05L7&3547/88 s3.00 + .oo Pergamon Presspk.
Sprctrochimica Acro.Vol.43B.No. 8,pp.955-962. 1988. Printedin GreatBritain.
Continuum
background in inductively coupled plasma mass spectrometry HIROSHI KAWAGUCHI, TOMOKAZU TANAKA and ATSUSHI MIZUIKE
Faculty of Engineering, Nagoya University, Furo-cho, Chikusa-ku, Nagoya, 464 Japan (Received 24 November 1987; in revised form 1 March 1988) Abstract-A laboratory constructed ICP mass spectrometer has been studied to reduce the continuum background to achieve better signal to background ratios. An optical baffle plate of 6 mm diameter is found to be the best with a cone-like lens cap of 8 mm diameter. Varying the operating conditions for the plasma and mass spectrometer affect the background intensity. Increasing the voltages applied to the ion lens elements always increases the background. A weak discharge in the lens elements is considered as the cause of the increase. The variation of background with the voltage applied to the ion deflector placed in front of the detector suggests that charged species partly contribute continuum background. Detection limits for several elements under compromised conditions are reported.
1. INTRODUCTION
REDUCING background intensity as much as possible is important in developing an instrumental analytical method, because the detection capability of the method is usually limited by the signal-to-noise ratio and the noise generally increases with the background intensity. This is also true in inductively coupled plasma mass spectrometry (ICP-MS). The basic background spectral features and polyatomic species observed in ICP-MS are fairly well documented in the literature [l, 21. In the background spectra, there are several strong ion peaks derived from argon and water vapor such as Ar +, ArH ‘, 0 +, OH + and OH:. Besides these ion peaks, there is always a continuum in the background spectra. Although vacuum ultraviolet (v.u.v.) photons from the plasma are considered to be the main origin of the continuum background in ICP-MS [3,4], the nature of the background has not been reported in detail. The channel electron multiplier, which has been generally used in ICP mass spectrometers, is blind to visible and U.V.radiation but quite sensitive to photons in the V.U.V. region, especially below 120 nm [S]. Argon has strong resonance lines in this wavelength region, i.e. Ar I 104.82 nm and Ar I 106.67 nm [6]. A disk-like optical balIle plate is usually placed on the axis of the ion lens to prevent these photons from striking the detector. Furthermore, the ion detector is offset from the quadrupole axis in most of the instruments described. With commercial instruments, background count rates are suppressed down to l-20 [7] or 30-100 [8] counts/s (cps). OLIVARES and HOUK [9] constructed an ICP mass spectrometer with an off-axis detector using no optical balIle plate. In the absence of the photon stop, they obtained a high ion transmission efficiency: the analyte count rates were (l-5) x lo6 cps for pg/ml analyte solutions. The background, however, was lOOO-1OOOOcps with their instrument and, therefore, the detection limits were reported to be poorer than those of commercial ones. Previously, we constructed an ICP mass spectrometer [lo, 1 l] similar to that of OLIVARES and HOUK [9] but with an optical baflle plate of 2 mm diameter in the axis of the ion lens. The background, however, was still high, i.e. about 5000 cps. In this paper, we will describe optimization of the sizes of the optical baffle plate and the lens cap cone which was reported previously [l 11. Behavior of the continuum background as well as analyte signals was studied with the variation of the operating parameters for the ion lens, the mass analyzer and the ICP.
2.
EXPERIMENTAL
2.1. Instrumentation
The ICP mass spectrometer used in this work has been described elsewhere [ 10, 111. For the present setup, a copper sampling orifice of 0.95 mm diameter and a brass skimmer of 0.5 mm diameter were used. The distance between them was 9 mm. A scale diagram of the electrostatic cylindrical ion lens as 955
