JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 42 1 Comparison of Normal and Low-flow Torches for Inductively Coupled Plasma Mass Spectrometry Using Optimized Operating Conditions E. Hywel Evans* and Les Ebdon Plymouth Analytical Chemistry Research Unit Department of Environmental Sciences Polytechnic South West Drake Circus Plymouth Devon PL4 8AA UK Simplex optimization has been used in order to optimize inductively coupled plasma (ICP) operating conditions for ICP mass spectrometry (namely nebulizer auxiliary and coolant gas flows and forward power) using a standard and a low-flow torch. The signal intensity for In+ at m/z 115 was used as the criterion of merit. Analyte signals were lower by a factor of approximately 3 for the low-flow torch compared with the standard torch.The Ba2+:Ba+ ratio was greater by a factor of 7 for the low-flow compared with the standard torch while the BaO+:Ba+ ratios were low for both torches. The ArO+:ln+ and ArN+:ln+ ratios were higher for the low-flow torch and the ArAr+:ln+ ratios were low for both torches. A relatively flat mass response curve was obtained using the low-flow torch indicating that mass discrimination effects were less pronounced for ions extracted from the plasma formed using this torch. Plasma stability was better with the low-flow torch compared with the standard torch when introducing volatile organic solvents. Keywords Inductively coupled plasma mass spectrometry; simplex optimization; low-flow torch; organic solvent Reports of optimization studies for inductively coupled plasma mass spectrometry (ICP-MS)*-6 have mainly de- tailed the effects of individual operating parameters on analytical performance.Whereas optimization studies for ICP atomic emission spectrometry (ICP-AES) generally use signal-to-background ratio or signal-to-noise ratio as the criterion of merit owing to the relatively large varia- tions in continuum background optimization studies for ICP-MS generally use maximum signal as the criterion of merit because of the low continuum background. More important from the point of view of ICP-MS is the need to optimize the instrument for maximum analyte re- sponse while at the same time minimizing potential interferences such as doubly charged and polyatomic ions. Several workers have optimized plasma operating con- ditions and ion lens voltages for the two main commer- cially available systems namely the Sciex Elan (Sciex Thornhill Ontario Canada)'~~.~ and the VG PlasmaQuad (VG Elemental Winsford Cheshire UK).2,3*6 These work- ers optimized the systems for maximum signal using univariate optimization techniques and subsequently chose operating conditions which gave large analyte sig- nals but also minimal signals due to doubly-charged and polyatomic ions.In general they concluded that suitable compromise operating conditions could be found for different elements with widely differing masses ioniza- tion potentials and chemistries and that these conditions yielded tolerably low levels of doubly charged and oxide ions. However it is possible that other workers may find completely contrary results since it has been our experi- ence that there is a great deal of variability in the levels of polyatomic ions from day-to-day with the same instru- ment let alone between different instruments.Such varia- bility may depend to a great extent on the condition of the sampling and skimmer cones which gradually deter- iorate with time. The trends observed for the two instru- ments were broadly similar although differences have been noted in the behaviour of doubly charged ions,3 probably owing to important differences in the design of the load coil interface and ion lenses which have a great influence on the ion energies of extracted ions and their subsequent transmission into the quadrupole analyser. All workers identified the most important para- * Present address University of Cincinnati Department of Chemistry Mail Location 1 72 Cincinnati OH 4522 1-0 172 USA.meters to be nebulizer gas flow forward power and sampling depth. The optimization studies mentioned above have been applied to ICP-MS systems operating with standard compo- nents and the preferred optimization technique has been to perform a series of univariate searches. However a more rigorous optimization regime is necessary for instrumental systems that have operating parameters that are interdepen- dent variables. Recently Schmit and Chauvette5 and Evans and Caruso6 have demonstrated the applicability of a multivariate simplex optimization technique for tuning of the ion lens voltages. Since the operating variables are known to interact univariate methods alone are unsuitable for system optimization.Therefore a multivariate optimi- zation technique such as simplex optimization is necessary in order to locate the true optimum. Low-flow torches have been shown to allow more stable plasma operation for ICP-AES.' It has been observed in this