loa
J. Chem. Eng. Data 1983, 28, 108-112
Total-Pressure Vapor-Liquid Equilibrium Data for Binary Systems of Aniline with 1-Chlorobutane and Ethyl Acetate Jagjlt
R. Khurma, 01 Muthu, Sarat
Munjal, and Buford D. Smith”
Thermodynamics Research Laboratory, Washington University, St. Louis, Missouri 63 130
Table I. Chemicals Used
Total-pressure vapor-llquld equillbrium (VLE) data are reported for the following two Mnary systems: 1-chiorobutane anlllne at 298.20 K, and ethyl acetate aniline at 297.49, 348.23, and 397.89 K. The experlmental PTx data were reduced to y,, Y,, and G E values by both the Mlxon-GumowskdCarpenter and the Barker methods, but only the Mlxon et al. results are reported in thelr entlrety. Six GE correlatlons were tested in the Barker data reduction. Varlous equations of state were used to estimate the vapor-phase fugacity coeffIcients.
+
+
component
vendor
stated purity, 7%
1-chlorobutane ethyl acetate aniline
Burdick and Jackson Burdick and Jackson Aldrich Chem. Co.
99.9 t 99.9 99.9+%
Table 11. Experimental P vs. x , Values for the 1 Chlorobutane ( 1 ) t Aniline (2) System
Introduction
.oooo
This is the third paper reporting total-pressure vapor-liquid equilibrium (VLE) data on binary systems containing aniline. The first paper ( 7 ) described in detail the apparatus and techniques used for the experimental measurements and reported data on the ethanol 4- aniline system. The second paper (2) covered aniline with acetone, acetonitrile, chlorobenzene, methanol, and 1-pentene. The defining equation for the activity coefficient and the definition of the standard states used are given in the first paper cited ( 7 ) .
Chemlcals Used The sources and purities of the chemicals used are listed in Table I. Activated molecular sieves (either 3A or 4A) were put into the chemical containers as they were received. Just prior to being loaded into the VLE cells, the chemicals were poured into distillation flasks and distilled through a Vigreux column (25” 0.d. and 470 mm long). I t has been found useful to distill aniline for the second time just before loading the cells. The first and last portions of the distillates were discarded. The retained portions were caught in amber bottles and back-flushed with dry nitrogen for transfer to the cell-loading operation. The stated purities of the chemicals were verified by gas-liquid chromatography at this point. None of the compounds exhibited any degradation during the VLE measurements. The cell pressures were stable with respect to time, and all liquids were perfectly clear when removed from the cells at the end of the last isotherm.
Experimental Data Tables I 1 and 111 present the experimental PTx data. The “smooth” pressure values reported are from the least-squares cubic splined fits used to provide the evenly spaced values required by the finitedifference Mixon-Gumowski-Carpenter method (3) for reduction of PTx data. Figures 1 and 2 show the experimental data in terms of the pressure deviation P, from Raoult’s law P, = P
-
[P2‘
+ x l ( P 1 ’- P 2 ’ ) ]
04 78 0953 1657 2232 3108 .4444 5259 ,6205 72 19 8065 8576 ,9282 ,9617 I 0000 4
0.077 1.880 3.354 5.042 6.106 7,337 8.747 9.416 10.136 10.895 11.582 12.047 12.784 13.177 13.657
0.077 1.891 3.35 1 5.046 6.104 7.336 8.739 9.424 10.13R 10.892 11.580
12.049
12.785 13.176 13.657
