Evaluation of several commercially available spray chambers for use in inductively coupled plasma atomic emission spectrometry Salvador Maestre, Juan Mora, Jose�-Luis Todolý� and Antonio Canals* Departamento de Quý�mica Analý�tica, Universidad de Alicante, 03071 Alicante, Spain Received 19th August 1998, Accepted 5th November 1998 Four diVerent spray chambers were compared for the elemental analysis of liquid samples by ICP-AES: a doublepass Scott-type spray chamber made from Ryton and three cyclonic spray chambers manufactured from various materials [i.e., glass, poly(propylene), PP, and poly(tetrafluoroethylene), PTFE].A glass concentric pneumatic nebulizer was used in conjunction with all four chambers. The parameters evaluated were: the characteristics of the aerosols at the exit of each chamber (i.e., tertiary aerosols); the solvent (Stot) and analyte (Wtot) transported through each chamber; and the ICP-AES analytical parameters (i.e., net emission intensities, limits of detection, LOD, and background equivalent concentrations, BEC).The interference produced by the presence of a widely used matrix (i.e., acids) was also evaluated for the four chambers. The results indicated that the cyclonic glass and PP spray chambers gave rise to coarser tertiary aerosols, higher solution transport rates, higher emission signals and lower LOD and BEC values than the other two spray chambers. For the cyclonic spray chambers, the position of the nebulizer proved to be of critical importance.With regard to the acid eVects, these were more pronounced as the tertiary aerosols became finer. Hence, for the Scott-type spray chamber, the signal reduction induced by the presence of acids was enhanced compared with the cyclonic spray chambers. The introduction of the sample is a critical step in atomic as for example the vortex type,3,8–10,14 Sturman Masters type15 and vertical rotary;16,17 and (iii) the single-pass or cylindrical spectrometric analysis.1,2 The characteristics of the analytical type8,21–23 or direct spray chamber.8 signal finally obtained depend, to a great extent, on the quality Among the designs mentioned above, the most widely used (i.e., eYciency and reproducibility) of the sample introduction in ICP-AES are the double-pass (Scott-type) and the cyclonic system.Usually, in inductively coupled plasma atomic emission designs.On comparing both designs, it has been reported that spectrometry (ICP-AES), the sample is introduced as a liquid the former gives rise to lower analyte transport eYciencies and solution.In most cases, the main components of the higher memory eVects than the latter.1,4,7,11,16,24–26 liquid sample introduction system are: (i) a nebulizer, that Another aspect to take into account is the matrix eVect (i.e., spreads out the liquid bulk generating an aerosol; (ii) a spray the change in performance induced by the main sample chamber, that filters the aerosol and selects the maximum components). Acids are among the most commonly present droplet size entering the plasma; and (iii) an injector tube, matrices.Several workers have reported that acids cause an used to introduce the aerosol into the plasma. ICP-AES signal depression with respect to water,27–38 although The spray chamber is a component of major interest.3,4 It at low acid concentration a signal enhancement has been is recognized that, when less of 5% of the analyte nebulized is observed.27 All these eVects are complex and diVerent reasons transported to the atomizer, the spray chamber, rather than have been proposed in order to explain them, such as modifi- the nebulizer (i.e., pneumatic nebulizers), determines the cations in the nebulization process28,32 due to a change in the characteristics (fineness and amount) of the aerosol injected solution viscosity and/or surface tension; changes in the aerosol into the plasma.4 The processes taking place within the spray transport;27,29,32 and deterioration in the plasma thermal chamber are many and complex in nature.Hence, the aerosol characteristics.30–32 In addition, acids induce increases in suVers from solvent evaporation, agglomeration, impact losses memory eVects through ‘transient’ mechanisms.39,40 Therefore, due to turbulence and/or inertia.5 Also, the aerosol thermal it is obvious that the spray chamber plays a very important and charge equilibria are established and the turbulence associrole also from the point of view of matrix eVects, as has been ated with the nebulization process is reduced inside the spray recently anticipated.37 chamber.The aim of this work was, thus, to evaluate the behavior of Some characteristics of an ideal spray chamber are: (i) it several spray chambers in ICP-AES. In this study, a Scott- should be able to transport as much analyte mass to the type spray chamber and three cyclonic spray chambers built