Analytica
Chimica
Acta,
67 (1973) 79-87 Company,
0 Elscvier Scientific Publishing
THE DETERMINATION SPECTROMETRY WITH D. J. JOHNSON Chemistry
Amsterdam
- Printed
OF GERMANIUM A GRAPHITE TUBE
in The Netherlands
BY ATOMIC ATOMIZER
79
ABSORPTION
and T. S. WEST
Department,
Imperial
College
of Scierlce and Technology,
Londotl
S W7 2.4 Y (England)
R. M. DAGNALL Biochemistry (Received
Division, 25th April
Htmtirlgdon
Research
Cemrc.
Htmtingdotl
(England)
.
1973)
The determination of germanium by atomic spectrometric methods is difficult, owing to the formation of a highly stable oxide species which precludes the efficient production of germanium atoms. Amos and Willis’ reported a sensitivity of only 120 pg Ge ml -’ for 1% absorption at the germanium 256.12, 256.16-nm doublet in a fuel-rich air-acetylene flame, and almost identical results in an airand nitrous oxide-acetylene hydrogen flame. Using nitrogen-oxygen-acetylene flames, the same aqhors obtained sensitivities (by atomic absorption spectrometry) of. 5.5 and 1.5 pg. Ge ml-‘, respectively. Manningz, using a nitrous oxideacetylene flame found a sensitivity of 2.5 pg Ge ml-’ and a limit of detection of 1 c(g Ge ml- i. Popham and Schrenk3 found a limit of detection of 0.5 pg Ge ml-’ with the same flame system, but with a (1 -t 1) acetone-water solvent system. With a reducing oxy/acetylene flame, the limit of detection WCS 5 /cg Ge ml-‘. The low background of the argon-separated nitrous oxide-acetylene flame has enabled the limit of detection to be lowered to 0.2 fig Ge ml-l (for a signal-noise ratio of 2)4. Other workers5-8 have measured the thermal emission at the 256.2-nm doublet of germanium in nitrogen-oxygen-acetylene and nitrous oxide-acetylene flames, Pickett and Koirtyohann 5*6 finding a detection limit of 0.5 pg Ge ml-‘. Bratzel et aLg demonstrated the feasibility of determining germanium by atomic fluorescence spectrometry, and Dagnall et ~1.’ using a germanium electrodeless discharge lamp light source and an argon-separated nitrous oxide-acetylene flame obtained a limit of detection of 0.13 fig Ge ml-l (S/N = 2). A non-dispersive atomic fluorescence system has also been used”. tube atomizer and the carbon lilament atom reservoir have The graph been used by c;mvous WC,kcrs 11-14. The detection limits (in terms of concentration) by this metLoJ have generally been as good as or better than those found for flame methods, with the advantage that only very small (~1) quantities of sample are required. These systems are also able to atomize such refractory elements as vanadiumil* 15, molybdenum”* 16*” and titaniumi’, and it would seem that this technique would lend itself to the determination of germanium. In this work, a study is made of the use of both the graphite tube and the carbon filament reservoir as methods for the determination of germanium.
80
D. J. JOHNSON,
T. S. WEST,
R. M. DAGNALL
EXPERIMENTAL
Apparatus The carbon filament assembly used was, apart from minor modifications to the cooling water supply, identical to that described by Maines’s and by West et al.1g-22. The graphite tube atomizer was designed to fit directly on top of the stainless steel pillars of the carbon filament reservoir, in place of the normal filament clamps. These two assemblies were easily interchanged. The graphite tube assembly is shown in Figs. 1 and 2. The graphite tube was machined from Ringsdorff 6-mm spectrographic graphite rod. The internal diameter was 4.8 mm (no. 12 drill) and the overall length 20 mm. A 2-mm hole, drilled at the midpoint, enabled a sample to be introduced into the tube. Two transverse slots ca. 1 mm deep, were cut in the tube 4 mm from the mid-point; the effect of these was to produce a localized heating at the centre of the tube. The ends of the tube were supported in graphite (Morganite EY 110) split rings (6 mm i.d., 9 mm o.d. x 3 mm long), which were clamped firmly in the stainless steel end-supports by means of two pinch bolts. The split rings were found necessary, in order that good electrical contact was made at the tube ends, and also to prevent overheating and melting of the end-supports at the point of contact. Only one of the supports was screwed to the pillar; the other was free-floating, electrical connection being made via flexible copper braiding (not shown). This arrangement prevented breakage of the tube by misalignment or expansion on heating. The tube assembly was enclosed in a Pyrex cell, which rested on the base plate. Two diametrically opposed holes (+-in. diameter) allowed transmission of the light beam. The cell top was removable, enabling a sample to be placed easily in the tube. The cell was flushed with nitrogen via the gas box normally used for the carbon filament, and also via a second gas inlet in the base plate. A total nitrogen flow rate of ca. 3 1 min- ’ ensured that the tube operated in an inert atmosphere. Power