T f I E JOI:R,\'dL
Sept., 1911
OF I.Y.DUSTRI.4L A4.1'D E S G I N E E R I N G C H E h f I S T R Y .
b y the bismuthate method as usual. Th:.s method is very similar to one which the writer devised and has used for this class of materials. There is very little choice between the two, apparently, inasmuch as the time consumed, degree of accuracy, etc., is about the same in each case. However, the method to be described might give better results than that; of TVatters on high vanadium products, inasmuch as the vanadium in such cases might not always be completely precipitated by his method and this would cause high results. Watters' method was tested as to this point with the Bureau of Standards chrome-vanadium standard, containing about 1.32 per cent. of chromium and 0 . 2 0 per cent, of vanadium, and no vanadium was found in the filtrate with the manganese. The percentage of manganese in the standard was found to be identical by the two methods. It has already been shown1 t h a t large amounts of chromium and vanadium are completely precipitated from solutions of steel without coprecipitation of manganese, provided the iron is kept mainly in the ferrous condition while the solution is being boiled with the precipitant. To carry out the method for manganese, I or z grams of steel are dissolved in sulphuric acid (IO per cent. b y volume), observing the precautions given in the last-quoted paper, and the chromium and vanadium precipitated by cadmium carbonate as described therein. To the filtrate from this precipitate add 2 5 cc. of concentrated nitric acid and boil till free from fumes. Cool, oxidize with bismuthate, filter through asbestos, reduce with a measured excess (of ferrous solution and titrate as usual. The method is quite rapid, and its use, 01'the use of similar methods which eliminate chromium and vanadium during the bismuthate oxidation, seems called for, inasmuch as the results of the cooperating analysts on the above standard were several hundredths of a per cent. high where the bismuthate method was used. Further, the Ford-TVilliams method also gives high results, apparently due to occlusion o f chromic acid. Some precipitates of manganese obtained by the Ford-Williams method from the chrome-vanadium standard were dissolved in sulphuric acid, neutralized with cadmium carbonate and boiled. The precipitate showed appreciable amounts of chromium when dissolved in nitric acid and oxidized with potassium chlorate. A bismuthate determination of manganese in the filtrate gave a result agreeing with the determinations b y the method of M'atters and that of the writer. B a R B A L r O F ST.4SD.iKDS.
~~.4SHINGTOS.
ON THE SURFACE TENSION OF MOLTEN GLASSES. By
EDWINTVARD T I L L O T S O h . ,
JR
Rweived June 16, 1911
Surface tension has been, up t o the present'time, one of the properties of a glass or enamel of which a comparatively small amount of information is a t hand. Owing to the lack of convenient methods for its del
niIs
JOURNAL.
3, 476 (1911)
631
termination it has been impossible to measure its value and the characteristic effects upon it of each constituent of the glass. This paper is a description of a simple, convenient, and fairly accurate method whereby the surface tension of glasses may be compared. r The method is a variation of that commonly used in the case of liquids, in which the surface tension is calculated from the weight of a drop falling from a tube or a surface of definite size. This method was used by QuinckeI for determining surface tension of easily fusible metals, and a modification of it for those having a high melting point. I n the latter case, small metallic rods or wires were lowered vertically into the horizontal flame of a blast lamp. From the weight of the drop which was formed and which fell, and the diameter of the wire, the surface tension was calculated by means of the following equation, l?i = 2n?'T, in which T is the surface tension, W is the weight of the drop, and 2 7 is the diameter of the rod. Quincke applied this method t o a large number of elements and salts, and recorded one experiment with glass fibers. The method has, however, not been extensively used since i t is not extremely accurate, but it is useful in instances where other methods are not applicable, especially where simple and rapid measurements are t o be made. The ability of this method to give absolute values for the surface tension is doubtful. At first sight i t would appear t h a t the weight of the drop increases until it just overcomes the upward pull of the surface tension, becomes detached and falls. If the problem were as simple as that, the equation given above would hold and the weight of the drops would be proportional to the diameter of the rod, or, in the case of liquids, to the diameter of the tube from which they fall. Tate stated* this as a law, but Lord Rayleigh showed3 that when water is allowed t o drop from various sized glass or metallic tubes, the weight of the drop increases relatively faster than the diameter of the tube up to a certain point, when the weight of the drop remains constant no matter how large the tube or surface from which it falls. Lord Rayleigh also pointed out t h a t the weight of the drop of water is influenced by a number of factors, such as the difference in pressure within the liquid drop from the atmospheric pressure, the physical character of the surface from which the drop is suspended, and the relation of the inner t o the outer diameter of the tube. I t would seem, therefore, that if metallic or glass rods or fibers were used the last two factors, a t least, would be eliminated. In the experiments described in this paper the glass fibers were lowered in a vertical position with the aid of the machine shown in Fig. I . I n this figure, A is a 3/,-inch iron rod 30 inches long, supported by a n iron plate 8 inches in diameter and I / ~inch thick, which is inch iron rod, secured firmly t o the table. B is a
' Pow. A n n . , 184,
356 (1868); I b i d . , 136. 621 (1868).
