Oct., 1939
THERMODYNAMICS OF ANILINE-NITROBENZENE MIXTURES
mary phases are (Na2,NaLi)SiOa solid solutions, Li2SiO3, NazSizO~,and LizSizOa, all three of which form solid solutions of limited extent, and the three modifications of silicon dioxide, namely, quartz, tridymite, and cristobalite. The liquidus fields meet at two ternary eutectics: one a t 697O, with (Naz,NaLi)SiOa, Li2SiOa and NazSiz05, the other a t 637') with Li~Si205,Na2Si205 and quartz as the eutectic constituents. LizSiz05melts incongruently throughout its region of existence in the system, the reaction temperature descending from 1033" in the binary system, LiSiOa-SiOz, to 641 O , the peritectic end-point in the ternary system, with Li~Si205,Li2Si03, NazSiiO5 and liquid in coexistence.
[CONTRIBUTION FROM THE
2877
The inversion temperature of quartz and tridymite has been redetermined. The temperature of 870 * loo, given by Fenner in 1913, is confirmed, the value obtained being 867 + 3 '. Refractive indices of glasses of various compositions in the system were measured. A discussion of solid solution relationships of sodium and lithium compounds in general is given, with particular reference to the theoretical aspects of the subject. Minor revisions of the phase relations in the systems, Na2SiO&3Oz and Li~Si03-SiOz, particularly with respect to the polymorphic behavior of NazSi206 and LizSi206, are presented. WASHINGTON, D.
GEOPHYSICAL LABORATORY, CARNEGIE
C.
RECEIVED AUGUST2, 1939
INSTITUTION OF WASHINGTON]
Pressure-Volume-Temperature Relations in Solutions. III. Some Thermodynamic Properties of Mixtures of Aniline and Nitrobenzene BY R. E. GIBSONAND 0. H. LOEFFLER A few months ago we noted' that the absorption of light by mixtures of aniline and nitrobenzene was strongly influenced by pressure changes. In investigating this effect we thought it desirable to examine thoroughly the volume changes which take place when these two liquids are mixed in different proportions a t different pressures and temperatures. Furthermore, we have noticed in the pure liquids2 that, whereas the internal pressure ( b E / b V ) T varies with temperature a t constant volume, another quantity, P A , computed from the P-V-T data is independent of temperature a t constant volume and is expressible as a'/ v". We have tentatively identified PA as the attractive internal pressure of the liquid. In this paper we shall give data from which the volumes of aniline-nitrobenzene mixtures may be determined a t all concentrations, between 25 and 85" and between 1 and 1000 bars, and we shall examine the behavior of PA in these solutions.
Experimental Results The solutions were all made up independently from weighed amounts of samples of aniline and nitrobenzene purified to a degree which already has been described.2 The specific volumes of (1) R. E. Gibson and 0. H. Loeffler. J . Phgs. Chum., 48, 207
(1939). (2) R . R . Gibson and 0 T I . T,oeffler, THISJ O V R N A I . , 81, 2.515 (1939).
