M
Thermal
ANY thorough researches upon the thermal decomposi-
tion of alkanes a t atmospheric pressure have been published recently, but nearly all of this work has been carried out upon the relatively simple gaseous members of the series. A certain amount of attention has been given to much larger alkanes. Hexadecane was extensively studied by Gault (17) and an approximately thirty-carbon paraffin by Waterman (50). However, because of the large number of gaseous and particularly of liquid products which result, complete and accurate analyses of all products have not been obtained as in the case of the small homologs. With the notable exception of the investigations of Frey and Hepp (15), little work has been done upon the pyrolysis of alkanes of intermediate carbon content. This is somewhat surprising, since efficient columns for the fractionation of gases (36) and liquids (IS),permitting analysis of the complex mixtures resulting from the decomposition of relatively small amounts of hydrocarbon, have been available for several years. T o bridge this gap, the present work was undertaken; the compound chosen was n-octane. Several workers have reported preliminary studies upon n-octane pyrolysis; their results are condensed in Table I. The data are of little value because of the few accurate determinations of time and extent of decomposition. Such gas analyses as are reported show little concordance. Moldavskir (SO) observed that pressure exerted a marked influence on the pyrolysis and that iron exhibited a catalytic effect. Hugel and Szayna (19) obtained evidence of splitting a t all carbon-to-carbon linkages of the molecule in nickel and iron apparatus. Decomposition in the presence of added catalysts has also been reported (IO,28, 99, 35, 64). Much more thorough work by Dintzes and associates (6-8) coincided with the present study; similar experimental conditions were employed, with the single exception that decomposition was carried out in copper-lined instead of steel or Pyrex apparatus (6). Because Dintzes' papers are not readily available and because they are closely related to the present work, his data are summarized in Table 11. The gaseous products were partially analyzed, but no attempt was made to examine the liquid products which also resulted. The velocity constants are consequently smaller than the true values for thermal decomposition of n-octane. The primary object of the present work was to determine the amounts of all reaction products obtained under various conditions. The free-radical theory (39,S9A), which explains nicely the proportions of products obtained under mild conditions from small alkanes, has not been adequately tested for larger hydrocarbons. It reasonably assumes that the saturated decomposition products obtained from normal paraffins retain the linear structure and that the unsaturated products are 1-alkenes. Frey and Hepp (15)found only olefins with double bonds a t the point of scission of smaller products from the initial paraffin, and the butene obtained by thermal decomposition of n-pentane has been identified (15, 31) as 1-butene. In view of the interconversion of 1- and 2-butene (21) and of 1- and 2-pentene (92) during partial decomposition, it appeared possible that other than 1-alkenes would result from pyrolysis of n-octane. Identification of the chemical individuals formed was therefore undertaken to prove or disprove the absence of isomerized decomposition products.
Decomposition
of n-Octane ROBERT F. MARSCHNER Standard Oil Company (Indiana), Whiting, Ind.
Because of the lack of complete and accurate thermal decomposition data upon normal paraffins between hexane and hexadecane, the behavior of n-octane in the neighborhood of 570" C. and at atmospheric pressure was studied. The noctane was isolated from petroleum in quantities large enough to permit the determination of the amounts of essentially all its pyrolysis products. The effect of temperature, extent of decomposition, and nature of the apparatus in which pyrolysis was effected were studied. A slight revision of the free-radical theory improves somewhat its agreement with the experimental data. The effect of a carbonized surface upon the pyrolysis was studied, and the bearing of these results upon the theory of thermal decomposition is discussed. The liquid products obtained were thoroughly examined for evidences of rearrangement of olefins and of isomerization of the carbon skeleton. A semiquantitative analysis of all the chemical individuals found among the liquid hydrocarbons is presented. The probable manner in which they were produced, as indicated by the evidence available, is outlined.
Isolation of n-Octane It was necessary in the present work to employ somewhat greater than the customary amounts of starting material. To avoid the inconvenience of synthesis, n-octane was isolated with unexpectedly little difficulty from petroleum by methods similar to those of Leslie and Schicktanz (26). Four 554
MAY, 1938
INDUSTRIAL AND ENGINEERING CHEMISTRY
555
low. Although not 100 per cent n-octane, this product compared favorably with the pure hydrocarbon boiling a t 125.6125.7" C. with an nY value of 1.3970-1.3976 (96,4 1 ) . The difficulty of purification by crystallizationwithout a solvent indicated an impurity of high freezing point. The rela-
barrels (630 liters) of a sweetened refinery naphtha containing essentially seven-, eight-, and nine-carbon hydrocarbons of a composite mid-continent crude were roughly distilled through a 2-meter packed column; a cut with an apparent boiling range of 120-130" C. was retained. Sixty liters of this n-octane concentrate were carefully distilled a t a constant reflux ratio of 30 to 1 through a fractionating column measuring 6 cm. X 2.5 meters, packed with screened metal turnings, and equivalent to sixteen theoretical plates. A summary of the data obtained is presented in Figure 1. Two-liter portions of the high-melting material were partially frozen by means of acetone and solid carbon dioxide, and the liquid was removed by filtration under reduced pressure. Fractions originally melting near -56" C. contained 80-90 per cent n-octane and gave a solid phase 90-95 per cent pure, but the higher and lower boiling fractions required two to four freezings to bring the n-octane to this concentration. The refractive index was used to follow further separation. Solid fractions above n2: 1.4055 were brought below this refractive index by refreezing; further solid material was systematically removed from the mother liquors until no further crystallization occurred a t -70" C. and was then reworked until a refractive index of 1.4055 was reached. Further purification of the 7 liters of crude paraffin was accomplished by recrystallization from chloroform; the n-octane was frozen in several portions from two successive batches (1.0 and 0.5 liters) of the solvent. Fractionation through column 3 (see "Apparatus") gave 6.5 liters of n-octane boiling within 0.5" a t 126" C . and with an nF value of 1.3990 or be-
k
REFRACTIVE INDEX
'-.140b
I
LITERS DISTILLED FIQURE
1.