Phil. M a g . , 27, 176 (1864). Ibid , [.5] 48, 321.
,
T H E J O U R N A L OF I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y .
D
E C
I3
N A
passing through the guides C and D and threaded for one-half of its length. It is raised and lowered by the threaded gear wheel E, which is actuated through gears b y means of the pulley P. The lower end of B is split, forming a clamp adjusted by the milled head N. The glass fiber is held in this clamp and is kept from swinging by a loop of wire held on the stage S, which is above the blast lamp. The drops of glass were caught in a porcelain dish beneath. Figs. 2-7 show the results obtained with a variety of silicate glasses. The approximate composition of these glasses expressed in equivalents are as follows: A. SiO, 75.3 per cent., CaO 4.4per cent., Na,O 20.3 per cent. B. SiO, 56.1per cent., P,O, 3 . 7 per cent., BaO 10.2 per cent., CaO 17.4 per cent., Na,O 10.7 per cent., K,O 2 . 0 per cent. C. Composition was not definitely known, except t h a t it contained large percentages of lead and boric oxides. D. SiO, 80.1 per cent., BaO 4.7 per cent., Na,O 15.2 per cent. E. SiO, 71.4 per cent., BaO 11.4 per cent., Na,O 17.2 per cent. F. A beaker of Jena laboratory glass. In the figures given the ordinates represent the weight of the drop expressed in grams and the abscissas the diameter of the fiber in millimeters. If Tate’s law were correct, a straight line would be produced, i. e . , the weight of the drop would be directly proportional to the diameter of the fiber. I t is evident from the results obtained t h a t this is not the case, since for larger fibers the weight of the drop is greater than this law demands. If, however, an empirical factor, proportional to the area of the cross section of the fiber, be introduced, a curve may be obtained which closely approximates the experimental results. Since this new factor is proportional t o the areas of the cross section or to the square of the diameter, it is evidently due t o the cohesive power which exists throughout the molten glass as distinguished from the phenomena appearing a t the surface. Before the drop can fall this cohesion, in addition to the surface tension, must be overcome. The weight of the drop is therefore increasingly greater with the larger fibers. The continuous curves which are shown in the figures are calculated from the following equation, W = a-D P-D*, in which W is the weight of the drop in milligrams, D is the diameter of the fiber in millimeters, and a and P are constants. This equation was derived by Quinckel from one given by Karmarsch and was employed for calculating the surface tension of solids. However, he did not apply it t o his results with molten substances. It must be remembered that, while this equation appears to satisfy the condition for small fibers,of glass, it cannot be considered to be an absolutely general equation. For, as Lord Rayleighr found in the case of drops of water, there will doubtless be a limit t o the size of rod, above which the drops will all be of the same size.
+
S 1 L
Fig. 1.
I
Sept., 1911
1 LOC.
cit
-
Sept., 191I
T H E J O U R N A L OF I N D U S T R I A L A N D ENGINEERING C H E M I S T R Y
Fig.
