I
JOHN J. MADISON and RICHARD M. ROBERTS Shell Development Co., Emeryville, Calif.
Pyrolysis of Aromatics and Related Heterocyclics Molecular structure can be correlated with the ability to condense and form products of higher molecular weight
THERMAL
cracking of heavier fractions of petroleum is generally accompanied by the formation of products of higher molecular weight than the starting material. Of the constituents separable by chromatography, the aromatics and resins are especially prone to undergo condensation reactions (70). The work reported here was part of a program of thermal cracking studies in these laboratories and was undertaken to obtain information on the features of molecular structure responsible for the products of high molecular weight resulting from the cracking of aromatics. T h e condensation of 43 aromatics and related heterocyclics was measured by subjecting them to pyrolysis in sealed borosilicate glass capillary ampoules in the absence of air. Experiments with anthracene showed that concentration had a profound effect on rate of condensation. Condensation was rapid only in liquid phase pyrolysis, and for this reason, and because practical thermal cracking of heavy petroleum stocks is carried out in liquid phase, nearly all the tests reported were carried out in liquid phase. T h e compounds studied were of high purity, available in limited amounts. Therefore about 100 mg. were used in each experiment. After thermal treatment, gas, CB-CIO, and heavy residue production were measured. In some experiments, ultraviolet and mass spectrometric analyses were performed on the products between Clo and heavy residue. Previous work in this field was done almost entirely by Tilicheev (27), who reported autoclave pyrolyses of 11 of the compounds in the present study.
Comparisons with Tilicheev's results are made below where possible.
Experimental Procedure Liquid Phase Pyrolysis. Liquid phase pyrolysis was carried out in sealed borosilicate glass capillary tubes of about 2mm? inside diameter, and about 8- or 9mm: outside diameter. Prior to loading, the capillary tube was drawn down at one end and sealed off, and the drawn end was bent into a crook to facilitate breaking of the ampoule after pyrolysis. The quantity of feed weighed into the prepared capillary and the length of the tube were so chosen that the tube was about half full of liquid during pyrolysis, thus minimizing the fraction of feed in the vapor phase. T h e loaded ampoule was attached to the vacuum line with heavy-walled gum rubber tubing, degassed, and sealed under high vacuum while in liquid nitrogen. T h e sealed ampoule, together with the residue of capillary tubing left on the vacuum line, was reweighed to make certain no reactant was lost during the evacuation and sealing operation. Very few capillary tubes failed in this investigation. The heating arrangement consisted of a six-element, 36-inch Hevi-Duty electric furnace, fitted with a steel liner, within which was an aluminum cylinder. Through this cylinder three longitudinal 7 / ~ - i n c hholes were drilled symmetrically about the axis. In one hole was placed a platinum resistance thermometer for measuring the temperature within the zone of the aluminum cylinder in which pyrolysis was carried out. The following method was employed to determine rates of heating and quench-
ing: By means of a stirrup and pulley, the reaction capillary ampoule and a dummy ampoule could be lowered simultaneously into two holes of the cylinder. The dummy ampoule was of about the same dimensions as the reaction ampoule and was loaded with a quantity of ground glass. A thermocouple was placed in the ground glass of the dummy ampoule, and, as both ampoules had very nearly the same dimensions and the same heat capacity, the temperature indicated by the thermocouple should have been close to the temperature of the charge in the reaction ampoule. The furnace was heated so as to provide two temperature zones-a lower (reaction) zone and an upper (preheating) zone about 25' C. hotter than the reaction zone. T o carry out a run, the reaction ampoule and the dummy tube were lowered simultaneously into the preheating zone, and after the temperature of the thermocouple reached the reaction temperature, the two tubes were lowered simultaneously into the reaction zone. After pyrolysis the ampoules were removed from the furnace and cooled in a blast of cold air. T h e ampoules could be heated to reaction temperature very quickly without overheating. The cooling rate was even more rapid. A typical temperaturetime curve is shown. Attached to the lower end of the furnace was a viewing chamber into which the ampoules could be lowered 'momentarily, to determine the presence or absence of a meniscus and its position in the tube. The viewing operation could be carried out with a momentary temperature drop of only a few degrees. VOL. 50, NO. 2
FEBRUARY 1958
237
10
0
20
30
TIME. M I N U T E S
Typical heating curve in liquid phase pyrolysis
T o separate the products the reaction ampoule (cooled to room temperature) was sealed into the separation apparatus shown. A steel rod was allowed to rest on the crook of the ampoule so that the ampoule could be broken open by raising the rod with a solenoid and dropping it on the crook. The bottom of the ampoule was embedded in aluminum granules, which served to cushion the shock at the bottom and to facilitate heat transfer to the ampoule. The rest of the product separation train consisted of a series of traps, manometer,