these solutions were determined a t 25.00 O in 55-ml. U-tube pycnometers, and the thermal expansions were measured8 a t 10' intervals from 25 to 85" in a weight dilatometer of 18 ml. capacity made of vitreous silica. The volumes of the solutions were expressed as functions of the temperature by empirical equations of the form =
v65
f a(t
- 55) + b(t - 55)'
+ c(t - 55)'
(1)
We already have published for the pure components such equations whose coefficients we may call u:, ui, b:, bi, etc. From them we computed4 ATv! and ATv; at each temperature, using 55" as the base temperature, and thence the quantity ATv - (xIAT~: x ~ A ~ = o ~8 ) ~ . This function measures the departures of the thermal expansions
+
(3) For details of the procedure in these measurements the reader is referred to the second paper of this series (ref. 2). (4) The symbols used here are as follows. The subscripts 1 and 2 refer to nitrobenzene and aniline in solution, respectively, the superscript 0 implies the pure components, and symbols without subscripts refer to the solutions. The weight and mole fractions are given by x and X ,respectively, Rz = Xz/Xi, and Y and V mean the specific and molal volumes. For a solution V = v/(xi/Mi x * / M z ) , M being the molecular weight. 9 is the apparent molal volume of a component. Av and AV are the volume changes on mixing per gram and per mole of solution. All volumes are given in milliliters. The pressure in kilobars is given by P, the temperature by 1 (centigrade scale) or T (absolute), and the total energy by E . AT and A p denote the finite changes with temperature and pressure, respectively, of the quantities to which they are prefixed. k = - a p v / v p = ~ . zI 1/(1 R%fl/Vp), z.2 = ( 1 - 21) are thevolume frartioas of the coniponentr
+
-
+
R.E. GIBSONAND 0. H.LOEFFLER
287%
Vol. 61 TABLEI
REPRESENTING THE ANILINE-NITROBENZENE MIXTLTRES AS FUNCTIONS OF THE TEMPERATURE (EQUATION 1) The last column gives the value of the principal coeficient in equation ( 2 ) . COEFFICIENTS IN
THE EQUATIONS
SPECIFIC VOLUMES OF
Mole fraction of aniline
Weight fraction of iniline
?
0 0000 13153 24925 36438 49517' 62103 74972 86062 1 0000
B O
ca -
I +
$ 0
%
k+
. I
i
.$ 0 aW
+ -
45 65 85 Temperature, "C. Fig. 1.-Deviations of the values of the specific volumes of aniiine-nitrobenzene solutions as calculated from equation (1) from the observed ones a i different temperatures 25
from the ideal law of mixtures, and it may be represented accurately by the relation 871
= i
+ sn(t - .i.i)
10 b 6 931 7 241 7 519 7 791 8 100 8 398 8 701 8 963 9 292
10Qc 1 528 1 601 1 666 1 730 1 802 1 873 1 944 2 006 2 083
lOgm
-0 0191 0303 - 0382 - 0413
-
-
0201 0483 0358
ponents2 and the relative volume changes were expressed as functions of the pressure by the Tait equation k = C log [ ( B P ) l ( B Po)1 (3) where C has the same value as for aniline and nitrobenzene, uiz., 0.21591 at all temperatures. The constant B was determined at each temperature from the compressions to 1000 bars. The compressions to 500 bars computed by the appropriate Tait equation always agreed with the observed ones within experimental error. The compressions were measured a t 25, 45, 65, and 85", and the constant B was expressed as a function of the temperature by equations of the type
=; .ix
+;.x, +-
+
ri
))I,
= ci/,'t 4-
r,c;
fX2C;
?!;,
=
zohrd.
( 5 ) See Ostwald-Luther, "Phpsikochemischer Messungen." 4th Akademische Verla~sgesellschaft,Leipzig, 1925, p 39.
=
R,i
+
-+ a ( f - 25) + B(t
- 25)2
(4)
which were fitted to thc data at 2.i, 4.5, aiid hi' In Table [I we ,give the values of 73?,, (Y and /? for 1 ABLE I1 COEFPICIENrS IN EQVATTON (4) REPRSbBNTINL r U E C O N STAN7 B Ih' TAE TAITEQUATlON A \ A T.'UNCTION Or rEh1 ITRATLIRE
xi/>:
-t 1 The coefficients a, b, c, and w are given 111 'Lable I aiid the deviation curves in Fig. 1 indicate how well the equations fit the data. It is of interest to note that the coefficients of the second and third powers of the temperature are found directly from the simple law of mixtures. The compressions of the solutions from 1 to 500 and from 1 to 1000 bars were measured in exactly the same way as those of the pure comc =
n
(2)
In these solutions [.XI bTzf: x2ATz1:) is always numerically greater thaii Alv. The coefficients 1 and fit were determined by a simplified method of least squares5 It will be seen that the coefficients in equation (1) were theretorr, obtainable From the following relations
I
16684 30506 43117 56461 68422 79842 89088 1 0000
Em 10% 0 855751 7 274 877160 7 465 895974 7 643 914104 7 819 934311 8 025 95339gb 8 218 972652 8 425 988850 8 615 1 008748 8 874
+
0
ed
0 0000
The thermal expansions of this solution were measured only to 65 ', and the compressions onIy a t 25" * Mean of two entirely independent determiaations.