2.
Fig. 3.
633
634
T H E J O U R N A L OF I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y .
5 Fig 4
Sept.,
lulLlll Fig. 5
1911
Sept.,
1911
T H E J O V R N A L OF IA'DUSTRIALP . A N D E N G I N E E R I N G C H E M I S T R Y .
Fig. 6 .
Fig. 7.
63 5
T H E J O U R N A L OF I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y .
636
In the expression given above, a is a factor proportional to the surface tension, and from which the surface tension may be calculated, as shown by the following considerations: From the equation W = a-D P-D’ it is evident t h a t the forces in action a t the moment when the drop falls, consist of the mass of the drop, which, under the influence of gravity, acts downwards ; the surface tension, proportional t o the diameter (circumference) of the fiber, which acts upwards; and a third force, proportional t o the square of the diameter, which also acts upwards. The weight of the fallen drop is therefore resolvable into two factors, one of which, a-D, is characteristic of the surface tension. If this fraction of W be denoted b y W‘, we have W f = a-D. Since the surface tension is t o be recorded in dynes, the relation between the several variables is expressed by the following equation: 9.8 W’ = nDT in which T is the surface tension, 9.8 is the value in dynes per centimeter of I mg. acting a t right angles to a line I mm. long, and in which W and D are expressed in milligrams and millimeters respectively. By eliminating W-’ and D in the last two equations, and solving 9.8a for T we find: T =
+
-.
A
Table I gives the summary of results obtained. Column I shows the glass used, column 2 values of a column 3 values of P, and column 4 the value in dynes per centimeter of the surface tension as calculated from a. TABLEI. Glass.
A , . .. . . . . . . . . . . . . . .
B..................
c. . . . . . . . . . . . . . . . . . D
..................
E.................. F . . .. . . . . . . . . . . . . . .
(2.
50.0 40.0 35.0 50.0 46.0 53.0
3. 4.5 2.5 . o 15 .o 15 .o 15 . 0 10.0
,, I. 156 .O 125.0 109.0 156.0 143 .O 165 .O
I t must be remembered t h a t these values obtained for the surface tension are probably not the absolute values, for, as Lord Rayleigh showed, in the case of water, this method gave values only slightly greater than one-half of the true value. The value of this method lies in the facility with which surface tension of two glasses may be compared, and it is not improbable t h a t an empirical factor, such as Lord Rayleigh used, may be employed when it is desired to know the true surface tension. * From the results of Table I , it is evident t h a t glass with higher percentages of silica show the higher surface tension, and also t h a t lead and boric oxides tend t o decrease surface tension. It appears also t h a t silicates differing widely in composition do not give extremely different values for the surface tension. I n employing this method for the determination of the surface tension of borate glass several difficulties were encountered, I t was almost impossible to draw large fibers with circular cross sections. Experiments were therefore usually limited to fibers smaller than 0.6 of a mm. in diameter. The results obtained with the same size of fiber were not so concordant .and showed in gefieral a greater divergence from the theoretical straight line than those obtained with
Fig. 8.
Sept., 1911
Sept., r q r
I
TT'E J O U R N A L OF I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y .
silicates, Fig. 8 shows a typical experiment. I n Table XI is shonn the composition of glasses employed and the results obtained.
30 70
20 80
10 9o
Bz03 00 P b O 100 Equivalents.
40 60
11.
T dynes
1..
2.. 3
Bz08 BaO Equiv. Equiy.
.
.. ,
.
,
.
4,.... 5 . .,., 6,.... i . ,, , ,
66.7 72.8 62.1 59.0 81.1
.. .. ..
68.1
..
50.7 49.1 37.2 25.3 80.0 78.0
,
,.
.. .
and it has been shown t h a t for a variety of silicate glasses the surface tension lies within not very wide limits. Borate glasses possess a surface tension which
50 50
60 40
70
30
80 20
90 10
100 00
Fig. 9. y.4Bl.E
No.