atomization cell as possible without deteriorating its excitation from diVerent materials [i.e., glass, poly(propylene) and poly- properties; (ii) the aerosol entering the plasma should be as (tetrafluoroethylene)] were used.A systematic investigation of fine as possible, and less turbulent than the aerosol generated the characteristics of the aerosol exiting each chamber (i.e., in the nebulization step. As a consequence, the analyte will be tertiary) the transport of solution, analytical parameters and more eYciently excited; (iii) the spray chamber should be matrix (acid) eVects was carried out.robust (i.e., minimum variation of its performance with samples of very diVerent nature and composition); (iv) the memory eVects should be minimal in order to increase the Experimental sample throughput; and (v) it should have mechanical simplicity and low production cost. A pneumatic concentric nebulizer (Model AR-35–07-C2, Glass To this end, several spray chamber designs have been Expansion, Australia) was used throughout.Its gas outlet proposed in the literature:6–26 (i) the double-pass (so-called cross-section area was 3.1×10-2 mm2, giving a back-pressure of 19 psig for a 0.8 l min-1 gas flow rate. The aerosol was, Scott-type) or reverse flow type;6–9 (ii) the cyclone3,8–10,14–20 J. Anal. At. Spectrom., 1999, 14, 61–67 61Table 1 Plasma instrumental conditions Rf power/kW 1.3 Integration time/ms 100 Sampling time/ms 1000 Outer gas flow rate/l min-1 15 Intermediate gas flow rate/l min-1 0.5 Central gas flow rate/l min-1 Variable Viewing height above load coil/mm From 10 to 12 mm, optimized in each case Torch Fassel-type Injector id/mm 2 Sample uptake rate/ml min-1 Variable Table 2 lists the element wavelengths as well as the ionic line Fig. 1 Schematic diagram and critical dimensions of the cyclonic energy sum values, Esum (i.e., sum of ionization and excitation spray chambers used.The parameters A, B, C, D and E are expressed energies). Solutions of two acids (i.e., nitric and sulfuric) at in mm, while a is given in degrees. concentrations ranging from 0.4 to 3.0 mol l-1 were also prepared. therefore, generated by the exposure of the liquid sample to a high-velocity gas stream. The four spray chambers used were: Results and discussion a Ryton Scott-type (Perkin-Elmer, U� berlingen, Germany) with an inner volume of 100 cm3 and three cyclonic spray chambers Spray chamber characterization with water made from diVerent materials: glass, poly(propylene) (PP), First, the performance of the four spray chambers was and poly(tetrafluoroethylene) (PTFE) (Glass Expansion); compared with water solutions. To this end, the characteristics their inner volumes were 47.5, 62 and 57.3 cm3, respectively.of the tertiary aerosols, amount of solution transported and Fig. 1 shows an outline of the cyclonic spray chamber design analytical figures of merit in ICP-AES were evaluated.as well as the critical dimensions of the three cyclonic spray chambers employed. The nebulizer was positioned tangentially Aerosol characterization. Fig. 2 shows the variation of the to the wall of the cyclonic spray chamber. Sauter mean diameter (D3,2) of the tertiary aerosols versus the The liquid sample was supplied to the by means nebulizing gas, Qg [Fig. 2(a)] and liquid, Ql [Fig. 2(b)] flow of a Gilson Minipuls 3 peristaltic pump ( Villiers-Le-Bel, rates for water and the four spray chambers evaluated.As can France). The gas flow, in turn, was varied by means of a mass be derived from Fig. 2(a), an increase in the gas flow rate led flow controller, Model FC260 (Tylan, Torrance, CA, USA). to finer tertiary aerosols for the four chambers tested. This Argon was employed as the nebulizer gas in all cases. might be because the nebulizer generated finer primary aerosols Drop size distributions (DSD) of the aerosols generated by on increasing Qg, since the amount of available kinetic energy the nebulizer (primary) and those exiting the spray chamber of the gas stream became higher.5,43,44 Hence, at Ql= (tertiary) were characterized by means of a Model 2600c 0.6 ml min-1, D3,2 values of the primary aerosols were 12.8 Fraunhofer laser diVraction system (Malvern Instruments, and 5.8 mm on switching from Qg=0.4 to 0.8 l min-1. Malvern, Worcestershire, UK) equipped with a lens of 63 mm In addition to the mean diameters of the DSD, the laser focal length that allowed for the measurement of drop diamdi Vraction instrument provides some parameters related to the eters ranging from 1.2 to 118 mm.The software employed was aerosol liquid volume contained in the sampling zone. In this the B.OD version. A model-independent algorithm was used work, the volume concentration (VC) was taken as an indi- to calculate the whole DSD from the energy data.The primary cator of this aerosol property. The VC is referred to the aerosols were