediate and simple excitation to excited levels, viz. Anl + Ar+ + An,+ + Arl+ + e- (+ AE) where Anl = analyte in the ground state and An,+ = analyte in the excited ion state. Only heavy particles are assumed to take part in this mechanism. Presumably this is because of the relatively long residence time of slow, heavy particles in the vicinity of neighbouring particles when compared with the residence time of high velocity low mass particles, e.g., an electron. (For a given energy, the difference in velocities between an electron and an argon ion is approximately two orders of magnitude.) Raaijmakers et al.also show that the conclusion by other workers30,56,57 on the non-LTE character of the population of excited ions versus excited neutral atoms may also be premature. In addition, they dismiss radiation trapping of resonant radiation to the ground state and therefore consider that there is a significant overpopulation of this state. The result of the deliberations of Raaijmkaers et a f . is that the plasma can be divided into two regions, the active zone (in and around the load coil) where the plasma is in an ionising state and the recombination zone (above and around the active zone) where the plasma is in a recombining state.This seems to be in direct contrast to the experimental findings of Horlick and other workers who show clearly that peak ion emission occurs well above the load coil region, a phenomenon not expected in a recombining region. They conclude that in the active zone the ground-state argon atoms are overpopulated by a factor less than ten where the system is close to LTE. In the recombination zone analyte excited states are in LTE with the ground level of the analyte ion, except for the excited ion states, which are resonant with the argon ion. The presentation of the excitation models discussed so far in the literature have been contradictory rather than complimen- tary and their failure to explain the characteristics of the ICP is, in our opinion, a consequence of their restricted considera- tion of just one or two excitation mechanisms.A more recent paper that considers various possible excitation mechanisms that occur in the ICP is that of de Galan.51 His main objection to the models previously proposedl5,30,40~47~50~53 is that they ignore the fact that in the analytical observation zone, AOZ, of the ICP the plasma is decaying from a spatially inhom- ogeneous zone of the primary input region towards a state of equilibrium. Nevertheless, the excitation models previously proposed are not without merit. Although de Galan is not convinced of the significance of Arm he postulates that the analyte atoms are excited through collisions with electrons, argon ions and, possibly, metastable argon atoms. The rate model of L ~ v e t t , ~ ~ although not applicable to the AOZ is, he suggests, appro- piate to the energy input region where the conditions of Lovett’s model are more likely to exist.Thus de Galan considers the ICP as an annular plasma within the load coil with a uniform T, and a corresponding LTE electron concentration (Lovett53). This persists for “several” mm above the coil region where it is in a state of pLTE. Analyte excitation occurs by an exchange of energy from the hot electrons and argon ions in the surrounding plasma to the cool argon and analyte atoms centrally injected (Aeschbach40), which finally becomes equilibrated beyond the AOZ. Finally, de Galan also suggests that perhaps molecules and molecular radicals serve to raise the excited- state populations by lowering the average kinetic temperature of heavy particles that control the de-excitation rates of excited atoms and ions.Hieftje et al.58 have sought to correlate the spatial behav- iour of analyte emission intensities with the concentration product of reacting species assuming a kinetically controlled mechanism. This steady-state approach led them to suggest that for atom excitation, ion - electron recombination is important while ion generation and excitation occurs by Penning ionisation in the outer plasma regions and excitation by electron impact in the analyte channel. The authors describe some 20 possible plasma reactions that might occur in the ICP and they try to derive a single equation incorporating the kinetic rate parameters of all of these equations.By their own admission this is a formidable task and the papers findings should only be considered as tentative. 6. Conclusions It is abundantly clear that we are by no means at the point of understanding completely the processes of analyte excitation and emission that occur in the ICP. Moreoever , it is also clear that no one excitation model is able to explain all the characteristics of the ICP. Rather it would appear that the330 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 spatial characteristics of the ICP result from several, if not many, different reactions and mechanisms that occur to varying degrees in different regions of the plasma. Undoubt- edly Penning ionisation does occur in the ICP as well as radiation trapping but the significance of these processes in their contribution to high ion - atom line intensity ratios is doubtful.Certainly in the case of radiation trapping it has been shown52 that such a process is not significant beyond the load coil region. What must not be forgotten, however, is that though a process may not be significant beyond the load coil or if it does not occur in the AOZ, it does not follow that it has no consequence on the processes that do occur in the AOZ, i.e., a process which only occurs within the energy input region may be a significant precursor to a relevant excitation mechanism in the AOZ. Generally we take the view of de Galan that the plasma is spatially inhomogenous decaying to equilibrium above the load coil. If we accept this premise then it is probably not appropriate to fit one particular excitation model to the observed spatial ion and atom emission profiles in the AOZ.The development of two-, or even three-region plasma models with boundaries between these regions are not really of any value as the plasma decays gradually from the energy input region and the gradient of decay is critically affected by the design of the plasma torch, physical size of the plasma and the operating parameters. It is for this reason that comparisons between emission profiles of plasmas in different laboratories operating under different conditions cannot be entirely reliable. In our opinion it is more appropriate to compare local electron densities and temperatures in different plasmas because whatever mechanisms occur these electrons play a major role.Characterisation of the ICP by measurements of the electron density that are independent of LTE is therefore useful and would enable workers to compare conditions in different plasmas by comparison of analyte behaviour in regions of similar electron density and T,. Finally, we believe that although the development of models to fit experimental observations is acceptable it would be preferable to develop the theory first and then design experiments to provide substantial and conclusive evidence for or against the developed theory, e.g., Batal and Mermet.46 In such a complex and inhomogeneous system such as a free-flowing atmospheric pressure ICP this would be a formidable but worthy task. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.13. 14. 7. References Greenfield, S . , Jones, I. LI., and Berry, C. T., Analyst, 1964, 89, 713. Wendt, R. H., and Fassel, V. A., Anal. Chem., 1965,37, 920. Snook, R. D., Lab. Equip. Dig., 1979, Nov., 81. Edmonds, T. E., and Horlick, G., Appl. Spectrosc., 1977, 31, 536. Blades, M. W., and Horlick, G., Spectrochim. Acta, Part B , 1981, 36, 861. Horlick, G., and Blades, M. W., Appl. Spectrosc., 1980, 34, 229. Blades, M. W., and Horlick, G., Appl. Spectrosc., 1980, 34, 696. Blades, M. W., and Horlick, G., Spectrochim. Acta, Part B , 1981, 36, 881. Horlick, G . , and Furuta, N., Spectrochim. Acta, Part B , 1982, 37, 999. Kawaguchi, H., Ito, T., and Mizuike, A., Spectrochim. Acta, Part B , 1981,36,615 Roederer, J. E., Bastiaans, G. J., Fernandez, M. A., and Freeden, K.J., Appl. Spectrosc., 1982,36,383. Furuta, N . , and Horlick, G., Spectrochim. Acta, Part B , 1982, 37, 53. Eckert, H. U., and Danielsson, A., Spectrochim. Acta, Part B , 1983, 38, 15. Caughlin, B. L., and Blades, M. W., Spectrochim. Acta, Part B, 1984, 39, 1583. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. Boumans, P. W. J. M., and DeBoer, F. J., Spectrochim. Acta, Part B , 1977,32,365. Rybarczyk, J. P., Jester, C. P., Yates, D. A., and Koirtyohann, S. R., Anal. Chem., 1982, 54,2162. Koirtyohann, S. R., Jones, J. S . , Jester, C. P., and Yates, D. A., Spectrochim. Acta, Part B , 1981, 36, 49. Koirtyohann, S. R., Jones, J. S . , and Yates, D.A . , Anal. Chem., 1980,52, 1965. Motooka, J. M., Mosier, E. L., Sutley, S. J., and Wets, J. G., Appl. Spectrosc., 1979, 33, 456. Anderson, T. A., Burns, D. W., and Parsons, M. L., Spectrochim. Acta, Part B, 1984, 39, 559. Raaijmakers, I. J. M. M., Boumans, P. W. J. M., vandersijde, B., and Schrom, D. C., Spectrochim. Acta, Part B , 1983, 38, 697. Schram, D. C., Raaijrnakers, I. J. M. M., van der Sijde, B., Schenkelaars, H. J. W., and Boumans, P. W. J. M., Spectro- chim. Acta, Part B, 1983, 38, 1545. Davies, J., and Snook, R. D., J . Anal. At. Spectrom., 1986, 1, 195. Lichte, F. E., and Koirtyohann, S. R., “Induction Coupled Plasma Emission from a Different Angle,” paper presented at the Federation of Analytical Chemistry and Spectroscopy Societies, Philadelphia, PA, 1976, paper 26.Danielsson, A., ICP Inf. Newslett., 1978, 4, 147. Demers, D. R., Appl. Spectrosc., 1979, 33, 584. Abdallah, M. H . , Diemiaszonek, R., Jarosz, J., Mermet, J. M., Robin, J., and Trassy, C., Anal. Chim. Acta, 1976, 84, 271. Davies, J., Dean, J. R., and Snook, R. D., Analyst, 1985,110, 535. Faires, L. M., Bieniewski, T. M., Apel, C. T., and Niemczyk, T. M., Appl. Spectrosc., 1985,39, 5. Mermet, J. M., C. R. Acad. Sci., Ser. B , 1975, 281,273. Mermet, J. M., Spectrochim. Acta, Part B , 1975, 30, 383. Walters, P. E., Chester, T. L., and Winefordner, J. D., Appl. Spectrosc., 1977,31, 1 . Kornblum, G. R . , and de Galan. L., Spectrochim. Acta, Part B , 1977,32,71. Eckert, H. U., Spectrochim. Acta, Part B , 1978, 33, 591. Alder, J. F., Bombelka, R. M., and Kirkbright, G. F., Spectrochim. Acta, Part B , 1980,35, 163. Gunter, W. H., Visser, K., and Zeernan, P. B., Spectrochim. Acta, Part B , 1982, 37, 571. Nojiri, Y., Tanabe, K., Uchida, H., Haraguchi, H., Fuwa, K., and Winefordner, J. D., Spectrochim. Acta, Part B , 1983, 38, 61. Jarosz, J., and Mermet, J. M., J. Quant. Spectrosc. Radiat. Transfer, 1977, 17, 237. Uchida, H., Tanabe, K., Nojiri, Y., Haraguchi, H., and Fuwa, K., Spectrochim. Acta, Part B , 1981, 36, 711. Aeschbach, F., Spectrochim. Acta, Part B , 1982,37, 987. Boumans, P. W. J. M., Spectrochim. Acra, Part B , 1982,37,75. Fujimoto, T., J. Phys. SOC. Jpn., 1979,47, 265. Fujimoto, T., J. Phys. SOC. Jpn., 1979, 47, 273. Fujimoto, T., J. Phys. SOC. Jpn., 1980, 49, 1561. Fujimoto, T., J . Phys. SOC. Jpn., 1980,49, 1569. Batal, 0. O., and Merrnet, J. M., Can. J . Spectrosc., 1982,27, 37. Blades, M. W., and Hieftje, G. M., Spectrochim. Acta, Part B , 1982,37, 191. Holstein, T., Phys. Rev., 1947, 72, 1212. Holstein, T., Phys. Rev., 1951, 83, 1159. Blades, M. W., Spectrochim. Acta, Part B , 1982, 37, 869. de Galan, L., Spectrochim. Acta, Part B , 1984, 39, 537. Mills, J . W., and Hieftje, G. M., Spectrochim. Acta, Part B , 1984, 39, 859. Lovett, R. J., Spectrochim. Acta, Part B , 1982, 37, 969. Griern, H. R., “Plasma Spectroscopy,” McGraw Hill, New York, 1964. Griem, H. R., Adv. A t . Mol. Phys., 1975, 11, 331. Barnes, R. M., Crit. Rev. Anal. Chem., 1978, 7, 203. Kornblum, G. R., and de Galan, L., Spectrochim. Acta, Part B , 1977, 32, 455. Hieftje, G. M., Rayson, G. D., and Olesik, J. W., Spectro- chim. Acta, Part B, 1985,40, 167. Paper J5I5 7 Received December 9th, 1985 Accepted May 20th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100325

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 33 1 Multi-element Analysis of Ferrotungsten by Inductively Coupled Plasma Atomic Emission Spectrometry Ivan HlavaEek and lrena HlavaEkova Chemical Laboratories of Poldi, United Steelworks, Kladn 0, Czech oslo va kia An inductively coupled plasma atomic emission spectrometry (ICP-AES) procedure has been developed for the determination of aluminium, cobalt, chromium, copper, manganese, molybdenum, nickel, silicon, titanium and vanadium in ferrotungsten. The sample is dissolved by a combination oxalic acid and hydrogen peroxide and the decomposition is completed with a mixture of sulphuric and phosphoric acids in a PTFE vessel without application of hydrofluoric acid. In this way silicon is quantitatively converted into a solution form.No separation of elements from the matrix is required. Calibration graphs have been prepared using synthetic standard samples and verified by means of the Czechoslovak standard ferrotungsten and analysed ferrotungsten samples. For comparison, the samples were analysed by flame atomic absorption spectrometry. The silicon content was also verified by gravimetric and titrimetric methods. The results and accuracy of the method are discussed. Keywords: Inductively coupled plasma atomic emission spectrometry; flame atomic absorption spectrometry; ferrotungsten; multi-element analysis; silicon The analysis of ferrotungsten is often performed by classical analytical procedures that are laborious and time consuming. We have therefore been investigating suitable solution spec- tral methods in order to develop an analytical procedure that is appropriate for a wide assortment of analyte elements.Many ferroalloys have already been investigated by AAS, and also by ICP-AES. With ferrotungsten it has been observed that silicon remains in solution if phosphoric acid is used for the decomposition. Talvitiel has already employed this property of phosphoric acid for the decomposition of silicate minerals and Lucas and Ruprecht2 have also used it for the analysis of chrome ores and chrome - magnesite refractory materials by AAS. The possibilities of the multi-element analysis of ferrotungsten by ICP-AES have been verified and low concentrations of aluminium, cobalt, chromium, copper, manganese, molybdenum, nickel, titanium, vanadium and silicon have been determined, except where the aluminium and silicon are bonded as oxides.1 Decomposition procedures have been examined so that silicon can be taken into solution together with the other elements. Experimental Instruments An ARL 33000 LA sequential emission spectrometer with an inductively coupled argon plasma was used, equipped with a TI-59 programmable calculator (Texas Instruments, USA) connected on-line. A Henry Radio (USA) generator with an operating power of 1250 W, an operating frequency of 27.12 MHz and a maximum reflected power of 10 W was employed. The construction of the quartz plasma torch was designed according to Fassel. The pneumatic concentric nebuliser (Meinhard type) operates under a pressure of 0.3 MPa.The flow of carrier gas (argon) that passes through the nebuliser is moistened by means of a connected bubbler. The spectrometer consisted of a monochromator with a 1-m radius concave grating (1440 lines mm-1) in a Paschen - Runge mounting. The reciprocal dispersion is 0.695 nm mm-l in the first order. The atomic absorption spectrophotometer (Perkin-Elmer 503) was equipped with a source for electrodeless lamps. Reagents and Solutions All chemicals were of analytical-reagent grade and solutions were prepared with de-ionised water. All stock solutions of metals were prepared from high-purity metals of purity 99.9% or higher and stored in polyethylene bottles. Acids. The acids used were hydrochloric acid (36% mlm), nitric acid (65% rnlm), sulphuric acid (98% mlm), phosphoric acid (85% mlm) and oxalic acid.Hydrogen peroxide, 15% mlm. Aluminium stock solution, 1 mg ml-1. Dissolve 1.000 g of aluminium metal in 50 ml of hydrochloric acid (1 + 1) and a few drops of nitric acid (1 + 1). After dissolution, add 25 ml of concentrated hydrochloric acid and dilute to 1000 ml with water. Cobaltstock solution, 1 mg ml-1. Dissolve 1.000 g of cobalt metal in 20 ml of nitric acid (1 + 1) and dilute to 1000 ml with water. Chromium stock solution, 1 mg ml-1. Dissolve 1.000 g of chromium metal in 50 ml of hydrochloric acid (1 + l), add 10 ml of nitric acid (1 + 1) and dilute to 1000 ml with water. Copper stock solution, 1 mg ml-1. Dissolve 1.000 g of copper metal in 25 ml of nitric acid (1 + l), add 20 ml of concentrated hydrochloric acid and dilute to 1000 ml with water. Manganese stock solution, 1 mg ml-1.Dissolve 1.000 g of manganese metal in 50 ml of hydrochloric acid (1 + l), add 10 ml of nitric acid (1 + 1) and dilute to 1000 ml with water. Molybdenum stock solution, 1 mg ml-1. Dissolve 1.000 g of molybdenum metal in 50 ml of concentrated hydrochloric acid and 20 ml of nitric acid (1 + 1) and dilute to 1000 ml with water. Nickel stock solution, 1 mg ml-1. Dissolve 1.000 g of nickel metal in 20 ml of concentrated nitric acid, add 20 ml of concentrated hydrochloric acid and dilute to 1000 ml with water. Silicon stock solution, 1 mg ml-1. Fuse 1.070 g of annealed silicon dioxide with 10 g of sodium carbonate in a platinum crucible. Dissolve the melt in water and dilute to 500 ml. Titanium stock solution, 1 mg ml-1.Dissolve 1.000 g of titanium metal in 200 ml of hydrochloric acid (1 + 1). After dissolution, dilute to 1000 ml with water. Vanadium stock solution, 1 mg ml-1. Dissolve 1,000 g of vanadium metal in 10 ml of concentrated hydrochloric acid and 20 ml of nitric acid (1 + l), add 30 ml of sulphuric acid (1 + 1) and evaporate to fumes of sulphur trioxide. After dissolution in water, dilute to 1000 ml.332 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 Investigation and Choice of Instrumental Conditions for All measurements were performed using the spectral lines recommended by the manufacturer (for which secondary slits were mounted) except cobalt, for which the 350.23-nm line was used instead of the original 238.89-nm line.The cobalt line at 350.23 nm line was reached by displacement of the primary slit.3 The measurement was performed using the secondary slit determined originally for the ruthenium line at 349.89 nm. The original cobalt line at 238.89 nm was very strongly affected by iron (high background). The optimum signal to background ratio and the intensity of the line were used to determine the observation height (Fig. 1). The reproducibility of measurement was also taken into account. These measurements were performed in aqueous solutions of acids, i.e. , 250 ml of solution contained 5 ml of concentrated sulphuric acid and 25 ml of concentrated phosphoric acid , with additions of analyte elements at concentrations of 4 yg ml-1 for aluminium, cobalt, chromium, copper, manganese, nickel, titanium and vanadium, 8 pg ml-1 for molybdenum, 40 pg ml-1 for silicon, 0.6 mg ml-1 for iron and 1.4 mg ml-1 for tungsten. The instrumental conditions are given in Table 1.Table 2 shows the effects of interfering elements. All samples contained the same amounts of acids so that undesirable effects could be avoided (for example, viscosity, ICP-AES Table 1. Analyte elements and instrumental conditions (ICP) Wavelength/ Secondary slit Observation Element nm width/pm height/mm Al I . . , . 394.40 Co I* . . . . 350.23 Cr I . . . . 360.53 Cu I . . . . 324.75 Fe I1 . . . . 259.94 Mn I1 . . . . 257.61 Mo I . . . . 317.03 Ni I1 . . . . 231.60 Si I . . . . 251.61 Ti I . . . . 363.55 V I1 . . . . 311.07 W I . . . . 400.88 75 75 75 75 75 75 75 75 75 75 75 50 22 26 22 22 18 18 22 18 18 22 18 22 * Measured on secondary slit determined originally for Ru.Table 2. Interferences from added elements (ICP) Interfering element Background equivalent (concentration 1%) concentration, % Fe . . . . . . Al, 0.00005; Ni, 0.0001 ; Co, 0.00004; Si, 0.0005; Cr, 0.004; Ti, 0.0005; Cu, <0.00001; V, 0.00006; Mn, 0.00003; W, 0.0002; Mo, 0.001 0.0004; Cr, 0.00005; Si, 0.0015; Cu, 0.00001; Ti, 0.0004; Mn, 0.0005; V, 0.00025 Cr . . . . . . Ti, 0.0005 Mo . . . . . . Si, 0.0075; V, 0.0015; Ti, 0.080; W, 0.001; T i . . . . . . . . V, 0.010; W, 0.090 W . . . , . . Al, 0.0001; Mo, 0.0015; Co, 0.00014; Ni, density and surface tension effects). The matrix elements in both a blank sample and synthetic ferrotungsten samples were simulated from high-purity tungsten and iron metals.Analyses of samples with additions of analyte elements confirmed the linearity of the measurements within the considered range of concentrations in both aqueous solutions of acids and synthetic ferrotungsten samples. Investigation and Choice of Instrumental Conditions for FAAS FAAS was used as a checking method for verification of the accuracy of the results. The instrumental conditions are given in Table 3. The samples again contained the same amounts of acids and the matrix elements of synthetic ferrotungsten samples were simulated from high-purity metals. The linearity of the calibration graphs was verified with additions of analyte elements in both aqueous solutions of acids and synthetic ferrotungsten samples. 100 50 $? 2 u 100 2 .- $ 0 3 0) Y 2 50 0 ClJ c 0 a) 4- - .- ( n o 2 100 - 2 L 0 ' 50 >: 4- .- (n c 9) +- .- a ) O g 100 a .- - 9) 50 0 Al 1 1 1 1 pB 'x A cu ;"y Mo I I I 14 18 22 26 Fe 1 1 1 , \ A Ni ic\ 2B x A V 1 1 1 1 14 18 22 26 Cr 1 1 1 1 Mn , 1 1 1 Si - 14 18 22 26 Observation heightirnm Fig.1. Dependence of relative intensity on observation height (background corrected): A, relative intensity: B, relative signal to background ratio Table 3. Analyte elements and instrumental conditions (FAAS) Element c o . . . . c u . . . . Mn . . Mo . . Ni . . . . Si . . . . v . . . . Wavelengthhm Slit widthhm Flame . . 240.7 0.2 Air - C2H, . . 327.4 0.2 Air - C2Hz . . 280.1 0.2 Air - C2H2 . . 313.3 0.7 N 2 0 - C2H2, fuel rich . . 352.5 0.2 Air - C2H2 . . 251.6 0.2 N,O - C2H2, fuel rich .. 