H. KAWAGUCHIet al.
956
well as orifices are shown in Fig. 1. The first element of the ion lens was made of No. 16 mesh stainless steel screen to provide fast pumping of neutral species [3]. A lens cap cone made of brass or stainless steel was attached to the mouth of the first element as described previously [ 111. As a photon stop, a stainless steel disk was positioned in the center of the first element with stainless steel wire (0.5 mm diameter). The cone and the disk were kept at the same voltage as the first element. The fourth element consisting of a tube, 3 mm i.d. and 8 mm long, has another function as an aperture of the third stage of differential pumping. Unless otherwise stated, the ICP operating conditions given in Table 1 were used throughout this experiment. By properly grounding the upper end of the load coil close to the sampling orifice, the erosion of the orifice by sputtering and the split peaks in mass spectra reported previously [lo] were no longer observed. 2.2. Reagents Analytical grade reagents were used without further purification. A solution containing 0.1 pg/ml each of Co, La and Bi in 1 ‘A HNO, was used throughout the measurements of the effects of the various operating conditions. Detection limits were measured with solutions containing 0.03 pg/ml of each element in 1% HN03. 2.3. Procedure Effects of various operating conditions were measured by acquiring a mass spectrum over a range between about 53 and 212 m/z under each set of conditions and introducing an analyte solution. While the mass spectrometer was being scanned continuously rather than stepwise with a cycle time of 1.1 s, signals were acquired by a multichannel scaler (Canberra, Series 35, 2048 channels). The sweep of the scaler was triggered by the start pulse of the scanning of the mass spectrometer. A memory group of 512 channels of the scaler was used in this work with a channel dwell time of 2 ms. Integrated mass spectra
pomp
pump
Fig. 1. Scale diagram of ion extraction interface and lens system. 1 sampling orifice, 2 skimmer, cap cone, 4 optical baffle plate, 5-8 first to fourth lens elements.
Table 1. Instrumentation
and ICP operating
conditions
Mass spectrometer
ULVAC Corp., MSQ-400, quadrupole, range &400, rod bias ground, resolution amu of base width
Plasma
Shimadzu, HPS-2, frequency 27.12 MHz, forward power 1.2 kW, reflected power < 5 W
generator
Torch
Conventional
Gas flow rate Nebulizer
Outer 13, intermediate 1.0, carrier 0.9 l/min Glass concentric, with spray chamber
Sample uptake Sampling
depth
rate
three-tube
mass 0.95
torch
2.5 ml/min 8 mm from outer end of load coil
3 lens
957
Continuum background in ICP mass spectrometry
were recorded by carrying out 60 consecutive sweeps. The data were transferred to a 16-bit computer and stored on disks for processing at a later date. Count rates of the analyte ions were calculated by dividing the sum of the counts of three channels at each ion peak by the total counting time, i.e. 0.36 s, and subtracting background. Therefore, the count rates of the peaks thus calculated are smaller than those measured by stopping the scan at the peaks; the latter procedure was used only in the determination of detection limits. The background count rate for each ion peak was calculated similarly by using the sum of the counts of five channels near the corresponding analyte peak.
3. RESULTS AND DISCUSSION 3.1. Effect of size of the optical bafJle plate Although an optical batlle plate of 2 mm diameter was used in previous work [lo], it was considered too small to stop photons from the plasma effectively. Therefore, larger plates, 4, 5, 6 and 7 mm in diameter, were examined in the present experiments with a lens cap cone having an 8-mm orifice diameter. For each size of the bafAe plate, the voltage setting of each lens element was first roughly optimized for maximum signal-to-background ratio (SBR) of analyte ions by observing analog signals on a cathode-ray tube screen. Then the variation of ion count rates of analytes, Ar:, ArO+, and background was measured as a function of voltage of the first element of the ion lens. As an example of the results, variations of the cobalt ion signal, the background near the cobalt ion peak, and the SBR as a function of the voltage of the first lens element are shown in Fig. 2 for four sizes of the optical baffle plates. Generally, both the analyte and background count rates decrease with increasing diameter of the optical bafRe plate. The background, however, decreases more rapidly than the analyte signals. Although the SBR curves for Co, La and Bi differ slightly from each other, the best SBR was obtained around 6 mm for all analytes. After the voltage setting of the first element was readjusted for the best SBR, the voltage setting of the third element was examined. The voltage setting of the fourth element was then examined similarly. The second element was set always at 0 volt, where the best results were obtained. Variation of signals and background as a function of voltage settings of the first, third and fourth elements of the ion lens are shown in Fig. 3. Note that the count rates are plotted on a logarithmic scale. Although each analyte signal has a maximum response in the examined voltage range of all the lens elements, the background continuously increases with increasing negative voltage. As a cause of the increase in the background with increasing lens voltages, a discharge within the lens system must be considered. Although the discharge was not noticed by
X10’ 10 r
NET SIGNRL
B~CKCROUND4MM /
6-
i
x102
i
5 1
4
in
zit RRTLO
7MM
1st LENSELEMENT VOLTRGE. V Fig. 2. Effect of size of the optical baffle plate on signal to background ratio. Analyte: Co 0.1 pg/ml