measured at 10 mm from the nebulizer tip, percentage of the laser measurement volume that is occupied whereas, for the tertiary aerosols, the spray chamber exit was by droplets.45 The results indicated that VC for tertiary placed at 5 mm from the laser beam. A set of three replicates aerosols increased with Qg. Thus, for instance, for the double- was performed in each case; the precision (RSD) reached was pass spray chamber the VC values were 0.009 and 0.027% always lower than 2%. when Qg was raised from 0.4 to 0.8 l min-1.This fact was due The solvent transport rate (Stot) was measured, only for to both the generation of finer primary aerosols and the water, by means of a direct method (i.e., by the adsorption of increase in the carrying capability of the gas stream.21,44 the aerosol in a U-tube filled with silica gel for 10 min).40,41 On increasing Ql from 0.4 to 0.8 ml min-1, at Qg= By weighing the tube before and after aerosol exposure, the 0.5 l min-1, a slight growth in the D3,2 of the primary aerosol Stot values were easily derived.Analyte transport rate values were obtained by a direct method (i.e., by collecting the aerosol on a glass-fiber filter, Type A/E, 47 mm diameter, 0.3 mm pore Table 2 Elements and lines used, wavelengths and energy sum values size, Gelman Sciences, Ann Arbor, MI, USA).5,41,42 A Element Wavelength/nm Esum/eVa 500 mg ml-1 Mn solution was nebulized.The Mn retained after 10 min was extracted by washing the filters with 1.0% Al I 396.152 — (v/v) hot nitric acid solution. The total solution volume was Ba II 455.403 7.93 made up to 100 ml in a calibrated flask. Finally, the Mn Cd II 214.438 14.77 concentration in each solution was determined by flame atomic Cr II 205.560 12.80 absorption spectrometry (FAAS). The precision of transport Cu I 324.754 — Mg II 280.270 12.25 experiments was always lower than 5% (RSD from three Mg I 285.213 — replicates).Mn II 257.610 12.25 Emission signals were measured with a Perkin-Elmer Optima Ni II 221.647 14.27 3000 ICP-AES instrument. Table 1 summarizes the plasma Sr II 407.771 8.73 instrumental conditions. Samples containing ten elements Zn I 213.856 — (1 mg ml-1 each) were prepared in water from an ICP multi- aEsum (only for ionic lines)=ionization energy+excitation energy. element standard solution (IV, Merck, Darmstadt, Germany). 62 J. Anal. At. Spectrom., 1999, 14, 61–67results. Therefore, neither the inner volume of the chamber nor its geometry seems to be responsible for the observed behavior. A more complete study on this subject is necessary. A factor that has proved to be critical is the relative position of the nebulizer–spray chamber (i.e., the distance between the nebulizer tip and the wall of the cyclonic spray chamber). Each cyclonic spray chamber accepted a given (fixed) position for the nebulizer and the latter could not be introduced beyond this point (Fig. 1). However, a conspicuous variation in the tertiary aerosol characteristics was observed as a function of the nebulizer position. In this way, with the PTFE cyclonic spray chamber, the D3,2 values of the tertiary aerosols varied from 2.3 to 2.9 mm when the nebulizer was moved backward around 5 mm from its initial (fixed) position. As regards the VC, this parameter changed from 0.025 to 0.034% for the reported situations.These results clearly demonstrate that the extent of the aerosol transport processes ( likely impaction losses) was modified on changing the nebulizer position. According to the VC data, the established nebulizer position was not the optimum and more liquid solution was allowed to leave the spray chamber on bringing the nebulizer out by 5 mm. Therefore, it seems that the nebulizer position is an important factor that should be optimized for cyclonic spray chambers.In summary, the nebulizer position for each chamber might partially explain the diVerences found for the three cyclonic spray chambers (Fig. 1). The spray chamber material might play a role in the aerosol transport and filtering process, but, at this stage, this was not verified. Furthermore, small diVerences in some critical dimensions (i.e., angle of the Fig. 2 EVect of the gas flow rate, Qg (a), Ql=0.6 ml min-1, and the top and/or bottom of the spray chamber, inner diameter or liquid flow rate, Ql (b), Qg=0.5 l min-1, on the Sauter mean diameter height) might induce diVerences in the behavior of these (D3,2) of tertiary aerosols for the spray chambers studied. (A) cyclonic chambers (Fig. 1). glass, (B) cyclonic PP, (C) cyclonic PTFE and (D) double-pass. Transport parameters. Fig. 3 shows the variation of the solvent transport rate (Stot) with Qg [Fig. 3(a)] and Ql from 9.7 to 10.5 mm was observed. This fact was due to the drop in the gas kinetic energy to liquid (sample) mass ratio as [Fig. 3(b)] for the four spray chambers tested. As expected