Toepler pump, greaseless gas-measuring receiver (75), thermocouple vacuum gage, and connections to a high vacuum manifold through packless metal valves. The system was entirely greaseless and rubber-free. After the reaction ampoule had been cooled with liquid nitrogen and the traps immersed in liquid nitrogen, the separation train was evacuated to a pressure of 0.1 micron of mercury. After breaking the ampoule with the steel rod, the permanent gases were transferred to the gas-measuring reTO TOEPLER PUMP, GAS MEASURING RECEIVER, AND MANOMETER
-
t THERMOCOUPLE GAGE VACUUM
/-
m
I'
SOLENOID
E
REACT:-'-
SINTERED GLASS GRANULES
D
TFLAPBI
TK RAPC
To separate reaction products, ampoule was sealed into product separation train and broken when steel rod, actuated by solenoid, was allowed to fall
238
INDUSTRIAL AND ENGINEERING CHEMISTRY
ceiver by means of the Toepler pump, while the traps were kept at liquid nitrogen temperature. When the system was pumped out to a few tenths of a micron of mercury, the steel rod was raised to the top of the ampoule-breaking assembly and sealed off from the separation train at point E. The tube containing the ampoule was heated with a Glas-Col heating mantle to 3020' C. for 30 minutes, then to 3520" C. for 15 minutes, while the traps were rnaintained at liquid nitrogen temperature. Tests showed that this procedure completely removed Cdg paraffin wax from an ampoule. 4 n y additional incondensable gas released during this distillation was transferred to the gas-measuring receiver, where the total sample was measured, mixed, and sampled for mass spectrometric analysis. The tube containing the ampoule with any remaining residue was then sealed off from the separation train. Cz to Cd gas product was obtained by maintaining traps B and C at -78' C. and warming trap A to -30" C. The gas liberated was pumped into the gasometer until the pressure of the system fell to about 200 microns of mercury. The C r C 4 gas sample in the gas-measuring receiver was then measured, and a sampIe was condensed into a glass tube and sealed off for subsequent mass spectrometric analysis. Experience showed that this procedure removed Cd, with some of the CS products. A liquid fraction (CS-CIO)was taken by cooling the liquid product trap (trap D) in the liquid nitrogen, while C., trap maintaining trap A at -30' B at room temperature, and trap C a t -220' C. (to reduce transfer of mercury vapor to trap D). When all of the liquid was collected, as evidenced by a fall in pressure to a few microns of mercury, the product was transferred to the tip of the trap and sealed off. The weight of Cs-Clo product was determined by weighing the tube before and after removal of its contents for mass spectrometric analysis. As Cb through Clo products were separated from higher boiling products by a low pressure distillation without reflux, the separation was not sharp. The material remaining in trap A constituted the "high boiling" fraction, including unreacted feed. This product was measured by separating the trap from the train and weighing the trap before and after quantitative removal of its contents. Occasionally this product was analyzed spectrophotometrically. The weight of heavy residue remaining in the ampoule was determined by weighing the ampoule before and after removal of the residue with organic solvents or hot nitric and sulfuric acid mixture. The above experimental technique
P Y R O L Y S I S OF A R O M A T I C S I
was used for samples ranging from 25 to several hundred milligrams; the material balance very seldom was less than 98%. In some cases the sealed tubes containing the higher boiling fractions were not opened for analysis. Therefore weights of this fraction were not obtained in these cases and are indicated by dashes in the tables. Vapor Phase Pyrolysis. The single low pressure vapor phase experiment reported (Table I) was made in a borosilicate glass ampoule of about 190-cc. volume (shown). The reaction vessel consisted of a 32-mm. outside-diameter, standard wall borosilicate glass vessel fitted with a thermowell, a break-seal, and a side arm for attaching the vessel to the separation train. The feed sample was weighed into a small tube, which was then inserted into the ampoule through small tube A . The ampoule was then attached to the vacuum line through tube A , cooled, evacuated, and sealed off. Two recording thermocouples were placed in the thermowell of the ampoule and the pyrolysis was carried out as for liquid phase experiments (except that the aluminum cylinder and the platinum resistance thermometer were not used). Before separation of the products, the break-seal was opened, and the steel rod was removed from the system as described. The subsequent separation procedure was the same as for liquid phase pyrolysis. A high pressure (67 pounds per square inch absolute) vapor phase pyrolysis (Table I) was carried out by using a heavy-walled borosilicate glass tube of 23 cc. internal volume. This ampoule had a crook-shaped tip and was handled in the same manner as the capillary ampoules. Purification and Properties of Compounds Studied. About half of the compounds used in this study had been prepared for use as spectroscopic standards and their purities were determined calorimetrically (22). ACENAPHTHENE. Picrate was recrystallized twice from absolute ethyl alcohol and chromatographed on silica gel with a benzene and hexane mixture. The product was recrystallized twice from ethyl alcohol and sublimed; 99.9 mole % purity by calorimeter. ACRIDINE.Complex with trinitrofluorenone was recrystallized, decomposed with hydrochloric acid, and neutralized. The product was sublimed; 99.86 mole yo purity by calorimeter. AMYLNAPHTHALENE, distilled from Sharples Chemical Co. Pentalene 195 (over 90% 2-sec-amylnaphthalene) ; boiling point 293" to 294' C.; nZDO = 1.5737. Analysis calculated for C~sHls: C, 90.85; H, 9.15. Found: C, 90.85; H, 9.29. ANTHRACENE, Eastman Kodak Co. 480-X, "blue-violet fluorescence."