FRACTIONATION OF PETROLEUM OCTANE
tively high refractive index (nU, 1.411-1.412) of the small volume of hydrocarbon recovered from the chloroform mother liquors pointed to Max. an aromatic impurity. Treatment of the prod&paPresDe-Gas AnalysisReferratu Temp. sure5 Time compn.5 Gas5 HZ CnHznCnHzn+a ence uct with warm concentrated sulfuric acid halved C. Atm. Hr. % Wt. % Mole p e r cent the concentration of hydrocarbons of high re. (1s) Glass 280 ($y 7;:; fractive index, as would be expected from aro4:7 (50) Silver 325 . .. .. .. .. .. Pyrex 405 (28) 1.0 (1) matics. The high melting point (15" C.) and prox'68' Pyrex 420 0.5 (2) ... ... .. .. .. .. .. ... imity of p-xylene to n-octane in petroleum, to420 1.0 Pyrex b61 Pyrex 450 . . . . . . (4) gether with the tendency (9) of aromatics to Silver 450 99 2:0 (31) 1510 $0) Silver 450 69 2.0 (38) 25.0 ii:i 619 7 9 : ~ distill below their normal boiling points in hySilver 450 201 4.4 2,0 [&] 23.5 9.7 2.8 87.5 drocarbon mixtures, suggest the probable presIron 450 123 15 5 (90) Glass 465b 1 0.5 ... i:4 9 8 : ~ (44) ence of this substance. Since xylenes are unGlass 475b 1 0.5 .( 8 6 ) 60.0 0:o 0.6 1.9 97.5 (44) Iron 475 70 2.2 3.7 3.9 92.5 (90) affected by the thermal conditions employed in Glass 4856 1 0.5 (5) 0.8 1.7 97.5 (4.6) this work, the hydrocarbon impurities were Pyrex 490b 1 .. 4 (4) Fire clay 550 1 .. 78 (30) 3517 2315 39:4 (46) considered unimportant, and because of the . . . . . . Pyrex 560 1 .. 72 . . (4) Pyrex 580 1 .. 50 . . . . . . (4) possibility, however unlikely, of introducing imPyrex 630 1 .. .. (90) (4) purities less inert by acid treatment, the described Fire clay 800 1 ciao) (95) 52:8 ii:3 35:3 (46) n-octane was employed. Hydrocarbon which Parenthetical values oaloulated or estimated as accurately as possible from original data. b Mean tempersture. was recovered after partial decomposition showed the narrow boiling range of the starting material TABLE11. EXPERIMENTAL DATAOF DINTZES ON THERMAL DECOMPOSI-but a small increase in refractive index proporTION OF OCTANE tional to the extent of pyrolysis, bearing out the above evidence. The hydrocarbon used was Gas Analysis, Run DeCzHa + CsHs 4- IC1 Refnot less than 95-98 per cent n-octane. No. Temp. Time compn. HI CHI CsHs CZHI CiH8 X IOa erence TABLEI. DATAO N THERMAL DECOMPOSITION O F ?%-OCTANE COMPILED FROM THE LITERATURE
O
$1
cg
);8
I
.
5
7
C. 7-8 496 7-12 506 7-9 527 533 7-4 545 7-7 7-10 570 8-10 569 8-1 1 570 570 8-2 569 8-1 571 8-3 8-5 569 8-6 572 &8 572 8-7 572 13 591
% M o l e p e r cent ....... 26.2 19,6 2.77 . . . . . . .52,4. 31.2 28.2 10.5 . . . . . . .40.6. . . . . . . . 31.1 23.2 2.7 . . . . 43.0 . . . . 68:s 6.92 75.8 13.8 . . . . . . .46.5. . . . . . . . . . . . 53.5. . . 22.6 19.6 22.8 34.2 20.8 5.22 2.6 18.2 21.6 39.2 17.4 5.6 6.20 3.6 . . . . . . .52.2. . . . . . . . 30.2 17.6 3.65 7.9 31.5 18.5 4.20 5.7 . . . . . . .50.0. . . . . . . . 31.1 22.5 . . . . . . .46.4. . . . . . . . 8.85 13.1 32.8 27.1 . . . . . . . 46.5. . . . . . . . . . . . 53.5 . . . . 29.6 30.7 80.1 41.4 . . . . . . . 39.7. . . . . . . 109.5 52.5 . . . . . . . 41.1. . . . . . . . 29.4 30.4 59.5 ....... 41.0. . . . . . . . 157 29.8 29.2 . . . . . . .41,5.. . . . . . . 175 . . .58.5. . . . 59.3 ....... 61.1 ...... .45.6. 215 28.6 27.8 156 35.4 487 64.8 Sec. 132
.......
........
....
0.21
Apparatus
1.04 1.96 2.37 11.4 22.2
The apparatus (Figures 2 and 3) consisted of a feed system, a cracking unit, and condensing and collecting apparatus :
...
...
15.8 10.6 6.4 7.1 5.4 4.7 4.1
.... ...
The feed system included a calibrated storage reservoir, R1, a tidal reservoir, 82, and a pump which fed n-octane to the cracking unit at a constant rate. This pump, operated by a Merkle-Korff motor, W , geared down to 20 r. p. m., proved satisfactory. The Pyrex glass valves, V , have been described (@). Adjustment of screw B regulated the feed rate between 10 and 1000 ml. per
556
INDUSTRIAL AND ENGINEERlNG CHEMISTRY T
VOL. 30, NO. 5
ing larger quantities of hydrocarbon, column 1 was used. It measured 4 mm. X 1 meter and contained a continuous Nichrome wire spiral. The receivers for these columns were protected with solid carbon dioxide condensers to prevent evaporation losses.