318.4 0.2 N,O - C2H2, fuel rich Burner lengthkm 10 10 10 5 10 5 5JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 333 Sample Preparation example, both a blank sample and synthetic calibration samples contained 0.15 g of iron metal and 0.35 g of tungsten metal for ferrotungsten samples containing about 30 _+ 5% of Fe and 70 k 5% of W. The calibration samples also contained additions of the analyte elements. ICP-A ES Weigh 0.5 g of sample into a PTFE beaker, add a mixture of 35 ml of hydrogen peroxide and 2.5 g of oxalic acid, dissolve by heating at about 100°C. Add 5 ml of concentrated sulphuric acid and 25 ml of concentrated phosphoric acid, evaporate to fumes of sulphur trioxide at a temperature of between 180 and 220 "C, cool and dilute to 250 ml with water.Results and Discussion Certificated values and analytical results for ferrotungsten samples obtained by FAAS and ICP-AES are summarised in Tables 4 and 5 , including the matrix composition. For silicon the results of gravimetric and titrimetric analyses are also given in Tables 4 and 5. The certificated value for the silicon content in the Czechoslovak standard ferrotungsten sample CSN 4-4-03 is only informative, and a difference between this certificated value and the ICP result was found. The silicon content was therefore verified by FAAS and titrimetry (titration of potassium hexafluorosilicate with sodium hydroxide solution after separation by membrane filtration). The results obtained confirmed the ICP result. Similarly, the difference between the certificated value and the ICP result for manganese was verified by means of FAAS.A few ferrotungsten samples were analysed by ICP-AES and FAAS to give a better comparison and the results are shown in Tables 4 and 5. The reproducibility of determination was studied and the basic statistical data are presented in Table 4 for FAAS and FAAS Weigh 0.5 g of sample into a glass beaker, add a mixture of 35 ml of hydrogen peroxide and 2.5 g of oxalic acid, dissolve by heating at about 100 "C, cool and dilute to 100 ml with water. The solutions obtained, which were used for measurements by FAAS, were clear. The solutions were not stable for long periods because they became turbid owing to precipitation of salts (oxalates). Comparative determinations of copper, cobalt, manganese, molybdenum, nickel, silicon and va- nadium were performed by FAAS.The measurements of aluminium, chromium and titanium were very unreliable (high detection limits, low reproducibility of measurements), and therefore these elements were omitted. A blank sample and synthetic samples were prepared by the same procedure as used for real ferrotungsten samples. The matrix elements were simulated from high-purity metals. For Table 4. Reproducibility of analysis of Czechoslovak standard ferrotungsten sample CSN 4-4-03 by FAAS and ICP and laboratory ferrotungsten sample A by ICP ("/.) V Sample Method CSN 4-4-03 (70.07% W, -29% Fe) Certificate Titrimetric FAAS Parameter* A1 Co Cr Cu Mn Mo Ni Si Ti 0.22t - 0.17 - - 0.109 - - - - - - - - - - Average SD RSD Average SD RSD (5 analyses) - <0.01 - - - - - - - (5 analyses) <0.01 <0.01 <O.O - - - - - - 0.165 0.10 0.02 0.015 0.18 - 0.010 0.004 0.002 0.002 0.015 - 5.9 4.2 10.6 14.0 8.3 - ICP 0.165 0.09 0.03 0.01 0.17 <O.O 0.015 0.008 0.002 - 0.007 - 9.2 9.2 7.7 - - 4.2 A (68.60% W, -29'y0 Fe) Gravimetric Titrimetric FAAS ICP 1.10 - 1.05 - 0.07 0.10 0.455 0.015 1.05 - - - - - - - - - - - <0.01 " (6analyses) <0.01 <0.01 0.015 0.065 0.11 0.48 <0.01 1.03 0.035 <0.01 SD - - 0.002 0.004 0 0.004 - 0.014 0 - RSD - - 12.9 6.9 0 0.85 - 1.4 0 - t Informative value.* SD, standard deviation; RSD, relative standard deviation. Table 5. Comparison of results for ferrotungsten samples ("/o) Ni Si 0.10 0.095 0.09 0.095 0.53 0.53 0.55 0.545 0.30 0.25 0.26 0.10 0.11 0.515 0.55 0.49 0.50 V Sample Method Titrimetric FAAS ICP Titrimetric FAAS ICP FAAS ICP E (68.80% W, -30'y0 Fe) .. FAAS ICP F (73.90% W, -25% Fe) . . Gravimetric Titrimetric FAAS ICP B (74.10% W, -25% Fe) . . Gravimetric C (68.35% W, -30% Fe) . . Gravimetric D (68.55% W, -30'~0 Fe) . . Gravimetric co c u Mn Mo - 0.08 0.085 - 0.35 0.40 - 0.02 0.01 - <0.01 <0.01 - 0.08 0.07 CO.01 <0.01 - <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 - - 0.025 0.025 0.44 0.48 0.25 0.265 - - - 0.35 0.37 - 0.02 0.015 0.02 0.01 0.02 0.01 - - - 0.01 <0.01 - <0.01 <0.01 <0.01 <0.01 CO.01 <0.01 - 0.12 0.13 0.105 0.09 0.12 0.12 - 0.12 0.135 0.07 0.05 0.05 0.055 - - - 0.085 0.10 - (0.01 (0.01 - <0.01 <0.01 - 0.08 0.07334 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 ICP-AES. Incorrect results were eliminated by the Grubbs test. The detection limits (defined as three times of the back- ground noise) for all the analyte elements in pure aqueous acid solutions and synthetic ferrotungsten solutions containing 0.6 mg ml-1 of Fe and 1.4 mg ml-1 of W are given in Table 6 for the wavelengths listed in Table 1. The presence of the matrix significantly influences the detection limits for some elements (Cr, Mo and Ti). The detection limits obtained by ICP-AES were taken to be about O . O l % , i. e. , 0.2 pg ml-1, for all analyte elements, considering that real ferrotungsten samples are often inhomogeneous. The measured values of the back- ground interference equivalent concentration of the matrix solution are also reported in Table 6.For some of the elements (Cr, Mo, Si and Ti) the source of higher values is spectral interferences (see Table 2). The concentrations found for some elements were below the detection limits and therefore a standard additions method was applied. Synthetic additions were carried out by adding known amounts of analyte elements in the expected concen- tration range to the Czechoslovak standard ferrotungsten sample CSN 4-4-03. The recoveries are shown in Table 7. The matrix elements were determined by ICP-AES only for a check on the matrix composition of the analysed samples, but not on the analytical results. These results were for information only. For practical use, the matrix composition of synthetic ferrotungsten samples was simulated with a preci- sion of +5% for both tungsten and iron (see the preparation of a blank sample and synthetic calibration samples).The contents of analyte elements were corrected for differences in the matrix composition of the samples if necessary (inter- ferences in Table 2). Table 6. Detection limits and background equivalent concentration (FLg m1-l) Detection limit Element A l . . . . . . co . . 1 . . . Cr _ . . . . . c u . . . . . . Mn . . . . . . Mo . . . . . . Ni . . . . . . Si . . . . . . Ti . . . . . . v . . . . . . Aqueous 0.03 0.02 0.02 0.01 0.01 0.03 0.10 0.06 0.04 0.01 Matrix 0.07 0.03 0.22 0.02 0.02 0.13 0.10 0.16 0.15 0.02 Background equivalent concentration 0.24 0.24 2.30 0.04 0.76 3.20 0.70 2.50 1 .oo 0.44 The influence of the matrix on the slopes of the calibration graphs was also investigated. Calibration graphs for each element are shown in Fig.2. These graphs were measured for pure aqueous acid solutions of the elements and for the synthetic ferrotungsten solutions containing 0.6 mg mi-' of Fe and 1.4 mg ml-i of W. Corrections were applied for the increase in the spectral background due to the presence of the matrix. For Co, Cr, Cu, Mn, Mo, Ni, Si, Ti and V the slopes of 100 50 8 >r 2 0 4- .- 4- 5 100 e, .- 4- - e, U 50 0 100 50 0 - 0 1 2 3 4 - 0 1 2 3 4 I L I I I l l 1 1 I , 0 2 4 6 8 0 1 0 2 0 3 0 4 0 Concentrationipg ml- Fig. 2. Calibration graphs (background corrected): A, pure aqueous acid solutions; B, synthetic ferrotungsten solutions Table 7. Results of synthetic additions to standard ferrotungsten sample CSN 4-4-03 (%) Element Added Al .. . . . . 0.05 0.10 0.15 0.20 Co . . . . . . 0.05 0.10 0.15 0.20 Cr . . . . . . 0.05 0.10 0.15 0.20 Cu . . . . . . 0.05 0.10 0.15 0.20 Mn . . . . 0.05 0.10 0.15 0.20 Found 0.051 0.099 0.147 0.198 0.050 0.102 0.151 0.200 0.051 0.104 0.153 0.204 0.053 0.102 0.152 0.201 0.050 0.103 0.153 0.202 Element Added Mo . . . . 0.125 0.25 0.375 0.50 Ni . . . . . . 0.05 0.10 0.15 0.20 Si . . . . . . 0.50 1 .oo 1 S O 2.00 Ti . . . . . . 0.05 0.10 0.15 0.20 V . . . . . . 0.05 0.10 0.15 0.20 Found 0.124 0.252 0.371 0.503 0.051 0.102 0.151 0.202 0.508 1.014 1 SO4 2.007 0.051 0.099 0.148 0.199 0.050 0.101 0.151 0.200JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 335 the two calibration graphs were almost identical, the differ- ences being less than 1% relative. This suggests that no significant matrix effect occurred for these elements. Small but detectable differences in the slopes of the calibration graphs for A1 were observed.For practical use, the matrix effect was eliminated as the blank sample and calibration samples contained almost identical amounts of the matrix elements. The effect of sodium salts that contaminated synthetic calibration samples taken from the silicon stock solution was established using ICP-AES. The effect of sodium on the intensities of the spectral lines for analyte elements up to a silicon content of 2% (for a mass of sample of 0.5 g , i.e., about 0.35 mg ml-1 of Na) was negligible. The effect of sodium on the intensity of the silicon line is shown in Fig. 3. Conclusion ICP-AES can be applied effectively to the multi-element analysis of ferrotungsten. The decomposition of samples in the presence of a substantial amount of phosphoric acid is slower than the normally used decomposition with hydrofluoric acid but the solutions obtained are clear and stable for long periods. For measurement, a quartz plasma torch and a glass nebuliser are used; they are not corroded as they are in the presence of hydrofluoric acid. A PTFE nebuliser and a fused alumina torch4 are not necessary. The decomposition pro- cedure is widely applicable and has been used with various other types of alloys and metals, including steels. The procedure described is more advantageous for ICP-AES than FAAS, where depressive effects of sulphuric and phosphoric acids are significant. 0.91 , , , 0 0.5 1 .o 1.5 Sodium concentration/mg mi-’ Fig. 3. Dependence of relative intensity for the spectral line Si 1251.61 nm on sodium concentration (background corrected). Silicon concen- tration, 29 pg ml-1; observation height, 18 mm References 1. Talvitie, N. A., Anal. Chem., 1951, 23, 623. 2. Lucas, R. P., and Ruprecht, B. C . , Anal. Chern., 1971, 43, 1013. 3. HlavBCek, I., and Hlavatkovi, I . , Hutn. Lisry, 1983,38, 512. 4. “Atomic Spectroscopy ICP/5500 System,” Perkin-Elmer, Norwalk, CT, September 1982, 16 pp. Paper J5l53 Received November 4th, 1985 Accepted May 8th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100331

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 337 Electrothermal Vaporisation Sample Introduction into an Atmospheric Pressure Helium Microwave-induced Plasma for the Determination of Iodine in Hydrochloric Acid Neil W. Barnett" and (the late) Gordon F. Kirkbright Department of Instrumentation and Analytical Science, UMIST, PO Box 88, Manchester M60 IQD, UK An inexpensive tantalum boat vaporisation device has been coupled directly to a microwave-induced plasma for the determination of iodine in hydrochloric acid. The vaporisation device was fabricated from polytetrafluoroethylene and powered by a Shandon Southern A3370 electrothermal atomiser. Detection limits of 2 x 10-10 g at 206.163 nm and 1 x 10-9 g at 506.120 nm were obtained for aqueous solutions. Keywords: Atomic emission spectrometry; electrothermal vaporisation; tantalum vaporiser; micro- wave-induced plasma; iodine The utilisation of the microwave-induced plasma (MIP) for spectrochemical analysis has recently been the subject of an extensive review by Matousek et al.1 In this paper, over 130 of the 286 references cited were directly concerned with sample introduction techniques for the MIP.Despite the considerable body of work and the spectroscopic advantages of the MIP, especially for the determination of the non-metallic ele- ments,'?* the technique does not enjoy the same degree of acceptance for elemental analysis as does the inductively coupled plasma (ICP).3 The determination of iodine using the MIP has been reported in numerous publications.4-l3 These reports either used gas chromatography (GC)4-10 for sample introduction or electrothermal vaporisation (ETV) .11-13 The iodine detection limits for GC-MIP varied considerably from 7 X 19-14 g s-l by McCormack et al.,4 which has never been approached closer than two orders of magnitude,9 to 1.4 x lO-9g s-1 reported by Dingjan and de Jong. lo In the above studies both arg0n~3~ and helium610 plasmas were used, operating at a t m o s p h e r i ~ ~ , ~ > ~ - ~ and reduced6.10 pressure, to excite iodine emission at two wavelengths, 206.1634>5,9 and 516.120 nm.6-8JO The three previously reported studiesll-13 concerned with iodine determination using ETV-MIP all employed atmo- spheric pressure plasmas and metallic vaporisation devices. As with the GC-MIP papers, the detection limits reported for ETV-MIP varied considerably, together with the iodine spectral lines that were utilised, that is, 1.2 x 10-9 g at 206.163 nm,ll2.6 x 10-6 g at 608.243 nm12 and 5 x g at 804.374 nm.13 These workers reported the use of platinum," tan- talum12 and tungsten13 vaporisers for the introduction of iodide and organoiodide compounds.However, Abdillahi and Snook14 have used a graphite ETV-MIP system for the determination of bromide at nanogram levels. This paper describes the development of analytical instrumentation and methodology to improve the previously reported detection capability of ETV-MIP for the determi- nation of iodine.11-13 In order to illustrate the utility of the analytical system developed, it was used to determine the iodide concentration in Aristar-grade hydrochloric acid.Experimental Instrumentation and Apparatus In 1977 Kirkbright's described a commercially available simultaneous multi-element ETV-MIP system. This instru- ment utilised an atmospheric pressure argon plasma, a * Present address: Department of Environmental Sciences, Plymouth Polytechnic, Drake Circus, Plymouth, Devon PL4 8AA, UK. tantalum vaporiser and a six-channel direct-reading spec- trometer. A PTFE electrothermal vaporisation chamber similar to the type used on the above system was provided for the present study by the manufacturer (EDT Research UK) and is shown diagramatically in Fig. 1. The specifications of the instrumental components used in this investigation are listed in Table 1 and a schematic representation of the complete analytical system is given in Fig.2. Wavelength selection and calibration were achieved using the apparatus shown in Fig. 3. Small amounts (ca. 2 mg) of crystalline iodine were placed in the conical flask (10 ml) in order to introduce iodine vapour into the helium plasma gas Table 1. Description and specifications of the components in the analytical system Component Specifications Microwave generator and reflected power meter . . Microwavecavity . . . . Monochromator . . . Optical arrangement . . Readoutsystem . . . . Electrothermal vaporiser and power supply . Plasmagascontrol . . . Dischargetube . . Microtron 200 MkII, 2450 MHz, 200 W (EMS, Wantage, UK) TMolo type as described by Beenakkerl7( EMS) D330,300 mm focal length Czerny - Turner type, 1200 lines mm-1 plane grating blazed at 300 nm, reciprocal linear dispersion 2.57-2.40 nm mm-l in the range 200-600 nm,f/5.5 aperture (Hilger Analytical, UK) Cavity and monochromator mounted on the same optical bar.Plasma viewed axially and imaged 1: 1 on to the entrance slit using a fused-silica lens of 40 mm diameter, 70 mm focal length 6256B photomultiplier tube (EM1 Electronics, UK), 476R power supply (Brandenburg, UK), Model 3000 250-ms f.s.d. chart recorder (Oxford Instruments, UK) Tantalum (99.9%) foil, 25 X 8 X 0.125 mm (Goodfellow Metals, Cambridge, UK), 100 A, 12 V transformer controlled by an A3370 electrothermal atomiser (Shandon Southern Instruments, UK) R-2-15-D volume flow controller with sapphire float, 60-650 ml min-1 He (Brooks Instruments, The Netherlands) Fused-silica tubes, 6 mm 0.d.X 2 mm i.d. (Hereaus Silica and Metals, Merseyside, UK)338 A J 1986, VOL. 1 200 mm H Fig. 1. Diagram of the PTFE electrothermal vaporisation chamber. A, Aluminium retaining ring; B, brass electrodes; C, air-tight PTFE “bung”; D , stainless-steel vaporiser mountings; E, helium carrier gas inlet tube; F, sample introduction port; G, PTFE vaporisation chamber (volume ca. 30 ml); H, observation window; I, air-tight torch adaptor; J , locating pin; and K, fastening screws for Ta boat I . 1 Microtron 200 Mkll u R;t;gd I I Vaporisation ALlial;luGl u--w, OSi,ica lens u TMOlO cavity A3370 electrothermal vaporisation power supply flow controller volul - D330 monochromator Fig. 2. Schematic representation of the ETV-MIP analytical system illustrating all the major instrumental components and subsequently into the MIP.The wavelength calibration device was connected to the plasma discharge tube in the position normally occupied by the vaporisation chamber. This type of wavelength calibration has been successfully utilised in a previous paper.16 Sample aliquots (15 pl) were delivered to the tantalum boat using a MA-SV microsyringe pipette with disposable PTFE tips (SGE, Australia). All volumetric apparatus was soaked in a solution of Decon 90 (5% VIV) for 24 h, thoroughly rinsed in distilled water and then soaked in dilute nitric acid (1 + 9). When the equipment was required for use it was removed from the acid and rinsed with distilled water. Reagents Potassium iodide (99.8% pure) and iodine (99.9% pure) were of AnalaR grade (BDH Chemicals, UK).Singly distilled water was used for all aqueous solutions. Chloroform (density 1.444-1.445 g ml-1) was of Aristar grade (BDH Chemicals). A stock solution of iodide (1000 mg 1- l) was prepared from potassium iodide as described by Dean and Rains.18 From this stock solution working standard solutions were prepared as required by serial dilution. A stock solution of iodine in chloroform (100 mg 1-1) was also prepared for serial dilution with chloroform as required. Aristar-grade hydrochloric acid (BDH Chemicals) was used as a sample for iodine determination. Helium was supplied by British Oxygen Company, Special Gases, UK. Calibration Procedures For the determination of iodide in hydrochloric acid, 2-ml aliquots of the acid were delivered into calibrated flasks (5 ml) that contained standard additions of iodide and made up to the mark.Aliquots of these solutions (15 p1) were introduced to the ETV-MIP for iodine determination. Analytical figures of merit and calibration graphs were obtained using both aqueous iodide and iodine in chloroform solutions. Instrumental Parameters Some of the instrumental parameters, such as wavelength selection vaporisation conditions and plasma gas flow-rate, were determined experimentally and are detailed under Results and Discussion.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 339 The forward power to the cavity was maintained at 130 W for all experiments with a resultant reflected power reading of 17 W.The forward and reflected power readings did not vary significantly over periods in excess of 6 h, provided that the helium flow-rate remained constant. However, it was neces- sary to re-adjust the tuning stubs of the cavity following any variations in plasma gas flow-rate. Prior to the inclusion of the volume flow controller in the ETV-MIP system, the forward and reflected power settings appeared to be markedly less stable. The entrance and exit slits were set and maintained at 25 ym, which correspond to a spectral band pass of 0.06 nm. The photomultiplier tube was operated at 1.2 kV, the output from which was monitored using the chart recorder. The sample introduction volume was kept constant at 15 yl, as this volume could be reproducibly delivered from the microsyringe and was accommodated adequately within the depression in the tantalum boat.A small V-shaped depres- sion was pressed into each tantalum boat by gently striking the handle of a screwdriver while the blade was centrally located on the tantalum strip supported by a wooden block. Results and Discussion The selection of the most sensitive spectral line for iodine determination was achieved using the device shown in Fig. 3. With tap A closed and taps B and C open, the 10-ml conical flask was purged with helium (250 ml min-1); the flask also contained crystalline iodine (ca. 2 mg). After the purging was complete (2 min), taps B and C were closed and tap A was opened, with a helium flow-rate of 250 ml min-1; the plasma was initiated by the insertion of a length of tungsten wire into the discharge tube.After setting the forward power (130 W) and tuning the cavity to give minimum reflected power (17 W), the discharge was allowed to stabilise for 20 min, after which time tap C was opened very slightly to allow a sufficient amount of iodine vapour into the plasma so as to produce the type of signals shown in Fig. 4. These five iodine spectral lines in the range 20Cb600 nm were located with respect to the position of helium spectral lines in the MIP.19 The five lines were then scanned using the motor drive on the monochroma- tor, at 2.5 nm min-1. The results of these scans are reproduced in Fig. 4. The relative signal to noise ratios (SNRs) of four of A Helium- - To MIP Fig. 3. Wavelength calibration device the five spectral lines were determined by taking three scans of each line at the respective chart recorder sensitivities shown in Fig.4. These signals were ratioed against the base-line peak-to-peak noise level measured at 10 mV f.s.d. in the region above and below the spectral line. All signals were normalised to a sensitivity of 100 mV f.s.d. and the SNRs subsequently calculated; these values are given in Table 2 together with some possible assignments for the observed spectral interferences. The above spectral interferences were thought to be the result of the concomitant impurities in the helium as they were observed at the same wavelength and intensity with or without iodine in the plasma. The choice of spectral lines for iodine determination was made on the basis of the SNRs.Clearly, the 206.163-nm line was the first choice using SNR as the criterion. Tanabe et a1.21 have suggested that there is the possibility of interference from NO and CO bands at this wavelength. However, Pearse and Gaydon20 have designated the emission intensity as “weak” from these band heads in the region of the 206.163-nm line. During the course of this study no such interference was observed at this line. The 516.120-nm line was also evaluated for analytical utility as many earlier workers”8J0 had used this wavelength successfully for iodine determinations. As the Shandon Southern A3370 was originally designed to control a graphite atomiser for AAS, the control of a tantalum 206.163 nrn 516.120 nrn 258.2 256.624 nm ! 9 nrn 1 100 rnV f.s.d.50 rnV f.s.d. OH band + 100 rnV f.s.d. 5 50 rnV f.s.d. Fig. 4. Scans in the vicinity of the five iodine emission lines identified by their respective wavelengths showing the recorder sensitivity as mV full-scale deflection (f.s.d.) Table 2. Signal to noise ratios for various iodine emission lines in the MIP with possible molecular interfering species Spectra line Relative Possible interfering wavelengthhm SNR species20 206.163 560: 1 256.624 45 : 1 NO, CO+ NO, CO+ 258.279 65 : 1 - OH 287.863 516.120 68 : 1 c2340 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 % 1000 v! > 500 E z c c m E 100 E 50 . .- v, .- 5 g 10 .- -0 -0 5 - foil vaporiser presented a problem owing to the difference in resistivity between graphite and tantalum. As a consequence of the much lower resistance of the tantalum boat, only the low-power settings were used for the vaporisation programme in order to avoid drawing the maximum rated current from the transformer.Puschel et ~ 1 . 