The Analysis of Magnesia Determination of Loss on Ignition, Silica, Iron, Alumina, Calcium, Manganese, and Phosphorus GILBERT W. SMIT€I1,Westvaco Chemical Division, Food Machinery & Chemical Corporation, Newark, Calg. Analytical methods for loss on ignition, silica, iron, alumina, calcium, manganese, and phosphorus i n magnesia are described. Loss on ignition and silica are determined by classical gravimetric methods; iron is determined colorimetrically by the o-phenanthroline method; alumina is determined colorimetrically by the aluminon method after separation of the iron by extraction with dichloroethyl ether from a 6 N hydrochloric acid solution; calcium is separated from magnesia by the method of Caley and Elving, after which it is converted to the oxalate and determined by titration with ceric sulfate; manganese is determined colorimetrically as the permanganate; phosphorus is determined colorimetrically as the reduced phosphomolybdate. Satisfactory precisions were obtained by these methods.
0
VER a period of fifteen years, a careful study has been made by this laboratory of methods for chemical analysis of magnesia. These methods include the determination of silica, iron, alumina, calcium, manganese, and phosphorus, several of which are among the more difficult determinations of inorganic quantitative analysis. Since the procedures noted have been successfully used over a period of several years, it is believed that their publication, together with a critical evaluation of the important variables, will constitute a contribution of definite value to those who make and use magnesia. The procedures described are based on published methods and are applicable to magnesia derived from either sea water or mine sources and ranging in respect to proportions of impurities as follows: Silica (SiOz) Iron oxide (FezOa) Alumina (AlzOs) Calcium (CaO) Manganese (Mn) Phosphorus (PzOs) Loss on ignition
0 . 0 5 to ZO+% 0,0001to 0 . 5 % 0.01 t o 2.0% 0.01 t o 5 . 0 % 0.0003 t o O . a % 0.0001t o 0 . 5 7 0
...
As the methods are also used in routine analysis, unnecessary manipulations have been eliminated to shorten the required operating time. Following a detailed description of the methods, data on factors affecting them are given, together with the precision and accuracy as determined in these laboratories. The term “magnesia” refers to partially or completely calcined magnesite or brucite as well as to magnesia from sea water. The magnesium oxide present may be partially combined with water, with carbon dioxide, or both. All analyses are reported on the “as is” basis. However a method for loss on ignition is included from which analyses may be calculated on the basis of the magnesium oxide content. REAGENTS
Mixed Acids (for Silica Determination). Dilute 500 ml. of C.P. 12 S hydrochloric acid with 500 ml. of distilled water. Add 700 ml. of C.P. 70 to 727, perchloric acid and mix. Hydrochloric Acid, 0.3 N . Dilute 25 ml. of C.P. 12 N hydrochloric acid to 1 liter with distilled water. Dichloroethyl Ether, Carbide and Carbon Chemicals Corporation. Aluminon Reagent. Dissolve 92 g r a m of C.P. ammonium acetate, 0.200 gram of Eastman P4468 aluminon, and 1 gram of gum arabic in separate portions of distilled water. Pour into a 1liter volumetric flask in the order given, dilute to the mark, and mix. Stock Standard Aluminum Solution. Dissolve 4.74 grams of C.P. aluminum chloride hexahydrate in distilled water, dilute to 1 liter, and mix. Store in a rubber or wax bottle. This solution is equivalent to approximately 1 mg. of aluminum oxide per milliliter. Standardize gravimetrically according to the method of Blum (3). This is a stock solution and will keep indefinitely. 1
Present address, Oak Ridge Xational Laboratory, Oak Ridge, Tenn.
