I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
2332
LITERATURE CITED (l)
c. K., and
R*MS.
(7) Naragon, E. A., and Lewis, C.J., IND.ENC.CHEM.,ANAL.ED., 18, 448, (1946).
Am. Pet701eum2nst’,
(8)
(3) (4) (5) (G)
Egloff,Gustav, “Physical Constants of Hydrocarbons,” vel. 1, New York, Reinhold Publishing Corp., 1939. Fenske, M. R., IND. ENC.CHEM.,24, 482 (1932). Gilliland, E. R., Ibid., 26, 681 (1934). Kuhn, W . , HeEw. Chim. Acta, 25, 252 (1942). Kuhn, W.,and Ryffel, A., Ibid., 26, 1693 (1943).
Selker, M. L., Burk, R.E., and Lankelma, H,p.,Ibid., 12, 352 (1940).
2 6 , I I I , 2 3 (1946).
(2)
Vol. 42, No. 11
Snyder, J. C., and Steuber W., Ibid., 16, 454 (1944). (10)Westhaver, C. J., IND. ENO.CHEM,,34,126 (1942). (11) Willingham, C. B., and Rossini, F. D., J . Research Natl. Bur. (9)
Standards, 37, 15 (1946). RECEIVEDDecember 14, 1949. Presented at the Meeting-in-Miniature, Philadelphia Section, AMERICAP;CHEMICAL SOCIETY, January 20, 1949.
ESTIMATION OF SURFACE TENSIONS Temarr Liauid Mixtures 1
J
ALAN S. MICHAELS, RICHARD S. ALEXANDER,
AND
COLMAN L. BECKER
Massachusetts Institute of Technology, Cambridge, Mass.
A
relation derived by Meissner and Michaels (7) for estimating surface tensions of liquid mixtures has been tested on three ternary mixtures-benzene-nitromethanen-propyl alcohol, toluene-isopropyl alcohol-furfural, and toluene-ethyl acetate-benzyl alcohol. Both the surface tensions and refractive indexes of these systems were determined experimentally; the refractive index data were then used, in conjunction with calculated values of the pa .ichor and molar refraction of the individual comporwnts, to calculate surface tensions by means of this relation. Comparison of the experimental and calculated \ alues of surface tension indicates that the parachor
molar refraction relation is as applicable to ternary mixtures as it is to binaries. The maximum deviation observed was about 19%; this occurred in the binary isopropyl alcohol-furfural, which appears to be abnormal because of strong adsorption of the alcohol a t the liquidvapor interface. The parachormolar refraction-refractive index relationship is, in general, more satisfactory for estimating surface tensions of ternary mixtures than linear interpolation of values for the pure components. As is true with binaries, however, the method fails in cases where strong surface adsorption of one or more components takes place.
I
(2)
as a function of composition for each mixture. The surface tensions of the pure liquid compounds selected for the mixtures differed widely, in order to provide an adequate test of the relations. Properties of the pure components employed are presented in Table I. Parachors for the components were calculated from the atomic and structural values of Sugden (8-8). Molar refractions were calculated similarly from the data of Eisenlohr (1, 6, 7). The experimentally determined refractive indexes of the mixtures were used directly in Equation 1.
(3)
EXPERIMENTAL
S A recent article ( 7 ) , Meissner and Michaels, using a modifi-
cation of a relation first proposed by Tripathi (9),were able to estimate accurately &face tensions of binary mixtures of organic liqriids from a knowledge of the refractive indexes of the liquids. The ,elations employed were as follows:
vh/,f
=
\PImtx * n L x
[Rim,, . 71.a,,,
-
1
+2
+ . . , etc. IRInLlA = [ R l ~ a+ [Rim + . . . , etc. +
[PIm2.= [PIISI [Plm
Khere y = surface tension of mixture in dynes per cm.; [PI = arachor; [ R ] = molar refraction; refractive index fsodium D-line); and z = mole fraction of component in liquid. Subscripts 1, 2, etc., refer to individual components.