Decomposition and Analysis Each run consisted of the necessary preliminary operations, a longer significant period, and a stabilization of the liquid products. After heating the furnace to the desired temperature, nitrogen flow through the unit was stopped; measured volumes of n-octane were drained at timed intervals from R1 into R2, and were pumped continuously into the cracking unit. The preliminary products were removed after steady operation had been reached, and the run products were then collected. Whenever R3 filled with liquids, they were drained into R5, but the condensate in R4 was allowed to accumulate until the end of the run. At this time the products were effectively isolated from the cracking unit by emptying R3 and turning 88. Nitrogen was passed until the appearance of fog at X 2 ceased and was then replaced with oxygen to remove aossible carbon and tar devosits. With the- head of K filled with solid Earbon FIGURE2. APPARATUSFOR THERMAL DECOMPOSITION OF %-OCTANE dioxide and acetone, the contents of R4 were slowly drained into R5. Little hydrocarbon hour :by restricting the stroke of the cam-motivated beam, E . distilled below -45" C. (about 800 mm.) as R5 was slowly Inlet N for the introduction of gases was closed by stopcock S2 warmed and the distillation was stopped a t + 2 (about 860 since the outlet of valve P1 bearing manometer MI was always open to the cracking unit. The cracking unit consisted of a preheater and a cracking coil. A helix, H1, of electrically heated copper tubing, C, and a thermos2 couple block, P 2 , formed the preheater. Its volume was about 40 ml. and its tem erature was maintained 50-100" C. below that of the cracking coif Temperatures were determined by thermocouples TC1 and TC2. Cracking coil H 2 consisted of 5.2 meters of 8.7-mm. i. d KAzS tubing, F , with a measured heated volume of 327 ml. It was immersed in a lead-filled furnace (Figure 3 ) , electrically maintained a t the desired temperature and thoroughly insulated, A . The Pyrex cracking unit was somewhat simpler. Metal-to-glass connections and block fittings ( U and P of Figure 2 ) were unnecessary in the furnace since the preheater, H 3 (Figure 3 ) , was sealed directly to the feed system and to the cracking coil, H4. Resistance Z vaporized n-octane in H 3 . The dimensions of H 4 approximated those of H 2 (Figure 2 ) , the corresponding volume being 315 ml. The lead bath was replaced with a packing of Monel metal shavings, Y (Figure 3), which was more difficult to maintain during~. operation at the exact temperature desired. The same condensing and collecting apparatus (Figure 2) served for both Pyrex, G, and steel, F, cracking units. A water condenser, D,retained unchanged n-octane and most of the liquid decomposition products in a graduated reservoir R 3 , and a solid carbon dioxide condenser, J , held condensabie gases in R4. Stabilizer K was an integral part of the apparatus; low-boiling products were distilled from R5 by means of an internal resistance coil, and their boiling points were registered by thermometer, T. Gases from J and K were metered and collected beyond exit XI. Waste exit X 2 opened to the atmosphere. Column 2, employed for the analyticai distillations of the liquid decomposition products, was similar to larger total condensation columns which have been described (5.2). Its packed section measured only 1.1 X 26 cm. The packing consisted of 16 cm. of slightly less than one-turn glass helices (6S), averaging 3.3 and 4.2 mm. in inside and outside diameters; graduated one- to six-turn helices above and below served to keep the fine central material in place. Tests with a benzene-carbon tetrachloride mixture (IS) showed column 2 to be equivalent to ten theoretical plates. For fractionat-
FIGURE 3. DETAILSOF CRACKINGUNIT AND ' PYREX COIL
mm.). All three-carbon products were thus removed from the liquids. Finally, the four-carbon products in the lines and meter were forced into the holder with air. About 1.5 moles of the gas sample were fractionated in a Podbielniak column (36)from which one-, two-, three-, and
MAY, 1938
INDUSTRIAL AND ENGINEERING CHEMISTRY
four-carbon cuts were taken. Traces of heavier material which often remained were considered five-carbon cuts. The unfractionated gas and the cuts were analyzed for unsaturation and for oxygen in a Burrell apparatus fitted with Francis pipets. Olefins were determined by a procedure (45) using 25 per cent oleum, and the additive unsaturation of the several cuts nearly always agreed with that of the entire gas within 1 per cent. This served as a check upon both the fractionation and the cut analyses. I n one run the acid method was compared with bromine absorption, the two methods giving values differing by only 0.2 per cent; such close agreement was probably fortuitous. Because the one-carbon cut was collected without condensation, it contained 5 to 10 per cent ethylene besides the expected methane, air, and traces of hydrogen. The latter was determined by passage over copper oxide a t 275-300” C. Air was estimated from the oxygen absorption in “Oxsorbent.” The molecular weight of the unfractionated gas was determined by means of an effusiometer ; this value verified the calculated additive value, which was employed in determining the total weight of air-free gas. The liquid sample was removed from €25 without escape of four-carbon hydrocarbons by firmly attaching a flask to the bottom of the stabilizer, cooling i t thoroughly with solid carbon dioxide and acetone, and draining the product through 87. The flask with its cold contents was fastened immediately to column 1. No attempt was made to separate the products in this column, since its purpose was to clear the liquid and to remove the major portion of the unchanged noctane. With the distilled cracked liquids was collected an equal volume of eight-carbon hydrocarbons to ensure the passage of the last traces of lighter material overhead. The unchanged hydrocarbon was collected separately; a small residue was left which was subsequently freed of n-octane in a small distilling flask. The colorless cracked distillate was transferred from the receiver to column 2 ; the same precautions were taken to prevent loss of volatile products as before, and the distillate was carefully fractionated a t a rate of 3-6 ml. per hour. The first fraction comprised the four-carbon hydrocarbons, but because the column operated above the boiling point of this material before reflux started, a varying small amount of five-carbon hydrocarbons also distilled. The amount was estimated by evaporating the cold liquid distillate into a warm collecting bottle and determining the molecular weight in a warm effusiometer. This method was not entirely satisfactory, and consequently the cut between four- and fivecarbon material was much less accurate than any other. Higher boiling materials were separated without difficulty. The five-, six-, and seven-carbon fractions were taken overhead, and an eight-carbon residue remained. A modified (32) Francis method (14) was used to determine the olefin content of each liquid fraction. I n certain cases the unsaturation exceeded 100 per cent; this was not considered an indication of di-olefins since various pure alkenes give either high or low characteristic values (32, 48). Chloroform was added to the five-carbon cut to reduce volatilization during the determination. Duplicate analyses checked within 1 per cent, but the accuracy of the method was probably not better than * 5 per cent.
Distribution of Products Table I11 presents condensed data for five runs. The per cent; decomposition was obtained from the weight of cracked products and the weight of n-octane charged to the cracking unit rather than that recovered although, as may be seen from the fair material balances, either method would be acceptable. The time of contact was calculated by
557
averaging the calculated volumes of hydrocarbons entering and leaving the cracking coil a t run conditions. The velocity constants assume a monomolecular reaction (15).