Standard Aluminum Solution (for Calibration Curve). Dilute 5 ml. of the stock standard alumina solution to 500 ml. with distilled water and mix. This solution is equivalent to approximately 10 micrograms of aluminum oxide per milliliter. Hydroxylamine Hydrochloride. Dissolve 20 grams of C.P. hydroxylamine hydrochloride in distilled water and dilute to 1 liter. a-Phenanthroline. Dissolve 4.0 grams of reagent grade ophenanthroline monohydrate in 200 ml. of 95% alcohol and dilute to 1liter with distilled water. Ammonium Acetate, 6 N . Dissolve 460 grams of C.P. ammonium acetate in sufficient distilled water to make 1 liter of solution. Standard Iron Solution. Dissolve about 10 grams of C.P. ferrous ammonium sulfate in distilled water, and add 5 ml. of c.P., 6 iV sulfuric acid and sufficient distilled water to make 1liter. Standardize this solution volumetrically and calculate the strength of the solution in terms of ferric oxide. Dilute the appropriate aliquot of this solution t o 1 liter to obtain a solution containing 0.0200 mg. of ferric oxide per ml. Methanol, Carbide and Carbon Chemicals Corporation, 99% grade. Ammonium Oxalate, 0.5 W . Dissolve 35 grams of C.P. ammonium oxalate monohydrate in distilled water and dilute to 1 liter. Ammonium Oxalate, 0.1 gram per liter. Dissolve 0.1 gram of C.P. ammonium oxalate monohydrate in 1 liter of distilled water. Phosphoric Acid, 2.5 N . Dilute 500 ml. of C.P. S570 phosphoric acid to 1 liter with distilled water. Phosphoric Acid, Treated. Dilute 100 ml. of C.P. 8570phosphoric acid to 1 liter with distilled water. Add 0.4 gram of sodium periodate and boil for 5 minutes. Cool before using. Do not add water after boiling. Ammonium Molybdate Reagent. Diqsolve 15 grams of C.P. ammonium molybdate in 200 ml. of distilled water. Add 700 ml. of 5.00 ,V hydrochloric acid slowly, mixing well between additions. Cool to room temperature and dilute to 1 liter with distilled water. Store in a brown, glass-stoppered bottle. This solution is stable for 2 months. Stannous Chloride, Stock Solution for Phosphorus. Dissolve 10 grams of C.P. stannous chloride dihydrate in 25 ml. of C.P. 12 N hydrochloric acid. Add a few pieces of metallic tin to the solution and store in a brown, glm-stoppered bottle. This solution is stable in contact with metallic tin. Stannous Chloride, Reagent for Phosphorus. Dilute 3 ml. of stannous chloride stock solution to 1 liter with distilled water Make up fresh before use. Standard Phosphate Solution. Dissolve 0.1915 gram of C.P. potassium dihydrogen phosphate in distilled water and dilute to 1 liter. One milliliter of this solution contains 0.100 mg. - of phos. phorus pentoxide. Quinaldine Red Indicator. Dissolve 0.2 gram of the dye in 100 ml. of distilled water. APPARATUS
Platinum Crucibles, l>ml. capacity, with covers. Electric Mume. Use any muffle capable of maintaining a temperature of 1000’ C. Blast Burner. Use compressed air blast burner similar to
1085
1086
ANALYTICAL CHEMISTRY
Fisher KO.3-910, capable of producing a temperature of 1150' to 1200' C. in a platinum crucible. Separatory Funnel, 125-ml., similar to Fisher Catalog No. 10437. Centrifuge, International, Size 1, Type C, Fisher Catalog No. 5-120 or similar. Colorimeter. Fisher Electrophotometer and 23-ml. test tube (sellsare used for routine testing. An Eimer & Amend colorimeter (no longer on the market) or a Coleman double monochromator spectrophotometer was used in the development of the methods. Any other photometer or spectrophotometer of equal quality could be used. PROCEDURES
~~
loss
Silica. Keigh accurately a 1-gram, - 100-mesh sample of magnesia and transfer quantitatively to a dry 400-ml. beaker. Add 5 to 10 nil. of distilled water to slurry and 30 ml. of mixed acids. Swirl over a l o r , free flame until the sample is completely dissolved, moving steadily to prevent caking. Continue heating until particles of silicic acid begin to form and coagulate. Place the beaker on the hot plate and allow to evaporate slorvly for at least 15 minutes, fuming until first crystallization of solids, but avoiding baking. Remove from the hot plate and cool. Add 5 ml. of C.P. 12 -V hydrochloric acid and then about 150 ml. of hot distilled water. Heat the solution to boiling and filter the