March 1947
INDUSTRIAL AND ENGINEERING CHEMISTRY
acid were used. T h e more concentrated acid solutions Tvere prepared in pure silver beakers from anhydrous hydrofluoric acid (Pennsylvania Salt Manufacturing Co.) b y dilution and by reaction with powdered silica (Baker's analyzed, C.P. grade). T h e hydrofluoric acid solutions were analyzed by titration with standard sodium hydroxide solutions (silica- and carbonatefree), using phenolphthalein as indicator. T h e hydrofluoric and hydrofluosilicic acid mixtures were analyzed by a method of cold and hot titrations with standard sodium hydroxide, using phenolphthalein as indicator, according to the procedure described by Scott ( 0 ) and Sherry et al. ( 1 0 ) . DISCUSSIOS OF RESCLTS
Table I1 presents the liquid and vapor compositions and related boiling points as determined for the 5ystem HF-HZO. T h e normal boiling points in column 5 were cnlculated from the experimental d a t a on the assumption of a n average entropy of vaporization of 23 calories per degree-mole for these solutions. The liquid and vapor compositions, in veight per cent, and the normal boiling points are plotted in Figure 6. Table 111 lists t h e boiling points and related liquid and vapor compositions as determined for t h e ternary system HF-H2SiFsHs0. These data are plotted in Figure 7 along with the approsimate liquid isotherms. T h e constant-boiling mixture of the system HF-H20 T K I ~ found to have a composition of 3 8 . 2 6 5 hydrofluoric acid and a boiling point of 112.0" C. a t 750.2 mm. pressure. These values are compared with previously published d a t a in Table IV and are in good agreement with results secured by Muehlberger ( 7 ) . T h e boiling point curve for the system HF-H,O as determined in thi. work is a t considerable variance TTith t h a t determined hy Fredenhagen and Wellmann ( d ) , except in the regions approaching the pure components. Lacking suitable corrohorntive dat:i, i t is concluded t h a t the boiling point curve as determined in this work is the more reliable in view of: (a) the agreement secured between the boiling points of the test liquids as determined in the still (Table I) Kith the published values, ( b ) the good agreement of t h e d a t a for the constant-boiling mixture with those secured b y Muehlberger, and (c) the smooth curve given b y the boiling points and t h e close extrapolation of this curve t o the boiling point of pure hydrofluoric acid. In the ternary system HF-H2SiF6-H20, a masimum boiling point was found having a cornposition of 10yc hydrofluoric acid, 36% hydrofluosilicic acid, and 54% Kater, and a boiling point of 116.1" C. a t 759.7 mm. pressure. This point lies in a vapor pressure trough which extends across the phase sy=tem from the
43 1
L3zeotropic mixture of HF-H20 to t h a t of H2SiF6-H&. as indicated by the dotted line on Figure 7. The composit,ion ( 4 1 5 !iydrofluosilicic acid and 59% water) and the boiling point, 111.5" C.) indicated for the aaeotropic mixture of the system HzSiF6-H20 are only approximate values; attempts t o determine these d a t a accurately were unsuccessful, because of the tendency of such solutions with lon- free hydrofluoric acid content to deposit silica in the boiler or in the condenser of the still d u r ing distillation. T h e boiling point ridge divides the phase diagram into two regions so t h a t a solution of a composition lying in one region cannot be distilled under normal pressure to yield ,z condensate having a composition lying in t h e other region. T h e results secured for the liquid and vapor compositions of solutions in the more acid regions of the ternary system (that ib, for solutions containing over 60% free hydrofluoric acid or over 30"' free hydrofluoric and 30% free hydrofluosilicic acids) are not equilibrium values since t h e still condenser did not condense all the vapor phase. Severt'heless, since in t,he method of sampling employed, practically all of the vapor and the eondensate y e r e collected for analysis, it was concluded t h a t the compositions found gave a reliable approximation of tlir liquid-vapor equilibria of such solutions. .iCBSOW LEDGhI E S T
