Anal. Chem. 1981, 53, 1536-1538
1536
Table IV. Effect of Eluent Composition on Anion Retention
compositiona O.OOl/O.OOl
0.003/0.001 0.003/0.002 0.003/0.005 0.006/0.005 0.009/0.005 0.003/0.007 0.012/0.005 0.009/0.007 0.003/0.009 O.Ol2/0.009 0.23/0.019
eluent properties ionic strength,b pH M C110.10 9.87 10.11 10.38 10.22 10.03 10.56 9.89 10.17 10.74 10.17 10.25
0.0050 0.0063 0.0099 0.0171 0.0198 0.0025 0.0237 0.0252 0.0292 0.0315 0.0396 0.0792
NO3-
SO-;
6.4 6.0 5.2 4.5 4.4 4.3 4.4 4.3 4.2 4.1 3.9 3.5
22.0 20.4 16.0 12.4 11.8
26.8 14.7 13.8 12.7 10.8 12.2 10.0 8.8
11.1
7.8
8.8
5.2
7.4
a (mol/L of NaHCO,)/(mol/L of Na,CO,). assuming no significant ion pair formation.
10.8 10.7 10.2 9.9
Calculated
Table V. Ion Size Parameters for Selected Anions ion size, ion size, anion a anion a c1NO,'
1.81a
1.8gb
Pauling crystal radii (7). ical radii (8). a
so:
2.30b co,21.85; HCO 31.63 Yatsimirskii thermochem-
first two instances above are tentatively explained as illustrating ionic sizelcharge effects. The sequence of ionic size (S042> NO, > C1-) from Table V accurately mirrors the relative effect that increased carbonate and bicarbonate
concentration in the eluent has on the species retention characteristics. This ionic size phenomenon has been attributed to the buildup of steric stress in the resin structure (9). The observation that increased carbonate ion concentration effects elution more than bicarbonate concentration reflects the electrostatic difference between the two species. Observation 3 allows us to conclude that the p H per se has no significance in determining the retention time of the species studied. It is proposed that pH effects the elution characteristics only by effecting the overall ionic strength of the eluent; in the systems studied in this paper, the effect of the OH- ion on the total ionic strength was minimal and therefore no pH dependent behavior was observed.
LITERATURE CITED Small, H.; Stevens, T.; Bauman, W. Anal. Chem. 1975, 47, 1801. Wetzel, R. Envlron. Sci. Techno/. 1979, 73, 1214. Wetzel, R.; Anderson, C.; Schlelcher, H.; Crook, D. Anal. Chem. 1979, 57. 1532. Hansen, L.: Richter, B.: Roiiins, D.:Lamb, J.: Eutouah, D. Anal. Chem. 1979, 51, 633. Lathouse, J.; Coutant, R. I n "Ion Chromatographic Analysis of Envlronmental Pollutants"; Ann Arbor Science Publishers: Ann Arbor, MI, 1978. Gjerde, D.; Schmuckier G.; Fritz, J. J. Chromatogr. 1980, 187, 35. Pauiing, L. "The Nature of the Chemical Bond"; Corneii University Press: Ithaca, NY, 1960; p 514. Waddington, T. Adv. Inorg. Chem. 1959, 7 , 160. Peters, D.; Hayes J.; Heftje, G. "Chemical Separations and Measurements"; W. B. Saunders: Philadelphia, PA, 1974; p 583.
Dennis Jenke Montana Energy and MHD Research and Development Institute P.O. Box 3809 Butte, Montana 59702
RECEIVED for review March 23, 1981. Accepted May 7,1981.
