3364
1'. J . FLOKY, L. AIANDELKEKN, J. U. KINSINGERAND IT,U.SIuxrz
/cUl. I R I B U I I U S BKOM
Vol. 74
I H E UEPARlMENT OF CHEMISTRY O F CORNELL UNIVERSITY]
Molecular Dimensions of Polydimethylsiloxanesl
Intrinsic viscosities of carefully fractiouated polydirnet1iylsilox.aiic.s have been dcterniined in methyl ethyl ketone and in phenetole at the limiting critical niiscibility temperatures ( e )for . l l = m , 20 and 83" for the two solvents, respectively. Molecular weights were established by osmotic pressure measurenicnts. 'Tlic intrinsic viscosity [?]ein methyl ethyl ketone a t 20' is accurately proportional t o X ' / 2 from AI = 53,500 to 685,000. The cotlstant K = [77]e/.2f1/2has the value 0.80 X at 20'. Less extensive measurements in phLiietole a t 83" indicate t h a t K is nearly independent of the temperature. 2 , calculated from K exceeds the free rotation value by 50%. The root-mean-square end-to-end length ratio, ( r 2 ~ / . 1 1 ) ' / as This disparity is somewhat less than has been found for other polymers.
Introduction
intriusic viscosity measurements were 2.2 X lo6 and 0.66 X lo6, respectivcly. All four samples dissolved completely in
Tlic deterniination of the dimensions of polynicr toluene and in iricthyl ethyl ketone without evidence of a molecules from dilute solution viscosity nieasurc- gcl compouent. This observation constitutes strong evim i l t s has been demonstrated in a series of recent dence for a preponderantly linear structure essentially unhy branching. papers.2-5 This method depends in the first place interrupted The solvents used for viscosity and osmotic measurements on the specification of a solvent medium in which were reagent grade. They were carefully dried and disthe osmotic forces acting on the polymer molecule tilled before use. Fractionation .-Polyniers A and B were fractionated are exactly zero. When this condition is fulfilled, solutions in ethyl acetate at concentrations of 1 g./100 thc average configuration of the polymer molecule from i d . by the addition of methanol. After a sufficient quanin solution and its average dimensions are "un- tity of this precipitant had been added to produce a perperturbed" in the sense that they depend only on manent turbidity a t 30°, the solution was warmed until it the bond structure of the polymer chain including became homogeneous, then placed in a constant temperature bath at 30.0'. After stirring gently for 30 minutes, the influence of hindrances to free rotation about the precipitate was allowed t o settle for 24 hours, and the these bonds. Since the intrinsic viscosity depends supernatant liquor was separated by decantation. The directly on the volume occupied by the chain, its precipitated fraction was washed with methanol, dissolved in determination in an "ideal" solvent as specified ethyl acetate, filtered to remove accumulated dirt, then above affords a means for ascertaining the unper- transferred t o a weighing bottle from which the solvent was removed by evaporation, the process being completed in turbed dimensions of the chain. 2*6 vacuum a t 60 to 80'. LYashings were transferred t o the The intrinsic viscosities of polyisobutylene, * main solution, and the next fraction was separated in the polystyrene,2~5natural r ~ b b e r , gutta ~ p ~ r c h a , ~same rnauner. Thrcc fractions rvere separated similarly from polymer C, cellulose trihutyrate4 and cellulose tricaprylate the third being retained for use in the precipitation measurehave been shown to be in quantitative accord with iricnts only. A lower concentration of about 0.3 g./100 ml. the above interpretation, and their unperturbed of ethyl acetate was used in order t o effect adequate fracdimensions have been deduced in this manner. tionation a t the higher molecular weight. Polymer E was This paper presents the results of similar studies 011 suhjectctl to a rough preliminary fractionation. Three iniddlc fractions consisting of about 30% of the original ~)olydimet-hylsiluxanefractions. wcre combined and subjected to careful refractionation from
Experimental Materials.-Four polydimethylsiloxanes said t o have Ixen prepared under conditions which should avoid insofar a s possible the introduction of units of functionality exceeding two were used in this study. Two of these, having viscosity-average molecular weights of about 100,000 and 125,000, designated A and B, respectively, were made available to LIS by the General Electric Laboratories, Silicone llivision. The other two, designated C and E , were prepared by the Dow-Corning research group a t the Mellon Institute.? Their average molecular weights according t o
_____
(1) A portion of this investigation was carried out i n connection nith the Government Research Program on Synthetic Rubber under contract with the Synthetic Rubber Division, Reconstruction Finance Corporation. It is a pleasure to acknowledge also the receipt of partial support through the research program sponsored at Cornel1 University by the Alleaany Ballistics Laboratory, a n establishment owned by the United States Navy a n d operated b y the Hercules Powder Company under Contract NOrd 10431. (2) (a) T. G. Fox,Jr., and P. J. Iilory, J . Phys. Colloid Chem., 63, 107 (1949); THISJOURNAL, 73, 1909 (1951); (b) ibid., 73, 1915 (1951); ( c ) 1. Polymer Sci.. I, 745 (1050). (3) H. L. Wagner and P. J. ];lory, THIS J O U R N A 74, L , 196 (1952). (4) L. Mandelkern and P. J. Flory, ibid., 74, 2517 (1952). (5) L. H. Cragg, T. E. Dumitru and J. E. Simkins, ibid., 74, 1977 (1952). (6) P. J. Flory, J . Chem. Phys., 17, 303 (1949); P.J. Flory and '1'. 0. Fox, Jr., Tiirs JOUKNAL, 73, 1904 (1951). (7) We are indebted Lo Dr. R. 0. Sauer of the Silicoue Products Dept., General Electric Company and to Dr. E. L. Warrick of the Mel-
Ion Institute for generously furnishing us with the various samples.
