Studies with desolvation in inductively coupled plasma-mass ~~~trorn~t~y
Central Research Laboratories,
RYOICWI TWKAHARA Sumitomo Metal Mining Co., Ltd., 3”18-5, Nakakokubun. Chiba, 272 Japan
Lchikawa-shi,
and MASAAKI KUBOTA” NationaL Chemical Laboratory for Industry, I-1, Higashi, Tsukuba-shi, fbaragi. 305 Japan (Received 31 July 1989; in revised form 24 October 1989) Abstract-The influence of water toading in the plasma on the signals of ions has been studied using an inductively coupted plasmamass spectrometer with a sample introduction system consisting of a concentric nebulizer. a spray chamber and a desotvator (i.e. a heater and a condenser). The amount of water vapor mtrodueed into the plasma was changed by changing the cooling temperature of the condenser. The ratios of Ba’+/Ba”, BaO+/Ba+ and ArO+/Co+ decreased with decreasing amount of water vapor. Spectroscopic interferences arismg from rare earth oxide and hydroxide ions were lowered by approximately one order of magnitude with the condenser cooted to WC, compared with those obtained using conventional sample introduction. Rare earth element impurities. such as Trn and Lu suffering interference from GdGH’, in Gd@, were determined with the desotvator. Ion kinetic energies are also given with and without desolvation.
1. ~NTRtXXJCTiON Srwca inductively coupIed plasma-mass spectrometry (KY-MS) is a highty sensitive technique of analysis, the interest in this field has rapidty expanded. Since ICP-MS instrumentation has become commercially available, the number of publications dealing with the influence of operating parameters on ICP-MS performance has greatly increased (l-6]. Only a few authors [7-lo], however, have reported the effect of desolvation in ICI’-MS, although it should decrease the water toading in the ptasma and consequently reduce interferences from species such as oxide ions (MU’) and hydroxide ions (MQH’)[Sf. Electrothermal vaporization (ETV) was reported as a means of sample introduction in ICP-MS by GRAYand DATE [7], and GRAY [8]. They found that polyatomic ion signals associated with argon and/or solvent species such as ArQ’, ArN’, Ar2+ and 02+ observed using ETV introduction were much smaller than those obtained using the conventional pneumatic nebu~ization method. They have not published further investigations of the use of ETV for ICP-MS. HWIBN and EATON [9] investigated the effecr of solvent water loading in the plasma by means of varying the wall temperature of a water-cooled spray chamber. The spectra from singly-charged (M’), oxide (MO +) and doubly charged (M’ +) species showed a strong dependence on water loading. ZHU and BROWNER [IO] also reported the influence of varying water loading in the plasma by a similar manner to that employed by HUTTON and EATON 191. The study was carried out with two different commerciai ICP-MS instruments, a VG Plasma Quad and a Sciex Elan. The water loading exerted a strong influence on the signals of M' , MO +’and M”, although the behavior of these species observed with the VG instrument was very different from that with the Sciex instrument. In this paper, the effect of desolvation on spectral characteristics observed with a commercially avaitabte KY-MS instrument is described.