Comparative Precision of Silver Inquartation Techniques Sir: Local assayers have recently noted a decrease in the precision of results obtained by using manufactured solid inquarts and have switched to solution inquartation, claiming vastly superior precision, The manufacturer contended that the solid inquarts gave results as good as can be obtained by fire assaying ( I ) . It had been our observation that the precision of inquartation was limited by the fire assay process and that precision equal to the precision of pipetting could not be obtained, as some assayers were claiming. The precision is of interest, since the average weight of inquarts assayed as blanks is subtracted from the bead weight to obtain the weight of silver in the assayed sample of ore. The precision thus determines the detection limit for gravimetric determination of silver. We undertook comparative inquartation studies to resolve the conflicting statements. EXPERIMENTAL SECTION To study the precision of solution inquartation, we added 5-mL aliquots of a freshly prepared solution of 400 ppm silver (as silver nitrate) in 1% (v/v) nitric acid to a series of crucibles containing weighed portions (85 g) of a sample-flux-reducing agent mixture prepared in large batches (3000 g of silver-free litharge, 150 g of borax glass, 900 g of sodium carbonate, 900 g of silica, and 150 g of flour). The portion of the mixture taken corresponded t o
a sample of 15 g of silica, using a flux very close to that recommended by Bugbee for a bisilicate slag after production of the recommended size button (2). The solution inquarted crucibles were dried in a 95 "C oven for several hours and at 105 "C overnight. To study the precision of the solid inquarts, we obtained new vials of inquarts from the manufacturer, older vials from local assayers, and very old boxes (in some cases more than 10 years past manufacture) obtained from forgotten storeroom supplies. Studies were performed on unbiased splits of 10 inquarts obtained by reduction of sample size by repeated employment of a Soiltest 10-in. splitter starting with the original box or vial of 1000 pieces. These inquarts were weighed and subjected to either of two studies, fusion and cupellation or dissolution and atomic absorption. The fusions and cupellations were performed at 1800 OF by using a large Denver Fireclay Co. electric assay furnace. Beads were weighed on a Cahn Model 26 electronic microbalance. The atomic absorption used was a Perkin-Elmer Model 303 at lox scale expansion, equipped with recorder readout, an active 0.5-Hz filter, using large zero offset, and air-Mapp gas (3). The solutions for atomic absorption were prepared in 125 mL of 20% (v/v) nitric acid, The acid was dispensed with a measured precision of 10.035% relative standard deviation, equivalent to 10.0007 mg silver. A final phase of the study used 1000-pL spikes of silver nitrate solutions of appropriate concentration in addition to either solid
0003-2700/81/0353-1536$01.25/00 1981 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 53, NO. 9, AUGUST 1981
1537
Table I. Results of Atomic Absorption Analyses of Sets of Ten Inquarts (Mean f Standard Deviation) manufacturer lot no.
M M351 M351 M2151 M5666 M5666 C0688
our bottle no.
inquart wt, mg
silver wt, mg
silver concn, %
JF R S M RM
101.77i 1.10 103.50i: 1.07 103.47r 01.78 99.60f 2.31 100.70i 1.16 99.51 f 5.16 93.13 f 2.53 91.16 f 4.50 95.81 i 1.94 97.19 f 1.78 100.04 f 0.41 100.64?; 0.26 100.31 f 0.73 100.90 f 0.37 100.38 f 0.27 100.01i 0.29 100.42 f 0.27 100.21 f 0.52 100.32 I 0.44 100.47 f 0.15 d.n.a. d.n.a.
2.061 f 0.028 2.206 f 0.023 2.192 ?: 0.067 2.221 f 0.057 2.141 f 0.037 2.215 f 0.132 1.429 ?: 0.489 1.382f 0.444 1.951 i: 0.052 1.976 i- 0.043 2.050 f 0.016 2.052 f 0.014 2.052 i: 0.016 2.052 i. 0.015 2.053 f 0.012 2.049i. 0.008 2.046 ?: 0.010 2.045 f 0.016 2.049 f 0.013 2.049 r 0.018 1.990?; 0.007 2.211 ?: 0.004
2.021 f 0.069 2.132i: 0.018 2.120f 0.067 2.230f 0.027 2.126I0.035 2.159 f 0.024 1.455 f 0.682 1.525f 0.511 2.036 i. 0.030 2.034% 0.035 2.049 f 0.016 2.042 f 0.014 2.046 f 0.020 2.034f 0.019 2.045f 0.011 2.048f 0.012 2.037 f 0.010 2.041 f 0.015 2.042 f 0.011 2.039 f 0.018 d.n.a. d.n.a.
