The measurement of silver in road salt by electrothermal atomic absorption spectrometry

Abstract

The strong complexes formed between silver (Ag⁺) and chloride (Cl⁻) may give rise to salt deposits from ancient seas containing elevated Ag concentrations. Since salt from these deposits is frequently used for road de-icing in the northern United States, elevated Ag concentrations in salt represent a potential source of Ag to the areas in which de-icing is performed. In this paper we describe the methodology used to determine the concentration of Ag in salt stored for road application in Madison, Wisconsin, USA. Road salt was dissolved in ultrapure water and analyzed by electrothermal atomic absorption spectrometry. A combination of low Ag concentrations and interferents in the matrix led to pre-concentration and matrix removal via co-precipitation with cobalt pyrrolidinedithiocarbamate and subsequent re-solubilization. Concentrations of Ag in road salt of 102–141 pg g⁻¹ were determined. Method recovery spikes and blanks indicated that neither analytical interferences nor contamination during sample preparation were significant. Calculations based on these concentrations have led to two important conclusions: (1) based on average ocean salinity values, Ag concentrations in open ocean water and in salt deposits are similar; and (2) compared to other sources, road salt is unlikely to represent an important source of Ag to areas in which salt is used for de-icing.

---

1 Introduction

Although silver is a naturally occurring element with an average crustal concentration of 0.1 mg kg⁻¹, the major sources of environmental silver are anthropogenic.¹ The toxicity of Ag, in its free ionic form (Ag⁺), to a variety of aquatic organisms has been documented.² A possible source of silver to watersheds is salt applied to roads during the winter months for de-icing. Environmental impacts resulting from road salt application have been studied extensively. The impacts include increases in ground water chloride concentrations,³ and mobilization of organic matter and metals including Cr, Pb, Ni, Fe, Cd and Cu from roadside soils.⁴ Relatively little information is available, however, on potential contamination of surface and ground water by trace constituents in the salt itself. One study of impurities in rock salt reported Ag concentrations ranging from <0.005 to 0.09 mg kg⁻¹,⁵ but it was performed well before the introduction of ‘clean’ sampling and analytical techniques.⁶

In the state of Wisconsin (USA), salt for road application primarily consists of halite (NaCl) mined from evaporite beds. The majority of this salt is extracted from underground deposits rather than more recent evaporites (e.g., those around the Great Salt Lake, USA). While the specific mechanism by which these thick beds of salt were deposited is a matter of controversy,⁷ there is general agreement that the salt beds formed through precipitation of salts as ancient water bodies evaporated. We hypothesized that, due to the affinity of silver ions for chloride ions (log K_f (AgCl_aq) = 3.31, log K_sp (AgCl_s) = −9.75),⁸ Ag might be concentrated in salt deposits as a result of geologic processes, or that salt might be contaminated with small amounts of Ag during extraction, transport and storage. If such enrichment occurred, road salt application could represent an important vector for Ag contamination of water bodies near sources of concentrated application. We tested this hypothesis by analyzing samples of road salt for Ag concentration.

The analysis of salt for trace levels of silver by electrothermal atomic absorption spectrometry (ETAAS) presents a number of difficulties because of interference from the inorganic matrix and the high volatility of silver. For a direct measurement on an aqueous solution of the sample, these effects may be largely removed by the use of appropriate matrix modifiers.⁹ At the ng g⁻¹ level, direct determination is unlikely to be successful because of the need to dissolve large amounts of salt, thereby increasing matrix interference effects. To circumvent these problems, a pre-concentration method has been used that extracts silver from sea water at the ng l⁻¹ level using co-precipitation with cobalt(II) pyrrolidinedithiocarbamate.¹⁰ This not only provides an enrichment of 100 but also separates the silver from the salt matrix.

In this paper we report the application of this method to the determination of total silver in road salt samples at pg g⁻¹ levels, and discuss the level of concern this concentration represents as an environmental contaminant.

2 Experimental

2.1 Samples

Road salt samples were obtained from two city storage facilities (west and east) in Madison, Wisconsin. The salt is stored under cover in large piles. Samples were removed from different parts of a pile near its surface after removing the crusted surface layer. Samples were placed in acid-cleaned (see below) 125 ml Teflon bottles and double bagged in new polyethylene bags. The bottles were handled using polyethylene gloves and “clean hands, dirty hands” techniques⁶ during sampling, as well as during storage and analysis.

2.2 Sample preparation

For direct measurement salt samples were dissolved in 0.5% nitric acid (TraceMetal Grade, Fisher, Pittsburgh, PA) to give a road salt concentration of approximately 6 g l⁻¹. Solutions were then filtered through acid-washed polycarbonate filters before analysis by ETAAS.

