181
Anal. Chem. 1984, 56, 181-185
the old six curve method this would have required approximately 60 min.
ACKNOWLEDGMENT The authors thank F. E. Greene for her assistance with the line source analyses of the standard reference materials. Registry No. Mn, 7439-96-5;Zn, 7440-66-6;Fe, 7439-89-6;Cu, 7440-50-8; Mg, 7439-95-4. (1) (2)
LITERATURE CITED Harnly, J. M.; O'Haver, T. C. Anal. Chern. 1981, 53, 1291-1298. Harnly, J. M.; O'Haver, T. C.; Golden, B.; Wolf, W. R. Anal. Chern. 1979, 51, 2007-2014.
(3) Harnly, J. M.; Miller-Ihli, N. J.; O'Haver. T. C. J . Autom. Chem. 1982, 4 , 54-60. (4) Harnly, J. M.; Kane, J. S.; Mlller-Ihli, N. J. Appl. Spectrosc. 1982, 3 6 , 637-643. (5) Wolf, W. R. In "Human Nutrition Research, Beltsville Symposia In Agricultural Research"; Allanheld, Osmun and Co.: Totowa, NJ, 1981; Volume 4, pp 175-196. (6) Kuennen, R. W.; Wolnlk, K. A,; Fricke, F. L.; Caruso, J. A. Anal. Cbem. 1982, 5 4 , 2146-2150.
RECEIVED for review June 30, 1983. Accepted October 20, 1983. Presented in part a t the 1982 Federation of Analytical Chemistry and Spectroscopy Societies Meeting (Paper No. 301), Philadelphia, PA.
Molecular Photoluminescence Spectrometry with Hydride Generation for Determination of Trace Amounts of Antimony and Arsenic Hiroaki Tao,* Akira Miyazaki, Kenji Bansho, and Yoshimi Umezaki' National Research Institute for Pollution and Resources, Yatabe, Zbaraki 305,Japan
Molecular photoluminescence detection was developed for the determination of antimony and arsenic. Antimony and arsenic were generated as hydrides and Irradiated with ultraviolet light. The broad continuous emission bands were observed in the ranges about 240-750 nm and 220-720 nm, and the detection ilmits were 0.6 ng and 9.0 ng for Sb and As, respectively. Some characterlstlcs of photolumlnescence phenomenon were made clear from spectroscoplc observations. After the Interference study, the method was successfully applied to the determlnatlon of antimony In river water and seawater.
The hydride generation technique has become widely used in spectrometric analysis. It allows the separation and preconcentration of an analyte from the sample matrix, thereby providing an improved detection limit and relative freedom from detector interference. For the final measurement, atomic spectrometries, e.g., atomic absorption ( I ) , emission (2-4), and fluorescence (5) spectrometry, have been mainly used, although molecular emission spectrometry (6, 7), mass spectrometry @),and thermal conductivity detectors (9) have also been employed. In addition, chemiluminescence spectrometry with ozone oxidation of hydrides has been proposed recently (10, 12). Molecular photoluminescence spectrometry has good sensitivity and selectivity for some organic or inorganic materials and has been applied to the biochemical and environmental spheres as fluorescence spectrometry (12). However, to our knowledge, it has not been applied so far to the determination of hydrides. In the present paper, molecular photoluminescence detection of some hydrides is investigated. Intense luminescence spectra are observed in the region 240-750 nm and 220-720 nm when stibine and arsine are exposed to ultraviolet light of wavelength 200-240 nm, respectively. The aim of this work is to demonstrate that molecular photoluPresent address: Government Industrial Research Institute,
Chugoku, Hiro-machi, Kure-shi, Hiroshima 737-01, Japan.
minescence spectrometry is a very sensitive and selective method for the determination of some hydrides. In the case of stibine, the molecular species which luminesces is also discussed and an interference study is carried out. The method has been successfully applied to the determination of antimony in river water and seawater. Although the present report is concerned only with antimony determination, the interference study shows that other hydrides, especially arsenic, can be determined in a similar way. Since this method does not need a flame, it is suitable for an automated system and will provide simple sequential multielement analysis capability of nanogram amounts of some hydrides by coupling to gas chromatography.
