The Absolute Isotopic Abundance and Atomic Weight of a Reference Sample of Silver

Abstract

The atomic weight of a reference sample of silver has been determined by mass speetrometry, with an uncertainty of one part in 10⁶, using a single filament silica gel procedure. Accurately known quantities of chemically pure ¹⁰⁷Ag and ^l09Ag were mixed to produce standards of known isotopic composition for calibration of the mass spectrometer. The absolute isotopic ratio of the reference sample of silver is ¹⁰⁷Ag/¹⁰⁹Ag = 1.07638 ± 0.00022 yielding an atomic weight of 107.86815 ± 0.00011. The indicated uncertainties represent an overall limit of error at the 95 percent confidence level which is the sum of the uncertainty components for the ratio determined and the components covering effects of known sources of possible systematic error.

1. Introduction

The inorganic mass spectrometry group of the National Bureau of Standards has been conducting a long term program of absolute isotopic abundance ratio and atomic weight determinations using high precision isotope ratio mass spectrometry. Previous atomic weight determinations include silver [1],¹ chlorine [2], copper [3], bromine [4], chromium [5], magnesium [6], lead [7], boron [8], rubidium [9], rhenium [10], silicon [11], potassium [12], thallium [13], and strontium [14].

The present work, a redetermination of the atomic weight of silver, was undertaken in conjunction with the calculation of a more accurate Faraday constant. The Faraday is directly related to other physical constants including the Avogadro constant, the proton gyromagnetic ratio, the magnetic moment of the proton in uuelear magnetons, and the ratio of the NBS as-maintained ampere to the absolute or SI ampere. Over the years a problem arose iu assigning a recommended value for the Faraday due to apparent discrepancies between the Faraday determined from electrochemical experiments and the Faraday calculated from other fundamental constants. Unfortunately, earlier electrochemical experiments were not sufficiently precise (6.8 ppm, one standard deviation, for the best silver determination) to either prove or disprove the existence of this discrepancy. As a result, the Faraday was excluded from the most recent (1973) least squares adjustment of the fundamental constants [28].

In 1980, Bower and Davis [15], using the same source of silver as was analyzed in this work, NBS Standard Reference Material (SRM) 748, published a new value for the electrochemical equivalent of silver with an accuracy of 1.28 ppm (one standard deviation). The calculation of the Faraday using this value and the 1962 atomic weight of silver [1], indicated that significant differences between the various Faraday calculations might indeed exist. In order to assess the significance of these differences, however, it became necessary to significantly reduce the uncertainty in the atomic weight of silver.

The method used for the determination of atomic weights at NBS, which has been briefly described in previous publications [1–14], and may be described as calibrated mass spectrometry, combines the techniques of high precision chemical assay with high precision mass spectrometry. The mass spectrometers to be used for the isotopic abundance measurements are calibrated for bias using synthetic mixes of known isotopic composition, prepared from nearly pure separated isotopes. Extensive research has demonstrated that this bias is due to non-linearities in the measurement circuit and mass-dependent isotopic fractionation [16]. In the case of ratios near one, this bias reduces to isotopic fractionation. These measured biases are then used to calculate the absolute isotopic abundance ratio and, ultimately, the atomic weight of the reference sample. A block diagram characterizing the atomic weight method is shown in figure 1.

Figure 1.

Mass Spectrometric Atomic Weight Method

The chemical research requires the development of an assay procedure for the pure element or reference sample precise precise to at least one part in 10⁴, which will ultimately be used to assay the separated isotope solutions. In addition, a procedure must be developed to purify the separated isotopes to eliminate impurities which could interfere with either the chemical or mass spectrometric procedures. Once these methods have been thoroughly tested using the reference sample, the separated isotopes are purified and dissolved. Aliquots of each isotope are then taken to produce calibration mixes which are blended to braekel the natural isotopic abundance ratio of the element. In addition, aliquots of each isotope are taken for assay so that accurate solution concentrations can be ascertained. A knowledge of the solution concentrations combined with aliquoting data allows the actual ratios of the calibration mixes to be calculated.

The mass spectrometric research involves the development of a high precision method for the analysis of the relative isotopic abundance ratios of the reference sample, followed by a comprehensive study to identify and control any sources of bias which conld affect the final measurements.

After the above tasks have been accomplished, two complete sets of analyses of the calibration mixes and the SRM are made by two different operators using different mass spectrometers. The calibration mixes are randomly selected and are analyzed in an alternating pattern with the reference sample. A comparison of the relative isotopic abundance ratios obtained for the calibration mixes with the calculated or true value produces a calibration factor. This factor is applied to the relative isotopic ratio obtained for the SRM to produce the absolute isotopic abundance ratio. The atomic weight is then obtained by multiplying the fractional abundance of each isotope by its nuclidic mass² and summing the resultant products.