silica on a 9-cm. \Thatman No. 41 filter paper. Carefully remove the silica adhering to the beaker by means of a rubber policeman. Kash the beaker and then the silica on the filter paper five times with small portions of 0.3 ,V hydrochloric acid. If the silica is high, more washing may be necessary at this stage. Follow by washing the silica five times with hot water. Transfer the paper and silica to a clean, unrveighed platinum crucible. Carefully char the paper in the front of the muffle and then slowlv increase the heat by moving the crucible toward the back of the muffle. When the carbon has been ignited, place a tightly fitting cover on the crucible and heat a t 1000" C. for 15 minutes. Finally heat over the blast burner for 5 minutes. Cool and weigh to 0.1 mg. Moisten the crude silica with a drop or two of distilled water and add 2 or 3 ml. of C.P. 487'0 hydrofluoric acid. Evaporate to dryness on a w-rll ventilated hot plate. Heat cautiouslv in the muffle to remove the last of the hydrofluoric acid, place the cover on the crucible, and then blast for 5 minutes. Cool and reweigh. Repeat the procedure from the beginning of this paragraph until constant weight is obtained.
Loss in weight in grams
x
5 Fe20s
To prepare the calibration curve, pipet 1-, 2-, 5-,7-, lo-, and 12ml. portions of the standard iron solution into 250-ml. flasks and carry through the above procedure beginning with the dilution t o 86 ml. Compare the colored solutions with a reagent blank. Plot log scale reading against milligrams of ferric oxide. Alumina. Weigh accurately 2 grams of - 100-mesh magnesia and transfer quantitatively into a 250-ml. beaker. Slurry with a small amount of distilled water and cautiously add 30 ml. of C . P . 12 N hydrochloric acid. Evaporate the resulting solution just to incipient crystallization on a hot plate. Avoid further heating, as aluminum chloride is appreciably volatile above 150 C. -4fter cooling, add 2 ml. of 12 N hydrochloric acid and 75 to 100 ml. of distilled mater. Add 1 ml. of C.P. 307, hydrogen peroxide, heat to dissolve the solids, and continue heating until evolution of oxygen ceases. Filter through an 11-cm. Whatman S o . 41 filter paper into a 400-ml. beaker. Wash the bulk of the silica onto the filter by washing the beaker twice with small portions of water. Wash the filter paper twice with distilled \later. Evaporate the filtrate to 25 to 50 ml. and retain. Ignite the filter paper a t about 1000" C. until the carbon is burned off. Remove from the muffle, cool, and moisten the residue with 2 to 3 ml. of hvdrofluoric acid. Evaporate the hydrofluoric acid, then ignite the crucible a t 1000" C. once more for 5 minutes. Cool and fuse the residue a t a low temperature n i t h about 0.5 gram of fused potassium bisulfate until the melt is clear and free from solids. Cool the melt and dissolve i t in the concentrated filtrate set aside above. Transfer the solution to a 250-ml. volumetric flask, add 125 ml. of 12 S hydrochloric acid by means of a graduate, cool, dilute to the mark, and mix. Using a graduate, measure 40 nil. of this solution into a 125-ml. separatory funnel. Add 40 ml. of dichloroethyl ether by means of a graduate, shake the mixture for 1 minute, and allow to settle. Withdraw the bulk of the dichloroethyl ether, add 40 ml. more, and repeat the extraction. Save the used dichloroethyl ether to be recovered later by distillation. Centrifuge the aqueous phase a t a speed of about 2000 r.p.m. for 1 minute. Pipet 10 ml. of the clarified aqueous phase into a 100-ml. beaker and evaporate just to incipient crystallization. Cool the beaker and add 10 ml. of 5 Ah\-drochloric acid and 50 to 75 ml. of distilled water. When the salts are completely dissolved, transfer the solution to a 250-nil. volumetric flask, cool, dilute to the mark, and mix. Pipet 25 ml. of this solution into a 100-ml. Erlenmeyer flask and add 25 ml. of the aluminon reagent while swirling. Avoid direct sunlight in handling the colored solution. Heat the solution on a boiling water bath for exactly 10 minutes, cool to room temperature in a stream of running xvater, and compare in a Fisher Electrophotometer with a reagent blank prepared by heating 25 ml. of distilled water plus 1 ml. of 5 IV hydrochloric acid and 25 ml. of aluminon reagent in the same manner. Use a Wratten 40 filter with light intensity set a t B. From the log scale reading, obtain micrograms alumina from the calibration curve. O
