520
INDUSTRIAL AND ENGINEERING CHEMISTRY FLUID FRICTION IN HYDRAULIC CIRCUITS
In evaluating a fluid as a coolant some attention should be given t o the flow characteristics of the fluid. Fluids requiring a large amount of pump work t o circulate them through the cylinder jacket, radiator, and auxiliary piping should not be rated so high as those requiring less work. For this reason careful consideration should be given t o the changes in viscosity with temperature, especially a t the lower temperatures. Since the quantities of fluid available were very small, it was not possible to study the fluid flow characteristics by experimental means. In the analysis pressure drop computations were made and compared for turbulent flow of fluids through straight tubes. CHEMICAL AND MISCELLANEOUS REQUIREMENTS
The corrosive action of the test fluid on aluminum alloys, brass, copper, solder, and steel wa,s studied. The corrosion apparatus consisted of a rotating sample holder, which was arranged so that the samples under test were rotated in the fluid. An electrical heater was provided to maintain the temperature of the fluid a t a given value. The apparatus was provided with a water-cooled cover t o reduce the evaporation loss from the apparatus a t elevated temperatures. The corrosion was determined by noting the loss between initial and final weights of the samples during a test. Becawe of limited time the samples were allowed to remain in the apparatus for 120 hours.
Vol. 40, No. 3
The test results obtained using commercial coolants were not the same as the results in actual engine tests using the same fluids. However, by using a commercial coolant as a standard, it was possible t o grade the corrosive action of the test fluids with respect t o the standard. In this way it was possible t o eliminate the fluids which were more corrosive than standard commercial coolants. Factors such as flash and fire points, spontaneous ignition temperature, freezing point, and molecular weight Kere also given consideration in determining the suitability of the fluid for use as a coolant for liquid-cooled aircraft engines. Other factors which are normally considered in the evaluation of a fluid are cost, availability, toxicity, and the effect of fluid on packing and gasket materials. These factors were not considered in the investigation described here. The final evaluation of the small quantities of new coolants was based on the results of (a) the hot wire tests, (b) theoretical computations for the heat transfer in the radiator, (c) flow characteristics of the fluid, and (d) corrosion tests. In general each fluid tested was compared with water as the basic fluid. LITERATURE CITED
(1) Colburn, A. P., Trans. Am. Inst. Chem. Engrs., 29, 174-210 (1933). (2) McAdams. W. H., “Heat Transmission,” 2nd ed., John Wiley & Sons, Inc., 1942. (3) Parsons, P. W., Trans. Am. Inst. Chem. Engrs., 40, 655-73 (1944).
RECBIVED October 3, 1946.
Heat Stability of Molybdena-Alumina Dehydrocyclization Catalysts ALLEK S. RUSSELL AND JOHN J. STQKES, JR. Aluminum Research Laboratories, New’Kensington, P a . T h e heat stability in dry air of molybdena-alumina catalysts has been measured as a function of alumina type, alumina area, and molybdena concentration. Activity of these catalysts for the dehydrocyclization of nheptane to toluene increased at calcination temperatures of 600” to 700” C., but decreased at higher temperaturesr Activated alumina of the H type was more stable than activated alumina of the F type. Stability decreased with increase of molybdena concentration. For molybdenaimpregnated activated alumina F and low-silica H, activity was stable towards loss of area on calcination until the area was reduced to a value just sufficient to accommodate the molybdena in a monolayer and thereafter activity decreased linearly with further loss of area.