2. Experimental Procedure³

2.1. Mass Spectrometry

The isotopic ratio measurements were made using two nearly identical 90°, 30 cm radius of curvature solid sample mass spectrometers equipped with a “Z” lens focusing source [18]. The collector was a deep bucket Faraday cup equipped with a 50 percent transmission grid shadowing a series of suppression grids [18–20]. The measuring eireuit consisted of two vibrating reed electrometers (VRE) and a voltage-to-frequency conversion system which transmits data to a computer for data acquisition. Prior to initiating the atomic weight ratio determinations, the measuring circuits of the mass spectrometers were calibrated and found to be linear to within one part in 10⁴ over a range of 30–80 percent of full scale for each VRE scale. However, the linearity of the measurement electronics becomes insignificant because of the equal atom composition of the standard and most of the mixes. Under these conditions any non-linearities in the ratio measurement will cancel, and any possible effects would be indicated by the endpoint calibration mixes.

The relative isotopic composition of the calibration mixes and the reference sample were determined by surface ionization mass spectrometry using a platinum single filament silica gel technique.

The filaments used in this work were fabricated from 99.9 percent pure platinum ribbon (0.025 × 0.76 mm) and were cleaned by heating at 2.5 A for 1 h(r) in a vacuum and under a potential field.

The sample mounting was carried out in two stages, hereafter referred to as the low temperature drying stage and the high temperature drying stage. The low temperature drying stage was similar to the lead-silica gel method described by Barnes et al., [21]. A 5 μL drop of silica gel suspension was placed on the filament surface and dried at 1.0 A for 5 min. A second drop of silica gel was added and the drying repeated. A 5 μL drop of silver as AgNO₃ in (1 + 9)⁴ HNO₃ was placed on top of the silica gel and dried at 1.0 A for 5 min. A drop of 0.75 N H₃PO₄ was then added and dried at 1.5 A for 5 min then at 2.0 A for 5 min. This drying procedure was carried out in a class 100 clean air hood with an airflow of 100 linear ft/s per s using a programmable sample drier designed by Gramlich and Shideler [22]. Throughout the procedure a heat lamp, whose intensity was adjusted to yield a temperature of 70 °C at the filament surface, was used to aid the drying process.

The high temperature drying stage was accomplished using pyrometer adjustment of the filament temperature. This method was first used for the determination of the atomic weight of thallium [13] and has since proved to be an important key to controlling the reproducibility of isotopic ratios of a number of elements. The filament was covered with a bell jar and purged with nitrogen gas for 5 min. The nitrogen flow was stopped and the filament was adjusted to a temperature of 1040 °C for 60 s.

After loading the sample into the source of the mass spectrometer, the system was allowed to pump down to a pressure of at least 2 × 10⁻⁶ torr (300 μPa) before starting the analysis. Liquid nitrogen was then added to the source cold finger to reduce the pressure to less than 1 × 10⁻⁷ torr (10 μPa). At t = 0 min, the temperature of the filament was increased to 760 °C. This yielded a ¹⁰⁷Ag ion beam intensity 1 − 2 × 10⁻¹² A. At t = 25 min, the temperature of the filament was increased to 790 °C. At this temperature the intensity of the ¹⁰⁷Ag ion beam was between 5 × 10 ⁻¹² A and 1 × 10⁻¹¹ A. If the intensity of the ¹⁰⁷Ag ion beam was outside of these limits, the analysis was rejected. Baseliue data were taken and ratio data were collected between t = 30 and 50 min.

2.2. Purification of Separated Silver Samples

Electromagnetically separated ¹⁰⁷Ag and ¹⁰⁹Ag isotopes in the form of silver metal powder were obtained from the Nuclear Division, Oak Ridge National Laboratory of the Union Carbide Nuclear Company. The ¹⁰⁷Ag was designated series R&D, sample 000101 and the ¹⁰⁹Ag was designated series R&D, sample 000201. The certificate which accompanied each sample showed enrichment to better than 99.9 percent for the major isotope. No information on chemical purity was given.

Since the method developed for the assay of silver was based on the precipitation of AgI from ammonical solution, it was necessary to develop a purification procedure that would reduce possible impurities to a level low enough so they would not cause a significant error.

Elements that form relatively insoluble iodides or hydroxides could possibly interfere. Included are Pb, and Tl which form insoluble iodides, and Al, Cr, Fe, Ga, In, Ti, Zr, and the rare earths which form insoluble hydroxides.

Each separated silver isotope was purified as follows: The silver isotope, as powdered silver metal (about 1g) was transferred to a 150 mL Teflon-FEP beaker, dissolved in 20 mL of (1 + 4) HNO₃, and the resulting solution evaporated to dryness. The residue was taken up in 50 mL of water and enough concentrated NH₄OH solution to dissolve the precipitated AgOH. The solution was digested on a warm hot plate for about one hour, allowed to cool, and filtered through a close textured paper. After washing the filter paper with dilute (1 + 49) NH₄OH the filtrate was heated to drive most of the NH₃ from the solution. The solution was made acidie with dilute (1 + 9) HNO₃ and 25 mL of (1 + 9) HCl was added to precipitate silver chloride. The solution and precipitate were digested on a hot plate for three hours in darkness and allowed to stand overnight at room temperature, also in darkness. Most of the supernatant slution was removed from the precipitated silver chloride by decantation and the last 10–15 mL was withdrawn by means of a plastic syringe equipped with a platinum needle. The precipitated AgCl was washed three times with a few mL of H₂O and the washings were removed with the syringe and needle.