LOSSon Ignition. Clean, ignite, cool, and accurately weigh a platinum crucible and cover. Transfer 2.5 to 3.5 grams of the sample to the crucible, immediately cover the crucible, and weigh accurately. Place the crucible in the front of an open muffle furnace. Move the crucible to the hottest part of the furnace during an interval of about 10 minutes. Close the door of the muffle and ignite for 1 hour at 1000 * 25" C. Remove the crucible from the muffle and, with the cover in place, cool as rapidly as possible. The best way to cool the crucible is to place it on a piece of unglazed tile and, after a fern minutes, transfer it to a metal plate to cool it to room temperature. Weigh accurately and calculate thr per cent loss on ignition. Gram loss on ignition X 100 = grams of sample
Mg. of Fe203 X 100 = 1000 X grams of sample
100 =
70silica
Iron. Weigh accurately a 0.5- to 3.0-gram sample of -100mesh magnesia and transfer quantitatively to a 250-ml. beaker. Add 15 ml. of C.P. 12 h'hydrochloric acid for each gram of magnesia taken and heat over a free flame until solution is complete. Evaporate to incipient crrstallization. Cooland add 1ml. of 12-1hydrochloric acid and 15 ml. of distilled water. Cover tTith a watch glass and boil. Filter through Khatman No. 42 filter paper, and wash the beaker and paper three times with small portions of hot 0.3 S hydrochloric acid and finally with water. If necessary, dilute accurately to a definite volume and take an aliquot (not over 25 ml.) containing not more than 0.25 mg. of ferric oxide. Dilute to 86 ml. in a 100-ml. graduate, mix, and transfer to an Erlenmeyer flask. .4dd 2 ml. of hvdroxylamine hydrochloride solution, 2 ml. of o-phenanthroline, and 10 ml. of 6 S ammonium acetate solution in the order given, mixing after each addition. After 15 minutes and before 60 minutes, compare the colored solution with a reagent blank using the Fisher Electrophotometer, Wratten 58A filter, and light intensity a t B. From the log scale reading obtain milligrams of ferric oxide from the calibration curve.
1Iierograms of A1120d X 250 ' 1,000,000 x 2
=
C'
Al>0
,electedas being the most suitable in this application. The effect of pH on color intensity was studied to determine the most suitable means of adjusting the acidity. This was done by adding varying amounts of hydrochloric acid to a series of solutions containing equal amounts of ammonium acetate. The data indicated that color was constant as long as sufficient ammonium acetate was present to convert all of the hydrochloric acid to acetic acid. The presence of free hydrochloric acid in the solution caused fading of the color. The practice has been adopted of adding a large excess (10 ml. of 6 N ) ammonium acetate to avoid the necessity of careful adjustment of the pH. The amount of hydroxylamine hydrochloride used is sufficient to reduce many times the amount of iron that is present in the sample. It was shown experimentally that a fivefold variation in the amount of hydroxylamine hydrochloride had no effect on the intensity of the color. The formula for the ferrous o-phenanthroline complex indicates that 10 micrograms of o-phenanthroline are required for each microgram of iron. Thus, the 8 mg. of o-phenanthroline which are used per analysis are sufficient to react with 0.8 mg. of iron, which is several times the amount ever found. Varying the ratio of o-phenanthroline to iron from 30:l to 180:l had no effect on the intensity of the color. Although little interference was expected from elements present in magnesia, especially that manufactured from sea water, the effect of a number of possible impurities was determined by adding the C.P. salts to 1-gram samples of a magnesia known to contain only traces of these impurities. The data indicate that phosphates, borates, manganese, and aluminum have no effect on the determination of iron. Titanium and sulfate appear to cause slightly low recoveries of iron (Table 11). Alumina. The phenylhydrazine method (1,9) was used in this laboratory for several years for the petermination of alumina in magnesia but gave results that were far from satisfactory from the standpoint of accuracy and precision as well as time consumption. The relatively low concentration of alumina sug-