TABLE111. THERVAL DECOMPOSITION OF n-OCTANE IN STAINLESS STEEL AND PYREX COILS F9 Recorded Gats: Temp., C. 538 + 2 Time. hr. 5.37 n-Octane charged, grams 324 Total gas volume (standard temp. and pressure) : T.iters 14 n Mol. weight 28.6 Analytical data: Gas fractionation, mole %: Ci ( f H z ) 33.6 CZ 44.0 16.2 CS C4 6.2 CS 0.0 Liquid fractionation, grams: C4 2.8 C6 2.9 CS 3.6 C7 1.6 “CS” 17.6 - 1 . -
Unsaturation: 7.2 C1 mole 7 cz‘ mole . 35.0 ~ a mole ‘ 86.8 ~ a mole : 93.1 $6, $e, C!, weight % b Cs, weight % 6.9 Complete analysis, mole %: Hydrogen a Methane 26.2 Ethene 22.3 Ethane 16.6 Propene 11.8 1.8 Propane Butene 9.5 Butane 0.6 4.1 Pentene 0.0 Pentane Hexene 4.2 Hexane 0.0 Heptene 2.9 Heptane 0.0 Calculated data: Feed rate, grams/hr. 60.3 Total products, grams 36.4 Recovered n-octane.. =rams 285.4 Tar, grams 3.0 Material balance, % 100.2 Decomposition, % 11.2 Contact time, ti sec. 30.4 3.90 Velocity constant, ?aX 103 c( Moles products 3.16 Mole n-octane decomposed
P ..P 4
5
c
d
F2
F3
F4
G2
570 *4 4.32 480
571 *2 2.91 368
571 *l 4.73 281
572 =4=8 4.33 148
50.1 28.4
39.2 28.8
50.5 30.5
31.0 29.4
29.2 53.4 12.4 4.6 0.4
29.4 48.4 18.6 3.3 0.3
27.2 46.2 17.3 8.6 0.7
30.4 45.3 17.4 6.9 0.0
3.2 5.7 8.7 4.2 41.3
3.4 6.1 7.2 2.6 36.4
1.7 7.2 8.1 4.3 27.3
1.4 4.8 4.2 2.8 14.4
7.5 59.5 89.2 92;O 100 7.5
6.4 59.5 8$5
5.6 58.5 88.6 95,. 2 100 9.3
9.6 58.7 89.0 95.4
0.1 23.6 29.8 18.9 9.7 1.1 5.8 0.5 3.6 0.0 4.0 0.0 2.9 0.0 111.0 88.5 341.7 1.5 89.9 18.4 14.6 14.0 3.30
lOOC 5.0 d
0.1 22.6 25.2 16.9 13.6 1.7 8.4 0.4 4.6
23.8 26.5 16.9 14.1 2.0 5.4 0.4 4.5 0.0 4.2
b
16.4 d
0.0
3.8 0.0 2.7 0.0
24.1 25.8 16.3 13.6 1.6 7.2 0.4 4.4 0.0 3.2 0.0 3.4 0.0
126.3 71.6 292,6 1.0 99.2 19.4 12.7 17.0
59.5 92.8 173.2 2.5 95.5 33.0 24.4 16.2
34.2 56.2 82.5 0.5 93.7 38.0 38.3 15.4
0.0
0.0
2.2
3.24
3.18
3.11
Assumed 94%. b Assumed 100%. Approximate; actual values varied from 98 t o 106%. Included in value for methane.
In these five runs the material in which the hydrocarbon was decomposed, the temperature, and the extent of decomposition were varied. Runs F3 and F4 illustrate the effect of doubling the heating time in steel apparatus a t 571 O C . Runs F4 in stainless steel and G2 in Pyrex show the effect of the coil material a t otherwise similar conditions. Runs F3 and F9 demonstrate the effect of temperature in the steel coil although the extent of decomposition also differed. Run F3 was made to check F2, which was somewhat invalidated by the low material balance, but is included for completeness. Examination of the data shows immediately that there are no pronounced differences in the products of any run. Change of coil material and variation of temperature or decomposition within limits do not materially alter the relative amounts of the resulting smaller hydrocarbons, The molecular weights and distillation analyses of the gases and the unsaturation of the cuts agree well. I n the complete analyses the agreement is good although certain irregular but probably experimental deviations are apparent. Finally, 3.2 * 0.1 molecules of smaller hydrocarbons were always obtained from a single n-octane molecule. The data picture convincingly the distribution of the fragments formed when
one of several molecules of n-octane is decomposed a t atmospheric pressure by heat alone a t temperatures near 570" C. The fairly smooth gradation in the amounts of paraffins and olefins with size is evident. Inspection will show that these analyses do not vary greatly from the corresponding runs of Table 11.
Kinetics The velocity constants of the runs a t 571" C. vary only within the experimental error; hence there is no reason to believe that the reaction is not monomolecular over the range 18-33 per cent decomposition. This is difficult to reconcile
-
0
0
THIS PAPER
11 .8
1.22
1.26
1.30
1000/T
FIGURE4. TEMPERATURE COEFFICIENT OF VELOCITY CONSTANT
with the work of Dintzes and Zherko (8) who observed that the velocity constant fell nearly 50 per cent by increasing the extent of decomposition through this range. Variation of the velocity constant with temperature is shown in Figure 4. The activation energy is in better agreement with the value obtained by Dintzes (6,7) than are those of other simi1a.r pairs-of investigations: E,kcal.
VOL. 30. NO. 5
INDUSTRIAL AND ENGINEERING CHEMISTRY
558
-n-Hexane-. 64.5 57.8
Reference
(6)
(16)
-n-Heptane63.2 46.5 (38)
(34)
-n-Octane64.5 60.1 (7) Present work
The smaller value of the present work should be expected since-the liquid as well as the gaseous products were considered.
methane production. Because in these runs the feed rate was maintained approximately constant, the extent of decomposition varied as well as the temperature. Deviations among the products of runs E18, E19, and E20 might therefore be attributed to either factor, but the logical explanation involves both. At 522' and 539" C. normal pyrolysis is partially hidden by a methane-producing reaction which has a smaller temperature coefficient than thermal decomposition. That the methane is not formed by synthesis from the elements is shown by the absence of hydrogen; the equilibrium vapor of the reaction
C
CHI
comprises nearly equal volumes of hydrogen and methane a t 540" C. and one atmosphere pressure (IO). The reverse reaction may occur at 571' C., as indicated by the considerable amount of hydrogen found in run E19. I n an attempted run a t 593" C. in the conditioned coil, 75 per cent decomposition occurred in 14 seconds contact time, and gaseous products resulted almost exclusively. These were only 6.3 per cent unsaturated and had a molecular weight of 12.0. Assuming that the olefin was ethylene, the other products were 60 per cent methane, 34 per cent hydrogen, and the carbon which quickly and completely plugged the outlet of the coil. The methane and hydrogen were in this case much closer to the methane-synthesis equilibrium values (IO). Since these runs were made previous to those of Table 111, the catalytic effect of carbon or, more probably, metallic carbides was obviously eliminated by burning the apparatus with oxygen. The catalytic effects of similar steels have been both reported (3, 11) and denied (18). No such carbonization or surface conditioning was observed in the Pyrex coil. The question of the source of the methane makes further examination of Table IV interesting. The parenthetical values show the analysis of products other than methane and
THERMAL DECOMPOSITION OF WOCTANEIN PRESENCE OF CARBON
TABLEIV.