EXPERIMENTAL SECTION Apparatus. The hydride generation system with a liquid nitrogen trap is shown in Figure 1. The hydride generated in the reaction vessel was swept into the liquid Nz U-trap (25 cm X 3.3 mm id., fully packed with 24-35 mesh glass beads). A U-tube with Drierite (25 cm X 1 cm id., 8 mesh) was used to remove the water vapor from the reaction solution. The hydride frozen in the liquid N2 trap was then released, by inserting the trap in hot water, and was introduced into the cell with a helium flow of 1 mL/min by a peristaltic pump (Technicon, Model 11). The hydride was then irradiated with light of 210 nm, and the resulting luminescence intensity at 325 nm was monitored with a fluorescence spectrometer. The luminescence detection was made by a Hitachi Model, 650-60 fluorescence spectrometer with a 150-W xenon lamp (Ushio, type UXL-157) and R-928F photomultiplier (Hamamatsu Photonics). The maximum slit width (spectral band-pass 20 nm) was used unless otherwise stated. The cell was a conventional fluorescence cell and the volume (4.4 mL) was reduced to 1.4 mL with a brass spacer to avoid the large dead volume. Chemicals. A stock solution of Sb (1000 pg/mL) was prepared by dissolving SbC1, in 3 M HCl solution. A stock solution of As (1000 pg/mL) was prepared by dissolving Asz03in a minimum amount of 0.2 M NaOH and diluting the solution with a small amount of concentrated HzS04to the final H,S04 concentration of 0.01 M. Stock solutions of Sn and Te (1000 pg/mL) were prepared by dissolving SnCl2.2H20and NazTeOsin diluted HCl solutions. Stock solutions of Ge and Se (1000 pg/mL) were prepared by dissolving K4Ge04and SeO, in diluted KOH solution
0003-2700/84/0356-0181$01.50/00 1984 American Chemical Society
182
ANALYTICAL CHEMISTRY, VOL. 56, NO. 2, FEBRUARY 1984
W W Reaction
(Drierite)
vessel
Flgure 1. Hydride generation system for photoluminescence. and distilled-deionizedwater, respectively. Stock solutions of Bi and Pb (1000 pg/mL) were prepared by dissolving the metals in a minimum amount of HNOBto lower the concentration of interfering ”OB. A 4% NaBH, solution and 1 M HCl were used for hydride generation. A 4% NaBH, solution was fiitered through a 0.45-pm Millipore filter. Stock solutions of salts used in interference studies were prepared from chloride salts for cation interference and sodium salts for anion interference. Procedure. A sample of about 20 mL (acidity was adjusted to 1M HCl) was placed in the reaction vessel. After the solution was bubbled for 30 s with He gas for degassing, the three-way stopcocks were opened to the positions shown in Figure 1 and the hydride trap was immersed in liquid nitrogen bath. After the trap was completely cooled, a 4% NaBH4solution was injected into the reaction vessel for 2.5 min at a flow rate of 2 mL/min by a syringe. The flow rate of He carrier gas through the sample bubbler was 200 mL/min. The reaction was continued for 30 s more in order to completely trap the hydride. Then, the pump was operated and the trap was transferred into a water bath at about 50 OC. By this procedure, the hydride was vaporized from the trap and was introduced into the luminescence cell with a He flow of 1mL/min. The luminescence intensity was monitored by the fluorescence Spectrometer. Measurement of Luminescence and Absorption Spectra. In order to select the best wavelength for analysis, the luminescence spectra were measured by sealing the hydrides in the cell. According to the procedure stated above, about 0.5 mg of each hydride-forming element (As, Sb, Bi, Ge, Se, Te, Sn, and Pb) was converted into the hydride. After the hydride was introduced into the cell, the pump was switched off and the hydride was sealed in the cell. The luminescence spectrum was measured by both scanning three times from 200 to 850 nm and averaging the intensities when the hydride was exposed to light of 210 nm. The measurement was carried out with a spectral band-pass of 5 nm. The absorption spectrum was measured before and after the measurement of the luminescence spectrum to evaluate the extent of decomposition of the hydride by the irradiation at 210 nm. The measurement was carried out by using a Shimadzu UV-240 spectrophotometer with a spectral band-pass of 1 nm. Measurement of Mass Spectra. Mass spectra were measured to identify the luminescing species by a JEOL Ltd., type JMS-Q10, gas chromatography/mass spectrometer. The GC separation column was removed, and the hydride in the luminescence cell was directly introduced into the mass spectrometer through a Teflon tube (60 cm X 0.33 mm i.d.). One part of the Teflon tube was replaced by a short poly(viny1chloride) tube (2 cm X 1mm i.d.) with a clip. The clip is necessary as the gas flow controller since the production of luminescing species needed irradiation with light for a relatively longer period as mentioned later in “Optimization for the Measurement of Antimony”. Otherwise, the hydride would be introduced into mass spectrometer before producing a sufficient quantity of luminescing species. About 0.1 mg of Sb was converted into the hydride and exposed to ultraviolet light of 210 nm. When the luminescencesignal became
Wavelength (nm) E
(b)
1 i
Wavelength (nm)
Figure 2. Photoluminescence spectra of (a) stibine and (b) arsine. Excitation wavelength was 210 nm. maximum, the hydride was introduced quickly into the mass spectrometer by removing the clip. The mass spectrum observed in this way was compared with that observed in the case of nonirradiation to detect the luminescing species or its resulting products. The operating conditions of the mass spectrometerwere as follows: emission current, 0.7 mA; ionizing voltage, 70 eV; preamplification lom9A/100 mV.
