failed to destroy the organic material completely; the 10-min furnace treatment described above was successful. Initial results with base metal samples, salted with gold, indicated low recovery values; for instance, gold salted with 65 g ferric chloride, 7 g nickelous chloride, and 5 g cupric chloride in a liter of solution gave average recoveries of 85% of gold. Subsequent experiments with individual base metal samples showed a definite dependence of gold recovery on the base metal chloride concentration in the influent; as shown in Table I, 0 to 0.05M base metal solutions resulted in complete recoveries, higher concentrations resulted in gradually larger losses of gold on the cation exchanger. This single factor has been responsible for all previous failures to separate gold quantitatively by cation exchange. The nature of this anomalous adsorption of anionic gold species is uncertain. Wang et a / . ( 7 ) believed that the mechanism involved the basic properties of the oxygen atoms of the neutral sulfo group and that “by addition of proton or metal ions, the functional groups of the resin acquire a positive charge.” Boyd, Lindenbaum, and Larson ( 5 ) concluded that a similar absorption of Fe(II1) resulted from the formation of ion association complexes which were “salted” into the cation exchanger phase “by essentially a solvent extraction mechanism.” This behavior is currently being studied further.
Table I1 shows results obtained for various quantities of gold separated from 25-g samples of base metal chlorides. By controlling the influent concentration, essentially complete recoveries of gold are obtained. The wet extraction procedure outlined above was used to determine the gold content of four ores. The values are listed in Table I11 and are similar to those reported for fire assay and other wet methods (2). Large deviations may be expected due to lack of homogeneity in the ores and the relatively small sample size. In summary, the successful separation of gold by cation exchange is a very useful tool for the wet determination in gold ores. It provides an alternative to fire assay, especially if the assay values are in doubt; certain pitfalls such as unknown ore composition and the nature of the noble metal minerals present would be eliminated. Finally, the procedure enables gold to be incorporated into the iron-coppernickel collection of platinum metals ( I I ) . Until the present time, gold collection by this method has not been investigated because of substantial losses of gold in the cation exchange separation. RECEIVED for review January 24, 1969. Accepted March 24, 1969. ~
~~
~~~
(11) K. C. Agrawal and F. E. Bearnish, Cun. Met. Quart., 3(4), 285 (1964).
Quantitative Analysis for Trace Chromium in Ferrous Alloys by Electron Capture Gas Chromatography W. D. Ross Monsanto Research Corporation, Dayton, Ohio 45407
R. E. Sievers Aerospace Research Laboratories, Wright-Patterson Air Force Base, Ohio 45433
THEANALYSIS for metals by gas chromatography can be performed on many types of samples through a wide range of concentration from macro to ultra trace quantities. The advantages of this technique over more conventional chemical or instrumental methods are the sensitivity, selectivity, speed, and relatively inexpensive equipment involved. This technique utilizes the conversion of metals to volatile metal chelates and subsequent analysis by gas chromatography. The general approach was described in a recent book on the subject ( I ) . Previously reported quantitative analytical methods (2-5) have used acid digestion and solvent extraction techniques for the conversion of the metals to volatile metal chelates. In the present study, we have utilized a direct reaction technique (6, 7) which is considerably shorter and more convenient ( I ) R. W. Moshier and R. E. Sievers, “Gas Chromatography of Metal Chelates,” Pergarnon Press, Oxford, 1965. (2) W. D. Ross and R. E. Sievers, Tuluntu, 15, 87 (1968). (3) R. W. Moshier and J. E. Schwarberg, ibid.. 13, 445 (1966). (4) G. P. Morie and T. S . Sweet, ANAL.CHEM., 37, 1552 (1965). (5) Ihid.,34, 314 (1966). ( 6 ) R . E. Sievers. J. W. Connolly. and W. D. Ross, J. Gus Chromutog.. 5, 241 (1967). (7) 0. Kamrnori, K. Sato, K. Takimoto, and K. Arakawa, Bumeki Kuguku, 15, 561 (1966).