LIQUIDCOMPOSITIONS. To cover adequately the field of each ternary mixture, the following procedure was employed: Binary mixtures of components A and B were prepared containing approximately 20, 40, 60, and 80 mole A , and the propertiesthat is, refractive index and surface tension-were determined. To each of these binaries, incremental amounts of C were added, and the desired properties measured for ternary mixtures con-
This relation was found to apply to a number of binary systems with considerable accuracy, although serious deviations were observed in systems where one c o m p o n e n t showed strong capillary activity with respect TABLEI. PROPERTIES AXD CONSTANTS OF PURELIQUIDCOMPOUNDS USED IN TERNARY to the other. MIXTURES The object of this inveetiga[RIt nD [PI, [PI’ Exptl. tion w&s to test the appliExptl. nd, Lit. Cited 36 78 Y5. Compound Caled. (8) dikd. (6) Exptl. (8) Caycd. Expa. Lit. Cited ( 4 ) cability of these relations to Nitromethane 131.1 132.1 12.44 12.49 1.3800*’.* 1.3818*0 35.5 35.5 36.1 A 0 . 5 ternary mixtures of organic Benzene 207.1 206.2(av.) 26.31 26.18 1.4970”,‘ 1.5014s 28.2 28.1 28.42 AO.06 n-Propanol 165.4 165.4 17.58 17.52 1.3842*6*2 1.3854m 23.5 23.5 23.4 *0.4 liquids. E i g h t common Isopro yl alooBol 165.4 165.8 17.58 17.54 1.3749” 1.3776” 21.5 21.3 21.6 t0.1 organic liquids (c.P. grade) Toluene 246.1 246.6(s~.) 30.93 31.06 1.4938’6.’ 1.4978’6.‘ 28.8 28.0 27.5 t0.1 were used in the preparation Furfural 210.5 212.9 25.54 25.43 1.5239“ 1.5261” 40.4 43.0 43.2 t 0 . 4 Benzyl of three ternaries; surface alcohol 260.3 259.6 32.45 32.41 1.5388“.8 1.5396” 39.8 39.2 38.4 1.0 tension and refractive index Ethylacetate 216.0 216.9(av.) 22.21 22.25 1.3700”.0 1.3722’V 23.4 23.4 22.9 k 0 . 4 1, were measured experimentally
November 1940
INDUSTRIAL AND ENGINEERING CHEMISTRY
2333
MOLE 46 B€UZCNC
Figure 1.
Surface Tension-Composition Diagram for System Benzene-Nitromethanen-Propyl Alcohol at 25 O C.
taining 20, 40, 60,and 80 mole % ' C. Properties of the other binaries (A-C and B-C) were also measured. All compositions were determined gravimetrically, using an analytical balance. Adequate precautions were taken to minimize errora due to evaporation loases, and it is believed that the accuracy of composition determinations is within 1%. REFRACTIVE INDEX. Refractive indexes were measured with an adjusted and calibrated Abbe refractometer, using a sodium-vapor lamp as the source of illumination. Temperature was maintained at 25" C. (*0.4' C. maximum deviation) by means of a circulating water bath. SURFACETENSION. Surface tension measurements were made by the du NoUy (ring) method. Although this technique is less acourate than others available for these determinations, it is probably the simplest, fastest, and most convenient where large numbers of measurements are required. The tensiometer was calibrated by plotting the instrument reading against the literature value of surface tension (4) for each of the pure components used in the investigation. This method of calibration gave better results than the absolute method of Harkins and Jordan (.a), apparently because of imperfections in the ring used. I n any case, the measured values of surface tension are believed to be accurate within 0.5 dyne per cm.; such accuracy was more than adequate for the purposes of this work. Great care waa taken to prevent evaporation or contamination of samples. Measurements were made a t room temperature (25' * 4' C , ) , All surface tension values were corrected to 25.0' C. by interpolating the temperature coefficients of surface tension of the pure components reported in the literature (4). In most instances, the temperature corrections were within the accuracy of the measurements.