2 2 overcame this difficulty by modifying a commercial graphite furnace power supply for use with a tungsten tube atomiser, the major modification being the replacement of the original transformer (12 V, 150 A) by one with a lower voltage secondary winding and higher current carrying capacity (4 V, 450 A). A satisfactory three-stage heating programme was achieved using the system described above. For the determination of iodide in aqueous solution and dilute hydrochloric acid, the low settings on the two drying stages and one ashing stage were employed.Owing to the much lower resistivity of the tantalum boat there was no need to use the atomise stage of the A3370. The settings used for iodide were as follows: Dry 1, 0.5 for 90 s; Dry 2, 1.0 for 45 s; and Ash, 3.0 for 3 s (iodine vaporisation stage). During the initial experiments with the tantalum boat vaporiser, the plasma discharge tube was disconnected from the vaporisation chamber during the first two stages of the heating programme, after which the chamber was re-connec- ted to the discharge tube, the plasma initiated and finally the sample was vaporised and transported into the MIP. This procedure resulted in poor reproducibility owing to switching the plasma on and off about every 3 min and thus the plasma was not stabilised. Further experiments showed that the disconnection of the discharge during the first two stages was unnecessary as the plasma was not extinguished by the evolved vapours.The background emission intensity did rise appre- ciably during the Dry 1 and Dry 2 cycles, but returned to its initial level prior to the final stage of vaporisation programme. The plasma was also tolerant towards the ingress of air when the samples were delivered to the tantalum boat. A small and reproducible signal was observed when the vaporisation cycle was carried out without any sample in the boat. This may have been the result of a small pulse of hot helium that physically perturbed the discharge, as has been reported by other workers.23 The optimisation of the helium flow-rate was achieved by a univariate search in which aliquots of aqueous iodide solution (containing 75 ng of I) were introduced into the plasma at various helium flow-rates.The results are shown graphically in Fig. 5 and indicate that the optimum helium flow-rate was in the range 200-300 ml min-1. Consequently, the plasma gas flow-rate was maintained at 250 ml min-1. Using the ETV-MIP system and the previously stated instrumental parameters, calibration graphs and analytical figures of merit were obtained for aqueous iodide standards at the two selected wavelengths. Some typical signal responses for various masses of iodide at 206.163 nm are shown in Fig. 6. The off-line signal in Fig. 6 was measured at 0.10 nm above the spectral line. The two calibration graphs shown in Fig. 7 illustrate the linear dynamic ranges of the two spectral lines used in this study.However, no samples were introduced that contained more than 1500 ng of iodide and therefore the upper limits of the linear dynamic ranges were not determined. The analytical figures of merit are listed in Table 3. The determination of the iodide content of Aristar-grade hydrochloric acid was performed using the method of standard additions calibration procedure and the results are plotted graphicaly in Fig. 8. From this graph the concentration of iodide in Aristar hydrochloric acid was found to be 0.94 mg 1-1, which is consistent with the BDH specification of less than 5 mg I-'. The signal response for iodide in the hydrochloric acid solutions was depressed by <1% compared with that recorded for pure aqueous iodide standards shown in Fig.6. Standard solutions of iodine were also prepared in chloro- form in the range 0.05-1.00 mg 1-1 and introduced to the - - - - - 'E I I I 100 200 300 400 500 .- Helium flow-rateirnl rnin-1 -a 0 - Fig. 5. Iodine emission intensity at 206.163 nm versus helium flow-rate for 75-ng injections of iodide in aqueous solution Off-line - 3.75 h 37.5 Blank Fig. 6. Some typical signal responses for various masses of iodide in aqueous solution measured at 206.163 nm and 10 mV f.s.d. The numbers above each peak relate to the mass of iodide injected in nanograms 5000 nm 500 1000 5( ( 1 1 1 5 10 50 100 Mass of iodidehg 30 Fig. 7. Calibration graphs for aqueous iodide solutions at the two wavelengths indicated, showing linear dynamic ranges in excess of 2.5 orders of magnitudeJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 341 Table 3. Analytical figures of merit for aqueous iodide solutions in the ETV-MIP Relative standard Wavelengthhm Detection limit/g deviation, % 206.163 2 x 10-1" 1.9* 516.120 1 x 10-9 Not determined * Calculated for twelve determinations of 15 ng of iodide. 1.25 0.5 0 0.5 1.25 2.50 5.( Iodide found/ Iodide addedipg ml-' pg ml - l Fig. 8. Method of standard additions calibration graph for the determination of iodide in Aristar hydrochloric acid using ETV-MIP ETV-MIP system. However, only the first stage of the heating cycle was required to vaporise completely the chloroform and molecular iodine from the tantalum surface. With the power set at 1.5 for 7 s and all other instrumental parameters maintained as for the aqueous samples, a calibration graph was obtained (Fig.9) for the 206.163-nm iodine spectral line. As the iodine and chloroform were introduced simultaneously into the plasma there was a significant change in the background emission level as a result of the presence of a relatively large loading of chloroform in the plasma either physically perturbing the discharge or causing increased molecular band emission from species such as CO as already mentioned.20.21 As the sample entered the MIP the discharge became distinctly green, probably owing to the emission from the Swan system of the C2 species.2" Evidence to support this idea was obtained when signals from the iodine in chloroform standards were monitored at 516.120 nm.At this wavelength the background emission from 1 pl of pure chloroform was off the recorder scale at 1.0 V f.s.d. This observation is consistent with the wavelength of the most intense bandhead of the Swan C2 system being at 516.52 nm.20 As a consequence, the attempt to produce a calibration graph for iodine in chloro- form at 516.120 nm was abandoned. As the background signal due to the chloroform at 206.163 nm was significant, the contribution of this signal was quantified. By introducing twelve 15-pl aliquots of pure chloroform into the MIP and calculating the mean and standard deviation of the resultant signals, the magnitude and reproducibility of the background contribution were deter- mined. The results of this experiment showed that the background signal was averaged at 1.2 mV with a standard deviation of 0.1 mV, which corresponded to a relative standard deviation of 8.3%.0 5 10 1 Mass of iodinehg Fig. 9. at 206.163 nm Calibration graph for iodine in chloroform standard solutions Whereas the background emission from the chloroform degraded the analytical precision of the technique, the emission intensity for iodine at 206.163 nm corrected for the background emission showed a marked improvement in sensitivity over the aqueous iodide standard solutions. The degraded precision together with the increased sensitivity can be observed in the calibration graph in Fig. 9. On average, the sensitivity compared with the aqueous calibration was calcu- lated to be a factor of four higher. The 2a detection limit for the iodine in chloroform standards was not calculated owing to the relative irreproducibility of the background signal.It was felt that the improvement in sensitivity was a consequence of the increased transport efficiency of the analyte from the tantalum boat to the plasma. It would seem reasonable that at temperatures below 300 "C iodine mole- cules would be virtually unreactive compared with the species vaporised from the aqueous iodide solutions at temperatures approaching 2000 "C. This argument is supported by the large variation in the detection limits of previous workers using GC-MIP4-10 compared with those using ETV-MIPl1-13 for iodine determinations. As indicated in the Introduction, the best detection limit for ETV-MIP" represents the inferior end of the range for GC-MIP.1" Conclusion This study illustrates the analytical utility of the atmospheric pressure helium MIP for low-level iodine determinations using relatively inexpensive instrumentation.It has been shown that the detection capability for iodine in the ETV-MIP system was improved by an order of magnitude. N. W. B. gratefully acknowledges the support, encourage- ment and opportunity to undertake this study, all of which were provided by his late friend and colleague Gordon Kirkbright. References 1. 2. 3. 4. 5. Matousek, J . P., Orr, B. J., and Selby, M., Prog. Anal. At. Spectrosc., 1984,7,275. Zander, A. T., and Hieftje, G. M. , Appl. Spectrosc. , 1981,35, 357. Barnes, R. M., CRC Crit. Rev. Anal. Chem., 1978,7, 203. McCormack, A. J., Tong, S.C., and Cooke, W. D., Anal. Chem., 1965, 37, 1470. Bache, C. A., and Lisk, D. J., Anal. Chem., 1966, 38,783.342 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. McLean, W. R., Stanton, D. L., and Penketh, G. E., Analyst, 1973, 98, 432. Beenakker, C. I. M., Spectrochim. Acta, Part B , 1977,32,173. Quimby, B. D., Delaney, M. F., Uden, P. C., and Barnes, R. M., Anal. Chem., 1979, 51, 875. Chiba, K., and Haraguchi, H., Anal. Chem., 1983, 55, 1504. Dingjan, H. A., and de Jong, H. J., Spectrochim. Acta, Part B , 1983, 38, 777. Aldous, K. M., Dagnall, R. M., Sharp, B. L., and West, T. S., Anal. Chim. Acta, 1971, 54, 233. van Dalen, H. P. J., Kwee, B. G., and de Galan, L., Anal. Chim. Acta, 1982, 142, 159. Rait, N., Golightly, D. W., and Massoni, C. J., Spectrochim. Acta, Part B, 1984, 39, 931. Abdillahi, M., and Snook, R. D., Analyst, 1986, 111,265. Kirkbright, G. F., paper presented at the Pittsburg Conference on Analytical Chemistry and Applied Spectroscopy, Cleve- land, OH, March 1977, paper P-454. Barnett, N. W., Chen, L. S., and Kirkbright, G. F., Spectro- chim. Acta, Part B , 1984,39, 1141. Beenakker, C. I. M., Spectrochim. Acta, Part B , 1976,31,483. 18. Dean, J. A., and Rains, T. C., in Mavrodineanu, R., Editor, “Procedures Used at the National Bureau of Standards to Determine Selected Trace Elements in Biological and Botan- ical Materials,” Special Publication No. 492, National Bureau of Standards, Washington, DC, 1977, p. 254. Weast, R. C., and Astle, M. J., Editors, “CRC Handbook of Chemistry and Physics,” Sixtieth Edition, CRC Press, Boca Raton, FL, 1981. 20. Pearse, R. W. B., and Gaydon, A. G., “The Identification of Molecular Spectra,” Fourth Edition, Chapman and Hall, London, 1976. 21. Tanabe, K., Haraguchi, H., and Fuwa, K., Spectrochim. Acta, Part B , 1981,36, 119. 22. Puschel, P., Formanek, Z., Hlavac, R., Kolihova, D., and Sychra, V., Anal. Chim. Acta, 1981, 127, 109. 23. Carnahan, J. W., and Caruso, J. A., Anal. Chim. Acta, 1982, 136,261. 19. Paper J6l4 Received February 13th, 1986 Accepted April 14th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100337

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 343 Inductively Coupled Plasma Atomic Fluorescence Spectrometric Determination of Cadmium, Copper, Iron, Lead, Manganese and Zinc Richard F. Sanzolone US Geological Survey, Box 25046, Denver Federal Center, MS 955, Denver, CO 80225, USA An inductively coupled plasma atomic fluorescence spectrometric method is described for the determination of six elements in a variety of geological materials. Sixteen reference materials are analysed by this technique to demonstrate its use in geochemical exploration. Samples are decomposed with nitric, hydrofluoric and hydrochloric acids, and the residue dissolved in hydrochloric acid and diluted to volume, The elements are determined in two groups based on compatibility of instrument operating conditions and consideration of crustal abundance levels. Cadmium, Cur Pb and Zn are determined as a group in the 50-ml sample solution under one set of instrument conditions with the use of scatter correction. Limitations of the scatter correction technique used with the fluorescence instrument are discussed.Iron and Mn are determined together using another set of instrumental conditions on a 1-50 dilution of the sample solution without the use of scatter correction. The ranges of concentration (pg g-l) of these elements in the sample that can be determined are: Cd, 0.3-500; Cur 0.4-500; Fe, 85-250 000; Mn, 45-1 00 000; Pb, 5-1 0 000; and Zn, 0.4-300. The precision of the method is usually less than 5% relative standard deviation (RSD) over a wide concentration range and acceptable accuracy is shown by the agreement between values obtained and those recommended for the reference mate ri a I s.Keywords: Inductive1 y coupled plasma atomic fluorescence spectrometry; geological materials; major and trace element determinations Quantitative trace element procedures useful for geochemical exploration purposes have been, in the past, mostly single- element determinations. However, the development of instru- ments such as the inductively coupled plasma atomic emission spectrometer1 and more recently the development of the inductively coupled plasma atomic fluorescence spec- trometer2-3 has brought simultaneous, multi-element capabili- ties to geochemical analysis. This paper describes a procedure for multi-element analysis of geological materials using a commercially available inductively coupled plasma atomic fluorescence spectrometer.It consists of an array of up to 12 element modules surrounding an ICP atomisation source. Each module contains a hollow-cathode lamp as an excitation source, an optical filter and a photomultiplier tube. The hollow-cathode lamps are pulsed in sequence and the detec- tion electronics are synchronously grated so that at any given time only one atomic fluorescence signal is produced and detected. Demers and Allemand3 and Demers et aL4 des- cribed the operational configuration of the instrument. The procedure uses a nitric, hydrofluoric and hydrochloric acid dissolution technique prior to determination of six elements by fluorescence spectrometry.A discussion of the conditions that can cause scattered-light problems, as well as the operation and limitations of the scatter-correction tech- nique is presented. Experimental Apparatus A commercially available inductively coupled plasma atomic fluorescence spectrometer, the Plasma/AFS, manufactured by the Baird Corporation,* was used for all fluorescence measurements in this study. The operational configuration of the instrument was described by Demers et d4 The instru- ment is capable of determining up to 12 elements simul- taneously using self-contained, interchangeable element modules. Six element modules were used in this study (Table 1.) * Use of trade names is for descriptive purposes only and does not imply endorsement by the US Geological Survey.The spectrometer was interfaced with an Apple 11+ computer and an Epson MX-80 printer for the collection and reduction of data.5 Line voltage regulation to the instrument was provided by a Sola CVS constant voltage transformer as recommended by the manufacturer. The sample introduction system incorporated a Gilson Minipuls I1 pump set at 500 units, a cross-flow nebuliser with 0.013 in i.d. capillary tubing and the standard quartz plasma torch supplied by the manufacturer. The delivery rate of sample solution was ca. 1 ml min-1. Measurements were made using a 5-s preburn and 20-s integration time. A Varian AA-1475 atomic absorption spectrometer was also used in this study. An atomic absorption determination of manganese in GXR-1 reference material was made using the standard procedures and instrument settings recommended by the manufacturer.This was carried out because no adequate quantitative reference value for manganese could be found in the literature. Reagents and Standards Stock solutions, each containing 1000 pg ml-1 in 5% VIV nitric acid, were prepared from metals or oxides for Cd, Cu, Fe, Mn, Pb and Zn (SPEX Industries, Metuchen, NJ, USA). A combined standard solution containing 1 pg ml-1 of the above metals was prepared daily by dilution of the stock solutions with de-mineralised water and was used for instrument calibration. A 1000 pg ml-1 solution of aluminium in 10% V/V hydrochloric acid was prepared from (SPEX Industries) and used for calibration of the scatter channel. Reference Materials Geological reference materials used in this study came from three sources.The six GXR samples are geochemical refer- ence materials prepared by the US Geological Survey.6 The eight GSD samples are stream and pond sediments obtained from the Institute of Geophysical and Geochemical Explora- tion, the Peoples Republic of China.’ The BCSS and MESS samples are marine sediments purchased from the National Research Council of Canada.8344 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 Table 1. Instrument module settings for elements in this study Observation height*/ Wavelength/ Lamp current/ Element mm nm mA c u . . . . . . 140 324.8 Cd . . . . . . 135 228.8 Pb . . . . . . 130 283.3 Zn . . . . . . 120 213.9 W (scatter monitor) . . . . 130 295.6 Fe 125 279.5 Mn .. . . . . 140 248.3 * From tip of injection nozzle of ICP torch. t Super Lamp, with 100-mA boost current used (manufactured by Photron, Australia) . . . . . . 10 6 6 t 11 20 18 18 Photomultiplier/ V 560 730 700 720 650 660 650 Table 2. Detection limits, linear ranges and determination ranges in the samples considered. Detection limit calculations based on twice the standard deviation of the background noise Detection limit Detection limit Upper linear Determination range range in with scatter without scatter correction/ correction/ limit/ sample/ c u . . . . . . 0.004 0.002 15 0 . 4 1 500 Element pg ml-' pgml-1 pgml-1 w g - ' Cd . . . . . . 0.003 0.001 5 0.3-500 Pb . . . . . . 0.050 0.034 100 5- 10 000 Zn . . . . . . 0.004 0.002 3 0.4-300 Fe . . . .. . N.D.* 0.017 50 85-250 000 Mn . . . . . . N.D.* 0.009 20 45-100 000 * N.D., not determined. Procedure Sample decomposition The digestion is a modification of a published procedure incorporating greater sample mass and reagent volumes.9 Weigh 0.500 g of <lOO-mesh soil, rock or sediment into a 100-ml PTFE beaker and moisten the sample with a few drops of water. Add 10 ml of concentrated nitric acid and place the beaker on an oscillating hot-plate, pre-set at 135-140 "C, for 10 min. Add 15 ml of concentrated hydrofluoric acid and evaporate to dryness. Add 10 ml of concentrated hydrochloric acid and again evaporate to dryness. Remove the beaker from the hot-plate, add 10 ml of 50% V/V hydrochloric acid, loosen the residue with a PTFE spatula to aid dissolution and return to the hot-plate for 2-3 min.Transfer the solution and any remaining residue into a 50-ml calibrated flask and dilute to volume with water. Allow the residue to settle before aspirating the solution into the spectrometer. Sample analysis Element determinations were performed using the menu mode programme of the inductively coupled plasma atomic fluorescence spectrometer. The instrument was calibrated in this mode using a de-mineralised water blank solution and the 1 yg ml-1 mixed element standard solution for two-point calibration. Two-point calibration is possible because the 1 pg ml-1 standard solution is well within the large linear range of about four orders of magnitude of the atomic fluorescence technique (Table 2). This wide calibration range is an advantage over atomic absorption techniques that have a much smaller linear range.The tungsten element module was used as the scatter monitor and was calibrated using a de-mineralised water blank and the 1000 yg ml-1 aluminium solution. Each element module was independently optimised for lamp current, photomultiplier voltage and observation height according to optimisation procedures previously established to maximise sensitivity.10 Table 1 lists the optimum module settings. Consideration of both expected element crustal abundance and the compromise plasma conditions necessary for simul- taneous multi-element determination of the elements studied necessitated dividing the elements into two groups. Copper, Cd, Pb and Zn were determined as a group using the 50-ml sample solution from the digestion with the use of scatter correction. The plasma conditions for this group were: r.f.power, 400 W; carrier flow, 1.60 1 min-1; coolant flow, 10 1 min-1; and propane flow, 20 ml min-1. The addition of propane, through the nebuliser spray chamber, promotes the dissociation of any metal oxides into free atoms and thereby improves sensitivity. l1?l2 Iron and manganese benefitted most from the introduction of propane. Iron and Mn were determined together using a 1 to 50 dilution of the sample solution without the use of scatter correction. The plasma conditions for these elements were: r.f. power, 550 W; carrier flow, 1.60 1 min-1; coolant flow, 11 1 min-1; and propane flow, 50 ml min-1. Copper and Zn were also determined under these instrument conditions when the concentration of these elements was too high to be determined in the ariginal 50-ml sample solution.Results and Discussion Sensitivity Detection limits were determined with and without the use of scatter correction using the module settings listed in Table 1 and the instrument operating conditions listed under Sample analysis. Table 2 lists the detection limits for elements determined in this study. The use of scatter correction increases the detection limit values by a factor of 2-3 owing to the noise from the scatter monitor channel being added to the noise of the other channels during scatter correction. Table 2 also lists linear range values for elements deter- mined in this study. Linear range values are based on a 5% deviation from linearity in the calibration graph.The fluores- cence technique can suffer from reversals in the analytical graph, which results in the same signal intensity for two vastlyJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 345 Table 3. Cobalt and nickel values (pg g-1) in three reference materials with and without scatter correction and comparison with reference values Value without Value with Sample Element value* correction correction Reference scatter scatter GXR-1 . . . . . . c o Ni GXR-2.. . . . . c o Ni GXR-3.. . . . . co Ni * Sanzolone et a1.9 8.8 * 42.5 8.3 14.4 59.2 60.4 80 60 21 27 101 88 33 32 17 16 47 48 different element concentrations. To eliminate this possibility, the instrument is designed to register a signal high (SH) response when the analyte signal is on the flat or downward side of the curve.This is carried out by means of a sensing circuit in the detection electronic system of the spectrometer that measures the rise time of the signal. To test the operation of this sytem for the elements studied, a range of concentra- tions from three times the upper limit of the linear range to as high as three orders of magnitude above the linear range was tested. In all instances the instrument gave the SH response, confirming the usefulness of this circuit for these elements. Scattering Scattering of the incident source radiation from the surface of salt particles present in the atomisation cell is a well-known drawback to the fluorescence technique." The use of an inductively coupled plasma as an atomisation source has reduced the problem of scattering; however, it can still be a significant problem in some instances.Samples with a high content of certain matrix elements such as Al, Ca, Mg and Si can cause considerable scattering. Silicon is not a problem in this study because it is volatilised in the sample decomposition procedure. However, Al, Ca and Mg at concentrations above ca. 1000 pg ml-1 (equivalent to 10% in the sample) do cause scattering problems. These elements are not a problem when determining Fe and Mn because the concentration of scatter- ing salts in the 1 to 50 dilution of the sample solution is low enough not to cause scattering. Scattering caused by Al, Ca and Mg is, however, a potential problem for determinations in the original 50-ml sample solution. To compensate for false analyte signals caused by scattering salts, the instrument manufacturer has provided a technique of correction based on the dual line method.lS'5 A module of an element known to be absent from the sample solution at detectable levels is installed in a channel of the instrument and used for scatter correction.A tungsten module was chosen for scatter correction in this study. The observation height (ca. 130 mm) was set to match that of the other modules. At this high observation height no tungsten atomic fluorescence signal is obtainable at any concentration, as the fluorescence