case is distinctly superior to HC1 as eluting agent because Ca forms the complex CaN03+ in "03 which decreases its distribution coefficient, while the distribution coefficient of A1 is distinctly higher in "03 than in HC1. Figure 4 shows the improvement of the separation when using 1.25M "03 instead of 1.50M. A further decrease in acid concentration would lead to inconveniently large elution volumes for Ca. From Figure 4, it was decided to elute Ca with 450 ml of 1.25M "03. A1 can then be eluted with 250 ml of 3.OM HCl. Figure 4 shows that the amount of A1 which can be present is also limited. The actual maximum amount allowed will depend on the accuracy required and the safety margin employed. I t has been shown above that the maximum amounts for some of the elements are limited. The elements and their limit amounts are summed up in Table VIII. Na and K can be present in considerably larger amounts without endangering their separation from other elements, but they may then not be present quantitatively in their fractions anywill have to be more, and the whole 950 ml of 0.5M "03 taken for their determination in the presence of each other. Furthermore, the total amount of elements present should not be larger than about 25 exchange equivalents, otherwise the Na and K peaks could be shifted too much and these elements partially be eluted outside their given fractions. Zirconium could be determined satisfactorily only in the basalt W-1. This was also the only rock sample which dissolved completely, in the HF-HCl-HC104 mixture. For the other rock samples, the recovery of Zr was erratic and low. An investigation using the syenite S-1 revealed that only about 30% of the Zr was dissolved in the HF-HCl-HC104 treatment. About 70% remained in the insoluble residue which was less than 1 mg in weight. When this residue was fused with the phosphoric acid mixture and, after dissolution, passed through a separate column, a large part of the Zr appeared in the V(V) fraction and the remaining part in the Ti(1V) fraction. The Na K fraction in between was free from Zr. Part of the Zr therefore seems to form an insoluble colloidal phosphate which is uncharged and passes
+
through the column. Alternatively, a polyphosphate complex may be formed. The phosphoric acid fusion will have to be omitted, when zirconium is to be included in the analysis, though it has great advantages because of its low addition to blank values, especially in case of the alkali metals. Various makes of lithium borate are investigated a t present for their background of alkali metals and other elements with the aim to select an alternative fusion reagent which can replace the phosphoric acid without endangering the determination of some of the other elements, when these are present in very low concentrations. With the exception of zirconium, the precision and accuracy of the described method is very high, as is indicated by the results presented in Tables I and 11. The precision for the results presented for actual rock samples in Table VI1 was comparable to that obtained for the synthetic standards. An evaluation of the accuracy of these results is difficult because the differences between the results obtained and the preferred standard values are not larger than the differences between preferred standard values from different workers as is shown for AGV-1 in Table VI1 ( 1 4 , 15). If it is assumed that the accuracy for actual rock samples matches the precision, the method would be suitable to investigate small variations in compositions due to sample inhomogeneities. The described method requires less time and is more accurate than the classical wet chemical analysis of rocks. I t is also more precise and accurate than modern instrumental methods of rock analysis such as X-ray fluorescence and atomic absorption spectrometry when these are used without effective separation methods, though it is more timeconsuming. I t seems to be very well suited for accurate standardization work and as a control method for routine instrumental determinations.
RECEIVED for review July 23, 1973. Accepted April 16, 1974. Part of a DSc. thesis by A.H.V. a t the Department of Inorganic and Analytical Chemistry, University of Pretoria.
Application of a Demountable Hollow Cathode Lamp as a Source for the Direct Determination of Sulfur, Iodine, Arsenic, Selenium, and Mercury by Atomic Absorption Flame Spectrometry G. F. Kirkbright and P. J. Wilson Chemistry Department, Imperial College, London, S. W. 7. U.K
A water-cooled demountable hollow cathode lamp has been investigated as a source for the elements, S, I, As, Se, and Hg. The high emission intensity, good signal-to-background and signal-to-noise levels observed, permit use of this type of source for flame atomic absorption spectrometry of these elements in the nitrogen shielded nitrous oxide-acetylene or air-acetylene flames at their resonance lines below 200 1414
nm. Self absorption line broadening from these sources is minimized for the volatlle elements at the low cathode temperatures and operating currents employed and using a continuous purge of argon at low pressure; favorable AAS sensitivitles and detection limits are obtained, compared with corresponding microwave excited electrodeless discharge lamp sources.