63 7
€'bo
NazO
Equiv. Equiv. ,
,
, ,
9.7 16.7 18.9 31.9 49.3 50.9 62.8 74 7
33.3 27.2 28.2 24.3
.. .. ..
a.
38.0 34.0 34.0 30.0 19.0 31.0 37.0 37.0 28.0 29.0 22.0 25.0 40.0
:?,
15.0 15.0 13.0 15.0
.. ..
15.0 15.0 15 0 15.0 15.0 15.0 15.0
per ceniimeter. 118.0 116.0 106.0 94.0 59.0 96.0 115.0 115.0 87.0 90.0 69.0 78.0 125.0
is lower than t h a t of the silicates and which is markedly different for the several lead borates. DEFT. INDUSTRIAL RESEARCH, UNIVERSITYOF KANSAS.
A RAPID AND ACCURATE METHOD FOR THE ANALYSIS OF WHITE METALS. By J. C. BENEKER,Metallurgist, Cincinnati Metal Products Company.
After investigating and trying many methods for the analysis of bearing and type metals, I have de10 , . . . . ,. vised a composite scheme which embodies the best I1 ..... 20.0 . features of other researches and several ideas which 12 . . . . . , 22.0 .. ,. 13... , . 65.5 34.5 . , , I believe are original. The method is rapid and at the same time quite accurate. Although these data are not complete, they bring Dissolve 1 / 2 gram of the fine drillings in a mixture out some interesting points. The values of p are the of 2 5 cc. HCl $. 5 cc. H N O or 2 5 cc. HCl saturated same for all the glasses in which enough meitsurements with liquid bromine (not bromine water), warm on were made t o determine the nature of the curve. the steam bath and add about one gram tartaric acid The addition of boric oxide t o borax, as shown in Ex- t o hold up the antimony: Dilute with hot water periments I and 2 , lowers the surface tension. The t o about 400 cc. and slowly add an excess of a solution addition of lead borate has the same effect, as shown of NaOH. Now add a n excess (about 20 cc.) of a in Experiments I , 3 and 4. I n Experiments 1 1 , 1 2 , I O per cent. solution of Na,S. This is made by satuand 1 3 , in which barium borates were used, an in- rating half of a I O per cent. NaOH solution with H,S crease in amount of boric oxide also causes a decrease and then adding the other half. Allow t o settle and in surface tension. On comparing these with Ex- filter by decantation a couple of times, using hot, periments 5-10, in which lead borates were used, we dilute Na,S solution for washing. Transfer the find boric acid showing the same effect when present precipitate t o the filter and wash with a hot, dilute in more than one molecule of B,O,t o one of PbO. Na,S solution several times. The residue consists of I n glasses which contain less' than one molecule of the sulphides of Cu, Pb, Fe and Zn, while the filtrate B,O,t o one of PbO the surface tension behaves irreg- contains Sn and Sb also As if present (Method of ularly, showing in general a lowering of the surface A. Rossing) .I tension with a decrease in the amount of boric oxide. Dissolve the residue in HNO, and run down with Further work will be necessary t o determine whether H,SO, for the lead. The Cu, Fe and Zn are deterthe surface tension may be relied upon t o indicate mined as usual, the copper preferably b y A. H. Low's definite compounds in a glass. The series of lead lbdometric method.2 Add t o the filtrate a small borates, as given above and plotted in Fig. g , is suffi- excess of HC1 and allow t o settle on the steam bath cient t o show t h a t the presence of compounds has con- about an hour. Filter the precipitate, washing well siderable influence on surface tension of the glass. with luke warm water t o remove all H,S. Place the The purpose of this paper has been t o bring forward paper containing the sulphides in an Erlenmeyer a simple and rapid method measuring the surface flask. Add a measured excess of N / I O iodine solutension of certain high-melting substances, especially Journal Society Chemical Industry, 1902, 191. glasses. A number of experiments have been recorded "Technical Methods of Ore Analysis 8_ . . . . 9,....
,.
.. ..
,.
_. , .
"