from the VC data, Stot increased on increasing both variables, Ql was raised.44,46,47 In spite of this, D3,2 values of tertiary aerosols did not show noticeable variations with Ql, irrespec- its variation with Qg being more pronounced than with the liquid flow rate. Again, on comparing the diVerent spray tive of the spray chamber employed [Fig. 2(b)].Therefore, from Fig. 2(b) it seems that each spray chamber imposed its chambers, a correlation between Stot and VC was found (i.e., Stot followed the order: Scott#PTFE<PP<glass). own cut-oV diameter, masking the variations produced in the primary aerosols.4 In contrast, the VC values slightly increased, The results concerning the analyte transport rate (Wtot) confirmed the variations observed in Stot (i.e., the analyte although the magnitude of the increase was lower than for Qg.For the double-pass spray chamber, at Qg=0.5 l min-1, transport rate became higher on increasing both gas and liquid flow rates, irrespective of the spray chamber used). Hence, for VC was 0.012 and 0.019% for Ql=0.4 and 0.8 ml min-1, respectively. The slight increase in VC values might be due to the Scott-type spray chamber at Ql=0.6 ml min-1, Wtot was 0.04 and 0.18 mg s-1 when Qg was switched from 0.4 to the intensification of solution losses due to the generation of coarser primary aerosols, that counterbalanced, at least in 0.8 l min-1.The increase in the analyte transported was less pronounced when the liquid flow rate was raised. Thus, at part, the sample uptake rate increase. From the results discussed above, it can be anticipated that Qg=0.5 l min-1 for the Scott-type spray chamber, Wtot was 0.06 and 0.07 mg s-1 for Ql=0.4 and 0.8 ml min-1, an increase in both Qg and Ql should produce an increase in the amount of solution exiting the spray chambers.respectively. As regards the spray chamber used, at Qg=0.5 l min-1 and With regard to the spray chamber design and material, Fig. 2 reveals that the tertiary aerosols obtained with the Ql=0.6 ml min-1, Wtot was 0.15, 0.11, 0.05 and 0.06 mg s-1 for the glass, PP, PTFE and Scott-type spray chambers, double-pass Scott-type spray chamber had lower D3,2 values than those generated by the glass and PP cyclonic spray respectively, which agreed with the results obtained for Stot (Fig. 3). As was observed for tertiary aerosols,Wtot also varied chambers but slightly higher values than those for the PTFE cyclonic spray chamber. Among the cyclonic spray chambers, on changing the nebulizer position. Hence, for the PTFE chamber, at Qg=0.5 l min-1 and Ql=0.6 ml min-1, Wtot D3,2 values followed the order: glass>PP>PTFE. The VC values, in turn, varied as follows: VC(Scott)# values of 0.06 and 0.07 mg s-1 were obtained on moving the nebulizer from the fixed position to 5 mm behind this point VC(PTFE)<VC(PP)<VC(glass).The reasons that might be argued to explain the diVerent behavior are based on changes (Fig. 1). in the inner volume and geometry of each spray chamber.9,16,22,23 Nevertheless, for the PP and PTFE chambers Analytical figures of merit. Fig. 4 shows the variation of the net emission signal of Mn when the gas flow [Fig. 4(a)] and the D3,2 values were very diVerent, whereas their inner volumes were similar.In addition, the inner volume of the PTFE sample uptake rate [Fig. 4(b)] were increased. From Fig. 4(a) it can be seen that the emission signal increased with Qg up chamber was smaller than that of the Scott-type chamber by a factor of 0.56, whereas both chambers gave rise to similar to a value of 0.5 l min-1. Further increases in the gas flow J. Anal. At. Spectrom., 1999, 14, 61–67 63resulted in a sharp signal decrease. This last fact was explained by the reduction in the analyte residence time inside the plasma as well as a deterioration in its thermal characteristics.48–50 On increasing the sample uptake rate a slight growth in emission signal was obtained [Fig. 4(b)]. These results were predictable from the variation in VC and transport parameters. As can be seen in Fig. 4, the cyclonic spray chamber made from glass aVorded higher emission signals than the remaining spray chambers. An interesting point is that, under the optimum conditions of Fig. 4(a), the maximum signal increase factor for the glass cyclonic spray chamber with respect to the Scott-type chamber (e.g., 2.3) was lower than the corresponding Wtot (e.g., around 3). This eVect could be accounted for by a reduction in the plasma thermal capability. In order to evaluate whether the plasma was deteriorated, the Mg II-to- Mg I (see Table 2) emission intensity ratio was measured. This parameter has been considered a good indicator of the plasma excitation properties.49–51 Actually, this ratio was lower for the glass cyclonic spray chamber than for the double-pass spray chamber.Hence, at Qg=0.5 l min-1 and Ql= 0.6 ml min-1, this ratio