where P is the experimental mixture pressure and the Pi’ values are the pure-component vapor pressures. The deviation pressure plot emphasizes the scatter of the P vs. x , data but does not show whether or not an azeotrope exists. The point symbols in Figures 1 and 2 denote the experimental data points exactly. The curves approximate the cubic splined fits of those data points. I t was not possible to measure data for the 1-chlorobutane aniline system at 348 and 398 K because of a chemical reaction between the two compounds. At 298.20 K, the deviation from Raoult’s law was positive at all compositions. No azeotrope was formed. At 297.49 K, the ethyl acetate aniline system deviates from Raoult’s law in the negative direction from x , = 0.0 to about x 1 = 0.92, and in the positive direction above x , = 0.92. The deviation is negative at all compositions at 348.23 and 397.89 K and the negative deviation becomes progressively stronger as the temperature increases. I t is likely that the deviation for this system is entirely positive at temperatures somewhat below 297.49 K. The system did not form an azeotrope at any of the three temperatures studied
+
+
Reduced Data The y,, y,,and GE values selected for publication are in Tables I V and V. Those values were obtained with the Mixon et al. data reduction method. The Peng-Robinson equation of state ( 4 ) was used to estimate the vapor-phase fugacity coefficients. The parameters used for the Peng-Robinson equation are in Table V I . The “experimental” pressure values tabulated in Tables I V
0021-9568/83/1728-0108$01.50/00 1983 American Chemical Society
Journal of Chemical and Engineering Data, Voi. 28, No. 1, 1983
Table IV. Calculated Data for the I-Chlorobutane (1) L I 9 U I O HOLAK VOLUMES,
CC/HOL:
TOTPL PRESSURE KP4 EXPTL. ClLCa
x1
.oooo .IO00 .2000 3000 4 000 5000 6000 a 7 000 a8000 9000 1.0000
Oa077 3.481
0.077 3,491 5.707 7.203 8.319 9.216 9.9R7 10.725 11.524 12.478 13.657
5.707
7.203 8.31’3 3.216 9.387 10.725 11.524
f5:G
1-CHLBRBBUTRNE ( 1 1
+
+ Aniline (2) System at 298.20 K
VL(1) =
105.120
VL(2)
MIXTURE FUGACITY COEFFICIENTS
1
la0000 a9981 a9970 e3962 a9956 a9951 e9947 a9943 -9939 -9934 a9927
RNlLlNE
109
Y1 a 0000 .9798 a 9887 a9917 .9933 a9944 9954 9962 9972 .9984 la0000
2 e9999 a9973 -9956 a9945 s9936 a9929 e9923 a9918 e9912 a9904 a9895
91.530
A C T I V I T Y COEFFICIENTS
1
2
1 a0000
3.0752 2.5120 2.0753 1 7498 1.5173
!:!it! 1 1067
E!6EaSS
FUNCTIONe J/HOLE 0.00
1.19~~1 1 a 3183 f 2.0910 2 a 5882 3.5469
ii#;
:;I$’:
la0531 1.0142 1*0000
252.36 450.36 592.07 678.31 710.57 687.84 607.92 468.36 267. 20 0.00
[21 ETHYL RCETRTE [ l l
R 298.20 K
+
RNILINE
[21
R 297.49 K C 3 9~78..8 2 E 9 3K
O W 0 o
S
0 0 9
U
e
u a.
Y
..
Y
23 :m
W
a
Lo
3
Ln
(0
W
Ln
W
[f
e z
[f
Q 3 h ;
e z
U
+-
0
=
E3
+
5
3
> w o
o
5
-
w
W 0
m 0 0
k
q
0
,
m
0
10
0.20
0.60
0.40
0.80
30
0.00
XI Flgure 1. Deviation from RaouR’s law for the 1-chlorobutane (1) aniline (2) system.
,
,
I
0.20
0.110
0.60
0.80
.oo
Xl
+
Figure 2. Deviation from Raoun’s law for the ethyl acetate (1) (2) system.
+ aniline
110 Journal of Chemical and Engineering Data, Vol. 28, No. 7, 7983
Table V. Calculated Data for the Ethyl Acetate (1) L I O U l D MOLAR VOLUMES, CC/MUL:
X I .0000 .1000 ,2000 .3000 .4000
,5000 6000 ,7000
.
HOUO
YO00
1.0000
I U l A L PRE SSURE, K P A E kPTL CALC. 0.074 0.074 1.230 1.230 2.321 2.321 3.422 3.‘t22 4.592 4.582 5.803 5..J03 7.0 19 7.079 8.396 8.1Y6 3.722 9.722 ll.Jl7 11.017 12.241 12.241
L l U U l C MULAR VOLUMtS,
.