signal for tungsten occurs between 85 and 95 mm. Conse- quently the presence of tungsten in any of the samples will not cause the module to overcorrect for scatter by detecting a tungsten analyte signal from the sample solution.To calibrate the scatter channel, a solution of 1000 pg ml-l of aluminium is used. During this calibration, the scatter is measured on the signal produced by the aluminium on all instrument channels and the ratios of these signals to the signal measured by the tungsten-monitoring channel are calculated. The instrument is then calibrated by the normal calibration process for the elements to be determined. The ratios established during the calibration of the scatter channel are used to adjust the signal of each analyte channel and, thus, to correct for scattering when samples are analysed. The system works well if the scattering is a result of a single component of the matrix and the scatter channel is calibrated against that component ( L e ., Al). However, the system may not ade- quately compensate for the scattering if it is produced by several matrix components, as in geological materials. This inability to compensate adequately may be because the light-scattering characteristics of other matrix salts differ from those of the salt used for scatter calibration. An additional observation is that the light-scattering charac- teristics of the various matrix components are not necessarily additive. For example, a 1000 pg ml-1 solution of a matrix component may generate a signal equivalent to x pg ml-1 in a given analyte channel. A 1000 pg ml-1 solution of another matrix component may generate a signal equivalent to 2x pg ml-1 in the same analyte channel, whereas a combined solution containing 1000 pg ml-1 of both matrix components may generate a signal in the analyte channel that is either greater or less than the expected 3x signal.This non-additive nature can cause errors when using scatter correction. These observations have been made previously by Naranjit et al. ' 6 The accuracy of the scatter-correction system depends not only on the similarity of the calibrated and actual scattering materials, but also on the fluorescence sensitivity of the analyte and its concentration in the sample solution. The correction system works well for elements such as Cu, Cd and Zn where the analyte fluorescence signal is large relative to the scatter signal. Values obtained for these elements in this study and their good agreement with literature values confirm this observation.However, when the scatter signal approaches or surpasses the fluorescence signal, small errors in the scatter correction system can give rise to large errors in scatter-corrected results. This is so for Co and Ni, two moderately sensitive elements originally included in this study. Table 3 lists Co and Ni values obtained on three reference materials with and without the use of scatter correction and compares them with the literature values. Although uncorrected values for Co and Ni are consistently high, as expected, the values obtained by using scatter correction are either over or under compensated. Therefore, neither Co or Ni could be adequately determined under the conditions of this study.Other techniques and procedures to eliminate or reduce scatter problems were not tried during this study. Haarsma et al. 17 have reported large decreases in scatter for Cd and Zn when HC104 is added to scatter-producing samples. Subse- quent to this study, it was learned that Demersl* has observed the almost complete elimination of scatter signals from 0.5% Ca and Mg on Ni and Co in the plasma when the samples are spiked with 10% or more of HC104. It is believed that in the presence of HC104 the salt particles burst into particles too small to cause scatter.19 Hydrofluoric acid and ammonium bifluoride are often effective also, especially for silicon, l8 because they produce volatile fluoride salts of many elements. Solvent extraction techniques may also be useful in atomic fluorescence spectrometry to isolate the analytes of interest from scatter-producing matrices.w P m Table 4.Mean concentrations (n = 6) of Fe, Mn, Cu, Cd, Pb and Zn in geological reference samples and comparison with literature values. All values are in pg g-1 except for those for Fe, which are O h Element Fe Mn Pb Zn c u Cd This work Literature This work Literature 1345 f 7 1300*, 1175t, 2.8 f 0.1 3.0t 72.0 f 1.0 69*, 68t, 3.5 _+ 0.05 4.0t 9.4 f 0.2 10.8*, l l t , <0.3 0.4t 7090 f 71 6500*, 6850t, t 0 . 3 0.35t 342 f 6 360*, 3501, <0.3 0.15t 64.9 f 0.3 105*, 66t, <0.3 0.15t 1129$, lo000 63.4$, 749 14.2$, 95 6780$ , 68604 308$, 3609 63.9$, 720 19.2 f 0.3 21.8 f 2.07 t 0 . 3 0.088 f 0.0227 Sample Material This work GXR-1 .. Jasperoid 25.9 f 0.6 Literature 24.7* This work Literature 1030 k 30 1 OOO** This work 732 f 32 Literature 670*, 930t, 82% 615*, 730t, 725 § 15*, 14t, 209 46t, 54§ This work 740 k 10 447 k 12 198 5 7 72.0 k 3.0 49.8 ? 2.0 121 k 4 72.2 f 1.5 46.1 f 0.4 44.6 f 1.6 94.7 f 6.2 233 f 13 120 f 3 211 f 4 43.6 k 5.1 98.2 k 1.3 170 ? 5 Literature 740*, 700t, 6750 500*,445t, 470§ 220*, 180t, 2040 64* , 69t, 76§ 50*, 507, 5l§ 120*, 92t, 1009 79 k 101 GXR-2. . Soil 1.92 f 0.08 1.90* 1 040 f 20 960* 731 f 30 19.5 k 1.4 63.0 k 3.7 GXR-3 . . FeandMn 20.5 k 1.6 GXR-4 . . Cu mill head 3.50 k 0.17 rich 18.6* 2.97* 3.19* 5.58* 27 950 k 950 22 300* 155 f 40 150* 370 k 30 280’ 1 OOO* 1 135 k 30 GXR-5 . . Soil 3.71 k 0.13 20.2 f 3.1 22*, 16t, 16§ 110*, 95f, 1000 24.4 k 4.77 GXR-6.. Soil 6.21 k 0.24 124 ? 5 25 f 2.2 28.5 k 0.3 GSD-1 . . GSD-2 . . GSD-3 . . GSD-4 . . GSD-5 . . GSD-6 . . GSD-7 . . GSD-8 . . BCSS , . MESS . . Stream Stream Stream Pond Pond Stream Stream Stream Marine Marine sediment sediment sediment sediment sediment sediment sediment sediment sediment sedimen t 4.90 k 0.14 5.157 1 076 2 56 920 L 607 1.27 f 0.02 4.51 f 0.15 4.05 k 0.21 1.237 4.567 4.137 233 k 14 451 f 19 857 k 31 240 k 307 400 k 357 825 k 507 4.3 k 0.9 4.9 f 77 t 0 . 3 0.065 k 0.0167 32 k 87 44 f 77 170 k 13 177 f 117 <0.3 0.1 f 0.0237 46.8 k 3.3 40 f 4.47 52 k 67 34.8 f 1.6 37.3 f 3.21 <0.3 0.19 f 0.003y[ 30.4 f 6.97 101 k 157 38.5 k 5.3 139 f 10 18.5 k 1.4 r 3.97 f 0.08 4.107 1 246 k 42 1 096 & 66 784 k 38 338 f 30 1 160 k 607 970 f 607 690 5 507 335 f 257 128 f 1.3 137 f 107 0.70 _+ 0.1 0.82 _+ 0.071 112 k 13$ 243 f 237 3.99 k 0.20 4.38 k 0.15 1 S O f 0.07 4.127 4.547 1.547 363 k 6 383 f 187 <0.3 0.43 +_ 0.047 27 f 57 144 k 107 33.5 k 0.3 38 f 27 1.0 k 0.1 1.05 f 0.081 392 L 11 350 k 26 238 f 191 4.1 f 0.2 4.1 f 0.71 <0.3 0.081 k 0.0177 15.8 f 0.6 18.5 f 2.711 4 .3 0.25 f 0.0411 24.4 f 1.1 25.1 2 3.811 0.63 k 0.08 0.59 f 0.1011 43 f 47 24.6 f 2.5 20.9 f 2.6 30.6 k 3 21 k 47 3.30 k 0.13 3.2911 3.0511 252 f 24 523 f 38 229 f 1511 513 _+ 251) 22.7 f 3.411 119 -t 1211 191 k 1711 3.01 k 0.10 34.0 + 6.111 * Gladney et a1.20 t Viets.21 $ Sanzolone et al.’ Q Sanzolone and Chao.22 7 Xie et a1.7 11 Berman.8 * * AA determination. this work. c 0 rJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 347 Analytical Data Accuracy The proposed method was applied to 16 geological reference materials. The results are well within the accuracy require- ments of geochemical exploration as shown by the good agreement with the literature reference values listed in Table 4. Precision Six replicates of the 16 geological reference materials were digested and analysed by the proposed method to establish precision. The relative standard deviations are less than 5% except in a few instances where the limit of determination is approached (Table 4). Conclusion A simultaneous, multi-element , atomic fluorescence method is described for the determination of six elements over a wide range of concentrations in a variety of geological materials. Precision and accuracy of data obtained on geological reference materials by this technique demonstrate that it is a viable alternative to atomic absorption and inductively coupled plasma atomic emission spectrometric techniques.The method is a rapid, simultaneous, multi-element proce- dure that has a greater linear range than AAS, and has a more stable base line and greater freedom from spectral interfer- ences than ICP-AES. References 1. 2. 3. Church, S. E., Geostand. Newsl., 1981, 2 , 133. Ullman, A. H., Prog. Andy. At. Spectrosc., 1980, 3, 87. Demers, D. R., and Allemand, C. D., Anal. Chem., 1981,53, 1915. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. Demers, D. R., Busch, D. A., and Allemand, C. D., Am. Lab., 1982, 14, No. 3, 167. Meier, A. L., and Bigelow, R. C., US Geol. Surv. OFR, 1984, 84-698. Alcott, G. H., and Lakin, H. W., in Elliott, I. L., and Fletcher, W. K., Editors, “Geochemical Exploration, Proceedings of the Fifth International Geochemical Symposium, Vancouver, BC, Canada, April 1-4 1974,” 1975, p. 659. Xie, X., Yan, M., Li, L., and Shen, H., Geostand. Newsl., 1985,9, 83. Berman, S., “Marine Sediment Reference Materials,” Division of Chemistry National Research Council, Ottawa, Canada, 1981. Sanzolone, R. F., Chao, T. T., and Crenshaw, G. L., Anal. Chim. Acta, 1979, 105, 247. Sanzolone, R. F., and Meier, A. L., Analyst, 1986, 111, 645. Demers, D. R., Spectrochim. Acta, Part B, 1985,40, 93. Long, G. L., and Winefordner, J. D., Appl. Spectrosc. 1984, 38, 563. Omenetto, N., Crabi, G., Nesti, A., Cavalli, P., and Rossi, G., Spectrochim. Acta, Part B, 1983,38, 549. Demers, D. R., “AFS/2000 Software Revisions,” Baird Spec- trochemical Products Division, Bedford, MA, 1985. Larkin, P. L., and Willis, J. B., Spectrochim. Acta, Part B, 1974, 29, 319. Naranjit, D. A., Radziuk, B. H., Rylaarsdam, J. C., Larkins, P. L., and VanLoon, J. C., Appl. Spectrosc., 1985,39, 128. Haarsma, J. P. S., Vlogtman, J., and Agterdenbos, J., Spectrochim. Acta, Part B, 1976,31, 129. Demers, D. R., personal communication, 1985. Herromann, R., and Alkemade, C. Th. V., Editors, “Chemical Analysis by Flame Photometry,” Interscience, New York, 1963, p. 304. Gladney, E. S., Perrin, D. R., Owens, J. W., and Knob, D., Anal. Chem., 1979,51, 1557. Viets, J. G., Anal. Chem., 1978, 50, 1097. Sanzolone, R. F., and Chao, T. T., Anal. Chim. Acta, 1976,86, 163. Paper J6l13 Received March 4th, 1986 Accepted May 15th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100343

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 349 Improved Determination of Cadmium in Blood by Flame Atomic Fluorescence Spectrometry Edet. J. Ekanem,* Charles L. R. Barnardt and John. M. Ottaway Department of Pure and Applied Chemistry, University of Strathclyde, 295 Cathedral Street, Glasgow GI IXL, UK Gordon S. Fell Department of Clinical Biochemistry, Royal Infirmary, Glasgow G4 OSF, UK A procedure is described for the determination of normal levels of cadmium in whole blood using flame atomic fluorescence spectrometry. This method represents an advance on previous work in that better control of the electrodeless discharge lamp source temperature has facilitated greater stability and thus better detection limits. Furthermore, sample pre-treatment by deproteinisation reduces the dilution factor from 1 + 4 to 1 + 1 so that the whole blood detection limit is improved to twice that of aqueous solutions. Keywords: Flame atomic fluorescence spectrometry; electrodeless discharge lamps; blood cadmium determination; protein precipitation The determination of cadmium in human blood has presented two main problems to the analytical chemist. Firstly, the normal levels, which have yet to be agreed upon, are at or close to the limits of detection of most atomic spectrometric techniques.This places an onerous task on the analyst to provide both accuracy and precision for such samples. Secondly, and more important, the matrix is complex and therefore produces difficult interference problems. At present the most widely applied method for the determination of cadmium in blood involves the use of carbon furnace atomic absorption spectrometry (CFAAS), coupled with a sample pre-treatment. Among the many pre-treatments employed are wet ashingl with acids, low-temperature plasma ashing2 and protein precipitation .3.4 In 1979 the development, in this laboratory, of a flame atomic fluorescence spectrometric (FAFS) method for the determination of cadmium in blood and urine was described by Michel et al.5 In this procedure blood samples were haemolysed with Triton X-100, acidified to 0.04 M with hydrochloric acid, mixed and centrifuged.The supernatant liquor was then aspirated into a nitrogen-separated air - acetylene flame supported on a capillary burner6 using a Perkin-Elmer nebuliser and spray chamber. Aqueous cad- mium calibration standards were also acidified to 0.04 M HCl. Before interpolation on the calibration graph, sample results were adjusted by an uptake rate correction factor to compen- sate for viscosity differences between the aqueous standards and blood samples.Blood detection limits of 1.4 pg 1-1 were reported for a 1-s count time. Although this level of sensitivity is adequate for the determination of raised blood cadmium levels in cases of proved intoxication, it is inadequate for determining normal levels of cadmium in blood. A reasonable estimate of the normal levels of cadmium in blood is provided by the lower end of concentration ranges that have been determined for reference populations of unexposed persons. Such a range of 1.1-6.4 pg 1-1 with a mean of 3.1 (k1.5) yg 1-1, obtained by FAFS, has been reported from this laboratory.5 A range of 0.3-7.9 pg 1-1 with a mean of 2 pg 1-1 based on a similar reference population and a range 0.3-6.0 yg 1-1 with a mean of 1.3 yg 1-1 based on a non-smoking population have also been reported using CFAAS .7 Sub-pg 1- 1 blood cadmium levels have frequently * Present address: School of Basic Studies, Ahmadu Bello Univer- i Present address: Department of Chemistry, Glasgow College of sity, Zaria, Nigeria.Technology. Cowcaddens Road, Glasgow, UK. been reported.Sl0 An FAFS procedure with a much lower detection limit than the previously reported 1.4 pg 1-1 is required for the accurate determination of low or normal blood levels. Modification of the blood matrix by protein precipitation has been applied to the determination of cadmium in blood by CFAAS, giving good accuracy for a range of precipitants.11J2 Aim The limit of determination of the previous procedures was hampered by a high aqueous detection limit (0.2 yg l-l), relative to the lower values in the working range, and the need for a five-fold dilution of blood samples, which further raised detection limits.These investigations were carried out in order to overcome these problems by reducing the need for sample dilution by using protein precipitation, and by decreasing the aqueous detection limits by using better temperature control of the source. Experimental Instrumentation The instrumentation used in this work, (summarised in Table 1) has been described in detail previously.5J3314 Radiation from a thermostated microwave-excited cadmium electrode- less discharge lamp (EDL) source ,15716 mechanically modu- - Table 1.Instrument operating conditions Light sources: (1)CdEDL . Broida 21OL cavity (2)Eimac . . . . Modulation. . . . Air - acetylene . . Type. . . . . . . Grating . . . . Wavelength . . Bandpass . . . . Type . . . . . . Photomultiplier type EHTvoltage . . Dutycycle . . . . Measurement period Flame: Monochromator: Photon counter: . . . . . . . . 534 "C (air stream temperature); Microwave power 60 W Operating current varied to match 300 Hz (both sources) corresponds to 180 "C in quartz EDL EDL scatter characteristics Slightly fuel rich, nitrogen-sheathed . . Spex Doublematefl4 . . . . 228.8nm . . 1.0nm Blazed at 300 nm 1200 grooves mm-' .. Ortec Brookdeal 5C1 . . EMI9789QB . . 115OV . . 55% . . I s350 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 lated at 300 Hz, is used to excite cadmium atomic fluores- cence. Samples are aspirated into a nitrogen-separated air - acetylene flame supported on a Perkin-Elmer nebuliser - spray chamber fitted with a capillary burner head. Fluorescence signals were monitored using a double monochromator photon counting system. An intensity- modulated (300 Hz) co-focused xenon-arc (Varian-EMI) continuum source was used for simultaneous background and scatter correction. Signals from the two sources were phase- locked using the synchronous sampling (5C21) module of the photon counter (Brookdeal). Thermostated Environment of EDLs It has been established that the best radiant performance for the cadmium EDL is obtained if the lamp is placed in an environment of appropriate, constant, high temperature.In the p-evious work5 the EDL temperature was monitored with a thermometer placed in the neck of the quartz lamp. A stream of hot air was applied to the outside surface of the lamp, the temperature of which was controlled by a Variac power supply. This approach to thermostating the EDL suffers from two main drawbacks. Firstly, there is a long time lag between any change in EDL temperature and the response of the thermometer, so that short term temperature fluctu- ations would not be noticed. Secondly, the heating power supplied to the air stream was not controlled by changes of the lamp temperature.The radiant output from an EDL is highly responsive to even minor fluctuations in temperature. Such fluctuations result in spectroscopic noise and ultimately in poor detection limits. For these reasons a new type of heater controller was constructed in which the power to the heater was governed by a voltage feedback from a thermocouple placed in the air stream immediately adjacent to the cavity. The placing of a thermocouple probe inside the cavity (closer to the EDL) was precluded because it destabilised the microwave field. This feedback-controlled heater gave a significant improvement in the control of EDL temperature, to within 0.5% at 500°C. Collection and Storage of Blood Blood samples were collected by venepuncture and stored in plastic sample tubes containing anticoagulant (potassium EDTA or lithium heparin).These modes of sample collec- tion17J8 and storage5 have previously been shown to be free from cadmium contamination. The polypropylene centrifuge tubes (Henleys Medical Supplies, London) in which protein precipitation was performed19 were screened for cadmium contamination by shaking the same aliquot of 2 M nitric acid consecutively in a large number of fresh tubes and observing the cadmium atomic fluorescence generated from this wash solution. Blank signal levels were always obtained. Containers and pipettes, used for preparing standards, were soaked overnight in 50% nitric acid, washed out and rinsed with de-ionised water before use. Where necessary, blood samples were stored temporarily at 4°C or for longer periods at -22 "C.Reagents Only high-purity (AnalaR) reagents were used and contami- nation checks were made by measuring reagent blanks for cadmium atomic fluorescence. All reagent solutions were prepared in de-ionised water. Aqueous calibration standards were prepared by serial dilution from a 1000 pg 1-1 stock solution. All standards were matrix matched by addition of appropriate amounts of the precipitant. Procedure for Protein Precipitation The protein precipitants considered were hydrochloric acid, chloroacetic acid (CAA) , trichloracetic acid (TCA) and nitric acid. All of these reagents have previously been used as protein precipitants .11912J9 The method used for each was essentially the same. A 2-ml aliquot of blood was dispensed into an equal volume of precipitant solution in a poly- propylene centrifuge tube.The resulting slurry was centri- fuged for 30 s at 3000 rev min-1 and the protein-free supernatant aspirated into the flame. A reagent blank was measured in each instance by replacing the blood with a 2-ml aliquot of de-ionised water. The concentration of each precipitant was optimised to yield the highest cadmium AFS signal from 2 ml of the same blood. The procedure for each was optimised using out-dated blood bank samples. The criteria used to identify the most appropriate reagent were the Cd fluorescence signal, supernatant volume and uptake rate. This procedure was compared with the Triton X-100 - HC1 procedures in terms of the detection limit and the signal obtained from 2 ml of the same blood when treated by each procedure.Analytical Cadmium Recoveries The recovery of cadmium from whole blood following protein precipitation was assessed by adding aliquots of inorganic cadmium standards to a 2-ml blood sample before deprotein- isation. The percentage recovery of total cadmium was acceptable for the range 0-10 pg 1-1 of added cadmium. Precision An assessment of the analytical precision of the method was obtained using a set of blood samples of known cadmium concentration that spanned the working range. Each of ten 2-ml aliquots of the blood samples were deproteinised, centrifuged and analysed, and the over-all precision of the technique was calculated from the signals obtained. Accuracy The accuracy of the results obtained by protein precipitation followed by FAFS was assessed by direct comparison of a set of samples previously analysed by CFAAS in another labora- tory.The accuracy of the FAFS procedure was also tested by the analysis of quality control blood samples from the Supra-Regional Assay Service (organised by Surrey Univer- sity) that had been analysed independently by many labora- tories. Interferences The experimental observations of Michel et al. 5 were con- firmed by the present investigation when nitric acid was used as the protein precipitant. Results and Discussion The concentrations of precipitants yielding the most efficient release of cadmium from the blood were optimised in terms of signal, uptake rate and supernatant volume.19 These con- ditions are summarised in Table 2.The minimum dilution of blood gave a maximum signal, although this resulted in a highly viscous supernatant for both the chloroacetic acids. The variation of cadmium FAFS signal with increasing nitric acid precipitant concentration is shown in Table 3. The optimum release of cadmium from blood is achieved with 2 M nitric acid, which gave the best over-all performance in terms of the highest signal (Tables 3 and 4) with acceptable supernatant volume (Table 5) and an uptake rate equal to that of theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 351 Table 2. Optimisation of reagent concentration Table 6. Sensitivities of procedures for 2-ml aliquots of same samples Concentration Optimum Signal-blank/ Reagent range observed concentration counts Triton X-100 - HCI .. As published5 As published5 106 Nitricacid . . . . 1 . G 4 . 0 ~ 2.0 M 340 C A A . . . . . . 1.0-4.0% 1.5% 101 TCA . . . . . . 1.0-4.0% 1 .O% 94 HCI . . . . . . 1.0-4.0 M 2.0 M 119 Table 3. Optimisation of nitric acid concentration HN03 Signal-blank/ concentrationh counts s-* 1 .o 1.5 2.0 2.5 3.0 4.0 274 289 340 3 30 337 339 HN03 signal Triton X-100 Signal-blankkounts s-1 Sample No. Triton X-100 HN03 signal 1 106 340 3.2 2 47 110 2.3 3 164 538 3.3 Table 7. Sample uptake rates Sample type De-ionised water . . . . . . . . . . . . 9.0 1 M nitric acid . . . . . . . . . . . . . . 9.0 2 M HN03 blood deproteinate (1 + 1 dilution) 9.0 Triton X-100 - HC1 blood deproteinate (1 + 4 dilution) . . . . . . . . . . . . 6.8 Triton X-100 - HC1 blood deproteinate (1 + 1 dilution) .. . . . . . . . . . . 5.0 . . Table 4. Comparison of reagent sensitivities Table 8. Typical calibration results Signal-blank/ Sensitivity relative Reagent counts s-l to Triton X-100 TritonX-100 - HCI . . 47 ~ M H N O ~ . . . . . . 110 1.5'IoCAA . . . . . . 