a -
v
II
x2
*2
+
1l)dH iobsrl
B26 I,
0 0000
1335.: 24925 30438 ti2103 74972 ,86062 I 0000
(kilobar.;)
calcd )
-
IO'n
10 3
at fii"
1 8652 1 8740 1.8843
9 ti3
1 18 1 72
$(I
2
-
>
9 06
1 86
1.8975
10 fi8
iz 7 tl 2
1.9320 1.9536 1.9768 2 0066
10 21 10 48 10 82
1 07 1 64 182 1 82 1 99
9 8.3
10 09
-1 1 -3 b fl i; -1: .;
the various solutions together with the differences between the observed and calculated values of B at 65" as an illustration of the fit of the empirical equations
Oct., 1939
THERMODYNAMICS OF ANILINE-NITROBENZENE MIXTURES
0.76
2879
plotted log P A vs. log V for all the solutions including the end members, and in order to facilitate a comparison of Figs. 2 and 3, we have used the same scale in both diagrams. It will be seen that in Fig. 3 the points for the different temperatures all fall on the same straight line for each solution and that the lines for the different solutions all have the same slopes. Over the region we have studied we may, therefore, write
0.72 0.68
80a70
e3 0.66 v
8
4 0.69
0.65
PA =
a'/VZ.737
(5)
2.737 being the constant slope of the 0.68 lines in Fig. 3. The constants a' were determined from the data and average values are given in Table 111. 0.64 It should be noted that the points a t Log v. 65 and 85" for nitrobenzene lie on a Fig. 2.-The logarithm of the internal pressure ( d E l d V ) ~of some line which is slightly below that d r a m aniline-nitrobenzene solutions as a function of the logarithm of the mold though the points for the lower volume. Note the different curves for the different temperatures.
The Internal Pressures ( b E / b I r )and ~ Related Quantities From equations (1) and (3) the expansion coefficients (bv/bT)pand the compressibilities were computed a t different pressures and temperature9 and hence the coefficients ( b P / b T ) , and (bE/bV)Twere 0.92 calculated readily for each solution a t 25, 45, 65, and 85", and a t 1, 250, 500,750, and 1000 bars. The curves of (bE/bV)Tvs. V , or of ( b P / b T ) , Ao.H vs. V , for the solutions resembled 4 those for the pure liquids in all major
peratures. We are not certain that this deviation is a real one and that it does not arise in the equations used for computing the derivatives; but it is significant that the same deviations are found in the solutions rich in nitro-
+ ~
respects2; in particular, (bE/bV )T decreased with rise of temperature a t constant volume. A sample of our results illustrating this temperature effect is given in Fig. 2 where log ( b E / b V ) Tis plotted against log V for several solutionp. Applying the method of analysis that we developed for the pure liquids to the results for the solutions,2 we computed the quantity (T(dP/bT)V -k B ) = P A which we have interpreted as the attractive internal pressure set up by the intermolecular forces, (B + p) being assumed to be a meas-
5 3 0.76
(1) NITROBENZENE
0.68
1.95
1.99 2.01 2.03 Log Fig. 3.-The logarithm of the attractive internal pressureP.4 = ( T y B ) of aniline-nitrobenzene solutions as a function of the logarithm of the molal volume. Note that the points for different temperatures fall on the same line for each solution. The scale in this diagram is the same as in Fig. 2. y represents the coefficient (dP/dT)v.
ure of the repulsive pressure, In Fig. 3 we have (6) For details of theam calculations, see ref. 2.
1.97
v.
+
benzene and gradually vanish as the concentration of aniline increases.
