Pulsed radiofrequency-excited electrodeless discharge lamps for

Pulsed radio-frequency electrodeless discharge lamps for atomic absorption and atomic ... Burst Radio-Frequency Excited Pulsed Hollow-Cathode Lamp...
0 downloads 0 Views 689KB Size
ANALYTICAL CHEMISTRY, VOL. 50, NO. 3, MARCH 1978

Table 111. Comparison of Methods Using Process Samples, ppm U,O, Sample No.

This method

PAR

DBM-TOP0

84.5 11.4 85.2 39.1 27.4 4.2 25.2

85.8 11.4 85.2 39.1 27.4 4.2 25.2

87.0 10.8 83.0 36.0 20.0

1

2 3

4 5 6 I

...

25.4

Table IV. Statistics of Results, No Carbonate Present

Trial No. 1

2 3

4 5 Mean Std dev RSD, % Lower limit

Uranium found, ppm 4.40 4.40 4.17 4.85 4.85 4.58 0.286

Trial No. 1

2 3 4 5

Mean Std dev 6.2 RSD, % of detection, ppm

Uranium found, PPm 82.7 82.7 83.0 83.0 82.4

sensitivity improved by a similar amount. Unfortunately, the linear region for this method extends only to 60 ppm. T h e equation of the least-squares fit of the calibration curve is Abs = 2.03 X (ppm U) + 1.84 x T h e correlation coefficient is 0.993 and Student's t (5 degrees of freedom) is 18.7. Precision was evaluated to be 6.2% at 4.5 ppm and 0.3% a t 83 ppm uranium. Data pertaining to the precision of the method in the absence of carbonate are included in Table IV. The proposed flow method gives rapid and precise analyses for uranium in the presence of large amounts of carbonate. The method is simple and does not require that the carbonate be destroyed by lowering the pH to the acid range, nor does the method require extraction of the uranium sample into an organic solvent.

ACKNOWLEDGMENT T h e authors express their gratitude to P. W. Carr for the use of the flow cell, to Perkin-Elmer Corporation for the use of a spectrometer, and to W. Lemons for his assistance. LITERATURE CITED C. J. Rodden, Anal. Cbem.. 25, 1958 (1953). C. J. Rodden, Ed., "Analytical Chemistry of the Manhattan Project", McGraw-Hill, New York, N.Y., 1950 pp 3-159. R. Pribil, "Analytical Applications of EDTA and Related Compounds", Pergamon, New York, N.Y., 1972, pp 263-279. R. J. N. Brits and M. C. B. Smit, Anal. Chem., 49, 67 (1977). L. R . Hathaway and C. W. Jones, Anal. Cbem., 47, 2035 (1975). M. A. Flashka and A. J. Barnard, "Chelates in Analytical Chemistry", Vol. 1, Marcel Dekker, New York, N.Y., 1967, pp 251-255. H. W. Bertram, M. W. Lerner, G. J. Petretic, E. S.Roszcowski, and C. J. Rodden, Anal. Chem., 3 0 , 354 (1958). T. V. Iorn. paper presented at 7th Technicon International Congress, New York, N.Y., 13 December 1976. L. Sommer, Z. Anal. Chem., 187, 263 (1962). L. Sommer, CoNect. Czech. Chem. Commun., 30, 3426 (1965). A. R. Main, Anal. Cbem., 26, 1507 (1954). L. A. Maclaine, "The Carbonate Chemistry of Uranium: Theory and Applications", in Rococ.Int. Conf. Peaceful Uses At. Energy, 8, 26-27 (1956). Fathi kbashi, "Principles of Hydrometallurgy", Vol. 2, Gordon and Breach, New York, N.Y., 1970, p 217. T. M. Florence and Y. Farrar, Anal. Cbem., 35, 1913 (1963).

82.8 0.273 0.3 0.59

and agreement is quite good. A statistical test (pair data experiment) was applied to the data set to see if the methods give statistically equivalent results. At a 95% confidence level, the d a t a can be said to belong to the same population. T h e applicability of the flow method to the analysis of uranium in natural waters was examined by performing the analysis on synthetic uranium samples in the absence of carbonate. Since carbonate is not present a t the high levels associated with solution mining, there is no need to add NaOH or (NH4)&03. As a result, improved sensitivity and detection limit are anticipated. The pump manifold was altered by eliminating pump channel No. 3, that is, the mixture of NaOH, (NH&C03, and N2H4was not employed. The detection limit improved to 0.59 ppm uranium as indicated in Table I\'. The

407

RECEIVED for review May 13,1977. Accepted December 12, 1977. This work was supported in part by Research Grant CHE-7618385 from the National Science Foundation.