DISCUSSION OF RESULTS
The results of this investigation are shown in Figures 1to 3, and in Tables I1 to IV. Deviations between experimental and calculated surface tensions are presented graphically on the rectangular-coordinate plots accompanying each triangular diagram; each of the smaller graphs shows the variation of surface tension (both experimental and calculated) with mole fraction of one component while the mole ratio of the other two components is held constant. BENZENE-NITROMETHANE-PROPYL ALCOHOL. Inspection of Figure 1 (and Table 11) shows that Equations 1, 2, and 3 yield values for the surface tension of this ternary mixture which are in close agreement with the experimental, except for mixtures rich in nitromethane. Data on the binary n-propyl alcohol-nitromethane (section A-0, Figure l),indicate capillary activity of the alcohol with respect to the nitromethane, thus causing abnormally low surface tensions when small amounts of alcohol are present. The disagreement between the calculated and experimental values is greatest for this binary and vanishes rapidly as the percentage of nitromethane in the ternary decreases. Average deviation between experimental and calculated surface tensions is about 2%. TOLUENE-ISOPROPYL ALCOHOL-FURFURAL. Figure 2 and Table 111 show that, as was true with the preceding ternary, there is appreciable disagreement between experimental and calculated surface tensions for one binary (isopropyl alcohol-furfural). This discrepancy propagates throughout the entire field of the ternary. However, this is probably because the other binaries (furfural-toluene and toluene-isopropyl alcohol) deviate similarly, although to a lesser extent. The average deviation is about
INDUSTRIAL AND ENGINEERING CHEMISTRY
2334
Vol. 42, No. 11
TABLE11. SYSTEM
NITROMETHANE-BENZENE--R-PROPYI~-4LCOHOL = nitromethane x = mole fraction 2 benzene y = dynedcni. 3 = n-propyl alcohol T = C. 1
XI
21
1.Oooo
0.0 1.ooM) 0.0
0.0 0.c
n~
28
0.0 0.0
1.oooc
1.3800 1.4970 1.3842
TnD, Y, C. Exptl. 25.3 25.4 25.2
35.3 28.1 23.5
T
-
8.
26.0 23.5 26.0
Y,
Calcd.
Deviation,
%
35.5 28.2 23.5
0.0 +0.3 0.0 0.0 f3.7
A 0.175 0.352 0.555 0.749
0.825 0.648 0.445 0.251
0.0 0.0 0.0 0.0
1.3829 1.3818 1,3602 1.3800
20.3 20.3 20.4
....
24.3 25.2 26.0 28.0
25.3 25.5 25.6 25.0
24.3 26.1 28.2 30.9
0.796 0.699 0.394 0.201
0.204 0.401 0.606 0.799
1,4752 1 ,4639 1.4319 1 ,4090
25.0 26.0
0.202 0.158 0.109 0.094 0.040
0.798 0.632 0.472 0.372 0.160
0.0 0.210 0.408 0.534 0.800
1.4809 1.4613 1.4426 1.4309 1.4042
2.5.2 25.1 25.1 25.1 25.1
. ,.,
27.2 26.1 25.1 24.5
21.9 26.1 26.0 26.2
27.6 26.1 25.3 24.4
+l.5 0.0 -0.4 -0.4
....
29.1 27.9 26.6 26.0 24.6
24.2 24.2 23.7 23.A 23.6
29.0 27.8 26.7 26.1 24.6
-0.3 -0.3 +0.3 f0.3 0.0
x2
xa
?ID
C
9%. In this system, capillary activity of both alcohol and toluene in the presence of a more polar compound (furfural) appears to be responsible for the major deviations. TOLUENE-ETHYL ACETATE-BENZYL ALCOHOL. In this system (Figure 3 and Table IV) the major discrepancies occur in the binaries toluene-benzyl alcohol and ethyl acetate-benzyl alco-
0.367 0.299 0.226 0.147 0.077
0.633 0.514 0,391 0,254 0,132
0.0 0.189 0.383 0.599 0.791
1.4643 1.4489 1.4333 1.4162 1,4011
25.2
0.601 0.477 0.336 0.237
0.399 0.317 0.237 0.157 0.079
0.0 0.206 0.407 0.606 0.802
1.4381 1.4259 1.4143 1.4049 1.3942
26.2 25.3 25.3 25.4 25.4
0.119
0.801 0.644 0.493 0.328 0,164
0.199 0.160 0.123 0.081 0.041
0.0 0.196 0.384 0,591 0.795
1.4119 1.4033 1.3991 1t.3946 1,3890
6.
y,
Calcd.
I)cviation.
%
.. .. .. . ,
.. . .