R. E. GIBSONAND 0. H. LOEPFLER
2880
VOl. 61
TABLE111 therefore, assumeds that Plz was equal to a conGENERALIZED FRACTION YI, AND A COaaPARISON OF stant, viz., (PllP22)*,and, using the relation VALUES OF a' DETERMINED DIRECTLY FOR EACH Pi = (YlP? Y2PjJ4 (8) SOLVTION (EQUATION 6) AND THE VALUESOF a' ComPUTED FROM EQUATION (9) we computed U,/U, from OUT results at 25" and AT x i n '
THE TnE
+
31
0.0000 .13153 24925 ,36438 .62103 ,74972 86062 1.u000
1.0000 0.90398 ,81113, .71325 .46527 .32250 .E3761
1 ( I R a ' (obsd J
LO%' (calcd 1
at 26"
2.079 2.011) 1.05,J 1.841 1.796 1.764.
134 218 272 279 235 155
2.163 2.071 2.004 1 9-17 1.840 1.797 1.766 1.739
1 bar and found it to be practically independent of concentration having the average value 0.5263.9 Combining equations ( 5 ) , (6), and (8) we get equation (9)
which should enable us to calculate a' for any solution in terms of the constants all, an, V,",and 1:'. for the pure liquids, the ideal molal volume of the The Variation of P A with Concentration solution and the fraction y1 which contains the We shall now attempt to calculate PA and a' for disposable constant U2/U1. In Table I11 we list the solutions from a knowledge of the properties values of yl computed on the assumption that of the pure components. If V is the volume of 1 U2/U1 = 0.5304 (the reason for the choice of this mole of solution,d Vo the volume of the correfigure, which differs slightly from that given sponding amounts of the components (i. e., the above, will appear under the discussion of the ideal molal volume of the solution) and if PA is volume change on mixing) together with the the observed attractive pressure in the solution, values of a' obtained directly from our data for then Pi,the attractive pressure of the solution the solutions and values of a' computed from a i rolume Vo,is given by equation (9) when the volumes at 25" and 1 bar p~(~~00)2.731 = pAV8.737 = a' (6) were used. The agreement between these two series of figures indicates that equation (9) does If, furthermore, we suppose the attractive presgive an adequate representation of a' as a function sures set up between like molecules in the solution of concentration. Moreover, Vo/17: changes very to be the same as in the pure liquids, viz., Pil for little with pressure and temperature, the extreme nitrobenzene and P22for aniline, and call the atvariation over the range we have studied being tractive pressure between the unlike molecules less than 0.4%. We may, therefore, conclude PIP, we may write that yl, y2 and hence U2/Ul are independent of Pi = Y;Pu YiPaz 2ylyePlr (7) pressure and temperature within the limits of where y1 and y p axe concentration functions that accuracy of our results. indicate the probabilities of significant interactions between the molecules,' and are defined as The Volume Changes on Mixing yI = 1/(1 R2U21UJand y z = (1 - 4'1). The The changes in volume which occur when the parameter Uz/UI may be regarded as an empirical components are mixed in different proportions a i constant which is equal io unity when y is the different pressures and temperatures may be mole fraction and is @/V! when y is the volume expressed conveniently in terms of the apparent fraction. ( 8 ) See G. Scatchard, Tram Faraday Soc.. 58, 162 (1937); S . E. 'When the mole fractions or the volume frac- Wood, THISJOURNAL, 59, 1511 (1937); J. H. Hildebrand and J. M tions were used in equation (7) no results of any Carter, % b a d ,54, 3599 (1932). (9) I n terms of the principle of independent surface or group acspecial interest were found, and the computed tion [I. Langmuir, Colloid Symposrum Monograbh, 111, 48 (1925). values of P ~ zvaried with concentration. We, J. A. V. Butler, D. W.Thomson. a n d W . H.MacLennan, J. Chem.
