Characterization of a hydrogen flame as an ion source for mass spectrometry Lee L. Yu,* Gregory C. Turk and S. Roy Koirtyohann† Analytical Chemistry Division, National Institute of Standards and Technology, Gaithersburg, MD 20899, USA Received 16th December 1998, Accepted 16th December 1998 A commercial inductively coupled plasma (ICP) mass spectrometer was modified to employ an air–hydrogen flame in place of the ICP as an ion source. A liquid nitrogen trap was placed in the vacuum line to remove water.A very simple intrinsic mass spectral background was obtained with the hydrogen flame ionization mass spectrometry (FIMS). Molecular ions such as K(H2O)+, Na(H2O)+, Ca(H2O)+ and CaOH(H2O)x+ (x=0–2) were observed when solutions containing Na, K or Ca were aspirated. Although the presence of the molecular ions complicated the mass spectra, it also provided a wider choice of analytical masses for an analyte. Isotope ratio measurements of Ca were made with both Ca+ and CaOH+ species at masses 40, 44, 57 and 61.Better isotope ratio precision was obtained at CaOH+ masses relative to those for Ca+ because the sensitivity was about 10 times higher. Isotope ratio measurement of K was made at masses 39 and 41. A ratio precision of about 0.2 and 0.5% was obtained for K and Ca, respectively. The results suggest that the FIMS is suitable for the isotope ratio measurement of K and Ca in simple matrices, and that the air–hydrogen flame is a more desirable ion source than an air–acetylene flame for FIMS.populations cannot be dramatically reduced without Introduction extinguishing the plasma. The spectral background of inductively coupled plasma mass Replacing the plasma with a diVerent ion source is another spectrometry (ICP-MS) has been well documented.1,2 While possible solution to the background problem. Zhang et al.10,11 many interferences from the plasma background can be mini- developed a helium inductively coupled plasma mass specmized by desolvation3,4 or by using an Ar–N2 mixed gas trometer.Elements such as K and Fe that suVer from spectral plasma,5,6 solutions to the interferences from Ar+ are surpris- interferences in Ar ICP-MS were detected at 0.1 and ingly few. As a result, isotope ratio measurements of K and 4 pg mL-1, respectively. Mass spectrometers were extensively Ca are severely impaired. One approach that has gained used to study the molecular species in the combustion flame limited success is to minimize the amount of Ar+ being in order to elucidate the combustion mechanism.12–14 extracted by the mass spectrometer.In this eVort, Smith et al.7 Conversely, flames can serve as alternative ion sources to the used an Ar–Xe supported plasma and a high resolution mode plasma for elemental mass spectrometry. Taylor et al.15 used of the mass spectrometer. The background at mass 40 was an air–acetylene flame to this eVect. A lower detection limit greatly reduced and its eVect on K masses was minimized.of K was reported in comparison with that obtained by The interference of 38ArH+ at mass 39 was negligible; neverthe- ICP-MS. Isotope ratio measurements of K were made, and less, the 40ArH+ interference at mass 41 was still the equivalent the feasibility of flame ionization mass spectrometry (FIMS) of about 0.9 ppm K. Despite the 40ArH+ background, the for ground water studies by using 41K as a stable tracer was measured K isotope ratio was reasonably close to the demonstrated. Unfortunately, the organic combustion prodaccepted value.7 ucts of an acetylene flame resulted in instrument contami- Jiang et al.8 virtually eliminated the Ar+ and ArH+ species nation.A significant amount of a wax-like residue was in the spectral background by operating an ICP-MS under observed inside the interface of the instrument after several cool plasma conditions. A detailed study of the ‘cold plasma’ hours of operation.It is likely that this accumulation has a conditions by Tanner9 showed that interferences from Ar+ deleterious