Spectrographic Determinatio n of PaIIadium, PIatinum, Iridium, and Rhodium Porous Cup Technique GILBERT H. AYRES AND EUGENE W. BERG The University of Texas, Austin, Tex.
For use in a study of the sharpness of separation of palladium from the other platinum metals, by dimethylglyoxime precipitation, a rapid and specific method was required. A spectrographic method for determination of palladium and for simultaneous determination of platinum, iridium, and rhodium in solution has been developed, utilizing a high voltage spark source and the porous cup electrode technique. The platinum metal chlorides were used in hydrochloric acid solution. Cobalt(I1) was added as the
internal standard, and the intensity ratios Pd 3242.71 Co 3354.4, Pt 2830.3/Co 3354.4, Ir 3220.7/Co 3354.4, and Rh 3323.1/Co 3354.4 were calculated. These line pairs gave normal spectrographic working curves. The working curve for one element is practically independent of variations in the concentration of the other platinum metals present. The w-orking range for palladium is about 5 to 160 p.p.m. The average relative analysis error was 2.8% for samples with widely varying ratios of the platinum metals.
T
HE quantitative determination of the platinum metals has
centration were needed; the great sensitivity of spectrographic methods, coupled with their feature of being rapid and specific, offered a possible approach to the solution of the problem. Although poor accuracy is commonly regarded as characteristic of spectrographic methods, such was not found to be the case in the work reported here; a relative analysys error of 2.8% compares favorably with other methods of analysis a t such low concentrations. Absorption spectrophotometric methods, while capable of good accuracy, lose their advantage for the purpose of this study because they are not specific. The porous cup electrode technique, reported by Feldman ( 1 ) for the spectrographic analysis of solutions, appeared to be well adapted to the study proposed. Some of the advantages it offered over other spectrographic techniques were the ease of obtaining a homogeneous sample and the simplicity of adding an internal standard; sampling was also greatly facilitated, i n that aliquots could be taken of concentrated solutions, or if the sample were too dilute. it could be concentrated by evaporation. In planning a detailed study of the separation of palladium, platinum, iridium, and rhodium from each other by precipitation methods, it was decided t o investigate, as the first portion of the work, the sharpness of the separation of palladium from the other platinum metals by precipitation of palladium with dimethylglyoxime in hydrochloric acid solution (4). At first thought, the procedure would appear to be an analysis of the precipitate (after its dissolution by destruction of the organic matter) not only for the recovery of the palladium taken, but also for the amounts of the other platinum metals that might be carried dovn with the palladium dimethylglyosimate. As it turned out in the preliminary experiments, however, the sensitivity of the spectrographir determination of the other platinum metals was not great enough to permit the determination of trace amounts of these metals. Therefore, this approach would not be satisfactory unless the palladium precipitate were contaminated with rather large amounts of the other metals, or unless the solution of a large amount of the precipitate were extensively concentrated by evaporation. The material reported in this paper is therefore centered about the proposed analysis of the palladium precipitate for a comparison of the amount of palladium found with the amount taken, and the analysis of the filtrate for platinum, iridium, and rhodium for a similar comparison.