X l 0000 .lo00 ,2000 ,3000 ,4000
,5000 .6000
.. .
7000 H000 YO00 1 .0000
.
x1 0000 .lo00 ,2000 .3000
.4000
5000 6000
.
./000
4
u000 7000
1. 0000
CC/MOL:
I I J I A L PPE S S U S t c KPA
EXPlL. l.JY0 11.052
19.734
ZH. 763
37.702
4 6 e‘J90
56.302
65.978
75.538
CALC. 1.890 11.052
19.937
28.763 37. 7 0 2 46.88Y
56.302
65.0 74 15.558
8 5 . 2 7’3 Y $ . 97’)
I I J I A L PRE S S U R t , KPA E kPTL CALC. 15.760 15.960 53.208 5 3 . 21)6 t l Y . 3H9 89.385 125.101 125. O Y Y 16 0 2 6 150. 924 196.986 196. 9.94 233.7Y9 233.7eY 2 11.392 2 11. 392 3 O Y ‘1 5 0 309.950 349.521 349.621 390.557 390.559 13
Aniline (2) System at 297.49, 348.23, and 397.89 K
VL(1) =
98.361
MIXTURE FUGACITY CO5FF 1CIEN:S 1.J600 e9994 .39R8
.9982 ,9976 .9970 ,9964 ,9957 ,9950 ,9743 e9937
VL(1) =
.oooo
105.R60
VL(2) =
MIXTURE FUG4CITY C ‘ l E FF I C I C h( 1 S 1 2 ,9966 ,9940 e9079 ,9806 eq?Yb ,9677 .+I14 e9551 ,9632 ,7425 49550 -9298 e9466 ,9169 ,9381 .YO37 eY294 ,9903 ,7204 a8765 -9111 e3623
,
Y l
.9959 ,9975 ,9986 ,9994 1.0000
MIXTURE FUGACITY c J E F F 1c 1E N I s 1 2 eY9YV aYY90 ,9363 .9Y43 ,9733 ,91398 ,9904 ,9953 .7974 .?YO7 eYY43 ,9761 e9811 ,9713 e9779 ,9464 ,9747 e9615 ,9715 ,7566 ,9682 .Y517
91.491 EXCESS
.94Y9 ,9991 e9982 .9174 ,9765 .9955 .Y946 ,9936 ,9925 .Y916 ,9906
and V are actually interpolated values from the cubic splined fits of the experimental P vs. x , values. (The fidelity with which the splined fits represent the actual experimental P values is shown in Tables I1 and 111.) The “calculated” pressure values are from the Mixon et al. data reduction method. That method usually can be made to reproduce the input (experimental) pressure values to any desired precision. The PTx data were also reduced with the Barker (5) method using six different GEcorrelations-“absolute” Van Laar, Wilson, NRTL, modified Margules, UNIQUAC, and the fiveconstant Redlich-Kister equation. Usually, the five-constant RedlichKister equation reproduces the experimental P vs. x values best but, for the ethyl acetate aniline system, the modified Margules equation (6)performed equally well. The Barker fits of the experimental P values (based on the Redlich-Kister equation) are compared to the Mixon et al. fiis in Table VII. I t is usually difficult for the Barker method to equal the Mixon et al. pressure fits but it actually does better for the 348.23 K set of the ethyl acetate -t aniline data. The calculated activity coefficient curves are shown in Figures 3 and 4. The points are from the Mixon et al. method while the curves approximate the Barker results. The curves
+
VL(2) =
,9453 a9741 9646 9902 .9936
Y1
.oooo
ma451 s 9 2 30 9529 .9690 9792 9861 ,9911 9349 .997? 1.0000
4CT I V ! T Y
COEFF I C i E N T S
I
1 0261 0.9557 0.3284 0.9220 0.9306 0.9455 0.9626 0.9795 0.9928 1.0001
1.0000
L
1.0000 1,0032
1 0091 1 1.0048 .0101
0.9916 0 9699 0.9388 0.9014 0 9652 0 9280
CI9BS
FUNCTION e J/HOLE
0.00
-4.02 -20. 8 2 -42.89 -64.10 -79.75 -86.84 -82.73 -65.61 -35.57 0.00
95.@91
4 C T I V f T Y COFFF I C i E N T S
I 6505 1.0149 0.9966 C.9863 0.9827 0.9846 0.9897 0.9930
0 9966 0.9991 1.0000