gested the possibility of using a colorimetric method. A preliminary investigation of several colorimetric procedures indicated that the aluminon method was the most promising. Because iron interferes with the aluminon method it was necessary to remove iron prior to the addition of the reagent. Several. methods were tested for the removal of iron but the one which proved most satisfactory in this application was extraction by means of dichloroethyl ether as suggested by Axelrod and Swift ( 2 ) . Their work showed that dichloroethyl ether quantitatively extracted 0.01 formal iron from 7.5 N hydrochloric acid, but that lower acid concentrations resulted in incomplete extraction, and that increasing the concentration of iron made it necessary t o use higher concentrations of acid. It was found that the relatively low concentrations of iron which were present in this work could be extracted essentially quantitatively from 6 S hydrochloric acid. Because of the close proximity of the specific gravities of the aqueous and organic solutions, separation by gravity is extremely slow. Although not essential to the successful application of the method, a centrifuge is a desirable adjunct and is indeed indispensable for rapid routine work. The aluminon method was suggested by Hammett and Sottery (8). Olsen, Gee, and McGlendon (IS) improved the method by stabilizing the lake, by adding the buffer, aluminon, and stabilizer as a single composite reagent, and by a more reproducible method of heating the solution to develop the colored lake. 80
o BLANK AGAINST DISTILLED WATER A ALUMINON AGAINST DISTILLED WATER 0
60
I
ALUMINON AGAINST BLANK I
?
u
z 4
a a W
t;
40
1
a
0 0
20
To determine the effect of variations in the amount of acid remaining in the solution, a series of blanks and of samples containing 50 micrograms of aluminum was buffered with 60 milliequivalents of ammonium acetate and varying amounts of hydrochloric acid or sodium hydroxide were added, after which the color was developed in the usual manner. Both the blanks and the samples were compared against distilled water as a standard and each sample was compared against the blank having the corresponding pH as the standard. The color of the blank approached that of the sample a t both high and lox pH values. As was to be expected, when the sample was compared against the blank the maximum difference was observed a t intermediate pH values. The data plotted in Figure 1 show that while considerable latitude is permissible in the pH it is essential that the pH of the sample be very close to that of the blank against which it is compared. The color intensity increased slightly on standing a t room temperature but the increase was so small as to be negligible. Three samples, to each of which 50.0 miprograms of aluminum oxide were added, were found to contain 49.6 to 50.5 microgranis immediately after preparation. After standing 27 days 51.0
V O L U M E 2 0 , N O . 11, N O V E M B E R 1 9 4 8 to 52.0 micrograms were found for an average increase of 3% of the amount present. The L U I of the method as applied to a sample of periclase derived from sea water was found to be *O.OF% aluminum oxide for a sample analyzing 0.40Y0 aluminum oxide. Similarly, a sample of light-burned magnesia gave values of 0.17 *0.017c aluminum oxide. The accuracy of the procedure as applied to dead-burned magnesia was tested by the analysis of Bureau of Standards magnesite 104, which has a stated value of 0.84Ye aluminum oxide. Four analyses made by the above method gave values of 0.87, 0.84, 0.84, and 0.86%. In the case of light-burned magnesia, accuracy was established by using the classical etherhydrogen chloride procedure ( 7 ) , precipitating as aluminum chloride, and finally neighing as aluminum oxide. Results on samples analyzed by both the colorimetric procedure and the ether-hydrogen chloride method are given in Table I11 and substantiate the accuracy of the colorimetric procedure. Calcium. Prior to the work of Caley and Elving ( 4 ) no satisfactory method was available for the determination of small