Microdetermination of Silver Using Ion Exchange Concentration N. S. McNUTT and R. H. MAIER Department o f Agricultural Chemistry and Soils, University o f Arizona, Tucson, A r k .
b A method using ordinary laboratory equipment i s described for the microdetermination of silver ion in dilute solutions containing less than 0.01 2 p.p.m. of manganese. The silver ion i s concentrated on a cation exchange resin bed and then removed from the resin with boiling 1M sodium sulfite. The silver ion in the eluent is determined colorimetrically, utilizing the color of the permanganate ion produced when manganese ion, persulfate ion, and catalytic amounts of silver ion are heated together above 50" C. This procedure i s used for the detection of 0.05 to 0.30 i 0.02 p g . of silver. Few interferences are encountered. High total ionic salt concentrations tend to retard silver adsorption on the resin. Duplicate or triplicate determinations should b e run and the results averaged, due to the necessity of extreme care to prevent silver contamination and loss.
T
HE existing methods for the determination of silver either require large amounts of silver or are subject to various interferences. Many methods have been published for the determination of Ag+ involving the precipitated halide ( I , 5). Several solvent estraction methods have been used to remove silver from aqueous solution ( 2 , 6, 8). Some very sensitive determinations using electrochemical methods (3, 4, 7 , 10. 11) h a w been developed. A nicthod for the determination of AgI in aerosols utilizes the electron micro"cope to measure crystal volumes (9). These methods all involve the use of c'\p m i v e and operationally difficult equipment. A method applicable to simple laboratory equipment was sought. The most promising colorimetric analysis for silver was based on the catalytic effect of Ag' in the osidation of divalent manganese to permanganate ion by the persulfate anion (13). The application of this method t o the determination of silver in rain water samples Jyas an objective of this study.
APPARATUS
A borosilicate glass ion ex change column (1.6 cm. X 19 cm.) equipped with a Teflon stopcock was used. A borosilicate glass n-001 pad was placed 276
ANALYTICAL CHEMISTRY
in the bottom of this glass tube just above the stopcock. Eluents were collected in conical graduated Pyrex centrifuge tubes (2.5 em. X 13 cm., capacity 40 ml.). A constant temperature ( + l o C.) water bath (hlagniTVhirl, Blue ill. Electric Co., Blue Island, 111.) was used in the digestion of samples. REAGENTS
All chemicals were reagent grade. Silver standard solutions nere prepared from AgKOo. The salt was dried a t 70" C. before weighing. The stock solution (1000 pg. per ml.) was placed in a polyethylene bottle in a dark place. All dilutions were made from this solution. LIanganese sulfate solution (0.006X) was prepared from AInS04 HzO, dried a t 70" C. Sodium sulfite solution (IJI) was prepared from the crystal. This solution was prepared fresh for each use, since hot ldf SazSOaoxidizes rapidly in air. The ion exchange resin used was Illco-211 W,hydrogen form, distributed by Illinois Water Treatment Co., 840 Cedar Street, Rockford, Ill. Deionized water was used for all dilutions. Deionized water was prepared by passing distilled n ater through a niised bed cation-anion exchange system and was stored in a polyethylene bottle. Only borosilicate glnssn-are was used and all except pipets were washed by ordinary laboratory washing procedure, then rinsed n i t h 0 . 1 s K C S , and approximately 10 times TT ith deionized 17-ater. The pipets TTere soaked in 15% (by vol.) K H 4 0 H solution. rinsed before and after use with deionized H20, and replaced in the IVH40H solution. Occasionally the centrifuge tubes nere soaked in 0.LV K C S for 24 hours. Parafilm dust covers n-ere used on each piece of glassnare used in the determination. RECOMMENDED PROCEDURE
T o approximately 100 grams of Illco-211 T T resin (H- form) add about 100 ml. of 6A7 NaOH. Keep the slurry cool as excess heating with strong base can cause premature decomposition of the resin. Drain the resin and add 0.1S K C S in exress, rinse several times with O . l N KCK letting each rinse remain in contact n i t h the resin for several minutes with occasional stirring.