’ To whom correspondence
should be addressed. 58f
582
R. TSUKAHARA and M. KUB~TA
2. EXPERIMENTAL The instrument used in this work is a SEIKO SPQ-6100s (SEIKO Instruments Inc., Tokyo Japan). The system, with a conventional sample introduction device, is described elsewhere [ll]. The ICP is essentially the same as that employed in atomic emission spectrometry (AES), but a Fasset torch was mounted ho~~ontaIly, similarly to other commercial instruments. A schematic diagram of the sample introduction system used in this work is shown in Fig. 1. Aqueous solutions were nebuiized into a Scott-type spray chamber with a Meinhard-type TR-30-C2 concentric nebuli2er, and desolvated by passing through a desolvator consisting of a heated glass tube and a modified Liebig condenser. The heater temperature was kept constant at 110°C. Water containing 40% ethylene glycol was used for cooling the condenser. The temperature of the cooling water, which was controlled from a recirculator, was varied from -7 to 25°C. Unless otherwise stated, the operating conditions given in Table 1 were used. The greatest ion signals could be obtained at central gas flow rates (flow rates of carrier gas to the plasma) of 1.3 to 1.5 1 mini, which are higher than those for a conventional sample introduction device (CSID) consisting of a Meinhard-type concentric nebulizer and a Scott-type spray chamber without a desolvator, The nebulizer used in this study worked satisfactorily at a lower gas flow rate (1.0 1 min-i). Thus, to attain the greatest signals, another argon gas flow line was attached to the drain tube of the spray chamber as shown in Fig. 1. This did not disturb the drain flow and nebulization. Prior to signal measurements, the central gas flow rate. i.e. sum of the nebulizer gas and the additive gas flow rates, was measured at the exit of the desolvator at various combinations of the two argon gas flow rates. Solutions of Co, Ba and Pb were prepared by dilution of 1000 pg ml-’ commer~ialty avaiIable standard solutions. A mixed solution containing 100 ng ml-t of each element was used for most
Fig. 1. Sample jo~rod~c~on system with a desolvator combined with a concentric nebulizer and spray chamber. (A) Nebulizer, (3) spray chamber. (C) heater controiler, (D) ribbon heater. (E) thermometer, (F) condenser, (C) coolant recirculator.
Table
I Typical operating conditions
ICP radio frequency power (kW)
I.25
refrected power (W) Central gas flow rate with desolvatlon (I min-‘) without desolvation (I min-‘) Outer gas flow rate (1 min. ‘) Auxiljary gas flow rate (r min-‘) Sampling depth* {mm) Cooling temperature of desolvatlon (“CI Ion lens settings
5
* Distance between sampfmg orifice and top of load coil.
I.3 10 15 1.1 5.9 0 optlmised for T’b
Studieswith desolvationin ICP-MS
583
experiments. Rare earth element solutions were prepared by dissofving high-purity rare earth oxides delivered by Shinetsu Chemical Co., Ltd. with a small amount of nitric acid.
3. RESJLTSAND DISCU~SX~N 3-l. Amount of water loading in plmma The water delivered to the plasma with the sample introduction system shown in Fig. 1 was collected with a glass tube filled with sifica gel which was attached to the exit of the desolvator, and the amount of the water was determined from the difference between the weights of the glass tube before and after collection, as reported by FUJBHIRO 1121. The amount of the water loading with a CSID was also investigated by means of the measurement of the decreasing amount of water in a recycling nebulization system, where the drain tube of the spray chamber was connected to the solution uptake tube. The plots in Fig. 2 show the change in the amount of the water loading with varying desoivator cooling temperatures. The amount of the water loading with the desohator set at a temperature of 0°C (5.0 mg min-‘) was ca. 10% of that observed with the CSID (47 mg min-‘1. so 45
____________-_ E
Fig. 2. Dependence of amount of water loading cm cooling temperature of the desotvator (0). The amount of water loading without the desolvator is shown as a dotted line.
The signal response behavior of s9Co+,r%a+ and Z’“Pb” with varying central gas fIow rates was investigated using the sample introdu~ion system shown in Fig. 1, Cobah and Pb were chosen as test elements because of the large difference in their mass. Another reason for the selection of 59Co is its closeness in mass to ‘@Ar’Q. Barium was chosen because its mass is almost midway between Co and Pb, and it forms measurable amounts of M2+ and MO’ species. The plots in Fig. 3 show the behavior of these anaiysis ion signals with varying central gas flow rate, at different cooling temperatures of the desoivator. With a low central gas flow rate, such as 1.00 1 min- I, the most intense signals were obtained at a higher cooling temperature. With a central gas flow rate of 1.31 1 min-I, the cooling temperature giving the most intense signal reduced as the analytic mass increased. According to an investigation with a VG instrument reported by ZHU and BROWNER flfl], a similar mass dependence on spray chamber cooling temperature was found, but data with a Sciex instrument showed that signals simply increased as the cooling temperature increased from 0 to 35°C and the response characte~stics for ail elements examined (I.& Mg, Cr, Co, MO, Cd, W and Pb) were very similar. HUTTON and EATON ?A@)Q:ko
r
Central
gas flow rate,
Fig. 3. Dependence of ion signals of Wo+, 93a* and zcwPW on central gas flow rate at various cooling temperatures of rhe desolvator: 0°C fO), 7°C f# and 21°C fN).