was supplied by a 20: 1 stepdown transformer, the primary voltage of which was controlled by a Variac transformer from the mains supply. This supply was originally designed for use with the carbon filament and was rated at 1.2 kW (continuous). Current and voltage measurements made with the graphite tube indicated that the power dissipated by the tube was ca. 3.5 kW but, as the tube was generally only switched on for 2 s or less, it was felt that this overload was acceptable. A Southern Analytical A1740 grating photometer, fitted with an EMI 9662B photomultiplier. was used. The slit-width was 0.06 mm, corresponding to a bandpass of 0.35 nm. The normal integration and readout facilities of this instrument were by-passed and the output was taken directly from the photomultiplier to the Y-axis input of a Telequipment DM53A Storage Oscilloscope, fitted with a “K” type amplifier. A 22,000 pF capacitor was connected across the input terminals, in order to reduce high-frequency noise; this increased the effective response time of the oscilloscope to ca. 75 ms f.s.d. A germanium electrodeless discharge lamp was used as a spectral source. This was made from germanium metal and iodine (amounts estimated as less than 1 mg) sealed under a pressure of 3 Torr of argon. Microwave power was provided
GERMANIUM
BY GRAPHITE-TUBE
A.A.S.
Shield
Fig 1. Graphite
tube nsscmbly.
Stainless End
Steel
Supports
Carbon
Fig. 2. Graphite
Gas
Split
tube assembly in detail.
by a Beckman “Microgen” (2450 MHz) unit, and coupling was via a fore-shortened s-wave cavity. The source was initiated by a spark from the EHT supply of the “Microgen”. In order to obtain good rejection of radiation emitted from the heated tube, the following optical arrangement was found satisfactory. The radiation ‘from the source was focused by a lens at the mid-point of the tube, and a second lens focused the radiation onto the monochromator slit. Two light stops (in the form of 2-mm diameter holes) were placed either side of the tube, so that only a narrow “pencil” of radiation was transmitted.
82
D. J. JOHNSON,
T. S. WEST.
R. M. DAGNALL
Solutions
A lOOO-p.p.m. germanium stock solution was made by dissolving 0.380 g of germanium dioxide in the minimum of 1 M sodium hydroxide. This was acidified with 1 M hydrochloric acid and diluted to 250 ml with deionized water. All other solutions were made from AnalaR-grade materials. PRELIMINARY
STUDY
Carbotr filament
atom reservoir
With the carbon filament, the germanium atomic absorption at 265.2 nm was investigated. A lOOO-p.p.m. solution ( 1 ~1) gave a large (> 990/o absorption) signal, but no signal at all was observed with a 500-p.p.m. solution. Absorption signals could be obtained from a lOO-p.p.m. solution, provided that the filament was pre-treated by heating to red heat in air before each sample was placed on the filament. However, the reproducibility of the signals was extremely poor (typically > 50%), and the filament had a lifetime of only about ten such treatments. It was concluded that the carbon filament did not provide a usable technique for the determination of germanium. Workers who have studied the Ge/GeO,/C systems23-25 indicate that the reaction GeO,+C
-+ GeO+CO
predominates when germanium dioxide is heated in the presence of carbon. Germanium monoxide is volatile at the temperature of the reaction, and sublimes as such without undergoing further reduction. If a thermal degradation of the germanium sample to Ge02 first occurs, then it would be expected that germanium will be largely lost as volatile GeO; little atomization will occur and hence little absorption will be observed. Graphite tube
One of the major differences bFt!ween the carbon filament and the graphite tube is the residence time of atoms in a high-temperature environment. With the carbon filament, this time is of the order of lo-’ s, whereas with a tube it can be of the order of seconds. It was thought that the use of a tube would facilitate the formation of germanium atoms by retaining the GeO species within the tube until the temperature had risen sufficiently for breakage of the Ge-0 bond to occur. Preliminary investigations confirmed this; a reproducible germanium atomic absorption signal was observed from samples containing less than 10 ng of germanium. It was decided, therefore, to evaluate fully the graphite tube technique for the determination of germanium. OPTIMIZATION
OF OPERATING
PARAMETERS
The germanium electrodeless discharge lamp emitted 23 germanium lines between 204.3 and 422.7 nm. Of these 12 were resonance lines. The absorbances of a lo-ng germanium sample at some of the most intense of these lines are shown in Table I. The unresolved 265.12, 265.16-nm line pair was chosen as the most
GERMANIUM TABLE
BY GRAPHITE-TUBE
I
WAVELENGTHS, GERMANIUM Wuvelettgth
209.4 249.8 253.3 258.9 259.2 265.161 265.121 269.1 270.9 275.5
83
A.A.S.