been considered among the more difficult analytical proced u r a As generally carried out, the analysis has been a tedious process requiring careful attention to details. Outstanding among separations of the platinum metals from each other and their detei niination has been the work of Gilchrist and Wichers ( 8 ) . Osnuum and ruthenium were quantitatively removed by distillation of their volatile tetroxides, followed by precipitation of the hydious oxides which were ignited and weighed as the metals. Of the remaining four platinum metals, palladium, rhodium, and iridium were separated from platinum by precipitation of their hydrous oxides. The platinum was precipitated as the sulfide, ignited, and weighed as metal. After the hydrous oxides of palladium, rhodium, and iridium had been dissolved, the palladium n as precipitated Rith dimethylglyouime, and the precipitate either weighed as such, or ignited and weighed as the metal. Rhodium was next removed by selective reduction to the metal, and subsequent redissolution and precipitation as sulfide, which w i s ignited to the metal. Iridium was precipitated as the hydrous ouide, ignited, and weighed as metal. The time-consuming techniques of gravimetric methods involve added difficulties if low concentrations are to be handled. Often one of the platinum metals must be determined in the presence of one or more of the others, and it is not always known if the separations are quantitative. K h a t appears in much gravimetric work to be precise and accurate determinations may be due only to compensating errors of incomplete precipitation of one element and coprecipitation or occlusion of the others. Several spectrographic determinations of the platinum metals have been reported; in general, they have dealt with the determination of alloy percentages or of impurities in a relatively pure metal sample. -4method for the direct determination of the platinum metals in solution would oftentimes be useful. It is not always feasible to precipitate the metals and then redissolve them; this is particularly true in the cases of rhodium and iridium, which require drastic treatment for their dissolution. Colorimetric or spectrophotometric methods that are entirely satisfactory have not yet been developed for the determination of unknown solutions with widely varying ratios of concentrations of the platinum metals. The spectrographic determinations reported in this paper are a part of a study of the separations of the platinum metals from a quantitative viewpoint. It was a t once apparent that for this purpose methods of determining the platinum metals in low con-
OPTIMUM WORKING CONDITIONS
I n order to establish the optimum excitation conditions for greatest sensitivity and reproducibility, and the most desirable
465
466
ANALYTICAL CHEMISTRY
line pair for the internal standard and the substance being analyzed, a great many spectra were made under widely varying circuit constants. An alternating current arc was fhbattempted to see if the normally high sensitivity and fair precision could be obtained. However, it was immediately evident that the heat produced in the discharge was too great and caused the solution to boil over long before a spectrum of the element in dilute solution could be recorded on the film. If the porous cup electrode technique were to be used, this left only a high-voltage spark unit source which was available. An increase in the inductance of the spark circuit to give more arclike excitation raised the limit of detection considerably. With the maximum inductance available in the unit plus the full amount of inductance in an inductance box applied to the circuit it was possible to increase the power input to the 4/3 kv.-amp. step without causing overheating of the electrode on long exposures. Even with this increase in power, the intensity of the palladium lines was found to be less than if 2/3 kv.-amp. were supplied with the minimum value of inductance in the circuit. Variation of the rotary gap setting had little effect on the sensitivity, but in the maximum position it increased the temperature of the electrode noticeably. A 180-second exposure proved necessary to give the desired sensitivity. This length of exposure is governed more by the sensitivity of the element under study than by other factors, because even a 180-second exposure resulted in a very light background. Presparking of the electrode was found necessary because of the variable porosity of the graphite rods. Although the thickness of the electrode floor was carefully controlled, the solutions often did not feed through immediately. Before the end of the prespark period the solution always fed through evenly and maintained only a thin film of solution on the surface of the electrode. The intensities of several lines of each of the platinum metals and of cobalt were read from spectra made up from standard solutions, and the intensity ratios of all possible combinations of the lines, using cobalt as the internal standard, were calculated. The selection of the Co 3354.4 line as the internal standard for all four of the platinum metals was based upon many determinatiom; the lines used for the platinum metals were Pd 3242.7, P t 2830.3, Ir 3220.7, and Rh 3323.1. Careful consideration was given to the establishment of a film calibration curve that would give a true relationship 9f the light intensity to the line density under the conditions of analysis excitation. The spectra obtained by this high voltage spark technique gave lines that were sharp and had no indication of pressure broadening. Even in the case of the Pd 3242.7 line, listed in the “hI.1.T. Wavelength Tables” as being hazy, a sharp spectral line was obtained. Because this is one of the more sensitive palladium lines, it was the one chosen for the palladium analysis. That this was a good selection was substantiated in that its use with the Co 3354.4 line as the internal standard gave a normal spectrographic working curve (45’ slope on logarithmic coordinates, when background correction was made) that was straight over a concentration range of 5 to 160 p.p.m. of palladium. A normal working curve is indicative that no self-absorption of the lines is occurring and that no other elements present are interfering with the line intensities (9, page 241). In view of the known adsorptive characteristics of the platinum metals, some concern was felt over the possible loss of the metals from solution in contact with the large surface area of the porous graphite electrode. The normal plot appears also to rule out any preferential adsorption of the platinum metals by the electrode. APPARATUS