The authors hereby express their thank5 to the PenwylvLtniu Salt llanufncturing Co. for permission t o puhlilh these esperimental data. An nclinowledgmeiit is made of the :issistance of R.H. Rnlston in sume preliminary studies on the subject systems. LlTERATURE CITED
Baur. E., Z . phuaik. Chem., 48, 453 (1904). Bineau, -L,Ann. chim. phys., [3] 7 , 257 (1843). Deussen. E., 2.anorg. Chem., 49, 297 (1906). Fredenhagen, IC., and Wellmann, M., 2.p h y s i k . Cheni., A162, 454 (1932). ( 5 ) Gore, G., Landolt-Bornstein Tabellen, p. 273 (1905). (6) Jacobson, C. X., J . Phus. Chern., 27, 577, 761 (1923): 28, 506 (1942). (7) hluehlberger, C. IT., Ibid., 32, 1858 (1928). (8) Roscoe, H., J . Chem. SOC.,13, 162 (1860); - i n n . Chern. 1 1 . Pharrn., 116, 218 (1860). (9) Scott, TI'. W,, "Standard Methods of Chemical .lnalysis", 5th ed., Vol., 11, p. 2209, New York, D. Van Nostrand Co., 1939. (10) Sherry, W.B., Swinehart, C. F., Dunphy, R. A., and Oghum S. C., IXD. ENG.CHEM.,ANAL.ED.,16,483 (1044). (11) Truchot, C., Compt. rend., 98, 821 (1884). PRESESTEO before the Symposium o n Fluorine Cheniistry as paper 93. Division of Industrial a n d Engineering Chemistry, 110th Meeting of tlie .$\IEHIC.4\. C H E \ l I C h t SOCIETY, Chicago, 111.
HYDROGEN-FLUORINE TORCH Homer F. Priest and Aristid V. Grossel WAR R E S E i R C H LABORATORIES, COLU\IB14 U \ I \ ERSITY. A E W YORB, N. Y .
A
S SOOK as the problem was solved of Dtoriiig fliiori:it=
under pressure ( 5 ) , the means was avrtilable for studying fluorine flames and constructing a fluorine torch. Since fluorine is the most reactive element known, we expected it t o produce the highest possible flame temperatures. This follows for two interdependent reasons: First,, reactions n.ith fluorine show the highest molar heat evolution, and second, because of the high strength of fluorine bonds, t h e dissociation of the reaction products is curtailed. The first point is illustrated by a comparison of the heat of 1 Present
address, Houdry Process Corporation, Marcus H o o k . Pa.
combustion ( f ) of one mole of hydrogen in fluorine and in oxygen, as shon-n below:
+ F: +2HF(,,,) + 128.0 lig.-cd. Hz + HnO(,ss, + 57.8 kg.-cal. H?
1,1202 -t
(1) (2)
Thus, in the case of fluorine the heat generated, for the sam(> volume of hydrogen, is 2.2 times greater than for oxygen. Similarly, tlie heats of combustion of methane or acetylene with fluorine are tTTice :IS high as those with oxygen, for
CH4
+ iF, --+
CF4
+ 4 H F + 415 1rg.-cal.
(3,
432
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 39, No. 3
4 TORCH was constructed of the concentric-tube tipmixing type with fluorine in t h e inner tip. i n intense bluish-white flame was produced with which copper could be welded w ith considerable ease, since the copper fluoride formed melts a t a lower temperature than copper and t h e +I elding operation is therefore self-refluxing. Nickel, \Ionel, and steel were welded with equal facilit?, but attempts to weld aluminum were unsuccessful. By stepping u p the fluorine suppll a t t h e moment when t h e copper fused, a cutting action similar to t h a t on steel is obtained, to gi\e a clean, uniform, hnifelihe cut an copper.
They are compared in Figure 1 wit11 the corresponding values for the reaction, 2H: taken e:dy 1410 3.500'
+ 02 e 2H20
from Lewis and Elbe's book (2). From these curves it can be calculated that l r h dissociation is reached for water at I