and which eliminates the acid digestion of the alloy and solvent extraction steps. We have also introduced a new method of accelerating reaction between the ligand and the metal, viz., the use of microwave radio frequency (RF) energy, which stimulates reaction at the ligand-metal interface. This approach offers the advantage of accelerating the reaction without overheating the major part of the unreacted ligand, thereby reducing the likelihood of thermal decomposition of the ligand. It was further expected that delivery of the energy could be more efficiently controlled when R F energy was used. The feasibility of these new approaches was demonstrated by developing a technique for the determination of minute amounts of chromium in two ferrous alloys (NBS standards 170A and 106b) containing 0.014 and l.lSyo chromium, respectively. EXPERIMENTAL
This research consisted of three experimental parts: analysis for 1 . 1 8 x chromium in NBS 106b by reacting the alloy and trifluoroacetylacetone [H(tfa)l in an R F field ; analysis for 0.014x chromium in NBS 170A by reacting with H(tfa) in an R F field; and analysis for chromium in NBS 170A by reacting the alloy with H(tfa) using conventional heating (heating mantle). VOL. 41, NO. 8, JULY 1969
1109
Keagents. Standard samples used in these analyses were: NRS 10613 steel (Nitralloy G ) containing 1.18% Cr, 1.07% Al, 0.326% C, 0.5OGZ Mn, 0.008% P, 0.016% S, 0.274% Si, 0.117% Cu, 0.217% Ni, 0.003% V, and 0.199% Mo; and, 170.4 (Basic Open Hearth Steel) containing 0.014% Cr, 0.046% Al, 0.052% C, 0.325% Mn, 0.005% P, 0.021% S, 0.036% Si, 0.059% Cu, 0.026% Ni, 0.009% V, 0.005% Mo, 0.037% Zr, 0.281% Ti, 0.006% Sn,and0.005% N. The l,l,ltrifluoro-2,4-pentanedione [H(tfa)] reagent (Pierce Chemical Co., Rockford, Ill.) was redistilled in our laboratories and stored in Teflon bottles at 0 “C until used. Instrumentation. The radio frequency generator used was a Diathermy Model CMD-IO microwave generator, manufactured by Raytheon. with an output of 110 W at a frequency of 2450 Mc. A Barber-Colman Model 20 gas chromatograph with modifications described earlier ( 2 ) was used in the electron capture mode of operation. The following instrument operating conditions were used in the analyses: column-Teflon, 7 ft X 0.06-inch i d . , packed with 15% SE-52 on Anakrom ABS; column temp-150 “C; electrometer gain-5 X 1000, 5 X 300; detector voltage-20 V ; Nz carrier26 psig; scavenger flow-120 ml/min; column flow-33 ml/ min; detector temp-200 “C; injection port temp-175 “C. Preparation of Standard Chromium(II1) Trifluoroacetylacetonate Solutions. The standard solutions of Cr(II1) trifluoroacetylacetonate [Cr(tfa)J were made by weighing 0.005064 g of this compound on a Cahn electrobalance and dissolving the sample in 100 ml of Mallinckrodt nanograde benzene, producing a solution of 5.06 X g/ml. This solution was further diluted 1 :10. The actual concentration of Cr present in this solution was 5.15 X lo-’ g/ml. Sample sizes of 1-pl, containing 5.15 X 1O-Io g of Cr, were injected into the chromatograph for calibration in the determination of Cr in NBS 106b. This amount of sample produced a peak height of 140 mm with the electrometer gain at 1500. The standard solution of Cr(tfa)3 used in the analysis for chromium in NBS 170A reacted in the R F field was 5.15 X lo-’ g/ml Cr diluted 1:lO; consequently. 1 p1 contained 5.15 X 10-11 g. One microliter of this standard solution produced a peak height of 50 mm with an electrometer gain of 5000. A standard solution containing 9.28 X lo-’ g of Cr per ml of benzene was used in the analysis of NBS 170A reacted with H(tfa) in a heating mantle. Procedure. In the analysis of NBS 170A, 2 to 4 mg of the sample (average size of a single turning particle, if available a much smaller particle can also be used) was weighed on a Cahn electrobalance. The sample was transferred to a 2-ml borosilicate glass test tube and approximately 0.3 ml of H(tfa) was added. [The volume of H(tfa) is not critical, except that there must be a surplus amount of ligand to react quantitatively with the alloy.] One drop (ca. 0.05 g) of 35$ H N 0 3 was added. The