eVect on instrument performance, making an air– were minimized whereas those from ArH+ persisted. A non- acetylene flame less suitable as an ion source. Alternatively, linear relationship was observed between the count rate and an air–hydrogen flame can be used for mass spectrometry, as the concentration of K, but the non-linearity was not expected reported by Turk et al.16 Unlike the residue from an acetylene to aVect the isotope ratio measurements.8 A 9% mass bias was flame, the water resulting from combustion was easily observed for the 39K/41K ratio.8 Despite the fact that the removed from the vacuum system with a liquid nitrogen count rate for 39K+ was lowered by about two orders of trap.16 In that work,16 the flame was used as an atom source magnitude, the detection limit was similar to that obtainable and ionization of ground state analyte atoms was with conventional instrument settings.8 Argon ions are an accomplished by laser-enhanced ionization (LEI) with subintegral part of energy transfer to the plasma; therefore, their sequent mass spectrometric detection.At a temperature of about 2300K,17 an air–hydrogen flame is capable of ionizing most of the alkali and alkaline earth elements to a significant degree.18 Therefore, it can also be an eVective ion source for †Present address: Department of Chemistry, University of Missouri at Columbia, Columbia, MO 65211, USA.these elements. J. Anal. At. Spectrom., 1999, 14, 669–674 669the National Institute of Standards and Technology (Gaithersburg, MD, USA) in 0.2 M HNO3 prepared locally by sub-boiling distillation. Sample solutions contained a similar acid concentration. Procedure An air–hydrogen flame was used for all the elements studied except Ca, for which an oxygen-enriched air–hydrogen flame was used. The gas flow rates for the air–hydrogen flame were 4.3 L min-1 for air and 3.7 L min-1 for hydrogen.For the spectral background studies, a mass scan consisting of six sweeps over the entire mass range (1–250 u) was performed with a blank solution. The total dwell time at each mass was 1.8 s. For the molecular ion studies, four single element solutions containing 100 ppm Na, 10 ppm K, 40 ppm Ca and Fig. 1 Schematic diagram of a hydrogen burner and MS interface. 10 ppm Fe were used.Mass scans from 39 to 97 u were performed for each solution. A calibration curve for K was constructed using solutions Table 1 Instrument parameters containing 0, 0.078, 0.38, 1.9, 9.3, 31 and 100 ppm K. For K Mass spectrometer Perkin-Elmer SCIEX Elan 5000 isotope ratio measurements, five replicate measurements were Spectral resolution 0.8 u (normal ) made on solutions containing 32 and 100 ppm K. The total Sample uptake rate 0.94 mL min-1 dwell time for each measurement was 36 and 144 s at masses Ni sampler and Ni skimmer 0.5 mm 39 and 41, respectively.orifice diameter An oxygen-enriched air–hydrogen flame was used for the Spacing between the burner 20 mm studies on Ca. The flow of the oxygen was regulated using the and the sampler Interface pressure 230 Pa (1.7 Torr) oxygen channel of the mass flow controller of the ELAN 5000 Quadrupole pressure 2.9 mPa (2.2×10-5 Torr) ICP-MS. The oxygen entered the burner through the auxiliary oxidant port of the spray chamber.Typical flow rates were 0.20 L min-1 of O2, 3.6 L min-1 of air and 4.6 L min-1 of H2. A calibration curve for Ca was constructed with solutions Experimental containing 0, 1.0, 2.0, 2.7, 3.9, 4.9, 6.5, 8.4, 9.3, 11, 23, 32 and Reagents and equipment‡ 40 ppm Ca. Isotope ratio measurements were made using 10 replicates of a 40 ppm Ca solution. The total dwell time for An Elan 5000 ICP-MS (Perkin-Elmer SCIEX, Thornhill, each measurement was 5, 35, 5 and 35 s at masses 40, 44, 57 Ontario, Canada) and a pre-mix burner and nebulization and 61, respectively. For studies of ionization interferences, a system designed for flame atomic absorption spectrometry blank and five samples were prepared.The five samples from Perkin-Elmer (Norwalk, CT, USA) were adapted for contained 10 ppm Ca, 10 ppm K, 10 ppm K plus 10 ppm Ca, this work. A schematic diagram of the FIMS is shown in 100 ppm Na and 100 ppm Na plus 10 ppm Ca. The count Fig. 1.Model 604 and 603 tapered rotameter flow meters from rates at masses 40, 44, 57 and 61 were measured. Matheson Gas Products (Secaucus, NJ, USA) were used to The detection limit was determined as the concentration control the flows of the air and the hydrogen, respectively. A giving a signal equivalent to three times the noise, calculated capillary burner head was used rather than the normal slotfrom the standard deviation of 11 repetitive measurements type burner head. It was constructed with a bundle of 43, with 3 s integration of the background intensity. 4 cm×1 mm id stainless steel capillaries clustered into 1 cm diameter and mounted in a water-cooled holder in the position where the ICP torch normally is located.The burner head was Results and discussion connected to the spray chamber with a 3 cm id flexible hose. General properties of the hydrogen FIMS A Gilson (Middleton, WI, USA) peristaltic pump was used to regulate the sample uptake.The vacuum pressure of the When a commercial ICP-MS instrument is retrofitted with a instrument was maintained by decreasing the orifice of both sampler of a smaller orifice, the geometry of the supersonic the sampler and the skimmer to 0.5 mm, and by increasing jet changes. It is important to ensure that the skimmer is the pumping capacity by using a RUVAC WAU-250 Roots within the boundary of the Mach disk to minimize interactions blower from Leybold Vacuum Products (Export, PA, USA) that could alter the chemical composition of the sample in series with the instrument roughing pump.Water was extracted from the ion source.14 Some researchers have sugremoved from the vacuum line by a Reliance Glass Works gested that the optimum sampler–skimmer separation for an (Bensenville, IL, USA) Model R-7210 liquid nitrogen trap Ar plasma ion source is two thirds of the Mach disk location installed upstream of the Roots blower. A Convectron Model relative to the sampler.19 The position of the Mach disk was 275 vacuum gauge from Granville-Phillips (Boulder, CO, calculated19 using the measured interface pressure (see Table 1) USA) was also installed to monitor the pressure in the interface and the diameter of the sampler orifice. This calculation places region.Control over the interlocks of the mass spectrometer the location of the Mach disk approximately 7 mm behind the was achieved by using the service mode of the Elantools orifice.Therefore, the skimmer, which was about 6.1 mm from software. Table 1 gives the instrument parameters used. the sampler orifice, was inside the Mach disk. All standard solutions were prepared by appropriate The mass spectrum of a 0.2 M nitric acid solution is shown dilutions of 3100 Series Standard Reference Materials from in Fig. 2. The H3O+, CH2O+, NO+ and NO2H+ peaks found in the background spectrum of an acetylene FIMS were not ‡Certain commercial instruments are identified in this paper to specify observed in the hydrogen FIMS.15 The temperature of an air– adequately the experimental procedure.Such identification does not hydrogen flame is about 150 K lower than that of an imply recommendation or endorsement by the National Institute of air–acetylene flame.17 The lower temperature and a lack of Standards and Technology, nor does it imply that the equipment identified is necessarily the best for the purpose. flame carbon relative to the air–acetylene flame are probably 670 J.Anal. At. Spectrom., 1999, 14, 669–674sensitivity of 4 counts s-1 mM-1 [the equivalent of 100 counts s-1 (ppm)-1 at mass 39] is needed for an analyte. Substitute Ia, Vai and ba in eqn. (6) with the sensitivity of 350 counts s-1 mM-1, the ionization potential of 4.3 eV and atomization factor of 0.4 for K.17 Substituting Ib in eqn. (6) with the limit sensitivity of 4 