The precipitated AgCl was then dissolved in 15 mL of NH₄OH and diluted with about 50 mL of water. The solution was heated to drive off excess NH₃ and made acid with (1 + 9) HNO₃ to repreeipitate AgCl. A few drops of (1 + 9) HCl were added and the solution was allowed to stand overnight in darkness. The supernatant liquid was then removed in the manner previously described. The precipitated AgCl was then dissolved in 20 mL of NH₄OH and transferred to a 350 mL platinum dish. A solution prepared by reacting 25 mL of (1 + 4) HNO₃ with 50 mL of (1 +4) NH₄OH yielding NH₄NO₃, was added as an electrolyte and the resulting solution was diluted to approximately 250 mL. The platinum dish was placed on aluminum foil which was connected to the cathode of a de-power source. The dish was covered with a Teflon-TFE cover fitted with a platinum rod through the middle which reached to within 1 cm of the dish bottom. This rod was connected to the anode of the dc-power source. Silver was then elcctrodeposited onto the dish by plating with a potential of 1.0 V for 24 h.

The electrolyte was then poured from the dish into a Teflon beaker and the electrodeposited silver in the dish was washed with dilute (1 + 9) NH₄OH.

The electrodeposited silver was dissolved with (1 + 4) HNO₃ diluted to about 150 mL and the solution was made ammonical with concentrated NH₄OH. Additional NH₄NO₃ electrolyte was added and the silver was again electrodeposited by plating at 1.0 V for 24 h.

The electrolyte was then poured into the Teflon beaker, and the silver was washed with water. The dish and contents were then dried and weighed. The electrodeposited silver was then dissolved with (1 + 4) HNO₃ and transferred to a Teflon beaker. The dish was washed with water, dried, and weighed to obtain the approximate weight of purified silver.

Calculations based on the starting weight of silver, and the weight of purified silver showed that more than 99 percent of the silver was recovered in each case.

The effectiveness of this purification procedure was first tested by purifying a solution of natural silver which had been doped at the 1000 μg/g level with each of thirty common impurity elements. The results of the analysis of this purified silver are shown in table 1. All of the impurity elements were reduced to 10 μg/g or less (the analytical method will also be described later).

Table 1.

Analysis of impurities in silver separated isotopes.

Element	Spike Isotope	Ag Dopeda and Purified (μg/g)	107Ag (μg/g)	109Ag (μg/g)
Al	26Mg	4	<14	<20
As	82Se	0.3	—b	—
Ba	137Ba	0.1	3.1	0.9
Bi	203Ti	0.05	—	—
Ca	44Ca	2.6	1.5	1
Cd	111Cd	0.4	0.3	1.2
Co	62Ni	0.5	—	—
Cr	53Cr	0.2	0.2	0.1
Cu	65Cu	1.2	2.2	0.7
Fe	54Fe	5	6	6
Ga	71Ga	0.2	<1.1	<1,6
K	41K	2.7	5	1.2
Mg	26Mg	1.8	0.8	1.2
Mn	54Fe	0.03	—	—
Mo	97Mo	0.01	0.06	0.06
Na	41K	9	<50	<60
Nd	145Nd	0.2	0.6	0.4
Ni	62Ni	0.4	2	0.4
Pb	206Pb	1.6	1.5	1.6
Pd	110Pd	0.2	0.8	0.2
Sb	117Sn	0.6	—	—
Se	82Se	0.2	0.3	0.4
Sn	117Sn	0.07	0.3	0.4
Sr	86Sr	0.03	0.1	0.1
Te	125Te	0.002	0.1	0.07
Tl	203Tl	0.1	0.5	0.5
V	53Cr	0.03	—	—
Zn	67Zn	1	0.3	0.6
Zr	91Zr	0.06	0.5	0.6

^aOriginal concentration—1000 μg/g each.

^b— denotes data not reported.

Since the method was effective in purifying grossly contaminated silver, it was applied to each of the silver separated isotopes, which were of much higher initial purity.

The acids and water used in these purifications were produced at NBS by sub-boiling distillation [23] and have been shown to be extremely low in trace metal contamination. The NH₄OH was produced by saturating high purity cold water with high-purity NH₃ gas and allowing the resulting solution to warm to room temperature. Apparatus such as beakers and filters were cleaned with ACS Reagent Grade acids and rinsed with high purity water before use.

2.3. Preparation and Analysis of Separated Isotope Solutions

Each isotope (¹⁰⁷Ag and ¹⁰⁹Ag) was transferred to a specially constructed and tared 500 mL quartz flask. These flasks were constructed from a standard taper joint 500 mL quartz flask by cutting the neck of the flask about 1 cm from the body and tooling the neck to fit a number zero polyethylene stopper and a standard aluminum serum cap. Final solutions were diluted to about 350 mL and final acidity was about 0.5 N HNO₃. The solution of ¹⁰⁷Ag was labeled “Ag-107” and the solution of ¹⁰⁹Ag was labeled “Ag-109”.

