Download as pdf or txt
Download as pdf or txt
You are on page 1of 2

Rapid Gravimetric Determination of Silicon in

Aluminum Alloys
PHILIP LI SAN1 and HENRY L. KATZ
Industrial Test Laboratory, Philadelphia Naval Shipyard, Philadelphia 12, Pa.

A rapid gravimetric method is described for the determination of silicon in


aluminum alloys for amounts exceeding 1.5%. An acid attack is employed using
phosphoric, nitric, and sulfuric acids. The silicon is oxidized by evaporation of
the solution, an excess of perchloric acid is added, and the silicic acid is dehydrated
by moderate boiling of the perchloric acid. The usual procedure of handling
dehydrated silicic acid is then followed. The precision and accuracy of the new
method are determined by a statistical study of the data.

INVESTIGATION was recently undertaken at this labo- addition of perchloric acid to the solution after the conversion of
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

AN ratory to
L obtain a rapid and accurate gravimetric method silicon to silicic acid, followed by a moderate boiling of the per-
for the determination of silicon in aluminum alloys. Although chloric acid, appeared to solve the problem. Subsequent experi-
Downloaded via UNIV OF PENNSYLVANIA on May 2, 2020 at 15:57:08 (UTC).

many colorimetric procedures which have the necessary speed and ments showed that when sufficient perchloric acid was present,
accuracy have appeared in the literature (7, 8, 9, 16), the control the boiling time was not critical and could be extended from 5 to
of pH, temperature, and time was so critical as to render these 15 minutes without causing a significant change in the silicon re-
methods unsuccessful in the hands of routine analysts running sults. In the course of the investigation the authors confirmed
large numbers of determinations. This laboratory usually deter- the lower limit on the percentage of silicon (1.5%) that could be
mines lower amounts of silicon (under 5%) spectrographically, determined by means of a phosphoric, nitric, and sulfuric acid
but the need still existéd for a rapid chemical procedure for de- mixture (11). Further observations on alloys containing up to
termining silicon contents of 5% and above. 1% magnesium and more than 1.5% silicon showed no apparent
Two gravimetric procedures have been used in this laboratory loss of silicon due to the possible presence of magnesium silicide.
for the determination of silicon in aluminum alloys. One is a To establish the fact that the acid mixture used in the proposed
modified A.S.T.M. method (2), employing an alkali attack of the method effected a complete oxidation of all the silicon present in
aluminum alloy followed by a single evaporation with hydrogen the alloy, four samples in duplicate (types 43, 355, 356, and 32)
peroxide and a subsequent single dehydration with perchloric were dissolved and the silicon was oxidized in the manner de-
acid; the other is the Alcoa procedure (1), similar to the A.S.T.M. scribed below. The solutions were then diluted and filtered, and
method but using a triple evaporation with hydrogen peroxide the amount of residue was determined. In practically every case
followed by a double dehydration with a perchloric-sulfuric acid the residue was negligible and the greatest amount found repre-
mixture. Since the alkali attack common to both procedures is sented less than 0.05% silicon.
tedious and time-consuming, efforts were directed toward evolv- The procedure finally evolved employs an acid attack of the
ing an accurate and rapid gravimetric method employing an acid alloy by a phosphoric, nitric, and sulfuric acid mixture in a ratio
attack of the aluminum alloy. of 3-4-1, respectively. The solution is evaporated until the
Many papers have been published on the use of acid mixtures silicon is completely oxidized, as evidenced by a clearing of the
(S, IS, 14,15,18,19) for attacking aluminum alloys in an attempt solution. Perchloric acid is then added and the solution boiled
to speed up the analysis. In general these procedures tend to moderately. Gelatin is added to facilitate the filtration of the
give low results which have been attributed to two factors—the silica (10) and the usual procedure for handling dehydrated silicic
incomplete oxidation of silicon in the metal to silica and loss of acid is followed.
volatile silicon compounds, such as silicon hydride, incurred dur-
REAGENTS
ing acid decomposition (4, 5, 17), indicating that the proper com-
bination of acids had not been attained. Fuchshuber, however, An acid mixture of 150 ml. of phosphoric acid (sp. gr. 1.71),
proposed a method (6) using a mixture of phosphoric, nitric, and 200 ml. of nitric acid (sp. gr. 1.42), and 50 ml. of sulfuric acid
sulfuric acids which, besides effectively dissolving the alloy with- (sp. gr. 1.84).
Ammonium nitrate, c.p. Perchloric acid, 70%. Gelatin
out any apparent loss of silicon by volatilization, oxidized all the solution, 1%. Sulfuric acid (1to 99). Sulfuric acid (1 to 1).
silicon present directly to silicic acid upon evaporation of the Hydrofluoric acid, 48%.
solution. Hydrochloric acid was then added and the solution
evaporated again until the silicic acid was dehydrated. The main PROCEDURE
objection to this procedure is that continuance of the dehydration Dissolve 2 grams of the sample (1.5 to 4.5% silicon) or 1 gram
for as little as one minute longer than is necessary tends to cause of the sample (4.5 to 10.0% silicon) in an 800-ml. beaker in 80 ml.
dissolution of silica in phosphoric acid. A later modification of of acid mixture. For samples containing more than 10% silicon,
dissolve 0.5 gram of the sample in 60 ml. of acid mixture. After
this method (11) introduced cobalt nitrate as a means of indicat- the initial reaction has subsided and most of the alloy has dis-
ing exactly when the silicic acid was completely dehydrated. The solved, cover the beaker and warm until the sample is completely
authors, however, have found that the change in color of the co- in solution. Remove the cover glass and evaporate until the
balt nitrate is not so sharp as suggested and have experienced solution clears. Immediately add a pinch of solid ammonium
nitrate to oxidize any sulfur, swirl, and remove the beaker from
difficulty in obtaining precise and accurate results. the hot plate. Allow to cool slightly, add 60 ml. of perchloric
acid (70%), and evaporate to fumes of perchloric acid. Cover
EXPERIMENTAL the beaker and moderately boil the solution for 8 or 9 minutes.
Investigations were undertaken by the authors to eliminate Cool somewhat, cautiously add 300 ml. of hot distilled water and
40 ml. of gelatin solution (1%), and stir well. Filter immediately
the critical time factor inherent in the above procedures. The on an 11-cm. No. 41 Whatman filter paper containing some paper