[9] reported that M” ion intensities increased dramatically on cooling the spray chamber from 30 to O*C. This trend seems to be consistent with our results observed at a higher gas flow rate (1.44 1 mini”’ in Fig. 3). There may be little value, however, in comparing the experimentaI results between the systems, because the sample ~~trod~~t~on systems are quite different- The desolvator in this work, for example, has a heater while that of Hvno~ and EATONdid not. 3.3. Pdyatomic ion interference reductian Although generally spectra in ICP-MS are much simpler than in ICP-AES, a number of peaks of M 2tt MO’ and ~lyatorni~ ion species associated with argon and solvent species occur, and these may often interfere with the M” peaks Spectral ~nterfe~n~e problems that arise from pofyatomic ions especially, in the low mass region, are well known [8,13], and MO+ and MOH+ species also result in the same problems [3,14]. Desolvation using a desolvator, such as in this work, or using an ETV is expected to be effective in reducing the levels of these species [S]. In this study, the signal responses of rJgBaZ*, 13xBaW-r- and 4~~Ar~~O~,as well as singly-charged anafyte ions were investigated with varying central gas flow rate or cooling temperature. The behavior of the ratios of Ba2+/Ba+, BaO+/Ba+ and ArO+/ Co+ is shown as a function of central gasflow rate in Fig. 4 and as a function of cooling temperature in Fig. 5. Doz.&y &urged ions. Increasing central gas flow rate resulted in an increase in the ratio of BaZ”iBa’ as shown in Fig. 4, This trend is consistent with that ~nvest~~ted with a CSID. The behavior of W’iM’ in ICP-MS without desolvation has been reported by various authors. GRAY et a/. fl5] found that increasing central gas flow rate or water loading in the plasma at a constant radio frequency power generally induced an increase in plasma potential and a corresponding increase in M2+ abundances, This agrees with the result of Mz’IM’ observed in this work. VAUCI~A~J and FIORLKKf3fl on the other hand, reported an opposite trend and they sngg~sted that the cooling effect of central gas on the plasma can perhaps explain the behavior they observed. Although the reasons for such disagreement are not immediately explainable, it should be taken into account that the instrumental systems and conditions used were different. Our instrumentation is similar to the VG type used by GRAYef al. ]15] rather than the Sciex type used by VAUGHANand H~RLICK f3]. Under the conditions used and with the ~ns~rne~t employed in this work, the effect of plasma potential might be mure dominant over Ba2+ abundance. As seen in Fig. 5, the ratio Ba’+/Ba+ decreased as the cooling temperature dropped from 15 to SC, consistent with the decrease in the amount of water loading in the plasma. This trend agrees with the behavior of the ratio of Ba*+/Ba+ with varying spray chamber temperature,
585
Studies with desolvation in ICP-MS
.”
0.078
‘;,
0004 0
:,
-
c
0 04
(’
B$’
-
Central
Fig. 4. Dependence
gas
tlow
rate,
I
min”
of 4”Ar1hO+/59Co+. ‘3XBa’60+/‘3XBa+ and ‘zxBa*+/‘3xBa+ on central gas flow rate.
