Docombor 15, 1943
ANALYTICAL EDITION
758
Cholak, J., Ibid., 7, 287 (1935). Cholak, J., and Bambsch, K., Ibid., 13,683 (1941). Cholak, J., and Story,R. V., J . Optical Soc. Am., 31, 730 (1941). Cowling, H., and Miller, E. J., IN’D.ENO.CHEM.,ANAL.ED., 13,
precipitator samples of foundry atmospheres and for the analysis of street dusts. They should be readily adaptable to the analysis of pharmaceuticals and soil samples. Either method can be used conveniently over a wide range of concentrations, but the great sensitivity of the colorimetric method makes it the one of choice when the quantity of zinc present is known t o be very small. Relatively large amounts can be handled more conveniently by polarographic means. The latter method is also more convenient when only occasional samples need be handled and is superior when cadmium is known to be present. .Another point in favor of the polarographic method is its superiority whenever it may be desirable to determine lead, bismuth, and cadmium as well as zinc in the material under investigation. In such cases carbamate should not be added in the initial extraction step.
145 (1941).
Deckert, W., 2. o d . Chem., 100, 386 (1935). Feicht, F. L., Schrenk, H. H., and Brown, C. E., U.9. Dept. Interior, Bureau of Mines, Rept. Investigation 3639 (1942). Hibbard, P. L., H i l ~ a r d i a 13, , 1 (1940). Hibbard, P. L., IND.ENO.CH~M., ANAL.ED.,9, 127 (1937). Hohn, H.. “Chemical Analyses with the Polarograph”, Berlin, Julius Springer, 1937. Holland, E. B., and Ritchie, W. S., J . Assoc. Oficial Ag7. Cham., 22, 333 (1939).
Hubbard, D. M., IND.ENQ.CHEM.,ANAL.ED.,12, 768 (1940). Hubbard, D. M., and Scott, E. W., J . Am. C h m . Soc., in press. O’Connor, R. T., IND.ENQ.CHEM.,ANAL.ED.,13,597 (1941). Reed, J. F., and Cummings, R. W., Ibid., 12, 489 (1940). Rogers, L. H., and Gall, 0. E.,Ibid.,9,42 (1937). Stout, P. R., Levy, J.. and Williams, L. C., Collection Ctachosloo.
Literature Cited
C h m . Commun., 10, 129 (1938).
Suprunovich, I. B., J . Gm. Chem. (U.5. S. R.), 8, 839 (1938).
(1) Bambach, K., and Burkey, R. E., IND.ENQ.CHEM.,ANAL.ED., 14, 904 (1942). (2) Biefeld, L. P., and Patrick, T. M., Ibid., 14, 275 (1942).
Walkley, A., Auatralia Cham. Inst. J . Proc., 9, 29 (1942). Wichmsnn, H. J.,
[email protected].,ANAL.ED.,11, 66 (1939).
Determination of Copper in Cast Iron by Direct Microelectrolysis WILLIAM M. MAcNEVIN r n RAYMOND ~ A. BOURNIQUE’, The Ohio State University, Columbus, Ohio This invwtigation of micromethods for the electrolytic determination of copper in nitric acid solutions of cast iron includes: the use of different kinds of apparatus and techniquemi the effects of other metals; the effect of nitric, sulfuric, and perchloric acids; the effect of insoluble siliceous and graphitic residues; the molybdenum interference; and the error in sampling. A recommended procedure is given.