A
K
C0688 C0688 040316
X Y 1 2 3
4 5 6 7 8 9 10
solution solution
Table 11. Results of Fiire Assay of Sets of Ten Inquarts (Mean f Standard Deviation) manufacturer lot no.
our bottle no.
inquart wt, mg
C0688
K
C0688
X
92.85 f 5.83 90.63 f 6.06 94.71 ?: 1.60 96.75 i: 1.89
or solution inquarts. All spiked samples were dried as described above for the solution inquarts.
silver wt, mg
silver concn, %
1.419 f 1.431 t 1.822f 1.909f
1.429 f 1.596 f 1.929 f 1.964 f
I
0.620 0.478 0.034 0.061
Table 111. Comparison of Fire Assay of Solid Inquarts and Solution Inquarts (Mean f Standard Deviation)
RESULTS The results of the atomic absorption study are shown in Table I in approximate1.y the order manufactured. Sporadic evidence of historical problems with control of inquarts weights is evident from comparison of the,newest lot (no. 040316,which showed a relative standard deviation of weight of the inquart itself of 10.1% to 10.7%) with the earlier manufactured lots. The earliest lots were found to have typical weight variations of approximately 11% or 2% rsd. However, one vial of each of two more recent lots showed relative standard deviations of as much as 15%. In one case, the concentration of silver in the inquart (obtained by dividing individual silver bead weight by the individual inquart weight before calculation of the mean and standard deviation) showed a relative standard deviation as small as expected, based on the experience with earlier lots. In another case, however, the relative standard deviation of silver concentration was approximately 140% This was ascribed to a second problem with manufacture, error^ in control of the homogeneity of the nominally 2% silver-lead alloy. These results were confirmed by our preliminary fire assay study, the results of which are shown in Table 11. That the error is not limited by the atomic absorption can be ascertained by comparison with the results for the solutions a t the bottom of Table I. In Table I11 are results of fire assays of sets of both solution in quarts and the newest manufactured lot of solid inquarts. The small sample size experiments (n = 3 and n = 5) were undertaken to illustrate the apparent precision which would be measured if commoii assay practice {a large number of samples and a small number of blank inquarts) were followed. The five sets of 10 were taken to allow more detailed statistical analysis. None of these experiments were paired. The results
0.429 0.397 0.031 0.031
n
10
a
silver bead weight, mg solida solution
1.899 ?: 0.017 1.910f 0.028 1.895f 0.023 1.904?: 0.019 1.905f 0.027 5 1.883 f 0.023 1.890 ?; 0.024 1.913f 0.030 3 1.898i 0.003 1.888 i 0.016 1.882f 0.015 1.887 c 0.017 1.902f 0.030 Lot 040316 bottle 1.
1.901 f 0.017 1.908 f 0.031 1.914 f 0.037 1.927+- 0.022 1.916 f 0.019 1.889 ?: 0.034 1.888f 0.020 1.873 f 0.029 1.908f 0.016 1.905i 0.006 1.917 t 0.049 1.898 i 0.043 1.893 % 0.031
of assays of five sets of 10 inquarts were found to have homogeneous variances and means at the 95% level of confidence when examined by using Bartletts x2 and the F test, respectively ( 4 , 5 ) . The grand standard deviation for these five sets of 10 each (45 degrees of freedom) was found to be 10.022 mg for the solid inquarts and 10.027for the solution inquarted samples. Overall, the standard deviations (67 degrees of freedom) for the data in Table I11 were found to be 10.022 for solid inquart9 and 10.028mg for solution inquartation afbr x2 and F testing revealed homogeneity throughout. These values are indistinguishable from one another at the 95% level. In Table IV are results of spiking both solution and solid inquarted sample-flux-reducing agent mixtures. The experiments with the lower level spikes were repeated. All four experiments were separate and unpaired. The data were
1538
ANALYTICAL CHEMISTRY, VOL. 53, NO. 9, AUGUST 1981
Table IV. Experimental Verification of Detection Limits, Using Spiked Samples amt of silver added, mg
solid inauarts
solution inquarts
no. assayed
mean, mg
range, mg
blank
detected
mean, mg
range, mg
blank
detected
5 5
1.883 1.879
0.056 0.039