M
OLYBDENA-alumina dchydrocyclization catalysts for conversion of n-heptane to toluene lose activity in use. One type of loss is caused by the deposition of coke on the active surface during the reaction. This loss is reversible, and the catalysts can be restored by appropriate burning and reduction. A second type of activity loss, and the one with which this report is concerned, is irreversible and is caused by structural changes in the catalyst; these changes are accelerated a t the temperatures produced during coke burning. The ability to maintain activity after high temperature treatment is an important advantage gained for molybdena by its impregnation onto
alumina, but not all aluminas impart the same heat stability to the composite catalyst. In a previous report (3) the initial activity of molybdena-alumina dehydrocyclization catalysts was discussed. The results indicated that activity increased linearly with the amount of “impregnated area”-that is, the amount of alumina area covered with molybdena. The highest activity resulted from impregnation of a large quantity of molybdena on high area alumina; neither a large quantity of molybdena on low area alumina nor large area alumina without adequate molybdena was effective. This report gives data on the effects of alumina type, alumina area, and molybdena concentration on the heat stability of molybdena-alumina catalysts. The change in activity on high temperature treatment is correlated with the extent of the impregnated area. MATERIALS, APPARATUS, AND ANALYSIS
The activated aluminas were the same purified products of the Aluminum Company of America, Chemicals Division, which have been described (3) except for the H alumina, low-silica, whoso analysis was: 37, loss on ignition, 0.4y0 NazO, 0.3% SiOz, 0.1% FezOa,0.3y0 CaO, 0.2% MgO, and 0.3y0SO3. Unless otherwise designated, the catalysts vxre prepared by the regular procedure of impregnating alumina with ammonium molybdate. “Coprecipitated” molybdena-aluminas were prepared by coagulating, with ammonia and ammonium molybdate, an alumina sol result-
March 1948
'
INDUSTRIAL AND ENGINEERING CHEMISTRY
ing from the action of water on amalgamated aluminum, according to the procedure of H e a d (2). The catalytic apparatus, which has been shown in diagram (a) was an atmospheric pressure system for passing n-heptane a t a constant feed rate through a fixed bed of catalyst at a controlled temperature, condensing and collecting the liquid, and measuring the gas and coke. 1he catalysts were calcined in a 150-ml. Vycor flask in a well insulated crucible furnace while dry air or steam was passed a t the rate of 0.2 cubic foot per hour through an inlet tube sealed to the bottom of the flask. Temperature was controlled within 3" C. by a thermoregulator. The sample was about 15" C. cooler a t the top than a t the bottom. The liquid product from n-heptane dehydrocyclization was assumed to be toluene, unreacted heptane, and a small quantity of olefinic hydrocarbons. Toluene was determined by density or refractive index; liquid olefinic hydrocarbons were determined by bromine titration. The carbon and hydrogen deposited on the catalyst were measured by combustion. EXPERIMENTAL PROCEDURE
Each new catalyst sample was screened t o 4 to 8 mesh, made t o a 100-ml. volume in a graduated cylinder, and heated 16 to 20 hours at 475" t o 500" C. in a furnace with air qirculation. To determine heat stability in air, the catalyst was calcined 16 hours a t temperature, dehydrocyclization was measured, and a representative 7-ml. sample was selected for coke analyeis and surface area measurement and replaced with fresh catalyst. The catalyst was then recalcined, dehydrocyclization measured, and the process repeated. The limited air supply t o the catalyst during calcination prevented overheating during combustion of the coke. For the heat stability data in steam, the procedure was the same except that the sample was calcined only 6 hours a t temperature and then was reheated 16 hours in air at 500" C. After dehydrocyclization the sample was reheated 16 hours in air a t 500" C. t o remove coke before recalcination in steam. To measure dehydrocyclization, the hot oxidized catalyst was transferred quickly t o a reaction bulb, and the bulb was connected t o the catalytic system, heated in 0.75 hour t o 500" C., and held a t 500" C. for 1.5 hours while hydrogen was passed through the catalyst a t 1.5t o 2 cubic feet per hour. At the end of the reduction period n-heptane was passed over the catalyst a t 497" C. for 1.5 hours a t a flow rate of 23 ml. per hour. Since the catalyst volume was 100 ml., the liquid hourly space velocity was 0.23. The system was flushed with hydrogen after the run. The measurements included volume of feed, volume of liquid reaction product, integral volume percentage toluene and olefinic hydrocarbons in the liquid product, gas volume and molecular weight, and carbon and hydrogen content of the coked catalyst. Surfade areas of the oxidized catalysts were measured by the method of Brunauer, Emmett, and Teller (1)employing the sorption of n-butane at 0" C. Results are expressed as the millimoles of n-butane which just cover the area of 1 gram of alumina in the sample (m.b./g.). Values of area in square meters per gram are obtained by multiplying the foregoing valuesoby 193 if the area of the n-butane molecule is taken as 32 square A.