test tube was placed in the radio frequency field l/r-inch from the antenna. The generator was adjusted to 40$ of full power until the solution began bubbling and became red. To prevent the reaction from becoming too violent, the generator was then turned back to 15% power. The bubbling stopped after about 5 minutes when the reaction was almost complete. The heating was continued 15 minutes to ensure quantitative reaction. The deep red solution was carefully dissolved in benzene and then transferred to a 10-ml volumetric flask. The reaction tube was carefully rinsed with benzene and the washings were also added to the flask. Occasionally a very small amount of grey residue was noticed remaining on the walls of the test tube or suspended in the drop of acid (which should remain in the reaction flask), but this did not affect the quantitative determination of this alloy. The volumetric flask was filled to the mark with Mallinckrodt nanograde quality benzene. A 1-ml aliquot of this solution was transferred to a 10-ml vial. NaOH solution (3 ml, 0.15M) was added to the vial. The vial was shaken vigorously for 2 minutes, resulting in a red flocculant precipitate. The organic layer was separated 1110
*
ANALYTICAL CHEMISTRY
by centrifugation with the precipitate suspended between the aqueous and organic layers. (Another washing should be performed if the organic layer is not clear and colorless.) The organic phase, which was withdrawn carefully with a medicine dropper, was then placed in a clean vial for sampling. Five 1-ml aliquots from the organic layer of the “unknown” were injected to obtain statistical peak height averages. The amount of chromium in the “unknown” was determined from ralibration curves prepared from analyses of standard solutions of Cr(tfa)s. (If desired, an internal standard can be used to reduce errors caused by variability in sample introduction and in detector sensitivity; however, our experience has shown that this is not necessary.) Log-log calibration curves (peak height DS. amount Cr present) were linear over two orders of magnitude (1O-Io and lo-” g of Cr). New curves should be drawn each day from fresh standard solutions for most accurate results. The analytical procedure for Cr in high carbon steels-e.g., NBS 106b-is slightly different because small, yet significant amounts of carbon and other occluded elements remain undissolved. In the dissolution step, an appreciable quantity of grey residue adhered to the inside walls of the test tube or was suspended in the solution. This residue contained about 80$ C ; the rest consisted of metals found in the original alloy. An analysis without digesting this residue showed only ca. 80% recovery of the total Cr. An emission analysis of the residue revealed that the rest of the Cr remained in the undissolved material. Therefore, a residue digestion step must be added to the procedure when large percentages of C are present. For the analysis of a high carbon steel (such as 106b), following the reaction with H(tfa), the benzene-soluble reaction products were transferred to a 100-ml volumetric flask, taking care not to transfer any of the residue. Because the Cr content is so high in this alloy, the reaction products were diluted to this volume to avoid detector overload. If a smaller particle is available for analysis, dilution is unnecessary. The residue was separated from the benzene by centrifugation. The remaining residue in the reaction tube was dried by placing the tube in the R F field at 40% power for 5-10 minutes. Six drops (ca. 0.3 g) of 70% H N 0 3was added to the residue; then the RF unit was again set at 40% power to permit slow refluxing. Any remaining residue can be scraped off the sides of the vessel with the sealed end of a melting point capillary or a small glass rod. It is essential that all of the residue be dissolved. After ca. 15 minutes, the solution became clear and amber colored. The acid solution was taken to dryness by heating with 40% R F power. The drying process was accelerated by withdrawing the vapor with a medicine dropper (with an elongated tip placed just above the surface of the refluxing