The pre-concentration method was used as described by Bloom and Crecelius.¹⁰ A known mass of the road salt was dissolved in a measured mass of Milli-Q (Millipone-waters, Bedford, MA, USA) water (either 500 or 1000 g) and the pH adjusted to 2.0 using concentrated nitric acid. Aliquots (200 g) of the solution were poured into 250 ml Teflon bottles along with 1 ml of 2% ammonium pyrrolidinedithiocarbamate (APDC) solution and 1 ml of 200 mg l⁻¹ cobalt nitrate solution. The APDC solution was purified by extracting any metal complexes into chloroform. The cobalt solution was a dilution of a high-purity ICP-MS cobalt standard (High Purity Standards, Inc., Charleston, SC). While no reference material is available for a matrix of this sort, the effect of the road salt matrix on analytical recovery could be tested by spiking road salt solutions with Ag prior to preconcentration. For these method recovery experiments a 500 μl spike of a 2.0 μg l⁻¹ silver nitrate standard solution was added to the solution prior to the addition of the APDC and the cobalt nitrate solutions.

The solutions were allowed to stand for approximately 1 h and then filtered through 0.4 μm polycarbonate filters. The filters were rinsed with Milli-Q water and then folded into quarters and placed in clean dry vials. The precipitate was digested by adding 300 μl of nitric acid and warming on a hotplate for 2 h. The nitric acid was allowed to evaporate, and the residue was re-dissolved in 2 ml of a solution containing 0.2% (m/v) ammonium dihydrogen phosphate and 5% (v/v) nitric acid. The vials were capped and placed in an oven at 60 °C for 8 h. Solutions were analyzed by ETAAS. Acid and ammonium dihydrogen phosphate used during pre-concentration were Ultrex grade (VWR Scientific, West Chester, PA).

2.3 ETAAS analysis

The analysis was carried out on a Perkin-Elmer (Norwalk, CT, USA) 5100 atomic absorption spectrometer fitted with Zeeman background correction, an HGA 600 furnace and an AS-60 autosampler. The instrument was situated in a trace-metal clean laboratory. The 328 nm line of a silver hollow cathode lamp was used for all the absorption measurements and a pyro-coated graphite tube with L’Vov platform was installed in the furnace. The diluent used was a 0.5% (v/v) nitric acid solution and the matrix modifier was a 26 g l⁻¹ ammonium dihydrogen phosphate solution. Sample, diluent and matrix modifier (0.1 mg per analysis) were pipetted into the graphite tube together. When concentrations were low, multiple pipettings and evaporations were used to increase the Ag concentration before atomization. Reagents were Ultrex grade (VWR Scientific).

2.4 Plasticware

All bottles, filtration equipment and autosampler cups were made from Teflon (Savillex and Nalgene). Inside surfaces of bottles were cleaned by contacting them with 50% hydrochloric acid (HCl; Fisher, ACS grade) for 24 h, and then contacting them with 50% nitric acid (HNO₃; Fisher Scientific, ACS grade) for 24 h. Finally, the inside surfaces were contacted with freshly made 1% HNO₃ (Fisher Scientific, TraceMetal grade) for 12 h. Outside surfaces were cleaned by soaking them in 20% HNO₃ for 24 h. All surfaces were rinsed at least three times with Milli-Q water between steps and before use, and bottles were dried in a laminar flow hood situated in a trace-metal clean laboratory. This cleaning method is similar to that used by Shafer et al..¹¹

3 Results and discussion

The fraction of Ag spikes recovered from ETAAS analysis of reagent-grade NaCl solution matrices was determined to establish whether a direct analysis for Ag in dissolved salt solutions would be possible. Effects of three matrix modifiers, ammonium dihydrogen phosphate, palladium alone, and palladium in combination with ascorbic acid⁹ were compared. For all modifiers tested, however, the analyses gave instrumental spike recoveries that were unacceptably variable (44–90%). When NaCl was added to the calibration standards at a concentration similar to those in samples to be analyzed (6 g l⁻¹ salt), recoveries were more acceptable (98–105%), but Ag concentration measurements in road salt solutions were indistinguishable from blanks. Increasing road salt concentration to levels where Ag might have been quantifiable by direct analysis (e.g., 60 g l⁻¹) led to background peak heights greater than those recommended for Zeeman background correction. This level of salt in the analyte solution also led to the formation of deposits on the contact cylinders and optical windows. It was therefore determined that Ag in aqueous solutions of road salt must be concentrated before analysis. The Co–APDC coprecipitation procedure of Bloom and Crecelius¹⁰ was chosen for this purpose.