BENZ[UIANTHRACENE, recrystallized three times from a mixture of benzene and methanol. 11-H-BENZO[b]FLUORENE, 99.6 mole % purity by calorimeter. BENZO [U~PYRENE,recrystallized three times from n-heptane. Analysis calculated for CZoH12: C, 95.2; H, 4.78. Found: C, 95.1; H, 4.9. BIBENZYL,recrystallized three times from 95% ethyl alcohol and sublimed; 99.97 mole 7 0 purity by calorimeter. BIPHENYL,recrystallized three times from 75 to 25 ethyl alcohol-water, chromatographed on silica gel, recrystallized three times again from aqueous ethyl alcohol. Meltins Doint 69.0' to 69.3' C. (corr.). CARBAZOLE.Picrate recrvstallized three times from ethyl alcoho1,'chromatographed on silica gel, recrystallized three times from benzene, sublimed. CHRYSENE, recrystallized once from xylene, once from benzene, then sublimed. Melting point 254' to 255' C . Analysis calculated for ClaH12: C, 94.70; H, 5.30. Found: C, 94.42; H, 5.37. DECACYCLENE, recrystallized three times from cumene. DIAMYLNAPHTHALENE, distilled from Sharples Chemical Co. Pentalene 195 (probably over 90% sec-amyl). Boiling point 183' C. a t 5 mm. of mercury, nz$ = 1.5523. DIBENZ [U,h]ANTHRACENE, Eastman Kodak Co. DIBENZOFURAN, recrystallized seven times from 90 to 10 methanol-benzene; 99.88 mole 70purity by calorimeter. DIBENZOTHIOPHENE,recrystallized three times from Skelly-Solve B; 99.5 mole % purity calorimeter.
1 J4
c / BREAK-SEAL-
.,-
Table I.
A
Ampoule for gas phase pyrolysis
9,10-DIHYDROANTHRACENE, reCryStallized three times from 75 to 25 ethyl alcohol-water. Chromatographed on silica gel, recrystallized three times from ethyl alcohol-water, and sublimed. Melting point 108.6' to 108.9' C. ' purity by calorim(corr.) ; 99.6 mole % eter. Contained 0.06% anthracene. 7,12 -DIMETHYL - BENZ [UIANTHRACENE, Eastman Kodak Co. 2,3-DIMETHYLNAPHTHALENE, picrate recrystallized from ethyl alcohol and
Comparison of Gas Phase and Liquid Phase Pyrolysis of Anthracene" Gas
Initial pressure, lb./sq. in.abs. Initial conc., moles/liter Products, wt. % ' charge Gas (through C4) c5 to ClO CIO' distillated/ Residue
Loss
3* 0.0031 0.2 0.1 100.4
0.0 -0.7
Phase Gas
Liquid
67b 0.072 0.03 0.1 99.9 0.2 -0.2
157O 1.7 5.9 1.9 15.4 77.6 -0.8
Gas, mole % Hydrogen Methane Ethylene Ethane Propylene Propane Butenes Butanes Weight of charge, mg. Reaction tube volume, ml. Run No. a
91 9
64 12 2 4
... ...
e
...
e
0
... ... ...
e
16 1 1 101.1 182 126
299.5 23.5 144
19 62 0 13 0
5 0 1 298.3 ca. 1.0 112
500° C., 90 minutes.
* Calculated, assuming feed t o behave as ideal gas.
Calculated from vapor pressure data, extrapolated. Including feed. Distilled at 300" t o 350' C., 1 micron of mercury. See experimental section. e Insufficient gas for sampling.
VOL. 50, NO. 2
FEBRUARY 1958
239
Table 11.
Unsubstituted Aromatics Not Possessing the Anthracene Configuration
0 Temperature,
c.
Time, min.
Biphenyl
Naphthalene=
Phenanthrene
Chrysene
Triphenylene
Pyrene
Fluoranthene
Decacyclene
CizHia
CioHs
Ci4Hio
CisHiz
CisHiz
Cl6HlO
C16H10
CoGHis
474 60
500 90
500 90
475. 60
500 90
475 60
500 90
475 60
475 60
475 30
Products, wt. % ' charge Gas [through
c41
cs to ClO
G o f distillatec Residue Total
Gas product Moles gas/ mole charge Composition, mole % Hydrogen Methane
co
0.0 0.0 99.7 0.0 99.7
...
-
0.0
0.0020
0.0015
67
7
0.87 267
-
0.0 101.1
-.
0.0029
0.0046
..-
... ...
. a .
...
...
264.6
0.02 0.06 101.2 0.0 101.3
0.0 101.1
95 5
27
Weight of charge, 155.4 mg. Reaction tube 0.64 vol., cc. Run No. 177
0.0
0.0
Traceb
...
0.04
Trace
...
0.6
0.0218 85 15
...
Trace Trace
0.0 0.6 99.9 __ 0.0 100.5
...
0.0
0.0024
0.0069
65 9 27
0.0 0.03 100.5 0.0 100.5
0.0011
..
91 9
...
.. ..
Trace Trace
...
0.0
0.0017 97 3
0.01
Trace I
.
.
99.3d 99.3
0.0013 86 14 * . I
87.1
82.7
116.2
77.1
116.3
180.7
243.9
190.2
2.16 125
0.47 181
0.71 269
0.53 180
0.73 268
0.61 183
0.78 254
0.87 255
a
Liquid phase not present during pyrolysis in this run. Critical temperature is 480' C.
A,
Distilled at 300' t o 350' C., 1 micron of mercury. See experimental section. Unconverted feed. See text.
* Solid, probably naphthalene.
chromatographed on alumina. Product recrystallized twice from ethyl alcohol and sublimed. Melting point 104.0' to 104.4' C. (corr.). 2,6-DIMETHYLNAPHTHALENE, recrystallized twice from ethyl alcohol; 99.8 mole % ' purity by calorimeter.
DIPHEXYLMETHANE, melting point 24.8" to 25.0' C.; 99.1 mole yo purity by calorimeter. FLUORANTHENE. Picrate recrystallized three times from ethyl alcohol, chromatographed, recrystallized again three times from ethyl alcohol, and sublimed; 99.92 mole % purity by calorimeter.