RESULT AND DISCUSSION Photoluminescence Spectra. Figure 2 shows the photoluminescence spectra obtained for antimony and arsenic when their hydrides were irradiated with light of 210 nm. Broad bands with maxima at 320-400 nm and 300-520 nm for antimony and arsenic, respectively, were obtained. With the instrument used in this experiment, second-order light gave greater signals than first-order light, and so the intense peak around 420 nm for arsenic, was a second order of the peak around 210 nm. Compared with antimony, arsenic exhibited a luminescence band beginning a t a slightly shorter wavelength and a distinguishable peak around 240 nm. Photoluminescence spectra were somewhat similar to the
ANALYTICAL CHEMISTRY, VOL. 56, NO. 2, FEBRUARY 1984
0'
0
183
1 2 3 He Gas Flow Rate (rnLlmln)
Effect of He gas flow rate for vaporization on the Iuminescence intensity. Water temperature for vaporization was 50 'C. Figure 3.
chemiluminescence spectra in a flow-type furnace hydrogen diffusion flame reported by Fujiwara et al. (13), although wavelengths of maxima intensities of photoluminescence spectra were slightly shorter than those of chemiluminescence spectra for both antimony and arsenic and the photoluminescence band around 240 nm for arsenic was much broader. For analytical antimony measurement, the peak at 325 nm was chosen. Optimization for the Measurement of Antimony. T o optimize the operating conditions, effects of several parameters, i.e., HCl concentration, amount of NaBH4, He gas flow rate for bubbling, trapping time, and temperature of water bath, and He gas flow rate for vaporization on the luminescence intensity were examined. Among these parameters, He gas flow rate for vaporization was very characteristic to the present method. The relationship between signal intensity and He gas flow rate for vaporization is shown in Figure 3. The optimum value was 1.0 mL/min, although it was usually several hundred milliliters per minute for conventional atomic absorption, atomic emission, or chemiluminescencetechniques. At such a high flow rate, the present method gave no signal. This means that it is necessary to expose the hydride to ultraviolet light for a relatively longer period in order to obtain the intense luminescence signal. Therefore, it is suggested that the luminescing species is not SbH3 itself but the decomposition products of SbHs or the reaction products of SbH3 with photon-induced species such as ozone. It took about 1.5 min until a signal appeared after immersing the hydride trap in hot water and further about 1.5 min to reach maximum signal intensity. Other optimum conditions were determined as follows: HC1 concentration, 1 M; amount of NaBH, (injection flow rate of NaBH4 solution), 5 mL of 4% NaBH4 (2 mL/min); He gas flow rate for bubbling, 200 mL/min; trapping time, 3 min; temperature of the water bath for vaporization, 50 "C. For a drying agent, Drierite (CaS04)was superior to CaClz giving rise to higher signals. Since a small amount of COz contained as an impurity in NaBH4 agent caused no interference, there was no need for placing a soda lime or NaOH trap before the liquid N2 trap. Such a COz trap caused the adsorption or decomposition of stibine. It is noted that a Teflon tube was used for the passage of stibine from the liquid N2 trap to the cell. If a Tygon or a poly(viny1 chloride) tube is used instead of a Teflon tube, stibine is adsorbed to the tube, and the signal is split into several peaks. Even for a Teflon tube, a shoulder of the sigrial peak was observed, but the extent of adsorption was fairly small. If a shorter Teflon tube, whose length was forced to be relatively long in this experiment due to the structural hindrance between hydride trap and fluorescence spectrometer, could be used, adsorption would be even less. Detection Limit and Analytical Calibration Curve. The analytical calibration curve is shown in Figure 4. The detection limit was 0.6 ng, which corresponded to 30 pg/mL when a 20-mL sample solution was ulsed. The detection limit was based upon three times the standard deviation of the blank signals (for 10 repetitive measurements). The blank
1
102
10
lo4
103
Amount of Antimony (ng)
Analytical calibration curve for antimony: HCI concentration, NaBH,, 5 mL; He gas flow rate for bubbling, 200 mL/min; trapping time, 3 min; He gas flow rate for vaporization, 1.O mLlmin. Figure 4. 1 M; 4 %
Table I. Photoluminescence Intensities of Other Hydride-Forming Ions at the Optimum Conditions for Antimony Determination re1 intensa ion re1 intensa ion As3
+
Bi3+ Se4 Te4 +
+
6.7 0.048
Ge4 Sn2+
n.d.b n.d.
