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