FLUORENE, chromatographed on alumina, recrystallized twice, and sublimed; 97.3 mole % ' purity by calorimeter.
HEXAETHYLBENZENE, recrystallized
240
twice from isopentane; 99.3 mole purity by calorimeter.
%
HEXAMETHYLBENZESE, recrystallized from ethyl alcohol and sublimed; 99.6 mole yo purity by calorimeter. INDOLE, recrystallized five times from n-heptane. -4nalysis calculated for CsH7N: C, 82.0; H, 5.97; N, 11.96. Found: C, 82.03; H, 5.75; N, 12.1. ISOQUINOLINE, twice distilled; 99.6
yo purity by calorimeter. 1-METHYLANTHRACENE, prepared from 1-methylanthraquinone. Product was distilled and chromatographed over alumina. Maleic anhydride value: 0.513 mole per 100 grams found, 0.521 mole per 100 grams theoretical. Analysis calculated for ClbH12: C, 33.71; H, 6.29. Found: C, 93.8; H, 6.7. 2-METHYLASTHRACESE, prepared from mole
INDUSTRIAL AND ENGINEERING CHEMISTRY
2-methylanthraquinone according to method of Martin ( 7 7). Recrystallized from benzene and from toluene and sublimed. %METHYLCHOLANTHRENE, Eastman Kodak Co. 1-METHYLNAPHTHALENE, American Petroleum Institute, Project 46, Sample 578; 99.92 mole % purity by calorimeter. &METHYLNAPHTHALENE,National Bureau of Standards special sample; 99.7 mole % purity by calorimeter. NAPHTHACENE,recrystallized three times from xylene and sublimed. Sealed in vacuum. Analysis calculated for CISH l z : C, 94.70; H, 5.30. Found: C, 94.62; H, 5.36. NAPHTHALENE, American Petroleum Institute, Project 46, Sample 577; 99.76 mole yo purity by calorimeter. PHENANTHRENE. Picrate recrystallized
P Y R O L Y S I S OF A R O M A T I C S from absolute ethyl alcohol, chromatographed on silica gel, product recrystallized twice from ethyl alcohol, sublimed. Melting point 98.8' to 99.4' C. (corr.); 99.96 mole yo purity by calorimeter. 2-PHENYLQUINOLINE. Sample obtained from the late F. W. Bergstrom of Stanford University. Recrystallized twice from petroleum ether. PYRENE. Picrate recrystallized from allyl alcohol, chromatographed first on silica gel and then on alumina. Product was recrystallized twice from allyl alcohol and sublimed. Melting point 150.8' to 151.3' C.(corr.); 99.9 mole yo purity by calorimeter. QUINALDINE, twice distilled; 98.9 mole % ' purity by calorimeter. QUINOLINE, distilled. Boiling point 128.7' C.; 99.4 mole % purity by calorimeter. An unusual observation with quinoline was noted. As in the usual procedure for preparing samples for pyrolysis, quinoline was placed in a capillary ampoule, degassed a t liquid nitrogen temperature, and sealed off. O n warming the ampoule to room temperature, the drawn-out tip cracked. This procedure was repeated twice with the same result, although tube failures in loading isoquinoline or liquid hydrocarbons never occurred. I t is concluded that quinoline expands on freezing or that the sample employed had absorbed water from the atmosphere and ice formation in interstices gave rise to pressure. Although the quinoline was kept in a closed vessel after distillation, this compound is known to be very hygroscopic. There appear to be no data in the literature on the volume change of quinoline on melting. RETENE. Recrystallized from ethyl alcohol. Picrate was recrystallized twice, chromatographed. Product was recrystallized from ethyl alcohol and sublimed 1,2,4,5-TETRAISOPROPYLBENZENE, recrystallized twice from acetone. Melting point 117.6' C.; 99.7 mole % purijy by calorimeter. THIANAPHTHENE, melting point 31' C.; 99.86 mole % purity by calorimeter. TRIPHENYLENE. Picrate recrystallized four times from benzene, chromatographed on silica gel. Product was recrystallized three times from benzene. Melting point 198.2' to 198.6' C. (corr.).
phase test was made in which the initial pressure was 157 pounds per square inch absolute. The results of these experiments, shown in Table I, demonstrate that, a t either low or high pressure, in the vapor phase anthracene was virtually unaffected by exposure to a temperature of 500' C. for 90 minutes. On the other hand, pyrolysis in the liquid phase resulted in over 77% conversion to heavy residue and 7.8% conversion to products in the gas and liquid range. The gas from the vapor phase pyrolysis was principally hydrogen, while that from the liquid phase experiment was mainly methane, showing disruption of the ring system in the latter case. The abrapt onset of extensive condensation and cracking encountered in the liquid phase test is probably related to the approximately 50-fold increase in molecular concentration (in run 144 us. run 112) resulting from the liquefaction of anthracene. These observations indicate that the condensation reaction is higher than first order. Jaquiss and Szwarc ( 5 ) have discussed the reaction of phenyl radicals with toluene in solution a t 80' C. and in the gas phase at 600' C. at low pressure. They suggest that in solution phenyl radicals add to the aromatic nucleus, while in the gas phase the radicals remove a hydrogen atom from the methyl group. In the present experiments, reactions in the liquid phase in which radicals add to anthracene may be of importance in the formation of heavier products. As the object of this investigation was to study condensation reactions, tests were made on other compounds in the liquid phase.