(ml)
SOURCE INTENSITIES IN A HEATED GRAPHITE
AND TUBE
ATOMIC ABSORPTION ATOMIZER
Ahsorhar~~ (10 r1g Ge)
Semitivity (1% crhsorptiort) (y)
3 7 7
0.155 0.013 0.000
2.8. lo-‘” 3.4.10-g
53
0.180
2.4. IO-”
100
0.305
1.5. lo-
41 49 47
0.123 0.178 0.129
3.6. IO-” 2.5. lo-lo 3.4*10-‘0
Relative of lirw
htensiry
SENSITIVITY
FOR
‘O
useful because of its high sensitivity and high signal-background pair is subsequently referred to as the 265.2-nm “line”.
ratio.
This line
Electrodeless discharge lamp powe, The absorbance of a germanium sample at 265.2 nm was measured at various microwave powers. This indicated that a low power was desirable. The absorbance was approximately constant below a microwave generator current of 10 mA but dropped by cn. 25”/, at 20 mA and cu. 35% at 25 mA. A current of 10 mA was normally used. Variac
setting The shape of the germanium absorption signal varied with Variac setting. Below CU. 70% power (cu. 8 V), no absorption peak was observed although the sample was lost from the tube, as was evidenced by the lack of an absorption peak on re-heating the tube at maximal power. This supports the view that loss of germanium as GeO is important; in this instance, the temperature was too low to effect atomization of the oxide. Above this power setting, an absorption signal was observed which increased in peak height and decreased in width as the power was increased. As all measurements were made from the peak height, the maximal power available was used in order to obtain maximal sensitivity. At the highest power setting, the tube reached maximal temperature (cu. 3200”) in less than 2 s. The germanium absorption peak commenced 0.14 s after the power was switched on and reached a maximum within 0.2 s. The overall peak-width was cu. 1 s. After the tube had been used a number of times, a second absorption peak occurred which commenced cu. 1 s after the power had been switched on. This was also observed at the germanium 303.9-nm (non-resonance) line and at 253.6 nm when a mercury electrodeless discharge lamp was substituted as light source; it was concluded that this peak was due to scattering of the light by ‘carbon particles, produced by sublimation from the tube walls.
D. J. JOHNSON,
84
T. S. WEST,
R. M. DAGNALL
Recommended procedure In order to obtain good reproducibility of results, the following timing procedure was employed. (i) t =O s. Switch on power at cu. 5”/, and set nitrogen shield gas at 3 1 min- r. This “pre-heat” of the tube to 70-80” helps to prevent the sample spreading excessively. (ii) t=45 s. Place sample in tube. Increase power to ca. 7”/0 to evaporate water. Check oscilloscope traces viz. 100°/Oand 0% absorption levels. (iii) t = 75 s. Switch power off. Set Variac to 100°A power. Switch on power for cu. 1.5 s, to volatilize sample. Measure height of absorption peak. (iv) t= 5 min. Repeat procedure. This long period was necessary to allow the tube and end-supports to cool; particularly the support which was “freefloating” and which made poor thermal connection with the water-cooled pillar. In order to reduce the time for one measurement, it would be preferable to water-cool the end-supports directly. Red
ts
A summary of results is shown in Table II. The calibration curves for l-, lo- and 20-,~l samples,‘except for a slight curvature at the lower end, were linear as far as useful measurements could be made (cu. 950/,absorption). This corresponded to a maximum sample of ca. 50 ng (5. 10e8 g). A slight dependence of sensitivity upon sample size was observed. The maximal absorbance from 10 ng of germanium was observed with a l-p1 sample (10 p.p.m.); the absorbance of a 5-,~l sample (2 p.p.m.) was less by 4%, a lo-p1 sample (1 p.p.m.) by 6%, and a 20-~1 sample (0.5 p.p.m.) by 10%. A 50-~1 (0.2 p.p.m.) sample showed an approximately 15% reduction in signal, but this size of sample was difficult to use because of “bumping” and consequent loss of sample on evaporation. The standard deviation (15 consecutive measurements) for a lo-ng sample (1 ,ul x 10 p.p.m.) was 2.7% and for a 2-ng sample (20 ~1 x 0,l p.p.m.) was 6.5%. TABLE
II
SENSITIVITY
AND DETECTION
Sample oolttme (IO ng Gr)
1 Pl
10 /.ll 29 Pl a Scale expansion
LIMIT
Sensitivity (for 10/0 absorptiotl) B
p.p.01.