counts s-1 (ppm)-1 and using a flame temperature of 2300 K, eqn. (6) is reduced to Vbi=6.3+0.46 log bb (7) The ionization potential Vbi has a maximum of 6.3 eV since bb cannot be greater than 1.Hence, the usefulness of the hydrogen FIMS is limited to elements with ionization poten- Fig. 2 Mass spectral background of a blank containing 0.2 M nitric tials below 6.3 eV, which include most of the alkali and acid. alkaline earth elements. If the atomization factor of an analyte is less than 1, the corresponding Vbi for the element is lowered. An example is Ca, which has an atomization factor of 0.15 in directly or indirectly20 responsible for the absence of these an air–hydrogen flame.17 Vbi for Ca is calculated to be 5.9 eV, background species.The background intensity of the hydrogen which is less than the first ionization potential of 6.1 eV for FIMS at any mass was below 14 counts s-1. Data from the the element. This suggests that Ca+ is not sensitive enough to six replicates of a scan showed that the average of the be useful for isotope ratio measurement with the hydrogen intensities at any given mass was smaller than the standard FIMS, and therefore the measurement of Ca has to be made deviation of the intensities at the mass, suggesting that the with other alternatives, as will be discussed later.The limited features observed in the background scan were due primary ionization capability of the hydrogen FIMS is an advantage to statistical noise. The hydrogen flame is, therefore, a clean rather than a disadvantage for the determination of the alkali source for mass spectrometry.and alkaline earth elements, because it promises a simpler The signal intensity of a 10 ppm K solution was about mass spectrum relative to that of ICP-MS for samples of 9×104 counts s-1 at mass 39, yielding a sensitivity of 350 complex matrices. counts s-1 mM-1. With some reasonable assumptions, the sensitivity and hence the eYcacy of the FIMS for the other Molecular ions elements can be estimated by using the ionization potential of the analyte.The ion number density in the hydrogen flame is The primary application of the hydrogen FIMS research is related to the ionization potential of the element by the Saha the isotope ratio measurements of K and Ca for which the equation:21 conventional ICP-MS technique is inadequate because of the isobaric interferences. Na and Fe are present in most sample log [M+]2 [M] =-5040 Vi T +2.5 logT-6.5 (1) matrices; therefore, they were included in the study.Spectra of the four elements using single element solutions are shown where [M+] and [M] are the ion and atom number densities, in Fig. 3. respectively, of the analyte, Vi is the ionization potential of The mass spectrum of a 100 ppm Na solution showed two the analyte in electronvolts and T is the flame temperature peaks at K masses, one at mass 39 with 700 counts s-1 and (2300 K). The mass spectrometric signal intensity is directly the other at mass 41 with 3000 counts s-1 [Fig. 3(a)]. The proportional to the ion number density of the isotope: peak at mass 39 was probably K from the contamination of the burner head after the prolonged use with high concen- [M+]=kI (2) trations of K solutions. The more intense peak at mass 41 where I is the signal intensity and k is the FIMS response cannot be explained by the K contamination because of the constant for the element. The atom number density is directly low natural abundance of 41K.The probable designation for proportional to the free atom fraction of the analyte and the this peak was Na(H2O)+. The background from the molecular molarity of the analyte solution aspirated: ions of the 100 ppm Na solution was the equivalent of about 2 ppm natural K at mass 41. A separation may be necessary [M]=fbC (3) for isotope ratio measurement of K in samples containing where f is a constant of the flame, b is the atomization factor high concentrations of Na.Alternatively, the interference on of the analyte and