-10
0 Cooling
Fig. 5. Dependence of J”Ar’hO+/“Co+,
10 temperature,
20
30 ‘C
‘3XBa’60+/lZXBa+and I’xBaz+/lzxBa+ on cooling temperature of the desolvator.
which was observed using a VG instrument [9,10], while the data investigated with a Sciex instrument showed that the ratio Ba*‘/Ba+ was independent of spray chamber temperature [lo]. It is known that the plasma potential in an ICP is more negative when water is not introduced [15]. In both our instrument and the VG instrument, the change in plasma potential caused by the decrease in the water loading may affect the ion extraction process, which is obviously of importance to the formation of various species including doubly-charged ions. The ratio Ba’+/Ba’ obtained under the conditions given in Table 1 was 0.015, which was approximately one half that obtained with the CSID. Oxide ions. On increasing the central gas flow rate from 1.3 to 1.5 1 min-‘, the ratio BaO+/Ba+ increased significantly as shown in Fig. 4. This trend is in agreement with that observed with ICP-MS systems without a desolvator [l-6,8]. The ratio BaO+/ Ba’ fell with decreasing cooling temperature as shown in Fig. 5. The cooling effect of central gas, or water perhaps, causes an increase in BaO+ abundance as VAUGHAN
R.
586
TSIJKAHARAand M. KUBOTA
and HORLICK[3] suggested. While decreasing water, which is a source of oxygen, is expected to result in a decrease in BaO+ formation. The ratio BaO+/Ba+ attained under typical conditions was 1 x 10e4, which was one order of magnitude lower than that obtained with the CSID. This improvement factor is almost the same as the reduction ratio in the degree of water loading with and without the desolvator. Consequently, the ratio BaO+/Ba+ seems to be directly related to the water loading in the plasma. Polyatomic ion species. 40Ar160+ is one of the most undesirable species of those associated with argon and solvent because it interferes with the measurement of 56Fe+. The behavior of ArO+/Co’ with varying central gas flow rate (Fig. 4) and varying cooling temperature (Fig. 5) is similar to that of BaO+/Ba+, and is perhaps explainable by the same logic. Under typical conditions, the equivalent concentration of the signal ratio of 4’1Ar160+ to 59Co+ was 20 ng ml-‘, while that observed using the CSID was 120 ng ml-l. 3.4. Determination of rare earth elements Since desolvation was found to reduce interferences from co-existing ions such as MO+, the effectiveness of this technique on the analysis of rare earth elements was investigated. Determination of rare earth elements in the presence of other rare earth elements is generally difficult because they are difficult to separate owing to their similarity in chemical properties. ICP-MS is expected to enable easier determinations without chemical separation and/or pre-concentration because of its high selectivity and sensitivity. However, MO+ and MOH’ species arising from a matrix of rare earth elements produce serious interference problems, and consequently make limits of determination less favorable than expected. The plots in Fig. 6 show the influence of sampling depth on the ratios 14*NdZ+/ 14*Nd+ , 142Nd1hO’/14*Nd+ and 148Nd1601H+/14XNd+ for a 1 pg ml-’ solution of Nd with and without the desolvator. The signals of NdOH’ at larger sampling depths were comparable to the background level, whence the ratio NdOH+/Nd+ plotted is only approximate. With the desolvator, increasing sampling depth influences the ratios in the following manner: Nd’+/Nd+ decreases and then increases with increasing sampling depth. NdO+/Nd+ decreases at smaller sampling depths. NdOH+/Nd+ is almost constant. The behavior of the ratios obtained with the CSID (Fig. 6B) is similar to that obtained with the desolvator, except that the ratio Nd*‘/Nd+ simply increased with increasing sampling depth. The observation, i.e. an increase in the ratio Nd’+/Nd+ with increasing sampling depth, is likely to come from the temperature distribution existing in the plasma. FURUTAand HORLICK[16] reported that the excitation temperature rises with increasing observation height from 5 to 15 mm above load coil.
0
6.0
6.0
10.0
121)
Sampling
6.0 depth, mm
6.0
100
12.0
Fig. 6. Dependence of ‘4zNd’+i’42Nd+ (O), 14’NdihO+/142Nd+ (0) and ‘4XNd’hO’H+/‘JKNd+ (W) on sampling depth (A) with and (B) without desolvation.
