In summary, it appears overwheliningly probable that the S and R data apply not to an alloy of approximate composition Na4Pb but to some other substance or substances as yet unidentifieri, ant1 possibly of uukriowri constitution and structurc We are grateful to the Ethyl Corporation for financial support, and to hIr. S.AT. Rlitzer aitd r h . H. Shapiro for valuable discussions. DEPARTMENT OF CHEMISTRY MASSACHUSETTS ISSTITUTE OF TECHXOLOGY CAWHRIDGE, MASS
The Reaction of Chlorosilanes with Benzaldehyde BY ALBRECHTZAPPEL RECEIVED FEBRUARY 28, 1965
Iiochow and Gingold' recently stated that they did not succeed in combining chlorosilanes with aldehydes, including benzaldehyde, to form siloxanes, even on boiling for 3-3 days, the starting products being obtained again in unchanged form. I n this respect I wish to report on work carried out on this reaction, though with different scope and aims. Silicon tetrachloride reacts with benzaldehyde on being allowed to stand a t room temperature, the reaction being expressed by the gross equation Sic14
+ C,H,COH +( S i C l ~ 0 )+~ GHaCHCI?
If Sic14 is allowed to stand together with benzaldehyde, after a few weeks these liquids which inis well show an increase in viscosity which is somewhat accelerated b y exposure t o daylight. Dernixing occurs in two phases after 2-3 months. This phenomenon is caused more rapidly a t elevated temperatures. Likewise, in the gaseous phase Sic14 reacts with benzaldehyde within the sense of the above equation. The lighter one of the two layers consisted chiefly of excess benzaldehyde and benzal chloride as well as a small quantity of chloropolysiloxane while the heavy phase consisted to the major part of chloropolysiloxane. The latter was obtained as a residue in the fractional distillation. 'The determination of the molar weight estimate resulted in values between lS00 and 2000. The chlorine values of these polysiloxanes, calculated on hydrolyzable chlorine, were correspondingly around 60m0 with the silicon values around 3FCC. The index n in the formula (SiCl,O), thus appears to be in the order of 1.5. ii-hile always an excess of benzaldehyde reacted with Sic&in all the cases examined, the reaction did not proceed beyond the formation of a chloropolysiloxane. I t appears to be probable, however, that the reaction will proceed to the formation of Si02 with sufficiently long periods of eyposure or more critical conditions. ;\s shown by orientating qualitative experiments, the reaction is not limited to S i c & but also proceeds in a corresponding manner in the case of organochlorosilanes, e.g., phenyltrichlorosilane. The reaction appears to take place in two stages. (1) I(1074)
G
Rochox,
and
h' (:m,rold
?HI5
Jol
R\IAl,
76, - i Y i ?
111 a first stage the Sic14 will add on tlic CO tloul)lt~ link of the aldehyde
I'I.ISiCl
+ OP-CIIC611i--d Cl.,SiOCIICICeIln ( 1 )
This product seeins to react, relatively rapidly with a further Sic1 link t o form CitSiOC€IC1C611~ j
.