Pb"
+
a The intensity for S b 3 +is defined to 100. detected.
n.d. 0.042 0.040
n.d. = not
signal was due to the antimony contamination in the hydrochloric acid used in this experiment. The detection limit of this method was almost equal to the detection limits of traditional methods reviewed in ref 14. Interference Study. Firstly, spectral interference from other hydride-forming elements (As, Bi, Ge, Se, Te, Sn, and Pb) were examined at the optimum conditions for antimony determination. Table I shows to what extent each of these elements luminesces. Except for As, these elements gave no significant signals. In general, the concentration of As is higher than that of Sb in actual samples like natural waters, so As could cause severe positive interference. Conversely, if the two hydrides would be separated by gas chromatographic methods, as reported in ref 10, arsenic also could be detected with a relatively high sensitivity. The detection limit for As was calculated to 9.0 ng (0.45 ng/mL) at least by simply comparing the luminescence intensity of As observed at optilhum conditions for Sb determination with that of Sb. Tin and lead gave p l y weak luminescence signals. However, since optimum conditions of hydride generation for Sn and Pb were widely different from those of Sb (1, 15), the quantit,ative detection of these elements might be possible, if the conditions of hydride generation and wavelengths of excitation and luminescence could be optimized. Secondly, the effects of various inorganic cations, anions, and volatile organic compounds were investigated with 10 ng/mL Sb solutions. The results are shown in Table 11. Negative interferences by transition-metal cations such as Ni2+ afid Cu2+and nitrite were observed. However, these interferences have also been reported for the hydride generation atomic absorption methad (16) and are due to the inhibition of hydride generation. There was no interference from four volatile organic compounds. Nitrate has been reported to have the inhibition effect (2, 14),but in the present method NO3- enhanced the lumines-
184
ANALYTICAL CHEMISTRY, VOL. 56, NO. 2, FEBRUARY 1984
Table XI. Interferences of Inorganic Ions and Volatile Organic Compounds with Determination of Antimonya concomitant none Na
K' MgZt
Ca2+ NH,+ Zn2t
10000
loa
10000 1000
101
re1
99
100 101 100
1000 100 10 10 10 10 10 10 10 10 10 1000 100 10 100 10 1 0.8 0.8
MnZ+
Mo6+ V5' COZ+
cuz+ Cr6+ Fe3+ NiZ+ 2-
c0,zNO,NO,' iCH,)ZCO CH,OH
98
CHCI,
97 69 68 65 57 100
49 98 127
2 73 100
99 100 101
1.5
Interferences were examined with a 1 0 ng/mL Sb
I
Wavelength ( n m )
104
s o h tion.
140
Wavelength ( n m )
102
0.9
C6H6
a
intens 100
+
so,
concn, Pg/mL
0
Ratio
Of
Absorbance (ASb/AAs)
Figure 6. Absorption spectra of (a) stibine and (b) arsine before (-) and after (---) irradlation at 210 nm for about 10 min. (c)Relationship between the ratio of absorbance (at 10 nm intervals of excitation wavelength of 200-230 nm) and the ratio of luminescence intensity at 325 nm of SbH, and ASH,: (A)200 nm, (A)210 nm, (0)220 nm, (0) 230 nm.