was 6.3 and 8.8 for the glass cyclonic and Scott-type spray chambers, respectively. When a sample introduction system is evaluated, another interesting analytical parameter is the signal stability. In this study, the so-called short-term stability was evaluated by means of the RSD of 20 replicates of the ICP-AES emission intensity.The results indicated that the RSD values were, in general, higher for the Scott-type spray chamber than for the cyclonic spray chambers. Hence, at Qg=0.5 l min-1 and Ql= 0.6 ml min-1, this parameter was 1.7, 1.1, 1.7 and 2.9% for the glass, PP, PTFE and Scott-type chambers, respectively. Fig. 3 EVect of the gas flow rate, Qg (a), Ql=0.6 ml min-1, and the Note that these values represent the mean of the RSD for the liquid flow rate, Ql (b), Qg=0.5 l min-1, on the solvent mass transport rate (Stot) for the spray chambers studied.(A) Cyclonic glass, (B) ten elements evaluated (Table 2). These results indicated that, cyclonic PP, (C) cyclonic PTFE and (D) double-pass. in relative terms, cyclonic spray chambers exhibited more stable emission signals than the Scott-type chamber. Fig. 5 summarizes the limits of detection (LOD) obtained for the diVerent spray chambers and elements studied. In this case, LOD values were calculated according to the 3sB criterion, where sB is the standard deviation of 20 replicates of the background.As can be observed, the glass cyclonic chamber gave the lowest LOD values for all the elements tested and, depending on the element, the PTFE cyclonic or Scott-type chamber aVorded the highest LOD values. These data mainly reflected the behavior of the emission signal. Table 3 lists the background equivalent concentration (BEC) values for the diVerent elements and spray chambers tested.It is interesting that, as happened with the LOD values (Fig. 5), the cyclonic spray chamber manufactured from glass gave rise to the lowest BEC values, whereas the Scott-type and the PTFE chambers provided the highest BEC values. Fig. 4 EVect of the gas flow rate, Qg (a), Ql=0.6 ml min-1, and the liquid flow rate, Ql (b), Qg=0.5 l min-1, on the net emission signal Fig. 5 Limits of detection (LOD) for the elements studied with all the for a 1 mg ml-1 Mn solution.(A) Cyclonic glass, (B) cyclonic PP, (C) cyclonic PTFE and (D) double-pass spray chamber. spray chambers evaluated. Qg=0.5 l min-1; Ql=0.6 ml min-1. 64 J. Anal. At. Spectrom., 1999, 14, 61–67Table 3 Background equivalent concentration (BEC) values for the elements and spray chambers tested BEC/mg ml-1 Element Glass PP PTFE Scott Al 0.83 1.56 3.10 2.30 Ba 0.02 0.04 0.07 0.05 Cd 0.09 0.11 0.21 0.19 Cr 0.11 0.16 0.30 0.24 Cu 0.17 0.27 0.52 0.43 Mg 0.01 0.02 0.03 0.02 Mn 0.03 0.05 0.08 0.07 Ni 0.22 0.31 0.60 0.47 Sr Saturation 0.02 0.03 0.02 Zn 0.040 0.06 0.10 0.10 The wash-out times (i.e., the time spent by the system to achieve 1% of the steady-state signal value) were measured for a 1mg ml-1 multi-elemental aqueous solution.The results indicated that the wash-out times were, in general, lower for the cyclonic than for the Scott-type chamber. As a consequence, at Qg=0.5 l min-1 and Ql=0.6 ml min-1, 20, 24, 22 and 30 s were required to diminish the signal to 1% of its initial value for the glass, PP, PTFE and double-pass spray chambers, respectively.This eVect has been discussed previously by other workers and can be explained in terms of the smaller inner volume and simpler geometry of the aerosol path for the cyclonic spray chambers.4,7,16,25 The study of the memory eVects was carried out with Pd. This element seems to be preferentially adsorbed on some Fig. 6 EVect of acid concentration on Sauter mean diameter of the polymer surfaces. It was found that the wash-out times were tertiary aerosol (D3,2).(A) Cyclonic glass, (B) cyclonic PP, (C) 26, 36, 38 and 44 s for the glass, PP, PTFE and Scott-type cyclonic PTFE and (D) double-pass spray chamber. (a) Nitric acid; spray chambers, respectively. These results showed an import- (b) sulfuric acid. Qg=0.5 l min-1; Ql=0.6 ml min-1. ant diVerence based on the spray chamber material; the lowest memory eVect was observed for the glass spray chamber, the The trends in D3,2 values of the tertiary aerosol versus acid memory eVect being more severe for the polymer-based spray concentration were the same irrespective of the spray chamber chambers.Other workers found Pd adsorption on the inner material and design and acid type used. Among the two acids walls of polystyrene digestion vessels.52 Therefore, from the studied, sulfuric acid generated the finer tertiary aerosols. The point of view of memory eVects, it seems that glass is the best Sauter mean diameters of the tertiary aerosols obtained with choice as the spray chamber material.the diVerent spray chambers followed the same order as those reported for pure water in Fig. 2. Spray