liquid in the tube) attached to a filter pump. An orange to yellow solid resulted. About 0.3 g of fresh H(tfa) and one drop of 35% ”01 were added and allowed to react with the solid for 15 minutes while heating at 40% power. The solution became light red. (If any unreacted solid remains, it may be necessary to scrape it loose from the sides or bottom of the tube and permit a longer reaction time.) The solution was added to the previously transferred material in the 100-ml volumetric flask and diluted with benzene to the 100-ml mark. The washing procedure and the chromatographic analyses are identical to the 170A alloy described earlier, except that because of the greater concentration of Cr, an electrometer gain of 1500 was employed and a more concentrated standard was used (5.15 X g Cr/pl). In the experiment to determine if the quantitative data could be recovered by reacting the ferrous alloy (NBS 170A) with H(tfa) using a mantle for a heat source, all steps were identical to the R F experiment except the heating procedure. The reaction was performed by placing the reaction test tube in a small 25-W heating mantle and heating to 130 “C for 30 minutes to ensure quantitative reaction of the alloy. This ex-
Table I. Analysis for Chromium in Ferrous Alloy NBS Standard 170A (Reacted in R F Field) Cr = 0.014%
Amount of chromium, g Wt of NBS sample, mg 2.208 4.958 3,280 3.989 2.991
Total, in NBS sample, x 10-7 3.09 6.94 4.59 5.58 4.19
Injected, x 10-11 3.09 6.94 4.59 5.58 4.19
Detd by GC, X 10-l1
Cr detd in NBS sample, 0.013 0.014 0.014 0.016 0.014 Mean 0.014
2.85 6.70 4.55 6.30 4.10
Deviation -0.001
z
Deviation, 7.1
0.000 0.000 0.002 0.000 Mean error 0.0002
Re1 error 1.4
Deviation -0.04 -0.03 0.09 -0.05 -0.07 Mean error -0.02
Deviation, 3.4 2.5 7.6 4.2 5.9 Re1 error 1.7
0 0
14.2 0
Table 11. Analysis for Chromium in Steel NBS Standard 106b Cr = 1.187,
Amount of chromium, g Wt of NBS sample, mg 2.512 3,464 3.318 3.158 3 I445
Total, in NBS sample x 10-5 2.96 4.09 3.92 3.73 4.06
Injected, X 10-lo
2.96 4.09 3.92 3.73 4.06
Detd by GC, X 10-lo 2.86 3.98 4.22 3.57 3.83
Cr detd in NBS sample, 1.14 1.15 1.27 1.13 1.11 Mean 1.16
z
Table 111. Analysis for Chromium in Ferrous Alloy NBS Standard 170A (Reaction in Heating Mantle) Cr
Wt of
NBS sample, mg 4.292 3,078 3.628 3.794 4.042
=
Amount of chromium, g Total, in NBS sample, Injected, Detd by x 10-7 x 10-11 GC, X 5.47 6.00 6.00 4.31 4.31 4.55 5.08 5.08 5.26 5.31 5.31 5.34 5.66 5.66 5.66
periment was also used to determine the stability of very dilute standard Cr(tfa), solutions (lo-' g/ml) and to ascertain whether there were significant day-to-day variations in instrument sensitivity. A reagent blank should be run prior to analysis of unknowns. A very small peak for Cr(tfa), (3-5 mm) appeared when the blank was analyzed and this peak height was subtracted in subsequent analysis of the sample. RESULTS AND DISCUSSION
NBS samples of ferrous alloys have been used to demonstrate the feasibility of analyzing for trace amounts of C r by gas chromatography. Tables I through 111 illustrate the results of analyses of samples prepared by direct reaction of the ligand with the alloys. The reactions by which the samples are dissolved are depicted by the following representative equations:
+ H(tfa) F e + H(tfa) Cr
0.0147G
-
+
+ H2 Fe(tfa), + HB Cr(tfa),
(1)
(2)
Cr detd in NBS sample, 0.013 0.015 0.015 0.014 0.014 Mean 0.014
Deviation -0.001
0.001 0.001 0.000
0.000 Mean error 0.0002
Deviation, 7.1 7.1 7.1 0 0 Re1 error 1.4
In these and other similar reactions (6, 7), the ligand functions both as an oxidizing agent and solvent. The addition of catalytic amounts of a n inorganic acid greatly accelerates the rate of reaction between the ligand and the metal. The particular acid used is important. Hydrochloric acid inhibits the direct reaction of iron or ferrous alloys with H(tfa), while nitric acid accelerates the reaction. The converse is true for pure powdered metallic chromium. However, the chromium present in low amounts in the ferrous alloys studied reacts quantitatively in the presence of nitric acid, which is needed for the rapid dissolution of the principal constituent. Table I shows the analysis of a low-carbon steel which utilizes a very rapid and simple dissolution procedure. Chromium was present at the 140-ppm level, and it was found that the electron capture detector provided more than adequate sensitivity and excellent agreement with the analytical results of independent wet methods certified by the NBS. Table I1 shows that a high carbon steel can be analyzed with equal accuracy but a n additional digestion step is required VOL. 41, NO. 8,JULY 1969