a n ethyl acctate solution containing about 0.15 g. of polymer per 100 nil. The third fraction thus obtained was used in the osmotic and viscosity studies (Table I). Intrinsic Viscosities.-Dilute solution viscosities were measured according t o procedures described previously, aiid the extrapolation of the specific viscosity-concentration c 100ml./g.), toinfinitedilution wascarriedout ratio, ~ ~ , , /(in by established methods. The constant k in the Hugginss relationship %PlL
=
[v I
+k
[7712c
(1)
assumed abnormally high values of 0.7 to 0.8 in the poor solvents methyl ethyl ketone (MEK) and phenetole. Xormal values of about 0.35 were observed in toluene, a good solvent. Osmotic Pressure.-Osmometers of the Bureau of Standards design9 were Membranes were of gel cellophane. All Feasurements were carried out in methyl ethyl The surface tension of solutions in this solketone at 30 vent containing up t o 0.5 g./100 ml. of polysiloxane did not differ detectably from that of the pure solvent, according to the capillary rise method. Thus, the procedure adopted by Scott,Io involving the use of polysiloxane solutions differing in concentration on either side of the membrane proved to be uniiecessitry. Thorough cleaning of the capillaries with strong aqueous caustic solution after each measurement was essential, however; results obtained otherwise displayed anomalous irregularities. T h e fin21 osmotic pressure W'ZLSrecorded only after t h v - . . - ._ -. -
.
(81 XZ. L. Huggins, THIS JOURNAL, 64, 271G (1942). (9) S. G. Weissberg and G. A. Hanks, unpublished. (10) D. W. Scott, THISJ O U R N A L , 68, 1877 (1946).
MOLECULAR DIMENSIONS 01'POLYDIMETIIYLSILOXANES
July 3, 1952
TABLE I POLYDIMETHYLSILOXANE FRACTIONS; PRIXCIPAL DATA Polymer fraction
M X
Original 17 I polymer, yo MBK at 30°
IO-J osmotic
M X 10-J calculated from [ 7 ] in MEK a t 30'
To in MEK, OC.
c-3 E-3 B-2 A-5 A-6
0
14 2.24 4670 17.0 9 0.766 685 ( f 2 5 ) 664 16 .485 278 ( & l o ) 290 7.8 15 .350 160 5.5 13 .290 112 ( & 3 ) 113 A-7 8 .259 93 ( 1 3 ) 92.3 -1.2 A-8 11 .337 78.7 A-9 7 .192 53.6 Based on molecular weights interpolated from intrinsic viscosities measured
difference in height remained constant within ltO.2 mm. for 12 hours. After completion of a run the osmometer was removed from the solvent, the solution capillary was disconnected and the solution was removed. The cell was filled with MEK and allowed to stand overnight. I t was rinsed twice with portions of the next solution to be run, then filled and reassembled using a clean capillary.
Results The individual fractions used in this work are characterized by the data given in the first five columns of Table I. Molecular weights of four of the fractions were measured osmotically. The osmotic results are shown in Fig. 1where the experimental data are plotted as log (a/c) against log c. The curves correspond to the A/c
=
(A/ca)
[I
+ rlc + (j/g) rP2c*]
(2)
c being expressed in g./lOO ml. and a in g . / c n 2 . The theoretical curves have been matched to the experimental points by rectilinearly shifting a standard plot of log [l r2c 4- (5/8)I'2 c2] us. log (rZc) in the manner previously described."
+
[?
TO
1
h.1 EK
phenetole
a t 20'
KzoO
10'
0.815 (ztO.02) .815(=t .02) .785" . 8 2 0 ( f .01)
0.674 .427 .314 ,274
74.7 70.7
x
66.8
.78Y
.180 a t 30' in MEK.