All spectra for this work were obtained on an Applied Research Laboratories-H. W. Dietert 1.5-meter grating spectrograph having a 24,400-line-per-inch original concave grating. An
average dispersion of 6.95 A. per mm. is obtained with this instrument. The power for excitation was supplied by a high voltage spark unit. The spectra were recorded on 35-mm. Eastman spectrum analysis No. 1 film. Applied Research Laboratories film developing machine of the rocking type, and comparator-densitometer were used for the film processing and photometric measurements, respectively. REAGENTS
Palladium metal (A. D. Mackay, Inc.) was subjected to a direct current arc analysis to determine the presence of any impurities. Only the most sensitive lines of platinum silver, and copper could be detected, indicating only a trace of these elements as impurities. One gram of the metal was dissolved in aqua regia and evaporated to sirupy consistency. The solution was treated with concentrated hydrochloric acid and evaporated again. This process was repeated and the palladium( 11)chloride was taken up in distilled water and diluted to 1 liter, to give a solution containing 1000 p.p.m of palladium. Platinum wire, No. 1 grade, was used to make the platinum solution. A direct current arc analysis revealed only the most sensitive lines of gold and silver as trace im urities. One gram of the metal was dissolved in aqua regia anfevaporated nearly to dryness. After repeated evaporation with concentrated hydrochloric acid, the residual liquid was taken up in distilled water and diluted to 1 liter to give a solution containing 1000 p.p.m. of platinum. Iridium(1V) chloride, stated by the supplier (A. D. Mackay, Inc.) to be of purity 99.5% or better, was subjected to direct current arc analysis, which revealed the presence of calcium, aluminum, nickel, cobalt, rhodium, palladium, silver, platinum, iron, silicon, magnesium, lead, and gold. Of the platinum metals detected, platinum appeared to be present in the largest amount and was estimated to be of the order of 0.03%. All the other elements were present in trace quantities except silicon and lead, which were estimated a t 0.1 and 1%,respectively. The iridium(IV) chloride (1.729 grams) was dissolved in 100 ml. of concentrated hydrochloric acid and diluted, with distilled water to 1 liter, to give a concentration of 1000 p.p.m. of iridium. Rhodium(II1) chloride tetrahydrate (A. D. hfackay, Inc.) showed, by direct current arc analysis, the presence of trace quantities of calcium, silver, copper, platinum, iridium, magnesium, palladium, and lead; semiquantitative estimates of the amounts of these impurities indicated the purity of the compound to be 99.5% or better. The salt (2.113 grams) was dissolved in a small amount of concentrated hydrochloric acid and diluted with distilled water to 772.7 ml., to give a solution containing 1000 p.p.m. of rhodium. Cobalt(I1) chloride hexahydrate, spectrographically free from the platinum metals, was made up into a standard stock solution containing 8000 p.p.m. of cobalt, to be used as the internal standard. A 1% dimethylglyoxime solution in 95% ethyl alcohol (precipitant for palladium) was used as a diluent in standards and samples containing platinum, iridium, and rhodium, so as to approximate closely the composition of solutions to be analyzed for these elements in the study of their separation from palladium. EXPERIMENTAL
Preparation of Electrodes. One-quarter inch graphite electrodes (National spectrographic carbons) were used in forming the porous cup electrodes. The graphite rods were cut into 2inch lengths and one end was cut flat with a facing tool in a revolving mandrel similar to that used for shaping electrodes. The faced electrodes were then placed in a jig and drilled out with a No. 30 (0.1285-inch) twist drill leaving a floor thickness of 0.025 inch. This then constituted the upper electrode used for solution analysis. The lower electrode was made from ‘/cinch graphite rod that had been sharpened in a pencil sharpener. Shaped lower electrodes were found to give a more uniform spark and more r e producible intensity ratios. Each processed porous cup electrode was subjected to a 3ampere direct current arc for 10 seconds with a 4-mm. arc gap before the electrode mas filled with solution. This prearc period gave the desired amount of porosity to the electrode for use with the type of solution under study, With this degree of porosity the hydrochloric acid solution of the sample seeped through slon~lyand was completely volatilized as it reached the electrode surface. The energy of the spark was dissipated in this volatilization and did not result in heating of the electrode. A suppression of the intensity of the cyanogen bands also occurred. The processing and subsequent treatment of the porous cup electrode ensured a 270-second exposure time without fear of puncturing the electrode. It was found that so long as,the solu-
467
V O L U M E 24, NO. 3, M A R C H 1 9 5 2 Table I.
