evaporated until fumes of H2SO4 were visible. The solutions were diluted to 60 g with water, 1 mL of 10 percent K2CO3 was added to each, and the solutions were digested for 11⁄2 h. Glass stirring rods were placed in each beaker and the thallium was precipitated by adding 2 g of concentrated. NH,OH, followed by the dropwise addition of 1 g of 10 percent K2CrO4 to each with constant stirring.

The solutions were allowed to stand at room temperature for approximately 18 h. Each solution was then filtered through a tared 15-mL fine fritted glass crucible. The filtrate containing the soluble thallium was collected in a 100mL Teflon beaker. After all the solution had been filtered, the Tl2CrO4 precipitate was washed three times with approximately 30-mL of 50 percent (v/v) ethanol-water mixture. The precipitate was dried for 2 h. at 125 °C and reweighed. Further drying at 125 °C yielded no change in weight of the Tl2CrO4 precipitate.

The crucibles were weighed to ± 0.002 mg on a microbalance. To eliminate any errors due to static charge, the crucibles and tares were reweighed cyclically until the reproducibility was within ± 0.005 mg. A buoyancy correction for the glass crucibles was made by averaging the change in weight of two empty tare crucibles. The air weight of the Tl2CrO1 was converted to vacuum weight using a measured value of 6.983 as the density of the precipitate at 22 °C. The millimoles of thallium present in the precipitate were determined using the calculated atomic weight for thallium and the 1975 atomic weight values for chromium and oxygen. The formula weights used were for 203T12CrO4 and for 205 Tl2CrO4.

After filtration of Th2CrO4 was complete, the soluble portion and washings were returned to the original 400 mL beaker and evaporated to a volume of approximately 10 mL. The solutions were made acidic with concentrated HNO3 (color change from yellow to orange) and a small amount of ethanol was added to reduce Cr* to Cr3. The solutions were transferred to weighed polyethylene bottles, diluted to 80 to 100 g and aliquoted. The aliquots were spiked by weight with 203T1, and the resulting solutions were evaporated to dryness. One gram of aqua regia was added to oxidize the thallium and, after evaporation, the residues were taken up in IN HBr. The thallium, as HT1C14, was extracted into methyl isobutyl ketone (MIBK) and evaporated

This method of determining the concentration of thallium solutions was previously tested on solutions containing a known amount of "natural" thallium. A thallium master solution was prepared from high purity (99.99%) thallium metal (SRM 997) and seven sets of four samples were withdrawn from this master solution, each on a different day over a period of one month. In addition, one more set of four was determined just before the assay work was begun on the separated isotope solutions. This extra set allowed the analyst to be certain that the experimental conditions were still under control. The final set which was completed 11 months after the first set was assayed, showed no evidence of any bias. The uncertainty (ts) of 31 individual determinations is 0.029 percent and the ts of the set averages is 0.014 percent. Comparison of the calculated and measured concentrations indicated a positive bias of 0.028 percent which would have a negligible effect on the ratios.

Pooling the results of the analysis of the separated isotope solutions shown in table 5 with the results of the eight sets described above, yields a value of ± 0.0000030 mmol Tl/g solution for the standard deviation of an individual determination (7 deg of freedom). The standard error of the average of four determinations is ± 0.0000015 mmol Tl/g solution.

'Student T test at a 95 percent confidence limit.

Most of the chemical determinations of the atomic weight thallium were performed prior to 1934. In all cases, the atomic weight was determined by ratioing the weight thallium or one of its compounds to an equivalent weight another element or compound. The most common metho involved the conversion of a known amount of thalliu chloride to silver chloride. The weight of thallium chloric was then ratioed to the weight of silver consumed, and the atomic weight of thallium was calculated using the accepte atomic weights of silver and chlorine.

An example of one of the more accurate chemical atom weight determinations is that of Baxter and Thomas [16]. 1 this experiment, thallous sulfate was recrystallized severa times and converted to the chloride. The chloride was the recrystallized several times and prepared for weighing distillation in nitrogen followed by refusion in nitrogen. Thị purified thallous chloride was weighed and dissolved in hu water. After dissolution was complete, a nearly equivalen amount of pure silver was added to precipitate the fre chloride. The end point was determined nephelometrical through the addition of hundredth normal solutions of si ver and chloride.

A history of the chemically determined atomic weight thallium is given in table 1. The only determination after, 1934 was that of Rodriques and Magdelena [26] in 196 This group employed high precision density determinations" of thallous chloride as the basis of their atomic weigh determination.

Table 2 lists the atomic weight determinations based o mass spectrometric measurements of the relative isotop abundance of natural thallium. A search of the literatur yielded only five published isotope abundance measur! ments. The measurement made by White and Cameron [3] is still listed as the best measurement from a single natur source by the International Commission on Atomic Weights [27]. The present work marks the first time that calibrate. mass spectrometry has been used for a determination of th atomic weight of thallium.