0.076 0.080
C
2 2
1.908 1.868
0.002 0.047
1.8'75 1.908
0.031 0.033
C
0.010
2 2
1.928 1.939
0.071 0.011
1.916 1.912
0.055 0.050
C
0.025
2 2
1.924 1.918
0.009 0.032
1.925 1.934
0.004 0.059
C
0.050
2 2
1.915 1.947
0.016 0.028
1.987 1.950
0.021 0.025
C
d.n.a. d.n.a. N N N N N N N
0.075
2 L
0.093 0.041
1.980 1.988
0.009 0.021
C
n
1.860 2.001
d.n.a. d.n.a. N N N N N N N N N
1.878 1.888
0.000
a b a b a b a
0.100
2 2
1.981 2.025
0.021 0.013
1.998 2.020
0.019 0.002
C
0.250
2
2.155
0.042
2.162
0.087
C
0.500
2
2.371
0.031
2.346
0.003
C
0.750
2
2.586
0.036
2.586
0.024
C
1.000
2
2.848
0.026
2.838
0.041
C
blank
b
a b a b a b a a a a
Y Y Y Y Y Y Y
d d d
d d d d
N N N Y Y Y Y Y Y _ _ _ I _
tested by using the substitute t test employing ranges and means (6). The appropriate mean from the applicable set of five blanks (e.g., blank a) was compared to that from the pair of spiked samples run with the blanks (e.g., those means labeled a in each case), to see if they were the same (the null hypothesis) or whether there was a significant difference between them (the silver was detected). The columns headed "detected" summarize the results of these substitute t tests a t the 0.05 level of significance.
ACKNOWLEDGMENT We are indebted to Don Book of the College of Agriculture, University of Nevada, Reno, for advice on statistics, to Robert De Palma of Bico-Braun International for supplying inquarts from the new lot and for candid conversations, and to Mary Kilby, of Rocky Mountain Geochemical Corp., and to Dan Kappes, Registered Professional Engineer, both of whom supplied some of the older inquarts used.
DISCUSSION The historical evidence of problems with control of both the weights and homogeneity of composition of solid inquarts shows that examination of each new vial and lot is warranted, before assuming that the inquarts can be used as received, as has been common assay practice. The results of the work support our observation that the standard deviation of solution inquartation was substantially larger at f0.028 mg than the error of iO.0009 mg which the pipetting error (i0.043% relative) would predict for the 2-mg beads, suggesting control by factors inherent in the fusion and cupellation steps. The comparison of solid and solution inquarts supports the contention of the manufacturer that solid inquarts give results as good as can be obtained by fire assaying. The use of inquarts of current manufacture (which we found to give i0.022 mg standard deviation) and solution inquartation (i0.028 mg) gave statistically indistinguishable results. Experiments with spiked samples showed that estimates of detection limits in routine assay (duplicate samples and five blanks) are such that results are best reported positive only when above 0.1 mg. Below this level, direct instrumental determination of silver by atomic absorption, perhaps incorporating fire assay preconcentration with arrested cupellation or using other precious metals (such as gold, platinum, or palladium (7) as the inquart material. (See, for example, ref 8.)
LITERATURE CITED de Palma, R., Bico-Braun International, Burbank, CA, private communicatlon. Bugbee, E. "A Textbook of Fire Assaying"; Wlley: New York, 1922; p 163. Mansell, R. At, Absorpt. News/. 1967, 6, 6-8. Ostle, H. "Statlstics In Research", 2nd ed.; The Iowa State University Press: Ames, IA, 1963; p 136. Crow, E.; Davis, F.; Maxfield, M. "Statistics Manual"; Dover Publlcatlons: New York, 1960; p 124. Bauer, E. "A Statistical Manual for Chemists", 2nd ed.; Academic Press: New York, 1971; p 69. Beaulieu, P. L. "Palladium Inquarts in the Classical Gravlrnetric Fire Assay and Fire Assay Collection-Atomic Absorption"; in progress. Haffty, J.; Riley, L.; Goss, W. Geol. Surv. Bull. ( U . S . ) 1977, No. 1445. 'Current address: Amoco Research Center, P.O. Box 400, Naperville, IL
60540.
Patrick L. Beaulieu* Brenda A. Keller Tamela R. Wortman Nevada Mining Analytical Laboratory Mackay School of Mines University of Nevada, Reno Reno, Nevada 89557 RECEIVED for review January 16,1981. Resubmitted March 20, 1981. Accepted March 20, 1981.