amounts of calcium in the presence of large amounts of magnesium. Caley and Elving suggested separation of calcium from magnesium as the sulfate in 90y0 methanol, followed by gravimetric determination of the calcium sulfate. This method has been applied to the determination of calcium in magnesia with the modification that calcium is determined volumetrically. The method consists in dissolution of the sample in hydrochloric acid, evaporation to remove excess acid and water, precipitation of the calcium as the sulfate in methanol, dissolution of the calcium sulfate in dilute acid, precipitation of the calcium as the oxalate, and finally titration of the oxalate with ceric sulfate. Caley and Elving suggest the use of 1 ml. of 36 N sulfuric acid to convert the calcium to the sulfate but state that a much smaller amount (twofold excess) is actually the minimum required. The use of twofold excess sulfuric acid in the author’s hands gave results that were 0.150/, low on Bureau of Standards magnesite 104 for which Caley and Elving had obtained good agreement with the stated value. The following experiment indicates that the minimum sulfuric acid requirement is actually near the amount which Caley and Elving suggested to be used, at least under the other conditions chosen for this test. Synthetic mixtures containing 850 mg. (21 millimoles) of magnesium oxide, 70 mg. of ferric oxide, and 34 mg. (0.6 millimole) of calcium oxide, were precipitated in the presence of varying amounts of sulfuric acid and the analyses completed as described above. The results indicate that it is desirable to have a t least
Table 111. Accuracy of Colorimetric Procedure for Alumina Light-Burned Magnesia No.
Per Cent Aluminum Colorimetric Ether-HC1 method method 0.17 0.18 0.16 0.18 0.18 0.20
0.19
Table IV.
0.16
o:i7 0.18
..
Effect of Sulfuric Acid Concentration on Recovery of Calcium CaO taken) CaO Found
(34.0 mg. of
Sulfuric Acid Millimoles 2 4.5 4.5 9 9 13.5 18 18 36
Error
1MQ.
Me.
32.6 32.8 32.9 33.3 33.4 33.6 33.5 33.6 33.5
-1.4 -1.2
-1.1 -0.7 -0.6 -0.4 -0.5 -0.4 -0.5
1089 13.5 millimoles of sulfuric acid present for maximum recovery and that 36 millimoles do no harm (Table IV). Thus, under the conditions chosen for this experiment, the minimum excess was some twentyfold. The effect of iron was determined in a similar manner, using 18 millimoles of sulfuric acid. The data given in Table V indicate that iron does not begin to cause low results until the amount exceeds 140 mg. of ferric oxide. As iron was not expected to exceed this amount, no procedure for its removal was included. Table V.
Effect of Iron on Recobery of Calcium (34 mg. of CaO taken)
Fez03 Taken
CaO Found
Error
MQ.
Mg. 33.4 33.4 33.3 32.6
Mg. -0.6 -0.6 -0.7 -1.4
0 70 140 210
Caley and Elving state that magnesium retards the precipitation of calcium and imply, a t least, that a substantial amount of time is required for complete precipitation of calcium from magnesia. Because it was desired to make the method as rapid a5 possible, the effect of time was determined by precipitating calcium (34 mg. of calcium oxide) from 100 ml. of 9570 methanol in the presence of 70 mg. of ferric oxide, 850 mg. of magnesium oxide, and 18 millimoles of sulfuric acid. Varying the time from the addition of the sulfuric acid to the start of the filtration from 1 minute to 60 minutes had no detectable effect upon the results. In practice this means that filtration can be started as soon as the sulfuric acid has been added and the filter prepared. The average of some forty determinations on Bureau of Standards magnesite 104 over a period of a year gave a value of 3.34y0 as compared with the stated value of 3.357,. The method may therefore be said to be accurate. The sample used for the precision test analyzed 1.93yc calcium oxide and gave an LUI value of *0.057,. Using the same sample the corresponding values for the LG2test were 1.96 * 0.12%. Manganese. This element is determined by an adaptation of the periodate procedure of Willard and Greathouse (16). Nitric acid is used as the solvent for the sample prior to this analysis on the basis of the finding of Mehlig ( 2 1 ) that free nitric acid in concentrations of 1.3 i$- or I’ess a t the time of measurement (3 N a t the time of boiling) does