C is the analyte molar concentration. Let 41K may be avoided by preparing the samples in D2O. us assume that the mass response curve of the spectrometer is The mass scan of a 10 ppm K solution [Fig. 3(b)] showed flat and therefore the response constant k is approximately a peak at mass 57 with 1000 counts s-1 and another at mass the same for all elements. Then for isotopes of element a and 59 with 100 counts s-1 in addition to the peaks at K masses.element b of the same molarity (Ca=Cb), the following applies: The probable species responsible for the peaks were the hydrated analyte ions 39K(H2O)+ and 41K(H2O)+. The intenlog [Ma+]2 [Ma] =log (kIa)2 fCaba =-5040 Vai T +2.5 log T-6.5 sities of K+ at masses 39 and 41 were 1×105 and 1×104 counts s-1, respectively. Assuming a flat mass response curve (4) of the FIMS in the mass range of this study the K+ species were estimated to account for over 99% of the total K log [Mb+]2 [Mb] =log (kIb)2 fCbbb =-5040 Vbi T +2.5 log T-6.5 containing ions.The mass scan of a 40 ppm Ca solution [Fig. 3(c)] showed (5) four groups of peaks in the mass range studied. The first group from masses 39 to 44 included peaks of 39K+ as a Subtracting eq. (4) from eq. (5) we obtain contaminant at 900 counts s-1, 40Ca+ at 3000 counts s-1 and 44Ca+ at 80 counts s-1. The second group included peaks log Ib2 Ia2 =-5040 Vbi-Vai T +log bb ba (6) from masses 57 to 65.Based on the intensities of these peaks relative to one another, the probable identifications of the peaks were CaOH+ species at masses 57 (30 000 counts s-1), For isotope ratio measurements to be practical, a minimum J. Anal. At. Spectrom., 1999, 14, 669–674 671ure relative to the local dew point.22 The gas may become supersaturated at some stage of the expansion, resulting in homogeneous nucleation and the formation of the hydrate ions.22 These hydrate ion species are rarely observed in ICP-MS, probably owing to the high temperature of the plasma and the low partial pressure of water relative to that of Ar.The presence of the hydrates of Na+, K+, Ca+ and CaOH+ complicates the mass spectra of the hydrogen FIMS. Analyte ion–solvent clusters are frequently observed23,24 in the spectra obtained using electrospray mass spectrometry (ES-MS). The extent of the analyte ion–solvent cluster depends on the electric field strength between the sampler and the skimmer.Increasing the voltage promotes collisions in the free jet,23 resulting in diminished hydrate ion intensities relative to singly charged analyte ions.23,24 This de-clustering principle of ES-MS should also be applicable to the hydrogen FIMS, and the hydrate ions in the hydrogen FIMS probably could be minimized by appropriate biasing of the sampler and the skimmer. No attempt was made to do so in the present work. Analytical performance with potassium Taylor et al.15 reported that with an air–acetylene flame, a 100 ppm Cs solution aspirating between each sample analysis was essential to keep the K signal stable.Without the Cs solution the K signal would drift down to the background level in 10–15 min.15 This phenomenon was not observed with the air–hydrogen flame. Calibration standards of K in 0.2 M nitric acid were measured and the intensities were plotted against the concentrations of K. A linear curve (correlation 0.9997) spanning over three orders of magnitude was indicated.The detection limit for K was about 0.2 ppb. This is comparable to the 0.2–0.3 ppb reported by Taylor et al.15 using an acetylene FIMS and it is slightly better than the 1 ppb obtained with ICP-MS;25 however, a much better detection limit of 0.1 pg mL-1 can be obtained with helium ICP-MS.10 Two solutions containing natural K were used for isotope ratio measurements. The mass bias calculated from the measured ratio of 13.51 was about 2.5% lower relative to the accepted ratio of 13.86, compared with the 9% bias observed in a cool Fig. 3 Mass spectra of (a) 100 ppm Na, (b) 10 ppm K, (c) 40 ppm Ca plasma.8 The results for five replicate measurements gave 0.19 and (d) 10 ppm Fe. and 0.14% relative standard deviation (RSD) for 32 and 100 ppm K, respectively. A mass scan of a 1000 ppm K solution is shown in Fig. 4. 