I
Comparison of liquid and Gas Phase Pyrolysis
As a preliminary to the investigation described, the effect of concentration on condensation reactions was studied. Tests were made with anthracene a t 500' C. for 90 minutes. Two gas phase experiments were made in which the initial pressures were 3 pounds per square inch absolute, and 67 pounds per square inch absolute, and a liquid
Unsubstituted Aromatics Compounds lacking the anthracene configuration in their molecular structure proved very resistant to pyrolysis. Eight unsubstituted aromatics in this category were tested : biphenyl, naphthalene, phenanthrene, chrysene, triphenylene, pyrene, fluoranthene, and decacyclene. In four cases tests were made under severe conditions, 90 minutes at 500' C.; the remaining compounds were pyrolyzed at 475' C. The severity of these conditions may be judged by the fact that about 50% conversion of n-hexadecane is calculated to occur in 4 minutes at 475' C. and in 1.5 minutes at 500' C. Results are shown in Table 11. Except for chrysene, no heavy residue was detected with any of the compounds. In the runs in which enough gas was obtained for analysis, hydrogen was the predominant gas product, showing that
the main reaction is intermolecular condensation by elimination of hydrogen. In two cases (Table 11) carbon monoxide was indicated by the mass spectrometric analysis. However the amount found was very small (less than 0.001 mole per mole of charge) and probably indicates traces of oxygen or oxygenated compounds in the charge. From the values of the moles of gas produced per mole of feed (Table 11) chrysene was the most reactive of the eight compounds. The relative reactivities of these compounds are discussed below. Phenanthrene, unlike its isomer anthracene, was not measurably reactive in 90-minute tests a t 450' and 500' C. The colorless starting material took on a faint yellowish green tinge at the higher temperature; spectroscopic examination showed some general absorption from 4000 to 2100 A. not due to phenanthrene. From the amount of gas produced, no more than 0.6% of the phenanthrene could have been converted. The result with phenanthrene given here is to be compared with that of Tilicheev (27), who reported 55% conversion of phenanthrene in 179 minutes a t 500' C. The much greater reactivity found by Tilicheev probably arose from impurities present or from heterogeneous catalysis in the autoclave. The large amount of "heavy residue" found in the run with decacyclene was incompatible with the very small gas production from this compound. A test showed that decacyclene distilled extremely slowly under the conditions employed for distillation of heavy products (300' to 350' C. at 1 micron of mercury). Hence, the heavy residue shown for decacyclene in the table was essentially unconverted feed. The extraordinary resistance to pyrolysis of the above class of compounds suggested their use as solvents for studying reactions in solution a t high temperature. In other work not reported here it was found that when other compounds were pyrolyzed in naphthalene or phenanthrene solution, the latter compounds behaved as inert diluents and tended to decrease the extent of condensation reactions.
Unsubstituted Aromatics Possessing Anthracene Configuration Five compounds possessing the anthracene configuration as part of their molecular structure were tested (Table 111). These compounds were much more reactive than those not possessing this configuration (Table 11). In the condensation of aromatic molecules, one molecule of gas is eliminated for every bond formed between aromatic nuclei. Thus the condensation reVOL. 50, NO. 2
*
FEBRUARY 1958
241
Table 111.
Unsubstituted Aromatics Possessing the Anthracene Configuration
Temperature, C. Time, min.
450 90
Products, wt. % charge Gas [through C4] cs to ClO CIO+distillate* Residue Total
0.03 2.4a 99.5 0.5 102.4
0 * 009
a
299.8
ca. 1.0 106
0.12
Trace
0.014 89 11
... ...
...
10.3
0.044
0.022
... ... ... ...
... ...
... ...
... ... ... ...
205.5 0.68 261
282.9 0.90 264
279.2 1.31 145
...
...
0.015 99 1
... ... ... ... ...
. . I
.*. ... ... ... ... . . I
...
e . .
. . e
110.3 0.81 259
158.2 0.83 262
CIBHIZ
3.3 2.3 31.1 64.9 101.6
-
6.2
86 14
... ...
Naphthacene 475 30
0.01 0.0
98.8
89 11
.**
. . a
0.18 0.5 99.5
e . .
477 60
Trace
-
0.037
...
476 60
0.09 0.1 93.6 6.1 99.9
73 27
. . e
... ...
co con
500 20
0.03 1.4
86 8 1 3 0 1 0 1
476 60
Trace
...
-
Gas product Moles gas/mole charge Composition, mole % Hydrogen Methane Ethylene Ethane Propylene Propane Butenes Butanes
Weight of charge, mg. Reaction tube, vol., cc. Run No.
476 30
Dibenz[a,h]anthracene C22H14
Benz[a]anthracene CisHiz
Anthracene CI&D
0.414 35 42 0 11 0.5
7 0 3 1 0.5 128.5 0.75 215
Benzo[a]pyrene C2oH12 476 50
0.04
Trace
...
2.5
0.010 66 34
...
... ... ... *.. ... ... ..#
55.5 0.51 272
Believed spurious. Technique improved later. C., 1micron. See experimental section.
* Distilled at 300Oto 350'
Table IV.