ANALYTICAL CHEMISTRY, VOL. 46, NO. 11, SEPTEMBER 1974
The sealed-off hollow-cathode lamp has been the principal sharp line source employed for analytical atomic absorption spectrometry (AAS) since its introduction for this purpose by Walsh ( I ) . In the past ten years, the performance characteristics (line intensity, line-to-background intensity ratio, stability, etc.) attainable with sealed-off hollow cathode lamp sources have been improved considerably. Although a number of alternative types of source, such as microwave excited electrodeless discharge lamps and continuum sources have been proposed for AAS, the simple hollow cathode lamp source has continued to find the most widespread application in analytical practice. The conventional hollow cathode lamp sources of elements which show appreciable volatility a t low temperatures, however, frequently exhibit low intensity a t the most useful resonance lines, poor 1ine:background intensity ratios and stability and limited operating lifetime. These effects result from both the high rate of sputtering of material from the cathode and the volatilization of cathode material to form a relatively dense atomic vapor in front of the cathode; loss of signal then occurs by self absorption, and cathode material is deposited rapidly on the lamp envelope to decrease the operating lifetime. These effects were exemplified in early sealed off hollow cathode lamp sources for arsenic and selenium and also prevent production of satisfactory sealed off lamps for elements such as sulfur and iodine. Recent studies in our laboratory have been concerned with the development of direct AAS methods for the determination of a number of volatile elements whose useful resonance lines lie below 200 nm. At these short wavelengths, where high background absorption is encountered from the atmosphere and also the analytical flame or non-flame atom cell, it is necessary to employ an inert gas purged or evacuated optical path and an intense sharp line source in order to undertake AAS. The use of this type of experimental assembly has recently been described for the direct determination of sulfur ( 2 ) ,phosphorus ( 3 ) ,and iodine ( 4 ) by AAS using a nitrogen separated nitrous oxide-acetylene flame and microwave excited electrodeless discharge lamp (EDL) sources for these elements. While the latter sources may exhibit intense emission a t their short wavelength resonance lines, it is difficult with elements of high volatility to ensure the reproducible manufacture of stable sources; in addition, the EDL sources for sulfur and iodine must be operated a t low power to avoid excessive self-absorption of the emitted radiation. Intense phosphorus EDL’s are difficult to prepare. In the search for an alternative suitable line source for these elements, the possible advantages to be gained from the use of a demountable hollow cathode lamp source were considered. The use of such a source should minimize the disadvantage encountered in sealed off hollow cathode lamps from limited operating lifetime, and the use of a continuous flow of inert gas in such a lamp should result in greater freedom from the self absorption effected by the high concentration of atomic vapor otherwise present in front of the cathode in sealed-off sources. Under these conditions, higher operating currents might be possible to provide sharp line sources of high intensity. Many types of demountable hollow cathode lamp have been described in the literature for use in atomic emission and absorption spectrometry. In several recent studies of the application of these sources to analytical atomic absorption ( 5 , 6 ) and fluorescence (6-8) spectrometry, their (1) A . Waish. Spectrochim. Acta, 7 , 108 (1955). (2) G. F. Kirkbright and M. Marshall, Anal. Chem., 44, 1288 (1972). (3) G. F. Kirkbright and M. Marshall, Anal. Chem., 45, 1610 (1973). (4) G. F. Kirkbright, T. S. West, and P. J. Wilson, At. Absorption Newsleft., 11, 53 (1973).
Figure 1. Demountable hollow cathode lamp assembly advantages with respect to operation a t high current to produce intense line emission and minimal self absorption have been described. Most work, however, has been confined to the examination of their suitability as sources for AAS or AFS of the metallic elements. This paper describes an evaluation of a commercially available demountable hollow cathode lamp as a sharp line source for the determination of the volatile elements arsenic, selenium, mercury, sulfur, iodine, and phosphorus which have principal resonance lines a t wavelengths less than 200 nm. A comparison of this source with the corresponding microwave excited electrodeless discharge lamps for these elements is described.