.oooo
+
+
+
follows t h a t proposed by oes not limit Y I and YZ t o the volume Fractions based on the volumes of the liquids under ordinary conditions. W e also wish to point out that our method of analysis impIies that P A depends chiefly on the mean distance between the molecules and is not greatly inffuenced by their distribution in the liquid, whereas the repulsive internal pressure depends strongly on the distribution of the molecules in the liquid. This seems Dlausible because the attractive forces between the molecules vary 'fsrich b y 5 t$+)pidlywith 81stnnPa %he* ?e the repulsive fot@ae*
(7) This treatment, which Scatchard [Chem. Rev., 8, 321 (1
Soc., 674 (1933) 1 YI and yx may be regarded as the voiume fractions of the active groups in the nitrobenzene and aniline molecules. An examination of the parachors of benzene, nitrobenzene, and aniline suggests that the mold volume of the NHz group is approximately one-half that of the NOi group fLandolt-B6rnstein, "Tabellen," Ergantung.Jband II(a), p. 178). Roughly the same result i s obthe molal volumes of benzene, nitrobenzene, snres where the liquids have the same compressibilities (ref. 1). Them is, therefore, some justification for associating y i and yz with I he volume fractions of the Pins and N I T 2 graups in the ~nllittsna.
Oct., 1939
THERMODYNAMICS OF ANILINE-NITROBENZENE MIXTURES
2881
sum B' of the solutions and BI and Bz of the pure components by the relation A V = V - XI^ -I- X Z E )then A V = Xz(*2 E) AV/C(XiVi -I- QU;) - (21 log Bi zz log BJ log B' (10) In the aniline-nitrobenzene system an expansion It was assumed that, on forming the solution occurs when the components are mixed, (% - at atmospheric pressure, the pure components V t ) being always positive. Increase of pressure first expanded or contracted until their repulsive Vi), pressures came to the common value B' and and of temperature both diminish (@2 an observation to which special attention is then mixed without further change. We indidrawn. cated also that the values of log B' computed from A clue to the origin of these volume changes the observed values of Av by means of equation may be obtained from an examination of the effect (10) were smuUer than the observed values of of temperature on the volume changes on mixing log B'. As equation (10) represents the volume when the volumes of the unmixed components change on mixing on the assumption that mixing are kept constant. In Fig. 4 we have plotted introduces no change in the distribution of the A V against V o for various solutions. It will be molecules, it is suggested that the discrepancy seen that at constant volume, and presumably constant attractive pressure, A V decreases quite rapidly as the temperature rises, and that this decrease is roughly proportional to A V itself. E Having in mind the interpretation we 5 tgave of similar changes in the (bE/ 'ij; b V)TVcurves, we suggest the follow- .f!o-26 ing explanation of this decrease of A V with rise of temperature at constant 'g V. We assume that the change with 8 temperature in the randomness of the $0.18 distribution of the molecules in a liquid is proportional to the order already existing in the liquid, and that 81 in a solution of aniline in nitrobenzene 5i +O,JO the molecular distribution is more random than in the pure liquids at the same temperature (the increase 89 93 97 101 105 107 Molal volume, ml. in disorder determining to some extent the volume change on mixing). Fig. 4.-The volume changes on mixing per mole of solution in four solutions of aniline in nitrobenzene as functions of the volume at differWe should, therefore, expect that a ent temperatures. given rise of temperature would have a greater disordering effect on the unmixed liquids between the observed and calculated values of than on the solution, and that at high tempera- B' is due to an increase in randomness on mixing. tures the change in randomness on mixing should In other words, on the formation of the actual approach zero, and the volume change on mix- solution a given volume increase ( Av) results in a ing should decrease accordingly. The change of smaller decrease in log B' than would be expected (B P) with temperature a t constant volume if no increase in randomness accompanied the has been associated* with the change in the re- mixing of the components. It is of interest to note pulsive pressure due to increase in the random- that the values of B' computed by equation (10) ness of the molecular distribution. Our data show approach the observed ones more closely as the that (B P)v increases more rapidly with tem- temperature is raised. These effects are small perature for the pure liquids than for the solutions. but they contribute qualitatively to the evidence In a recent paper' we attempted to express Av correlating an increase in the randomness of the in terms of the net internal (or repulsive) prec- molecular distribution with an increase in the volumes of one of the components. It will be recalled that if
-
-
,
!.
4
+
+
+
R. E. GIBSONAND 0. H. LOEFFLER
2882
+
quantity B' or (B P ) which we have tentatively identified as the repulsive internal pressure of the liquid or solution.
-!