laboratory8 that stable operation of the plasma is more difficult for ICP-MS compared with ICP-AES on the introduction of an organic solvent probably because of the perturbing effect of the interface which is in contact with the plasma in the former instrument. There- fore a low-flow torch may have certain advantages over the standard torch in ICP-MS for the introduction of organic solvents. Gordon et aL9 have optimized a water cooled low-flow torch for ICP-MS using a univariate optimization tech- nique and arrived at the optimum operating conditions of plasma gas 2.1 1 min-l; neublizer gas 0.20 1 min-l; and forward power 1100 W.Using this torch they observed analyte signals comparable with those obtained with the standard torch. Ross et a1.lo have studied a 9 mm torch and found it to give similar sensitivities and doubly charged ion and oxide ratios to a conventional plasma. However in the last two s t u d i e ~ ~ ? ~ ~ the operating condi- tions were optimized using univariate approaches. Simplex optimization has so far not been applied to ICP-MS for the optimization of the plasma operating conditions but has been shown to offer particular advantages for ICP-AES.11-13 In the present work plasma operating conditions using both a standard torch and a low-flow torch have been optimized by using simplex optimization and the two torches subsequently evaluated with regard to plasma stability for the introduction of volatile organic sol- vents.422 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL.6 Experimental Instrumentation All experiments were performed using an inductively coupled plasma mass spectrometer (VG PlasmaQuad 2). Torch dimensions are given in Table 1. The low-flow torch differed from the standard torch in that the spacing between the intermediate and outer tubes and the internal diameter of the injector tube were smaller in the former. Addition- ally the inlet tubes for the auxiliary (intermediate) and coolant (outer) gases were constricted in the low-flow torch thereby increasing the velocity of these gases through the intermediate and outer tubes respectively. A standard concentric nebulizer was used in conjunction with the low- flow torch and a V-groove nebulizer (Ebdon nebulizer PS Analytical Sevenoaks Kent UK) with the standard torch.Simplex Optimization Parameters The parameters which were optimized are listed in Table 2 with the ranges over which the optimization experiments were conducted. The sampling depth was not included because adjustment of this parameter caused arcing be- tween the torch box and sampling cone probably owing to a loose radiofrequency connection. The sampling depth was maintained at 11 mm and the spray chamber at 2 "C. Simplex optimization experiments were performed using a software package developed previously14 and run on a microcomputer (Apple IIe Apple Computer Cupertino CA USA).The algorithm used was based on that of Yarbro and Derning,ls and initially covered the whole of the factor space then contracting in step size by 50% for each contraction as the optimum was approached. The optimum was deemed to have been reached when the relative standard deviation of the response factors for the vertices retained in the simplex was within the medium term precision of the technique which was given as 5% for this experiment. The signal intensity for In+ in area counts s-l at m/z 1 15 was taken as the criterion of merit. Whenever operating conditions were altered the ion lenses (extraction collec- tor L1 L2 L3 and L4) and the pole-bias were re-adjusted manually by the operator in order to obtain the maxi- mum signal.Additionally the position of the torch in the horizontal and vertical planes was re-adjusted in the same way so as to obtain a maximum signal. This was particularly critical for the low-flow torch since the small internal diameter of the injector meant that accurate alignment with the sampler orifice was important in order to achieve maximum analyte signal. After each optimization was completed univariate searches were performed for each parameter in turn while holding the others at the optimum established by using the simplex procedure. Mass Spectrometer Operating Conditions The mass spectrometer was operated in survey scanning mode throughout. Data acquisition parameters are listed in Table 3.Reagents and Standards A multi-element standard solution of Be Co In Ba and Pb 100 ng ml-l was prepared by dilution of a 10 pg ml-l multi-element stock solution using 2% v/v nitric acid (Aristar BDH Chemicals Poole Dorset UK) in distilled de-ionized water. All multi-element stock solutions were made up using single element standard solutions (Spectro- SOL BDH Chemicals). Results and Discussion Optimization experiments were completed in 25-30 steps. The only problem encountered was that caused by extreme sets of operating conditions at the boundary limits which made tuning of the ion lenses difficult. Table 4 shows the optimum