This should remove trace metal contaminants. Rinse n ell n i t h deionized H20 (ten 100-ml. portions) making sure to remove all cyanide from the resin. -4dd three 100-ml. portions of 131 HE03 to produce the hydrogen form of the resin. Air dry the resin in a cyanide-cleaned beaker. Weigh out 1.00 gram of the dried resin and place in the ion escliange column, add 10 ml. of deionized H20, swirl and rinse the walls of the column to obtain a compact resin bed. Add the sample to the resin bed and permit the eluent to drain a t approximately 4 in1 per minute (for samples with a Ag concentration on the order of 2 p.p.b.) Rinse the walls of the column with two 10-ml. portions of deionized H20 maintaining the same eluting rate. Add 5 ml. of boiling 1-11 Ka2S03to the column and let stand for 1 hour (initial equilibrium temperatures are approximately 50" C.), and drain a t the rate of 4 ml. per minute. Add 3 ml. of boiling 1M Sa2S03,let stand for 10 minutes, then drain a t the snme eluting rate. These eluents should be drained into conical Pyrex centrifuge tubes. Add exactly 1 ml. of 0.00631 llInSO1. then add 0.5 ml. of 7.551 H3P01. The p H should be approximately 0.9. Adjust the final volume of the solution to 10 nil. by the addition of deionized HzO if necessary. Add to this solution 2 grams of solid K2Sz0,. Place the tubes in a compartmentalized wirc rack and place this rack in the constant temperature water bath whirh is standardized a t some temperature between 97" and 100" C. Keep the tubes in the water bath for exactly 15 minutes. At the end of 15 minutes, quickly place the rack in an ice water bath. The oxidation of Nn+2 is started by heating the solution above 50" C. and stopped by cooling below 50" C. Crystals may precipitate in the cooled solution. For a large volume of crystals, bring the total volume of tube contents to 15 nil. nith deionized H20, stir IT ith a polyethylene rod, then centrifuge the solution. If there are fen. crystals, simply let them settle. If crystals present no problem, a total volume of 10 nil. may be used to obtain more sensitive standard curves. Decant the clear supernatant solution into a spectrophotometer tube and determine the per cent transmittance in a Bausch &- Lomb Spectronic 20 a t a wavelength of 525 nlp. Prepare a standard curve for silver by following exactly the above treatments and procedure.
To clean the used resin column and bed, rinse with 10 ml. of deionized HzO, two 10-ml. portions of 13’ HNOI, then follow this with 10 ml. of deionized H20. Do not use the same resin bed for more than two determinations because of the deleterious effects of heat. DISCUSSION O F PROCEDURE
This method is very sensitive and is therefore subject t o many sources of crratic behavior. Kith a method of this type every piece of glassware and polyethylene ware must be standardized for size and degree of cleanliness; for example. pipets should be rinsed n-ith the solution to be pipetted four times bcforc delivering solutions in the procedure. These conditions are arbitrary but should be chosen and strictly folloned. I n carrying through the same procedure several times (in the same n a y as far as can be observed), the last set of detcvminations was less wratic than the first w t of detcrminations and a gradient can almost be set for erratic behavior. il possililc cause of this tlccrease in erratic behavior might lie some factor such as the glassware “getting ujcd to” a change in contents. I t should bc stressed that all determinations should be run in dupliratc, or in triplicate if practical, and the resulti avcrdged. rl standard curve waq obtaincd for Ag+ in thc colorinic.tric procedure alone. No concentration ptep n i t h the ion exchange resin ‘rl as involved. The appropriate Ag+ stxndard aliquots were pipetted directly into the centrifuge tubrs; l l n S 0 4 , &P04, and K2Sz0, were added and the determination completed as in the recommended procedurc. The pH of the solution n a s appro\imately 1.2. The curve obtained is shown as line I1 in Figure 1 vhich reprebents the avtwige of four complete calibration curves completed on differe n t days. To determine the interference due to Llf Sa2S03,the following n as carried out:
The standard Ag+ solution (0.1 p g . per nil.) was made 124 n-ith respect t o Sa2S03. This standard was pipetted directly into the centrifuge tubes. The hfnS04 and H3P04nere added and the solution volume increased t o 10 ml. n i t h 1Jf ?r’apSO,. The K&08 solid n a s then added and digestion begun a s in the recommended procedure.
A graph of the average of six separate standard curve determinations is shown as line I in Figure 1. The displacement of line 11 from line I shows the interference of the addition of 121 Sa2S03. This interference is reproducible if the concentration of Na2S03in the centrifuge tubes remains constant. A standard curve was obtained for Ag+ from a standard 0.1 pg. per nil.
30_t-
;r-C
035
,
010
I
I
015
020
025
030
Pg.A g Figure 1 .