EXPERIMENTAL Apparatus. A “Miniglow” demountable hollow cathode lamp system (Spectro Products Inc., North Haven, Conn.) was employed. The construction of the lamp is shown diagrammatically in Figure 1. This unit was operated from a stabilized power supply (Spectro Products Inc.) capable of operation between 0 and 100 mA, and in conjunction with a gas control unit for precise control of fill gas pressure between 0.1 and 10 Torr. The cathode assembly and lamp body may be water cooled and the inert filler gas (argon) flow rate may be adjusted uia a needle valve fitted to the gas control unit. In the present work, a graphite cathode (Poco Graphite Inc., Decatur, Texas; porous cup electrode 25.4 mm in length, 6.15-mm 0.d. and 3.96-mm i d . ) and a brass anode were employed. For excitation of the atomic line spectra of the elements investigated, these cathodes were filled with elemental sulfur (laboratory grade, resublimed), iodine (analytical reagent grade), arsenic (laboratory reagent grade), and selenium (Specpure, Johnson and Matthey, Royston, Herts). A Techtron Model AA4 flame spectrophotometer (Varian-Techtron Pty., Melbourne, Australia) was modified for operation at wavelengths less than 200 mm as described elsewhere (9). The photomultiplier used was an EM1 9783A side window type selected for maximum gain. A nitrous oxide-acetylene slot burner (50 mm in length) with provision for flame separation by inert gas shielding was used in conjunction with the indirect nebulizer and expansion chamber of the AA4 flame spectrometer. The source radiation was modulated a t 400 Hz using the internal modulation available from the ac amplifier (Spectro Products Inc.) employed. The optical path between the flame and monochromator and the monochromator itself was nitrogen purged. Electrodeless discharge lamp sources were operated in a nitrogen purged 3h-wave resonant cavity at 2460 MHz using a 200-W generator (Electromedical Supplies Ltd., Wantage, England). The source radiation was modulated a t 285 Hz from a Microtron modulator unit driven in synchronization with the ac amplifier of the Techtron AA4 spectrometer. Arsenic, selenium, and mercury atomic absorption studies were undertaken using a premixed air-acetylene flame supported at a long path (100 mm) burner. (5) G. Rossi and N. Omenetto. Appl Spectrosc., 21, 329 (1967). (6) R. A. Woodriff, G. V. Wheeler, and W. A. Ryden, Appl. Spectrosc., 22, 348 (1968). (7) J. I . Dinnin and A. W. Helz, Anal. Chem., 39, 1489 (1967). (8) G. Rossi and N. Omenetto, Talanta, 16, 263 (1969). (9) G. F. Kirkbright, T. S. West, and P. J. Wilson, Anal. Chim. Acta, 66, 130 (1973).
ANALYTICAL CHEMISTRY, VOL. 46, NO. 11, SEPTEMBER 1974
1415
./ 14
1L'2
-
10
:
l2
10-
1,.
10,
n
I
1
2
A
1 0
i 2'0
1'0
3 0
L O
50
60
'0
CU R R F NT hn N
L '0
3'0
20
Figure 3. Effect of lamp-operating current on emission intensities
ARGON PRFSSURF(Torr1
Figure 2. Effect of argon filler gas pressure on line emission intensi-
obtained for resonance lines of iodine, sulfur, and mercury
ties for elements studied
Table I. Sensitivities and Detection Limits Obtained Using the Demountable Hollow Cathode Lamp (DHCL) and EDL Sources for AAS as the Vola tile Elem en t s
Element
Wavelength, nin
Flaine
Source
As
193.7
Air-acetylene
Se
196,O
Air-acetylene
Hg
253.7
Air-acetylene
I
183.0
S
180.7
N, separated N,O-C.H, N, separated N,O-C,H?
DHCL EDL DHCL EDL DHCL EDL DHCL EDL DHCL EDL DHCL DHCL
182,O 182.6 a
!I
I,
I,
I ,
Table 11. Optimum Operating Conditions for t h e Demountable Hollow Cathode Lamp Source
Element
As
Sensitlvlty, fig ml Deteclpc tion absorp- limit, gg ml tion
0 9 2 0 053 2 09 1 6 2 5 14 0 12 0 1 0 9 0 3 0 5 0
Se
0 45
0 0 0 On 0 Ob 0 0
S
P
Fromreference 14). From reference ( 2 ) .