Portions of each separated isotope solution, equivalent to about 40 mg of silver, were taken for determination of impurities. Each sample was spiked with 10⁻⁷ g of ⁴⁴Ca, ¹¹¹Cd, ⁵³Cr, ⁶⁵Cu, ⁵⁴Fe, ⁷¹Ga, ²⁶Mg, ⁹⁷Mo, ¹⁴⁵Nd, ⁶²Ni, ²⁰⁶Pb, ¹¹⁰Pd, ⁸²Se, ¹¹⁷Sn, ⁸⁶Sr, ¹²⁵Te, ²⁰³TI, ⁶⁷Zn, and ⁹¹Zr. Each spiked solution was then treated as follows: the solution was diluted to about 40 mL with H₂O and a slight excess of HCl was added to precipitate AgCl. The beaker and contents were allowed to stand in the dark overnight. Most of the supernatant liquid was removed from the precipitated AgCl by decantation into a second Teflon beaker. The last approximately 10 mL was removed by means of a plastic syringe and platinum needle and added to the decanted solution. The precipitate was washed with a few mL of (1 + 9) HNO₃ which was also combined with the decanted solution by use of the syringe. This solution was then evaporated to a few mL, removed from any remaining AgCl with the syringe and transferred to another beaker. This solution was evaporated to a large drop and transferred to specially prepared gold wire electrodes for analysis by spark source mass spectrometry using a peak switching electronic detection system.

The results of the analysis of the purified separated silver isotopes are shown in table 1. In addition to the spiked elements, the concentrations of other elements were estimated by comparison to spiked nuclides using the assumption that the sensitivity of the natural element and spike nuclides are the same. The results show that most elements are at the low ppm level or less. The only elements found at concentrations of greater than 10 ppm were Al and Na. These contaminants may have resulted from the Pyrex glass flask used to store the spike isotopes. Even if they were present in the amounts indicated, the effect would be negligible since sodium would not interfere with the analysis at this level, and aluminum, as Al(OH)₃, would cause an error of only 0.001 percent in the assay of each isotope.

2.4. Assay of the Separated Isotope Solutions

The method developed for the high-precision assay of the silver solutions was based on a combination of gravimetry for the determination of most of the silver (greater than 99.5 percent) as AgI and isotope dilution mass spectrometry (IDMS) for the remainder. Silver iodide was selected as the gravimetric form since, of all the silver halides it is the least light sensitive, least soluble, and has the highest gravimetric factor. In fact, it was found that dry stoichiometric silver iodide was not detectably sensitive to fluorescent lighting since the material remained lemon yellow and no change in weight could be detected on prolonged exposure. The material was also found to be non-hygroscopic up to relative humidities of 90 percent.

The “Ag-107” solution and the “Ag-109” solution were each sampled for assay of silver in the following manner: four weighed portions of about 35 g, containing about 0.25 mmol Ag were withdrawn from each flask. A 10 cm platinum needle which had been inserted through a No. 0 polyethylene stopper was used to replace the stopper on the quartz flask, A 20 mL polyethylene hypodermic syringe, the plunger of which had been covered with a sheet of Teflon-FEP, was attached to the Kel-F hub of the needle and the desired amount of solution was withdrawn. The syringe was then disconnected from the hub and the tip was capped with a Kel-F cap. The syringe and contents were then weighed on a semi-micro balance to ±0.02 mg. (Any static charge on the syringe was dissipated by placing the syringe on the balance pan which was surrounded by several polonium anti-static sources and waiting 2 min before taking a weight.) The solution was then delivered to a 100 mL Pyrex beaker with a Teflon coated rim and the syringe was again capped and weighed. The weight of the sample taken was determined from the weight of the syringe before and after delivery of the sample. A second weighed portion was taken in the same manner and combined with the first portion for each assay sample. Two assay samples were withdrawn from each solution before and after withdrawing the calibration samples to ensure that no change in concentration had occurred during the time interval of withdrawing samples (about 6h).

Each sample was then assayed as follows: 20 mL of concentrated high-purity NH₄OH was added and approximately a 1 percent excess of 0.5 N NH₄I was added slowly from a small wash bottle. The solution was heated covered on a hot plate.

Precipitation from ammonical solution resulted in the formation of a white precipitate, AgI·l/2 NH₃ [24], Heating the solution drives off ammonia and results in the conversion of this white compound into yellow AgI. The AgI formed in this manner is composed of relatively large, easily filterable crystals that do not pass into the colloidal state when washed with water. The fact that AgI is slightly soluble in dilute NH₄OH also contributes to the formation of larger crystals.