25.2
29.7 28.1 26.7 26.0 24.7
24.5 24.2 23.8 23.5 23.4
30.1 28.7 27.4 26.0 24.8
4-1.3 f2.1 f2.6 0.0 f0.4
80.8 28.3 26.8 26.1 21.8
26.4 26.8 26.1 26.3 26.2
30.9 28.5 27.4 26.2 24.6
4-0.3 C0.7 f2.2 f0.4
32.7 29.0 27.0 25.7 24.8
25.2 23.8 23.6 24.0 43.0
33.6 30.4 28.7 26.9 25.1
+2.8 f4.8 C6.3 +4.7 +1.2
0.0
25.2 25.2 25.2 25.2 25.2
hol. The deviations for the ternary compositions can again be ascribed to the abnormal behavior of the binaries. However, in spite of the appreciable numerical deviations (average, 6 9 3 , the shapes of the calculated and experimental curves are quite similar. This similarity emphasizes the advantage of the parachhor.-niol:tr refraction-refractive index relrition over simple liiicnr
0
100
0
MOLC *k FURFURAL
FWFURAL
MOLE *& C- PROPANOL
10
Figure 2.
T
F
100
*.
y,
Exptl.
E
MOLE *h FURFURAL
MOLC
T,D, C.
D
4-8.1
+10.4
B 0.0 0.0 0.0 0.0
Xl
20
30
40 SO 60 70 k O L C PERCENT FURfllRAL
80
90
Surface Tension-Composition Diagram for System Toluene-Isopropyl .Alcohol-Furfural a t 25' C.
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
November 1950
0
2335
Io0 0 100 0 MOAE w ErnrL ACETATE
100 0 MOLE W ETUYL ACCTATE
MOLE
Figure 3.
*+TOLUCHE
Surface Tendon-Composition Diagram for System Toluene-Ethyl Acetate-Benzyl Alcohol at 25 O C.
interpolation as a means for estimating surface tensions of mixtures. REFRACTIVE INDEX. Tabled 11, 111, and IV show that the measured refractive indexes of the mixtures deviate but slightly from those calculated by the single additive relation: ltDm,,
=
121)171
+
n1@2
+
n&Zh
(4)
Theae minor deviations ttetween experimental and calaulltted re-
fractive index are not coincident with, and bear no apparent relation to,the observed deviations between experimental and calculated surface tensions. The general character of the variation of refractive index with composition for the mixtures studied indicates that no significant abnormalities in behavior of the bulk liquid phases have been encountered. Inasmuch aa surface tension is determined in the main by the properties of a thin liquid layer at the gas-liquid inter-
-
TABLE IrI. SYSTEM ISOPROPYL ~LCOHOL-~OLUENE-~~~RFr;RAi, isopropyl aloohol toluene 6 = furfural 4 5
ir4
25
1.0000
0.0
0.0
0.0
0.0
1.oo00
z*
0.0 0.0
1.oooO
nil
1.3749 1.4938 1.5239
TSD, 7, O C. Exptl. 21.3 28.0 25. I . . , . 43.0
....
T
8.
Y'
Calcd.
0.0 0.0 0.0
0.0
0.203 0.408 0,594 0.808
1.4089 1.4409 1.4681 1.4992
25.2 25.2 25.2 25.2
%
23.0 23.5 22.6
21.5 28.8 40.4
23.2 25.1 27.6 32.2
23.0 22.4 27.2 22.0
25.5 29.5 32.9 37.1
f17.a +19.2 +ll.6
0.792 0.598 0.266
0.208 0.402 0.734
1.4985 1.5039 1.5142
25.0 25.1 25.1
29.7 31.0 35.5
22.5 22.6 20.6
30.5 32.4 36.4
+2.7 f4.5 +2.5
0 800 0.639 0.480 0.319 0.162
0.200 0.160 0.124
0.0 0.201 0.400 0.601 0.798
1.4049 1.4299 1.4540 1.4790 1.6019
25.1 26.1 25.1 26.1 26.1
22.3 24.7 27.8 29.0 33.6
25.0 21.0 20.5 20.5 21.0
23.3 26.6 20.8 38.6 37.1
+4.S 4-7.7 +9.2
+2.n -0.0
+IO.?
C
O.OS0 0.040
24
ZL
E6
nu
+16.0
+10.4
O
C.
Y,
T
y,
Deviation,
8.
Calcd.
%
23.5 24.8 27.1 30.1 34.2
24.8 26.0 26.0 25.2 25.4
24.8 27.7 30.6 33.6 37.0
+5.5 +11.7 i-13.0 +11.7 f8.2 +7.4 f14.4 +9.9
Exptl.