T
HE discovery of important technical properties in copperiron alloys (4) has created a new interest in a rapid and
accurate method of analysis for copper. Previously this waa not a simple procedure but involved one or more preliminary chemical separations of the copper, followed by electrolyeis or other method of determination (6,6, 8). The chemical aeparation, in addition to beiig nlow, waa of questionable accuracy unlea extreme care was applied. Various r e m n s are given (9) for the necessity of separating other metals, parbicularly iron, from copper before it can be satisfactorily determined by electrolysis. (1) If much iron is present with the copper either the voltage used is not great enough to deposit all of the copper or, if it is, some iron usually deposita as well. (2) If nitric acid is present, the following phenomenon frequently occura (3). By the time most of the copper has been deposited, the solution will have acquired a brown color and then the brown color will suddenly disappear, accompanied by a rapid solution of the copper. Further increasing the voltage will not redeposit the copper. This effect has been attributed to the oxidizing effect of femc ions, nitric acid, ferric nitrate, and oxides of nitrogen. (3) Molybdenum, if present, deposita on the cathode as oxide. All the VariatiOM described above have been observed when conducting, on a macro wale, the electrolysis of copper in nitric acid solutions containing much iron. Preliminary experiments 1 Prment addrw, Watinghouee Elwtrio & Manufacturing Co.. Pitburgh. Pen-.
showed that most of the difficulties disappeared when the determination was carried out with microapparatus apd enough sample to give 0.3 to 2 mg. of copper. This suggested the present investigation of micromethods for the electrolytic determination of copper in nitric acid solutions of cast iron, There have been studied: (a) the use of different kinds of apparatus and techniques; (21) the effects of other metale; (c) the effect of nitric, sulfuric, and perchloric acids; (d) the effect of insoluble siliceous and graphitic residues; (e) the molybdenum interference; and cf) the error in sampling.
Samples A number of solutions containing various amounts of alloying met& were prepared in order to study the individual effect of each metal. After the method was developed it was tested with fifteen samples of cast iron or steel containing vsrying amounts of copper. These samples are described in Table I and are referred to below according to number.
OF SAMPLES ANALYZED TABLE 1. COMPOSITION
Sample No. Type % Cu 1 Cast iron 8.41 2 Cast iron 8.04‘ 3 Castiron 7.38 4 C a t irona 6.44 5 Curtiron 5.96 6 Cast iron 3.12 7 C a d iron 3.05 8 Cast iron 2.05 9 Castiron 1.91 10 Cast iron 1.50 11 CMtironb 1.44 12 Crst iron 1.09 13 Cast iron 0.38 14 Carbonsteele 0.267 I.6.
Chrome-
?& Cr 4.62 (7)
2.85 2.18
...
... o:ois 1.0
O:i6
vanadium steel (0.22% V) 0.13 1.03 Bureau of Standards sample 115. b Bureau of Standards sample 5p. 0 Bureau of Standard8 sample 358.
% Ni % C 20.20 2.56
% S i % Mn % Mo 1.96 1.22
lf:32 2:30 15.98 2.42 .. 3.0 2.8 ... 3 . 3 . . . 3.4 1.0 0.11 2.90 o : i 3 2.86
1:82 1.60 2.0 2.0 1.9 1.8
....
1:OS 1.01
. ..
1:iti 1.2
. ..
.. .
i:i i:i’ 1.a 0.61 . 1.0
0125
1103
0139 01345
.
0.28 0.292 0.21 0.50
. .. 1:: .. .. ..
0:s 0.2 015
... .. .
::: .
, ,
INDUSTRIAL AND ENGINEERING CHEMISTRY
760
TABLE 11. ELECTROLYSIS OF SYNTHETIC SAMPLES
Solution
(5-ml. aliquots, Ha801 f trace of "01) Mean No. of Wei ht Av. Theoretical Aliquotu of 8 u Deviation Wei ht Run Observed from Mean of 8 u
Vol. 15, No. 12
comparison was made in solutions containing sulfuric acid with only a trace of nitric, because it is well known that copper can be deposited from such a solution without interference. Deviation from theory could then be attributed to the effect of other metals.
Error
Effects of Nitric, Sulfuric, and Perchloric Acids 12 4 4 7 2 6
cu cu Cu-Fe Cu-Ni Cu-Cr Cu-Fe-Ni-Cr-Mn
1.391 1.492 1.388 1.410 1.413 1.382
0.015 0.015 0.006 0.007 0.000 0.017
1.398 1.502 1.398 1.398 1.398 1.398
-0,007 -0.010 -0.010 +0.012 4-0.015 -0.016
TABLE111. ELECTROLYSIS I N SULFURIC ACID (5-ml. aliquots, B. of 8. sample 115, 6.44% Cu) Mean Av. Theoretical Np. of Aliquota Wei ht Deviation Wei ht Sample Run of t u from Mean of
d
Error
Ma.