After digesting overnight, the cover was removed and the solution was evaporated to a volume of 20 mL. Removal of NH₃ was tested by holding moist p-Hydrion paper over the hot solutions. If the paper indicated NH₃ in the vapor the solution was heated until NH₃ was no longer detected. The beaker was then removed from the hot plate, covered, and allowed to stand overnight.

The solution was then filtered through a tared platinum-Munroe crucible and the precipitated AgI was eaught on the platinum mat. The beaker and precipitate were washed with several small increments of water. The filtrate and washing were transferred back to the original beaker and reserved for the determination of dissolved and untransferred silver. The erucible and contents were dried for 3 h at 200 °C. Pure AgI is only very slightly reduced by light when wet. To guard against any possible decomposition, the precipitation, filtration, and drying were carried out in orange light or darkness. (The pure compound is lemon yellow and even slight decomposition causes the AgI to darken.)

After heating, the crucible was cooled in a desiceator and transferred to the case of a micro-balance and allowed to stand for at least 2 h. The erucible and contents were then weighed on a micro-balance to ±0.002 mg An effective buoyancy and adsorption correction for the platinum crucible was made by averaging three empty tare crucibles. (The drying, cooling, and weighing were repeated to ensure constant weight.) The apparent mass of the AgI was then determined and converted to true mass using a 5.66 g·cm⁻³ for the density of ¹⁰⁷AgI and 5.71 g·cm⁻³ for the density of ¹⁰⁹AgI. These densities were calculated by assuming that they are proportional to the density of natural αAgI, D³⁰ = 5.68 g·cm⁻³, in the same ratio as their molecular weights. The vacuum weight of the silver iodide was converted to millimoles of silver using the calculated atomic weight for silver and 120.9045 for iodine. The formula weights used were 233.8909 for ¹⁰⁷AgI and 235.8087 for ¹⁰⁹AgI.

To determine the soluble and untransferred silver, the filtrate and washings, which had been reserved in the original beaker, were spiked with 1 to 2 μmol of either ¹⁰⁷Ag or ¹⁰⁹Ag. Five mL of ammonium hydroxide solution and 0.25 g of potassium cyanide (ACS Reagent Grade) were added to dissolve silver iodide and the solution was well mixed by stirring with a magnetic stirring bar for a few hours. Silver was then separated electrolytically by plating overnight with a potential of 2.6 V onto a platinum wire anode. Silver was then dissolved from the electrode with a few mL of (1 + 1) HNO₃ and the resulting solution was evaporated to dyrness. The residue was taken up in (1 + 49) HNO₃ and the 107/109 ratio was determined by surface ionization mass spectrometry. The amount of silver as μmol of Ag was then calculated and added to the silver from the gravimetric determination to yield the total silver in the sample.

This method of assaying silver solutions was first tested on solutions containing known amounts of silver. The material was SRM 748 which is 99.999 + percent silver. The solutions were prepared in the approximate concentration range of 0.10 to 0.16 mmol AgI. Four portions containing from 0.9 to 1.1 mmol Ag were withdrawn from each solution and silver was determined as described above. Nine sets of four samples were analyzed in this manner.

Statistical analysis of the data from these nine sets showed that the standard error of the mean of four values was 5.2 × 10⁻⁶ mmol/g and that the coefficient of variation (standard error/mean) for a set of four was 0.003 percent. Comparison of calculated to measured concentrations showed a small positive bias of about 0.005 percent, but this would have a negligible effect on the ratio of two assays.

The results of the assay of the separated isotope solutions are shown in table 2. Pooling the results of the assay of the separated isotope solutions and the nine sets of natural solutions, yields a value of 0.0000009 mmol Ag/g for the standard error of the average concentration of each separated isotope solution. The uncertainty associated with this value at the 95 percent confidence level is 1.8 × 10⁻⁶ mmol Ag/g.

Table 2.

Concentration of silver separated isotope solution

Solution	Sample No.	Weight Ag I (g)	Ag from Ag I (mmol)	Ag from Filtrate (mmol)	Total Ag (mmol)	Weight Sample (g)	Concentration (mmol Ag/g)
“Ag 107”	1	0.223445	0.955668	0.002583	0.958251	34.00905	0.0281760
	2	0.231677	0.990876	0.002358	0.993234	35.25092	0.0281761
	3	0.234055	1.001047	0.001176	1.002223	35.56893	0.0281769
	4	0.228265	0.976283	0.001211	0.977493	34.69020	0.0281778
						Average	0.0281767
“Ag 109”	1	0.229478	0.973153	0.003956	0.977109	35.78498	0.0273050
	2	0.228795	0.970257	0.002316	0.972573	35.62279	0.0273020
	3	0.226974	0.962534	0.001029	0.963563	35.28974	0.0273044
	4	0.231376	0.981202	0.001614	0.982814	35.99738	0.0273024
						Average	0.0273034

2.5. Isotopic Analyses of the Separated Isotope Solutions

Each of the separated isotope solutions was analyzed eight times on each of two mass spectrometers, MS #1 and MS #5, by Operators 1 and 2, respectively. The mass spectrometer sources were cleaned between the analyses of the ¹⁰⁷Ag and ¹⁰⁹Ag as a precaution against the possibility of contamination from source parts, although back-to-back analyses of the two separated isotopes on the same source failed to yield any evidence of contamination. The corrected isotopie compositions of the two isotopes are shown in table 3.