Reactivities of Unsubstituted Aromatics
Pyrolysis, Moles Gas Formed per Mole Hydrocarbon Charge, 475O C., 60 Min. Naphthacene Anthracene Benz[a ]anthracene Dibenz [a,h]anthracene Benzo [alpyrene Chrysene Triphenylene Fluoranthene Biphenyl Pyrene Decacyclene Phenanthrene Naphthalene
Relative Reactivity Toward CCls. at 91' CSn
Relative Reactivity Toward CH3.
k'r
at 85" C.b
SingletTriplet Excitation Energy,c Xcal./Mole
9520 820 468
42.0 47.2
...
52.3
0.414d 0.037 0.022
26 11 30
0.015 0.010 0.0046 0.0024 0.0017 0.0015 0.0011 0.0013d 0. 002Q6 0.0020
1.85 70 0.033 0.007
... 0.0004 0.3
...
0.004 0.01
...
57
... ...5
...
...
56.6 68.1
...
125
65.2 48.0
27 22
61.8 60.9
...
...
See p. 64 of (6) : k', is related to ratio of rates of addition of CC1v t o aromatic and t o an olefin. * See (8). Methyl reactivity to benzene is taken as unity. See (9) for all energies except that for biphenyl (18). 30 minutes. e 500' C., 90 minutes.
242
INDUSTRIAL AND ENGINEERING CHEMISTRY
action would lead to a minimum of 0.5 mole of gas per mole of reactant condensed. If condensation leads to more complex structures in which nuclei are connected by more than one bond, the moles of gas produced per mole of reactant converted will exceed 0.5. Within the limitations resulting from the uncertainty about the type of condensation reaction and in the absence of any cracking of the rings, the moles of gas produced per mole of reactant can be taken as a measure of conversion to condensed products. Appreciable amounts of methane were produced from the most reactive aromatics having the anthracene configuration (Table 111), an indication that in these cases a part of the gas production resulted from cracking of the rings. The moles of gas produced per mole of reactant are compared in Table IV with the reactivities of the corresponding hydrocarbons with trichloromethyl (6) and methyl ( 8 )radicals at moderate temperatures. The five most reactive compounds in pyrolysis were the most reactive toward CC13 radical, and the three
P Y R O L Y S I S OF A R O M A T I C S most reactive in pyrolysis were the most reactive toward CH3 radical. Szwarc (79) showed that reactivities of aromatics with methyl radicals correlated well with the energy of excitation of the aromatic from its singlet ground state to the first excited triplet state, in which the hydrocarbon exhibited diradical character and was reactive toward radicals. The available singlet-triplet excitation energies (9, 72) are listed in Table IV, and with the exception of pyrene, lower excitation energies are found in the hydrocarbons possessing the anthracene configuration. Hydrocarbons of the latter type are known to have dienelike properties. For example, anthracene and naphthacene form an addition product with one molecule of maleic anhydride (73). These reactions are believed to arise from easy activation to diradical structures-e.g., in anthracene, the structure in which there are unpaired electrons on the two central carbon atoms. Tilicheev tested anthracene, phenanthrene, biphenyl, and naphthalene and found the reactivity to decrease in that order (27). The condensation of these compounds in high temperature pyrolysis is not mere polymerization, as hydrogen is eliminated in the reaction. I t would seem that easy activation to a diradical would facilitate the formation of an activated complex between two molecules, which then might form a condensation product by elimination of a hydrogen molecule.
Alkyl Aromatics Polyalkylbenzenes.
Although ben-
zene is very refractory in the temperature range of these experiments, alkyl substitution greatly increases thermal reactivity. This follows from introduction into the molecule of bonds of lower dissociation energy than the C-H bond in benzene itself. Results obtained with hexamethylbenzene, hexaethylbenzene, and 1,2,4,5-tetraisopropylbenzene are shown in Table V. Reactivity increases with the size of the alkyl substituents in this series. Hexamethylbenzene was very slightly reactive a t 475' C., producing only a small amount of gas and no detectable residue. The gas product consisted of methane and hydrogen in 10 to 1 ratio. Substitution with ethyl instead of methyl groups leads to increased reactivity, and a significant extent of reaction was observed with hexaethylbenzene even a t 450' C. At 475' C. the gas production in 60 minutes was increased tenfold over that a t 450' C., and residue production increased from 0.3 to 1.3 weight yo of charge, while the composition of the gas was substantially unchanged, being mainly methane and ethane, at a ratio of 1.6 to 1. 1,2,4,5-Tetraisopropylbenzene proved more reactive than its isomer, hexaethylbenzene, with respect to both cracking and formation of residue. The main gas products were methane, ethane, and propane, in the ratio 4.7 to 1.6 to 1. The pattern of gas composition was unaffected by a considerable change in conversion, as in the case of hexaethylbenzene. The order of reactivity of these three alkylbenzenes is related to the relative
weakness of their aliphatic C-C bonds, although if chain reactions prevail the relationship is probably not a simple one. Szwarc gives the following bond dissociation energies (in kilocalories) in toluene and ethylbenzene (76).
In the xylenes the CoH&Hz-H bond is slightly weaker than in toluene. Szwarc found that the (chain) pyrolysis of ethylbenzene (77) was much faster than the pyrolysis of toluene, and it is not surprising that hexaethylbenzene, with six relatively weak (3 C-C bonds, decomposed faster than hexamethylbenzene. Again, 1,2,4,5- tetraisopropylbenzene has eight (3 C-C bonds and thus might be expected to crack faster than the isomeric hexaethylbenzene with six such bonds. In the present experiments hydrogen was in all cases a minor product, differing in this respect from Szwarc's results obtained a t low pressure and higher temperature. Szwarc found a ratio of hydrogen to methane of 1.5 to 1 from toluene and rn-xylene, 1.9 to 1 from pxylene, and 1 to 1 from o-xylene (78). Szwarc suggested that the lower hydrogen to methane ratio from o-xylene, compared with toluene and the other xylenes, was the result of a decreased aryl-CHa bond energy caused by steric __
Table V.