The vertical displacement gives log ( a l c ) ~ ,froiii which the molecular weights given in the fourth column of Table I have been computed according to van? Hoff's law. In Fig. 2 log [7]h4EK,3O0 is plotted against the logarithm of the osmotic molecular weight. From the straight line drawn through the points we obtain the empirical relationship bg[?]MEK.300
E
-3.318
+ 0.55 log
(3)
which has been used to calculate the molecular weights given in column five of Table I. The extrapolation to fraction A-9, having a molecular weight only a little more than half that of the lowest osmotically measured fraction, is unquestionably justified by the behavior of other linear polymers; the much longer extrapolation involved in assigning a molecular weight for the highest fraction, C-3, is subject to greater uncertainty. This fraction was used, however, only for the measurement of Tc in a region where the maximum conceivable error in its molecular weight would be inconsequential.
L
t
O.0
I
-041
,
,
,
-0.5
0.0
0.5
L o g c. Fig. 1.-Osmotic pressure measurements: log r / c vs. c, with A in g./cm.2 and c in g./lM) ml. Curves are best fitted according to equation (2). Note difference in ordinate scale for fraction A-7. (11) T. G. Fox, Jr., P. J. Flory and A. M. Buecbe, THIS JOURNAL, 73, 285 ( 1 9 5 1 ) . (12) P. J. Flory, J . Chcm. P h y s . , 17, 1347 ( 1 0 4 0 ) ; P. J . Flory and \V. R . Krigbaum, ibid., 18, 1086 (1960).
I
I
5.0
5.2
I
5.4
Log
5.6
5.8
M.
Fig. 2.-log [ 7 ] n ~ 1 < , ? o ovs. log M . Lcngth of bar iiidicating each point corresponds to the uncertainty log A4 ill each case.
Precipitation temperatures were measured on solutions of several of the fractions in MEK and in phenetole a t suitable concentrations. These were chasen to cover in each case the region of the critical pDint, whichoccurred a t about 1g./lOOml.for the highest fraction, C-3, in MEK, and increased to about 8 g./100 ml. for the lowest fraction measured.
The critical, or consolute, temperature T,- was taken as the maximum in the plot of the precipitation temperature against concentration. Results are tabulated in columns six and sex-e11 of Table I . In accordance with prevous procedure,? these are plotted against in Fig. :3 'The critical miscibility temperatures 0 in the liniit of infinite molecular weight, obtained from the extrapolation of these plots to -11= 0, are 8% and ,LXj0I lf 11 I ,7 'rAULE Ii W
W c
i
t \
CY
drld a5 -
CYc
= J ~ ' \ I \ # I (1 - 0 7
I ) E T E R M I S A I ' I O S C)F J
The linear distortion of the average polymer c o w figuration owing to solvent-polymer interaction 15 represented by a , and is the mean-square endto-end length for the unperturbed chain; the parameter @ is believed to be a universal constant, CM is a constant for a given solveni polymer system, and 'klis an entropy of dilution parameter AltT = 8 equation (6) gives a: = I a5 the phyiic-ally valid solution, as is required by the fdet thdt the osmotic forces are then Lern IC follows that
2
1 7 1 ~=
h 21'
k' A 1 83'
(111
( t
Kesults of intrinsic viscosity measureinelits rndclt. in MEK a t T = 0, i.e , a t 20°, given in the next to the last column of Table I, are shown graphically in Fig. 4. A straight line has been drawn through the points in accordance with equation (4') Given i n the last column of Table I 'ire the values obtained for K ~ o otake11 c j \ ratio5 of [ q ] X l r lor each fraction rl%c chrectly tlc>trriiitiicxtl 0%-
I'ractioii
E-% $7
Iqli;' 0,:jSl
,283
1:SISC PHENElOLE
i~lw
[q]ii3
0.W
,:3(19
interpolated
0 416
,298
I< X 1 0 3
O.';!+(l.O
--
0%)
.,;I
.4y. tl.77
Discussion Coinprisons of intrinsic viscosities of polyisubutylenes and polystyrenes' with the values of 7' obtained froin the dissyniiiietry of light scattered by their solutions lead to @ = 2.1 X 10" when \';2 is expressed in cm. and M in molecular weight units. Taking I< = OS03 X for the polydiniethylsiloxane chain, we obtain according t o equation (a), (ii/M)'l? X 10" = '730. This quantity is numerically equal to the root-mean-square unperturbed end-to-end length expressed in Bngstriiin units for .I/ = lo6. Values obtained similarly lor other linear polyriiers are inclutletl iii the sccontl C ~ ~ U I IofI ~Table I IIT.