Effect on Intensity Ratios of Varying One Concentration
Concentration, P.P.M. Rh Ir Pt 40 80 400 40 80 200 40 80 104 40 80 400 40 80 200 40 80 104 40 80 400 40 80 200 40 80 104 80 80 200 40 80 200 20 80 200 80 80 200 200 40 80 20 80 200 40 200 200 40 100 200 40 50 200
Rh 3323.1 c o 3354.4 1.23 1.24 1.24 1.27 1,25 1.26 1.27 1.26 1.25 1.98 1.24 1,oo 1.93 1.31 0.906 1.29 1.34 1.26
Intensity Ratios I r 3220 7 c o 3354 4 0 845 0.822 0.831 0,888 0.892 0.895 0.861 0.854 0.860 0.829 0.825 0.931 0.833 0.949 0.841 1.48 1.06 0.729
Pt 2 8 3 0 . 3 c o 3354 4 2.91 1.91 1.29 3.01 1.89 1.33 3.18 2.07 1.30 2.01 1.98 1.97 2.00 2.11 2.00 2.13 2 10 1.9'9
tion fed through the electrode evenly, the intensity ratios were independent of the bottom thickness of the cup. Thus the same electrode could be used for subsequent exposures of the same solution and identical intensity ratios were obtained. Exposure. -4primary slit width of 60 microns and spectrum analysis S o . 1 film were used for all exposures made. Excitation was provided by the 2/3 kv.-amp. step on an A.R.L.-Dietert high voltage spark unit with 0.045 mh. of inductance in the spark circuit. The upper electrode was filled with the sample by means of a long-nosed dropper. A magnetic shutter shielded the primary slit for the first 3 seconds of each exposure to allow time for the liquid to seep through and the spark to become uniform. A 180second exposure was made in two 90-second periods with adjustment of the lower electrode preceding each period. Refilling of the electrode was not necessary because it had a capacity capable of supplying solution for a 180-second exposure. Development. The film was developed in Eastman D-19 developer in a rocking-type developing machine that maintained a temperature of 18.5" C. New developer was diluted with used developer in a ratio of 2 to 1, to ensure a more constant development from one film to another. A 10-second 5% acetic acid stop bath was used preceding the fixing of the film in Eastman concentrated x-ray fixing solution. The film was left in the fiver only a few seconds after it had cleared. Photometry. All photometric measurements were made on an A.R.L.-Dietert comparator-densitometer. Excitation under the above-prescribed conditions resulted in a spectrum that had a light background and mas relatively uniform in the region under study. In all cases the backgrounds of the two lines were the same within several per cent transmittance and varied about an average value of SO%. This background was due to the sample, and if no correction \yere made the accuracy should not have been seriously affected (S, page 238), because only a proportionality constant was introduced. A background correction resulted only in a rotation of the curve into a position such that it became a normal spectrographic working curve. Film Calibration. Emphasis was placed on obtaining a calibration curve under excitation conditions closely approximating the analytical procedure. The same conditions of time and circuit constants were used for all exposures. For film calibration a stock iron(II1) solution made up to 10% by volume of concentrated hydrochloric acid was used, and a rotating step sector was placed in the path to relate accurately the relative intensity values of the light to the density of the lines. Calibration was made in the region 2800 to 3000 A.; the same curve could be used for all work in the region 2600 to 3400 A. (5, page 7 2 ) . This range was broad enough to include suitable lines for analysis of all the platinum metals. Study of System Variables. I n general, spectrographic methods of analysis are subject to the influence of a large number of variables. ,4 discussion of the variables due to the instrument and to the technique of exposure and development will not be attempted here A rigidly standardized procedure was adopted, and the data showed that these variables have been held to a minimum, or a t least have been maintained fairly constant, As pointed out earlier, a part of the problem a t hand was the
determination of platinum, iridium, and rhodium in hydrochloric acid solution, after the precipitation of palladium with dimethylglyoxime. The system under study therefore had, as variables, the concentrations of each of the three platinum metals, in addition to the hydrochloric acid and the organic reagent content. An acid content of 10% by volume of concentrated hydrochloric acid, and a content of 10% by volume of a 1% solution of dimethy1glyo.rinie m 95% ethyl alcohol proved to be suitable, the amounts not being critical. The system variables were thus liniited to the variation in concentrations of the three metals. The system was studied for any possible adverse effect on intensity ratios produced by changing concentrations of the constituents. This effect is one of the most common sample variables found in spectrographic analysis; one constituent in varying coneentration often shifts considerably the worhing curves of the other elements. Standard samples were prepared by taking aliquots of the stock solutions, adding cobalt(I1) as the internal standard, and concentrated hydrochloric acid, followed by dilution with distilled water to k n o m volume. Several groups of three series of standards were prepared, in each series the concentration of one metal was varied, and the others were held constant. From the spectra obtained, appropriate line intensity ratios were measured. Table I shows the values of the intensity ratios obtained; the data in Table I represent the averages of a large number of determinations and include any day-to-day and film-to-film variations. It is evident that varying the concentration of one element while holding the other two constant produced no significant effect on the intensity ratio either of the element varied or of the element whose concentration was held constant. The working curves plotted from these data were linear. After it had been established that the variation of the concen. tration of one element had no significant effect on the intensity ratios for the other two, data R-ere taken for establishing the working curves, with all three concentrations as variables. Groups of three series of standards were prepared with all three concentrations being varied within the desired concentration ranges. The concentrations and the intensity ranges are sho~vn in Table 11; these data are the averages of a large number of analyses, and include any day-to-day and film-to-film variations. The working curves established from the different series of standards were all linear, and the curves for a given element were practically coincident; any one of the series of standards listed in Table I1 could be used for establishing the working curves.