59 (300 counts s-1), 60 (80 counts s-1), 61 (1000 counts s-1) The magnitudes of the three peaks at masses 39, 40 and 41 and 65 (60 counts s-1), and 40Ca(H2O)+ at mass 58 (500 were 1.63×106, 238 and 1.19×105 counts s-1, respectively, counts s-1).The third group included peaks from masses 75 yielding abundances of 93, 0.014 and 6.8% (not corrected for to 83. The probable identifications for these peaks were mass bias), respectively. They were in good proportion relative CaOH(H2O)+ species at masses 75 (20 000 counts s-1), 77 to the accepted abundances of 93, 0.012 and 6.7% for the (200 counts s-1), 78 (30 counts s-1) and 79 (300 counts s-1) corresponding K isotopes.The fact that the peak at mass 40 and Ca(H2O)2+ at mass 76 (30 counts s-1). The fourth group is well resolved from those at 39 and 41 suggests that the included peaks from masses 93 to 97. The probable identifi- measurement of 40K is possible. cations for these peaks were CaOH(H2O)2+ at masses 93 Potassium-40 is sometimes used for mineral dating.26 The (1000 counts s-1), 95 (20 counts s-1) and 97 (40 counts s-1).Assuming a flat mass response curve of the FIMS, the relative abundance of the Ca species was about 55, 37, 5, 2 and 1% for CaOH+, CaOH(H2O)+, Ca+, CaOH(H2O)2+ and Ca(H2O)+, respectively. Therefore, CaOH+ was the dominant Ca species in the FIMS. The spectrum of a 10 ppm Fe solution [Fig. 3(d)] showed two major peaks at masses 39 and 56, respectively. The peak at mass 39 was probably K contaminant from the burner head discussed earlier.The peak at mass 56 with an intensity of about 120 counts s-1 was from Fe. Apparently, the high ionization potential of Fe (7.87 eV), relative to the limit of 6.3 eV discussed earlier, was responsible for the low sensitivity. The hydrate ions observed in Na, K and Ca spectra were probably formed during the supersonic jet expansion.22 As the gas expanded from the flame into the vacuum through the Fig. 4 Mass spectrum of a 1000 ppm K solution. sampler orifice, its temperature dropped faster than the press- 672 J.Anal. At. Spectrom., 1999, 14, 669–674previously. A measurement of a 2.7 ppm Ca solution with the air–hydrogen FIMS gave a count rate of about 160 counts s-1, or a sensitivity of about 60 counts s-1 (ppm)-1. Since the rate of ionization is exponentially proportional to the temperature of the flame, the higher the flame temperature, the higher is the sensitivity. An oxygen–hydrogen flame would be more desirable for Ca because of its higher temperature relative to an air–hydrogen flame; however, the oxygen–hydrogen flame is extremely susceptible to flash back.In this work, an oxygen enriched air–hydrogen flame was used to enhance the Ca sensitivity. As the flow of the air was decreased, the flow of the oxygen was increased, and the flow of hydrogen was also increased in order to maintain the linear flow velocity of the gas mixture. This procedure reduced the chance of a flash back. The sensitivity of Ca was increased by a factor of about Fig. 5 Calibration curve for Ca. 8 in the oxygen-enriched flame relative to that in the air– hydrogen flame. Fig. 5 shows the calibration curve of Ca using capability of the FIMS for the measurement of 40K can be signal intensities at mass 40. The curve is linear up to about compared with radiochemical scintillation counting. At the 10 ppm (correlation 0.9993), beyond which it bends toward signal intensity of the experiment, it would take about 70 min the concentration axis.The curvature at concentrations higher to collect 106 counts at mass 40. A 70 mL volume of 1000 ppm than 10 ppm is probably a result of auto-ionization suppres- K solution would be needed with a sample uptake of 1 sion, since the element is highly susceptible to ionization mL min-1 of the