chamber characterization with acid solutions The VC values of the tertiary aerosols increased with acid concentration. This growth was more pronounced for the Acids are one of the most commonly present matrices in ICP-AES analysis. Based on the results reported in the litera- Scott-type spray chamber than for the cyclonic spray chambers.Hence, with the former chamber, the VC value for water ture, there are two groups of inorganic acids:29,31,32,35,36 (i) those that interfere with the aerosol transport processes; and (0.011%) was improved by a factor of around two when 3 mol l-1 solutions of either nitric (0.022%) or sulfuric (ii) acids that, in addition, deteriorate aerosol production during the nebulization process.In this work, two acids (0.023%) acid were employed. In contrast, for the glass cyclonic spray chamber the VC values varied only slightly (0.073, 0.078 representative of each of these two groups were used (i.e., nitric and sulfuric) in solutions of diVerent concentration. and 0.074% for water, and 3 mol l-1 nitric and sulfuric acid, respectively). In summary, on increasing the acid concentration there was Aerosol characterization. Fig. 6 shows the variation of the D3,2 of the tertiary aerosols with the acid concentration for a higher (or at least the same) amount of aerosol in liquid form at the exit of the spray chamber and it was contained in nitric acid [Fig. 6(a)] and sulfuric acid [Fig. 6(b)]. The corresponding points for pure water are also shown. From Fig. 6 smaller droplets. These findings agreed well with those previously reported in the literature.32,35,36 it can be seen that the presence of acids in solution led to a decrease in the D3,2 values relative to pure water, the reduction being more important on switching from pure water to the Transport parameters.It was diYcult to measure the solvent transport rate by means of a continuous method with concen- 0.4 mol l-1 concentration. According to the data shown in Fig. 6, further acid concentration increases did not produce trated acid solutions, since a leak of acid (white vapors) was observed at the exit of the U-shaped tube even when two tubes significant variations in the D3,2 values.In contrast, no appreciable variations were found in the primary aerosol D3,2 values. were placed in series at the exit of the spray chamber. For this reason, some experiments were performed by applying an At Qg=0.5 l min-1 and Ql=0.6 ml min-1, the primary aerosol D3,2 values were 10.9, 10.1 and 10.9 mm for pure water and indirect method.53,54 Both solution and drain containers were continuously weighed. Then, the mass loss was plotted versus 3 mol l-1 nitric and sulfuric acid, respectively.All these results revealed that, as it was stated in a previous study, there is a time, giving a straight line, the slope of which corresponded to the Stot value. From these experiments no important vari- diVerent aerosol transport mechanism for water and for acids.35 ations in the solvent transport rate were observed between J. Anal. At. Spectrom., 1999, 14, 61–67 65pure water and acid solutions. Hence, at Qg=0.5 l min-1 and Ql=0.6 ml min-1 and with the PP cyclonic spray chamber, the relative Stot values, i.e., (Stot)acid/(Stot)water, were 0.97 and 1.09 for 3 mol l-1 nitric and sulfuric acid solutions, respectively.Note that the precision of this method was poorer than that of the direct method.40 Similar results were obtained for the other chambers. As regards the analyte transport values, the results indicated that when an acid was present the Wtot values decreased with respect to water. Nonetheless, as was mentioned previously, the aerosol liquid fraction at the exit of the spray chamber became greater as the acid concentration increased and Stot did not vary significantly.These results support the existence of the so-called aerosol ionic redistribution (AIR) phenomenon. 35,55,56 As has been demonstrated,35,56 this eVect contributes to a decrease in the analyte concentration in aerosol droplets exiting the spray chamber when ionic species are present. On comparing the diVerent spray chambers employed, it was observed that, for both nitric and sulfuric acid, the reduction in Wtot was, in general, less pronounced for the cyclonic spray chambers than for the double-pass Scott-type spray chamber.Hence, for nitric acid at Qg=0.5 l min-1 and Ql=0.6 ml min-1, the ratios (Wtot)nitric/(Wtot)water were 0.95 and 0.87 for the glass cyclonic and Scott-type spray chambers, respectively. The ratios for sulfuric acid were 0.80 and 0.70, respectively.Note that 3 mol l-1 acid solutions were employed. This result indicates that acid eVects are expected to be less severe for the cyclonic spray chamber than for the Scott-type spray chamber. Fig. 7 EVect of acid concentration on the relative net emission intensity, defined as the mean