1.5~10-‘0 1.7~10-‘0 1.8~10-‘0
0.15 0.017 0.009
DATA FOR GERMANIUM Limit of tletectioit Ct
p.p.lPl.
3.10-‘0 3*10-‘0 3. lo- ‘0
0.3 0.03 0.0 15”
Range (p.p.nl.)
0.3-50 0.03-5 0.015-2.5
( x 3) used.
Interferences The effects of a number of foreign anions and cations on the absorbance of a IO-ng (1 p.p.m. x 10 ~1) germanium sample are summarized in Table III. In the tests given as “Interference”, the germanium absorption signal was superimposed on a highly irregular background absorption caused by either molecular absorption
GERMANIUM
BY GRAPHITE-TUBE
A.A.S.
..
85
or light scatterby the interfering species (confirmed by using the mercury 253.6-nm line, when the background was still observed). It was not possible to resolve the germanium absorption peak from this high background absorption. Attempts to remove the interfering species by selective volatilization failed, because of the high volatility of germanium oxide which resulted in the sample being lost when more than 35”/, power was used. However, to remove, e.y. sodium chloride, it was uecessary to use cu. 60”/” power. TABLE
III
INTERFERENCE
DATA
(10 ng Ge in 10 111containing
foreign
1000-fold NZl CLl
M6 Mn co Zn Cd H6 Pb Sn S P i=l
NaCl CaCI, MgSG, MnCl, COCI, Zn( NO& CdS04 H&l2 Pb(NWz Sn/6 M HCI l&so4 Na,HPO, HCI
a lntcrfcrencc
caused
’
ions at specified
e.~ccss
level)
loo-$old
oucr Gr
over Ge
InterLU InterT.U Interf.” Intcrf.” -30 +25 0 -70 0 -85 0 - 10
0 0
by non-resolution
cxcP.ss
-45 0 0 +25 0 -25 0 0 - 10 0 oT signal
from
background
absorption.
DISCUSSION
It has been shown that the determination of germanium with a carbon filament atom reservoir is- not feasible, because of the pcor efhciency of atom formation. Other workers l7 have reported similar results where volatile oxide species are present, e.g. silica. As mentioned previously, a heated tube facilitates atomization owing to the longer residence time of species in it. However, the rate of increase in temperature of the tube is also important. If, say, germanium oxide is volatilized at cu. lOOO”, but no germanium atoms are produced below CQ. 3000”, then it is important that the time taken for the tube temperature to rise from 1000” to 3000” should be as short as possible to prevent loss of sample as germanium oxide. For this reason, the small slots were cut in the tube. A more rapid temperature rise, and an improvement in sensitivity were found with larger slots, but the higher localized temperature attained reduced the lifetime of a tube considerably. The normal useful life of a tube was 50-80 “flashes”, although it would be expected to be considerably longer for those elements not requiring such high temperatures as germanium.
56
D. J. JOHNSON.
T. S. WEST.
R. M. DAGNALL
A further advantage of the graphite tube technique over the carbon filament, is the possibility of using a larger sample. An often-quoted advantage of the carbon filament is the small sample required, but when the sensitivity (in p.p.m.) is insufficient for an analysis as in this instance, it is not possible to improve it by increasing the sample size. The tube does not suffer so severely from this limitation and even the largest sample volumes used (50 ~41) are small compared with the requirements of many other techniques. One of us (D.J.J.) wishes to thank Imperia1 Chemical Industries Limited, Agricultural Division, Billingham, Teeside for the award of a research grant to carry out this work. SUMMARY
A carbon filament atom reservoir and a graphite tube atomizer are compared as methods for the determination of germanium by atomic absorption spectrometry. It is shown that the filament technique is not applicable, probably because of loss of the sample as a volatile oxide species which results in inefficient atomization. The design of a small graphite tube atomizer is described; with this, a limit of detection for germanium of 3. IO- lo g was found. For a 20-4 sample, this corresponds to a concentration of 0.015 p.p.m. The interfering effect of 13 anions and cations is studied.