Ma.
Ma.
Mo.
1.292 1.291 1.275 1.288 1.291
0.005
1.291 1.291 1.291 1.291 1.291
$0.001 0.000 -0,016 -0.003 0.000
n. of S. cast iron
in Ha804 HNOi +++ 421mml.of l . of HNOi ml. of HNOa + 5ml. of HNOa
8 1 1 1 1
... ... .. ., ..
Apparatus All the early work was done with the Pregl (8) apparatus. A separate study was first made by the authors (7) on the precision of the microelectrolyticdetermination of copper. In doing so the Pregl and Hennance-Clarke (1) electrolytic cells were compared. Results with the Pregl cell varied more than could be accounted for and were generally lower than with the Hermance-Clarke a paratus. In the latter cell the solution wa.9 drained &ough a stopcock with continuous addition of wash water, thereb maintaining the circuit until washing was complete. With t k apparatus 1 to 2 mg. of copper could be deposited from 7 to 10 ml. of nitric or sulfunc acid solution with an accuracy of a0.01 mg., a figure that has been used in the present work as the accurac which ma be expected when no interferences occur. The !kermance-&rke cell was slightly modified. The surrounding water jacket was omitted and the cell made slightly larger to accommodate 10 to 12 ml. of solution. Otherwise the electrodes, electrode support, and air stirrer were the eame. For smaller percentages of copper, samples 13,14, and 15, it was necessary to use larger weights of samples. In order to avoid too high a concentration of other metals the solution of the sample was diluted to about 50 ml. and a correspondingly larger cell was used. It was constructed on the principle of the Clarke cell but with a funnel-shaped top to hold the extra solution. In such an apparatus all the solution circulates intimately between the electrodes. The apparatus was e uipped with voltmeter and ammeter with suitable resistances for Langing the applied voltage. The microbalance used was a Sartonus, air-damped type with pFotected beam. Its sensitivity was 1.35 micrograms per division.
General Procedure No unusual precautions were taken. Air was bubbled through the solution at a rate of 2 to 3 bubbles per second. Cathodes were cleaned in hot dilute nitric acid, washed with water and alcohol, dried a t 105' C. for 5 minutes; and allowed to stand 5 minutes before weighing. Bone-tipped forceps were used in handling the cathodes. Current and voltage were kept as nearly constant as possible during each electrolysis.
It is usually recommended that copper be electrolyzed from B sulfuric acid solution. This is inconvenient because the sample must first be dissolved in an oxidizing acid which is then removed by evaporation, The effect of nitric acid was therefore determined by comparing (Table 111) the results using sulfuric acid solutions with those containing various additions of nitric acid. The absence of any trend in the presence of nitric acid is apparent. Without any nitric acid, however, the evolution of hydrogen gas is exceasive and produces a spotty deposit. When nitric acid alone was used the results were equally good. In one set of experiments (Table IV) a copper solution was compared with one containing copper and iron, as a check on the frequent statement in the literature that low results are obtained with nitric acid solutions containing much iron. In the second set of experiments in nitric acid the copper wan determined in a cast iron. The results here (Table V) were slightly high (compareTable 111, aliquots in sulfuric acid), which may in part be due to a slight amount of carbon and silica which was not filtered from the individual samples. Finally, perchloric acid was tried aa a solvent for the sample. The rate of solution of the sample was notably slower and evolution of hydrogen was excessive during electrolysis. Some blisters and cavities were observed microscopically on the deposit. When a small addition of nitric acid was made, gas evolution diminished greatly and the deposit was clear. The results obtained (Table VI) have as good precision as those from sulfuric or nitric acid.
TABLE Iv. ELECTROLYSIS I N NITRICACID
Solution
(Synthetic samples. . _ . . 6-ml. diauota) Mean Aq. No. of Wei ht Dena- Theoretical Aliquota of tionfrom Wei ht Run Observed Mean of
cu Cu-Fe
&
7 7
8u
Error
.