Kothe [23]

The accepted atomic weight of thallium was 204.39 from 1925 until 1962. In 1962, when the 12C scale was adopted, the chemical combining weight ratios were recalculated as part of a general review of atomic weight data for all elements. The accepted atomic weight value of thallium has remained at 204.37 ± 0.03 since 1962. At this level of uncertainty the atomic weight becomes a limiting factor in high accuracy assay analyses.

2. Experimental Procedure

2.1 Mass Spectrometry

The isotope ratio measurements were made on two solid sample mass spectrometers. Both were nearly identical single stage 90°, 30 cm radius of curvature instruments equipped with a "Z" lens focusing source [33]. The collector was a deep bucket Faraday cup type equipped with a 50 percent transmission grid shadowing a series of suppression grids [33, 34, 35]. The measuring circuit consisted of two vibrating reed electrometers (VRE), a voltage to frequency converter, and a scaler-timer. Data acquisition was made by computer control. Prior to initiating the atomic weight ratio determinations, the digital measurement circuits of the mass spectrometers were calibrated and were found to be linear to within one part in 10 over a range of 20-100 percent of full scale for each VRE scale. Nonlinearities in the VRE and/or voltage to frequency converter can result in significant systematic biases in the correction factors unless the mixes closely bracket the isotopic ratio of the standard, and the signal intensities of the corresponding isotopes of the standard and mixes are measured at nearly the same point on the VRE scale. The linearities of the measurement electronics used in this work, combined with close matching of the isotopic ratio of the mixes to the standard, reduced the systematic biases introduced by the VRE and VF converter to less than a part in 105. Measurement circuits have been examined which exhibit nonlinear response approaching a part in 103. The use of such measurement systems combined with calibration mixes which differ in isotopic ratio significantly from the standard can result in errors in the correction factors which are larger than the precision of the ratio measurements.

The mass spectrometric procedure used in the determination of the thallium isotopic ratios employed a single filament tungsten ion source. The method was initially chosen based on the success of Gramlich and Machlan [36] in using a single filament tungsten approach for gallium, a member of the same periodic family. In addition, Huey et al. [37] had reported that the use of rhenium, a commonly used filament material for the analysis of thallium, opened up the possibility of interference from ReO* peaks at masses 201 to 205 with the thallium masses at 203 and 205. The potential

interference of ReO was examined using a bare rhenium filament ribbon at high temperatures (up to 2200 °C). Although no masses were found in the region from 201 to 205 at an ion current sensitivity of 2 × 10-A, the use of rhenium as a filament material was avoided since the potential for interference did exist.

To obtain highly precise ratios the fabrication and cleaning of the tungsten filaments had to be carefully controlled. The tungsten ribbon (0.025 × 0.76 mm) had to be mounted on the filament posts such that a nearly perfect square flat top filament surface was obtained. Filaments with either convex or concave surfaces affected the drying of the thallium on the filament and, thus the precision of the ratio measurement. The filament surface was cleaned by degassing at 3.0A for 12 h under a vacuum and in a potential field. The degassing parameters of current and time were especially critical to the precision of the thallium ratio measurements. If the degassing process was carried out at either higher currents or for longer periods of time, the 205T1/203T1 ratio could be shifted as much as 0.15 percent.

A 10 μL drop containing 1 μg of thallium as TICl, in (1 + 9, V/V) HNO3 was loaded onto a tungsten filament ribbon. The thallium solutions were stored in (1 oz) 30 mL screw cap Teflon' PFA beakers. This container material was chosen after considerable testing indicated that prolonged storage in either polyethylene or glass containers yielded inconsistancies in the analytical procedure.

The sample mounting procedure was performed in two stages, hereafter referred to as the low temperature and the high temperature drying phases. The low temperature phase was carried out on a Class 100 clean air bench using a programmable sample dryer designed by Gramlich and Shideler [38] which permitted this phase of the drying to be accomplished with a high level of reproducibility. The low temperature drying utilized currents of 1A and 3A and an infrared heat lamp to dry the sample. The intensity of the lamp was controlled to yield a temperature of 50 °C at the filament surface. Failure to control this parameter could cause a shift in the measured thallium isotopic ratio by as much as 0.05 percent. The sample size was controlled by monitoring the length of time required for the drop to dry on the filament surface since a drop which was either too large or too small would affect the precision of the ratio

measurement.

Another parameter which can greatly affect the precision of the ratio measurement is the air flow of the clean air bench where the low temperature drying was performed. A reduction in the air flow from 30 m/s to 12 m/s, which could

2 Certain trade names and company products are identified in order to adequately specify the experimental procedure. In no case does such identification imply recommendation or endorsement by the National Bureau of Standards, nor does it imply that the products are necessarily the best available for the purpose.