CI.,SiOSiCl:;
+ C6HsCIICll
(2)
S o such products as result from equation 1 were isolated. The apparent discrepancy with respect to the work of Rochow and Gingold can satisfactorily be accounted for by these two authors only having tried triphenylchlorosilanes with benzaldehyde.' In the case of this silane the reaction is strongly inhibited for steric reasons, and the reactivity of the Sic1 linkages is unfavorably influenced by t h e three phenyl groups. This suggestion is supported by the observed fact that already the PhSiC13 reacts with benzaldehycl:. somewhat slower than SiCI,. ( 2 ) Private report b y Prof. E . G. Rochow
BAYERi\KTIENGESSEI,I,SCHA~T I ~ ~ ~ E R K ~ - s E N - B (:ERRIAUY . ~ ~ - ~ R ~ ~ , ~ ~ ~ , FARBENFABRIKEN
Phase Diagram of the System KSbO3--KTaO3by the Methods of Differential Thermal and Resistance Analysis B Y ARs0r.D
REISMAX,S O L TRIEBWASSER AKn HOLTZBERC RECEIVEDA P R I L 15, 1953
FRXnERrC
K N b 0 3 and KTa03both exhibit the BBOn perovskite structure, but have widely separated Curie points. The unit cell size of both compounds, in the cubic state, differs only slightly.' On the basis of this, and the isomorphic nature of tantalum and niobium compounds, a strong possibility for solid solution interaction between the meta salts was seen to exist. If such an interaction takes place, it might be possible to prepare dielectric ceramics with Curie points ranging between 13 and 68S°K., the Curie temperatures of KTaOy and KNb03, re~ p e c t i v e l y . ~Resolution ,~ of the diagram woukl provide needed information for the preparation ( i f the necessary ceramic materials. The solidus curve of a solid solution is generally more difficult to determine than the liquidus curve. The graphical extrapolation method employed by Tammann4 and later modified by Campbell and Prodanj was found inapplicable because of the difficulties in attaining reasonable equilibrium throughout the cooling range. The standard heating curve techniques are a t best laborious, and if not performed with the utmost care tend to gi\?e low results.6 If applicable, quenching methods are invaluable, but in this system the tendency fur crystallization to occur, even with small charges, (1) P. Vousden, Acta C y y s l . , 4 , 373 (1951). ( 2 ) J. K. IIulm, B. T. LIatthias and E. -4. L o n g , P h y s . X l v , 7 9 , 885 (19.50). ( 3 ) G. Shirane, H. Danner, A. Pavlovic a n d I< P e p i n s k y . i!'i,i., 93,
672 (19nO). (4) C. 'l'ammann, Z . Q E O Y ~ . C h e m . , 37, 303 (1903). ( 5 ) A S . Campbell a n d L. A . Prodan, T H I SJ O U K N A I . . 70, 5.-A (1'1) C, \T, l f o r e y , Jr., Il'ash. A i n d . (1.f Sii., 13, :lf; (19231.
(Ill
IX!
Aug. 20, 1933
NOTES
was so great that quenches could not be employed to fix solidus points. Consequently a procedure was sought which would enable closer approximation to static conditions in order t o attain better equilibrium. Use of quasi static conditions would eliminate the one inherent difficulty attendant with presently employed methods. Several iiivestigators have reported the use of high temperature resistivity measurements for thermal The methods employed were applicable only to thin metal wires. Glaser and Moskowitzll have recently reported a method for measuring resistivities of refractory hard metals, but because of the inherent difficulties of the method very large temperature gradients existed in the specimens. A procedure has been developed in this Laboratory which enables precise determination of solidus points by continuously measuring the conductivity of ceramic samples as a function of temperature. If two closely spaced platinum wires are imbedded in a ceramic disc, a sharp change in conductivity should be observed when the solidus is reached. Since the conductivity-temperature data are independent of heating rate, quasi equilibrium can be maintained throughout a determination.
for eight hours, ground, repressed and refired until homogeneity and completeness of reaction were verified. The wires were imbedded in the final pressing. Differential Thermal Analysis.-The heating and cooling curves were obtained using the apparatus previously described.13 The only change in t h e equipment was t h a t the top plug of the platinum furnace was wired with a platinum spiral. This modification increased the length of the uniform temperature zone. Previously described seeding and stirring techniques were employed during cooling experiments permitting t h e use of samples weighing approximately 10 g. fur liquidus determinations. As the stirring w a s sufficient to prevent temperature gradients in t h e liquid, large heat effects could be obtained even with cooling ratcs of 0.7"/min. and Pt-Pt 1ORh couples. Two cooling curves were run a t each of seven different mole Yo. Heating curves were run using 15 g. of powdered solid solution. T h e thermocouples were centered in t h e powder and the heating rate was approximately l"/min. These points were used t o verify the d a t a obtained from conductivity measurements. Conductivity Analysis.-The measurements were made with a Mosely x-y recorder. T h e voltage developed across a small resistor in series with a d r y cell and the sample is applied t o t h e x-axis of the recorder as shown in Fig. l a . T h e output of a thermocouple monitoring the temperature of the sample is connected to the y-axis. A Kanthal wound furnace, controlled b y a variac drive was employed in these determinations and the temperature was rapidly brought u p t o about 25' below the solidus. T h e rate of temperature rise was then adjusted t o approximately 0.5"/min. or less. A typical conductivity curve is shown in Fig. l b . At least two determinations were made at each of the mole % studied, and the results obtained were within f 3 ' from one another.