PuIsed Radiof requency- Excit ed ElectrodeIess Discharge Lamps for Analytical Atomic Spectrometry John W. Novak, Jr. and Richard F. Browner" School of Chemistry, Georgia Institute of Technology, Altanta, Georgia 30332

Novel behavior of rf excited electrodeless discharge lamps is observed when these sources are operated in a pulsed mode. The radiant output of the EDLs is found to increase rapidly when short ( microwave; Cd: rf E microwave; Zn: rf < microwave.

T h e search for new and improved radiation sources for analytical atomic spectrometry, particularly atomic 0003-2700/78/0350-0407$01 .OO/O

fluorescence spectrometry, has recently received a great deal of attention (1-6). In particular, much recent work has been with pulsed sources, and the potential advantage of pulsed source operation in AFS has been thoroughly described by Omenetto et al. (3, 6, 7). However, experimental detection limits obtained for many elements using pulsed sources, including pulsed HCls ( 4 , 5 , 8 ) ,pulsed tunable dye lasers (2, 6) and pulsed continuum sources ( I ) , have been disappointing ( 7 ) . As a consequence of the problems observed with the previously mentioned sources, a study of the spectral properties of electrodeless discharge lamps (EDLs) when operated under high power pulsed conditions was initiated. While CW or modulated operation of metal EDLs (both microwave ( S 1 4 ) and rf excited (15-17) has been thoroughly examined, we know of no previous studies involving the high C 1978 American Chemical Society

408

ANALYTICAL CHEMISTRY, VOL. 50, NO. 3, MARCH 1978

Figure 1. Circuit diagram of rf generator power pulsed operation of these lamps, other than for wavelength assignments (18-21), although recently the high power C W operation of sealed rf excited lamps has been described (22).

EXPERIMENTAL Optical System and Electronics. All source radiance measurements were made on a 0.5-m Ebert monochromator (Jarrell-Ash model 82-020) using an 1180 grove/mm grating, blazed at 190.0 nm with a Varian model EM1 9783 B photomultiplier. Supply voltage to the PMT (-700 V) was from a Keithley Instruments model 244 High Voltage supply. Source emission signals were displayed on a Tektronix model 561A oscilloscope or, in a few instances, on a Keithley model 4145 Picoammeter connected to a Sargent-Welch model S26 recorder. Focusing lenses were of fused quartz with diameter = 5 cm, focal length = 12.5 cm. Electrodeless Discharge Lamps. Lamps were made from Vitreosil transparent quartz tubing 8 cm in length and 8-30 mm internal diameter. Lamps with an internal diameter between 13 and 20 mm proved most satisfactory. Tubes of 16-mm i.d. were found to be optimum, and were used primarily in this study. All lamps were single element (or element plus I*), prepared as described by Dagnall and West (12). The pure metal was used for most Zn, Cd, and Hg lamps, but the iodides of Zn and Cd were also investigated. Argon was used as the fill gas with pressures varying from 0.3 to 3.0 Torr. No other fill gas was studied, as Ar has been previously demonstrated to provided an optimum combination of long lamp lifetime and high output intensity (13, 23).

RF Generator and Pulsing Unit. A circuit diagram of the generator is shown in Figure 1. The generator was operated at a frequency of 13.5 MHz. The plate voltage was variable from 0 to 3000 V. During pulsing, the cathode current rarely exceeded 700 mA. In the pulsed mode the range of the generator was between 100 and 1600 W peak power. The excitation coil was made from 12 turns of 4-mm 0.d. copper tube and the lamp was positioned axially along the coil. The coil (6.4-cm i.d., 14 cm long) was widely spaced and radiation was measured at right angles to the tube axis along a cross-sectional diameter. An open plate capacitor was used in tuning the cavity, the optimum tuning criteria being a combination of maximum lamp emission intensity and minimum screen current. The pulsing unit (circuit diagram available on request) gave pulse repetition rates variable from about 1 s-' to 1.5 X s-'. Pulse width was variable from 10 to 0.2 ms. The total power coupled to the discharge was limited by the low Q of the system. It is unlikely that more than 20% of the rf energy coupled to the EDL although accurate coupling efficiency measurements were not attempted. The coupling efficiency could be improved by use of more carefully designed energy couplers (24, 25), but in order t o be effective these require a highly frequency stable rf source. The frequency stability of the rf generator was not adequate to take advantage of the higher Q possible with these couplers. The microwave generator used for comparison purposes was a 2.45-GHz Microtron 200 MK I11 unit (Electromedical Supplies.

**-Ti

Figure 2. Comparison of radiant output of pulsed and CW operated Ar EDLs as a function of incident rf power. (0)pulsed operation, (A) CW operation. In the pulsed mode, the signals recorded are peak values

Wantage, U.K.) with variable frequency source modulation. A 3/4 X Broida cavity was used for lamp coupling. Microwave excited lamps were optimized for temperature by operation in a thermostated mode. EDL Temperature Control. The lamp temperature was optimized for each measurement by controlling the temperature of the air from a fan/heater-coil arrangement. Air temperature was variable from 20 to 400 "C. Individual temperature optimization of the lamps was necessary at each power setting in the CW mode in order to compensate for the considerable inductive rf heating which occurred, particularly at high powers. However, in the pulsed mode one temperature setting of the heater was usually sufficient. At most, there was a need for only two temperature settings, one for high power and one for low power operation.