MOLECULAR DIMENSIONS OF POLYDIMET~IYLSILOXANES
July j,1922
3367
TABLEI11
ROOT-MEAS-SQUARE END-TO-END CHAIN LENGTHS COMPARED WITH
THOSECALCULATED FOR FREE ROTATION -
( r ; / M ) I / z x 10" Obsd. Calcd.
Polymer
Polydimethylsiloxane a t 20'; free rotation values calcd. for e2 = 130' Polydimethylsiloxane a t 20' ; free rotation value calcd. for ey = 160" Polyisobutylene2" a t 25" Polystyrene*ba t 34' Polymethyl inethacrylate2'at 31' S a t u r a l rubber3 at 0 to 60" Gutta percha3 a t 60' Cellulose tributyrate4 a t 45" Ccllulose tricaprylate4 a t 48"
-
($?,)I/*
Fig.5.-Structure 730
456L
1.60
730 795 725
530" 412' 302' 310' 485' 703" 408' 366'
1.38 1.93 2.4 2.2 1.7
680 830 1030 1090 850
1.5 2.7 2.3
= 1.65 A . and el = 1100. b~ = 1.54 A. and e = Calculated from appropriate bond lengths and 109.5". aiigles assuming free rotation about single bonds. C
~
Z
The polysiloxane chain consists of identical bonds of length 1 meeting a t angles 81 and 82 which alternate along the chain as shown in Fig. 3 . If free rotation is assumed about each Si-0 bond, it is possible to show that the mean-square end-to-end distance is given by
r;, = d 2 ( 1 -
e,)( 1 + COS on)/(1 - COS 8, COS 02) ( 7 ) when n, the number of chain bonds, is assumed to be large.13 (The added subscript included with ro refers to the assumption of free rotation.) XRay1? and electron diffraction studieslj on lower cyclic dimethylsiloxane pclymers give 1 = 1.63 i= 0.03 8. (1.64 f. 0.03 A. and 1.66 f 0.04 8., respectively), O1 Z 110' and 8 2 Z 130'. It may be argued that these values for the angles are of little significance for the linear polymers inasmuch as they were obtained on six-membered rings which COS
'If"
(13) The derivation of equation (7) parallels that employed to obtain the well known relation -
I{/ = 7212
(1
-
\
/o
cos 8 j / ( 1
+ cos 0 )
for a simple chain of identical atoms. See H. Eyring, Phrs. Rcu., 39, 746 (19321, and F. T. Wall, J . Chcm. Phys., 11, 67 (1943). When 81 = 8 2 , equation 17) reduces to this expression. (14) W. L. Roth, THISJOCRNAL, 69, 474 (1847); A d a K i i s f . , 1, 34 (1Y48). (15) E. H . Aggarwal and S. H. Bauer, J. Chcni. Phys., 18, 42 (1050).
of the siloxaiic chain.
are known to be strained.I6 The silicon angle 81 is quite certainly nearly tetrahedrallj in the absence of strain, hence the value 110' may be accepted as approximately correct. The Si-0-Si angle 82 probably is easily deformable,17 and the unstrained 8 2 therefore probably exceeds 130'. Sauer and Mead'si7 value of 160 f 15' for 82, based on dipole moments is considered too large, however.':, l8 Values of (i$M)'lz obtained from equation (7), with n replaced by 51/37, are given in the third column of Table 111 for two values of 8 2 , 130 and 160', which almost certainly are extreqies. The ratios of the observed root-mean-square unperturbed end-to-end lengths to those thus computed assuming free rotation are given in the last column of Table 111. ,41so included arecorresponding data obtained for other linear polymers. Although the average extension of the polysiloxane chain is considerably larger than that calculated assuming free rotation, the deviation is less than has been observed for most other polymers. This is in accord with the high degree of flexibility commonly attributed to the siloxane chain. I t should be borne in mind, however, that ordinary hindrance to rotation about the chain bonds is not the only - factor contributing to the increase of over r:f; "short range" steric interactions between near neighboring units of the chain will increase the observed dimensions over those calculated by the over-simplified approximation of independent free rotation about each bond. This factor alone may be largely responsible for the disparity bctween? and (calcd.) for the siloxane chain.
2
r5
ITHACA, XEK YORK (16) D. W. Scott, THISJ O U R N A L 68, , 2294 (1946). (17) R 0. Sauer and D. J. Mead, ibid.,68, 1794 (1946). (18) ADDED IN PRooa.-On the basis of X-ray studies on the cyclic octasiloxane and on frozen stretched silicone rubber, Dr. L. K. Frevel has selected the following as the best values: I = 1.70.k. and 0%= 130 * 10'. The authors are indebted t o Dr. Frevel for bringin,: theye results to their attention in advance of publication.