Table 11. Effect on Intensity Ratios of \-arying A11 Concentrations Concentration, P.P.31. Rh Ir Pt 400 80 200 200 40 100 100 20 50 400 20 100 200 80 50 100 40 200 400 40 50 200 20 100 100 80 200 400 40 50 200 20 100 100 80 200
Rh3323 1 co 3 3 5 4 . 4 1.88 1.33 0.963 0,954 1.90 1.34 1.28 0.919 1.88 1.26 0.965 1.86
Intensity Ratios Ir 3220 7 c o 3354 4 1.41 0.971 0.721 0.976 0.658 1.46 0.703 0.949 1.45 0 718 0.965 1 43
Pt 2830.3 c o 3354 4 3.12 2.02 1.30 3.13 1.92 1.30 3.26 2.02 1.35 3.17 1 98 1 36
The introduction of 10% by volume of 1% solution of dimethylglyoxime in 95% ethyl alcohol into the standard samples had no adverse effect on the intensity ratios and the working curves. The data of Table I11 are representative of many samples and include day-to-day and film-to-film variations. Of interest is the fact that ethyl alcohol lowers the surface tension of the solution, causing it to feed through the electrode more
ANALYTICAL CHEMISTRY
468 Table 111. Comparison of Intensity Ratios after Addition of Ethyl Alcohol Concentration, P.P.M. Rh Ir Pt E t h y l Alcohol Absent 40 50 400 20 100 200 80 200 100 E t h y l Alcohol Present 40 50 400 20 100 200 80 200 100
Table IV.
R h 3323.1 c o 3354,4
Intensity Ratios Ir 3220.7 c o 3354,4
P t 2830.3 co 3 3 5 4 , 4
1.26 0.955 1.86
0.718 0.965 1.43
3.17 1 98 1.36
1.29 1.07 1.83
0.799 1.06 1.45
3.06 2.08 1.43
II -
5001
t
If 3220.7 c o 3354.4
250
Pt 2 8 3 0 . 3 Go 3 3 5 4 . 4
/
,
I/ f I
/
1
/ /
1
/
Reproducibility of Palladium Data Intensity Ratio, P d 3242,7/Co 3354.4 0.313 1.0 . 0 5 8 0.688 10.027 1.51 f 0 . 0 3 3.25 f 0.10 6 . 5 6 f: 0.14 11.7 0.5
Pd. P.P.M. 5 10 20 40 80 160
*
IO
'OO0F--
1
1
1
I
1 1 1 1 1
I
I
I
1
I l l 1 1 1
t
dr 5 100
Figure 2. Working Curves for Platinum, Iridium, and Rhodium
ti
l
I-
o
r
t
L
acid in an amount to make the final concentration 10% by volume of concentrated acid. The standard sample solutions were run in triplicate, and the intensity ratios of Pd 3242.7/Co 3354.4 measured. Figure 1 shows the log-log plot of concentration against intensity ratio. For the working curves of platinum, iridium, and rhodium, standard solutions were prepared by taking aliquots of the stock solutions in such quantities as to vary randomly the ratio of the platinum metals to each other in the various solutions. Cobalt(11) was added to give a final concentration of 320 p.p.m., hydrochloric acid to give a final concentration of 10% by volume of concentrated acid, and 1% dimethylglyoxime in 95% ethyl alcohol to give 10% by volunie in the final solution. The working curve was prepared by making triplicate determinations of three
1.0.1
Pd 3242.7 "TENS'TY co 3354.4 Figure 1. Working Curve
rapidly. If desired, the floor thicltness of the electrodes could be increased, thus further reducing any possibility of puncturing the electrode during an exposure. From all the above data it was apparent that satisfactory working curves could be established for the determination of rhodium, iridium, and platinum in varying concentrations. Further confirmation of the validity of the calibration was furnished by the results of the analysis of 43 samples with widely varying concentrations of the three metals; the average relative analysis error was 2.8%. Preparation of Working Curves. I n preparing the standards for the working curve for palladium, aliquot portions of the stock palladium were taken so that upon dilution to volume the concentration would be in the range 5 to 160 p.p.m. of palladium; cobalt(I1) solution was added to give a final concentration of 320 p.p.m. of cobalt, and hydrochloric
Table V.