RESULTS A N D DISCUSSION The cathode was prepared for each element by placing a small quantity (ca. 1 mg) of the elemental form a t the base of the hollow graphite electrode. Each source was then operated for a few minutes at 7 mA until a stable discharge was obtained, and the variation of analytical line intensity with argon filler gas pressure and lamp operating current was determined. Figures 2 and 3 show typical results obtained in these experiments. The signal-to-background intensity ratio a t the primary resonance line and the long term stability were also studied for each source. T h e results are summarized below for each element. A study of the use of the demountable hollow cathode lamp as a source for AAS of each element was carried out, and the results are again summarized below with the sensitivities and detection limits obtained also shown in Table I. Arsenic. A wavelength scan of the lamp emission spectrum between 230 and 190 nm produced an arsenic spectrum with intense line emission at 193.7, 197.1, 200.3, and 228.8 nm. The spectrum in this wavelength range also exhibited a weak copper line a t 194.2 nm from the anode and, a t fill gas pressures greater than 4.0 Torr, the carbon 1416
Relative emission intensities1
193.7 197.1 228.8 200.3 196.0 206,3 204,.0 207.5
1. o 0.5 1.7 0.3 1.0 0.25 1.1 4.5 1.0 2.6 1.0 0.34 0.31 0.05 1.0 2.4 1.3 0.05 0.05 1.0 1.5
184.9
1 0 2 3 22 25 5 30 5 10
Emission wavelength, nm
a
253.7 183.0 184,4 187.6 178.2 180.7 182.0 182.6 177.5 178.3 213.6 253.6
Operating current, mA
Argon pressure, Torr
2
1.0
2
0.7
1.5
0.7
1.5
2.5
2.0
1.5
2.5
1.5
Uncorrected for response characteristics of photomultipler used.
193.0-nm line. The relative emission intensities of the arsenic lines and the optimum operating conditions are summarized in Table 11. The background a t the 193.7-nm line was very low, so that a signal-to-background intensity ratio of greater than 200:l was obtained at this wavelength. A sensitivity for the determination of arsenic in aqueous solution of 0.9 pg/ml (for 1%absorption) and a limit of detection of 0.45 pg/ml was obtained a t the 193.5-nm line using an air-acetylene flame. The sensitivity obtained compares favorably with that obtained using an electrodeless discharge lamp in the same instrumental arrangement (see Table I). Selenium. The selenium 196.0, 203.99, and 207.5 lines were emitted with high intensity from the source under the optimum conditions shown in Table 11. A signal-to-background intensity ratio of approximately 200:l was observed a t the primary resonance line at 196.0 nm. This compares very favorably with the signal-to-background intensity ratio observed from an electrodeless discharge lamp source a t this wavelength. Using the air-acetylene flame, a sensi-
ANALYTICAL CHEMISTRY, V O L . 46, NO. 11. SEPTEMBER 1974
183 O
I
I
387 6
ED1
I -
I
I
HOLLOW CATHODF
~
DHCL
SIGNAL 1807nm
182.6
LPL
Figure 4. Observed relative intensities of iodine 183.0, 184.4, and 187.6 lines emitted by demountable hollow cathode source and EDL tivity (1%absorption) of 0.53 pg/ml and a limit of detection of 1.0 pg/ml were obtained for the determination of selenium in aqueous solution using the 196.0-nm line, a slightly fuel-lean flame, and a height of observation between 2 and 7 mm above the burner. These again compare favorably with the corresponding values obtained using an EDL source under similar operating conditions. Mercury. A minimum (less than 0.5 mg) of mercury in the elemental form or a few milligrams of solid HgC12 added to the graphite cathode resulted in intense emission a t the mercury 253.7-nm and 184.9-nm lines. Table I1 shows the optimum operating conditions established using elemental mercury in the cathode shell. The high intensity of the line emission for mercury enabled operation of the instrumental gain and spectrometer slit width settings a t their minimum values. Under these conditions, no measurable background a t r'l 8, from the peak wavelength of the line was observed a t 253.7 nm and 184.9 nm. With a 10-cm air-acetylene flame a t the 253.7-nm line, a sensitivity (1% absorption) of 1.6 pglml for the determination of mercury in aqueous solutions of mercury(1) nitrate and a limit of detection of 2.0 ,ug/ml were achieved. These values compare with a sensitivity