experiment. The concentration of NO; in most natural waters is not so high that there is no problem in analyzing river water and seawater for Sb. However, it would be necessary to check the recovery of Sb for analyzing the acid (HC104-HN03, "0,-HC1-HF, etc.) sample digests. Identification of Luminescing Species. Because photoluminescence occurs only when the He gas flow rate for vaporization is very low, this is not a typical fluorescence spectrometric detection system. Photoluminescence spectra shown in Figure 2 were similar to the chemiluminescence spectra of SbO or As0 in ref 13. The (AsO), excimer and Te, excimer are also suggested as possible orgins of the luminescence in chemiluminescence of arsine (11)and in MECA of tellurium (20). By analogy to these phenomena, species such as SbO, Sb,, and (SbO), might be produced by the reaction of SbH3 and O3 (produced by UV irradiation of O2 which is an impurity in He gas and by leakage of air) or by the reaction of SbH3* (excited state) and 02.Subsequently, these species might be the emitting entities in the present method. However, these species were not detected in the mass spectrometric study. In addition, these reactions do not seem to occur very readily because of the low concentration of 02. There was no difference between the mass spectrum observed in the case of irradiation and that of nonirradiation, and other species were not observed except for SbH3and its fragments in both cases. Figure 6a,b shows absorption spectra of SbH3 and AsH3 before and after the irradiation at 210 nm. From the decrease of absorbance, the hydrides seem to decompose, and the rate of decomposition of stibine is considered to be faster than that of arsine. Figure 6c gives the relationship between the ratio of absorbance of the two hydrides at 10-nm intervals of excitation wavelength of 200-230 nm and the ratio of luminescence intensities a t 325 nm of them. If the absorption process of the hydrides (SbH3+ hu SbH3* and AsH3 + hu AsH3*) is responsible for the luminescence and other
L-_-
loo0 100
500
too0
Concentration of Nitrate
1ooooo
(pg/mL)
Flgure 5. Enhancing effect of NO3- on the luminescence intensity. All samples contained 10 ng/mL of Sb.
cence signal. The effect of NO;, which is added in the form of NaN03 at the acid concentration of 1 M HC1, on signal intensity is shown in Figure 5. The intensity in the presence of 1-10, mg/mL NO3- is constantly 1.3-1.4 times that in the absence of NO3-. However, if nitric acid is used instead of NaN03 and HC1, the plot of signal vs. concentration of nitric acid went through a maximum a t about 2 M H N 0 3 (ca. 12 mg/mL NO3- if perfectly ionized, but in fact not ionized perfectly (17)). Signal intensity at 4 M HN03 was about half signal intensity a t 2 M "OB. The decrease at 4 M HNOB was due to the inhibition effect of an increased acid concentration, and it was also observed at 4 M HCl. In the presence of NaN03 or HN03, the standard deviations of signal intensities were much greater than those observed in the absence of these compounds. Nitrous oxide is reported to be produced by the reduction of NO3- with NaBH, (9). Hence, as the reason of enhancing effect of NO;, photochemiluminescence between SbH3and NzO (SbH3+ N 2 0 H,SbO* + Nz + H, and H,SbO* H,SbO hv) is guessed by the analogy to Ba N 2 0 chemiluminescence (18) and Mg + N 2 0 chemiluminescence (19). However, the species like H,SbO was not detected from the mass spectrometric study. Therefore, the reason for the enhancing effect was not elucidated in this
+
-
+
-
-
-+
ANALYTICAL CHEMISTRY, VOL. 56, NO. 2, FEBRUARY 1984
-
-
processes such as ozone formation and subsequent chemiluminescence (30, hv 2O,, O3+ SbH3 H,SbO,* and O3 ASH, H&O,*) are not important, luminescence intensity L(A) (as a function of excitation wavelength (A)) is given as L(A) = AI(A)q(A)q’(325 nm) (1)
+
-
+
where M(A)is the intensity of light absorbed by the hydride, q(A) is the luminescence yield, and q’(325 nm) is the quantum yield of photomultiplier at 325 nm. M(A)is given as A I ( A ) = &)(A) - I ( A ) = Io(A) (1 - e-to(W)
= I o ( ~ ) (-l e-2.303A(X))
(2)
where l o ( A ) is the intensity of incident light, I ( A ) is the intensity of transmitted light, to(X) is the absorption coefficient of the hydride, c is the concentration of the hydride, 1 is the length of the cell, and A(A) is the absorbance. Because the concentration of the hydride was kept sufficiently low to make the quantity A(A)