Reproducibility of Rhodium, Iridium and Platinum Data
Concentration, P.P.M. R h 3323.1 Pt c o 3354.4 Rh Ir 1.29 1 . 0 . 0 3 40 50 400 200 0.971 1 0 031 20 100 2.00 & 0 0 8 80 200 100
0.730f0.054 0 968 k O . 0 4 3 1.48 f O . 0 7
P t 2830.3 c o 3354.4 3.27 & O 1 4 2.16 = t o 04 1.44+O.Oi
Table YI. Precision R h 3223.1 Co 3354.4 1.26f0.02 1.07 1.0.02 1.78 f 0 . 0 6 1.48 1 0 . 0 3 1.33 i 0 . 0 6
I r 3220.7 Co 3354.4 1.9910.03 2.85 1 0 . 0 6 1.49 f 0 . 0 4 2.42 f O . 0 9 2.18 1 0 . 0 6
Table VII. Rhodium Added Found 38.0 36.0 26.0 26.0 70.5 72.0 49.5 50.0 40.0 40.0
Ir 3220.1 c o 3354.4
Intensity Ratio P t 2830.3 Co 3354.4 0.952&0.017 1.07 1 0 . 0 2 0.780 1 . 0 . 0 2 4 1.34 1 0 . 0 2 0.892 1 0 . 0 2 4
P d 3242.7 Co 3354.4 0.82310.042 1.25 +0.04 2.10 1.0.07 3.30 i O . 0 9 4.34 A 0 . 1 6
% .kve,.age Devn., P d 4.9 3.2 3.3 2.7 3.7
..iccuracy
Concentration, P.P.M. Iridium Platinum Added Found Added Found 182 94.0 190 96.0 332 117 320 120 110 60.2 120 58.0 252 172 240 172 212 83.0 212 80.0
Palladium Added Found 8.1 7.8 15.4 15.6 33.8 33.4 66.5 64.6 101 96.4
Rh
% Relative Error Ir Pt Pd
5.56 2.08 4.21 3.65 0 . 0 0 2.50 3 . 7 5 1 . 2 8 2 . 0 8 3.09 8 . 3 3 1 . 2 0 1.00 0.00 5 . 0 0 2 . 9 4 0.00 3.75 0.00 4 . 6 7
V O L U M E 2 4 , NO. 3, M A R C H 1 9 5 2 standard samples, and measuring the intensity ratios of Rh 3323.1/Co 3354.4, Ir 3220.7/co 3354.4, and P t 2830.3/Co 3354.4. The log-log plots of concentrations against intensity ratios are shown in Figure 2. Reproducibilityof Working Curves. The reproducibility of the standard working curve for palladium is shown in Table IV. The data represent the averages of a number of working curves constructed over a period of about 2 weeks, and include any shifts that occurred in the working curve during this time. A plot of the individual sets of data invariably produced a normal spectrographic curve when background correction was made. Graphs of the data without background correction were also made, and in every instance the result was only a rotation of the working curve from the normal 45’ slope; there was no change from the straightline relationship. On the basis of these findings the decision was made to dispense with background correction; the curves of Figures 1 and 2 are drawn from data without the application of background correction. Table V shows the reproducibility of the standard working curves for rhodium, iridium, and platinum. The data were collected over a period of several days, and include any day-to-day and film-to-film variations, and any shifts in the working curve duiing this period. The plot of the data resulted in straight-line graphs. The validity of the working curves has been firmly established
469
by a large number of determinations; three standard samples differing sufficiently in concentration are adequate for the preparation of a standard curve. Precision and Accuracy. The precision and accuracy of the method were evaluated from the working curves by running nine determinations on each of five samples (statistically, little would have been gained by increasing the number of determinations on each sample). The nine spectra for one sample were recorded on the same film. The precision of the measurements is given in Table VI; the average deviation of the intensity ratios was 2.4%. The data of Table VI1 show the accuracy of the method; the average relative analysis error was 2.8%. Further work has shown that this relative error is unaffected if only thrce determinations are made on each sample. LITERATURE CITED