not affect the results. hlehlig carried out his test with nitric acid as the only acid present. I n the present application it was necessary to add phosphoric acid to eliminate iron interference, as well as nitric acid to dissolve the sample. Accordingly, the effect of nitric acid was determined on actual samples of magnesia in the presence of 5 ml. of 85% phosphoric acid and the volume of the solution a t the time of boiling was approximately the same as that when the color measurement was made. Under these conditions nitric acid had no effect upon the apparent amount of manganese a t concentrations up to 2 N. However, observation showed that increasing concentrations of nitric acid did retard the rate of color development. This effect became evident when the nitric acid concentration exceeded 0.8 W and this concentration was therefore chosen as the maximum to be used. Because the iron in magnesia is present in the ferrous state, in part a t least, it is necessary to boil the nitric acid solution to oxidize all the iron to the ferric state. It was found that 25 millimoles of phosphoric acid per 100 ml. are sufficient to eliminate interference by ferric iron and that six times this concentration does no harm. The precision of the method was determined using a representative sample of light-burned magnesia. The average and LU1 using the Coleman spectrophotometer were 0.0910 * 0.00097e manganese. The corresponding data for the LU2 test were determined by means of an Eimer & Amend colorimeter and were 0.0915 * 0.006070. The high ratio of the LU? to the LUI is some-
1090
A N A L Y T I C A L CHEMISTRY
what abnormal and is probably due, in part, to the fact that different photometers \\ere used for the two measurements. I n any case the precision is considered satisfactory for routine work. The accuracy was determined by adding known amounts of manganese to 2.5-gram portions of a sample of magnesia in which 0.0028% of manganese was found. The added manganese was recovered within the precision of the method (Table VI). Phosphorus. A colorimetric method was selected for determination of phosphorus because of the low concentration present. After a survey of the literature the reduced phosphomolybdate method of Dickman and Bray (5) was adopted because it was applicable in the presence of hydrochloric acid and iron. The author confirmed the findings of Dickman and Biay regarding the effect of acid concentration, ammonium molybdate concentration, and stannous chloride Concentration as well as the jtability of the color. Spectral absorption data indicated that the colored compound had a minimum transmittancy between 740 and 760 millimicrons. Accordingly, a red (Wratten F-29) filter was selected for the color measurement. Silica also reacts with molybdate to give a silicomolybdate which has properties similar to that of the phosphomolybdate. However, the acidity required for the formation of the reduced phosphomolybdate is much greater than that required for the formation of the reduced silicomolybdate and should be great enough to prevent the formation of the latter. &loreover,silica added in the form of a soluble silicate had no effect upon the intensity of the color when present in concentrations as high as 2 mg. Since the preparation of the sample includes a step for the separation of silica, higher concentrations of silica are not expected. Moreover, the accurate analysis of Bureau of Standards magnesite 104, which contains 2.54% silica, as described below, indicates that silica does not interfere. Dickman and Bray found that ferric iron did not interfere when present in concentrations of less than 15 p.p.m. but that it was riecessary to reduce higher concentrations of iron to the ferrous atate prior to color development. As iron is usually present in magnesia in concentrations sufficient to give a concentration of more than 15 p.p.m. in the final aliquot, hydroxylamine hydrorhloride is added to reduce the iron. I t was shown that this procedure eliminated iron interference even when the iron-phosphate ratio was as high as 100 to 1 by analysis of Bureau of Standard3 magnesite 104, which has a stated ferric oxide content of 7.07y0 and a phosphorus pentoxide content of 0.05770. Replicate analy-
Table VI.