Table 3.

Isotopic composition of the silver separated isotopes

Separated Isotope		Isotopic Composition (atom percent)
“Ag 107”	107Ag	99.97136 ± 0.00206a
	109Ag	0.02863 ± 0.00206
“Ag 109”	107Ag	0.02865 ± 0.00203a
	109Ag	99.97135 ± 0.00203

^aThe uncertainly of the ratio determination is taken to be 7 percent, which is much larger than the 95 percent confidence limit, to take into account possible biases and non-linear instrumental behavior for ratios as large as these.

The measured imprecision of the ratio measurements for each of the separated isotope solutions was less than 3 percent (2 standard deviations), however an uncertainty of 7 percent was placed on the ratios for the separated isotopes to cover possible measurement system nonlinearities and isobaric interferences. Several possible species have a potential for producing isobaric interferences in the silver mass region. These include ${\text{CaPO}}_2^{+}$, ${\text{SiPO}}_3^{+}$, and AsS ⁺ . Any interferences from ${\text{CaPO}}_2^{+}$ can be detected by monitoring mass 103 which results from the major isotope of calcium (⁴⁰Ca : 96.9 percent abundant). However, the major isotopes of silicon and sulfur would produce ${\text{SiPO}}_3^{+}$ and AsS⁺ interferences at mass 107. Thus extremely small quantities of either of these species could not be detected by examining other masses in the silver spectral region. A ${\text{SiPO}}_3^{+}$ species has not been reported or observed in a thermal ionization-silica gel analysis but is included herein because of the nearly infinite supply of these elements on the filament, and the fact that an intense and sustained ${\text{PO}}_3^{-}$ ion beam is observed in the negative ion spectra under the analytical conditions reported in this paper.

Kelly and Wasserburg [25] reported the existence of AsS ⁺ which, if present, would enhance the measured 107/109 ratio. The measurement of the ¹⁰⁹Ag separated isotope is particularly sensitive to this interference because AsS⁺ would produce a 107/109 ratio of 22. Efforts to generate as AsS⁺ ion beam by loading As₂S₃ and H₂SO₄ onto the filament using the procedure described in section 2 were unsuccessful. Since the silver was loaded in (1 + 9) HNO₃, it is felt that the nitric acid allowed oxidation of the sulfur to sulfate, thus reducing any AsS⁺ to an undetectable level.

Analysis under identical conditions of ¹⁰⁹Ag samples ranging in amount from 4 μg to 0.005 μg indicated a possible mass-107 isobaric interference at a level of 1 (± 1) × 10⁻¹⁶ A. This level of interference is below the direct measurement detection limit of the instrument but can be inferred from the slight decrease in the observed 109/107 ratio with decreasing sample size. This suspected interference would produce a bias of approximately 2 percent to the observed ratios of the ¹⁰⁹Ag separated isotope and a bias of approximately 0.002 percent to the observed ratios of the reference sample and the calibration mixes.

2.6. Preparation of Calibration Samples

Eight calibration mixes were prepared by blending weighed portions of the “Ag-107” and the “Ag-109” solutions to produce 107/109 ratios ranging from 0.5 to 2.0. Six of the calibration mixes were within a few percent of the natural ratio 107/109 of 1.076.

The portions for mixing were withdrawn from the flasks and weighed in the manner previously described for the assay of each solution. Each portion wieghed about 10 g and was weighed to ±0.05 mg. It is therefore estimated that the weighing error for each mix should not exceed one part in 10⁵. To minimize any significant possibility of change in concentration of the isotope solutions with time, the portions for the calibration mixes were withdrawn from the flasks between the samples taken far assay, over a period of about 6 h.

Each calibration mix was thoroughly mixed, the sides of the beaker were washed with water, and the solution was evaporated to dryness. The residue was taken up with dilute (1 + 9) HNO₃ and the mixing, washing, and evaporation were repeated. The calibration mixes were then taken up in (1 + 9) HNO₃ to a concentration of 0.8 mg Ag/mL and transferred to small Teflon bottles. The isotopic compositions of the calibration mixes are given in table 4.

Table 4.