H. KAWAGUCHI et al.
958
102. __-J 10’ 1st
.
LENS ELEMENT
lOJ-----0 - 50 -100 VOLTRGE. V Fig. 3. Effect of voltages
-150
B.G. .-v’ ..J
3rd LENS ELEMENT 100; 0
-50 -100 VOLTAGE. V
-150
4th
LENS ELEMENT
lOOL-----0 -100 -200 VOLTAGE. V
of ion lens elements on various ion signals and background. plate diameter: 6 mm, La: 0.1 pg/ml.
Optical
baRle
observing the lens current, there could be a weak discharge like a Townsend discharge. V.U.V. photons will be generated by the discharge and cause the background increase. Analyte signals, background readings and their ratios at the optimum lens settings are summarized in Table 2 for each diameter of the optical bafile plate. The continuum background decreased to less than 100 cps when the optical baffle plates are larger than 6 mm. Therefore, the 6-mm baffle plate was used hereafter. The background usually increased with decreasing m/z, probably due to the increase in scattered ions; abundant background ions derived from argon and water vapor are present below 41 m/z. 3.2. Effect of diameter of the lens cap cone As described previously [ll], a lens cap cone was effective to reduce the continuum background. In the present experiments, orifice diameters of the lens cap from 4 to 8 mm were tested for the best SBRs. With the 6-mm optical bafRe plate, there were no significant differences in the signal and background intensities with lens caps from 6 to 8 mm, but analyte intensities decreased considerably with the 4-mm lens cap. Therefore, the 8-mm cap was used throughout this work. 3.3. Effect of resolution setting of the quadrupole The resolution of the quadrupole mass spectrometer can be changed by varying the ratio of the d.c. to rf voltages applied to the quadrupole. This change also affects the SBRs. Variation of analyte signals and background as a function of the resolution setting are shown in Fig. 4. The resolution setting is calibrated by the base width at 5 “/, of the peak height for ion peaks in the mass spectra. As shown in Fig. 4, analyte signals increase with decreasing resolution, i.e. with increasing base width, while the background is almost independent of the resolution setting. This is reasonable because the transmission of the scattered ions as well as photons through the quadrupole may not be affected by the resolution setting. 3.4. Effect of ion deflector voltage of the detector A Channeltron multiplier Model 4870 (Galileo Electra-Optics Corp.), which was employed in the present instrument, is designed to be used with an off-axis arrangement and has a rectangular ion deflector at the ion entrance. Although the deflector was used usually at ground, the positive voltage was applied to observe the variations of analyte and background count rates. The results are shown in Fig. 5. As the deflector voltage was increased toward positive, both analyte and background count rates decreased in a similar way. The change of the background is not large enough to draw definitive explanations, but it may suggest that
959
Continuum background in ICP mass spectrometry Table 2. Effect of size of optical baffle plate on SBR. Analytes: 0.1 pg/ml
co
Bi
La -
Diameter pf ba!Ile plate (mm)
Signal @PSI
BG @PSI
SBR
Signal @PSI
BG (cps)
SBR
Signal (cps)
BG @PSI
SBR
4 5 6 7
69 500 59 500 39000 4450
212 147 72 10
330 405 540 445
18900 16000 9800 1800
130 83 20 8
145 190 490 225
7000 5700 5700 1175
108 60 17 5
65 95 335 235
B.G. ~~_*-.C__c_.+_-10’.
1oal 0.5
0.7
0.9
1.1
1.3
BRSE YIOTH AT 5% PERK HEIGHT. MASS
Fig. 4. Effect of resolution setting of mass spectrometer on various analyte signals and background. Analytes: 0.1 pg/ml.