E20 Recorded data: 52215 Temp., O C. 16.00 Time, hr. n-Octane oharged, grams 1654 Total zas voc (standard temD. ana pressurej 24.0 Liters 20.5 Mol., weight Total liquid product, grams 14.0
E18
E19
53912 5.75 744
57112 2.25 224
29.7
28.1 21.6 12.1
21.0
19.9
THE
Ca. mole % C4 mole % Ch' weight Cs' weight C,:weight
d 3.6 0.7 57.2 48.5 63.4 10 7 (29)b 12.4(29)b 19.6 (41)b 8.1 (19) 12.4 (26) 6.8 (18) 2.8(7) 2 . 6 (7) 3.8 ( 8 ) 2,l(5) 0.9 (3) 0.6 (1) 2.2 (5) 3.7 9) 4.1 (11) 0.1 ( 0 ) 0.4el) 0.2 (1) 3.0(7) 3.1 (7) 0.3 (1) 0.1( 0 ) 6.3 (15) 5.3 (15) 4.1 ( 8 ) 0.0( 0 ) 0.0( 0 ) 0.0( 0 ) 3.0(7) 1.9(4) 3.6 (10) 0.0( 0 ) 0.0( 0 ) 0.0(0) (I
Effect of Conditioned Surface
' Certain differences in the products of Table I11 and of Tables I and I1 are evident. The absence of appreciable hydrogen is in disagreement with the work of all previous investigators except Spanier (44, see also 10 and 11A). A possible explanation is that his analyses are the only ones reported for nonmetallic apparatus. Although the decomposition is exactly the same in stainless steel as in Pyrex, this is true only if the metal coil is thoroughly burned out between runs. The data of Table IV illustrate the effect upon the products of a reaction-conditioned surface, obtained by an extended period of hydrocarbon decomposition without intermediate burning. These runs were of a preliminary nature and in the absence of material balances cannot be considered as accurate as the runs of Table 111. The effect of the conditioned coil is almost entirely one of increasing
+ 2H1*
Propene Propane Butene Butane Pentene Pentane Hexene Hexane Heptene Heptane Calculated data: Feed rate grams hr. Deoornpo&tion, Time of contact, sec. Velocity constant, hi X l o s Moles products Mole n-octane decomposed
k
2;
103 2.2
20.0
1.14 3.97
129 6.4 14.3 4.6 3.79
100 17.5 15.5 12.4 4.11
a Included in value for methane. b Figures in parentheses are recomputed values, neglecting methane and hydrogen.
MAY, 1938 TABLEV.
INDUSTRIAL AND ENGINEERING CHEMISTRY EXPERIMENTAL AND PREDICTED PRODUCTS OF WOCTANEDECOMPOSITION -Predicted300' C. (69A) 0
Hydrogen Methane Ethene Ethane Propene Propane Butene Butane Pentene Pentane Hexene Hexane Heptene Heptane Moles products Mole n-octane decomposed
20 35 15 10
C. (67) 0 17
600'
1000' C. (39A)
44
15 8 0
0
5
0
7
0 4 0 4 0 4 0 4
4 0 4 0 4 0 4
0
5
0
5
0
5
0
0
14
15 49
3 2 ''
Obtained. 571' C. 0.1
23.5 26 0 16 5 13 5 1.8 7 2 0 5 4 3 0 0
3 8
0 0
0
2.8 0 0
..
3 2
559
Application to Free-Radical Theory Comparison of these results with those predicted by the free-radical theory shows a general similarity but important differences. An average analysis of the products obtained a t 571' C., weighted for the better runs, is listed in Table V with predicted analyses (37, S9A) a t various temperatures. The pronounced ethylene discrepancy is not readily explained; study will show that no simple revision will fit the theory to the experimental data. The differences are lessened if it is assumed that the 2-octyl radical is produced in larger amounts than the 3-octyl radical and this in turn in larger amounts than the 4-octyl radical. The higher olefins, which according to the theory have but one possible source, are not produced in the equal amounts demanded (39)by the prediction but form a graded series. This is not surprising; the nature of the groups R and R' in a radical such as
1 i hydrogen. These values and those of unsaturation show R-/-cH~-cH-cH~+R' much greater variation than the later runs (Table 111),and I I I only the most pronounced trends can be considered significant. At small decompositions the highest boiling products appear more abundant, TABLEVI. DIRECTION O F SPLITTING O F ALKYLRADICALS but this is partially if not entirely true because T ~ ~ CHz=CHCHzR' ~ . , the enhanced experimental losses are largely Source Radical R RJ ' C. RCHPCH=CHZ Reference concentrated in the volatile condensable prodn-Butane 2-Butyl Methyl H 575 4 7 (16) 600 3 9 ($3) ucts. That the decreased amount of three- to 600 3 0 (35) five-carbon material is the result of specific den-Pentane Ethyl 650 3 9 (33) 396 1 8 (15) generation to methane is therefore very doubtful. 560 8 8 (15) It might be expected that the unsaturation would 650 5 0 (31) 425 3 3 (15) fall greatly with increased degeneration of the n-Hexane 575 7 4 (16) product, since olefins are much more susceptible 3-Hexyl Ethyl Methyl 425 0 9 ( 15 ) 57 5 2 4 (15) to metal catalysis than paraffins (20). The unn-Octane 2-Octyl Amyl H 571 Very large This paper saturation data show with reasonable certainty 3-Octyl Butyl Methyl 571 2 6 This paper 4-Octyl Propyl Ethyl 571 1 1 This paper that this is not the case, a small decrease with temperature being expected (16). Thus the catalytic effect does not involve the products of would a priori be expected to influence the relative rates of decomposition; nor does it involve the n-octane molecule, since the velocity constants for its decomposition in the splitting a t the two dotted lines. The larger group should conditioned coil are, within the estimated experimental error, part from the radical more readily and several papers indicate that this is true. Table VI lists the ratios of the two oleidentical to those for the uncatalyzed reaction, as may be fins, CHF=CHCHZR' and RCH&H=CHZ, experimentally seen from Figure 4. This interpretation appears logical, and the conclusion that intermediate fragments, such as free produced from the theoretical radicals a t various temperaradicals, are involved in the degeneration to methane seems tures. Hydrogen is split from 2-alkyl radicals in decreasing inescapable. proportion as their length is increased. Radicals split from
I
I
I
I
I
I
I
I
100
MONO-OLEFINS
LIQUID PRODUCT I
I-
Ba
70
50
SATURATES
40
30
0
20
FIGURE 5.
40
60 0 2 MILLILITERS
4 6 0 DISTILLED
IO
20
70
FRACTIONATION O F LIQUID DECOMPOSITION PRODUCTS
40
INDUSTRIAL AND ENGINEERING CHEMISTRY
560
the 3- and 4-alkyl radicals directly in proportion to their lengths, and the smaller of the two olefins consequently predominates in the product. As it stands, the free radical theory cannot predict with accuracy the products of thermal decomposition of paraffins as large as n-octane, and considerable further experimental data upon which to base generalizations must be obtained before the theory can be developed and applied to these more complex cases.