1 e0016 1. boo0 1.004R 1.0082 1.0100 1 0083 1a 0 0 3 1
0.9949 0.9840
0 9704 0.9511
EXCESS c19es FUNCTION 9 JIHOLE 0.00 8.53 9.04 4.49 2.91 0.53 6.20 8.66 7.14 1.13 0.00
EXCESS Y1
.oooo
,7255 e9521 9356 ,9357 .7553 e9691 s 9796 .9978 ,9944 1.0000
ACTIV!TY
CDEFF I C i E N T S 1.6000 1.0017 1e 0 0 5 8 1.0118 1a 0 1 9 1 1.0270 1.0354 1eOQQ4 1 0539
CIBBS
FUNCTION
J/MOLE 0.00 31.84 53.40 66.52 72.23 71.69 65.89 55.45
40.81 0.00
1 eO6Ql 1e 0 7 6 9
22.29
are actually fits by the plotting software of closely spaced Barker results fed to the program and, in some cases, there can be a noticeable difference between the input values and the actual location of the curve drawn. The Barker results in Figures 3 and 4 are from the fiveconstant Redlich-Kister correlation for GE. The agreement between the Barker and Mixon et al. activity coefficients is very good, even for the complicated ethyl acetate system. The only serious deviations between the two occur at high x values for the 297.49 and 348.23 K isotherms. Table VI11 compares the pressure fiis and the infinitedilution activity coefficients from the Mixon et al. and the various Barker solutions. Note that the modified Margules (five constants) and the five-constant Redlich-Kister equations are the only ones which approach the Mixon et al. results in the accuracy of the P fits. As is usually the case, the Barker solutions which agree best with the Mixon et al. pressure fits will also agree best in the yimvalues obtained. Even the modified Margules and Redlich-Kister Barker sobtions deviate somewhat from the Mixon et al. y l mvalues in Tables VI11 (where the virial equation of state with the Tsonopoulos 8 correlation was used) and in Figure 4 (where the
,
Journal of Chemical and Engineering Data, Vol. 28, No. 1, 1983
Table VI. Parameters for Peng-Robinson Equationa
a
component
T,, K
P,, MPa
W
aniline ethyl acetate 1-chlorobutane
699.0 532.2 542.0
5.309 3.830 3.688
0.3820 0.3630 0.2180
111
E T H Y L R C E T R T E [ l l + P N I L I N E (21 A a 2 9 7 . ~ 9K
+
0 3 1 8 . 2 3 r( c 397.as K
x
Binary interaction constant was set at 0.0 for all systems.
Table VII. Comparison of the Barker and Mixon et al. Pressure Fits max % dev in Pa temp, K
Barker
Ethyl Acetate (1) + 297.49 0.902 348.23 0.244 397.89 0.199
rms for % devb
Mixon
Barker
Mixon
Aniline (2), Peng-Robinson 0.556 0.361 0.259 0.260 0.089 0.103 0.078 0.077 0.034
1-Chlorobutane (1) t Aniline (2), Peng-Robinson 298.20 0.319 0.091 0.107 0.049 a % dev = 100[ IP,,lcd -Pexptl 1/PeXptl]. [E”(% d e ~ ) ~ / n ] ~ ” .
l-CHLBROBUTRNE[lI
+
rms for % dev =
RNILINE [PI
R z9e.20 K
AA I
IO
0.20
I
I
0.110
0.60
I
0.80
A
1 .oo
Xl
Flgure 4. Activity coefficients for the ethyl acetate (1)
+
aniline (2) system. Curves are from Barker results; points are from Mixon et al. method. The X10-’ multiplier means that decimal must be moved one place to the left in ordinate scale values.
f
a
00
0.20
0.60
0.110
0.80
xi
+
Flgure 3. Activity coefficients for the 1-chlorobutane (1) aniline (2) system. Curves are from Barker results; points are from Mixon et al. method.