Hexamethylbenzene Temperature, C. Time, min. Products, wt. yocharge Gas [through C,] cs to ClO C d distillate' Residue Total Moles gas/mole charge Gas, mole % Hydrogen Methane Ethylene Ethane Propylene Propane Butenes Butanes
co
coz
476 60 0.24 0.. 0.
...
Trace 0.028 7.9 80.0
... 2.0
... ... ... ... 9.0
...
Polyalkyl Benzenes and Methyl- and Dimethylnaphthalenes 2,3-Di-
Hexaethylbenzene 45 1 60
0.6 0.3 99.4 0.3 100.6 0.082 5.1 53.5 1.4 34.1 0.8 1.2 0.6 0.6 1.2 0.8 152.3 0.68 224
476 60 6.4
Trace
... 1.3
1,2,4,5-Tetraisopropylbenzene 450 60 5.0 1.3
... '
Trace
476 30 11.9 0.6
... 4.1
1-Methylnaphthalene 475 30 0.2 0.2 99.8
2.3 57.8 1.0 37.4 0.0 1.0 0.0 0.5 0.0 0.1 190.4 0.87 263
0.54 3.9 55.8. 3.5 19.0
1.7 11.8 1.4 2.7 0.3 0.0 268.3 0.74 2 74
1.21 2.3 57.7 2.5 18.0 1.5 11.4 1.2 3.9 1.3 0.1
55.9 263.7 Weight of charge, mg. 0.27 1.09 Reaction tube vol., cc. 258 284 Run No. Distilled at 300Oto 350° C., 1micron of mercury. See Experimental section.
0.63 0.15 98.9 0.1 99.8
0.0 -
-
0.018
0.059
100.2
0.72
475 60
14 85 1
... ... ... ... ... ... ...
265.0 0.73 175
7 93
... ... ... ... ... 9 . .
... ..*
271.9
0.75 174
2-Methylnaphthalene 475 60 0.17 0.05 99.7 0.0 99.9
-
methylnaphthalene
475 60
476 60
0.54 0.2 99.6 0.0 100.3
0.3 0.1 99.7 0.1 100.2
... ...
181.4 0.58 176
29.9 0.16 179
VOL. 50, NO. 2
,
27
21 79
... ... ... ... ... ... ... ...
... ... ... ... ... ...
0.036
0.067
0.019 39 61
2,6-Di-
methylnaphthalene
70 3
... ... ... ... ...
... ...
297.6 1.06 191
FEBRUARY 1958
243
interaction of the two adjoining methyl groups. The hydrogen atom lost by the methyl group in the slow first step would then have an increased tendency to remove or displace a methyl group from the ring. In the case of hexamethylbenzene there is a maximum of steric i n t e r h i o n between methyl groups, and the aryl-CHs bond in this compound may actually be weaker than the aryl CH2-H. The small amounts of hydrogen from hexamethylbenzene may be an indication that the slow first step is the dissociation of a methyl radical from the hexamethylbenzene molecule. The high ethane content in the gas product from 1,2,4,5 - tetraisopropylbenzene suggests that a methyl radical, formed initially by cleavage of the weak p C-C bond, removes a methyl from another isopropyl group to form ethane. This reaction may be favored over methyl attack on the tertiary hydrogen because of steric interference with the latter process. The formation of a suhstantial amount of propane indicates cleavage of the aryl-isopropyl bond. This bond is probably weakened by the branched structure of the alkyl group and by steric interaction of the neighboring bulky isopropyl groups. The rates of thermal decomposition of the three alkylbenzenes above are in the reverse order of their reactivities with trichloromethyl radical a t 91' C. (7), where it was found that the rate of radical attack was decreased by steric hindrance in hexaethylbenzene and still more in 1,2,4,5-tetraisopropylbenzene. Methyl- and Dimethylnaphthalenes. Table V also gives the results of pyrolysis tests with the two methylnaphthalenes and with two of the 10 dimethylnaphthalenes. Unsubstituted naphthalene, like benzene, is a very refractory compound, but the presence of one or more methyl substituents increases the condensation rate. A point of interest in Table V is the effect of isomerism upon reaction rate. Methyl substitution on a-carbon atoms leads to the greater reactivity. Thus the condensation rate of l-methylnaphthalene was several times greater than that of 2-methylnaphthalene, an effect which may also be noted in Tilicheev's data (21). Greater quantum mechanical stability would be expected for the 1-naphthylmethyl radical as compared with the 2-isomer, leading to a lower alkyl C-H bond dissociation energy and greater reactivity in l-methylnaphthalene. 2,3-Dimethylnaphthalene and 2,6-dimethylnaphthalene, in which the two methyl groups are both on naphthalene 8-carbon atoms, were roughly three times and twice as reactive, respectively, as 2-methylnaphthalene. The higher boiling fractions from py-
244
rolysis of the methyl- and dimethylnaphthalenes were analyzed in an exploratory manner in a high molecular weight mass spectrometer at the Houston Research Laboratory of Shell Oil Co. ( 7 4 ) . The results are given in Tables V I and VII. Because compounds expected as products were not available for calibration, sensitivities had to he assumed, and the conversions calculated from the mass spectral results were subject to considerable uncertainty. This is shown by the fact that the ratio of the measured gas
compounds listed in Tables V I and VII, corresponding to the masses actually found, are intended merely to serve as examples, as the isomer or isomers present were not identified in the mass spectrometric analysis as carried out on these products. If breaking of all of the bonds except the C-C bonds in the naphthalene ring can occur, the product masses expected from radical coupling reactions are then 128, 156, 254, 268, and 282, as indicated below.