conditions established for the standard and low-flow torches. Univariate searches at the established optimum conditions for analyte signal Ba2+:Ba+ ratio BaO+:Ba+ ratio ArO+ ArN+ and ArAr+ are shown in Figs.1 - 1 5. Despite the fact that the optimization was performed using llsIn+ as the criterion of merit it was considered useful to compare the effects of operating conditions on the criteria mentioned above so that a valid comparison between the two torches could be made with respect to polyatomic and doubly charged ions. On completion of the optimization the simplex proce- dure retained the five vertices which gave the best response factors each vertex consisting of a set of operating conditions. The vertical arrows on Figs. 1 4 10 and 13 Table 1 Dimensions of the standard and low-flow torches in mm Torch dimension Standard torch Low-flow torch Injector tube i.d. 1.6 1 .o Intermediate tube 0.d.15.6 16.6 Outer tube 0.d. 20.4 20.4 Outer tube i.d. 18.0 18.0 Configuration factor 0.77 0.8 1 Gas inlets. i.d. 6 2 Table 2 Boundary limits of parameters studied during the simplex optimization Range Table 3 Data acquisition parameters used in survey scanning mode Value Parameter Mass range mlz Number of channels Number of scan sweeps Dwell timelps Time per scads Number of scans 8.0-215.5 2048 100 320 65.5 I Table 4 Optimum operating conditions established for the stan- dard and low-flow torches Optimum conditions Parameter Standard torch Low-flow torch Nebulizer gasll min-' 0.200-1.250 0-0.800 Auxiliary gasll min-' 0-2.5 0-2.0 Coolant gasll min-I 11-18 6-10 Forward powerlW 900- 1 800 500-900 Parameter Standard torch Low-flow torch Nebulizer gad1 min-I 0.842 0.670 Auxiliary gad1 min-I 0.8 1.6 Coolant gad1 min-l 11.5 6.8 Forward power/W 1583 865JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL.6 423 indicate the optimum condition for each of the operating conditions which was obtained in each instance by averag- ing the values for each of the parameters in the five vertices retained by the simplex procedure. The error bars represent the range of values obtained for each operating condition. The trends shown in the graphs for the analyte signal were repeatable from day-to-day and the optimum did not exhibit significant variation over a period of several months. The absolute magnitude of the signals varied with time and the polyatomic ion to indium ratios also varied considerably.However the general trends remained rela- tively similar and were sufficiently dissimilar for the two torches to warrant comparison. The effects of the different operating parameters on the criteria mentioned above are discussed below. Effect of Operating Parameters Nebulizer gas The effect of nebulizer gas flow on analyte signal for the elements Be Co In Ba and Pb using the isotopes at mlz 9 59 115 138 and 208 respectively is shown in Fig. l(a) for the standard torch and Fig. l(b) for the low-flow torch. It should be remembered that the optimization was per- formed using the In+ response as the criterion of merit hence the vertical arrows indicate the optimum conditions for this element only. The simplex procedure located the optimum successfully for both torches.However for the standard torch [Fig. l(a)] Be+ and Co+ signals peaked at higher nebulizer gas flow rates than In+ and Ba+ and Pb+ signals peaked at slightly lower flow rates the trend being that the lower the mass the higher the nebulizer flow rate 100000 (a) t 80000 60000 40000 - 20000 4- C 3 8 ( D o I I I Ag P\ 25000 20000 15000 10000 5000 n 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 Nebulizer gas flow/l min-’ Fig. 1 Effect of nebulizer gas flow on signals for A 9Be+; B 59C0+; C llsIn+- D 138Ba+; and E 208Pb+ using (a) the standard and (b) the low-flow torch that was required for maximum signal. Gray and co- workers2J6 have shown that the plasma potential and ion energies for elements of different mass are dependent on the nebulizer gas flow and forward power at least in the VG PlasmaQuad instrument for which plasma potentials be- tween + 5 and + 20 V are common.A consequence of this is that certain sets of operating conditions may result in ion energies which vary across the mass range making it difficult to optimize the ion lenses in order to achieve uniform response for elements of different mass. This may be the cause of the trends observed and shown in Fig. l(a) since the ion optics were optimized for In+ only. These results differ from those of Gray and Williams2 who found that one nebulizer gas flow was optimum for elements of different mass. One explanation for this may be that they performed experiments at 1300 W forward power whereas this work was performed at 1583 W forward power which was found to be the optimum.Douglas and French” have suggested that ion energies vary considerably more than those reported by Gray and Williams,2 owing to measure- ment of the ion energies after the ion optics which act as a filter