--
-
I----
'
. - ~ - ~ ~ .-_ - -
--EQUATION 12'
~
1
Vol. A 1
terms of the general fractions yl and y2. It should also be remarked that the usefulness of equation (1 1j is diminished somewhat by the fact that zl, 32, j , and q all vary with temperature and pressure. d trial of equation (12) (or its equivalent form equation 12') which is analogous to equation (11 ) showed that it gave a good but not altogether satisfactory representation of our data. l1 (@I
+
AV/(X,Ul XZUZ) = KYlY* = ky, = k / ( l R,VJUi)
-
E)
+
(12) (12')
The constants k and Uz/l71 in equation (12') determined from the data by the method of least squares1* are given in Table IV and the differTABLE Il' THECONSTANTS k AND us/ ui AS OBTAINED BY THE METHOD OF LEASTSQUARES FROM EQUATION (12') REPRESENTING THE VOLUME CHANGEON MIXINGAS A FUNCTION OF THE GENERALIZED VOLUMEFRACTIONS AT DIFFERENT PRESSURES AND TEMPERATURES
--
0.40 0.60 0 80 Weight fraction of aniline. Fig. 5.-Deviations of the values of the volume change on mixing in aniline-nitrobenzene solutions as computed from equation (12') or (13) from the observed values at different pressures, temperatures and concentration5 0.20
In order t o represent the volume changes on mixing as functions of the composition of the solutions, we used equations or the type"" A 1'/ ( X I I.:" + X?17;) = pziZz(1 + q z i ) (13 ) where z1 and z2 are the volume fractions, to represent our data. At 25" with p = 0.01230 and q = 0.341, equation (11) reproduced the results quite well, the average deviation and the largest deviation corresponding to errors of 1.6 and 3 X respectively, in the specific volume. Equally satisfactory fits were obtained at the other temperatures and pressures. The term p; in equation (11) shows that the curve of AI;/(XlV: X2E) vs. z1 is unsymmetrical, the maximum lying between zl = 0.5 and zl = 0. As this asymmetry effect of concentration on the absorption of light and on the attractive internal pressures of mixtures of nitrobenzene and aniline, it is of interest to examine equations which express the dependence of AV on concentration in
+
(10)G.Scatehard. S E. Waod and J. M. Mochel, J . Phys. Chcm., 411, 122 (1939)
t, 'C.
25 45 65 85 25 45 65 85
P (bars)
1 1
1 1 1000 1000 1000 1000
k
UP/ u t
1.0428 1.0297 1.0077 0.9826 .8694 .8023 .7851 .7649
0.6600 ,6922 ,7164 ,7299 ,6625 ,6265 ,6535 ,6746
Average AW (obsd. calcd.) X IO&
3.3 2.8 2.2 2.4 3.0 2.4
3.8 3.0
ences between the observed and calculated values of Av at 25" and 1 bar are plotted in Fig. 5 . The largest deviation corresponds to an error of - 7 X 10-6 in Av. The deviation curves for the results a t the other temperatures and pressures are the same as the curve we have given in Fig. 5 within in Av. In order to give an an error of 2 X idea of the numerical value of Av we have recorded Av a t 25' and 1 bar in the last column of Table 111. The parameter l J t / l J ~as obtained from equation (12) varies greatly with temperature and is considerably larger than we found from the attractive pressure equation. In view of our discussion of the effect of change in molecular distribution on the volume change on mixing, of the complicated nature of this quantity, and of the fact that the compressibility of the solution is an important factor in determining the volume
-
(11) From equation (12) it may be seen that (Vs V 3 = by2. Va being the partial molal volume. The analogy with Hildebrand's equation for regular solutions is evident [THIS JOURNAL, 61, 78 R2Uz/U1), this is the same (1929)l. Furthermore, as yi = 1/(1 type of equation that was proposed by van Laar [Z.Dhyatk. Chem 73, 723 (19io)j. (12) We are indebted to Dr. W. Edwards Deming of the U. S Department of Agriculture for valuable suggestions concerning these calculations.