of the ratio Iwith acid/Iwithout acid, obtained Emission signal. Fig. 7 shows the variation of the relative for the ionic lines; 1 mg ml-1 solution concentration. (A) Cyclonic net emission intensity (Irel) versus the acid concentration for glass, (B) cyclonic PP, (C) cyclonic PTFE and (D) double-pass spray the four spray chambers evaluated and nitric [Fig. 7(a)] and chamber. (a) Nitric acid; (b) sulfuric acid. Qg=0.5 l min-1; Ql= 0.6 ml min-1. sulfuric [Fig. 7(b)] acid. The parameter Irel refers to the mean of the ratio Iwith acid/Iwithout acid obtained for each ionic line. In addition to the mean points, in Fig. 7 the maximum and minimum Irel values have also been represented as bars.In chambers (those that aVorded the lowest Mg II/Mg I ratios), the acid eVects were less severe than for the Scott-type spray this way, the bar width indicates the degree of variability of the acid eVect for the diVerent ionic lines tested. As can be chamber. Therefore, the acid eVects found in this study were mainly of a physical nature (i.e., aerosol transport through the derived, no significant variations in Irel were found as a function of the ionic line. spray chamber, rather than analyte excitation deterioration, was the reason for the acid interference indicated in Fig. 7). From Fig. 7, it can be seen that, as expected, the higher the acid concentration, the lower the Irel values. As was anticipated by the Wtot measurements, for a given acid, the greatest Conclusions depressive eVect (i.e., lowest Irel value) was found with both acids for the Scott-type spray chamber, whereas a lower signal Cyclonic spray chambers manufactured from glass and PP generate coarser tertiary aerosols and higher solution transport reduction was produced for the glass cyclonic spray chamber.For a given spray chamber, sulfuric acid [Fig. 7(b)] provoked rates than the double-pass and PTFE spray chambers. As a consequence, higher ICP-AES emission intensities and lower the strongest matrix depressive eVect. From the results discussed above, it seems that there is a LOD are obtained with the former two spray chambers. Cyclonic spray chambers aVord a better short-term stability correlation between the aerosol drop size and the matrix eVect caused by acids; the finer the tertiary aerosols, the lower the of the signal (in relative terms) than the Scott-type spray chamber.The wash-out times for pure water solutions are, in Irel values (i.e., the more severe the acid eVects). This observation agrees with others in which the acid eVect was studied general, lower for the cyclonic spray chambers. The chamber material can influence the spray chamber for diVerent tertiary aerosol fractions.35 Deterioration of the plasma thermal characteristics has also performance, mainly in terms of memory eVects when elements that can be adsorbed on the walls (i.e., Pd) are present. In been suggested as a reason to explain the acid eVect, mainly if the plasma is working under so-called non-robust con- this instance, polymeric materials seem to be less appropriate than glass.ditions.30,31,40,57 These latter conditions prevail at low rf power levels and high gas flow rates.In this study, no appreciable Cyclonic spray chambers are less sensitive to acid eVects than double-pass spray chambers. Among the cyclonic spray variations in the Mg II-to-Mg I emission intensity ratio with the acid concentration were observed. The robustness of the chambers, that made from glass oVers the lowest matrix eVect and that made from PTFE the greatest matrix eVect. This plasma was also confirmed by the small variation in the Irel values versus the ionic line energy sum (Table 2 and Fig. 7),58 fact, previously observed for other ionic matrices (i.e., Na),59 seems to be related to the proportion of fine droplets present i.e., under the experimental conditions employed in this work, the acid matrix eVect might be mainly attributed to a decrease in the tertiary aerosol. The position of the nebulizer inside the chamber has a in the amount of analyte reaching the atomization cell.Another fact that supports this reasoning is that, for the cyclonic spray noticeable eVect on the performance of cyclonic spray cham- 66 J. Anal. At. Spectrom., 1999, 14, 61–6726 R. L. Lawrence and F. E. Lichte, Anal. Chem., 1982, 54, 638. bers. This is a parameter that must be optimized in order to 27 M. Marichy, M. Mermet and J. M. Mermet, Spectrochim. Acta, obtain the best performance in terms of ICP-AES sensitivity. Part B, 1990, 45, 1195. The nebulizer plume shape and spray chamber geometries 28 S.Greenfield, H. Mc. D.McGeachin and P. B. Smith, Anal. Chim. need to be matched. The matrix eVects should also be Acta, 1976, 84, 67. dependent on the nebulizer position, as it determines the 29 J. Farino, J. R. Miller, D. D. Smith and R. F. Browner, Anal. Chem., 1987, 59, 2303. characteristics of the tertiary aerosols. 