Preliminary Separation of Liquid Products The fractionated liquid products of like carbon content of these and several unreported runs were combined and refractionated through column 2 . The results are presented in Table VI1 and Figure 5 . The refractive index of the five-carbon product, half of which was lost previous to refractionation, traced a curve with a maximum, a minimum, and a sharp rise somewhat before “breaking” to six-carbon products, proving the presence of a t least three compounds. Fractionation B (Table VII) separated large amounts of a nearly pure hexene. I n fractionation C the refractive index curve again reflected the presence of three seven-carbon products. Although the product appeared relatively simple, unexpected hydrocarbons were definitely present. After doubling the amount of material available for further investigation by repeating the treatment upon fractions from similar runs, paraffinic hydrocarbons were separated from the large amounts of olefins by the general method of Brame and Hunter (1). Bromine was slowly added with agitation to the undiluted fractions a t 0 ” C., with no attempt to saturate the alkenes a t this point. The light yellow solution was distilled from a Claisen flask a t 20-30 mm.; the brominated olefin was distilled as completely as possible from the small residue to eliminate the possibility of reducing the concentration of higher boiling dibromides. Unbrominated hydrocarbons which distilled first were condensed in a solid carbon dioxide trap. It was then easy to complete the separation by repeating the bromination upon this material, distilling the unreacted paraffin, and combining the additional dibromide with the main portion.
Saturated Products Negligible reaction occurred when the paraffinic distillates were shaken with several portions of cold concentrated sulfuric acid. After being washed with water, the colorless products were fractionated through column 2 from a small amount of anhydrous alkali carbonate. The distillation curve of the five-carbon material before it was washed with
TABLEVII. Fraction No.
VOL. 30. NO. 5
acid is shown in Figure 5. A small amount of n-octane was purposely added to force the lighter hydrocarbon overhead. After being washed with acid, the hydrocarbon was again distilled (Table VIII-A). The pentane present was clearly n-pentane: B. P. (760 Mm.), C. ngo Product n-Pentane (41)
36.2 36.0
1,3576 1.3577
Fraction C2 showed the boiling point of n-hexane. However, instead of lowering the refractive index, the first bromination had raised it to a value far above that of n-hexane. The second bromination showed that 40 per cent hexenes remained after the first and increased the refractive index by an amount to be expected from 17 per cent of benzene in n-hexane. Comparison with experimental data on this system (47) showed that the behavior of the six-carbon material closely followed that of a mixture of n-hexane and benzene. Although the presence of benzene could not be proved, t h e behavior could be explained by no other hydrocarbon. Fractionation of the seven-carbon product revealed a doubtful plateau a t 97” C. which may well have been due to nheptane. TABLE VIII. Fraction
No.
1 2 3 1
2 3 4 5 1 2 3 4
FRACTIONAIION OF NONOLEFINIC HYDROCARBONS
Refractive Index Volume = c. n %, Ml. % A. Five-Carbon Product 3 4 6-35.6 1 ,3582 0.9 19 35.6-36.3 1.3576 3.7 81 Residue (n-octane) ,... 2.0 .. B. Six-Carbon Product (before Second Bromination) 46.0-62.4 1 3880 3.0 28 62.4-66 2 1.3903 3.0 28 66.2-69 1 1.3948 3.0 28 69.1-ca. 90 1.3964 1.7 16 Residue (n-octane) .... 3.6 .. C. Six-Carbon Product (after Second Bromination) 61 .4-67.6 1.3953 1.6 39 1.3965 1.4 34 67.6-69.6 69.6-94.4 1.4048 1.1 27 Residue 1.4031 ... ..
B. P. (760 Mm.)
Unsaturated Products
The olefins were regenerated from their dibromides by means of zinc in ethanol. The volatile pentene and hexene were distilled continuously with ethanol from the reaction flask into a solid carbon dioxide trap, and the heptenes were separated from the undistilled solution by the addition of water. Fractionations of the regenerated water-washed olefins are presented in Table IX and Figure 5. Fractions A1-3 FRACTIONATIOX OF COMBINED LIQUIDPRODUCTS were the expected 1-pentene; its best physical constants compared favorably with synthetic material.
B. P. (760 Mm.) Refractive Index
c. A.
n %o Five-Carbon Product
Volume
Mi.
B. P. (760 Mm.),
% Product 1-Pentene (48)
31.3 30.1
’ C.
n2,0
1.3728 1.3711
The high values were due to the presence of an isomer. The best constants of the latter indicated one of two pentenes: 1-2 3-8 9-10 11
1 2 3 4 5
B. Six-Carbon Product 1.3880 34.6-62.9 1.3887-1.3891 62.9-64.6 1.3895-1.3905 64.6-65.5 Residue ..... (added to C ) C. Seven-Carbon Product 61.8-67.2 1.3930 67.245.7 1.4030 85.7-92.4 1.4018 92.4-93.2 1.4030 Residue .....
9.2 49.5 11.8 4.4
16 63 15
4.0 3.0 3.9 8.6 4.0
17 13 16 37 17
6
Product 2-Pentene (4% 2-Methyl-2-butene(48)
36.5 36.4 38.4
1.3867 1.3797 1.3878
Although the refractive index indicated 2-methyl-Z-butene, extended refluxing of the dibromide with water failed to give methyl isopropyl ketone, the readily identified product (18) of the hydrolysis of 2,3-dibromo-2-methylbutane. T h e material must therefore have been 2-pentene, as indicated by the boiling point. The pentene mixture, up to this point, appeared very similar to that obtained by pyrolysis of 1-
MAY, 1938
INDUSTRIAL AND ENGINEERING CHEMISTRY
pentene itself @2). However, a trace of olefin of high index (fractions A7 and C1, Table IX) appeared before the hexenes distilled. Since the previous treatment had eliminated all substances except monoolefins, this must have been cyclopentene: B. P. (760 Mm.), C. n go About 43 44.2
Product Cy(-lopentene (9)
TABLE IX. FRACTIONATIOS OF MONOOLEFINS Fraction No.
B. P. (760 Mm.),
O
C.
1.4026 1.3997
The index was again high, owing to the probable presence of other straight-chain heptenes. Their physical constants (43) show that they were not present in large amounts. A further small amount of dibromoheptanes was obtained by recovering the water-insoluble products obtained in the Francis analyses of the “eight-carbon”residues of all runs and by treatment with bromine of the material not required for analysis. After regeneration, the olefins were distilled (Table IX E ) through a small column packed with glass helices (53) and capable of handling 2 to 5 ml. of material. The mixture evidently contained cycloalkenes, probably methylcyclohexenes : B. P. (760 Mm.), T o 103 102-112 121-125
TABLEX. 7 -
Hydrocarbon 1-Pentene 2-Pentene Cyelopentene n-Pentane
Cs Fraction
Vol. % i n fraction
Vol. % of total
66
37
3 24
2 13
7
4
C.