D
4-0.8
B 0.0 0.0 0.0
T.D,
Deviation,
A
0.797 0.592 0.400 0.192
fraction 5 E mole gydfedom.
z
0.599 0.478 0.359 0.240 0.122
1.4310 1.4498 1.4680 1 ,4865 1.5051
25.1 25.1 25.1 25.1 25.1
1.4539 1.4679 1.4804 1,4949 1.6092
26.1 25.2 25.2 25.0 25.0
24.4 25.1 28.3 31 .O 35.1
27.0 27.3 27.6 27.6 27.7
26.2 28.7 31.1 34.1 37.1
1.4742 1.4830 1.4930 1.6030 1.5129
25.1 25.2 25.2 25.2 26.1
25.7 27.7 29.5 32.0 36.5
26.0 26.6 26.9 27.2 27.3
27.4 29.4 31.9 34.5 37.2
E 0.397 0.317 0.238 0.159 0,078
0.603 0.481 0.361 0.242 0.119
0.198 0.158 0.119 0.079 0.040
0.802 0.640 0.480 0.320
0.160
0.0 0.202 0.401 0.601
0.800
+lO.O
+5.7
4-6.6 4-6.1
f8.1 +7.8 +4.8
INDUSTRIAL AND ENGINEERING CHEMISTRY
2336
TABLE IV. SYSTEMTOLUENE-BENZYL ALCOHOL-ETHYL ACETATE 5 = toluene 7 = benzyl alcohol 8 = ethylacetate X?
DL
zs
x = mole fraction 7 =
T
=
dynedcm. OC.
T,AD, Y . C . Exptl.
nu 1.4938 1.5388 1.3700
25.2 25.3 25.2
T
Y,
Deviation, %
8.
Calcd.
27.9 39.2 23.4
28.3 28.4 27.5
28.8 39.8 23.4
f3.2 +1.5 0.0
26.8 26.0 25.0 24.2
27.8 27.8 27.8 27.2
27.6 26.8 25.7 24.5
f2.9 +3.0 f2.7 +l.Z
34.7 28.6 26.3
27.2 27.2 27.2 27.2
37.3 34.5 31.2 27.8
+7.5 +9.2 +9,1 +5.7
0.0 1.OOO 0.0
0.0 0.0
0.0
0 798 0,601 0,400 0.203
0.0 0.0 0.0 0.0
0.202 1.4690 0.399 1.4462 0.600 1.4211 0.797 1.3960
A 25.3 25.2 25.4 25.3
0.0 0.0 0.0 0.0
0.816 0.642 0,442 0.232
0,184 0.358 0,558 0.768
1,5110 1.4825 1.4490 1.4129
25.3 25.3 25.3 25.3
0.213 0.161 0.122 0.081 0.041
0.787 0.622 0.470 0.315 0.158
0.0
0.217 0.408 0.604 0.801
1.5304 1.4985 1.4690 1.4373 1.4046
C 25.0 25.1 25.0 25.0 25.0
36.0 31.6 29.3 27.3 25.3
22.8 22.8 22.7 22.7 22.7
37.8 34.9 32.4 29.5 26.6
+4.4 +10.4 +10.6 +8.1 +5.1
0.407 0.325 0.244 0,159 0.082
0.593 0.474 0.358 0.231 0.119
0.0 0.201 0.400 0.610 0.799
1.5192 1.4443 1.4699 1.4317 1,4023
D 25.2 25.2 25.2 25.2 25.2
31.3 30.4 28.3 26.5 24.9
22.3 22.8 23.0 23.5 24.5
84.7 33.4 31.1 28.3 25.9
+10.9 f9.9 f9.9 +6.8 +4.0
0.606 0,483 0,364 0.241 0.122
0.394 0,314 0.236 0.157 0.079
0.0 0.203 0.400 0.602 0.799
1.5119 1.4856 1.4583 1.4296 1.4002
25.1
31.1 28.7 27.6 26.2 24.7
29.7 28.7 29.0 29.0 29.2
32.9 31.4 28.6 27.7 25.5
+5,8 f9.O f7.2 +5.7 +3.2
0.800 0.634 0.479 0.321 0.163
0.200 0.158 0.120 0.080 0.041
0.0 0.208 0.401 0.599 0.796
1.5029 1.4774 1.4520 1.4285 1.3992
25 1 25.1 25.1 25.1 25.1
28.7 27.9 26.9 25.5 24.7
28.4 28.2 28.3 28.2 28.2
30.8 29.6 28.1 27.3 25.2
3-7.3 f6.1 f4.5 f7.1 4-2.2
1,000 0.0
1.000
B 31.6
E 25.1 25.1 25.1
. ..