Peng-Robinson equation of state was used). Note the variation in the y l mvalues in Table VI11 with the Barker GE correlation used. Sometimes the Gautreaux-Coates equation (8, 9)can be used to provide further evidence concerning the most probable y i mvalues but for this system its component-2 values are not reasonable. Usually, when the (dPldx values needed by the Gautreaux-Coates equation are obtained from the splined fits, its y i mvalues agree well with the Mixon et al. resutts (which are also based on the splined fits). However, the Mixon et al. finitedifference method ”reaches” the x 1 = 0.0 and x = 1.0 points by a quadratic GEextrapolation based on
the GE = 0 values at x 1 = 0.0 or x 1 = 1.0 and the two adjacent GE points. That GE extrapolation will, in a few cases, generate (dPldx l)im values which differ appreciably from the splined-fit slopes. The Gautreaux-Coates values obtained from the x ,x,lP plots are equally unreasonable. (The xlX2/PD plots extrapolate better at x 1 = 1.0 than do the PDlx1x2plots in this instance.) I n both cases, the Gautreaux-Coates method fails to give useful values at 297.49 and 348.23 K because of some scatter in the measured P or x 1 values at x 1 = 0.925 and 0.960. That scatter is not apparent in Figure 2 but does become apparent on the more sensitive XIX2/PD plot. Because of the shapes of the P vs. x isotherms and the very small departures from Raoutt’s law in that region, those two points must be known with extreme accuracy if the yZmvalues calculated with the Gautreaux-Coates equations are to be accurate. Except for the Van b a r value at 297.49 K, the Barker values in Table V I I I agree reasonably well with the Mixon et al. values for yzm.The Barker results are less sensitive to the positions of the experimental P values near x 1 = 0.0 and 1.0 because the GE correlation constants are based on all the P points across the entire composition range. The Mixon et al. results are much more sensitive than the Barker method to the extreme x 1 points but the splined fits and the quadratic exptrapolation of the GE curve to the end points do tend to smooth the data somewhat. For this system, that smoothing action probably brings the calculated results closer to the truth, but in general it can obscure the true behavior of the activity coefficients at the very low and very high x 1 values. The uncertainty concerning the true values of the ylmand yZm values is increased by the effect of the equation of state values in Table used. A close comparison between the yZm V I I I (Redlich-Kister solution) and Figure 4 shows appreciable differences between the values obtained with the virial and
112 Journal of Chemical and Engineering Data, Vol. 28, No. 7, 1983
Table VIII. Effect of Calculation Method on yimValues for the Ethyl Acetate (1)
+ Aniline (2) Systema calcd yi’” values
accuracy of P fits, max % dev/rmsd calculation method hlixon et al. Barker: absolute Van Laar Wilson NRTL modified Margules UNIQUAC Redlich-Kister, five constants Gautreaux-Coa tes: splined fits X , X , / P D plots
component 1
component 2
297.49 K 348.23 K 397.89 K 297.49 K 348.23 K 397.89 K 297.49 K 348.23 K 397.89 K 0.6/0.3 0.3/0.1 O.liO.0 1.030 1.057 1.129 0.943 0.982 1.113 3.1/1.6 4.9/1.6 1.5/0.5 0.8i0.3 7.0/2.2 0.9/0.4
0.710.2 0.6/0.2 0.3/0.1 0.3/0.1 2.1/0.8 0.3/0.1
0.3/0.1 0.3/0.1 0.6/0.2 O.liO.0 0.3/0.2 O.liO.0
0.942 1.276 1.060 1.040 1.322 1.051
1.048 1.083 1.062 1.055 1.155 1.070
1.116 1.116 1.109 1.126 1.145 1.129