Several techniques were employed to test the efficacy of the Co–APDC method for Ag determination in a matrix of dissolved road salt. Results of many of these tests are summarized in Table 1. Silver concentration, when corrected for method blank values, has a roughly linear relationship with salt concentration in the unconcentrated solution, indicating that the fraction of Ag concentrated by the Co–APDC method was not affected by the concentration of road salt over the range tested. Instrument and method recoveries performed on road salt concentrates were all less than complete, and had no obvious correlation with salt concentration. Low recoveries might be expected if, for example, very small particles of AgCl precipitated from the original solutions and passed through the filter during the Co–APDC concentration process. Tests performed on spiked and unspiked samples of Ultrex NaCl indicate, however, that the NaCl in road salt is not the source of the incomplete recoveries of Ag in road salt samples. Chemical speciation modeling (performed using the equilibrium speciation model MINEQL¹²) confirmed that, at the concentrations of Ag and Cl present in the samples, saturation with respect to AgCl was not reached. The similarity of instrument and method recovery values for road salt samples suggests that some other component of the road salt acts as an interferent during the analysis process. A chemical species in the road salt may, for example, allow Ag to volatilize from the heated graphite tube in ionic rather than atomic form. ICP-MS analysis of a 200 mg l⁻¹ road salt solution performed using the procedure of Shafer et al.¹¹ indicated that road salt contained 0.15 to 0.4 μg g⁻¹ of Cu, Ni and Pb, all of which will be concentrated by the Co–APDC process.¹⁰ In addition, Fe and V are present at 10 to 100 times these levels. This leads to a relatively complex matrix in Co–APDC precipitates. It should be noted that, though our method recoveries were lower than those of Bloom and Crecelius, our solutions were spiked at Ag concentrations 2 to 20 times lower than those of Bloom and Crecelius. In addition, our spike recoveries were relatively consistent for a given salt concentration. Overall, our results lead us to agree with the conclusion of Bloom and Crecelius that the Co–APDC precipitation technique is effective for concentrating Ag and separating it from a complex matrix with high Na⁺ and Cl⁻ concentrations.

Table 1 Representative measurements of Ag recovery carried out on road salt and Ultrex sodium chloride. Instrument recovery refers to Ag spiked into road salt samples already concentrated via the Co–APDC method. Method recovery refers to Ag spiked into the road salt solution prior to application of Co-APDC concentration

Sample	Salt concentration/ g l^−1	Instrumental measurement/ μg l^−1	Instrument recovery	Method recovery
Road salt	40	0.332	0.775	0.746
Road salt	20	0.210	0.895	0.840
Road salt	6	0.104	0.928	0.802
Ultrex NaCl	20	0.046	1.033	0.960
Blank	0	0.042	0.909	0.912

---

In spite of observed variability in recovery values, instrument precision with respect to analysis of concentrated road salt samples was acceptable. Concentration measurement was performed 5 times in one analytical run on a single concentrate of a 40 g l⁻¹ road salt solution (Table 2), and variation was small (mean = 0.354, s = 0.011, n = 5). It was noted, however, that variation among concentrates from 40 g l⁻¹ samples, especially between-run variability was disproportionately large compared to those from 20 g l⁻¹ samples. Based on this observation, it was decided that the salt concentrations in solutions to be concentrated should not exceed 20 g l⁻¹.

Table 2 Precision data for the instrumental measurement on road salt solutions concentrated by the Co–APDC method. The solution analyzed was a concentrate from a 40 g l⁻¹ road salt solution

Measurement	Instrument concentration/ μg l^−1	Instrument recovery
1	0.343	0.844
2	0.354	0.798
3	0.368	0.661
4	0.361	0.731
5	0.344	0.777
Mean	0.354	—
s	0.011	—

---

A method limit of quantitation was calculated as 0.087 μg l⁻¹ using the formula LOQ = method blank + 10s,¹³ and it was decided that 20 g l⁻¹ would provide analytically quantifiable Co–APDC concentrates with respect to Ag (Table 3). Solutions of 20 g l⁻¹ were therefore used to establish the mean Ag concentration and the Ag concentration range in road salt.

Table 3 Mean concentrations, blanks and spike recoveries for Co–APDC concentrates. Concentration and recovery values are from 20 g l⁻¹ road salt solutions only

Ag in concentrates/ μg l^−1, uncorrected			Method blank/μg l^−1			Method recovery
Mean	s	n	Mean	s	n	Mean	s	n
0.23	0.02	6	0.047	0.004	5	0.79	0.024	3

---

Silver concentrations were determined on salt from the storage areas on the east and west sides of Madison. Composite samples were created for each location by combining approximately equal masses of salt from different locations on the pile. An effort was made to exclude insoluble matter such as small pebbles from these composites. Three separate 10-g masses of each of these composites were dissolved in 500 ml of ultrapure water. Each of these solutions was concentrated and analyzed separately (Table 4). The overall mean of these analyses is 119 pg g⁻¹, and the standard deviation (based on mathematically combining the standard deviation values for uncorrected concentrations, blanks, and recovery values) is 16 pg g⁻¹. The mean concentrations for the west and eastside storage locations, 107 and 131 pg g⁻¹, respectively, are each less than one standard deviation from the overall mean.