for ions of the same energy. For the low-flow torch with the exception of Be all elements exhibited maximum signals at more or less the same nebulizer gas flow [Fig. l(b)J. Gordon et aL9 have postulated that the plasma potential is greater in plasmas operated at lower than normal flows with a consequent increase in ion energies. However as long as the spread in ion energies for elements of different masses is small then one set of ion lens conditions such as those for In+ should be optimum for all ions. This seems to be true for the low- flow plasma studied in this work with the exception of Be.The major effect of a high plasma potential is to increase the amount of doubly charged i ~ n s . ~ J ~ This effect was observed for the low-flow torch studied in this work the Ba2+:Ba+ ratio showing a peak that coincided with the peaks in analyte signals and having the high value of approximately 0.5 at its maximum [Fig. 2(b)]. During operation of the low-flow torch what appeared to be a secondary discharge or boundary layer was visible at the sampler orifice. The large amounts of doubly charged ions formed might explain why the peak of the Ba2+:Ba+ ratio closely matched that for the analyte signal. For the 0.6 0.4 0.2 + (D 0 m +- 0 standard torch the Ba2+:Ba+ ratio was much (a) 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 Nebulizer gas flow/l min-’ Fig.2 Effect of nebulizer gas flow on the ratios for A Ba2+:Ba+; and B BaO+:Ba+ using (a) the standard and (b) low-flow torch424 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 lower at the optimum nebulizer gas flow [Fig. 2(a)] in comparison with the low-flow torch but increased as gas flow increased. These results are consistent with those obtained by other workers2v3 using a PlasmaQuad instru- ment with an asymmetrically grounded load coil. Gray et a1.16 have shown that as nebulizer gas flow increases so does the plasma potential resulting in an increase in the amount of doubly charged ions a decrease in ArO+:Co+ and ArAr+:Co+ ratios and a slight decrease in CeO+. For the Elan instrument with a centre tapped load coil the Ba2+:Ba+ ratio decreases with nebulizer gas flow.' The behaviour is different in this instrument because the plasma potential is relatively low,'* i.e.+0.5 to -3.5 V and less dependent on operating conditions. For both torches the BaO+:Ba+ ratios were low at optimum nebulizer gas flows [Fig. 2(a) and (b)] and therefore would not give rise to serious potential interfer- ences. The BaO+:Ba+ ratio increased slightly at high gas flow for the standard torch [Fig. 2(a)] in agreement with Gray and Williams,2 and the ratio showed a small peak at 0.75 1 min-l for the low-flow torch [Fig. 2(b)] which closely follows the trend observed for analyte signal suggesting that the observed trend was influenced by factors such as analyte or O+ transport into the plasma.The effect of nebulizer gas flow on ArO+ ArN+ and ArAr+ is shown in Fig. 3. For the standard torch the trends exhibited in Fig. 3(a) agree with those reported by Gray and Williams,* and each show a minimum at the optimum nebulizer gas flow though the response for ArAr+ was much lower in this study. In contrast for the low-flow torch the ArN+ and ArO+ signals peaked at nebulizer gas flows which were the same as and 0.01 1 min-l higher respectively than the optimum for In+ signal [Fig. 3(b)]. This difference 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 Nebulizer gas flow/l rnin-' Fig. 3 Effect of nebulizer gas flow on the signals for A ArO+; B ArN+; and C ArAr+ using (a) the standard and (b) the low-flow torch is significant because the data were acquired within the same experiment for all masses.It is likely that the presence of the O+ and N+ precursors influenced the formation of the polyatomic species to the greatest extent their concen- tration in the plasma being determined by the nebulizer gas flow. However Douglas and French17 have calculated that for low-flow argon plasmas a greater proportion of the plasma gas will be sampled through the sampler orifice possibly leading to greater entrainment of atmospheric gases which also influence the formation of ArO+ and ArN+. Whatever the situation more work is necessary in order to determine the mechanisms of formation of these species in a low-flow low-power plasma with a high plasma potential. The low-flow torch gave rise to a smaller signal for ArAr+ in comparison with the standard torch (Fig.3) though the difference was not great. The low-flow torch gave rise to a smaller signal for ArAr+ in comparison with the standard torch (Fig. 3) though the difference was not great. Forward power The effect of forward power on the analyte signal is shown in Fig. 4 for both torches. With regard to the standard torch [Fig. 4(a)] the simplex procedure has successfully located the optimum forward power for maximum In+ signal. However the maximum signals for Be+ and Co+ were at lower power while those for Ba+ and Pb+ were at higher power. As observed for nebulizer gas flow the optima were mass dependent although in this instance the order was reversed i. e. higher masses required higher power for maximum signal. A correlation between power and first 50000 40000 30000 20000 10000 B I 1 I m m C 22 P .