Isotopic composition of calibration mixes

Solution No.	Isotope Solution	Weight Solution (g)	Ag from Solution (mmol)	107Ag from Solution (mmol)	109Ag from Solution (mmol)	Total 107Ag (mmol)	Total 109Ag (mmol)	Ratio 107/109
1	“Ag 107”	10.02135	0.292369	0.282288	0.000081	0.282364	0.263439	0.07184
	“Ag 109”	9.64836	0.263433	0.000076	0.263358
2	“Ag 107”	8.95832	0.252416	0.252344	0.000072	0.252412	0.237226	1.06402
	“Ag 109”	8.68835	0.237222	0.000068	0.237154
3	“Ag 107”	9.24458	0.260482	0.260407	0.000075	0.260477	0.243049	1.07171
	“Ag 109”	8.90162	0.243044	0.000070	0.242974
4	“Ag 107”	9.75636	0.274902	0.274823	0.000079	0.274897	0.257233	0.06867
	“Ag 109”	9.42110	0.257228	0.000074	0.257154
5	“Ag 107”	9.59357	0.270315	0.270237	0.000078	0.270309	0.248906	1.08599
	“Ag 109”	9.11603	0.248899	0.000071	0.248828
6	“Ag 107”	10.11275	0.284944	0.284862	0.000082	0.284940	0.269592	1.05693
	“Ag 109”	9.87378	0.269588	0.000078	0.269511
7	“Ag 107”	10.01824	0.282281	0.282200	0.000031	0.282348	0.517264	0.545849
	“Ag 109”	10.01824	0.282281	0.282200	0.000031
8	“Ag 107”	9.90908	0.270552	0.520474	0.000149	0.520552	0.270624	1.92353
	“Ag 109”	9.90908	0.270552	0.000078	0.270475

2.7. Isotopic Analyses of the Calibration Mixes and the Standard Sample

Two complete sets of analyses of the calibration mixes and reference sample were performed by Operator 1 on MS #1 and Operator 2 on MS #5. Operator 1 performed three analyses of each calibration mix and 24 analyses of the reference sample. Operator 2 performed four analyses of each calibration mix and 32 analyses of the reference sample. The samples were run in a pattern alternating randomly selected mixes with the reference sample.

3. Results and Discussion

The results of the measurement of the eight calibration mixes are shown in table 5. The calibration factors for each analyst varied over a range of 0.011 percent for Operator 1 and 0.019 percent for Operator 2. In addition, the calibration factors for those mixes with a 107/109 = 2 and 107/109 = 0.5 were indistinguishable from those bracketing the reference sample, indicating an insignificant degree of nonlinearity over the measured range.

Table 5.

Determination of correction factors

Calibration Sample No.	Isotopic Ratio, 107Ag/109Ag			Correction Factors
	Calculated	Operator 1	Operator 2	Operator 1	Operator 2
1	1.071840	1.081566	1.081420	0.99100367	0.99114511
2	1.064016	1.073622	1.073579	0.99104924	0.99109658
3	1.071705	1.081345	1.081214	0.99108175	0.99120936
4	1.068671	1.078304	1.078236	0.99106309	0.99113312
5	1.085906	1.095850	1.095681	0.99100465	0.99116505
6	1.056931	1.066516	1.066407	0.99100878	0.99111770
7	0.5458493	0.5507895	0.5506505	0.99102567	0.99128580
8	1.923532	1.941050	1.940748	0.99097136	0.99113254
	Mean Values of Calibration Factors			0.9910261	0.9911607

Table 6 summarizes the observed and corrected ¹⁰⁷Ag/¹⁰⁹Ag values for the SRM for Operators 1 and 2 as well as the absolute isotopic abundance ratio for silver and its uncertainty.

Table 6.

Determination of corrected isotopic ratios

	Observed 107Ag/109Ag	Correction Factor	Corrected 107Ag/109Ag
Operator 1	1.0861089	0.9910261	1.076362
Operator 2	1.0859898	0.9911607	1.076390
		MEAN	1.07638
		±	0.00022
Uncertainty components:
95 percent confidence limits in ratio determination . . . .			±0.00003
Bounds due to possible systematic error in composition of separated isotopes . . . . . . . .. . . . . . . . . . . . . . .			±0.00009
Bounds due to possible systematic error in chemical analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .			±0.00010

Table 7 gives summary calculations of the reference sample. The atomic weight is calculated from the absolute isotopic abundance by summing the product of the nuclidic masses obtained from Wapstra and Bos [17] and the corresponding atom fractions.

Table 7.

Summary calculation of the atomic weight of silver

	Uncertainly Components
Values	Overall Limit of Errora	Mass Speclrometric Analytical Error	Possible Systematic Error in Composition of Separated Isotopes	Possible Systematic Error in Chemical Analysis
Atomic Weight = 107.86815	±0.00011b	±0.0000137	±0.0000408	±0.0000466
Nuclidic Masses (l2C = 12)
107 Ag = 106.905095	±0.000012
109Ag = 108.904754	±0.000010
Atom Percent
107Ag = 51.839170	±0.005057	±0.000687	±0.00204	±0.00233
109Ag = 48.160830	±0.005057	±0.000687	±0.00204	±0.00233
Isotopic Ratio
l07Ag/l09Ag = 1.07638	±0.00022	±0.0000296	±0.0000880	±0.0001006

^aThe overall limit of error is the sum of the 95 percent confidence limits and the terms covering effect of known sources of possible systematic error.

^bIncludes a component for uncertainty in nuclidic masses.