56 1.0% TCA . . . . . . 50 2.0M HCI . . . . . . 69 1.3 2.3 1.2 1.1 1.5 Table 5. Recovery of liquid fraction after deproteinisation Supernatant Wet residue Reagent volume/ml volume/ml TritonX-100 - HCI . . 8.0 2.0 2.0 M HNO3 . . . . . . 2.0 2.0 1.5%CAA . . . . . . 1.6 2.4 1.0% TCA . . . . . . 1.6 2.4 aqueous standards (Table 7). It is clear from these results that, even when an uptake rate correction (the factor is the ratio of the aqueous uptake rate to the supernatant uptake rate) is applied to each precipitant solution, as appropriate, nitric acid deproteinisation provides the most efficient release of cad- mium.Both the protein precipitation and Triton X-100 procedures yield two fractions after centrifugation. There is a wet, solid residue of cell protein and a supernatant liquor. In order to perform accurate FAFS analysis for blood cadmium it is necessary that at least 2 ml of supernatant be produced. Attempts to use hydrochloric acid proved unsuccessful as it produced effervescence and foaming when mixed with blood samples. Also, as it yielded inferior FAFS signals and variable supernatant volumes for the same sample, hydrochloric acid was considered unsuitable for protein precipitation. The results in Table 5 illustrate the breakdown of a typical sample when treated with each of the precipitants.In each instance a 2-ml aliquot of sample was treated and converted into supernatant and residue using an equal volume of precipitant, although the Triton X-100 procedure used 1 + 4 volume dilution. Apart from the relatively poor signals obtained with the chloroacetic acids, their supernatant liquids were so viscous that they blocked the capillary burner when aspirated into the flame. For these reasons these acids were not investigated further. This left only nitric acid and the Triton X-100 procedure as suitable methods for blood cadmium determi- nation. The relative sensitivities of the two procedures are compared in Table 6, from which it is clear that the use of nitric acid has distinct advantages over the Triton X-100 procedure.In view of the wide disparity between the dilution Average uptake ratelm1 min-l Parameter Aqueous Standard calibration additions Fittedline . . . . . . y = 141x+ 34 y = 141x+205 Correlation coefficient Blood detection limit Aqueous detection limit ( r ) . . . . . . . . 0.9995 0.9982 (20) . . . . . . . . 0.02 pg 1- ' (20) . . . . . . . . 0.01 pg I - ' factors used by each procedure the two reagents were compared at similar dilutions. It was found that the greater the dilution factor, the closer the results for the two procedures become. The nitric acid deproteinisation, however, yields superior sensitivity at all sample dilutions. The detection limits obtained for both procedures at the various dilutions also follow this trend.Even at the same sample dilution, signal levels obtained with the Triton X-100 procedure are about 35% lower than the corresponding signals obtained following the nitric acid deproteinisation. The liquid fraction resulting from the Triton X-100 treatment is more viscous and yields lower uptake rates than that obtained from nitric acid deproteinisation. When aqueous calibration standards are used for FAFS blood analysis it is essential to compensate for differences in uptake rates between samples and standards. The observed fluctuations in uptake rate associated with variable haemoglobin levels hampered the Triton X-100 procedures in that it necessitated the use of a variable uptake-rate correction factor, which could be sample depen- dent. In this respect the nitric acid deproteinisation procedure has an important advantage as no uptake rate correction is required, as can be seen from Table 7. This, together with the improved sensitivity of the 2 M nitric acid deproteinisation, make it a more suitable matrix modification procedure for the determination of blood cadmium levels by FAFS.Evaluation of the Use of Nitric Acid Deproteinisation for the Determination of Blood Cadmium Levels by FAFS The results obtained by direct analysis using aqueous calibra- tion standards showed good agreement with those obtained by standard additions. A typical example is presented in Table 8, which indicates close agreement between the slopes of aqueous and standard additions calibration graphs. The blood detection limit is expressed as twice the standard deviation of the aqueous signal multiplied by the dilution factor (f).352 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 Recoveries The efficiency of the protein precipitation procedure was assessed in terms of recovery of cadmium following addition of inorganic cadmium to blood samples prior to protein precipi- tation. Despite the difference in nature between the added cadmium and that present in blood, as a protein complex, the results shown in Table 9 indicate efficient release of total cadmium is effected by this procedure, as all the values are close to 100%. Analytical Precision The analytical precisions shown in Table 10 were obtained for the method as a whole, using normal blood samples containing less than 10 pg 1-1 of cadmium.Instrumental precision was measured using aqueous standards. The between-batch precision was assessed by measuring the same blood sample twice a week for 6 weeks. The detection limits are improved considerably (Table 11) by the modified procedure compared with our earlier method. Average values of 0.01 and 0.02 pg 1-1 have been reproducibly obtained for aqueous and whole blood samples, respectively. This rep- resents an almost two orders of magnitude improvement in the blood cadmium detection limit. This improvement has been achieved principally by a more precise EDL temperature control system. This has been further enhanced by the reduced sample dilution allowed by the nitric acid deproteinisation procedure. Good accuracy is indicated in Table 12, which represents a comparison of the FAFS protein precipitation procedure, for 16 samples, with a CFAAS procedure carried out in a different laboratory.The samples numbered 11-16 were provided by the Supra-Regional Assay Service (SAS) quality control service, operated by the Heavy Metals Laboratory at the University of Surrey. The accuracy of the procedure was confirmed by the close correlation (Table 13) between the inter-laboratory means and the FAFS results. Conclusions Nitric acid deproteinisation offers a simple means of matrix modification with minimum sample dilution, by which normal levels of cadmium in whole blood can be accurately deter- mined by FAFS. This procedure is in fact being adapted in this laboratory for the determination of the lower levels of cadmium in plasma and serum.Although FAFS systems of apposite sensitivity are not commercially available, the technique provides a rapid and convenient method for Table 9. Analytical recoveries of cadmium following protein precipitation of whole blood Cadmium concentrationlpg I-* Added Recovered Recovery, O/O 0 1.16 - 2 3.14 99 4 5.56 110 6 7.67 108 8 9.80 108 10 11.40 102 cadmium determinations in both blood and urine as both normal and elevated levels can now be determined with good precision in both matrices. It is of particular value for the rapid screening of samples for Table 11. Detection limits for the determination of cadmium in blood Blood detection Dilution limit/ Reagent Reference factor vgl-' Triton-X100 - HCI . . 5 5 1.14 2.0 M nitric acid .. . . This work 5 0.05 2.0~nitricacid . . . . Thiswork 2 0.02 Table 12. Comparison of FAFS with CFAAS Cadmium concentration/ Pi3 1- Sample No. FAFS CFAAS 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 15.4 8.0 10.0 28.1 0.8 7.6 9.9 9.2 9.6 6.6 24.2 14.9 3.6 31.4 11.2 18.6 15.1 7.8 9.3 27.9 0.7 6.1 8.0 7.5 9.0 4.8 23.8 15.7 3.3 27.3 8.0 16.2 Regression results: CFAAS results is The regression equation for the correlation between FAFS and FAFS = 1.02CFAAS -+ 0.91 The standard deviation of scatter20 of the FAFS results about this line is 1.32 pg I-'. Table 13. Inter-laboratory comparison of FAFS results Cadmium concentration/ Clg1-l Sample No. FAFS SAS means 11 24.2 24.9 12 14.9 17.6 13 3.6 2.9 14 31.4 32.4 15 11.2 10.5 16 18.6 18.1 Regression results: and the Supra-Regional Assay Service (SAS) is The standard deviation of scatter20 of the FAFS results about this line is 1.28 pg 1-1.The regression equation for the correlation between FAFS results FAFS 1.02SAS + 0.00 Table 10. Measurement precisions Precision (RSD), Yo Cadmium level/ Sample Pg I-' Total Instrumental Between-batch - - Blood (1) . . . . . . 1.8 6.6 Blood (2) 7.8 1.9 - Aqueous(1) . . . . . . 5.0 - 3.1 - Aqueous(2) . . . . . . 10.0 - 1.8 - . . . . . . 7.4JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 353 which results are urgently needed and for population surveys, as about 300 samples can be analysed daily. The blood detection limit of 0.02 pg 1-I is ideally suited to the determination of cadmium at the sub-pg 1-1 level. The advantages of the FAFS method with nitric acid deproteinisation include the following: ( i ) lower blood levels can be determined more precisely; (ii) analytical sample solutions have the same uptake rate as calibration standards, hence no uptake rate correction factor is required; (iii) the scatter of incident radiation by sample particles in the flame is negligible; background correction for scatter is therefore virtually unnecessary, but has still been used in this work to take account of any gross variations in matrix from sample to sample; (iv) only one reagent is involved, reducing the risk of con tamination.The authors thank the Scottish Home and Health Department for the award of a postdoctoral fellowship (to C. L. R. B.). They also acknowledge the continued support of the Eastern District, Greater Glasgow Health Board.The thanks are extended to Messrs. M. Porter and G. Brown of the Chemistry Workshop, Strathclyde University, for the construction of the heater and also to Mr. G. Henderson of the Biochemistry Workshop, Glasgow Royal Infirmary, for the design and construction of the heater controller. References Gorusch, T. T., “The Destruction of Organic Matter,” Pergamon Press, Oxford, 1970. Carter, G. F., and Yeoman, W. B., Analyst, 1980, 105, 295. Stoeppler, M., Brandt, K., and Rains, T. C., Analyst, 1978, 103, 714. Stoeppler, M., and Brandt, K . , Fresenius 2. Anal. Chem., 1980, 300, 372. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Michel, R. G . , Hall, M. L., Ottaway, J. M., and Fell, G. S . , Analyst, 1979, 104, 491. Aldous, K. M., Browner, R. F., Dagnall, R. M., and West, T. S., Anal. Chem., 1970,42, 939. Fell, G. S . , Chem. Brit., 1980, 16, 323. Friberg, L., Piscator, M., Nordberg, G., and Kjellstrom, T., “Cadmium in the Environment,” Second Edition, CRC Press, Cleveland, OH, 1974. Sharma, R. P., MacKenzie, J. M., and Kjellstrom, T., J . Anal. Toxicol., 1982, 6, 135. Subramanian, K. S., and Meranger, J. C., Clin. Chem., 1981, 27, 1866. Einarsson, O., and Lindstedt, S., Scand. 1. Clin. Lab. Invest., 1969, 23, 367. Baily, P., and Kilroe-Smith, T. A., Anal. Chim. Acta, 1975,77, 29. Michel, R. G., Sneddon, J., Hunter, J. K., Ottaway, J. M., and Fell, G. S . , Analyst, 1981, 106, 288. Sthapit, P. R., Ottaway, J. M., and Fell, G. S . , Analyst, 1983, 108, 235. Michel, R. G., Coleman, J., and Winefordner, J. D., Spectro- chim. Acta, Part B, 1978, 33, 195. Michel, R. G., Ottaway, J. M., Sneddon, J., and Fell, G. S., Analyst, 1978, 103, 1204. Fell, G. S . , Ottaway, J. M., Hussein, F. E. R., Michel, R. G . , and Hall, M. L., in Brown, S. S., Editor, “Clinical Chemistry and Chemical Toxicology of Metals,” Elsevier North-Holland, Amsterdam, 1977, p. 367. Fell, G. S., Ottaway, J. M., and Hussein, F. E. R., Br. J. Znd. Med., 1977,34, 106. Ekanem, E. J . , Barnard, C. L. R., Ottaway, J. M., and Fell, G. S., Talanta, 1986,33, 55. Davies, 0. L., “Statistical Methods in Research and Produc- tion,” Third Edition, Oliver and Boyd, London, 1967, p. 150. Paper 5619 Received February 24th, 1986 Accepted May 6th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100349

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 355 Behavior of Zeeman Corrected Atomic Fluorescence at High Source Currents Guo Tie-Zheng and Roger Stephens* Department of Chemistry, Dalhousie University, Halifax, NS B3H 4J3, Canada Magnetic fields are generated by a pulsed air-cored solenoid, a device which allows the field-on-source and field-on-atomiser configurations to be readily compared. The fluorescence intensity is observed to increase with increasing field strength for both configurations at high lamp currents. However, the increase is markedly greater for the field-on-source arrangement. The effect is explained in terms of distortion of the source emission profile by the combined effects of Zeeman splitting and self-absorption. Zeeman absorption scanning is used to examine this process.Keywords: Zeeman effect; atomic fluorescence; self-absorption; scatter correction Van Loon and co-workers have shown that use of the Zeeman effect can provide effective scatter correction for atomic fluorescence measurements. 1 In principle the technique is similar to the Zeeman background correction method, which is now quite widely used in atomic absorption spectrometry.2 Thus in both configurations the application of a magnetic field causes a frequency displacement of the o components of the associated Zeeman multiplet. The o displacement alters the overlap between the atomic line profiles of source and absorber, thereby changing the corresponding absorption or fluorescence signals. The magnitude of such changes provides a selective measure of the atomic concentration of interest.Despite their apparent similarity, the instrumental require- ments of Zeeman corrected absorption and fluorescence apparatus are markedly different. Polarisation selection of the n and o components is a less attractive option for fluorescence than for absorption because of light loss in the polariser, which can be serious for systems of large aperture, especially at low wavelengths, and because particle scatter may be polarised anyway, in which case a true correction is not necessarily attained. The latter problem does not arise when polarisation selection is used in absorption systems, as the distribution of scattered energy is symmetrical about the optical axis, and shows no polarisation dependence in toto. This symmetry is lost in the ordinary fluorescence configuration as only a single direction is used for detection.The problem is analogous to that caused by Schlieren effects in polarised Zeeman corrected absorption apparatus,3 and can be solved in a similar manner if necessary. The alternative to polarisation selection is a.c. modulation of the magnetic field. The method has been described in detail for absorption systems by de Loos-Vollebregt and co- workers,4-6 and for fluorescence measurements by van Loon and co-workers.1 Some observations on the use of the latter application are described in the present work. Particular attention is given to the behavior of the field-on-source configuration when the source is driven at high current levels. Experimental All measurements were made on a Varian Techtron AA5 spectrometer with a 1 cm diameter air - acetylene burner head mounted in the standard nebuliser assembly.Longitudinal magnetic fields were generated by a pulsed air-cored solenoid. The construction of this device and its application to absorption measurements have been described7; the design remained unchanged save for the power supply, which was * To whom correspondence should be addressed. modified for 220-V operation, thereby allowing field strengths up to 1.1 tesla to be obtained. Field-on-source measurements were made by inserting the appropriate hollow-cathode lamp through the coil former. Field-on-atomiser measurements were made by splitting the coil into two halves. These functioned as a pair of Helmholtz coils, providing sufficient room between them to accommodate the flame.The direction of the magnetic field axis remained the same in the two configurations: i.e., the optical axis of the lamp - atomiser system remained parallel to the magnetic field axis and perpendicular to the direction of observation irrespective of the location of the magnet (Fig. 1). The air-cored solenoid must be driven by high current pulses at a low duty cycle in order to obtain the necessary field strengths without excessive power dissipation. This in turn requires that the lamp be synchronously pulsed to maintain an adequate photon count. Such an arrangement is slightly artificial in an absorption system, but it becomes a natural part of an atomic fluorescence measurement as pulsing the source simultaneously improves the source to background emission ratio of the apparatus.The open construction of an air-cored coil is also helpful, allowing three optical axes to be retained and thereby avoiding the optical constraints encountered with an iron-cored magnet. 1 A Burner B head - 5 cm I \ a \ I i i Monochromator, detector Fig. 1. Optical arrangement. A, Coils joined for field-on-source configuration; asterisk marks the position of the back of the cathode. B, Coils separated for field-on-atomiser configuration. The inside surfaces of the coils were faced with 1/16 aluminium as a heat reflectorJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 356 8 > fn a C a e .- w .- 2 a .- c - a cc A . . 0 1 .o Field strengthn Fig.2. 300; B, 200; C, 100; and D, 50 mA Relative source intensity versus field strength for Mg: A, 10 7.5 fn C w .- 5.0 I e c .- - m .- k 2.5 v) 0 2.5 Magnetic fieldn Fig. 3. Fluorescence signals versus magnetic field strength for Mg, field-on-source confi uration. Lamp currents pulsed at: A, 400; B, 300; C , 190; D, 121; and E, 40 mA, respectively. Curves lying above/below the axis are phase shifted by 180” > c .- 15 a C a C a fn c .- z 3 a - c .- c - a a 0 Field-on- source Field-on- 400 Lamp current/mA Fig. 4. Relative fluorescence intensities for the field-on-source and field-on-atomiser configurations. Values are corrected for the intensity change of the field-on-source arrangement The method of operation in both field-on-source and field-on-atomiser configurations was the same: the required solution was aspirated into the flame; the magnet was pulsed over isolated half cycles of the a.c.line voltage; the source was synchronously pulsed at the start, at peak field strength and at the end of each current pulse. Thus the two outer pulses correspond to a conventional fluorescence measurement (zero magnetic field) while the centre pulse measures the fluorescence when either the hollow-cathode source or the flame absorption profile shows longitudinal Zeeman splitting. Thus in either configuration the interaction between the source emission and flame absorption profiles involves only the C J ~ components of the multiplet. Under these circumstances a genuine atomic fluorescence signal varies in magnitude when the zero and peak field signals are compared; a scatter signal does not.The signal processing circuits compared an integrated average of the two outer pulses with that of the centre pulse to give a scatter corrected output signal. Details have been given previously .7 Results Signal Dependence on Magnetic Field Strength The equivalence of the Zeeman splitting observed in emission and absorption suggests that the field-on-source and field-on- atomiser configurations should produce comparable results, and indicates that the choice between the two is simply one of technical convenience. This supposition was found to be correct at low lamp currents. However, at high lamp currents very different behavior was observed. A phase reversal of the signal occurred, such that the fluorescent intensity increased instead of decreased with increasing magnetic field strength.At the same time the lamp intensity (field-on-source) also increased with field strength, by a factor dependent on the actual lamp current. Partly as a result of this intensity increase the signal to noise ratio of the field-on-source configuration became markedly greater than that of the field-on-atomiser arrangement. This behavior was observed for every element examined. Experimental results for Mg are shown in Figs. 2-4. Some detection limits obtained in this “inverted” mode of operation (ie., field-on-source, maximum attainable lamp current) are given in Table 1. In all instances similar results were obtained to those discussed in detail for Mg provided that appropriately high current pulses were supplied (Table 1, column 3).The ability of the apparatus to correct for particle scatter is illustrated by the results shown in Fig. 5. It is apparent from these observations that satisfactory performance can be obtained when the equipment is used in the inverted mode. Origin of the Inverted Mode The fluorescence intensities shown in Fig. 3 increase more rapidly than do the associated source intensities of Fig. 2. This occurs because the overlap integral of the o components of the source with the flame absorption profile is also a variable, increasing with increasing field strength to a maximum at about 0.6 tesla. Such a result is not surprising when the combined effects of Zeeman splitting and self-absorption are considered (Fig. 6).In addition, however, the comparison between field-on-source and field-on-atomiser configurations shows that the effect is not symmetrical (Fig. 4): the fluorescence intensity associated with the appropriate o components increases more rapidly with field strength for the field-on-source configuration even when the change in source intensity is taken into account. The observation is of interest as it shows that not only does the magnitude of the overlap integral change with field strength, but the actual form of the source profile must be field dependent as well. Detailed measurement of the field strength along the axis of the air-cored solenoid showed its homogeniety to be quite good:JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 357 Scatter Fig.5. Corrected fluorescence signal given by 0.1 pg ml- * of Mg, 400 mA lamp current, field-on-source and corrected scatter signal observed under the same conditions. The scatter signal was generated by aspirating water with the flame shut off. The intensity ratios of the signals (scatter : Mg) was 2: 1 prior to correction Fig. 6. Schematic representation of the combined effects of self-absorption and longitudinal Zeeman splitting. A, u components with no self-absorption; B, effect of self-absorption on A (the CT+ and u- profiles are drawn separately as solid and broken lines); and C, the sum of u, and CT- curves in B, showing an over-all intensity increase at the zero field frequency relative to curve A. These curves are drawn on the assumption that the emission and self-absorption lines show identical Zeeman splitting.Comparison with the experimental profiles of Fig. 8 indicates this assumption to be incorrect the field strength varied by 4% across the inner diameter of the coil and by <2% axially along the length of the cathode bore. Therefore it was felt that a physical ejection of the self-absorbing atom cloud into a region of low magnetic field strength was an unlikely cause of the discrepancy between the two configurations. Hence it was concluded that a measurement of the associated source profiles would be of interest. The following observations were made. Source Profile Measurements Zeeman absorption scanning of a Varian Techtron Mg lamp was carried out following a method similar to that described by van Heek.8 The scanning furnace containing Mg vapour was located in the steady field of a conventional electromagnet, and a h/4 plate and linear polariser were placed after the furnace to select either the o+ or o- components in the usual way.In the present arrangement both emission and absorp- Table 1. Detection limits Detection Wavelength*/ Lamp limit/ Element nm currentt/mA pg ml- Cd . . . . 