RESULTS AND DISCUSSION T h e first study of pulsed EDLs consisted of a comparison of UV resonance line intensities, with the lamps operated first in a pulsed mode, then CW. T h e lines monitored were those commonly used for AFS: namely Zn, 213.8 nm; Cd, 228.8 nm; and Hg, 253.6 nm. Ar was monitored both a t its 420.1 n m and 33'7.1 n m resonance lines and also a t the 247.9 n m ion line. There were no noticeable differences in behavior between the resonance and ion lines and only curves for the 247.9 nm line are presented here. All lines measured were integrated over the entire line profile by the use of instrumental spectral band widths of 20.1 nm. Line profile effects would not, therefore, be observed from these data. Ar Electrodeless Discharge Lamps. (Influence of Power on Emission Intensity). T h e variation in Ar emission intensity as a function of incident rf power was first studied for a low pressure (- 1Torr) Ar lamp with both pulsed and C R operation (Figure 2). T h e S-type curves in Figure 2 are explained as follows: (i) when the lamp first ignites, the discharge is concentrated along the inner axis of the tube. (ii) As the power increases, t h e lamp suddenly develops a "fireball" type of emission, with a very intense area occupying about a third of the tube length. (iii) As the power is raised further, the intense discharge spreads more evenly over the entire tube. (iv) After the intense discharge fills the tube, an increase in power creates a much more moderate increase in lamp intensity. Heating effects of the inner plasma probably become significant when the rf power is increased further, resulting in decreasing intensity, especially in the CW mode, even when the lamp is cooled. From the two curves, one can see t h a t a t lower wattages the CW mode results in a higher source intensity. Also the intense glow occurs a t a lower power. However, as the power

ANALYTICAL CHEMISTRY, VOL. 50, NO. 3, hIARCH 1978

w

I 600

32I)P *

I

409

J

,800

b

Figure 3. Comparison of radiant output of pulsed and CW operated Ar EDLs as a function of incident rf power; no high intensity mode CW operation. I n the pulsed mode, occurring. (0)pulsed operation, (A) the signals recorded are peak values

is increased, the output of the pulsed lamps continues to rise while the CW lamp output levels off. T h e slope is not very steep, even for the pulsed lamps. A disadvantage with the rf generator used in these studies was the lack of peak power output advantage in t h e pulsed mode compared to the CW mode of operation. As a consequence of this, and the lack of our ability to couple rf energy more efficiently with the EDLs (see Experimental section), it proved impossible to determine the form of these curves a t higher powers. Consequently, it is not known whether the curve for pulsed source operation will ultimately reach a peak or a plateau region. Lamps made with argon a t fill pressures greater than approximately 3 Torr failed to make the transition to the “fireball” state and, consquently, radiant output at the Ar lines was much lower in both CW and pulsed modes of operation than with the 1-Torr lamps. However, the difference in intensity between the two modes was more marked than for 1-Torr lamps (Figure 3). In this case the pulsed lamp intensity at maximum rf power was about a third brighter than for the CW lamp, and was still rising sharply, whereas the CW operated lamp intensity became asymptotic to the power axis. Tube Diameter and Fill Gas Pressure. The effect of tube diameter and fill gas pressure was investigated for Ar and Hg EDLs. T h e optimum fill pressures for all the lamp diameters investigated (Le., 8-30 m m i.d.) was found to be in the range 0.3-3 Torr. There was little difference between t h e maximum intensities a t the different pressures, but the transition from diffuse glow to “fireball” emission occurred at lower powers with lower fill gas pressure. It was also found that t h e “fireball” emission occurred a t lower powers in the larger diameter tubes. Hg Electrodeless Discharge Lamps. A comparison of pulsed vs. CW operation for a typical Hg lamp (- 1 Torr Ar fill gas) showed a similar pattern to the Ar lamp (Figure 4). Initially, the glow was concentrated along the tube axis. Under these conditions there was probably strong self-reversal of the source output, resulting from reabsorption of radiation in the nonemitting outer layers of the lamp. Again, there was a clear transition between a region of low slope at powers below about 300 W. followed by a region of high slope. The CW operated lamp again went into its high intensity glow a t lower power than the pulsed lamp. However, a t higher powers the pulsed Hg EDL continued to increase in intensity while the CW operated lamp reached a maximum intensity, followed by a rapid drop in output with further increase of rf power. A likely explanation for this behavior is that adequate cooling of the CW operated lamp, which is necessary in order to maintain

i

l

l