- a 80000 60000 40000 20000 0 400 800 1200 1600 2000 Forward power/W Fig.4 Effect of forward power on signals for A 9Be+; B 59C0+; C llsIn+. D 138Ba+; and E 208Pb+ using (a) the standard and (b) low- flow torchJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 42 5 Table 5 Masses of the major isotopes first ionization potentials ( I ) and apparent optimum powers (Popl) for the elements studied Isotope I,/V PO,l/W 9Be 9.32 1300 5 9 c 0 7.86 1350 ll51n 5.79 1583 13*Ba 5.21 >1800 zo8Pb 7.42 >1800 ionization potential cannot be made i.e. those elements with highest first ionization potentials would require high- est power whereas this was not true (Table 5). A disparity in ion energies for elements of different mass may again explain the apparent differences in optima as discussed for nebulizer gas flow since the ion optics were optimized for In+ only.Gray et a l l 6 found that plasma potential and hence ion energies increase as nebulizer gas flow is increased but decrease as power is increased. Bearing this in mind the trends in mass dependent optima for nebulizer gas flow and power support each other Taking Pb as an example for one particular nebulizer gas flow or forward power Pb+ ions will have a greater energy than In+ ions due to the mass dependent acquisition of kinetic energy in the expansion region. Hence Pb+ ions with energies required for maximum transmission through ion optics optimized for In+ will be most prevalent at a lower nebulizer gas flow or a higher power than is optimal for In+.Such an effect was observed in this work however since no measurements or ion energy were made such an explana- tion remains tentative. For the low-flow torch maximum signal was not obtained for any of the elements even at the highest power studied [Fig. 4(b)]. The simplex procedure located the ‘optimum’ power for the In+ signal at a lower value than would be regarded as optimum i.e. the optimization failed. This was because at a power greater than 1000 W the torch started to melt requiring a false low response to be input to the computer whenever such a condition arose. This is the normally accepted procedure for the type of algorithm used but it can lead to poor results when the optimum is close to or beyond one of the boundary conditions. This had the effect of causing the simplex to locate an ‘optimum’ at a lower power.However the trends exhibited show a similar pattern for all elements indicating that the discrepancy in ion energies was less pronounced with the low-flow torch compared with the standard torch. The increase in analyte response with increasing power is probably due to a corresponding increase in ionization temperature. For the standard torch the Ba2+:Ba+ ratio decreased with increasing power to less than 0.1 at the optimum [Fig. 5(a)] in agreement with the results of Long and Brown,3 and possibly owing to a corresponding decrease in plasma potential. The BaO+:Ba+ ratio was less than 0.01 at all powers studied. For the low-flow torch the Ba*+:Ba+ ratio trend exhibited different behaviour showing a maximum at 825 W [Fig.5(b)]. The subsequent decrease in the ratio above 825 W could have been due to a decrease in plasma potential though this seems unlikely. As mentioned already a discharge was visible in the sampler orifice during oper- ation and this may well have had an effect. Gordon et aL9 observed Ce2+:Ce+ ratios of up to 1.0 and a much broader peak profile between 900 and 1400 W using a water cooled low-flow torch though it was operated at much lower gas flows and was different in design. Evidently it is desirable to operate this particular torch at the maximum power possible in order to reduce the Ba2+:Ba+ ratio while at the 1 .o 0.8 0.6 0.4 0.2 + l-0 0 rn i o 4d X (a) A 400 800 1200 1600 2000 Forward power/W Fig.5 Effect of forward power on the ratios for A Ba2+:Ba+; and B BaO+:Ba+ using (a) the standard and (b) the low-flow torch same time improving the analyte signals. The BaO+:Ba+ ratio was of the order of 0.0 1 between 800 and 1000 W [Fig. 5(b)] which was the analytically useful power range. The ArO+ and ArN+ signals exhibited contrasting behav- iour for the two torches (Fig. 6). Whereas for the standard 40000 20000 r UJ v) 4d = o s Forward powerIW Fig. 6 Effect of forward power on the signals for A A&+; B ArN+; and C ArAr+ using (a) the standard and (b) the low-flow torch426 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 0.8 0.6 0.4 torch the signals for these species decreased with increasing forward power [Fig. 6(a)] for the low-flow torch the signals decreased to a minimum at 700 W increased to maxima at 950 and 900 W respectively then