+
Oct., 1939
THERMODYNAMICS OF ANILINE-NITROBENZENE MIXTURES
2883
change on mixing,13it seemed legitimate to regard viation of approximately -7 X in Av in the equation (12) as too simple to be of much physical 62% solution. Experimental checks have failed significance and to examine the behavior of UZ/UI to show that there is any reason to suspect the in some empirical equations. original data and, as this point occurs near the The parameters k1 and UZ/U1 in equation (13) middle of the concentration range, its importance were evaluated by a least squares method at 25, cannot be minimized. However, the simplicity 456.5, and 85" both a t 1 and 1000 bars. of equation (13) together with the fact that it yields a value of U,/U1 which not only is inde( 9 2 - Vg) = Vkiyi/log (B' P) (13) pendent of temperature, pressure, and concenThe results are given in the third and fourth tration but also is the same as appears in the columns of Table V. Although UZ/UI obequations relating the attractive internal pressures and the optical absorption coefficient^'^ to the TABLE V THECONSTANTS kl AND u2/ulIN EQUATION (13) AS OB- compositions of the solutions, gives the equation TAWEDBYTHEMETHODOFLEASTSQUARESAT DIFFERENT a significance which justifies further study of its TEMPERATURES AND PRESSURES. IN THE LASTCOLUMN applicability. ARE GIVEN AVERAGEVALUES OF kl CONSISTENT WITH The quantity - ( A P @ z - ApV,") which is a A VALUE OF U2/Ul INDEPENDENT OF PRESSURE AND measure of the departures of the compressions TEMPERATURE ki X 10' 1 to 1000 bars) from the ideal law of mix(from t, O C . P (bars) ki X 108 u2/u1 (Us/Ui = 0.5304) tures is always positive in nitrobenzene-aniline 25 1 0.5207 2.744 2.780 solutions. This is an example of a case where the 45 1 .5346 2.201 2.199 solutions are more compressible than the pure 65 1 .5405 1.668 1.657 85 1 ,5258 1.152 1.162 components and therein it differs from aniline25 1000 .5609 4.033 3.940 benzene or aniline-chlorobenzene mixtures. As 45 1000 ,5308 3.444 3.454 far as we can tell from our data - (A&z - A P E ) 65 1000 ,5530 3.095 3.055 for each solution is independent of the tempera85 1000 ,5785 2.766 2.669 ture; and although - (A& - A P E ) may be caltained in this way is very sensitive to small fluc- culated satisfactorily from a simple modification tuations in the specific volume-concentration of equation (13), we may say that for interpolation data, the values we obtained were practically purposes the equation - - ( A & - ApV,") = constant over the whole temperature range a t 1 0.1899 X I gives a simple and adequate expression bar, and varied no more a t 1000 bars than might of the compression data a t all temperatures. be expected from the greater uncertainty of the The fact that when aniline and nitrobenzene experimental data at the higher pressures. The are mixed only a small amount of heat is abaverage value of U2/Ul a t 1 bar, which was found sorbed and an expansion occurs, together with the to be 0.5304, was assumed to be a constant char- fact that this expansion diminishes as the temacteristic of the liquid mixture and independent perature rises, is contributory evidence against of the pressure, temperature, and concentration. compound formation in the ordinary sense of the From it we computed the values of y1 given in term and the volume relations are such that OI Table 111. In the fifth column of Table V we any simple hypothesis we must consider that if a give the average values of kl consistent with compound were formed in these solutions, it U Z / U I= 0.5304; and in Fig. 5 we have plotted would be destroyed by rise of pressure. the differences between the observed values of Summary Av and those computed from equation (13) with these adjusted values of y1 and k1. The deviation We have measured the specific volumes a t 25', curves for the different pressures and tempera- the thermal expansions a t 10" intervals between 25 tures being very much alike, we have given only a and 85", and the compressions to various presfew examples. These curves indicate that equa- sures up to 1000 bars a t 25, 45, 65, and 85" of tion (13) does not represent the volume change on six solutions of aniline in nitrobenzene covering mixing as well as might be desired; indeed, the the whole range of concentration. These data deviations are only slightly smaller than those have been represented by suitable equations from given by equation (12'). There is always a de- which the volumes, the thermal expansibilities
+
(18) R. E. Gibson, THIS JOUXNAL, 69, 1621 (1937).