% 68 12 5 5 10
11 83 6 16 57 10 8 9 29 10 46 15 42 28 30
This product could not have been confused with the straightchain octenes, which would result from dehydrogenation of n-octane but were definitely absent.
Source of Liquid Products
Above 1,4164 1.4465
93.7 93.2
Product Methylcyclohexenes (Beilstein) Straight-chain octenes ( 6 7 )
1-2 3 4-6 7-8
1.3936 1,3928 1.3942
The largest amount of the seven-carbon olefins had physical constants similar to those of 1-heptene: Product 1-Heptene ( 6 4 )
5 6
1 2 3
Either or both were indicated, and the small amount of material was not further examined. The first “heptene” fraction (Dl-3) contained a six-carbon monoolefin which, because of its high index, must have been a cycloalkene and was probably cyclohexene: About 85 83 4
Volume MI.
Pentenes 1.3728-1.3744 29.1 30.6-33.2 33,2-36,0 1.3805 5.1 36.0-37,O 1.3869 2.1 37.0-39.6 1.3919 2.3 39.6-71.6 1.4080-1.3971 4.2 B. Main Hexenes 54.0-63.2 1.3860 4.4 63.2-64.1 1.3880-1.3885 32.4 Residue 1.3914 2.2 C. High- and Low-Boiling Hexenes 56.7-63.8 1.3914 4.2 63.8-64.7 1.3898-1.3905 15.0 64.7-65.4 1.3917 2.6 65.4-66.2 1.3936 2.2 Residue 1.4093 2.5 D . Main Heptenes 60.7-90.9 1.3896-1.4164 12.2 90.9-92.6 1.4074 4.1 92.6-94.2 1.4026-1.4037 19.5 94.2-96,2 1.4069-1.4168 6.6 E. High-Boiling Heptenes 91-99 1.4219 2.1 99-103 1.4304 1.4 Residue ..... 1.5
1 2-3 4
ny
66 67.9-68.1 66.6-67.0
Product Cyclohexene ( 2 4 )
ng
1 2-4 5
1.3884 1,3858
The higher and lower boiling azeotrope fractions were combined, thoroughly washed with water, and fractionated. Fractions C2-3 (Table IX) contained 1-hexene, but C5 and the generally high refractive indices reflected a definite isomer content I Product 2-Hexene (40) 3-Hexene ( 4 0 )
Refractive Index
c. 1-3 4 5 6 7-8
Above 1.4080 1.4224
63.8 63.4-63.7
B. P. (760 Mm.)
A.
The ethanol solution of the hexenes was fractionated, .giving first a ternary ethanol-water-hexene azeotrope boiling a t 42“ C. (747 mm.) and then the binary ethanol-hexene azeotrope, much of which boiled a t 56.0” C. (747 mm.) with an n’,” value of 1.3821 and contained 23 per cent ethanol. The constant-boiling portion of the azeotrope was washed with water to give olefin with a refractive index (nz$ of 1.3883. Refractionation of part of this showed that the hexene was fairly pure (Table IX-B) ; the best constants proved it to be mainly 1-hexene : Product 1-Hexene (40)
561
n %’
To 1.430 T o 1.450 1.409-1.415
Table X presents a semiquantitative analysis of the fivecarbon to seven-carbon decomposition products. Because of the varied conditions under which they were produced, it is not possible to state with finality the manner in which each was formed. The large amounts of 1-alkenes indicate that they are the initial products; rearrangement to isomeric olefins and cyclization to cycloalkenes appear to follow. Not unexpectedly, 1-heptene is transformed to a greater extent by subsequent reactions than the lower 1-alkenes. Other gradations with carbon content of the amounts of the various types of product are evident here as in Table V. Aromatic hydrocarbons may well be products of the following series of reactions : 1-alkene (+ 2-
and 3-alkenes)
+ cycloalkene +
aromatic hydrocarbon
This explanation differs from the many theories involving acetylene and 1,3-butadiene although it is similar in certain respects to that of Wheeler and Wood (61). The excess gaseous product from one run and the combined four-carbon cuts from several runs were brominated, and the dibromides fractionated under diminished pressure. I n neither case was a crystalline residue obtained which would have revealed the presence of tetrabromides of 1,&butadiene or acetylene. Aromatic compounds result in larger amounts as decomposition conditions are intensified, and in the present work, they were probably formed mainly during the runs in the conditioned coil (29). Conversely, if the data have been correctly
ANALYSIS OF LIQUIDMO OCTANE DECOXPOSITION PRODUCTS Hydrocarbon 1-Hexene 2- and 3-hexenes Cyclohexene Benzene (?) ?%-Hexane
CsFraction Vol. % in fraction
80
8 7 1 4
Vol. % of total 24 2 2 Trace 1
C i Fraction
Hydrocarbon 1-Haptene 2- and 3-heptenes (7) Methylcycloherenes n-Heptane ( 1 )
Vol. % in fraction
Vol. % of total
63 12 20
13 2
5
4
1
INDUSTRIAL AND ENGINEERING CHEMISTRY
562
interpreted, the initial products of decomposition of nalkanes are indeed 1-alkenes and smaller n-alkanes. High-boiling residues were obtained in all runs, but because of the small amounts they could not be thoroughly investigated. More than half of a composite sample distilled below 200” C. The variation in the amount produced in different runs is probably not significant, since a t least part of this material resulted from polymerization during stabilization and distillation of the liquid products.
Conclusions 1. The n-octane employed was obtained without difficulty and in satisfactory yield and purity from mid-continent petroleum. 2. Thermal decomposition a t atmospheric pressure and 571” C. led to a long but simple series of products which differed quantitatively from those predicted by the free radical theory in the amounts of ethylene and of the heavier olefins formed. The product analyses were the same when decomposition was increased from 18 to 33 per cent, when carried out in Pyrex instead of stainless steel apparatus, and differed but little from those obtained a t 538” C. 3. The ethylene deficiency cannot be readily explained, but the amounts of higher olefins can be accounted for by assuming that octyl radicals break unequally in the two possible directions. 4. At temperatures below 571 O C. stainless steel apparatus which had been in continuous use for an extended period gave products which contained abnormally large amounts of methane. The other products were not greatly changed. Above 571” C. both methane and hydrogen resulted to the exclusion of heavier products. 5 . Fractionation and chemical treatment of the liquid decomposition products showed that the saturated hydrocarbons retained the linear structure of the n-octane, as assumed by the free radical theory, but that the 1-alkenes apparently first produced were accompanied by small amounts of other unbranched alkenes and cycloalkenes.