Vol. 42, No. 11
face, and since the measured values of surface tension are almost always lower than those calculated, the conclusion that the devirttions observed are due to positive adsorption of one or more components seems justified. CONCLUSIONS
Investigation of three ternary mixtures leads to the tentative conclusion that Equations 1, 2, and 3 are useful for estimating surface tensions of multicomponent liquid systems. Deviations between such calculated and experimental values for ternaries are of the same nature and ma nitude as those observed for binaries. In most cases, the trends of surface tension with composition are more accurately described by use of the above equations than by linear interpolation of the surface tensions of the pure components. As has been found to be true with binaries, the parachor-molar refraction relation will yield accurate values of surface tension for ternary mixtures which do not euhiht adsor tion phenomena. Where one component of a mixture shows capiiary activity with respect to another, significant deviations between experimental and calculated surface tensions may be anticipated, ACKNOWLEDGMENT
The authors wish to express their appreciation to H. I?. Meissner for his generous advice and assistance during the course of this work and to H. H. Carter, who so kindly prepared the surface tension-composition diagrams. LITERATURE CITED
(1) Eisenlohr, F., Z. phys. Chem., 75, 585-607 (1910). ( 2 ) Harkins, W . D., and Jordan, H. F., J . Am. Chem. Soc., 52, 1751
(1930).
F
(3) Hodgman, C. D., editor, “Handbook of Chemistry and Physics,”
25th ed., Cleveland, Chemical Rubber Publishing Co., 1941. International Critical Tables, Vol. IV, p. 448 6,New York, hlcGraw-Hill Book Co., 1928. ( 5 ) Landholt, H., and Bornstein, R., “Physikalische-Chemiache Tabellen,” 5th ed., Berlin, Julius Springer, 1931. ( 6 ) Ibid., erg IIa, p. 173. (7) Meissner, H. P., and Michaels, A. S., IND.ENC;.C H m c . , 41, 2782 (1949). (8) Sugden, S.,J . Chem. SOC.(London),125, 1177 (1924). (9) Tripathi, R.C., J . Indian Chena. Soc.. 18, 411 (1941). (4)
Low Pressure Rotary J
Hydrogenator E. B. HERSHBERG, FREDERICK BERTSCH, HERBERT KAPLAN,
AND
HAYDEN C. BROWN
Schering Corporation, Bloomjield, N. J.
A design is described for a hydrogenator which employs new features of mixing, hydrogen feed, and operation. A study has been made of the several variables which affect the rate of hydrogenation in this equipment.
0
F THE several factors which influence the over-all rate of a
low pressure catalytic hydrogenation, the type apparatus employed plays a more important role than is commonly realized. Equilibrium must be reached quickly between liquid, solid, and gas, because a catalyst, particularly an active one, which is depleted of hydrogen or of the material to be reduced, loses its activity. The more critically differential hydrogenations are dependent on specific catalysts and a rapid reaction rate. I n certain cases, for example that of steroid ketones, a rapid hydrogenation often favors the production of the desired stereoisomer. It is often possible to hydiogenate selectively one of two double bonds or to reduce an acetylenic bond to an ethylenic bond. When the conditions of hydrogenat.ion favor the selection of one center pref-
erontially, a fast reaction usually increases the spread between the two rates of hydrogenation. Agitation determines the degree of success which any given design of hydrogenator will attain, both in keeping the catalyst in suspension and in mixing the liquid and gas phases. High speed propeller agitators or centrifugal pumps such as are used in large commercial installations (6, 6, 1 4 ) readily satisfy some of these requirements, but their use is contraindicated by the nature of the catalysts commonly used for hydrogenation at low pressure. Platinum catalyst ( 2 ) prepared by reduction of the oxide and Raney nickel (9) prepared by digestion of nickel-aluminum alloy with alkali both have a spongelike structure and are comparatively fragile. Under the impact of propeller blades these catalysts are hammered into denser and much less active forms. Both oscillating ( 1 , 4 ) and reciprocating shakers (1 1 ) are difficult t o adapt to a larger size. In one unusual experiment in which a reciprocating shaker was used the platinum catalyst was recovered in the form of small round pellets like lead shot. Mont-