tl Present address, Franklin Institute, Laboratories of Research and pulp. Swab the beaker and wash the paper 12 times with hot
Development, Philadelphia 3, Pa. dilute sulfuric acid (1 to 99), followed by a few washes of hot dis-
252
VOLUME 19, NO. 4, APRIL 19 4 7 253

tilled water. Transfer the paper and precipitate to a platinum not significant—i.e., all methods give substantially the same aver-
crucible, char off the paper completely, and ignite the residue at age. The proposed method values, on the other hand, are sig-
1100° C. to constant weight. Add 3 drops of dilute sulfuric acid
(1 to 1) and 5 ml. of hydrofluoric acid (48%), and evaporate care- nificantly different from those obtained with the umpire method
fully to dryness. Ignite at 1100° C. to constant weight. The in only one case (chemist 1, sample 32). In other words, the
difference in the two weights represents the weight of silica which Alcoa method and the proposed method gave substantially the
contains 46.72% silicon. same results, while the modified A.S.T.M. procedure tended to

PRECISION AND ACCURACY give low results on the high silicon-bearing samples.
In view of the demonstrated precision and accuracy, together
To test the reliability of the proposed method, six samples were with the large saving in operational time effected, the proposed
chosen as representative of the types of aluminum alloys usually method, in the opinion of the authors, constitutes an excellent
run for silicon in this laboratory ranging from 2.0 to 12.5% of procedure for routine laboratory use.

Table I. Comparison of Methods for Determining Silicon in Aluminum Alloys


Chemist 1 Chemist 2
Standardc Standardc
Sample® Method^ % Silicon Av. deviation % Silicon Av. deviation
195A I 2. 17 2. 09 2. 03 2,.10 0. 07 2. 02 2. 00 2, 00 2, 01 0. 01
II 2. 01 1. 99 2, 04 2,.01 0. 02 2. 01 1, 99 2. 08 2. 03 0. 05
108 I 3 66 3 .59 3. 61 3 ,62 0. 04 3, 56 3. 56 3, 61 3,.58 0, 03
II 3 .64 3 .63 3,.47 3 .58 0. 10 3 ,65 3, 61 3. 67 3. 64 0, 03
355 I 5 .42 5 .37 5 .31 5 .37 0,.05 5 .18 5 .28 5 26 5 .24 0 .05
II 5. 40 5. 42 43 5. 42 0. 01 5. 30 5. 31 5. 26 5. 29 0. 03
43 I 5..48 5, 47 '. 43 5 .46 0. 03 5..38 5. 42 5, 42 5. 41 0. 02
II 5, 50 5..57 5. 52 5,.53 0. 04 5. 66 5. 60 5..54 5. 60 0. 06
III 5. 52 5 55, 5.,54 5 .54 .0. 02 5 45
. 5 .53 5 49 5 49 0. 04
356 I 6, 69 6 .82 6, 68 6 .73 0 08 6 .63 6 .63 6 .61 6 .62 0 .01
II 6, 76 6 75 6 82 6 .78 0 ,04 6 .78 6 .74 6 .88 6 .80 0 .07
III 6 86 6 .83 6..84 6 .84 0, 01 6 .79 6 .78 6 .81 6 .79 0 .01
32 I 11, 94 11 95 11. 98 11 .96 0. 02 11 .60 11 .90 11 .82 11 .77 0 .16
II 12 .36 12 .28 12 32 12 .32 0 04 12 .35 12 .29 12 .15 12 .26 0 .10
III 12 .14 12 .16 12 .26 12 .19 0 .06 12 .20 12 .06 12 .04 12 .10 0 .09
All samples I 0. 07 d 0 .07d
All samples II 0. 05¿ 0 ,06d
®
Sample 355, 0.5% magnesium. Sample 43, 0.13% titanium. Sample 356, 0.10% titanium and
0.3% magnesium. Sample 32, 1.0% magnesium.
>
I, modified A.S.T.M. method. II, acid attack method. Ill, Alcoa method.
c Calculated from Standard deviation -
\l—.
\n -
1’
where 2d2 = sum of squares of deviations of each
of 3 results from their average and = 3.
d Calculated from where 2s2 is the sum of squares of standard deviations and N is the number
V:
of standard deviations