Feldman, Gyrus, *kNAL. CHEY., 21, 1041 (1949). (2) Gilchrist, R . , and Wichers, E., J . Am. Chem. SOC.,57,2565 (1935). (3) Harvey, C. E., “Spectrochemical Procedures,” Glendale, Calif., ilpplied Research Laboratories, 1950. (4) Hillebrand, W. F., and Lundell, G. E. F., “Applied Inorganic Analysis,” p. 292, New York, John Wiley & Sons, 1929. (1)
RECEIVED for review July 12, 1951. Accepted November 1 5 , 1951. Investigation supported jointly by the University of Texas a n d the United States Atomic Energy Commission, under the terms of Contract No. AT(40-1)-1037.
Spectrochemical Determination of the Mineral-Element Content of Beef ARTHUR J. RIITTELDORF’ AND DONALD 0. LANDON Armour Research Foundation of Illinois Institute of Technology, Chicago 16, I l l . A large number of elements in trace concentrations are nutritionally essential. This project was undertaken in order to determine the normal trace-element content of beef and incidentally to develop a spectrochemical method that would be potentially useful for the quantitative analysis of other organic materials. Seven composite samples of beef, representing all the grades, were analyzed. Nineteen elements, ranging in concentration from 0.001 p.p.m. of chromium to 270 p.p.m. of magnesium, were detected and determined. The spectrochemical method employed a direct current arc on the ashed material. Standards were prepared using a “synthetic meat ash” as a base. These and similar data w i l l aid in study of the role of trace elements in nutrition.
D
CRING the past few years, interest has grown in the socalled trace elements and their role in the nutritional processes of plants and animals. At present, it. is known that boron, cobalt, copper, iodine, iron, magnesium, manganese, molybdenum, and zinc are required by plants and,or animals. (These supplement the elenients present in relatively large amounts: phosphorus, chlorine, calcium, potassium, sodiuni, hydrogen, nitrogen, oxygen, carbon, and sulfur. ) *is refinements in analytical procedures develop, it is possible that a number of other elcments-below the present-day limit,s of sensitivity-will be established as essential. This study was undertaken to gain a:, much information as possible about the mineral composition of beef, using the eniission spectrograph as an analytical tool. Earlier investigations, employing mainly wet-chemical analytical methods ( I , 3,5-8)>had shown that a large number of elements were present in meats, but quantitative agreements were poor in many instances (see Table VII). Present address, .Jarrell-Ash Co.. Boston 16, M a r .
.\wrIIoD
Samples were analyzed by ashing and arcing the ash in highpurity graphite electrodes. Reference standards were prepared by grindin together known concentrations of the elements to be determinef with “synthetic meat ash,” prepared from potassium, phosphorus, sodium, and magnesium salts mixed in approsimately the same ratio as found in the ashes of the actual mcats. To correct for possible volatilization losses, the standards n.ere heated in a manner almost identical with that of the samples. To correct for possible matrix errors, as it was not possible to prepare standards with the same molecular structures as the samples, all standards and samples were mixed with aluminum sulfate and graphite in the ratio 1 : 2 : 3 prior to analysis. As in most traceelement work, it was necessary to check all reagents and cheniicals carefully before analysis and to purify where contaminating elements would interfere. APPARATUS
The spectrograph used in this investigation was a l.&meter, diffraction-grating instrument manufactured by Applied Research Laboratories. 4 n .4RL Multisource provided the selfignited, 12-ampere direct current arc. Both an ARL and a Leeds and Xorthrup densitometer were used. An ,4RL dcvelop-