30 E. Yosimura, H. Suzuki, S. Yamazaki and S. Toda, Analyst, 1990, 115, 167. The authors thank Mr. Alan Eastgate (Glass Expansion 31 A.Ferna�ndez, M. Murillo, V. Carrio�n and J. M. Mermet, J. Anal. Europe, Switzerland) for the loan of the nebulizer and spray At. Spectrom., 1994, 9, 217. chambers and for the suggestions given during the preparation 32 A. Canals, V. Hernandis, J. L. Todolý� and R. F. Browner, of the manuscript. Also, the authors acknowledge the DGICyT Spectrochim. Acta, Part B, 1995, 50, 305. 33 I. B. Brenner, J. M. Mermet, I. Segal and G. L. Long, Spectrochim. (Spain) for the financial support of this work (Project Acta, Part B, 1995, 50, 323.PB95–0693). 34 H. Ishii and K. Satoh, Talanta, 1983, 30, 111. 35 J. L. Todolý�, J. M. Mermet, A. Canals and V. Hernandis, J. Anal. At. Spectrom., 1998, 13, 55. References 36 M. Carre�, K. Lebas, M. Marichy, M. Mermet, E. Poussel and J. M. Mermet, Spectrochim. Acta, Part B, 1995, 50, 271. 1 R. F. Browner and A. V. Boorn, Anal. Chem., 1984, 56, 786A. 37 J. M. Mermet, J. Anal. At. Spectrom., 1998, 13, 419. 2 Sample Introduction in Atomic Spectroscopy, ed.J. Sneddon, 38 R. L. Dalhquist and J. W. Knoll, Appl. Spectrosc., 1978, 32, 1. Elsevier, New York, 1990, p. 1. 39 I. I. Stewart and J. W. Olesik, J. Anal. At. Spectrom., 1998, 13, 843. 3 A. Gustavsson, Spectrochim. Acta, Part B, 1984, 39, 85. 40 F. J. M. J. Maessen, J. Balke and J. L. M. de Boer, Spectrochim. 4 B. L. Sharp, J. Anal. At. Spectrom., 1988, 3, 939. Acta, Part B, 1982, 37, 517. 5 A. Canals, V. Hernandis and R. F. Browner, Spectrochim.Acta, 41 D. D. Smith and R. F. Browner, Anal. Chem., 1982, 54, 533. Part B, 1990, 45, 591. 42 R. F. Browner, A. Canals and V. Hernandis, Spectrochim. Acta, 6 R. H. Scott, V. A. Fassel, R. N. Kniseley and D. E. Nixon, Anal. Part B, 1992, 47, 659. Chem., 1974, 46, 75. 43 J. W. Olesik and L. C. Bates, Spectrochim. Acta, Part B, 1995, 7 C. Rivas, L. Ebdon and S. J. Hill, J. Anal. At. Spectrom., 1996, 50, 285. 11, 1147. 44 A. Canals, V. Hernandis and R. F. Browner, J.Anal. At. 8 L. Ebdon and R. Collier, Spectrochim. Acta, Part B, 1988, 43, 355. Spectrom., 1990, 5, 61. 9 M. H. Ramsey, M. Thompson and B. J. Coles, Anal. Chem., 1983, 45 Malvern Instruments, 2600 Laser DiVraction User Manual, 55, 1626. Malvern, UK, 1991. 10 X. Zhang, H. Li and Y. Yang, Talanta, 1995, 42, 1959. 46 B. L. Sharp, J. Anal. At. Spectrom., 1988, 3, 613. 11 L. Ebdon and M. R. Cave, Analyst, 1982, 107, 172. 47 J. Mora, J. L. Todolý�, A. Canals and V. Hernandis, J.Anal. At. 12 B. Thelin, Analyst, 1981, 106, 54. Spectrom., 1997, 12, 445. 13 S. Greenfield and P. B. Smith, Anal. Chim. Acta, 1972, 59, 341. 48 V. Hernandis, J. L. Todolý�, A. Canals and J. V. Sala, Spectrochim. 14 S. Greenfield and D. T. Burns, Anal. Chim. Acta, 1980, 113, 205. Acta, Part B, 1995, 50, 985. 15 M. Hoenig, H. Baeten, S. Vanhentenrijk, G. Ploegaerts and 49 Y. Novotny, J. C. Farinas, W. Jia-Liang, E. Poussel and T. Bertholet, Analusis, 1997, 25, 13. J. M. Mermet, Spectrochim. Acta, Part B, 1996, 51, 1517. 16 M. Wu and G. Hieftje, Appl. Spectrosc., 1992, 46, 1912. 50 J. M. Mermet, Anal. Chim. Acta, 1991, 250, 85. 17 A. Geiger, S. Kirschner, B. Ramacher and U. Telgheder, J. Anal. 51 J. M. Mermet, Spectrochim. Acta, Part B, 1989, 44, 1109. At. Spectrom., 1997, 12, 1087. 52 A. Eastgate, personal communication. 18 T. D. Hettipathirana and D. E. Davey, Appl. Spectrosc., 1996, 53 M. A. Tarr, G. Zhu and R. F. Browner, J. Anal. At. Spectrom., 50, 1015. 1992, 7, 813. 19 K. A. Taylor, B. L. Sharp, D. J. Lewis and H. M. Crews, J. Anal. 54 C. Pan, G. Zhu and R. F. Browner, J. Anal. At. Spectrom., 1990, At. Spectrom., 1998, 13, 1095. 5, 537. 20 S. A. Beres, P. H. Brueckner and E. R. Denoyer, At. Spectrosc., 55 J. A. Borowiec, A. W. Boorn, J. H. Dillard, M. S. Cresser, 1994, 15, 96. R. F. Browner and M. J. Matteson, Anal. Chem., 1980, 52, 1054. 21 L. Gras, J. Mora, J. L. Todolý�, V. Hernandis and A. Canals, 56 K. G. Kronholm and R. K. Skogerboe, Appl. Spectrosc., 1986, Spectrochim. Acta, Part B, 1997, 52, 1201. 8, 1161. 22 H. Isoyama, T. Uchida, T. Niwa, C. Lida and J. G. Nakagawa, 57 S. S. Berman, J. W. McLaren and S. N. Willie, Anal. Chem., 1980, J. Anal. At. Spectrom., 1989, 4, 351. 52, 488. 23 H. Isoyama, T. Uchida, C. Lida and J. G. Nakagawa, J. Anal. At. 58 C. Dubuisson, E. Poussel and J. M. Mermet, J. Anal. At. Spectrom., 1990, 5, 307. Spectrom., 1997, 12, 281. 24 A. Montaser, M. G. Minnich, J. A. McLean, H. Liu, J. A. Caruso 59 C. Dubuisson, E. Poussel, J. L. Todolý� and J. M. Mermet, and C. W. McLeod, in Inductively Coupled Plasma Mass Spectrochim. Acta, Part B, 1998, 53, 593. Spectrometry, ed. A. Montaser, Wiley-VCH, New York, 1998, p. 83. 25 D. R. LuVer and E. D. Salin, Anal. Chem., 1986, 58, 654. Paper 8/06550K J. Anal. At. Spectrom., 1999,