decreased again [Fig.6(b)]. In the latter instance i.e. the low-flow torch the initial decrease in polyatomic ions may have been due to a reduction in the entrainment of atmospheric gases as the plasma increased in size with increasing power. Subse- quently O+ and N+ precursors from the nebulizer gas may have predominated and the polyatomic ions increased as ionization temperature increased with power until the point at which other factors such as ion energy or decomposition had a greater influence. The ArAr+ signal varied very little for both torches over the power ranges studied. - - - Sampling depth The sampling depth parameter was not included in the simplex optimization but was held constant at 11.0 mm.However univariate searches were undertaken in an effort to ascertain whether this parameter had a significant effect on the criteria already mentioned. Sampling depth is defined as the distance between the tip of the sampling cone and the foremost coil of the load coil. For the standard torch analyte signals decreased as the sampling distance was increased [Fig. 7(a)] which would be expected as ions were sampled from successively cooler regions of the plasma where ion populations were lower. There was comparatively little change in analyte signals as the sampling distance of the low-flow torch was increased [Fig. 7(b)]. This could have been owing to the secondary discharge in the sampling orifice which probably influenced the degree of ionization of the elements to the greatest 50000 40000 30000 20000 1oooc v) v) t4 ; c (D m 1.m 2 120000 & 100000 .- v) 80000 60000 40000 20000 n ( b ) B 1 8 9 10 11 12 13 Sampling cone-load coil distance/mm Fig. 7 Effect of sampling depth on signals for A 9Be+; B s9C0+; C IlsIn+; D 13*Ba+; and E 2osPb+ using (a) the standard and (b) the low-flow torch 1 .o (a) 0.2 1 B % - n~ 0 I I 8 9 10 11 12 13 Sampling cone-load coil distance/mm Fig. 8 Effect of sampling depth on the ratios for A Ba2+:Ba+; and B BaO+:Ba+ using (a) the standard and (b) the low-flow torch extent rather than the sampling distance. However the discharge did change in intensity as the sampling distance was altered becoming more intense at greater distances. Similarly very little change was observed in the Ba2+:Ba+ and BaO+:Ba+ ratios for either torch (Fig.8) although a steady increase in the Ba2+:Ba+ ratio was observed using the low-flow torch as the distance between the sampling cone and the load coil was increased [Fig. 8(b)]. This was probably linked to the discharge in the sampler orifice which increased in intensity as the sampling position moved further away from the load coil. A gradual increase in the signals for ArO+ ArN+ and ArAr+ was observed as sampling distance from the load coil 40000 20000 - v) v) C 3 c 8 0 m 40000 1 C 0 v) .- 20000 (a) C A Y Y 8 9 10 11 12 13 Sampling cone-load coil distance/mm Fig. 9 Effect of sampling depth on the signals for A ArO+; B ArN+; C and ArAr+ using (a) the standard and (b) the low-flow torchJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1991 VOL.6 427 increased for the standard torch [Fig. 9(a)] probably because ions were sampled from a cooler region of the plasma. The trends exhibited by these ions when using the low-flow torch contrasted greatly with each other [Fig. 9(b)]. The signal for ArO+ decreased that for ArN+ increased slightly and that for ArAr+ remained constant. The con- trasting trends for ArO+ and ArN+ suggest that these ions may have been formed with precursors from different sources in the low-flow plasma either from the nebulizer gas or from entrained atmospheric gases. Auxiliary gas The simplex optimization successfully located the optimum gas flows for both torches. The trends in analyte signals observed for the standard torch [Fig.10(a)] agree with the results of Long and Brown3 who observed an identical pattern albeit at lower gas flows. For the low-flow torch auxiliary gas flow had little effect on analyte signals [Fig. 1 O(b)]. Likewise Ba2+:Ba+ and BaO+:Ba+ ratios were relatively unaffected (Fig. 11) for both torches. The ArO+ and ArN+ signals exhibited similar behaviour for the standard torch [Fig. 12(a)] although contrasting this with the behaviour for the low-flow torch [Fig. 12(b)] suggests again that these ions were formed with precursors from different sources with this torch. For both torches the optimum gas flows determined for In coincided with signals close to the minima observed for these ions. The ArAr+ signal varied very little over the ranges studied for either torch (Fig.12). 100000 80000 60000 40000 7 20000 fn 3 4- 8 0 l i (a) B 25000 20000 1 = 1 (11111 0 0.5 1.0 1.5 2.0 2.5 3.0 Auxiliary gas flow/l min-’ Fig. 10 Effect of auxiliary gas flow on signals for A 9Be+; B 59C0+; C IlsIn+; D 138Ba+; and E ***Pb+ using (a) the standard and (b) the low-flow torch 0.4 o . 