of 2.5 pg/ml and a limit of detection of 4.0 pg/ml for Hg(I)N03 using an EDL with the same spectrometer. Iodine. A wavelength scan of the lamp emission spectrum from iodine crystals or from mercury(I1) iodide in the graphite cathode showed strong iodine line emission a t 183.0, 184.4, 187.6, and 206.1 nm, as well as a weak line emission a t 178.2 nm. The spectral scan shown in Figure 4 was obtained using elemental iodine in the cathode and is compared with that from a microwave excited electrodeless discharge lamp source operating under optimized conditions. The figure shows that the relative intensities of the 183.0-, 184.4-, and 187.6-nm lines are different from those obtained from the EDL source; with the hollow cathode source, the useful iodine 183.0-nm ground state line shows the greatest intensity whereas in the EDL source the iodine 187.6-nm non-resonance line is the most intense. The background a t the 183.0-nm line was negligible with the hollow cathode lamp compared to that from the EDL where a background intensity as high as 8% of the peak line intensity a t 183.0 nm was obtained. With either iodine crystals or mercury(I1) iodide in the graphite cathode of the demountable lamp, a sensitivity (1%) absorption) of 1.4 ,ug/ml for the AAS determination of iodine in aqueous solution was observed using a nitrogen separated nitrous oxide-acetylene flame. The corresponding detection limit was 22 pg/ml. These values compare with a sensitivity of' 12 pg/ml and a detection limit of 25
i.....__
B A C U G W
.....
Figure 5. Observed relative intensities of sulfur ground state lines emitted by demountable hollow cathode source and EDL and signal: background ratio at 180.7 nm for each source with optimized operating conditions ,ug/ml using an EDL source reported previously using a similar spectrometer ( 4 ) . Sulfur. I t is for sulfur that the demountable hollow cathode lamp source shows the most marked improvement in characteristics for atomic absorption work when compared to the corresponding electrodeless discharge tube source. Using resublimed elemental sulfur (1 mg) in the graphite cathode, the lamp emission spectrum exhibited strong line intensity a t the sulfur 180.7-, 182.0-, and 182.6nm lines when a lamp current of only 2 to 5 mA was employed. The intensities of the line emissions a t these wavelengths from the lamp were four times greater than those obtained from an EDL under optimized conditions. A signal-to-background intensity ratio a t 180.7 nm of 50:l was observed compared to a ratio of 1O:l achieved with a sulfur EDL source (see Figure 5). The relative emission intensities of the three lines are different from those obtained with a sulfur EDL source and reflect the different excitation conditions in the hollow cathode source (see Table 11). The direct determination of sulfur by AAS has been reported by L'vov and Khartsyzov (10) and Kirkbright and Marshall (2). In the former work, AAS sensitivities and detection limits are reported a t three sulfur lines for a graphite furnace atom cell and a vacuum monochromator in conjunction with an EDL source. In the latter work, a nitrogen shielded nitrous oxide-acetylene flame, a fixed channel vacuum monochromator, and an EDL source were employed, and the determination was carried out a t the sulfur 180.7-nm line only. With the arrangement described above and a nitrogen shielded nitrous oxide-acetylene flame, the demountable hollow cathode lamp (DHCL)was employed as a source in the determination of sulfur by AAS at its 180.7-, 182.0-, and 182.6-nm lines. Calibration graphs for the three lines are shown in Figure 6, as is the calibration obtained by Kirkbright and Marshall ( 2 ) for sulfur determination a t 180.7 nm using an EDL and a vacuum monochromator. The graphs for 180.7, 182.0, and 182.6 nm are linear in the range 10-50 pg/ml, 10-200 pg/ml, and 10-350 pg/ml, respectively. The sensitivity (1% absorption) a t 180.7 nm obtained when sulfur was introduced into the flame as an aqueous solution of ammonium sulfate was found to be 1.0 pg/ml. A limit of detection of 5.0 pg/ml was obtained compared to 30 ,ug/ml obtained by Kirkbright and (10) B V L'vov and A. D. Khartsyzov, Zh. PrMSpectrosk, 11, 413 (1969)
ANALYTICAL CHEMISTRY, VOL. 46, NO. 1 1 . SEPTEMBER 1974
1417