Table 4 Concentration of Ag in salt from City of Madison storage piles. Concentrations have been corrected using mean method blanks and method recoveries

Ag concentration in road salt/pg g^−1
Replicate	West side pile	East side pile
1	114	126
2	104	141
3	102	126

---

Results of these analyses provide insight into the extraction and use of the salt that Madison, Wisconsin uses on its roads. Salt used on Madison roads is primarily extracted from mines in the mid-western US and Canada, which is present as a result of the evaporation of ancient water bodies. Based on the source of our samples, it is impossible to pinpoint the specific deposits from which the road salt was extracted. In spite of this fact, however, some interesting comparisons between salt-deposit Ag concentrations and ocean Ag concentrations can be made. Several researchers, using state-of-the-art clean sampling and analysis techniques have determined Ag concentration ranges in the open ocean; these values range from less than 0.26 to 580 pg l⁻¹ Ag. Deeper water generally contains a higher Ag concentration.^10,14–16 If an average salt concentration of 30 g l⁻¹ is assumed for ocean water, this leads to a range of concentrations from less than 0.9 to 190 pg Ag (g salt)⁻¹. The range of Ag concentrations determined for road salt fits within this range, suggesting that (1) the salt is not contaminated with Ag during extraction, transportation and storage, and (2) Ag concentrations in ancient water bodies was similar to that in modern ocean water.

The City of Madison purchases approximately 9000 tons of road salt each year, generally from a State of Wisconsin contract on which approximately 570 000 tons per year are purchased. The state purchases this salt for use on state highways and to offer for sale to municipalities with populations greater than 1500.¹⁷ Based on the average concentration of Ag in this salt as determined in this study, road salt introduces approximately 1 g of Ag to the Madison environment and approximately 60 g to the entire state annually, assuming that all 9000 tons are spread in Madison and 570 000 tons are spread statewide. In contrast, effluent from the publicly owned treatment work that serves the greater Madison area introduces approximately 12 kg of Ag to its receiving stream each year (based on a 1-d composite Ag concentration and an average plant volume discharge).¹¹ Thus, it can be assumed that the Ag introduced to the watersheds in the Madison area through road salt is a tiny fraction of the total anthropogenic input.

4 References

1. T. W. Purcell and J. J. Peters, Environ. Toxicol. Chem., 1998, 17, 539.
2. R. Eisler, a synoptic review, US Government of the Interior, Washington DC, Biological Report 32, National Biological Service 1996..
3. K. W. F. Howard and P. J. Beck, J. Contam. Hydrol., 1993, 12, 245.
4. C. Amrhein, J. E. Strong and P. A. Mosher, Environ. Sci. Technol., 1992, 26, 703.
5. A. Bloomberg and K. Ladenberg, J. Electrochem. Soc., 1959, 106, 54.
6. C. C. Patterson and D. M. Settle, Accuracy in trace analysis: sampling, sample handling, and analysis, Special Publication 422, US Bureau of Standards, Washington DC, 1976, pp. 321-351..
7. M. K. Jenyon, Salt Tectonics, Elsevier Applied Science Publishers, London, 1986..
8. A. E. Martell and R. M. Smith, Critically Selected Stability Constants of Metal Complexes Database, National Institute of Standards and Testing Standard Reference Database 46, Version 2.0, Gaithersburg, MD, 1995..
9. P. Bermejo-Barrera, J. Moreda-Pineiro, A. Moreda-Pineiro and A. Bermejo-Barrera, Talanta, 1996, 43, 35.
10. N. S. Bloom and E. A. Crecelius, Anal. Chim. Acta, 1984, 156, 139.
11. M. M. Shafer, J. T. Overdier and D. E. Armstrong, Environ. Toxicol. Chem., 1998, 17, 630.
12. W. D. Schecker and D. C. McAvoy, MINEQL+ Version 2.1; A chemical equilibrium program for personal computer, The Proctor and Gamble Co., Cincinnati, OH, 1991..
13. Standard Methods for the Examination of Water and Wastewater, ed. L. S. Clesceri, A. E. Greenberg and R. R. Trussel, 17th edn., American Public Health Association, Washington, DC, 1989..
14. J. H. Martin, G. A. Knauer and R. M. Gordon, Nature (London), 1983, 305, 306.
15. A. R. Flegal, S. Sanudo-Wilhelmy and G. M. Scelfo, Mar. Chem., 1995, 49, 315.
16. M. Murozomi, Japan Analyst, 1981, 30, S19.
17. Personal communication, Penny VanDeventer, State of Wisconsin..

---

Footnote

† Presented at SAC 99, Dublin, Ireland, July 25–30, 1999.

---