Experimental Procedure Reagents .-The tantalum pentoxide used mas purchased from the Fansteel Metallurgical Company. This reagent is the company's T-400 grade and contained as the maximum impurity O . O O l ~ o TiOn. The niobium pentoxide was purchased from t h e same company and contained a s the maximum impurity 0.2yo T a . Reagent Grade Mallinckrodt potassium carbonate was dried at 400' for one hour t o remove moisture. T h e potassium niobate was prepared according t o the procedure developed in this Laboratory,'Z and was a t least 99.9% pure. T h e potassium tantalate was prepared by a similar procedure t o the niobate except t h a t a final H F extraction was employed t o remove small traces of TanO;. Preparation of Samples.-Samples used for cooling curve c q x r i m e n t s were prepared by fusing NbzOs, TanOj and K 2 C 0 3until Con evolution was complete. The validity of this procedure was determined by running mired samples of pure KSb03 and KTZi03 a t the same mole percentages. For heating curve experiments, very intimate mixtures of the oxides were fired approximately 30" below the expected solidus for two-hour intervals, until weight luss determination showed completion of t h e reaction. The samples were then ground, dried and refired and reground until X-ray spectrometer traces showed complete homogeneity of t h e solid solution. The total firing time was approximately 40 houri : i d n ~ ) r n i a lonly l ~ ~ two grindings were necessary. As :LII dtcrnativc procedure, ceramics mere prepared from the pure meta salts. The discs mere ground to a fine powder and used for thermal aiialysis. Coiiductivity measurements were made with ceramics liaving two B & S #28 platinum vires imbedded in them. The ipdcing of the wires w:is approximately 2 mm. These samples were prepared directly from the pure meta salts, and completeness ( i f snlicl solution interaction was verified with X-ray spectrometer tracings. The intimately ground salts were prezsetl a t 10" 11~.,41i.~ without the use of binders. The sample5 \yere t l i i A i i lirc~l:ipproxiiiiately 30" below the solidus ~~
.-
(7) S I . K . Andrrws, J P h r s . C k e n i , , 27, 270 (1993). (8) K . Becker, "Hochschmelzende Hartstoffe und ihre technische Anwendung," Verlag Chemie, Berlin, 1923. (9) W Espe and XI. E C n d I , " WerkstuiTkunde der I l u c h v a k u u m t e c l l ~ nik," Springer Varlag, Berlin, 193Ii. (10) U'. H. Colner a n d 0. Zmeskal, T ~ n n sA. S M , 44, 1158 (1952). (11) F. W. Glaser a n d D . XIoskowitz, Powder M e t . B u i . , 6, 178
(lU53). (12) A. Reismnil, F. H d t z h c r g , S. Triehwasser, ?if. Derkenblit, 'l'rei>aratic~nof I'iire I'utnssiiim Metnniobate," t o Le piil,lished
42'0
Vx
VY
4
*.
n
2x10~
I -
A I i Fig. 1.-(a)
.
v, (bl
(01
conductivity circuit; (b) conductivity curvc.
Discussion of Experimental Results The data obtained by the different methods are listed in Table I, and are graphically depicted in Fig. 2 . TABLE I 'I'HERMAL A N I ) 1