6 ~ + 0.2 - 0 0.5 1.0 1.5 2.0 2.5 3.0 Auxiliary gas flow/l min-’ Fig. 11 Effect of auxiliary gas flow on the ratios for A BaZ+ Ba+; and B BaO+:Ba+ using (a) the standard and (b) the low-flow torch 60000 40000 20000 fn fn 3 4- C I 8 0 60000 m 1. C 0 m .- 40000 20000 0 (a) ( b ) I B 7 3 3 Y * .c Y U 1l u 0.5 1.0 1.5 2.0 2.5 3.0 Auxiliary gas flow/l min-’ Fig. 12 Effect of auxiliary gas flow on the signals for A ArO+; B ArN+; and C ArAr+ using (a) the standard and (b) the low-flow torch Coolant gas The simplex optimization located the ‘optimum’ coolant gas flow to be close to the lower boundary limit for both torches (Fig.13). Evidently the true optima were below the boundary limits for coolant gas flow at 1 1 and 6 1 min-l for the standard and low-flow torches respectively. The torches could not be operated below these flows because of the danger of melting. Analyte signals varied in a similar428 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1991 VOL. 6 :TA B manner for both torches i.e. decreasing as coolant gas flow increased. However a coolant gas flow of 10 1 min-' would be necessary for the low-flow torch if it was operated at the maximum power of 1000 W. The BaO+:Ba+ ratios remained relatively constant throughout the range of coolant gas flow studied for both torches (Fig. 14).The Ba2+:Ba+ ratios exhibited contrasting t 50000 40000 30000 20000 B " ( a ) m - Is v) .- 30000 t ' ' 20000 10000 0 1 - \ 5 10 15 20 Coolant gas fIow/I min-' Fig. 13 Effect of coolant gas flow on signals for A 9Be+ B 59C0+; C I151n+; D 138Ba+; and E 2osPb+ using (a) the standard and (b) the low-flow torch 0.8 0 2 0.6 .- 4- 0.4 0.2 0 5 10 15 20 Coolant gas fIow/I min-' Fig. 14 Effect of coolant gas flow on the ratios for A BaZ+:Ba+; and B BaO+:Ba+ using (a) the standard and (b) the low-flow torch behaviour for the two torches with respect to coolant gas flow (Fig. 14). It is difficult to assess the influence of this parameter on the formation of Ba2+ ions in the plasma.Increasing the coolant gas flow caused an increase in the intensity of the discharge in the sampler orifice using the low-flow torch which probably contributed to the increase in the Ba2+:Ba+ ratio. The signals for ArO+ ArN+ and ArAr+ showed very little change with respect to coolant gas flow for the standard torch [Fig. 15(a)]. However once again the ArO+ and ArN+ signals exhibited dissimilar trends for the low-flow torch [Fig. 15(b)]. This lends further strength to the argument that these species were formed with precursors from different sources in the low-flow plasma. More significantly such species were more prevalent in the low-flow plasma and were more sensitive to coolant gas flow than the standard torch. 40000 20000 - I cn c +I Z o (II m 22 (II C Is v) 40000 .- 20000 0 A- B $5 Coolant gas flow/! min-' Fig.15 Effect of coolant gas flow on the signals for A ArO+; B ArN+; and C ArAr+ using (a) the standard and (b) the low-flow torch Semi-quantitative Calibration Fig. 1 6 shows two semi-quantitative calibration graphs constructed when using the low-flow torch. Fig. 16(a) illustrates the effect of including Ba in the calibration. In m X 1191 1 (a) X I "---' 0 1 41.50 83.00 124.5 166.0 207.5 m/z Fig. 16 Semi-quantitative calibration graphs calculated using the low-flow torch with (a) 138Ba+ included; and (b) 138Ba+ excludedJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY. SEPTEMBER 199 1 VOL. 6 429 this instance the signal for I3*Ba+ was degraded because of the propensity for Ba2+ ions to form when this torch is used and the calibration was rendered useless.However when Ba was excluded a useful calibration was obtained [Fig. 16(b)]. Because of this effect it may be necessary to perform a separate calibration using elements with low first ioniza- tion potentials if such elements are to be determined semi- quantitatively. Another interesting property of the low-flow torch was the relatively high sensitivity achieved for 9Be+. A relatively flat mass response curve was obtained using the low-flow torch compared with that typically obtained with a stan- dard torch. The relatively high signals obtained for 9Be+ and 59C0+ were observed throughout the experiment although there was considerable variation in relative signal intensities for different operating conditions.It is suspected that the existence of the secondary discharge in the sampler orifice may have played an important role in that it increased the ion energies of the elements thereby reducing space charge effects in the ion beam resulting in less defocusing of the light elements. Effect of Organic Solvents Several organic solvents of differing volatility were intro- duced into the plasmas formed with the standard and low- flow torches and their effects on plasma stability and reflected power noted. The torches were operated using the conditions shown in Table 6. It was necessary to introduce oxygen into the nebulizer gas in order to prevent carbon deposition on the cones the amount depending on the solvent used. Also the sample uptake rate was reduced considerably