of Method for 3Ianganese hlanganese Recovered Error
4ccuracy
Manganese Added 70
70
o/o
0,0008 0,0020 0.0040
0.0014 0.0024
f O ,0006
0.0081
0.0037
- 0.0003
0.0080
-0.0001
+ O ,0004
ses by the above procedure gave 0.058, 0.033, 0.059, and 0.05970 phosphorus pentoxide in this sample. ACKNOWLEDGMENT
The author hereby acknonledges the contributions of H. H. Hartzell (deceased), T. Woodn ard, F. Melhase, H. Leitch, and F. R. Brooks to the development of most of the methods cited and to Dvr.ight Williams and George S.Haines of the Technical Department ‘of \Testvaco at South Charleston, IT. Va., for developmental data and advice on these procedures. Appreciation is also expressed to C. W.Redeker of the Xen-ark Control Laboratory for somr of the experimental data and summaries relating to precision. LITERATURE CITED
(1) Allen, E. T., J . Am. Chem. SOC.,25, 421 (1903). (2) Axelrod. J., arid Swift, E. H., I b i d . , 62,33-6 (1940). (3) Bluni, W., Bur. Standards Sci. Paper 286 ( M a y 10, 1916). (4) Caley, E. R., and Elving, P. J., ISD.ESG. CHEM.,ASAL.ED.,10, 264 (1938). (5) Dickman, S. R., and Bray, R. H . , Ibid., 12,665 (1940). (6) Fortune, W. B., and Mellon, M .G . , I b i d . , 10,60 (1938). (7) Gooch, F.h., a n d H a v e n s , F. S., Am. J . Science, 2 , 4 1 6 (189ti). (8) H a m m e t t , L. P., and Sottery, C. T., J . Am. Chem. Soc., 47, 142 (1925). (9) Hess, W.H., and Campbell, E. D., I b i d . , 21,776 (1899). (10) Hillebrand, W.F., and Lundell, G. E. F., “Applied Inorganic Analysis,” pp. 722-5, New York, John Wiley & Sons, 1929. (11) Mehlig, J. P., IND. ESG.C H E M . AXAL. , ED.,11, 274 (1939). (12) Moran, R. F., I b i d . , 15,361 (1943). (13) Olsen, A. L., Gee, E. A , , and McGlendon, V., I b i d . , 16, 169 (1944). (14) Saywell, L. G., and Cunningham, B. B . , I b i d . , 9, 67 (1937). (15) Willard, H. H., a n d Cake, W.E., J . Am. Chem. Soc., 42, 2208 (1920). (16) Willard, H. H., and Greathouse, L. H., I b i d . , 39, 2366 (1917).
RECEIVED February 24, 1948.
‘Chromatographic Behavior of Some Aldehydes ARTHUR L. LEROSEN
AND - 4 L E X A N D E R
MAY’
Louisiana S t a t e University, Baton Rouge, La.
T
HERE are many data in the literature concerning the effect of changes in the number of adsorbed groups per molecule on the adsorption sequence in chromatographic columns. Two examples will illustrate this point, the increase of adsorption affinity with increasing number of double bonds per molecule in the carotenoid series given by Zechmeister (S),and the increase of adsorbability due to lengthening the carbon chain in hydrocarbons adsorbed on charcoal which has been excellently worked out by Claesson ( I ) . I n the present work the authors have studied the relation between side chain and R values in a series where the aliphatic side chain is essentially not adsorbed ( R is the rate of movement of the zone relative to the developing solvent. 21. ,
I
IPresent address, Chemistry Degartment, B o u t h w estrln Loiii5ianh Institiit?, Lafayettr, La.
The aldehydes were selected for this purpose, as they show good rates of movement on silicic acid when benzene is used as the developing solvent; they are easily detected by a solution of pararosaniline bleached by sulfur dioxide (Schiff’s reagent). The data for the limited number of aliphatic aldehydes available during the progress of this study are given graphically in Figure 1. I t is evident that the side chain acts as a hindrance to adsorption and thereby leads to an increase in the value of RLwith The values of R, increasing number of carbon atoms for the trailing edge were not significantly different from R( and are not given. The data for all experiments (Table I) indicate that a double bond (crotonaldehyde) produces a stronger adsorption than would be observed for the corresponding saturated compound, 0.284 compared to an estimated value of 0.475