current nebulizer. The sample uptake can be interferences to be discussed later. The detection limit was reduced easily without sacrificing the sensitivity or the signal about 2 ppb at mass 40 compared with the 2 ppb at mass 44 stability by using a direct injection high-eYciency nebulizer by ICP-MS.25 As discussed earlier, the sensitivity of CaOH+ (DIHEN).27 An estimated 7 mL of the solution are needed was about 10 times higher than that of Ca+.Consequently, a with a sample uptake of 0.1 mL min-1 of the DIHEN;27 better detection limit of 0.2 ppb was obtained at mass 57. therefore, the calculated volume of 70 mL is the upper limit.A 40 ppm Ca solution was used to demonstrate an isotope The 40K in 70 mL of the solution is about 1.30×1017 atoms. ratio measurement. Signal intensities at masses 40, 44, 57 and Assuming a 100% detector eYciency, the time it would take 61 were recorded. Results of 10 replicate measurements are to collect N counts is summarized in Table 2. The ratios of 0.0241 and 0.0249 obtained for Ca+ and CaOH+ species are about 12–16% t= Nt1/2 N0 ln 2 (8) higher than the accepted ratio of 0.0215 for 44Ca/40Ca.On the per u basis the bias is about 3–4%. Although this mass bias is where t1/2 is the half-life of 40K (1.27×109 years) and N0 the larger than the per u bias of about 1% for K discussed number of 40K atoms in the K solution at the beginning. A previously, it is within the range of the mass bias reported in simple calculation shows that with 70 mL of 1000 ppm K, the literature.8 The discrepancy between the mass bias obtained scintillation counting would take over 5 d to collect the same with K and Ca in this work is probably attributable to the number of counts.Therefore, the mass spectrometric method diVerent flames used for the two elements. has the potential to be more practical than scintillation count- The feasibility of using CaOH+ species for isotope ratio ing for 40K determinations. analysis was evaluated by comparing the isotope ratio and the ratio precision at CaOH+ masses with those at Ca+ masses. Analytical performance with Ca It is expected that the intensity ratio at CaOH+ masses would be identical with that of the corresponding Ca+ masses.The According to the discussion on the general properties of the 3% diVerence between the two ratios in Table 2 is probably hydrogen FIMS, Ca has a sensitivity below the limit of 4 counts s-1 mM-1 [or about 100 counts s-1 (ppm)-1] set forth due to the mass bias since these ratios are not corrected for Table 2 Isotope ratio measurement of a 40 ppm Ca solution Average intensity at mass Ratio 40 44 57 61 44/40a 61/57a 6659 160.7 8.361×104 2085 2.41×10-2 2.49×10-2 RSD (%) 8 9 7 7 3 0.5 aNot corrected for mass bias.Table 3 Intensity and the standard deviation (n=5) of K, Na and Ca solutions at Ca and CaOH masses Samplea 40 (40Ca+) 44(44Ca+) 57(40CaOH+) 61(44CaOH+) Blank 1±0.3 0.8±0.2 4±0.8 1±0.4 Ca 3000±40 70±2 40000±200 900±200 Na 1±0.3 0.9±0.3 8±0.7 0.8±0.4 Na+Ca 50±4 2±0.5 700±10 20±2 K 100±3 0.8±0.3 3000±20 1±0.4 K+Ca 300±3 4±0.9 5000±80 50±2 aNa, 100 ppm Na; K, 10 ppm K; Ca, 10 ppm Ca.J. Anal. At. Spectrom., 1999, 14, 669–674 673the bias. The measured ratio precision of 0.5% RSD at CaOH+ References masses is much smaller than the 10% predicted by the law of 1 M. A. Vaughan and G. Horlick, Appl. Spectrosc., 1986, 40, 434. error propagation with 7% RSD each at masses 57 and 61 2 S. H. Tan and G. Horlick, Appl. Spectrosc., 1986, 40, 455. (Table 2), suggesting that the intensities at masses 57 and 61 3 A. Montaser, H.Tan, I. Ishii, S. Nam and S. Cai, Anal. Chem., are highly correlated. This high correlation substantiates the 1991, 63, 2660. viability of using CaOH+ species for isotope ratio analysis. 4 L. C. Alves, D. R. Wiederin and R. S. Houk, Anal. Chem., 1992, The ratio precision listed in Table 2 is apparently limited by 64, 1164. 