Une comparaison entre reservoir atomique a filament de carbone et atomiseur a tube de graphite est faite, en vue du dosage du germanium par spectrometrie d’absorption atomique. La technique du lilament nest pas applicable, probablement en raison de pertes dues a la formation d’un oxyde volatile. L’atomiseur a petit tube de graphite est decrit. I1 permet d’arriver a une limite de detection de 3 - lo-” g de germanium. Ce qui correspond a une concentration de 0.015 p.p.m. pour un Cchantillon de 20 jtl. On examine Cgalement l’influence de 13 anions et cations. ZUSAMMENFASSUNG
Eine Kohlenstoff-Heizfaden-Kiivette und ein Graphitrohr-Verdampfer werden als lvlethoden fur die Bestimmung von Germanium durch Atomabsorptions-Spektrometrie verglichen. Es wird gezeigt, dass das Heizfaden-Verfahren nicht anweudbar ist, wahrscheinlich weil Probenverluste in Form einer fliichtigen OxidSpezies auftreten und die Atomisierung unwirksam werden lassen. Der Aufbau eines kieinen Graphitrohr-Verdampfers wird beschrieben; hiermit wurde eine Nachweisgrenze fiir Germanium von 3 - lo- lo g erhalten. Fur eine Probe von 20 111 entspricht dies einer Konzentration von 0.015 p.p.m. Der stiirende Einfluss von 13 Anionen und Kationen wird untersucht. REFERENCES 1 M. D. Amos and J. B. Willis, Spccrrodtim
Acru.
22 (1966)
1325.
GERMANIUM
BY GRAPHITE-TUBE
A.A.S.
2 D. C. Manning, AL Absorption Netvslert., 6 (1967) 35. 3 R. E. Popham and W. G. Schrcnk. Spectrocltinr. Acra, 23B (1968) 543. 4 G. F. Kirkbright, M. Sargent and T. S. West, Trrlurw, 16 (1969) 1467. 5 E. E. Pickett and S. R. Koirtyohann. Specrrochim. Aclu, 23B (1968) 235. 6 E. E. Pickett and S. R. Koirtyohann. Spectrochh. Acra, 249 (1969) 325. 7 R. M. Dagnall. K. C. Thomson and T. S. West, Amrl. ChinI. Acra, 41 (1968) 551. 8 R. M. Dagnall. G. F. Kirkbright. T. S. West and R. Wood, Amlyst. 95 (1970) 425. 9 M. P. Bratzel. J. M. Mansfield. J. D. Winefordncr and K. E. Zacha, And. Chew.. 40 (1968) 10 P. L. Larkins and J. B. Willis, Specrrochinl. Acm, 26B (1971) 491. 11 B. V. L’Vov, Spectroclrim. Ackc, 24B (1969) 53. 12 H. Massman, Spctrochittr. Acrcr. 23B (1968) 215. 13 T. S. West and X. K. Williams. Ad. Chirn. Acto. 45 (1969) 27. 14 M. D. Amos, Amer. Lab., August (1970) 33. 15 K. W. Jackson, T. S. West and L. Balchin. Aml. Clwn., 45 (1973) 249. 16 R. M. Dagnall. D. J. Johnson and T. S. West, AII~J/. Chim. Acru. (1973) in press. 17 H. M. Doncga and T. E. Burgess. AMJ/. Chvn.. 42 (1970) 1521. 18 1. S. Maims. Ph.D. Thesis. University of London, 1970. 19 R. G. Anderson. I. S. Maincs and T. S. West. Anal. C&n. Actor. 51 (1970) 355. 20 J. F. Alder and T. S. West, Atd. Chim Acm. 51 (1970) 365. 21 J. Aggctt and T. S. West, AIJN/. Chim. Acru. 55 (1971) 349. 22 L. Ebdon. G. F. Kirkbright and T. S. West, Aml. Clrirn. Acra. 58 (1972) 39. 23 N. F. Turkalov, R. L. Mngunov and A. V. Zagorodnyuk. Ukr. Khint. Z/r.. 38 (1972) 215. 24 V. I. Davydov, B. V. Tcplyakov and G. K. Romanov. Z/I. Prikl. K/h.. 35 (1962) 1625. 25 L. Barton and C. A. Hcil, J. Less-Cornmor~ Mcr.. 22 (1970) Il.
87
1733.