2.982 0.870 0.843 0.898 0.886 0.885
0.967 0.975 0.970 1.079 0.994 0.994
1.087 1.087 1.088 1.116 1.097 1.104
1.025 0.983
1.058 1.074
1.129 1.125
8.046 9.666
1.208 1.819
1.244 1.336
Virial equation, Tsonopoulos correlation ( 7 ) . Table IX. Effect of Equation of State Choice on yjmValues Obtained with the Mixon e t al. Method for Ethyl Acetate (1) Aniline (2) at 397.89 K
+
yim
eq of state used
1
2
ideal gas vinal through B“: Tsonopoulous ( 7 ) Hayden-O’Connell (I 0) Redlich-Kwong : Lu modification (11) Peng-Robinson ( 4 )
1.0306
0.7003
1.1286 1.1452
1.1127 1.2045
1.1094 1.1224
1.0162 1.0769
The Peng-Robinson equation (4) is convenient to use because, unlike the various modifications to the Redlich-Kwong equations such as that due to Lu ( 7 7), only the acentric factor is needed beyond the critical temperature and pressure values. aniline system Also, as in the case of the ethyl acetate covered in this paper, the Peng-Robinson results tend to fall in the mtddle of the various equations tested. The values reported in Tables I V and V are based on the Peng-Robinson equation, while the comparisons shown in Table VI11 are based on the virial equation and the Tsonopoulos correlation.
+
Reghtry No. Aniline, 62-53-3; ethyl acetate, 141-78-6; lthbrobutane, 109-69-3.
Peng-Robinson (4) equations of state. Table I X compares several equations of state. I t should be remembered in making these comparisons that the system pressure is very low at the aniline end (seethe x , = 0.0 values in Table 111). At the ethyl acetate end (x = 1.O) at 397.89 K, the pressure is still only 390.5 kPa but that is high enough to bring out the differences between the various equations of state shown in Table IX. At 390.5 kPa, the virial equation of state truncated after the second coefficient should work well but the results depend upon the correlation used to predict the Bu and Bu values. Purecomponent data evaluation work done in the Laboratory has shown that for ketones and alcohols the Tsonopoulos correlation (7) will more closely approximate the rellable experimental B8values for more compounds than does the Hayden-OConnell correlation (70). Even when the Hayden-O’Connell correlation works best for a ketone or an alcohol, the Tsonopoulos values are also quite close but the reverse is not always true. Also, on the basis of limited experlence with comparisons such as the one shown in Table I X , our tendency now is to trust the Tsonopoulos results more.
,
Literature Cited (1) Maher, P. J.; Smith, B. D. J. Chem. Eng. Dsta 1079, 2 4 , 16. (2) Maher, P. J.; Smith, B. D. J. Chem. Eng. Dsta 1980, 2 5 , 61. (3) Mixon, F. 0.; Gumowski, B.; Carpenter, B. H. Ind. Eng. Chem. Fundam. 1965, 4 . 455. (4) Peng, D.-Y.; Robinson, D. B. Ind. Eng. Chem. Fundam. 1076, 4 , 455. (5) Barker, J. A. Aust. J. Chem. 1953, 6 , 207. (6) Abbott, M. M.;Van Ness, H, C. AIChEJ. 1975, 2 1 , 62. (7) TSOt’K~poulos.C. AIChE J. 1074, 2 0 , 263. (8) Gautreaux, M. F.; Coates, J. AIChE J. 1955, 1 , 496. (9) k h e r , P. J.; Smith, 8. D. Ind. Eng. Chem. Fundam. 1070, 18, 354. (10) Hayden, J. G.; O’Connell, J. P. Ind. Eng. Chem. Process Des. Dev. 1975, 14, 209. (11) Haman, S.E. M.; Chung, W. K.; Elshayai, I. M.; Lu, 8.C. Y. Ind. Eng. C h ” Process Des. Dev. 1977, 16, 51.
Received for review July 22, 1982. Accepted September 24, 1982. We gratefully acknowledge the financial support received from the National Science Foundation Grant CPE-60000534 and from the Industrial Participants in the Thermodynamics Research Laboratory.