5 A. Montaser and H. Zhang, in Inductively Coupled Plasma Mass the counting statistics since the total counts measured at each Spectrometry, ed.A. Montaser, Wiley-VCH, New York, 1998. mass are much less than 106. As a result, a better ratio 6 J. W. Lam and J. W. McLaren, J. Anal. At. Spectrom., 1990, precision was obtained at CaOH+ masses because of the 5, 419. higher sensitivity relative to that at Ca+ masses, thus making 7 F. G. Smith, D. R. Wiederin and R. S. Houk, Anal. Chem., 1991, CaOH+ the species of choice for isotope ratio measurements. 63, 1458. The eVects of easily ionized elements (EIE) Na and K on 8 S.Jiang, R. S. Houk and M. A. Stevens, Anal. Chem., 1988, 60, 1217. Ca signal intensities at masses 40, 44, 57 and 61 are summarized 9 S. D. Tanner, J. Anal. At. Spectrom., 1995, 10, 905. in Table 3. The intensities for single element solutions of Na 10 H. Zhang, S. Nam, M. Cai and A. Montaser, Appl. Spectrosc., and K are also included. The count rate of 100 and 3000 1996, 50, 427. observed at masses 40 and 57 for the 10 ppm K solution are 11 S. Nam, H.Zhang, M. Cai, J. Lim and A. Montaser, Fresenius’ probably due to 40K+ and 39K(H2O)+ species discussed J. Anal. Chem., 1996, 355, 510. earlier. The ionization suppression from Na and K is evident 12 G. C. Eltenton, J. Chem. Phys., 1947, 15, 455. 13 A. N. Hayhurst and T. M. Sugden, Proc. R. Soc. London, Ser. A, when the signal from the Ca solution is compared with those 1966, 293, 36. with alkali elements in addition. A 60-fold and a 20-fold 14 A. N. Hayhurst, D. B. Kittleson and N.R. Telford, Combust. suppression of Ca signal were observed when 100 ppm Na and Flame, 1977, 28, 123. 10 ppm K were added to the 10 ppm Ca solution, respectively. 15 H. E. Taylor, J. Garbarino and S. R. Koirtyohann, Appl. In addition to the multiplicative interferences due to ionization Spectrosc., 1991, 45, 886. suppression, the 10 ppm K in the Ca solution resulted in 16 G. C. Turk, L. Yu and S. R. Koirtyohann, Spectrochim. Acta, Part B, 1994, 49, 1537. additive interferences at masses 40 and 57 due to 40K+ and 17 Handbook of Flame Spectroscopy, ed.M. L. Parsons, B. W. Smith 39K(H2O)+ species. A separation may be necessary if Ca is to and G. E. Bentley, Plenum Press, New York, 1975. be determined in high concentrations of easily ionized elements. 18 Instrumental Methods of Analysis, ed. H. H. Willard, L. L. Merritt, Jr and J. A. Dean, 5th edn., Van Nostrand, New York, 1974. 19 D. J. Douglas and J. B. French, J. Anal. At. Spectrom., 1988, 3, 743. 20 J. A. Green and T. M. Sugden, in 9th International Symposium on Conclusion Combustion, ed. W. G. Berl, Academic Press, New York, 1963, The mass spectral background of the air–hydrogen flame was pp. 607–621. 21 M. N. Saha, Philos. Mag., 1920, 40, 472. very simple; however, the analyte spectra were complicated. 22 F. T. Greene and T. A. Milne, Adv. Mass Spectrom., 1964, 3, 841. The low flame temperature relative to an ICP resulted in a 23 D. J. Douglas, in Inductively Coupled Plasma in Analytical Atomic limited number of alkali and alkaline earth elements becoming Spectrometry, ed. A. Montaser and D. W. Golightly, VCH, New appreciably ionized. The isotope ratio measurement of K and York, 2nd edn., 1992. Ca was successfully demonstrated with the FIMS. Two major 24 G. R. Agnes and G. Horlick, Appl. Spectrosc., 1995, 49, 324. 25 Perkin-Elmer Technical Summary, TSMS-12, Perkin-Elmer, limitations relative to ICP-MS are severe EIE interferences Norwalk, CT, 1991. and the presence of analyte ion-solvent clusters in the spectra. 26 Nuclear and Radiochemistry, ed. G. Friedlander, J. W. Henedy, Although the molecular ions complicate the analytical spectra, E. S. Macias and J. M. Miller, Wiley, New York, 3rd edn., 1981. they can sometimes be used to advantage for elements such as 27 J. A. McLean, H. Zhang and A. Montaser, Anal. Chem., 1998, Ca. Compared with the acetylene flame, the air hydrogen 70, 1012. flame is equally capable while being more practical as an ion source for FIMS. Paper 8/09797F 674 J. Anal. At. Spectrom., 1999, 14, 669–674