Standard curves for silver
(Complete procedure blank at 100% T )
Ag- in deionized HzO (no complexing agent added) passed through the resin column, according to the recommended procedure. The silver was pipetted directly onto the resin bed-this eliminates any erratic behavior due to dilution to concentrations approximati,ly 1 p.p.b. or due to losses by silver adhering to the surfaces of polyethylene sample containers. Line I11 in Figure 1 shows the averages of siy curves obtained by this procedure. 9 standard curve was determined for simulated rain water samples, n-hich were aliquots of the silver standard containing 0 to 0.30 Ag+ diluted to 100 ml. in polyethylene bottles. The exact recommended procedure was carried out using these dilutions of the standard as samples. These results are shown in line I V in Figure 1. The erratic and lower recoveries represented b y line IV as compared to line I11 are not due to dilution but are probably due to Ag+ adsorption on the walls of the polyethylene bottles. I n support of this view, a n experiment was conducted which showed that it vas not dilution n hich caused the variation in precision and accuracy. The 100-ml. polyethylene bottles were partially filled with varied volumes and concentrations of Ag‘ solutions as shown in Table I. An investigation was conducted on the possible use of the Ag+-resin particle complex instead of AgT as the catalyst for the oxidation of to h l n 0 4 - by KzS208,the reaction which is t h r basis
for the color formation step in the recommended procedure. Substitution of the Ag+-resin particle complex as the catalyst would eliminate the need for an eluting agent to remove the silver from the resin. This procedure proved to be unsatisfactory and further investigation of this approach was halted in favor of the column technique with hot 1J4 NazSOa, after the value of SOs- as a complexing agent for silver was realized ( I @ . Interferences in the colorimetric digestion step were not encountered from 100 pg. Zn+2, CU+?.and Fe+3; or from Ca+2, Mg+2, KO3-, I-, NHs, EDT-4, and SOB-2. Preliminary investigation of equilibrium distribution constants, KD= (Amt. of solute on resin)(\’ol. of soln.) (G. of recin) (Amt. of solute in soln.)
detcrmined by equilibrating a certain amount of resin 111th a silver solution and determiniIiA the Ag- in solution bcforc and after resin addition, indicated t h a t 111 SaKO,, 0 . l M HSO,, 0.001.V KI, 0.1M N H 4 0 H , and 0.0121 EDTA iwrc less effective in producing a low K Dvalue than hot I N SapSOs. Several rain water samples \T ere carried through the recommended procedure using loo-, 250-, and 500-nil. volumes; no .4g+ FT-aL detectable in the range of 0.6 p.p.b. Addition of 4 ml. of 16N H N 0 3 to several 500-nil. samples and subsequent heating to 97’ C. inside the polyethylene containcrs for 40 minutes gave no detectable increase in cationic .4g+. This digestion should have decomposed any Ag12- complex and helped to free any adhering Sg’. The presence of ionic salts, such as H S O , in the sample solution, seemed to retard the concentration of Ag+ by the resin. Since this depression is probably due to ionic replacement of Ag+ on the resin by the H + from the HKOS in the sample solution, it would be advisable t o keep ion concentration in the sample to a minimum. Locses due to the large size polyethylene sample containers (at the level of 1 p.p.b. of Ag+) eliminate the advantages of concentrating any larger volumes than 250 ml. Solutions, used to rinse the such as 0.1JI ”,OH,
Table I.
Ag+, pg.
0 0.30 0.30 0.30 0 30 0.30
Influence of Dilution on Silver Recovery Soh. Ag+ Concn., H20 Wash hg+ Recovered, Volume, M1. /*g. Volume, 11.11. P.P.B. 100 10 30 50 70 100
0 30 10 6 4.3 3
10 100 80 60 40 10
Av.
0 0.31 0.26 0.32 0.28 0.29 0.29 + 0.02
VOL. 34, NO. 2, FEBRUARY 1962
277
polyethylene containers, in an attempt t o recover adhered Ag+, caused some losses of -4g+ from the resin. The use of such solutions for recovery may be facilitated by using more than 1 gram of Illco - 211W resin for the column. The obvious necessity for use of polyethylene in all places of contact of a Ag+ solution with a solid medium should emphasize the meticulous care required in sample collection. It certainly is not possible to allow rain water samples for Ag+ analysis to stand in metal containers for days before removal for analysis. The use of glass is highly unadvisable since AgT is tightly adsorbed on the surface (IS).
LITERATURE CITED
(1) Blaedel, W. J., Meloche, V.
y,,
“Elementary Quantitative Analysis,” pp. 271-308, Row, Peterson and Co., Evanston, Ill., 1957. ( 2 ) Bode, H., 2. anal. Chern. 144, 165
(1955). (3) Cave, G. C. B., Hume, D. S . , ANAL. CHEM.24,588 (1952).
(4) Diehl, Harvey, Butler, J. P., Analyst
77, 268-72 (1952). (5) Firsching, F. H., ANAL.CHEM. 32, 1876 (1960). (6) Friedeberg, H., Zbid., 27,305 (1955). (7) Griess, J. C., Lockhart, B. R. (to
U. S. Atomic Energy Comm.) U. S. Patent 2.612.470 (Seot. 30. 1952). (8) Handliy, T. H.; Dean, J. A.,‘ANAL. CHEW32, 1878 (1960). (9) Koenig, L. R., Zbid., 31, 1732 (1959).