The atomic weight of silver, 107.868 ± 0.001, as recommended by the IUPAC Commission on Atomic Weights, is based on the isotopic abundance measurements reported by Shields, Gamer, and Dibeler [1] in 1962. This ratio is, in fact, the average of their isotopic abundance measurements published in 1960 [26] and the results of the mineral survey described in the 1962 publication.

A reanalysis of sample C-140, a nugget of native silver from Ontario, Canada, which had yielded anomalous data in the 1962 study [1], was undertaken. One section of this nugget, referred to as position 1, had yielded a ratio of ¹⁰⁷Ag/¹⁰⁹Ag which was 0.3 percent higher than three other sections, designated positions 2, 3, and 4; all of which mirrored the ¹⁰⁷Ag/¹⁰⁹Ag ratio of the natural silver reference sample (SRM 948). At the time, little or no chemistry was performed to purify the samples prior to analysis, and it was believed that the “anomaly” might be due to natural differences or chemical impurities. A reanalysis of the four sections of C-140, confirmed that position 1 yielded a ¹⁰⁷Ag/¹⁰⁹Ag ratio that was higher than the ratios for positions 2, 3, and 4, which again yielded ratios indistinguishable from the natural silver reference standard. Subsequent analyses indicated the major impurity at position 1 was mercury. Following purification by electrodeposition and ion exchange chromatography, the ¹⁰⁷Ag/¹⁰⁹Ag ratio for position 1 matched the ratios obtained for positions 2, 3, and 4, and thus, the reference standard. This experiment clearly documents the critical importance of chemical purification in the determination of accurate isotopic ratios and in establishing the limits for natural isotopic variations. The data reflecting the effects of purification on sample C-140 are shown in table 8.

Table 8.

Effect of Purification on C-140

Chemical Procedure	Correcteda 107Ag/l09Ag
Initial Electrodeposition	1.07686
Additional Electrodeposition	1.07676
Anion Exchange, Precipitation and Electrodeposition	1.07634

^aAll ratios have been placed on an absolute basis using the absolute isotopic abundance ratio for silver from Table 7 (^l07Ag/¹⁰⁹Ag = 1.07638).

In an attempt to establish the limits of isotopic variability among commercially available silver, a series of analyses were performed on silver metal samples of varying purity, which had been obtained from refineries throughout the world. The results are shown in table 9. With the exception of the marz grade silver from Materials Research Corporation, all samples yielded isotopic ratios which were in excellent agreement with the natural reference standard. Although a deviation of 7.6 parts in l0⁴ is small, it is not known whether the difference is real or an artifact due to a chemical impurity. Additional testing of this material will be done in conjunction with a more thorough mineral survey in the future.

Table 9.

Isotope Ratios of Commercial Silver

Descriptiona	Correctedb 107Ag/l09Ag
Needle Ag, 99.995%, Engelhard, Indiana, USA	1.07650
AgNO3, 99.999%, Engelhard, Indiana, USA	1.07633
Ag, Marz Grade, Materials Research Corp., USA	1.07557
Ag, Single Crystal, Marz Grade, USA	1.07622
Refined Ag, Port Pirie, Australia	1.07634
Ag, Normal Purity, Johnson Malhey, Canada	1.07640
Ag, High Purity, Cominco, TADANAC, Canada	1.07648
Refined Ag, Mixture of Swedish Ores	1.07646
Electrolytic Ag, Boliden Mine, Sweden	1.07670
Ag, Crystalline, Mexico	1.07638
Ag, 99.99 + %, Lima, Peru	1.07653
Ag, Plata Fina, Granalla, Mexieo	1.07656
Average	1.07637
Standard Deviation	0.000282

^aAll descriptions are as stated by the producer. No further purity veri, verification was made.

^bAll ratios have been placed on an absolute basis using the absolute isotopic abundance ratio for silver from Table 7 (¹⁰⁷Ag/¹⁰⁹Ag = 1.07638).

The effect of this isotopic survey of commercial silver on the atomic weight is small. A recalculation of the atomic weight of silver based on the isotope ratios of these materials yields a value of 107.86816 ± 0.00032 at the 95 percent confidence level, which is nearly identical to the atomic weight of the reference sample (107.86815) as given in table 7, the major difference existing in the uncertainty.

The reference sample used in this atomic weight determination will be issued as both an isotopic and assay Standard Reference Material (SRM) and will be designated as SRM 978a. Its isotopic analysis was found to be experimentally identical to SRM 978, the major difference existing in the uncertainty placed on the measurement.

The redetermination of the atomic weight of silver has allowed the recalculation of the Faraday from the electrochemical equivalent of silver as determined by Bower and Davis [5], yielding a value of 96486.18 ± 0.13 A_NBS·s·mol^−l (one standard deviation). Further details of the calculation of this constant are given elsewhere in this issue of the Journal of Research [27].

The reduction of the overall uncertainty in the Faraday constant has allowed discernment of the discrepancies which existed between the Faraday determined from electrochemical experiments, and the Faraday calculated from other fundamental constants. It is expected that the new NBS value of the Faraday will be included in the 1981 CODATA least squares adjustment of the fundamental constants.