228.8 850 0.07 Cu . . . . 324.8 1000 0.1 Fe . . . . 248.3 1800 0.6 Mg . . . . 285.2 1000 0.004 Mn . . . . 279.5 1000 0.25 Detection limit$/ pg ml-1 0.1 0.9 5.0 0.02 0.4 * Monochromator slit width = 300 pm in all instances. t Field-on-source configuration. $ Conventional operation with the lamp under continuous drive at 20 mA, modulated at 285 Hz, and signal measurement by the standard AA5 amplifier. tion multiplets contain only the circularly polarised o? components; as their polarisation states are orthogonal the o+ emission line and o+ absorption line, or alternatively their o- counterparts, can be selectively transmitted using the appro- priate alignment of the A14 plate at & n/4 relative to the polarisation axis, i.e., the system does not mix o+ with o-.The lamp emission profiles observed when no field was applied across the source (Fig. 7) show no unusual features: the lines are symmetrical and remain so irrespective of the degree of self-absorption. By contrast the emission profiles observed after a magnetic field was applied to the source (Fig. 8) are clearly distorted. At low lamp currents normal Zeeman splitting of the source is clear, although the emission maxi- mum appears to be located at a value of the scanning field which falls slightly below the measured value of the field applied to the source. However, increasing the lamp current failed to produce a symmetrical self-absorption of the o emission components in a manner analogous to that seen in Fig. 7.Instead the profile simply becomes broader, while retaining its emission maximum at more or less the same position. The broadening is clearly asymmetric, with both of the o components showing a sharper slope on the low field side of the line. These results were repeated under different conditions of magnetic field strength, pulse length, lamp current and optical configuration. No substantial departures from the trends noted above were observed save only for the transition to the conventional self-absorbed profiles of Fig.7 as the field strength was reduced. A brief mention will be made of the relative behaviour of the o+ and o- components, When no field was applied across the lamp no difference could be observed between these components. This shows the absence of any systematic Doppler shift, and is in accordance with the observations of Falk for measurements on the Ca 422.7-nm line under similar conditions.9 However, in the presence of the magnetic field small differences between the o, and 0- signals occurred as the retarder was moved back and forth. The differences were reproducible over a time span of a few minutes, but not between runs: i.e., allowing the lamp to cool down appeared to destroy the quantitative condition responsible for this observation.A detailed examination of this effect has not been carried out. The o profiles shown in Fig. 8 have been drawn on data points that are the average of successive o+ and 0- measurements. Discussion and Conclusions It is suggested that a reasonable explanation for the nature of the lamp profiles in the field-on-source configurations is obtained if it is assumed that the emission and self-absorption lines are subject to magnetic fields of different strengths. This suggestion appears to be the most satisfactory way to explain the asymmetry of the profiles, which is diagnostic of a frequency displacement between the emission and358 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 1 .o 0 1 .o Scanning field strengthn Fig. 7. Experimental profiles, no field applied to the source. A-D Ire for lamp currents of 14, 50, 100 and 300 mA, respectively shielded from the external magnetic field. The extent of the shielding depends on the actual electron density. Therefore it is felt that the present results are not indicative of a single emission or absorption line located at a fixed magnetic displacement. Rather, it is suggested that, as the lamp current is increased, the emission and self-absorption lines become distributed over a range of frequencies that are associated with different magnetic displacements, and with the emission lines tending to be associated with lower values of the field strength. This basic model is complicated however by the effects of radiation trapping, which further modifies the frequency distribution of the emerging photons.9 Some support for these ideas was obtained by matching the emission and fluorescence data for Figs. 2-4 to a two-layer model for the source.This procedure gave reasonable agreement with the observed results provided that the self-absorbing layer was located in a region of higher magnetic field strength than the emitting layer. No other experimentally realisable conditions were found to allow even approximate agreement to be obtained. Irrespective of the mechanism that shapes the emission profile in the presence of the magnetic field, its experimental consequences are of potential interest. The associated line broadening it produces affords a method to magnetically de-tune the emission and self-absorption profiles of a hollow- cathode lamp, and allows useful signals to be obtained at lamp currents that are precluded under normal circumstances by the onset of self-absorption. Therefore enhanced intensities become available because of the higher lamp currents, and possibly because the de-tuning reduces self-absorption losses as well.This combination is potentially very attractive in an atomic fluorescence source, especially if, as the present data indicate, an effective scatter correction can be achieved at the same time. The results are also of interest because it is believed to be the first time that any sensitivity advantage has emerged to favour one magnetic configuration over the other in a Zeeman-corrected system.It is also the first indication to suggest that a Zeeman-corrected system might be able to offer a fundamentally higher signal to noise ratio than the equi- valent conventional apparatus; the result of the availability of an enhanced source intensity, which leads to a correspond- ingly improved signal to shot-noise ratio. This suggestion is distinct from the capability of Zeeman equipment to give better detection limits for real samples by virtue of its ability to discriminate accurately against high background signals, thereby allowing full use of existing sensitivity to be made. This aspect has been discussed elsewhere.2.3 1 .o 0 Scanning field strengthn 1 .o 1. The authors are indebted to the Natural Sciences and Engineering Research Council of Canada for support of this work.Fig. 8. Experimental profiles with a field of 0.4 tesla applied to the source. A, lamp current = 14 mA; comparison with curve A in Fig. 7 shows self-absorption to be small, indicating that the splitting of this profile is dominated by the Zeeman effect. B, C and D are for lamp currents of 50, 100 and 300 mA, respectively. Comparison with the corresponding curves in Fig. 7 shows these currents to produce marked self-absorption when no field is applied to the source self-absorption lines. It also accounts for the failure of the o components to develop a conventional self-reversed profile at high lamp currents. The diamagnetism of the plasma provides one mechanism that might allow such local field variations to occur. On this basis, regions of high electron density become centres of increased emission intensity that are simultaneously 2. 3. 4. 5. 6 . 7. 8. 9. References Naranjit, D. A., Radzink, B. H., and van Loon, J. C., Spectrochim. Acta, Part B , 1984, 39, $169. de Loos-Vollebregt, M. T. C., and de. Galan, L., Prog. Anal. A t . Spectrosc., 1985, 8, 47. Stephens, R., Talanta, 1978, 25, 435. de Loos-Vollebregt, M. T. C., and de Galan, L., Spectrochim. Acta, Part B , 1980, 35, 495. de Loos-Vollebregt, M. T. C., Van Uffelen, J. W. M., and de Galan, L . , Spectrochim. Acta, Part B. 1982, 37, 527. de Loos-Vollebregt, M. T. C., and de Galan, L., Spectrochim. Acta, Part B , 1982, 37, 659, Guo, T. Z . , and Stephens, R . , Anal. Chem., 1985, 57, 424. van Heek, H. F., Spectrochim. Acta, Part B , 1970, 25, 107. Falk, H., Prog. Anal. A t . Spectrmc., 1982, 5, 205. Paper of J6l22 Received March 26th, 1986 Accepted M a y 28th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100355

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 359 Atomisation in Graphite Furnace Atomic Absorption Spectrometry: Atmospheric Pressure vis-a-vis Vacuum Vaporisation* Ralph E. Sturgeon and James S. Arlow Division of Chemistry, National Research Council of Canada, Ottawa, Ontario KIA OR9, Canada Arrhenius plots were prepared from signals obtained for the atomisation of Pb, Bi and Au from a graphite surface at atmospheric pressure and in vacuum at lO-9atm. Spatially integrated absorbance was recorded for atmospheric pressure studies and mass spectral intensity for vacuum runs. Comparison of activation energies (€a) calculated from the leading edges of the signals showed no significant differences between atmospheric and vacuum vaporisation. The €a values are thus not influenced by potential radial inhomogeneities in the vapour densities within the tube.For the elements and conditions used in this study, Arrhenius data support a vaporisation model that assumes the analyte to be present on the graphite surface in the form of microparticles as opposed to a monolayer. Keywords: Arrhenius plots; atomisation; graphite furnace; vacuum vaporisation Atomisation processes in electrothermal atomisation atomic absorption spectrometry have been studied intensively over the past decade1-’ yet no concensus has developed as to the most likely mechanisms of atom formation let alone the influence of the thermochemical properties of the gas phase in the furnace7-” or the applicability of the models used to investigate these mechanisrns.4,6,12,13 The majority of such studies assume the analyte to be present initially as a monolayer or sub-monolayer on the atomiser ~ a I l .1 ~ 1 7 With such a linear or first-order depen- dence on the mass of analyte present on the surface, the rate of vaporisation is given by where m is the mass of analyte on the surface, IJ is a frequency factor, E, is the activation energy for the rate-limiting process and T(t) is the time-dependent temperature of the substrate (graphite tube wall). In an instance such as this the vaporisa- tion rate constant is given by dmldt rn . . . . . . . . (2) k = - - Such a simple “thermal desorption” scenario describing sample vaporisation is incompatible with many of the experimentally deduced mechanisms that are characterised as processes involving condensed phase species and changes in bulk thermophysical properties.5 In light of this, L’vov and Bayunovl3 assume the analyte to be distributed as a poly- disperse residue of spherical microparticles on the tube surface and within the furnace wall.In the simplest situation, for a spherical particle of radius r13. dm 4nMDrp dt R T( t) . . . . . . - (3) where M is the molar mass, D is the diffusion coefficient of analyte atoms in the inert gas and p is the equilibrium vapour pressure of the analyte at temperature T. The corresponding vaporisation rate constant is given by * . (4) 3MDp r2pRT k = - . . . . . . where p is the density of the analyte microparticle. Irrespective of the physical model selected for study (often dictated by mathematical convenience), evaluation of atomi- * NRCC No.26143. sation mechanisms requires isolation of the source function [ S ( t ) ] from the measured absorbance signal.2.15 The latter is determined by two simultaneously opposing processes that characterise the supply and removal of analyte vapour. The atom population [ N ( t ) ] in the furnace is expressed as a convolution integral of the form N(t) = /‘S(t’)R(t - t’)dt’ . . . . ( 5 ) where R(t) is the mass transport or removal function. Under the condition R(t) >> S ( t ) , which can be approximated at low temperatures (i.e., low vaporisation rates), low rates of heating and with rapid forced removal of atomic vapour from the furnace, the source function can be experimentally isolated for study.*,15.18 Analysis of resultant signal transients is based on the generation of Arrhenius plots that characterise the vaporisation rate constant.Such plots are invariably sigmoid in shape and numerous explanations have been advanced to account for this feature including multiple sequential or competitive mechanisms of atomisation (i.e., a coverage dependent activation energy for release), 14,15 dif- fusion of atoms from the porous structure of the tube wa11,6.16 changes in the mean radius of evaporating microparticles,13 decreased surface coverage of the analyte monolayer19 and incomplete separation of the source function as the tempera- ture increases.15 Reliable determination of the values of the thermochemical parameters of the atomisation process from Arrhenius plots appears possible provided measurements are limited to the low temperature leading edge of the signa1.2.6~13.~~J~ Holcombe and co-workers20J1 have recently obtained temporally- and spatially-resolved absorbance profiles in a graphite furnace and reported the absence of radial symmetry in the atom distributions of many elements.Based on a more generalised form of equation (l), these workers pointed out that the vaporisation (i.e., desorption) of surface material may be described by an equation of the form 0 dm Mdu M dt A dt A - -S-- = S-wuflexp[-E,/RT(t)] . . (6) where S is the surface area of the substrate, M the molar mass of the adsorbed species, A is Avogadro’s constant, u is the surface concentration (molecules per unit area) of analyte and n is the order of the desorption reaction.The kinetic order of a desorption process provides some information as to the mechanism of vaporisation. Zero-order kinetics often indicate desorption from a multilayer or three-dimensional structure where the rate of desorption is independent of surface coverage. First-order kinetics may be indicative of monolayer coverage. It follows that, if the order360 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 is greater than zero, the rate of desorption will be influenced by the amount of material present or the surface coverage. Holcombe and co-workers20.21 utilised equation (6) as a basis for explaining the radial concentration gradients in the furnace. During atomisation, analyte atoms released from their original site of desorption (i.e., from the lower wall) may adsorb on the cleaner upper walls of the furnace (where the sample coverage is initially zero).The rate of adsorption of vaporised analyte will exceed its rate of desorption from these clean surfaces, producing a radial concentration gradient. As vaporisation continues the surface coverage becomes more uniform as does the rate of desorption from all surfaces and the concentration gradient disappears. Studies by Holcombe and co-workers*0,21 suggested that E, values deduced from Arrhenius plots may be highly dependent on the zone of observation and that such values derived from spatially integrated absorbance measurements may be in error as a consequence of the contribution to the signal by secondary desorption of analyte from other surfaces.In such instances, desorption of isolated adatoms is not likely to be characterised by the same E, as that of atoms from the original sample residue. The present study was undertaken to investigate the possible errors that may arise through use of spatially integrated optical data to prepare Arrhenius plots from which E, values are calculated. Comparisons of E, values so obtained with those derived from mass spectral (MS) data characterising the vacuum vaporisation of analyte from a graphite surface permit evaluation of the significance of such effects as the free atom spatial distribution, source signal deconvolution and multiple atom - wall interactions on the resultant values. The advantages of mass spectral detection for mechanistic studies have been outlined earlier.22.23 The ability to use rapid heating of the furnace while simultaneously maintaining an isolated source function at all temperatures is particularly noteworthy for this study.Experimental Instrumentation Spatially integrated optical measurements of atomic absorp- tion were made using a fast response (tRC = 10 ms) detection system. Samples were atomised in a Perkin-Elmer HGA-2200 furnace mounted on a modified Varian Techtron Model AA-5 spectrometer. Signal transients were recorded on a digital storage oscilloscope (Gould, Model OS4000) and subse- quently re-plotted on a strip chart recorder for evaluation, A nominal heating rate of 480 K s-1 with an internal purge gas flow-rate of 200 ml min-1 was used. Vacuum vaporisation was achieved using a CRA-63 type atomiser and a quadrupole mass spectrometer was used for analyte detection as described earlier.23 The time constant of the undamped system was about 0.2 ms.Atomisation was performed at a nominal system pressure of 7 X 10-4 Pa (5 x 10-6 Torr) using a heating rate of 1160 K s-l. Temperature measurements were taken with a calibrated automatic optical pyrometer (Ircon Inc., Model 1100, Niles, IL, USA) by sighting through the sample dosing holes in the furnaces and assuming the radiator to be a black body. Temperature errors for these geometries are negligible.24 Reagents Stock solutions of Pb, Au and Bi were prepared by dissolution of the high-purity metals (Spex Industries Inc., Metuchen, NJ, USA) in dilute HN03 and/or aqua regia. Working standards of these elements were prepared by serial dilution of the concentrate with 1% V/V HN03.Procedure Samples were introduced into the HGA-2200 in 10-yl volumes using a Perkin-Elmer AS-1 autosampler. Samples were manually dispensed into the CRA-63 furnace in 2-1-11 volumes using an Eppendorf pipette. Similarly, 2-y1 sample volumes were applied to the outside surface of the CRA-63 by successively drying multiple 0.5-yl volumes as the furnace was suspended in an inverted vertical position. This ensured even drying on a well-defined spot on the convex surface. All optical absorbance measurements were made at the resonance lines of the elements. Mass spectral measurements were centred at 197, 208 and 209 a.m.u. for Au, Pb and Bi, respectively. Absolute masses of 3, 5 and 5 ng of Pb, Bi and Au, respectively, were atomised for absorbance measure- ments while 25, 15 and 30 ng of Pb, Bi and Au, respectively, were taken for mass spectral measurements.Samples were “ashed” at 720 K in Ar for optical measurements and at 700 K in air for mass spectral measurements with a further 700 K “ashing” stage completed irz vacuo prior to atomisation. The detection system was “tuned” to the correct mass by peaking the response arising from low temperature volatilisation of small chips of the ultrapure metal placed within the furnace.23 Arrhenius plots were constructed froin the absorbance (or mass spectral intensity) and temperature - time data according to the procedures outlined by Smets16 and Guerrieri et a1.19 Results and Discussion Table 1 summarises results for the determination of release energies obtained for the atomisation of Pb, Bi and Au at both atmospheric pressure and at 7 X 10-9 atm.In the latter case, results are given for vaporisation of sample from both the interior and exterior surfaces of the CRA-63. Also given in Table 1 are the enthalpies of potential reactions that may describe the over-all analyte vaporisation process. The E, values were calculated from the least-squares fitted slopzs of Arrhenius curves derived from the leading edges of the signal transients, as shown in Figs. 1-3. The initial linear portion of the plot spanned a 100-230 K temperature range, it being greatest for Au and least for Pb. Table 1. Activation energy of leading edge. Numbers in parentheses denote the number of trial runs contributing to the reported averages Activation energy/kcal mol- 1 7 X l0-9atm 1 atm Ar, Interior Exterior interior surface surface AHlkcal mol-1 Pb _ .. . . . 97+6(5) 89 k 2 (3) 100 _+ 3 (3) 94.5 (1300 K) Bi . . . . . . 94_+6(3) 93 2 2 (2) 99 k 5 (4) 100.6 (1 100 K) A u . . . . . . S9_+4(3) 83 2 (2) 77 _+ 7 ( 5 ) 84.5 (1600 K) Element surface PbO(,,+ Pb,,, + V 2 0 2 Y2Bi2O3(,)+ Bi,,, + %02 Au(l,-+ 4,)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 36 1 2.0 1.5 1 .o 0.5 c I cn 0, 2 s o -0.5 -1.0 i ‘4. 0 0 I I I I I ~ 6.5 7.0 7.5 8.0 8.5 9.0 Fig. 1. Arrhenius plot for atomisation of 25 ng of Pb in CRA-63 at 7 X 10-9 atm. Mass snectrnmetric detection at 208 a m 11 1 O ~ T - ~ I K - 1 -4.0 7.0 7.5 8.0 8.5 9.0 104 T- VK- 1 Fig.2. Arrhenius plot for atomisation of 15 ng of Bi in CRA-63 at 7 X lo-’ atm. Mass spectrometric detection at 209 a.m.u. Comparison of the data obtained in Ar and at 10-9 atm suggests that in our case E, values derived from the leading edge of the signals have not been compromised by the spatially integrated nature of the optical absorbance measurements. Agreement of our vacuum data with our data obtained at atmospheric pressure suggests that the flux of analyte atoms that may be desorbing from the upper surfaces of the furnace (operated at atmospheric pressure) is insignificant with -1.5 -2.0 -2.5 -3.0 38. + -3.5 0 0 0 \ 4.0 5.0 6.0 1047- 1 i ~ - 1 Fig. 3. 