587
Studies with desolvation in ICP-MS
The effect of desolvation on these interferences was a one order of magnitude in the ratios Nd2+/Nd+, NdO+/Nd+ and NdOH+/Nd+ compared with those of CSID. With solutions containing a matrix element (100 pg ml-’ as oxide) and 0, 10, 100 or 1000 ng ml-’ of analytes, analyte equivalent concentrations of interferents (IEC) were investigated with and without the desolvator. Here the IEC is the analyte concentration which gives a signal equivalent to an interfering signal. The values of IEC at analyte isotope masses selected to be least influenced by the MO+ or MOH+ species from the matrix elements, are shown in Table 2. The IEC values obtained with the desolvator were generally one order of magnitude lower than those obtained without desolvation. Rare earth element impurities in GdzO, were determined by ICP-MS with and without desolvation. The sample solution was prepared by digestion with HN03, and the solution finally contained 100 kg ml-l of sample and 1.4 M of HNO,. Among the elements, such as Tm, Yb and Lu, which are overlapped by GdO’ and GdOH+ peaks, Yb could not be determined even with desolvation (Table 3). However, the concentration of Lu was found to be lower than 20 pg g-’ by using the desolvator, while this determination was impossible without desolvation owing to a serious interference from reduction
Table 2. Effect of desolvation on the reduction of spectroscopic interferences in the determination earth elements in Nd,O,, Sm,O,, EuzOz and GdzOA
Matrix
Analyte
Nd Nd Nd Sm Sm Sm Eu Gd Gd
‘“9Tb ‘“‘DY insHo IhSHo
Interfering species
IEC* (ng ml-r) With desoivation Without desolvation 18 4.1 0.3, 4.2 1.0 0.7, 2.7 56 1.4
‘h’Er ‘&pTrn rhs’rn ‘7JYb 175Lu
of rare
156 63 4.9 83 24 19 122 637 20
* interference equivalent concentration expressed as the analyte concentration which gives a signal equivalent to an interfering signal of a 100 yg ml-’ matrix eiement oxide solution. Table 3. Analytical results of impurities in Gdz03 with and without desolvation Impurities (fkg g-r) Analyte J”Sc *vY ‘39La r4Ve ‘JrPr ‘**Nd ‘?irn 151EU lsVTb In3Dy lhsHo ‘=Er rTrn ‘=Yb “5Lu * Serious interference
With desolvation
Without desolvation
R. TSUKAHARA and M. KLJBOTA
588
15*Gd1601H+. Also, the value for Tm obtained with desolvation was lower than that obtained without desolvation because of the interference by rs2Gdi60’H+. These examples prove the usefulness of desolvation in routine analysis of these samples. 3.5. Ion kinetic energy The ion kinetic energy (KE) in the ICP-MS system was determined by observing the dependence of signals on pole bias, which is a positive potential applied to the quadrupole mass filter to reduce KE and to consequently obtain higher resolution in the mass spectrum. Ion signals of *5N2+, 3XAr+, 40Ar’60+, 59Co+, 40Ar2+, 138Ba+ and *08Pb+ were measured with varying pole bias and the pole bias giving a negative peak in the curve obtained by differentiating ion signals vs. pole bias was calculated as KE, as reported by FULFORDand DOUGLAS[17]. Under typical conditions, but with no central gas, the KE of Ar’ was found to be 3 eV. OLIVARESand HOUK [18] reported that the KE of Ar+ was 2 eV under almost the same conditions. They also found that the KE of Ar’ was 14 eV at a central gas flow rate of 1 1 min.