silicon. The series of six samples was analyzed for silicon by two ACKNOWLEDGMENT

operators, using the proposed method and the modified A.S.T.M. The authors wish to express their gratitude to M. Kaplan for
procedure, both operators running each series three separate the statistical analysis of the data, to G. Norwitz for assistance
times. The results obtained are shown in Table I. in the experimental work, and to K. L. Proctor for his continued
In studying precision, a distinction should be made between interest in the investigation.
the ability of a chemist to reproduce his own results with a given
method and the ability of different chemists, perhaps in different LITERATURE CITED
laboratories, to check each other. Since only two chemists were
used in this study, both from one laboratory, only the first can be (1) Aluminum Co. of America, “Chemical Analysis of Aluminum",
2nd ed., pp. 28-30, 1941.
estimated with any validity. In this sense, the results show the
(2) Am. Soc. Testing Materials, “Methods of Chemical Analysis of
two methods to be substantially the same, with an average stand- Metals", p. 131, 1943.
ard deviation of about 0.06% silicon. Using this figure, the pre- (3) Callender, L. H., Analyst, 58, 81 (1933).
cision may be expressed as follows: In the long run, a single de- (4) Churchill, . V., Bridges, R. W., and Lee, M. F., Ind. Eso.
termination on a given sample by a given chemist by either Chem., Anal. Ed., 9, 201, (1937).
(5) Ibid, 9, 533-4 (1937).
method would not be expected to differ from the average of a (6) Fuchshuber, ., Z. anal. Chem., 116, 421 (1939),
large number of determinations under the same conditions, by (7) Hadley, B. W., Analyst, 66, 486 (1941).
more than 0.12% silicon, 95% of the time. (8) Ibid., 67, 5 (1942).
Examination of the averages in Table I shows that in the case (9) Ibid., 70, 43 (1945).
(10) Nikolaev, N. S., Zavodskaya Lab., 10, 536-8 (1941).
of the high silicon-bearing alloys, the modified A.S.T.M. method (11) Osborn, G. H., and Clark, J., Metallurgia, 31, No. 185, 230
tends to give lower results than the proposed method (note es- (1945).
pecially sample 32 and chemist 2 results on samples 43 and 356). (12) Rider, P. R„ “Modern Statistical Methods”, New York, John
The testing of these differences by the t test {12) indicates that Wiley & Sons, 1939.
(13) Salzer, Erhard, and Theissig, F., Chem.-Ztg., 64, 468 (1940).
they are statistically significant on the 95% level. (14) Shishkin, B. M., Zavodskaya Lab. 7, 96 (1938).
To determine whether the higher or lower results are the more (15) Shkotova, S. N.,/bid., 8, 213 (1939).
reliable, three of the samples (43, 356, and 32) were reanalyzed by (16) Stross, W., Analyst, 69, 44 (1944).
the Alcoa method which, because of the additional peroxide evapo- (17) Term, A., and Moravia, G., Atti Xo congr. intern, chim., 3, 470
(1939).
rations and the double dehydration employed, may be consid- (18) Urech, P., Z, anorg. allgem. Chem., 214, 111 (1933).
ered as an umpire method. The values obtained by the latter (19) Vasil’ev, K. A., and Vegrin, M. L., Zavodskaya Lab,, 7, 263
procedure are higher than those obtained by the modified (1938).
A.S.T.M. method, as can be seen from Table I. In all cases ex-
The opinions expressed are those of the authors and are not to be construed
cept one, these differences are statistically significant on the 95% as reflecting the official views of the Navy Department, through whose
level; in the one case, chemist 1 on sample 356, the difference is permission this article is published.

You might also like