7 X Arrhenius plot for atomisation of 30 ng of Au in CRA-63 at atm. Mass spectrometric detection at 197 a.m.u.respect to that generated by the original sample over the leading edge of the signal and, hence, does not influence the initial slope of the Arrhenius plot. It may be possible that the 200 ml min-1 gas flow-rate through our furnace produces a more homogeneous distribution of atomic vapour than that measured by Holcombe and co-workers20J1 in their “static” CRA-type device. Unfortunately, spatially resolved absor- bance measurements could not be made in this study to verify the above. Despite the significant mean-free path of atoms released from the graphite surface in a vacuum (several metres at 10-9 atm), the possibility of their “sticking” to the upper wall of the furnace (upon collision) following initial vaporisation is real when the sample is initially deposited inside the furnace. However, for atomic vapour arising from the volatilisation of samples deposited on the exterior wall of the tube, such desorption - adsorption - re-desorption phenomena on the upper wall should, to a first approximation, be ruled out. In the simplest of cases, re-deposition on to the clean walls of the tube under vacuum conditions does not directly bear on the work of Holcombe and co-workers20~21 because mass trans- port times are extremely short at 10-8 atni and vapour density gradients that lead to artefacts in the measured spatially resolved absorbance measurements do not develop.However, the upper walls may have an additional influence on the over-all release processes. Gas phase oxide released from the initial analyte deposit may undergo reduction upon collision with the upper wall.At atmospheric pressure one should expect that such a reaction would produce observable atoms via a rate limiting step (RLS) corresponding to either the heat of vaporisation of the oxide, the dissociation energy of the surface adsorbed molecule or the desorption energy of the reduced metal from the upper surface. It is to be noted that such a secondary source of atoms has, as its origin, isolated adsorbed analyte species and not a three-dimensional multi- layer analyte deposit surrounded by a Langmuir film. With vaporisation in vacuum and, for brevity of consideration, with the sample desorbing from the exterior wall of the tube, it may be argued that the electron bombardment energy of the ionisation source (70 eV) could simulate the effect of a reduction reaction involving collision of an analyte oxide with the (virtual) graphite wall by providing the energy needed to362 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 produce the free metal. In this instance, however, it is not likely that dissociation by electron impact in the MS source is the RLS, leading to the conclusion that the RLS must be the vaporisation of the analyte oxide. Thus, in order that the measured E, values for both vacuum and atmospheric pressure runs be identical, as they are in these experiments, it is necessary that the RLS be the same in both instances, i.e., the A Hvap of the metal oxide. These possible reactions may be illustrated using lead as an example. At atmospheric pressure PbO(q -+ PbO(,) + [PbO(,d) + Pb(ad) O( I 1 Lad) Pb,,, + CO - - (7) where (ad) signifies an isolated surface adsorbed species; at 10-9 atm PbO(l)+ PbO,,) %-Pb&) + qti .. . . (8) In both instances the reactions produce measurable Pb(,). However, the over-all reaction is different. At atmospheric pressure the products would be Pb(,) and CO. At 10-9 atm Pb,,, and O(,) are produced. On the basis of the measured E, values for both cases, it seems probable that the reaction enthalpy for equation (9) is, in fact, measured (see Table l ) , PbO(l)+ Pb,,) + 9502 . . . . . . (9) with the conclusion that the influence of the upper wall in producing such reactions is insignificant during the time frame encompassing the initial portion of the leading edges of the signals of the elements studied.In addition to the upper wall surface, it is necessary to consider the effects of the bottom surface from which the sample initially vaporises. With desorption of mono- or sub-monolayers or single adsorbed species, the effect of the lower surface is “transparent” to the absorbance signal in that the measured desorption energy directly reflects adsorbate - substrate interaction energies. With “desorption” from multi- layer or bulk three-dimensional analyte deposits the surface roughness of the graphite must be considered since vaporised material may interact with graphitic surface structures that are in close proximity to the analyte deposit. Surface desorption commonly obeys the cosine law of emission, the greatest flux being directed perpendicular to the evaporating surface.However, the small amount of material released at grazing angles to the substrate will collide with it, stick and subsequently desorb. Depending upon the extent of penetration of the sample into pore structures in the tube wall or the surface roughness of the graphite, interaction of the gas phase analyte with the nearby graphite surface may occur before the analyte becomes a detected gas phase entity. This effect would be equally operative at 1 and at 10-9 atm. However, the good agreement of E, values with enthalpy changes for these reactions suggests that, over the initial part of the leading edge of the signal, the flux of observable atoms from such secondary desorption sites on the lower wall surface (characterised by different release energies) is insignificant compared with the major flux from the surface of the primary release site.These “surface effects” are independent of those which may be produced by the interaction of gas phase analyte with the upper or side wall surfaces, as discussed earlier and which may have a direct influence on E, data derived from spatially integrated absorbance measurements.20Jl The atomisation processes suggested in Table 1 are in agreement with those postulated by other workers. 1,6,7,10,13~16 The data suggest that atomisation occurs from “bulk” three- dimensional structures or multilayers possessing standard state thermodynamic properties. Agreement between measured E, data and enthalpy changes characterising the over-all atomisation reactions is not unexpected at atmospheric pressure.A pseudo- equilibrium may be established in a thin boundary or diffusion layer (Langmuir film) above the condensed phase analyte wherein multiple release - adsorption processes occur prior to final escape of the analyte vapour into the observation volume of the “free” gas phase. In such circumstances equilibrium processes are actually being measured and the Arrhenius plot yields a A H for the over-all reaction.19 Based on the good agreement between the atmospheric pressure and vacuum vaporisation data given in Table 1, it would appear that the same reactions are occurring at 10-9 atm. It is extremely difficult to envision a pseudo-equilibrium release - deposition process developing under these circum- stances such that the E, = AH.It is, likewise, difficult to imagine that the RLSs for desorption of Pb(,) and Bi,,) are fortuitously equal to the over-all enthalpy changes accorn- panying these reactions, as they are at atmospheric pressure. The alternative explanation of the data for both instances (1 and 10-9 atm) is to assume that the RLSs for desorption are associated with activation energies that fortuitously corre- spond to the A H of the proposed reactions. Thus, no pseudo-equilibrium is necessarily established even at 1 atm and the release process is purely under kinetic control. A comprehensive interpretation of the data remains elusive. The existence of microcrystals on the atomiser surface following sample dehydration is well documented by direct surface analytical studies,25-*7 as is the mobility of surface adsorbed atoms at high temperature28 and the tendency for monolayer films of some elements to “bead” or “roll-up” into droplets upon heating.29 Thus, it is not improbable that, at least during the initial stages of analyte release from the surface (i.e., over the leading edge of the signal), atomisation originates from three-dimensional structures. Such conditions also lead to the less than first-order desorption that charac- terises the atomisation of some elements from the graphite ~urface.3~5.20J~ It should be noted that vaporisation of the major amount of the sample may subsequently occur from two-dimensional structures or isolated adatoms once the three-dimensional structures have vapor.ised.5 This, of course, would result in a spectrum of E, values characterising release as the temperature increased. Alternatively, such pseudo-first order desorption kinetics often observeti20J1 may be the result of an equilibrium between different surface phases with one phase maintaining a constant surface concentration of the desorbing phase.30 Arrhenius plots for the vacuum vaporisation of Pb, Bi and Au are given in Figs. 1-3.These data pertain to atomisation of samples placed inside the furnace; virtually identical results were obtained when the samples were loaded on to the exterior wall. The Arrhenius curve in Fig. 1 was prepared according to Smet’s procedure16 whereas those in Figs. 2 and 3 illustrate the methodology suggested by Guerrieri et al. 19 Manipulation of the MS signal intensity - temperature data by either technique produces curves of the same slope, only the intercepts must be interpreted differently.The sigmoid shape, characteristic of all source func- tions2?5,6,13,19 is clearly evident in these figures. The bending of the curves toward the abscissa at high temperatures does not reflect incomplete separation of the supply and removal functions. The increase in slope corresponding to the extreme of the falling edge of the signal has been reported by others6.13.19 and is the most interesting feature of these curves. Guerrieri et al. 19 theoretically predicted such an increase in the high temperature portion of the Arrhenius plot. Accord- ing to this model,19 atomisation is assumed to be a reversible release process in which heterogeneous equilibrium is estab-1 lished between a monolayer or submonolayer surface film and a small interfacial region in the gas phase.A single binding energy is assumed to characterise all analyte atoms on the surface. By considering that the condensed phase sample takes the form of a polydisperse ensemble of spherical microparticles distributed over the tube surface, L’vov and Bayunovl3 alsoJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 363 correctly predicted the sigmoid shape of the Arrhenius curves. According to this model, the vaporisation rate constant [equation (4)] increases continuously as the particle vaporisa- tion proceeds. Only over the leading edge of the signal, in the low-temperature region of the Arrhenius plot, is a linear relationship expected because the size of the microparticles changes least in this temperature interval. The high-temperature region of the Arrhenius plots shown in Figs.1-3 has been approximated by a linear fit to the data. The signal in this extreme portion of the falling edge is noisy. Although there is no uncertainty as to the increase in the vaporisation rate constant with temperature in this region, the suggestion that the data describe a straight line is speculative. Extreme data points (not shown in these figures) increase faster than the linear relationship implied. This is in agree- ment with the predictions of the model of L’vov and Bayunov.13 However, the large error in estimating the signal in this region makes these data unreliable.A similar upward trend in the high-temperature portion of the Arrhenius curve is also evident in the data presented by Guerrieri et af.19 It is evident that the qualitative shape of the Arrhenius plots does not permit choice of a model describing analyte atomisation, i. e. , pseudo-equilibrium release from either monolayer19 or spherical three-dimensional deposits, l 3 as both predict much the same characteristics for the Arrhenius plot. Quantitative E, data, however, appear to be in good agreement with enthalpy changes governing over-all reactions of bulk three-dimensional deposits. This would not be so had vaporisation occurred from monolayers or isolated adsorbed species or islands.29 It is noteworthy that Arthur and Cho29 have characterised the energetics of desorption of monolayers and islands of Au from the basal plane of graphite and obtained coverage dependent release energies ranging from 52 to 74 Kcal mol-1.The value of 83 Kcal mol-l (average) obtained in these studies clearly suggests the presence of three-dimensional Au structures on the surface. Similarly, Guerrieri et a!. 19 obtained a release energy of 107 Kcal mol-1 for atomisation of (an assumed) monolayer deposit of PbO from a carbon filament atomiser. This release energy, characterising the enthalpy change accompanying the atom- isation process is also similar to the AH for reaction ( 9 ) . The question arises as to whether this is simply a fortuitous occurrence that the A H for monolayer desorption (charac- terising such a strong adsorbate - substrate interaction) phenomenon is so similar to the A H for reaction ( 9 ) involving multilayer three-dimensional PbO deposits.Conclusions Atomisation of samples in high-vacuum facilitates isolation of the source function in the graphite furnace because of the rapid removal of analyte vapour. Agreement between activa- tion energies obtained from vacuum vaporisation and atmos- pheric pressure vaporisation suggests that spatially integrated absorbance data, measured under conditions that effectively isolate the source function, are not influenced by the possible radial asymmetry of the vapour density in the furnace over the initial portion of the leading edge of the signal. The qualitative shape of Arrhenius plots resulting from vacuum vaporisation does not permit testing of the various models proposed for atom formation.However, based on data presented here, it would appear that simple vaporisation of microparticles , as opposed to pseudo-equilibrium vaporisation of a monolayer film, is a more acceptable hypothesis for the metals studied. The need for a more detailed understanding of the surface distribution of the initial analyte residue is obvious. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. References L’vov, B. V., Spectrochim. Acta, Part B , 1978, 33, 153. Paveri-Fontana, S . L., and Tessari, G., Prog. Anal. A t . Spectrosc., 1984, 7, 243. Holcombe, J. A., and Rayson, G. D., Prog. Anal. At. Spectrosc., 1983, 6, 225. Katskov, D. A., Zh. Prikl. Spektrosk., 1979, 30, 612.Sturgeon, R. E., Fresenius 2. Anal. Chem., in the press. L’vov, B. V., Bayunov, P. A., and Ryabchuk, G. N., Spectrochim. Acta, Part B , 1981, 36, 397. Frech, W., Lundberg, E . , and Cedergren, A., Prog. Anal. Ar. Spectrosc., 1985, 8, 257. Cedergren, A., Frech, W., and Lundberg, E . , Anal. Chem., 1984, 56, 1382. Sturgeon, R. E., Siu, K. W. M., and Berman, S. S . , Spectrochim. Acta, Part B , 1984, 39, 213. L’vov, B. V., and Ryabchuk, G. N . , Spectrochim. Acta, Part B , 1982, 37, 673. Salmon, S. G., and Holcombe, J . A., Anal. Chem., 1982, 54, 630. Frech, W., Zhou, N. G . , and Lundberg, E . , Spectrochim. Acta, Part B , 1982, 37, 691. L’vov, B. V., and Bayunov, P. A., Z h . Anal. Khim., 1985,40, 614. Sturgeon, R. E., Chakrabarti, C. L., and Langford, C. H., Anal. Chem., 1976, 48, 1792. van den Broek, W. M. G. T., and de Galan, L., Anal. Chem., 1977, 49, 2176. Smets, B., Spectrochim. Acta, Part B , 1980, 35, 33. Torsi, G., and Tessari, G., Anal. Chem., 1975, 47, 839. Gilmutdinov, A. Kh., and Fishman, I. S . , Spectrochim. Acta, Part B , 1984, 39, 171. Guerrieri, A., Lampugnani, L., and Tessari, G., Spectrochim. Acta, Part B , 1984, 39, 193. Rayson, G. D., and Holcombe, J . A., Spectrochim. Acta, Part B, 1983, 38, 987. Holcombe, J . A., Rayson, G. D., and Akerlind, N. J r . , Spectrochim. Acta, Part B, 1982, 37, 319. Styris, D. L., Anal. Chem., 1984, 56, 1070. Sturgeon, R. E., Mitchell, D. F . , and Berman, S. S., Anal. Chem., 1983, 55, 1059. Falk, H., Spectrochim. Acta, Part B , 1984, 39, 387. Muller-Vogt, G., Wendl, W., and Pfundstein, P., Fresenius 2. Anal. Chem., 1983, 314, 638. Muller-Vogt, G., and Wendl, W., Anal. Chem., 1981,53,651. Sire, J., and Voinovitch, I. A., Anafusis, 1979, 7, 275. Hayward, D. O., and Trapnell, B . M. W., “Chemisorption,” Butterworths, London, 1964. Arthur, J . R., and Cho, A. Y., Surf. Sci., 1973, 36, 641. Yates, J. T. Jr., in Park, R. L., and Lagally, M. G., Editors, “Solid State Physics: Surfaces,” Volume 22, Academic Press, New York, 1985, p. 425. Paper JA6l3 Received March 7th, 1986 Accepted May 27th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100359

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 365 Depth Concentration Profiles Obtained by Carbon Furnace Atomic Absorption Spectrometry for Nickel and Aluminium in Human Skin John F. Alder and Maria C. C. Batoreu" Department of Instrumentation and Analytical Science, University of Manchester Institute of Science and Technology, PO Box 88, Manchester M60 IQD, UK Anthony D. Pearse and Ronald Marks Department of Medicine, University of Wales College of Medicine, Heath Park, Cardiff CF4 4XN, UK Carbon furnace atomic absorption spectrometry was used to detect nickel and aluminium within the skin after topical application of materials containing these substances. Determination of the metals at particular depths within the skin was achieved by horizontal sectioning of the biopsies and analysis of the 20-30 pg skin samples obtained directly by dry ashing and atomising the solid skin sample in the furnace.The method was validated using wet ashed guinea pig skin samples and standard additions procedures and standard deviation data established using gelatine samples. Care was taken in developing the sample preparation procedure to assure no contamination of the skin sample by trace metals. The data obtained indicate that there is an apparent accumulation of nickel in fixed positions in the skin, perhaps associated with a higher density of Langerhans cells in these positions. Keywords: Skin; nickel determination; aluminium determination; atomic absorption spectrometry; Langerhans cells Nickel is one of the most common causes of allergic contact dermatitis,l.2 some of its water soluble salts, nickel chloride and nickel sulphate being strong sensitisers.Measurements of trace amounts of nickel in tissues by atomic absorption spectrometry (AAS) is difficult due to the low concentration present and poor sensitivity of AAS. Carbon furnace atomic absorption spectrometry (CFAAS) is the provisional refer- ence method used by IUPAC3 for comparative evaluation of other procedures for measurements of nickel in biological materials. Topically applied aluminium compounds are used in cosmetic and medicinal preparations for their astringent, antibacterial and antiperspirant properties. Aluminium salts reduce axillary sweating and are also active against bacteria that decompose macerated stratum corneum causing odour.The reduction of sweating has been ascribed to the aluminium complexing in the terminal sweat duct resulting in ductal obstruction4 with inhibition of secretion. Lansdowns observed epidermal damage after topical application of different aluminium salts, and noted that they could penetrate deeply and reach the dermis. The objectives of the present study were to determine the total amounts of nickel and aluminium that penetrated after application to skin, and to characterise the specific distribution patterns in the skin, using the technique described. Experimental Procedures An unmodified Perkin-Elmer 305 atomic absorption spectro- photometer with HGA-70 carbon furnace and deuterium arc background correction was employed. Output was to a Servoscribe 210 strip chart recorder (250-ms full-scale deflec- tion response time).A Perkin-Elmer AD2 digital electromag- netic ultramicrobalance (100-ng resolution) was used to weigh samples. The relative standard deviation of weighing was found to be ca. 2.4%. Biopsies (4 mm diameter) were obtained from normal human volunteer subjects who had given their informed consent and from a guinea pig. The preparative (doping) procedures for aluminium and nickel differ, but successive steps are common to both elements. * Present address: Faculdade de Farmacia, Universidade de Lisboa, Av. des Forces Armadas, 1699 Lisboa Codex, Portugal. Untreated skin was also biopsied and analysed in the same way, as a control. Instrument operating conditions are given in Table 1. The investigations were performed using guinea pig and human skin in vivo.A spray of 20% m/V aluminium chloride hexahydrate solution was applied to the shaved skin of a guinea pig and to the forearm of a human volunteer. The spray was left undisturbed for 4 h, after which time biopsies were taken. For the human skin, transparent adhesive tape (Sellotape) was applied several times prior to the biopsy to remove the excess of aluminium from the surface. Four nickel-sensitive and four nickel non-sensitive subjects were recruited. A 5% suspension of nickel sulphate in white soft paraffin was applied to the lower back of each volunteer using occlusive dressings. The dressings were left in place for 48 h, after which time the biopsies were taken. In the guinea pig a parallel study was carried out in order to obtain both small sections and larger samples of the same skin.The aluminium content of the larger samples was determined by wet digestion in order to validate the direct method. About 30 mg of skin were digested with 0.2 ml of concentrated nitric acid in a PTFE lined high-pressure digestion vessel heated for 2 h at 150°C. The solution was diluted to 10 ml and the aluminium and nickel contents determined by standard additions (Table 2). Skin biopsy samples were frozen in hexane that had been cooled to -70°C with an ethanol - solid carbon dioxide mixture. The skin tissue was surrounded by a viscous inert polymer (Tissue-Tek, R. A. Lamb & Co. Ltd) for correct orientation prior to freezing. Samples were then stored in liquid nitrogen prior to sectioning.Sections (20 pm) were cut through the samples on a Brights motor driven cryostat. The cabinet temperature was -20 "C and the temperature of the stainless-steel knife (Jung, FRG) was -70 "C. All the sections analysed for the penetration of Al and Ni into the skin were cut in the horizontal plane through the skin from the stratum corneum to the lower dermis. Every tenth section was stained with haematoxylin and eosin for microstructural evaluation. There was no evidence of nickel contamination from the stainless-steel knife. The 20-pm cut sections were mounted on glass slides that had been left overnight in 5% V/V nitric acid, thoroughly washed in doubly distilled water and then air dried.366 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 Table 1. Instrument operating conditions: Perkin-Elmer 305 AA spectrophotometer with background correction and an HGA-70 carbon furnace Drying Ashing Atomisat ion Slit - Argon Element Line/nm Time/s "C Time/s "C Time/s "C pass/nm 1 min-' Temperature/ Temperature/ Temperature/ band flow-rate/ A1 . . . . 309.2 30 110 30 1200 3 2700 0.3 3* Ni . . . . 232.0 30 170 30 1200 5 2650 0.3 2* * Gas flow interrupted during atomisation period. Table 2. Mean value of metal content in control skin (horizontal sections) A1 Ni No. subjects No. subjects Result/ from whom samples No. Result/ from whom samples No. Sample Pg g-' taken determinations I.18 g-' taken determinations - - - Guinea pig . . . . 13.6 k 1.9* 1 4 52 4.4 k 3.0 4 sensitive 50 4 non-sensitice Human skin .. 