-’ with desolvation, but details of the desolvation device were not reported. In this work, the KE of Ar+ was 11 eV at the same central gas flow rate with the desolvator operating at a cooling temperature of 20°C. Taking account of the differences in instrumentation and operating parameters, the KE values obtained in this work seem to be in reasonable agreement with those obtained by OLIVARESand HOUK [18]. Figure 7 gives the behavior of the KE at various masses obtained with desolvation at cooling temperatures of 0 and 20°C and also without the desolvator. In the experiment, the central gas flow rate with the desolvator was adjusted to 1.0 1 min-‘, which was the same as that of CSID. At a cooling temperature of 0°C slightly lower KE values were obtained at lower ion mass. This behavior cannot be clearly explained. The KE values observed with the desolvator at a cooling temperature of 0°C were smaller than those obtained at 20°C and much smaller than those obtained with the CSID. The KE is also likely to depend on the water loading. The dependence, however, is not a proportionality, because the difference of the KE values between operation with desolvation at 20°C and with the CSID is much greater than that between operation with desolvation at 20°C and at 0°C. However, it should be noted that the difference in the amount of water loading for the former case is only twice as large as for the latter one. One of the reasons for this might be the state of the water: the desolvator supplies water vapor, while the CSID system gives water aerosol and vapor. LONG and BROWNER[19] suggested that water aerosol requires energy for vaporization, affecting the properties of the plasma. The influence of water loading on KE was also discussed by OLIVARESand HOUK [18], who found that the KE values determined with the nebulizer on were quite similar to those obtained with the nebulizer off. If the
Moss,
Fig. 7. Ion kinetic
energies with the desolvator
m 2-l
at 0°C (0) (0).
and at 20°C (0)
and without the desolvator
Studies with desolvation in ICP-MS
589
data in the present work indicate that the KE is more dependent on the amount of water aerosol than on that of water vapor, the following explanation might be suggested. OLIVARES and HOUK employed an ultrasonic nebulizer with a desolvator and so the state of the water introduced into the plasma in their work must be virtually pure is operating. Even with the nebulizer turned off, water vapor, when the nebulizer vapor would probably still be supplied because of evaporation of residual water in the spray chamber. Since the amount of water vapor exerts less influence on KE than does aerosol, only a slight difference might be obtained when the nebulizer is either on and off.
REFERENCES [l] G. Horlick, S. H. Tan, M. A. Vaughan and C. A. Rose, Specrrochim. Acra 40B, 1555 (1985). [2] S. E. Long and R. M. Brown, Analyst 111, 901 (1986). [3] M. A. Vaughan and G. Horlick, Appl. Specrrosc. 40, 434 (1986). [4] G. Zhu and R. F. Browner, Appl. Specrrosc. 41, 349 (1987). [5] H. Kawaguchi, T. Tanaka and A. Mizuike, Specrrochim. Acra 43B, 955 (1988). [6] A. L. Gray and J. G. Williams, J. Anal. At. Specrrom. 2, 599 (1987). [7] A. L. Gray and A. R. Date, Analyst 108, 1033 (1983). [8] A. L. Gray, Specrrochim. Acra 41B, 151 (1986). [9] R. C. Hutton and A. N. Eaton, J. Anal. At. Specrrom. 2, 595 (1987). [lo] G. Zhu and R. F. Browner, J. Annl. Ar. Spectrom. 3, 781 (1988). [ll] J. Takahashi and R. Hara, Anal. Sci. 4, 331 (1988). [12] Y. Fujishiro, J. Specrrosc. Sot. Japan 25, 229 (1976). [13] S. H. Tan and G. Horlick, Appl. Specrrosc. 40, 445 (1986). [14] A. R. Date, Y. Y. Cheung and M. E. Stuart, Specrrochim. Acra 42B, 3 (1987). [15] A. L. Gray, R. S. Houk and J. G. Williams, J. Anal. Ar. Specrrom. 2, 13 (1987). [16] N. Furuta and G. Horlick, Specrrochim. Acra 37B, 53 (1982). [17] J. E. Fulford and D. J. Douglas, Appl. Specrrosc. 40, 971 (1986). [18] J. A. Olivares and R. S. Houk, Appl. Specrrosc. 39, 1070 (1985). [19] S. E. Long and R. F. Browner, Specrrochim. Acra 43B, 1461 (1988).