10’
----.--+
B.G. \.,.-
t lOOA 0
OEFL::ToR
“OL:&.
V
150
Fig. 5. Effect of voltage of ion deflector on various ion signals and background. Co and La: 0.005 pg/ml. sA(B) L3:8-F
960
H. KAWAGUCHI et al.
the background is caused in part by charged species, because photons cannot be deflected by the deflector voltage. 3.5. Effect of sampling depth and carrier gas Jlow rate Although the effect of the plasma operating parameters on the analyte signals in ICP-MS has been reported by various authors [l, 10, 12, 131, the results are not necessarily consistent with each other except that the carrier gas flow rate is the most important parameter. In Fig. 6, the variation of the analyte ion counts (Co’ and La+), the background of La+, the molecular ion counts (ArO+ and Ar: ) and the ratios of LaO+/La+ and La”/La+ vs carrier gas flow rate is shown for sampling depths (distance of sampling orifice from the end of the load coil) ranging from 4 to 10 mm. The behavior of analyte ion count vs carrier gas flow rate changes considerably as the sampling depth is changed. If the carrier gas flow rate is set at 0.7 l/min, the analyte signal increases as the sampling orifice is moved closer to the load coil. At 1.0 l/min flow rate, however, the signal decreases with the same movement of the orifice. Moreover, the behavior of Co+ is different from that of La+ at each sampling depth. A complex parameter.behavior for analyte signals was also reported by HORLICKet al. [12]. The continuum background, however, changes in a relatively simple manner: it decreases with increasing carrier gas flow rate for all the sampling depths examined. The molecular ions, ArO’ and At-:, behave differently from each other and also from analyte ions. The behavior probably reflects the mechanism by which these ions are generated. The behavior of the ratios of oxide and doubly charged ion counts to singly charged ion counts for lanthanum are interesting to note. Although the ratio of oxide to singly charged ions for Ba, La, Ce, Th and U was reported always to increase with carrier gas flow rate [l, 9, 10, 131, the ratio of doubly to singly charged ions for these elements was reported to increase by some authors [9, 10, 131 or to decrease by others [I]. The differences in grounding method of the load coil [ 143 and therefore, the differences in characteristics of the secondary discharge occurring at the sampling orifice may be one of the reasons of the inconsistency. In the present instrument, however, a tiny discharge was visible at the mouth of the orifice only when an extremely high carrier gas flow rate was used or the plasma was moved far away from the sampling orifice. Although both the ratios of oxide and doubly charged ion counts to singly charged ion counts increase with increasing carrier gas flow rate, the rate of the increase reduces with increasing sampling depth, as shown in Fig. 6. The operating conditions of the plasma for the best SBR are not necessarily the best ones for the least interference by the oxide and doubly charged ions. 3.6. Effect of the rf power Variations of various signals and ratios as a function of rf power are shown in Fig. 7. The flow rate of the carrier gas and sampling depth are fixed at 1.0 l/min and 10 mm, respectively. While the cobalt ion count decreases as the power increases, the lanthanum ion count increases. This is probably due to the decrease of the oxide and doubly charged ions of lanthanum with increasing rf power. The behavior of the ratio of doubly to singly charged ions as a function of rf power is consistent with those reported by OLIVARES and HOUK [9] and GRAY[13] but again contradictory to that reported by VAUGHAN and HORLICK[l]. With increasing rf power, the background increases and the SBR for Co+ decreases, but the ratio for La+ remains unchanged. The behavior of ArO+ somewhat resembles that of Lao’ but is different from that of Ar: also in this experiment, 3.7. Detection limits As shown in the previous sections, the sensitivity of analyte ions depends on various operating conditions of the plasma. Although the optimum conditions were different for each element, compromised conditions were chosen to measure the detection limits: sampling depth 10 mm, carrier gas flow rate 1.0 l/min, and rf power 1.3 kW. Other conditions are the same as those stated in Table 1. The detection limits for single ion monitoring were determined by using 10 s integrations
Continuum
background
;l_J ;k 0.7
-0330.9
1.0
0.7
0.8
0.9
%I
in ICP mass spectrometry
,;i___/( j,,--_rlam
1.0
C#IRIERGRSFLOU
0.7
0.e
0.9
1.0
‘0.7
0.8
0.9--t.,
RATE. L/MN
Fig. 6. Effect of carrier gas flow rate on various ion signals and ion count ratios. Rf power: 1.2 kW, Co and La: 0.1 &ml.