Ma.
Mg.
Ma.
Mo
2.024 1.997
0.004 0.005
2.027 2.003
-0.003 -0.006
TABLE V. ELECTBOLYSIS IN NITRICACID (Hermance-Clarke apparatua. Individual sam les Bureau of Standard8 cast iron 115, 6.44% Cup Sample Weight Weight of Cu Theoretical Error Ma.
Ma.
Ma.
Ma.
30.394 30.946 35.947 34,078 29.617 29,950 28.499 37.388
1.978 1.993 2,355 2.203 1.934 1.957 1.856 2.407
1.957 1.993 2.315 2.194 1.906 1.929 1.835 2.408
+0.021
0.000 +0.040 +0.009 +0.028 +0.028 f0.021 -0.001
T ~ LVI. E ELECTROLYSIS IN PERCHLORIC ACID
Effects of Other Metals In preliminary work with alloys, results of wide variability and poor accuracy were obtained, but with synthetic solutions containing copper and other metals corresponding in concentration to the samples of cast iron described in Table I, the results (Table 11) agreed well with the 0.01-mg. precision obtained with solutions containing only copper. The slighGly low results whenever iron was present are not regarded as significant. This
(5-ml. aliquots, B. of 8. cast iron 115, 6.44% Cu) No. of Mean Av. Theoretical Aliquots Wei ht Deviation Wei ht Solvent Run of 8 u from Mean of 8 u
Error
Ma.
Mg.
Ma.
MR.
3
1.248
0.003
1.292
-0.048
5
1.290
0.005
1.292
-0,002
2
1.282
0.003
1.292
-0,010
+ + +
HCIO! 3-minute boiling HCIO! 15-minute boihng HCIO, 15-minute boiling
ANALYTICAL EDITION
December 15, 1943
Effect of Insoluble Residues It was apparent, through most of the work, that results with aliquot samples tended to be lower and in better agreement with the theoretical than with individual samples. In the latter case, any insolubleresidue of silica and graphite was not filtered out but was added to the cell, in order to avoid filtration and consequent excessive volumes. When aliquot sampling was employed, the siliceous residue usually settled before samples were taken. It waa suspected, however, that some suspended silica or graphite was behg coprecipitated with the copper; therefore several experimenta were made to check this point. Clear solutions were compared with those in which the suspension was purposely maintained at a maximum (Table VII). It is apparent that high results may be due to this effect. In separate experiments it was found that results were excessively high if the sample was ground unusually h e . In an effort to overcome this effect without the necessity for filtering or allowing the suspensions to settle, additions of gelatin and other organic reagents were made with the purpose of changing any charge on the solid matter and perhaps affecting its attraction by the cathode. Such attempts have been without success so far. The coprecipitation may therefore be largely a mechanical effect.
761
TABLE X. DETERMINATION OF COPPER IN INDIVIDUAL SAMPLES O F STEELS AND CAST IRONS Sample
No. 1 2 4 5 9
No. of Samples 12 4 0 4 4
Ap roximate Av. gample Deviation Weight from Mean ME. MU. 0.008 20-24 0.007 2c-23 0.016 20-26 0.002 21-23 0.012 4c-43
Av. Cu Observed
Theoreti4 Cu
%
%
8.44 7.88 6.45 6.07 1.98
8.41 8.04 6.44 5.90 1.91
TABLE XI. DETERMINATION OF COPPERIN ALIQUOTSAMPLES OF STEELSAND CAST IRONS Sample
No. of
No.