13.1 k 4.9 1 16 2.9 k 1.65 * Wet digestion. The skin sections were recovered by removing the T-Tek with a plastic needle and passing the slide quickly through steam, and made into a ball in order to facilitate the subsequent handling procedure. Slides with the skin balls were dried in the oven at ca. 60°C for 2 h. The sections were then weighed on the ultramicrobalance using methyl cellulose capsules as containers. Transfer of the sample was achieved with a vacuum pipette; the sample mass range was 20-70 pg. The skin samples were introduced directly into the furnace with a vacuum micropipette. Owing to the very small size of the samples much care had to be taken in order not to lose them during these steps. Experience in handling was of the greatest importance and took some time to acquire.Inevitably, some samples were lost, mainly those corresponding to the first layers of dermis, which were not only smaller in diameter, but also of a poorer consistency. There was a very small amount of residue left in the furnace after atomisation, which could be removed by a puff of air. Control Procedures The level of nickel and aluminium in untreated skin was determined by the dry-ashing method (Table 1) in order to establish control values (Table 2). The aqueous calibration graphs for aluminium and nickel were linear in the range studied (0-1.5 and 0-60 ng, respectively). The metal concen- tration obtained by solid injection was determined from the corresponding aqueous calibration graphs; this approach had been validated in previous ~ o r k .6 3 ~ The fact that aqueous calibration standards could be employed in the procedure was not only most useful, but also served to prove that the background correction was able to compensate well for the residual amount of carbonaceous smoke that appeared during the atomisation step and that matrix effects on these elements were negligible. Results Fig. 1 shows the aluminium level in the treated skin of one subject, plotted as a function of depth from the skin surface. These concentrations represent the mean value of three individual determinations of each section mounted on the same slide. The individual values were not considered as it was not possible to give them an ordered value in relation to the skin surface distance. The 20-pm sections cut from this biopsy sample were mounted three to a slide and unfortunately the r P, 1 l4OIl 120 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Distance from skin surfaceimm Fig.1. biopsy non-sensitive; the broken line is the control value Penetration of aluminium in human skin after 4 h application; order in which they had been mounted was not noted at the time; because of the nature of the sampling, the experiment could not be repeated. Where the concentration of metal in the skin sections was becoming low, several sections had to be grouped together to give sufficient mass of metal for a strong signal to be obtained. At greater depths in the aluminium (Fig. 1) and in the nickel biopsy (Fig. 2) groups of three or more sections were analysed simultaneously and account for the grouped points in Fig.1 and larger blocks in Fig. 2. Control values for aluminium are those reported in Table 2. For nickel, four biopsies were analysed from sensitive and from non-sensitive subjects. The plots were all similar and typified by Fig. 2. The 10% relative standard deviation on the points is that found using gelatine as a microsolid standard. Control values for epidermis - dermis and dermis are reported in Table 3. The biopsies were mounted with one section on each slide until a depth of 200 pm was attained. The slides after 200 pm contained three sections that were injected at the same time, because the absorbance values obtained after 200 pm were much lower. Moreover, it was the only way of analysing the control nickel samples, as the values found were close to the detection limit (0.2 ng of nickel).The most significant value that reflects inhomogeneity in metal distribution, is the concentration of metal in the skin. It was not forgotten that in opting for concentration, apparent inhomogeneity in concentration itself reilects to some extent inhomogeneity in the density of skin. The volume of biopsyJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 .- 100 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 Distance from skin surface/mm Fig. 2. Penetration of nickel in human skin after 48 h application (assumed 10% RSD); biopsy 11, non-sensitive; the broken line is the control value Table 3. Nickel concentration in horizontal sections of control skin biopsies; values in parentheses are the number of determinations Epidermis - upper dermis/ Biopsy CLgg-' Dermis/pg g- Sensitive- I I1 111 IV 4.4 f 0.55 (2)* 3.8 k 1.85 (1)t 6.0 f: 0.61 (1)t 12.0 t 4.0 (3)* 1.53 f: 0.51 (12) 3.76 5 1.85 (14) 1.39 k 0.61 (11) 2.77 ? 0.31 (10) Non-sensitive- I 4.5 ? 1.0 (2)* 1.74 k 0.44 (10) I1 3.6 f: 0.53 (1)t 2.05 k 0.53 (9) 111 4.1 k2.9 (3)* 1.36k 0.51 (13) IV 3.9C 1.3 (3)* 1.70 ? 0.61 (13) * Error figure quoted is the range.t Error figure quoted is the standard deviation for the dermis results. sections from different regions could alter by as much as a factor of four. Any judgement must therefore be based on comparison between inhomogeneity in absolute amount of metal per biopsy section and inhomogeneity in its concentra- tion, taking into account inhomogeneity in density. Fig.3 shows this comparison for a biopsy from a nickel sensitive person. Some of the peaks occur in the same position in both the absolute quantity and concentration plots, clearly demonstrating that these are real inhomogeneities in metal content. A feature that distinguished the distributions in the sensitive and non-sensitive subjects is the presence of an inhomgeneity that occurs in approximately the same position within the epidermis. In some instances the deviations are quite large, suggesting an apparent accumulation of the metal. Discussion Similar apparent accumulation of metal in fixed positions in the skin has been found by other workers8.9 and onelo postulated a selective uptake of some metals, e.g., nickel, chromium, cobalt, mercury and gold, by Langerhans cells.Liden and Lundberg,8 when determining chromium in sec- tions of treated skin, also found higher deviations of chro- mium distribution and they concluded that the deviations occurred at similar sites at which the Langerhans cells appear in highest density. Molokhia and Portnoyll found regional variations in skin copper, manganese and zinc levels over the body. The lowest concentrations were found in the planar epidermis and the highest in the epidermis of the foreskin. They thought it possible that these variations were related to the dendritic cell population in these areas. 90 80 70 c I 60 % C 0 'E 50 2 2 40 . c C a 8 - a Y u 30 z .- 20 10 L 0 0.2 0.4 0.6 0.8 1.0 1.2 Distance from skin surface/mm Fig. 3. Comparison between the depth profile of absolute nickel content and its concentration in each skin sample, biopsy 111, nickel sensitive patient The results reported here demonstrate that CFAA can be utilised for measuring the metal content of minute samples of tissue and is capable of detecting the low concentrations of nickel and aluminium encountered in dermatological investi- gations.The sample preparation procedures are critically important in generating meaningful data and at each stage great care was taken in maintaining the integrity and purity of the samples. Contamination of samples by the microscope slides and packing materials was checked for by running blank samples through the same procedure. The samples were sent by post between the laboratory in the Welsh National School of Medicine where the biopsies were taken and sectioned, and the laboratory in UMIST where the analyses were performed. No evidence of contamination was found when the handling procedures and standards of cleanliness had been established. Clearly, the omnipresence of aluminium demands particular care in this respect. M.C.C.B. was supported by the Instituto Nacional de Investigacao Cientifica, Portugal. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. References Cavelier, C., Foussereau, J., and Massin, M., Contact Dermati- tis, 1985, 12, 65. Lacroix, J., Moran, C. L., and Collin, P. P., J . Pediatr., 1979, 95, 428. Sunderman, F. W., Pure Appl. Chem., 1980, 52, 529. Papa, C. M., and Kligman, A. M., J . Invest. Dermatol., 1967, 49, 139. Lansdown, A. B. G . , Br. J . Dermatol., 1973, 89, 67. Alder, J. F., and Batoreu, M. C . C., Anal. Chim. Actu, 1982, 135, 229. Alder, J. F., and Batoreu, M. C. C., Anal. Chim. Acta, 1983, 155, 199. Liden, S., and Lundberg, E., J . Invest. Dermatol., 1978,72,42. Shelley, W. B., and Juhlin, L., Nature (London), 1976,261,46. Shelley, W. B., and Juhlin, L., Arch. Dermatol., 1977, 113, 187. Molokhia, M. M., and Portnoy, B., Br. J . Dermatol., 1970,82, 254. Paper J6l27 Received April 21st, 1986 Accepted May 27th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100365

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 369 Determination of Cadmium in Blood Plasma by Graphite Furnace Atomic Absorption Spectrometry M. M. Black* and Gordon S. Fell Department of Clinical Biochemistry, Royal Infirmary, Glasgow G4 OSF, UK John M. Ottaway Department of Pure and Applied Chemistry, University of Strathclyde, 295 Cathedral Street, Glasgow GI IXL, UK A graphite furnace atomic absorption spectrometric (GFAAS) procedure is described for the determination of Cd in rat blood plasma. Blood plasma was deproteinised with nitric acid prior to injection into the furnace. Calibration was with Cd in acidic solution and absorption signals were read in the integration mode. The method has a detection limit of 1.26 pg of Cd, and an imprecision of 1-4% (relative standard deviation) at a plasma concentration of 1 l g 1-1 of Cd.Accuracy was established by quantitative recovery of Cd added to plasma and by intercomparison of results with other atomic spectrometric procedures. Plasma Cd determination and the radioimmunoassay of plasma metallothionein are useful for the study of the mechanism of Cd toxicity in animals. Eventual applications in human occupational medicine could provide a warning of Cd toxicity before irreversible renal damage has occurred. Keywords: Plasma cadmium determination; graphite furnace atomic absorption spectrometry; nitric acid dep ro teinisa tio n; signal integration Occupational and environmental exposure to cadmium is a known health hazard. Cadmium accumulates in body tissues, particularly the liver and kidneys, and has a long biological half-life.' A critical organ effect is seen in the kidney when a level of 100-200 yg g-1 of Cd (wet mass) is reached in the renal cortex.The resultant tubular and glomerular damage is observed as increased urinary excretion of protein, calcium, phosphate and other substances. The clinical consequences are varying degrees of bone disease and an increased tendency to form renal calculi. At present the laboratory monitoring of a workforce exposed to cadium dust or fume relies on the determination of whole blood Cd, urinary Cd and the detection of an excess of low relative molecular mass proteins in urine. These tests when positive indicate that a considerable body burden of Cd has already accumulated. The pathological changes induced are not then reversible, even after removal from further Cd exposure.2 Cadmium in whole blood is present in unexposed people3 at a concentration of ca.1 yg 1-1, cigarette smokers have up to 3 yg 1-1 and a proposed threshold4 for industrial workers is 10 pg 1-1. Studies of the distribution of Cd in blood show that about 90% is in the red cells and less than 10% is bound to plasma proteins. A metabolic model for Cd suggests, however, that the plasma Cd fraction is important in the distribution and uptake of the metal, especially in the transport of Cd from the liver to the kidney. Hitherto, measurement of plasma Cd has not been feasible. In unex- posed people the concentration is very low, probably less than 0.1 yg 1-1, but during environmental or occupational exposure when the whole blood Cd level is 10 yg 1-1 or greater, the plasma Cd level may be above 0.5 yg 1-1.During investi- gations of the effects of chronic Cd toxicity in the rat we have observed that the appearance in blood plasma of detectable amounts of Cd and of metallothionein coincide with early pathological changes in the kidney that may still be revers- ible.5 Metallothionein is measured by a radioimmunoassay6 that determines the protein but does not differentiate between the various metal-containing species (e.g. , between Zn - metallothionein and Cu - metallothionein). In this paper we * Present address, Robens Institute, University of Surrey, Guildford, Surrey, UK. describe the development of a graphite furnace atomic absorption spectrometric (GFAAS) method for the determi- nation of Cd in plasma using commercially available equip- ment.By measuring both plasma Cd and plasma metallo- thionein in samples obtained sequentially from rats dosed with Cd, we were able to show that the major Cd species in plasma is Cd - metallothionein, and that this is toxic to the kidney.5 If a similar mechanism occurs in humans, then the determi- nation of plasma Cd together with plasma metallothionein could be used in occupational medicine to provide an early indication of Cd toxicity before irreversible changes have occurred. Experimental Apparatus A Perkin-Elmer Model 2280 atomic absorption spectrometer was used, combined with a Perkin-Elmer HGA-500 graphite furnace equipped with a Perkin-Elmer AS-1 autosampler and a Perkin-Elmer Model 56 chart recorder.Argon was used as the purge gas. Measurements from the readout systems were manually entered into a DEC Professional 350 computer for prep- aration of calibration graphs, calculation of cadmium concen- trations and statistical analyses. Preparation of Samples and Standards Whole blood (5-10 ml) was collected from killed rats by abdominal aorta puncture in plastic tubes with heparin as anticoagulant. Plasma was obtained by centrifugation and transferred into plastic containers for storage at 4-10°C. All steps of this procedure were checked for possible Cd contamination. Plasma was deproteinised with equal volumes of acid (10% V/V Aristar nitric acid), in acid-washed plastic tubes. Samples were then allowed to stand €or 15 min at 4°C prior to centrifugation. (Refrigeration prior to centrifugation was found to give clearer supernatants.) Standards were prepared370 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 Table 1. Instrumental conditions Stage Temperature/"C Ramp/s Hold/s 120 1 20 Dry . . . . . . . . Ash . . . . . . . . 400 1 20 Atomise . . . . . . 1400 1 10GS,t= 1Os* Clean . . . . . . 2750 1 5 Injection volume . . . . . . . . 20 p1 Lamp current . . . . . . . . . . 4 mA Wavelength . . . . . . . . . . 228.88 nm Band pass . . . . . . . . . . . . 0.7 nm Operating mode . . . . . . . . . . Peak area Deuterium background correction Standard graphite tube * GS = gas stop; t = integration period. 0.3 rw Peak + area y v) a, C m $ 0.1 2 1.2 a, C m + a 0.1 2 I I I I I w 150 250 350 450 550 650 750 0- I TemperaturePC Fig.1. Effect of increasing ashing temperature on the peak-height and peak-area signals for Cd (3 pg 1-I) in deproteinised rat plasma (5% V/V nitric acid) that contained between 0 and 3 yg 1-1 of Cd in 5% V/V nitric acid in plastic containers. Any samples found to have Cd concentrations greater than 3 pg 1-1 were diluted with 5% V/V nitric acid and re-analysed. Analytical Procedure Dilution of plasma with either water or dilute Triton X-100 solution (0.1%) was ineffective in reducing the non-atomic background absorption to a level (<0.7 A) that could be compensated for by the deuterium arc background correction system. An alternative was to remove the major portion of the organic matrix, which caused smoke during the atomisation stage, before injection of the sample into the furnace.This was achieved by deproteinisation with acid as described above. Nitric acid also acts as a matrix modifier by reducing halide interferences. It is preferred to phosphate matrix modifiers because it can be obtained in a pure form with a low level of Cd contamination. Results and Discussion The optimum instrumental conditions are shown in Table 1. An optimum atomisation temperature was found for peak-height (1900 "C) and peak-area (1400 "C) measurements using Cd standards prepared in 5% V/V nitric acid. Peak- height measurements were obtained from the chart recorder whereas peak-area measurements were obtained from the digital display of the spectrometer. These values were used to optimise the ashing temperature.Fig. 1 shows an ashing profile for deproteinised plasma. 0.3 vl 73 s 8 0.2 a, C (0 + s 0.1 2 <\ Peak height I I I 0 1 5 10 HN03, Yo 3.2 a, f m 0.1 e s n 4 0 Fig. 2. Effect of increasing nitric acid concentration on the sensitivity of the measurement of Cd (3 pg 1-1) in pooled rat plasma samples Peak heighl n 0 5 10 HN03, "10 Fig. 3. Effect of increasing the concentration of nitric acid on the precision of measurement of Cd (3 pg 1-1) in pooled rat plasma samples The purpose of the ashing stage is to remove matrix constituents that may cause interference effects. Using the peak-area mode, temperatures of 450-550 "C can be utilised without losses of Cd. These temperatures are too low to have any effect on the removal of inorganic interferents, so an ashing temperature of 400°C was chosen as giving the best precision [0.97% relative standard deviation (RSD)].The optimum acid concentration is the lowest acid concen- tration that gives the best reproducibility and the lowest background signal. A low acid concentration is also preferred in order to increase graphite tube lifetimes. The effect of increasing nitric acid concentration on sensitivity is shown in Fig. 2. The rapid decrease in sensitivity for peak-height measurements is believed to be due to spreading effects, as nitric acid has a low surface tension and contributes to increased spreading as the droplet dries out during the drying stage and the residual nitric acid is effectively concentrated. Spreading results in lower peak- height values becaus? there is a temperature gradient along the graphite tube, which results in Cd being atomised at different times.Peak-area measurements are not affected so much, provided that the signal produced remains within the integration period, Fig. 3 shows that the optimum acid concentration on the basis of precision is a final nitric acid concentration of 5% V/V. The background absorbance was measured using the cadmium non-absorbing line of 226.5 nm (Fig. 4). A sharp decline in non-atomic background absorbance is found with increasing nitric acid concentration up to 5% V/V. At this concentration the background signal is reduced to a level which is well within the capability of the deuterium arc background correctionJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 37 1 v) U C s 4 - v) ; 3 - m e $ 2 - a a 1 - - 0 Peak height - Peakarea a, C (0 f 0.5 2 Q I ' ' 0 O ' il 5 10 HN03, O h Fig. 4. Effect of increasing acid concentration on the non-atomic background absorbance observed during measurement of Cd (3 pg I - l ) in pooled rat plasma samples (Cd line at 226.5 nm) 0.4 v) 0.3 U s a) 0.2 f 2 0.1 a cn 0 C m 0 B 0 1 2 3 Cdipg I-' Fig. 5. Comparison of the gradients of calibration graphs for (A) Cd added to deproteinised plasma (standard additions), (B) Cd standards in water and (C) Cd standards in dilute acid (5% V/V nitric acid) system. The accuracy of the plasma cadmium method was determined by standard additions plots, recovery experiments and the use of inter-laboratory comparisons.No certified quality control samples were available. As nitric acid causes a depression in the absorbance signal for Cd, standards were prepared in acid of the same concentration as used for the samples. Using the peak-area mode, the gradients of the standard additions plots were found to be similar to those obtained using standards prepared in 5% V/Vnitric acid (Fig. 5 ) . These were equal within the limits of imprecision of the measurement. With peak-height measurements there was a poorer corre- lation between the gradients of standard additions plots and those for standards in 5% V/V nitric acid. Recovery experiments were performed on pooled plasma, which was placed in acid-washed containers and spiked to give Cd concentrations of 0.9, 1.8 and 2.5 pg 1-1.Each sample was deproteinised and analysed in duplicate six times (Table 2). Recoveries of added Cd were excellent. To examine further the accuracy of the procedure, a comparison was made between the two laboratories involved in this work, using three different techniques for the determination of plasma Cd in five samples (Table 3) using a pool of rat plasma, obtained as described, from rats dosed with Cd in drinking water. Flame atomic fluorescence spectrometry (AFS) involved deproteinisation of the plasma with nitric acid, but required only aqueous Cd standards for preparing calibration graphs. It Table 2. Recovery experiments on cadmium in spiked plasma Cadmium Concentration added/ obtained/ Recovery Pg I - ' pg I-' ( n = 6) kSD, O/o 0 0.22.5 k 0.042 - 0.9 1.15 +_ 0.046 102.7 rt 5.1 99.7 rt 1.8 1.8 2.02 kO.034 102.0 * 1.9 2.5 2.73 k 0.049 Table 3.Inter-laboratory comparison of different techniques for the determination of cadmium in blood plasma Deprotenisation + Sample No. tube wall AAS AFS Probe AAS 1 2.9 3.4 3.8 2 1.7 2.2 4.0 3 1.5 2.0 2.2 4 3.8 3.2 3.8 5 2.6 2.8 3.5 is believed that this technique is an accurate and interference- free procedure.7 The third method used was probe atomisa- tion atomic absorption spectrometry.8 For this technique plasma was diluted 5-fold with water and 20 pl of the solution were injected on to the probe and then dried. During the ashing stage 2 p1 of 50% Aristar nitric acid were added. The probe was then removed from the furnace while the furnace was raised to isothermal conditions.8 After 10 s, the probe was re-inserted in the furnace.Calibration was effected with aqueous standards and signals were obtained as peak-height measurements on a chart recorder. Overall, the GFAAS method with deproteinisation gave slightly lower values than the AFS technique. The results obtained by the experimental probe atomisation method are higher than those given by other techniques, but most values were within k0.5 yg 1-1 of each other, which is reasonable at such low concentrations. Using the GFAAS method with deproteinisation the calibration graph was linear for Cd levels up to 3 pg 1-1. The Cd concentration giving an absorbance of 0.0044 A is 0.048 P8 I-' (0.96 pg). A detection limit of 0.063 yg 1-l (1.26 pg) of Cd was calculated from seven replicate analyses of plasma samples, using the convention of taking twice the standard deviation of the signal variation, using plasma samples with a Cd concentration of around 1.0 pg 1-1.The performance of the method was checked on a day-to-day basis by two methods: (1) using a pool of plasma as a control and (2) duplicate analyses of samples. Pooled plasma was prepared by spiking 100 ml of plasma with cadmium to give a concentration between 1 and 2 yg 1-1. This plasma was stored at 4 "C with 1% sodium azide as an antibacterial agent. Measurements of the cadmium concentration in this pool were 1.76 k 0.04 pg 1-1. Results deviating more than &lo% from this value required samples within a batch to be repeated. Duplicate samples were also compared from batch to batch as a check on the stability of the control plasma. The determination of plasma Cd could offer a more sensitive means of detecting impending Cd toxicity than the present laboratory procedures.This possibility should be investigated in workers known to be exposed to Cd and being carefully monitored. Initially, parallel measurement of plasma metallothionein would also be desirable to confirm the presence of Cd - metallothionein and to exclude the possibility of Cd contamination of the plasma samples prior to analysis.372 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 One of us (M. M. B.) was in receipt of a grant funded by the 3. 4. 5. Greater Glasgow Health Board (Unit 1 East). The authors acknowledge the assistance of P. R. Stahpit and S. K. Giri in carrying out Cd determinations by atomic fluorescence and probe atomic absorption methods. References 6. 7 . 1. 2. Friberg, L., Elinder, C.-G.. Kjellstrom, T., and Nordberg, 8. G. F., Editors, “Cadmium and Health: A Toxicological and Epidemiological Appraisal, Volume I, Exposure, Dose and Metabolism,” CRC Press, Boca Raton, FL, 1985. Roels, H., Djubang, J., Buchet, J.-P., Bernard, A., and Lauwerys, R., Scand. J . Work Environ., 1982, 8, 191. Mcintosh, M. J . , Moore, M. R., Goldberg, A , , Fell, G. S . , Halls, D. J., and Cunningham, C., Ecol. Dis., 1982, 1, 177. Rogenfelt, A . , Elinder, C. G., and Jarup, L. Znt. Arch. Occup. Environ. Health, 1984, 55, 43. Aughey, E., Fell, G. S . , Scott, R., and Black, M., Enuiron. Health Perspect., 1984, 54, 153. Mehra, R. K., and Bremner, I., Biochem. J . , 1983, 213,459. Michel, R. G., Hall, M. L., Ottaway, J. M., and Fell, G. S . , Analyst, 1979, 104, 491. Littlejohn, D., Cook, S., Durie, D., and Ottaway, J . M., Spectrochim. Acta, Part B , 1984, 39, 295. Paper 5611 7 Received March 12th, 1986 Accepted April 24th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100369

出版商:RSC    

年代:1986

数据来源: RSC