,4807 D H C L
A/
I1830nm As 1937nm S 180 7 n m
06
Hg 1 8 4 9 n m S e 1 9 6 Onm
m
01
-
100
2 00
3JO
tl
k DL
L 00
Figure 6. Calibration graphs for sulfur AAS in nitrogen-shielded nitrous oxide-acetylene flame. Calibration graph at 180.7 c m with EDL source from reference (2) Figure 7. ( A ) . Recorder tracings c stability of line intensity r each Effect of element (100% transmission setting for A A S work). (4. successive lamp ignitions on observed intensity of Hg 184.9-nm line ~
Marshall (2). Although the 1%absorption sensitivity a t the 182.0- and 182.6-nm lines was poorer than a t 180.7 nm, the limits of detection were very similar; this can be explained by the fact that the flame noise decreases with increasing wavelength. A summary of the sensitivities obtained a t each wavelength for sulfur is shown in Table I. Phosphorus. The phosphorus resonance lines a t 177.5, 178.3, and 178.8 nm were emitted only weakly from the hollow cathode source when 1 mg of red phosphorus was used in the graphite cathode. This may be due to the instrumental and optical arrangement, however, as the phosphorus 213.5-, 213.6- and 253.5-nm non-resonance lines are observed with appreciable intensity under the same conditions. I t is likely that a vacuum monochromator, a more sensitive photomultiplier, and calcium fluoride windows and lenses would be required in a further study of the direct determination of phosphorus by AAS a t the short wavelength phosphorus resonance lines; no study of phosphorus AAS with the demountable hollow cathode source was possible with the instrumental arrangement employed in this work.
CONCLUSION In recent studies in our laboratory concerned with the direct determination by AAS of the non-metallic elements whose resonance lines lie a t wavelengths less than 200 nm, much effort has been devoted to the construction of intense and stable EDL sources for these elements. For sulfur, phosphorus, and iodine, we have experienced some difficulty in achieving intense emission a t the low wavelength resonance lines, particularly if source line broadening due to self absorption is to be minimized. In addition, it has frequently proved difficult to produce high signal-to-background intensity ratios a t these wavelengths for sulfur. The long term stability and relatively poor reproducibility of the emission intensity each time the source is operated, may also prove troublesome. As described here, the de-
1418
mountable hollow cathode lamp source can provide intense emission and high signal-to-background intensity ratios a t the short wavelength resonance lines for arsenic, selenium, mercury, iodine, and sulfur. In addition, good long term stability is attainable for the demountable hollow cathode source with these elements. Figure 7 shows the stability of the 100% transmission setting-ie., zero absorbance-obtained over a period of 2% hours when the hollow cathode source was operated with the recommended conditions and with the spectrometer adjusted to give maximum sensitivity for atomic absorption studies. I t is evident that the output intensity for each element is stable to within ca. 1-2% over the period investigated. I t is probable that the stability is assisted by the use of a continuously purged system a t the optimum pressure within the lamp and the use of a water cooled cathode. Figure 7 also shows the reproducible output intensity obtained for mercury a t the 184.9-nm line when the lamp is alternately switched off and on. The demountable hollow cathode lamp sources provide readily for the rapid interchange of graphite cathodes containing different elements. No difficulties have been observed in operation with deposition of the cathode filler material on the internal surfaces of the lamp envelope with the water-cooled cathode and low operating currents employed; the operating lifetime under these conditions for a single cathode containing only milligram amounts of the element whose atomic line spectrum is required, therefore, is also considerable even for the volatile elements studied. Thus, for iodine and sulfur, a useful operating lifetime for each cathode fill is estimated as a t least fifty hours.
RECEIVEDfor review December 10, 1973. Accepted April 22, 1974.
ANALYTICAL CHEMISTRY, VOL. 46, NO. 11, SEPTEMBER 1974