RF WWER.
kW
Fig. 7. Effect of rf power on various ion signals and ion count ratios. Carrier gas flow rate: 1 I/min, sampling depth: 10 mm, Co and La: 0.1 &ml.
for signal and blank to facilitate the comparison with literature values. The standard deviation of the blank was calculated by measuring the blank 10 times for each ion. The detection limits were defined as the concentrations which yield signals of 3 times the standard deviations of the blanks. The detection limits shown in Table 3 are comparable to those reported by GRAY [13] except for Pb. The higher detection limits for Pb and Bi might be due to the lower transmission efficiency at higher m/z of the mass spectrometer used in this work, though further improvements of the sampling interface and lens system could improve the detection limits for heavy elements. 4. CONCLUSIONS Background behavior in the ICP mass spectrometer constructed in our laboratory was investigated. The larger the diameter of the optical baffle plate, the lower the continuum
H. KAWAGUCHI et al.
962
Table 3. Detection
limits Detection
ml2 52 55 59 64 89 115 139 208 209
Element Cr Mn co Zn Y In La Pb Bi
limit (q/ml)
Signal (cps/ppb)
(eps)
This work
Ref. [13]
483 310 310 2-l 143 124 115 33 22
337 902 61 96 36 35 31 34 26
0.04 0.06 0.03 0.3 0.03 0.04 0.05 0.2 0.3
0.06 0.1 0.05 0.2
BG
0.06 0.05 0.05
background, but the best SBR was obtained at 6 mm. The background always increased with increasing negative voltage of the ion lens elements, although analyte signals reached their maximum values at definite voltages. A weak discharge in the lens elements at the elevated lens voltage was suspected as the cause of the background increase. The variation of the background as a function of the ion deflector voltage suggests that charged species partly contributed the continuum background. While analyte signals varied in a complex way with the change of the ICP operating conditions, the background intensity varied in a relatively simple manner: it decreased with increasing carrier gas flow rate and decreasing rf power. Since the ICP operating conditions for the best SBR did not necessarily coincide with those for the least interference by oxide and doubly charged ions, compromised conditions had to be chosen depending on the analytical problem. REFERENCES [l] M. A. Vaughan and G. Horlick, Appl. Spectrosc. 40, 434 (1986). [Z] S. H. Tan and G. Horlick, Appl. Spectrosc. 40, 445 (1986). [3] R. S. Houk, V. A. Fassel, G. D. Flesch, H. .I. Svec, A. L. Gray and C. E. Taylor, [4] [S] [6] [‘I] [S] [9] [lo] [ll] [12] [13] [14]
Anal. Chem. 52, 2283 (1980).
A. R. Date and A. L. Gray, Analyst 106, 1255 (1981). M. C. Johnson, Rev. Sci. Instr. 40, 311 (1969). R. S. Houk, V. A. Fassel and B. R. LaFreniere, Appl. Spectrosc. 40, 94 (1986). H. P. Longerich, B. J. Fryer, D. F. Strong and C. J. Kantipuly, Specrrochim. Acta 42B, 75 (1987). G. P. Russ III and J. M. Bazan, Spectrochim. Acta 42B, 49 (1987). J. A. Olivares and R. S. Houk, Anal. Chem. 57, 2674 (1985). H. Kawaguchi, T. Tanaka, T. Nakamura and A. Mizuike, Bunseki Kagaku 36, 271 (1987). H. Kawaguchi, T. Tanaka, T. Nakamura, M. Morishita and A. Mizuike, Anal. Sci. 3, 305 (1987). G. Horlick, S. H. Tan, M. A. Vaughan and C. A. Rose, Spectrochim. Acta 408, 1555 (1985). A. L. Gray, Spectrochim. Actu 41B, 151 (1986). D. J. Douglas and J. B. French, Spectrochim. Acta 418, 197 (1986).