Aliquota
1 2 3 4 5 0 9 11 12 13 14 15
2 8 4 15 8 4 2 4 5 2 2 2
Weight of Sample per Ahquot
Ma. 20 20 20 20 20 2540 40 40-100 50-80
80 200 200
Av. Deviation from Mean Weight of Cu MO. 0.005 0.003 0.005 0.005 0.011 0.004 0.004 0.010 0.008 0.003 0.020 0.000
Av. Cu Theoretical Observed Cu
%
%
8.43 7.98 7.43 6.48 6.05 3.15 1.99 1.52 1.04 0.40 0.27 0.13
8.41 8.04 7.38 6.44 5.96 3.12 1.91 1.44 1.09 0.38 0.27 0.13
Interference by Molybdenum Molybdenum when present was found to deposit on the cathode
as the iridescent black sesquioxide, even at voltages as low as 2.2 volts. A solution of phosphoric acid or ammonium bifluoride prevented this. With bifluoride, however, a brownish precipitate formed on the cathode and additional nitric acid was necessary.
For a 5-ml. aliquot sample, 1.5 ml. of 85 per cent phosphoric acid or 0.2 gram of bifluoride plus 0.5 ml. of nitric acid was needed in order to prevent the deposition of molybdenum. A voltage of 2.5 volts was d c i e n t . Three samples of copper-molybdenum
cast irons were analyzed with this modification (Table VIII).
TABLE VII. EFFECTOF SILICAAND GRAPHITE Best chemical value, 8 . 4 1 7 Cu. aIlqu0t.g Mem We' h t No. of of% Solution Aliquot. Observed
(Sample 1.
1. Clear 1 SiOr C supended 2: CIeirA?) 2. Si01 suspended 3. c d r 4. Clear
0 1 4 3 3 2
Nitrio acid ~olution,5-ml. Theoretioal We' ht
&I
of ME.
Error
ME. 1.713 1.749 1.773 1.837 1.089 1.714
1.684 1.084 1.726 1.720 1.086 1.707
+0.027
Mo. +0.005
+0.003 4-0.078 +0.004 C0.007
OF COPPER IN MOLYBDENUU CAST TABLE WII. DETERMINATION IRON Sample
No.
Mo %
Reagent
Theoretiaal cu
Observed
%
%
TABLE IX. ANALYSISOF SIFTEDSAMPLEB
cu
-
Sampling Error The eampling error is not important if samples as large as 20 to 40 mg. of 20-mesh or h e r material are used. Samples were sifted and prepared in three sizes: coarser than 20-mesh, between 20- and 4O-mesh, and finer than &mesh. Several determinations were made on each of the sifted samples. The results shown in Table I X indicate that individual sampling under these conditions is not a serious cause of error.
Recommended Procedure Prepare a solution in the necessary solvent, using enough sample so that each 5 ml. will contain 0.3 to 2 mg. of copper. The sample should also contain 0.3 to 0.5 ml. of 70 per cent nitric acid. If the sample contains molybdenum add 0.15 ml. of 85 per cent hosphoric acid. Filter through a sintered-glass micro funnerinto the electrolytic cell, rinse, and make u the volume to 9 to 10 ml. Add 3 drops of 95 per cent ethyl alcoiol to eliminate spray, adjust the air stream used for stirrin to 2 to 3 bubbles per aecond, and electrolyze at 2.5 to 2.8 volts for 20 a n u m . Add 20 to 30 mg. of urea, rinse down the cell walls, and continue electrolysis for another 5 minutes. Slowly drain the cell, adding wash water simultaneously until the current has dropped to zero. Do not allow the voltage to exceed 3.5 volts during this ope" tion. Withdraw the cathode, rinse with alcohol, dry for 5 rmnutes a t 110' C., cool 5 minutes, and weigh.
Sample Weight
Observed
Weight of Cu Theoretical
Mo.
MU.
ME.
Moa
>2O-mwh
21 .03 20.77 22.09 19.92
1.774 1.748 1.845 1.680
1.769 1.747 1.858 1.070
4-0.005 +O.OOl
Table X gives a summary of results obtained in this way wing individual samples. Table XI shows similar results using aliquot samples.
20-40
20.74 22.39 22.60 21 . 0 8 24.34 22.80 21.40 23.58
1.753 1.888 1.921 1.771 2.063 1.918 1.810 2.014
1.744 1.882 1.900 1.773 2,047 1.917 1.800 1.983
+0.009
Discussion and Summary
Sieve Sire