521
mined from the graph. Activity was computed as tcst catalyst feed rate divided by the standard catalyst feed rate. I n establishing the graph of TO against feed rate, the standard catalyst was run a t each feed rate for such a duration that its coke deposition was 4.0 grams per 100 ml. of catalyst, and the test catalyst was run for the duration employed for standard catalyst at the test temperature and feed rate. The yields from a run are 100 times the milliliters of n-heptane reacted to form the given product, divided by the total milliliters of n-heptane reacted. MOLYBDENA UNIFORMITY ON ALUMINA
The volatility of molybdenum trioxide leads t o the supposition that it may be distributed uniformly on alumina at high temperatures, even though the alumina was not uniformly impregnated with ammonium molybdate from solution, To test this hypothesis, mixtures of activated alumina and molybdenum trioxide (from calcination of ammonium paramolybdate) were heated 16 hours in a closed crucible. A sample which was heated only to 500' C. had activity of 0.06, another sample heated similarly t o 650" C. had activity of 0.41, and one which was initially wet with water and heated to 650" C. had activity of 0.93. To create this activity, molybdenum trioxide must have diffused over the alumina surface for large distances. This diffusion increased with temperature and was promoted by water. Many of the samples discussed below showed an increase in activity after calcination a t 650" C. compared with the values after 500" C. It is probable that this increase in activity arose from a superior uniformity in the distribution of molybdenum trioxide on the alumina. EFFECT OF ALUMINA TYPE
In Figure 1 is shown the activity as a function of calcination temperature in dry air and in steam for activated alumina F and low-silica H both impregnated t o Mo/A10.05. At 500" C. the F alumina had greater activity than the H, but after calcination temperatures higher than 700" C. the H was definitely superior. For these materials, the activity was the same whether the calcination wm in steam or in dry air. Both materials showed a small increase in activity during calcination to 600" C. in dry air as discussed above, although the surface area diminished slowly in this treatment. It is evident in Figure 2, in which activity for these samples is plotted as a functi6n of surface prea, that the H alumina
CALCULATIONS 0 F DRY A I R
The activity of a test catalyst was defined (3)as the ratio of nheptane feed rate for the test catalyst to the feed rate for a standard catalyst when each catalyst produced the same TO, the integral volume percentage of toluene plus olefinic hydrocarbons in the liquid reaction product. The catalyst selected as standard was a n activated alumina F of area 0.38 millimole of n-butane per gram, impregnated t o an atomic ratio of molybdenum to aluminum of 0.05 (12.401, molybdenum trioxide). TO produced by standard catalyst was graphed a s a function of feed rate. TO from the test catalyst at a given feed rate was measured and the feed rate at which standard catalyst pr~ducedthis TO was deter-
0
F STEAM
A
H DRY A I R
0
H STEAM
.O 500
Figure 1.
600
700
800
900
C A L C INAT IO N T E M P E R A T UR E "C Effect of Calcination Temperature on Activity
Activated aluminas F and low-silica H, Mo/A10.05, in dry air and in steam
522
INDUSTRIAL AND ENGINEERING CHEMISTRY
*
Vol. 40, No. 3
100
*/. GASEOUS HYDROCARBONS 80
1.2
%COKE 1.0
n 6o d
% OLEFINIC HYDROCARBONS
41
2-
> .8
9
t
40
2 I-
v
i-
?
16
IV
4
12
I .8
1
.a
.4
1.4
a
A A
0.71 1.01
0.085 0.05
0
1.2
0
.3
12
9
.6
1.5
1.8
10 1.0
.e 1.4
.6
'1zEPia
.2
3.2 0500
600
700
800
1.2
>
1.0
t
-I2
.e
ooo
.6
0.4 7,
.4
0.71
lirmzal 1.01
'2 00
.3
.6
I
.9
SURFACE A R E A
MILLIMOLES N-BUTANE '
Figure 6.
A5
1.2
1.5
PER G R A M
Effect of Surface Area on Activity
UppeT. High-silica activated alumina H in dry air Lower. Activated alumina F in dry air
HIGH-SILICA ACTIVATED ALUMINA R
The activity of activated alumina H high-silica, impregnated to 0.02, 0.05, 0.10, and 0.15 is plotted in Figure 5 (lower) as a function of calcination temperature in dry air. At a calcination temperature of 500' C., activity increased progressively with increased molybdena concentration. After higher temperature calcination, the heavily impregnated aluminas lost activity much more rapidly than the alumina with less molybdena. Thus after
INDUSTRIAL AND ENGINEERING CHEMISTRY
524
700" C. calcination, the AIo/Al 0.05 sample had higher activity than the Mo/A10.15. The linear correlation of activity with surface area on caicination did not obtain for high-silica activated alumina H , although the activity of this surface has previously been shown to increase approximately linearly with amount of added molybdena t o Mo/ A1 0.15. The relation of activity to surface area on calcination for this material is shown in Figure 6 (upper), The sample impregnated to Mo/A10.15, a monolayer for this high surface area material, showed a linear decrease t o area about 0.6 millimole of nbutane per gram and then another linear decrease of lesser slope. A possible explanation for this behavior is that the high-silica alumina surface is not homogeneous, and the most stable area is least effective in distributing molybdena in an active layer. Samples impregnated with lesser amounts of molybdena declined slowly in activity with area until they intersected the monolaycr curve, after which they decreased with area a t the same rate as the higher molybdena samples. The transition from the uniform activity to the rapid decline with loss of area was apparently not sharp for the Mo/A1 0.05 sample. There was a small loss of molybdena on calcination of these samples, but this was much less than would be necessary l o explain the break in Figure 5 (Ion er). Coke and gas yields were increased several per cent by the silica. COPRECIPITATED MOLYBDENA-ALUMINA