for the more volatile solvents namely acetone cyclohexane and hexane al- though in general a greater sample uptake rate could be tolerated by the low-flow torch.The performance of the two torches with respect to reflected power and plasma stability is shown in Table 7. It is evident from Table 7 that the low-flow torch was much more tolerant of organic solvents especially the more volatile solvents than the standard torch. This may be an Table 6 Operating conditions for the low-flow and standard torches for organic solvent introduction Parameter Standard torch Low-flow torch Ar nebulizer gad1 min-' O2 nebulizer gad1 min-l Auxiliary gad1 min-l Coolant gad1 min-I Forward power/W Sampling depth/mm Spray chamber temperature/"C Sample uptake rate/ml min-' 0.802 1.3 0.025-0.071 15 1800 10.5 -2 0.12-1.46 0.655 0.025-0.1 15 1 .o 10 1000 10.5 - 2 0.12-1.46 Table 7 Reflected power and plasma stability (S stable; U unstable; and VU very unstable) compared for the low-flow and standard torches during organic solvent introduction Standard torch Low-flow torch Reflected Plasma Reflected Plasma Solvent power/W stability power/W stability Propan-2-01 45 S 10 S Ethanol 42 S 10 S Methanol 60 S 12 S Acetone 80 vu 25 S Cyclohexane 65 vu 20 S Hexane 67 vu 45 U Table 8 Figures of merit for the standard and low-flow torches operated at optimum conditions Figure of Standard Low-flow merit torch torch 'I5In+ signal/area counts s-I 47 591 17 508 Ba2+:Ba+ 0.078 0.510 0.008 BaO+:Ba+ ArO + In + 0.19 1.12 ArN+:In+ 0.12 0.27 ArAr+:Tn+ 0.001 0.003 0.020 advantage for the analysis of organic solvents and coupled high-performance liquid chromatography-ICP-MS studies where the mobile phase may be volatile. Conclusions A comparison of various figures of merit can be made for the two torches operating at conditions found to be optimum for maximum In+ signal by reference to Table 8.In general signals for a number of elements covering the mass range being studied were lower with the low-flow torch compared with the standard torch although considerable variation in the magnitude of signals for the low mass elements Be and Co was noted for the low-flow torch. The Ba2+:Ba+ ratio was much greater and was influenced differently by operating parameters for the low-flow com- pared with the standard torch probably because of a greater plasma potential in the low-flow plasma.The BaO+:Ba+ ratios were low for both torches with no significant differences with respect to operating parameters. The ArO+:In+ and ArN+:In+ ratios were higher for the low-flow compared with the standard torch probably owing to the greater air entrainment and results suggest that the ArO+ and ArN+ species were formed with precursors from different sources in the low-flow plasma and to a lesser extent in the plasma formed using the standard torch. The ArAr+:In+ ratio was lower for the low-flow compared with the standard torch. A relatively flat mass response curve was achieved using the low-flow torch and the plasma formed with this torch was much more stable for the introduction of organic solvents which may be an advantage in some applications involving volatile solvents.The support of E.H.E. by the Science and Engineering Research Council and ICI Materials Research Centre under the Co-operative Award in Science and Engineering (CASE) scheme is gratefully acknowledged. References Vaughan M. A. and Horlick G. Appl. Spectrosc. 1986 40 434. Gray A. L. and Williams J. G. J. Anal. At. Spectrom. 1987 2 599. Long S. E. and Brown R. M. Analyst 1986 111 901. Vaughan M. A. Horlick G. and Tan S. H. J. Anal. At. Spectrom. 1987 2 765. Schmit J.-P. and Chauvette A. J. Anal. At. Spectrum. 1989 4 755. Evans E. H. and Caruso J. A. presented at the 1991 Pittsburgh Conference Chicago USA 3rd-8th March 1991 paper No. 109 1. Ng R. C. Kaiser H. and Meddings B. Spectrochim. Acta Part B 1985 40 63.430 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 8 Evans E. H. Ebdon L. and Barnett N. W. presented at the 1989 European Conference on Plasma Spectrochemistry Reutte Austria 8th-14th January 1989 paper No. P2-44. 9 Gordon J. S. van der Plas P. S. C. and de Galan L. Anal. Chem. 1988 60 375. 10 Ross B. S. Chambers D. M. Vickers G. H. Yang P. and Hieftje G. M. J. Anal. At. Spectrom. 1990 5 351. 11 Ebdon L. Evans E. H. and Barnett N. W. J. Anal. At. Spectrom. 1989 4 505. 12 Ebdon L. Cave M. R. and Mowthorpe D. J. Anal. Chim. Acta. 1980 115 179. 13 Cave M. R. Kaminaris D. M. Ebdon L. and Mowthorpe D. J. Anal. Pruc. 1981 18 12. 14 Norman P. Ph.D. Thesis Plymouth Polytechnic Council for National Academic Awards 1987. 15 Yarbro L. A. and Deming S. N. Anal. Chim. Acta 1974,73 391. 16 Gray A. L. Houk R. S. and Williams J . G. J. Anal. At. Spectrom. 1987 2 13. 17 Douglas D. J. and French J. B. J. Anal. At. Spectrom. 1988 3,743. 18 Houk R. S. Schoer J. K. and Crain J. S. J. Anal. At. Spectrom. 1987 2 283. Paper 1/01 842F Received March 15th 1991 Accepted April 19th I991