(10) Lord, S. R., Jr., O’Seill, R. C., Rogers, L. B., Zbid., 24, 209-13 (1952). (11) Xorwitz, George, Anal. Chim. Acta 5, 106-8 (1951). (12) Schwenck, J. R., personal com-
munication, Department of Chemistry, Sarramento Junior College, Sacramento, Cnlif
(13) Underwood, A. L., Burrill, A. M., Rogers, L. B., ANAL. &EM. 24, 1597 (1952).
RECEIVEDfor review July 13, 1961. Accepted Kovember 20, 1961. Work partially supported through Grant No. G 8216 from the National Science Foundation to the Institute of iltmospheric Physics of the University of Arizona. Contribution from the Brizona Agricultural Experiment Station, Technical Paper KO.672.
Gas Chromatographic Identification of Major Constituents of Bubbles in Glass F. R. BRYAN and J. C. NEERMAN Scientific laboratory, Ford Motor Company, Dearborn, Mich. Experiments with a standard chromatographic system and column material indicate that gas chromatography offers a feasible procedure for identification of major constituents of bubbles in glass. Detectability approaches 0.01 PI. for either nitrogen or carbon dioxide. Sulfur dioxide i s detectable to 0.1 PI, The method described has quantitative usefulness for nitrogen and carbon dioxide determinations, but requires an order of magnitude improvement in sulfur dioxide sensitivity to compete favorably with mass spectrometry.
C
has recently been devoted to the sampling and analysis of bubbles in glass by mass spectrometry (2, 3). Mass spectrometric techniques are reported to provide quantitative accuracies for major constituents to within =t2% when bubbles of approximately 1 pl. are involved, and within = t l O % when dealing with 0.05-p1. bubbles. Detectability limits for various gaseous constituents b y mass spectrometry are reported to range from 0.001 t o 0.01 pl. Although mass spectrometry is adequate in respect to detectability and accuracy for the analysis of micro-sized bubbles, there is concurrently a need for a simple and inexpensive means of identifying the major constituent of typical bubbles of about 0.5-p1. size. This requires an analytical system with detectabilities of the order of 0.1 pl. or better, since most bubbles in glass
278
ONSIDERABLE EFFORT
0
ANALYTICAL CHEMISTRY
exhibit a gas pressure of about atmosphere (2). Because gas chromatography often provides advantages of simplicity and economy over mass spectrometry, the detectability capabilities of a standard chromatographic system was investigated in respect to gases frequently encountered in commercially produced plate glass. According to previous analyses of bubbles occurring in this same type of glass, the major constituents are most likely to be nitrogen, carbon dioxide, and sulfur dioxide (8). The literature indicates that both nitrogen and carbon dioxide are detectable chromatographically in volumes as small as 0.01 pl. (1). Corresponding detectability data for sulfur dioxide were not readily found. It seemed reasonable to expect, however, that all three gases might be resolved by one column material, and that any one of the three gases might be detected if it were the major constituent of a normal sized bubble. EXPERIMENTAL
Instrument Assembly. T h e chromatographic system is assembled principally from individual commercial components. Helium carrier gas flows through t h e system at a rate of 30 cc. per minute. A 6-foot column of tritolyl phosphate in a 1/4-inch diam. copper coil is submerged in a constant temperature oil bath held t o 26 i 0.5” C. A Gow-Mac thermal conductivity cell, operating at 330 ma.,
is used as detector. Signal is fed to a multirange recorder with maximum sensitivity of 1 mv. full scale. Recorder chart speed is 0.3 inch per minute. Immediately ahead of the column, a bubble-breaker sample introduction assembly is installed in the helium line. This assembly consists of a modified bellows valve described previously (2)’ Sampling Procedure. Glass samples containing bubbles are c u t t o a size slightly less than ‘/*-inch square, a n d scribed over the center of t h e bubble. T h e sample is placed in t h e bubblebreaker, and helium flow is established to purge t h e sample chamber of air. During this period, helium should be vented from t h e system prior t o reaching the column, since moisture from the air will otherwise be retained by the column and slowly eluted t o interfere with the subsequent sample analysis. After purging the sample chamber, the helium flow is directed through the column, and the signal base line is established on the recorder. At this stage, any residual air which may be issuing from the sample chamber will be evident within a 2-minute period as a nitrogen peak. With the base line established and the recorder on maximum sensitivity, the bubble is broken allowing the contents to be released into the helium stream. Column Resolution. T h e 6-foot tritolyl phosphate column was selected from those available a n d probably does not represent t h e best possible