Coprecipitated molybdena-alumina samples have shown low activity and low coke yield when calcined only at 500' C. They had the unique property of a very rapid increase in activity on calcination to about 700" C., and a t this temperature had as high activity as impregnated molybdena-aluminas. On further calcination the activity diminished with decrease of area below that capable of supporting a monolayer, according t o the rules deduced for other high purity molybdena-aluminas. There is some question of the reason for the low initial activity. I t might be associated with a deficiency of molybdena on the surface, but conductivity measurements of ammonia leachings of the catalyst indicated that the molybdena was as available for solution when the catalyst was originally precipitated as after 700" C. calcination. The coprecipitated catalysts had small pore size, as
Vol. 40, No. 3
measured by butane sorption isotherms, and the pores increased in size when the samples were calcined. Insufficient data on the influence of pore size on activity for this reaction have been assembled to test this as an explanation for the low initial activity. DISCUSSION
Impregnation of molybdena onto alumina produces a composite which retains much of the heat stability of the alumina, although the stability is smaller the greater the amount of molybdena. Since activity increases with added molybdena only up to monolayer formation on alumina, there is no advantage in the use of a greater quantity of molybdena. The most desirable fraction of a molybdena monolayer must be determined from an economic balance of activity and stability. Activated alumina H has a twofold advantage over activated alumina F; it has greater surface, and it is inherently more stable. However, if the H is impregnated to its highest activity, about 3, the sample is less stable than is the F impregnated to its highest activity, about 1.2 a t hlo/Al 0.05. At impregnation to Mo/Al 0.10, the activity is 2 and the stability exceeds that for the F, Mo/A10.05. The fact that an incompletely impregnated alumina maintains activity with loss of area until only enough area remains for a monolayer indicates that the molybdena either is not buried {uring recrystallization of the alumina or that it can rediffuse t o the alumina surface. The utility of the concept of impregnated area as a guide to activity is increased by the finding that it applies for calcined samp l y as well as for those freshly prepared. On the other hand, the anomalous result for silica containing alumina implies that considerable caution must be exercised in extending such a simple relation to the very complex phenomena observed in industrial catalysis. LITERATURE ClTED
(1) Brunauer, S., Emmett, P. H., and Teller, E., J . Am. Chem. Soc., 60, 309-19 (1938). (2) Heard, L., U. S. Patent Reissue 22,196 ( O c t . 6, 1942). (3) Russell, A. S., and Stokes, J. J., Jr., ISD. ENG.CHEM., 38, 1071 (1946).
RECEIVED January 13, 1947 .
Pvrolvsis of High Poly J
*
J
U
Hydrocarbons J
T h e stability of high polymeric hydrocarbons toward depolymerization as determined by the lowest temperature at which volatile products .are produced has been shown to be as follows: polyethylene > polystyrene > polyisobutylene. A mechanism of the depolymerization has been presented in which the role of free radicals is emphasized. The type of decomposition products obtained in the pyroIysis of polymeric hydrocarbons has been shown to be a function of the structure of the hydrocarbon chain and the volatility of the products of pyrolysis.
T
HE difference in the ease of polymerization of ethylene, isobutylene, and styrene requires the use of completely different polymerization techniques for the manufacture of commercial plastics from these monomers. Styrene polymerizes readily in the presence of heat, isobutylene requires the addition of catalysts, such as boron trifluoride, whereas ethylene does not
d
RAYMOND B. SEYMOUR Industrial Research Znstitute, University of Chattanooga, Chattanooga, Tenn.
polymerize under these conditions but requires heat and high pressure in the presence of oxygen as a catalyst. Styrene can be polymerized by all three techniques and isobutylene can be polymerized under the conditions used for ethylene, but there is a definite decrease in the ease of polymerization as one goes from styrene t o isobutylene to etliylene. Because of this difference in ease of polymerization and the structural differences of the polymers, it should be of interest to determine the ease of depolymerization of these compounds and the factors which determine the type of product obtained. D E W LYM ERIZATION
I n the depolymerization of high polymers in which heat labile groups are absent, there are three possible mechanisms that
'