(14) Unpubli8hed reaulta obtained in thio Laboratory.
T. r,.
2884
and the compressibilities may be computed a t any pressure and temperature within the range of our observations. Our data enable us to compute ( W / b T ) v , ( bE/bV)T and the volume changes on mixing for the various solutions, and to examine the variation of these quantities when the temperature is changed a t constant volume. In the light of the results of this analysis, we suggest a correlation between the effect of temperature and of mixing on the molecular distribution in the liquids on the one hand and the internal pressures and the volume changes on mixing on the other The attractive pressures of the solutions were calculated in the same way as those of the pure liquids and were reproduced by the formula PA = a’/V2.737 An equation was developed whereby the constants a’ for the solutions were represented as a function of the concentration in
~ C O N T K I B U T I O NI‘IIOM l”h
terms of constants for the pure components and ctions y~ and y2 which contain an empirical constant, lJz/ U1. Although the volume changes on mixing were well represented as functions of the ordinary volume fractions oi the components by equations of a conventional type, we also examined several equations expressing this quantity in terms of significant properties of the solutions and the fractions ?‘I and y p These equations reproduced the volume changes on mixing fairly well. One oi them had the form of the equations developed for regular solutions and another gave a value of the empirical constant, U2/Ul, which was independent of temperature and pressure and was approximately equal to the corresponding constant obtained from the attractive pressurecomposition equatioii. WASHINGTON. 1) C.
DBPAXIMENT O F PHYSICAL
Senun Albumin. I. The
Vol. til
MCMEEKIN
CHEMISTRY,
RECEIVED JULY 19, 1939
HARVARD MEDICAL
SCHOOL]
and Properties of Crystalline Horse Serum onstant Solubility BY T. L. MCMEEKIN
Since the crystaliization of serum albumin by Giirber in 1894, many investigators have doubted its homogeneity. Fractional crystallization from salt solutions has roducts that vary widely in solubility,2 composition,3v4and dielectric properties6 Despite this evidence of inhomogeneity, the recognition and separation of constituent proteins has proved difficult. Perhaps the most important advance in its purification was the separation of carbohydratecontaining globulins6 from the albumin fraction and the recognition that serum albumin does not contain carbohydrate. 3 , 4 Serum albumin has been reported t o be homogeneous in resp size7 and electrophoretic mobility.* Ultra gal studies have shown it to have a m eight of 67,500,’ independent of pH between 4 and 9. In solutions (1) DUrber, WUrsbur
Ges., 118 (1894). euldirg, 18, No. 5 , 1 (1930).
(6) Ecwitt, Biochem. I ,98, 26 (1938). (7) Svedberg and Sjogren, THISJOURNAL, SO, 3318 (1928); 114, 2856
(1930).
(8) Tiselius, Biockem. J., 81, 1464 (1987).
more acid than pH 4 it was decomposed into smaller molecules, the process being found to be reversible. This paper first describes our present procedure for the separation of the serum proteins and the preparation of crystalline carbohydrate-free serum albumin homogeneous with respect to electrophoresis at neutral reacti0ns.O This highly purified serum albumin can be separated further into two crystalline fractions differing in chemical composition and electrophoretic mobility in acid but not in neutral s o l ~ t i o n . ~The separation of serum albumin into two fractions is based on the discovery made in this Laboratory by J. I). Ferry,Io that a salt-free concentrated solution of serum albumin crystallizes when brought to $H 4.0 with sulfuric acid.’l The solubility of the (9) I am indebted to Dr. John A. Luetscher. Jr.. for these measurements. See also the second paper in this series, THIS JOURNAL, 61, 2888 (1939). (10) Reported by Ferry to the Division of Biological Chedstry at the 96th meeting of the American Chemical Society, Milwaukee, Wis., September 7, 1938. (11) Piettre, Comfit. rend. Acud. Sci., 188, 463 (1929). reports that among other proteins Serum albumin crystallizesfrom aqueous solution following removal of globulin with acetone, and of acetone by evaporation over sulfuric acid in vacuo.