Acknowledgment It is a pleasant duty to thank W. E. Kuentzel for assistance in assembling the apparatus, B. L. Evering for aid throughout the early work, R. F. Ruthruff for general advice, and J. C. Stauffer for fractionating the gaseous products.
VOL. 30, NO. 5
Friedmann, Ber., 49, 1344 (1916). Gault and Hessel. Ann. chim. phys., [IO]2, 31”9 (1924). Groll, ITID. ENG.CHEM.,25, 784 (1933). Huge1 and Szayna, Ann. combustibles liguides, 1, 781, 817, 833 (1926). Hurd, IND. ENG.CHEM.,26, 50 (1934). Hurd and Goldsby, J. Am. Chem. SOC.,56, 1812 (1934). Hurd, Goodyear, and Goldsby, Ibid., 58, 235 (1936). Hurd and Pilgrim, Ibid., 55, 4902 (1933). Kistiakowsky, Ruhoff, Smith, and Vaughan, Ibid., 58, 137 (1936). Leslie and Schicktanz, B u r . Standards J. Research, 6, 377 (1931). ENG.CHEM.,22, 953 (1930). McKee and Seayna, IND. Mavity, Ph.D. thesis, Ohio State Univ., 1931. Moldavskii and Kamuscher, Compt. rend. m a d . sci. U . R. S. S., IN. S.l 1936. 355. Moldavskii and others, J . Gen. Chem. (U. S. S. R.), 5, 1791 (1935); 7, 1840 (1937). Ibid., 6, 617 (1936). ENG.CHEM.,27, 1082 (1935). Morgan and Munday, IND. Mulliken and Wakeman, IND.ENG. CHEM..Anal. Ed., 7. 59 (1935). Neuhaus and Marek, IND. ENG. CHEM.,24, 400 (1932). Pease and Morton, J . Am. Chem. SOC., 55, 3190 (1933). Petrov, Meschtscherjakow, and Andrejev, Ber., 68, 1 (1935). Podbielniak, IND.ENG.CHEM.,Anal. Ed., 3, 177 (1931). Rice, IWD. ENG.CHEW,26, 259 (1934). Rice and Johnston, J . Am. Chem. Soc., 56, 214 (1934). Rice and Rice. “Aliohatic Free Radicals.” . D. - 64. Baltimore. Johns Hopkins Press, 1935. (39A) Ibid., pp. 91-107. (40) Schmitt and Boord, J. Am. Chem. SOC.,54, 754 (1932). Shepard, Henne, and Midgley, Ibid., 53, 1948 (1931). Sherrill and Walter, Ibid.,-58, 742 (1936). Soday and Boord, Ibid., 55, 3293 (1933). Spanier, dissertation, Karlsruhe, 1910. ENG.CHEM.,27, 1072 Sullivan, Ruthruff, and Kuentael, IND. (1935). Tocher, J. Soc. Chem. I n d . , 13, 231 (1894). Tongberg and Johnston, IND. ENG.C H ~ M25, . , 733 (1933). Tongberg, Nickels, Lawroski, and Fenske, Ibid., 29, 571 (1937). Tropsch and Mattox, Ibid., 26, 1338 (1934). Waterman and Perquin, J . I n s t . Petroleum Tech., 11, 42 (1925) ; 14, 318 (1928); 15, 369 (1929); 16, 29 (1930). Wheeler and Wood, J . Chem. SOC.,1930, 1819. Whitmore and Lux, J . Am. Chem. SOC., 54, 3448 (1932). Wilson. Parker. and Launhlin. Ibid.. 55. 2795 (1933); Roper, Wright, Ruhoff, and S&th, Ibid., 57, 954 (1935) ; Young and Jasaitis, Ibid., 58, 377 (1936). Yurev and Pavlov, J . Gen. Chem. (U. S . S. R.), 7, 97 (1937). (16) (17) (18) (19)
I
RH~CBIIVED November 8. 1937. Presented before the Division of Petroleum Chemistry e t the 94th Meeting of the American Chemical Society, Rochester, N. Y.,September 6 to 10, 1937.
Literature Cited (1) Brame and Hunter, J.I n s t . Petroleum Tech., 13, 794 (1927). (2) Bruun and others, B u r . Standards J.Research, 6, 363, 869 (1931) ; Payne and Lowy, IND.ENG.CHEM.,24, 432 (1932); Perrin 55,4136 (1933). and Bailey, J . Am. Chem. SOC., (3) Cambron and Bayley, Can. J. Research, 9, 583 (1933). (4) Conant and Mendum, in Hurd’s “Pyrolysis of Carbon Compounds,” A. C. S. Monograph 50, p. 73, New York, Chemical Catalog Co., 1929. (5) Dintaes, Compt. rend. acad. sci. U . R. S. S., [N. S.]1933, 153. (6) Dintees, K h i m . Tverdogo Topliva, 4, 381 (1933). (7) Dintaes and Frost, J . Gen. Chem. (U. S. S. R.), 4, 610 (1934). (8) Dintzes and Zherko, Ibid., 6, 68 (1936). (9) Dolliver, Gresham, Kistiakowsky, and Vaughan, J . Am. Chem. Soc., 59, 831 (1937). (10) Egloff, Schaad, and Lowry, J . Phys. Chem., 34, 1617 (1930). (11) Egloff, Thomas, and Linn, IND. ENQ.CHEM.,28, 1283 (1936). (11A) Engler-Hofer, “Das Erdol,” Vol. I, p. 574, Leipsig, Verlag von 8. Hirsel, 1913. (12) Evers, Rothrock, Woodburn, Stahly, and Whitmore, J. Am, Chem. SOC.,55, 1136 (1933). (13) Fenske, Tongberg, and Quiggle, IND.ENG.CHEM.,26, 1169, 1213 (1934). (14) Francis, Zbid., 18, 821 (1926). (16) Frey and Hepp, Ibid., 25, 441 (1933). ~
Economics of Some of the Less Familiar Elements-Correction In my article under the above title [IND. ENG.CHEM.,30, 4316 (1938) 1, some of the calculations based on geophysical approximations are erroneously given. The error rests in the computation of the number of tons per cubic mile and the subsequent misplacement of the decimal point. On page 431, column 2, lines 1 and 2, the figures should read: 71 869.100 2,966,038 ~- . tons of
titanium +-conium
466 313 tons of lithium 114:078 tons of beryllium
On page 432, column 1, line 3, the reference to bromine in sea water should read: ‘(.. . .(about 314,592 tons per cubic mile of brine).” Page 432, column 2, line 32 should read: “. . .an average cubic mile of the lithosphere contains 144,078 tons of beryllium metal.” H. CONRADMEYER
.