1111
P
J
U
Cr(tfa)3
1 ' 1 " " ' 1 1 ' 1 1 ' 1 1 1 1 1 ' 4 8 12 16 MINUTES
0
0
3 6 MINUTES
9
Figure 1. Microanalysis of reaction products of basic hearth steel with trifluoroacet ylacetone Sample injected contained 5.66 X lo-" g of Cr (col temp 150 "C) Sample was washed with 1M NaOH owing to the failure of the sample t o dissolve quantitatively in H(tfa). Data in Tables I and I1 were obtained on samples for which a n R F field was used t o accelerate the dissolution reaction. Conventional heat sources-e.g., a heating mantle-can also be used to accelerate dissolution with some sacrifice in convenience. Table I1 illustrates analytical data obtained following dissolution of samples heated by a mantle. The results compare quite favorably with those listed in Table I. The washing step must remove the excess H(tfa) and the Fe(tfa)a without destroying any Cr(tfa)~. The optimum concentration, the time of washing, and the ratio of washing solution to reaction solution must be carefully controlled to adequately remove the interferences. Being able to selectively eliminate or retain certain metal chelates by varying the washing procedure may be important in extensions of this method to other metals. Figures 1 and 2 illustrate this point. The reaction products shown in the chromatogram of Figure 1 were washed with 1.OM NaOH showing the removal of H(tfa), Fe(tfa)r, and most of the Al(tfa)a; whereas, a 0.15M solution of NaOH was used t o wash the reaction products shown in Figure 2. This figure shows quite adequate chromatographic separation of Al(tfa), and Cr(tfa)a (column temp 120 "C) and illustrates the complete removal of Fe(tfa)a and H(tfa). The use of these two NBS standards has proved that a wide
1112
ANALYTICAL CHEMISTRY
Figure 2. Reaction products of basic open hearth steel containing 0.014% Cr and 0.046% AI (col temp 120OC) Sample was washed with 0.15M NaOH range of chromium concentrations can be analyzed using this technique. The principal advantage of the direct reaction method is that extremely small initial sample sizes can be analyzed for trace metals. The actual manipulative time involved for the analysis of low carbon alloys is ca. 15 minutes. However, the full potential of the technique has not yet been exploited. Only 5-10 pl of sample are actually needed for chromatographic analysis. More than 99.9% of the sample is discarded in the present procedure; therefore, at least two orders of magnitude lower Cr concentration could be determined if the reaction products were diluted only t o 0.1 ml. The limiting factor of the technique is the manipulation of small volumes during the washing operation. Further miniaturization of the manipulative procedure or evaporative concentration of the benzene solution prior to chromatographic analysis will produce even greater effective sensitivities. The simplicity of the direct dissolution of the sample in a fluorocarbon beta-diketone leads us to suggest this technique for preparing samples for entirely different forms of instrumental analysis. In particular, it should be useful in atomic absorption where it is well known that sensitivity can often be enhanced by making measurements on organic rather than aqueous solutions. In the proposed method prior dissolution in aqueous acids would be unnecessary and contamination by blanks from this source would be avoided. RECEIVED for review March 14,1969. Accepted May 9,1969. Work presented in part at the 156th ACS meeting, Atlantic City, September 1968. Research was sponsored by the Aerospace Research Laboratories, Office of Aerospace Research, United States Air Force, Contract No. A F 33(615)1176.