Molecular Weights Resulting from Radical Combination Reactions in Pyrolysis of Methylnaphthalenes Reaction Product -+ Naphthalene Naphthyl * i- H 128 127 1 -+ Dimethylnaphthalene Methylnaphthyl CHa * 141 15 156 2-Napthyl4 Binaphthyl 254 127 Methylnaphthyl * or --c Methylbinaphthyl or dinaphthylmethane 268 naphthylmethyl naphthyl * 141 127 2-Methylnaphthyl - or 2-naphthylmethyl 4 Dimethylbinaphthyl or dinaphthylethane 282 141 6
-+ 9
+
production to the conversion of starting material indicated by mass spectrometry ranged from 0.1 to 0.4 mole per mole of feed converted, although at least 0.5 mole of gas (hydrogen or methane) should have been formed, from stoichiometry. The mass spectrometric analyses are largely of interest in showing the masses present in the product. The possible
Table VI.
All of these masses were actually found in the product except 254, that of binaphthyl. Instead of 254, mass 252, which corresponds to benzofluoranthenes, benzopyrenes, bisnaphth ylenes, or perylene, was found in all three experiments in Table V. Masses 408 and 422, which were found, would be expected from secondary reactions. Masses
Pyrolysis of Methylnaphthalenes". Mass Spectrometric Analysis of Products above C ~ O Masses
Mass 128 142 156 178 192 226 252 266 268
Formula 0.5hr. 1 hr. 0.2 7.5 CioHs CiiHlo 91.9 81.4 0.6 0.2 CizHi2 0.06 CiaHio ... 0.03 ClbHlZ 0.1 CiaHio 1.3 0.8 CzoHlz 0 . 4 c21H14 1.1 2.2 C2iHie
... ...
C12H16 CzzHis
280
282 338 348 358 406 408
CZ&6
414 422
CaaHis CaaHza
Ca&4
Run No. a
INDUSTRIAL AND ENGINEERING CHEMISTRY
475O
c.
1 hr.
0.9 97.1 0.1 0.01 0.01
0.02 0.23
...
...
...
...
... ... ... 0.1
0.10 0.18 0.18 0.08 0.15
... ... ... ...
0.1
...
0.4 4.7 1 . .
0.1
175
6.0
0.16
174
0.40 1*2
0.01
. . a
0.02
176
Possible Compounds Corresponding t o Formula Naphthalene Feed Dimethylnaphthalene Anthracene Methylanthracene
Expected from Simple Radical Reactions as Primary or Secondary Products Primary Primary
Benzofluoranthene Dibenzofluorene Methylbinaphthyls or dinaphthylmethane Dinaphthylethylene Dimethylbinaphthyls or dinaphthylethane
Secondary Secondary Primary
Di- [naphthylmethyllnaphthalene
Secondary
-
Secondary
Di- [naphthylmethyl) methylnaphthalene
Secondary Primary
P Y R O L Y S I S OF A R O M A T I C S not explainable by simple radical reactions constituted only a minor fraction of the products in each case. Mass 282 was a major product in all three tests and corresponds to dimethylbinaphthyls or dinaphthylethanes. The mass spectrometric information on higher boiling products from 2,3- and 2,6-dimethylnaphthalenes is shown in Table VII. Again, all but a minor fraction of the products are of masses which would be expected from simple radical reactions. Mass 170 (methylethylnaphthalene or trimethylnaphthalene) might be expected, but it was not found. The presence of masses 278, 292, 306, and 320 in the products from the dimethylnaphthalenes may indicate the occurrence of condensation reactions of the type
Table VII.
Mass 128 142 156 264 266 268
Pyrolysis of 2,3- and 2,6-DimethylnaphthalenesS. Mass Spectrometric Analysis of Products above C ~ O
Products > CIO,% (Vol.) 2,3-Dimethyl- 2,6-DimethylFormula naphthalene naphthalene
CioHs CiiHto CizHiz CziHiz CziHi4
. CziHia
27b 280 282
CZZHl4
292 294
CzaHia
1.00 3.50 90.3 a * .
0.07 0.14
0.40 1.80 94.2 0.20 0.02
...
0.57 0.43 0.14
...
CZ8Hl8
0.50 0.29
0.07 0.21
296
CzsHzo
0.22
0.90
306 308
c24H18 c24H20
0.21 0.29
0.12 0.21
310
C24HZZ
1.49
1.37
320 322
CZ6HZO
0.35 0.50
... ...
CzzHie CZZHU
CzbHzz
0.27 0.23
Masses Expected from Simple Radical Reactions as Primary or Secondary Products
Possible Compounds Corresponding to Formula Naphthalene Methylnaphthalene Feed
Secondary Primary
Dibenzofluorene Dinaphthylmethaneor methylbinaphthyls Pentacene Dinaphthylethylene Dinaphthylethane or dimethylbinaphthyls Methylpentacene Methyldinaphthylethylene Methyldinaphthylethane Dimethglpentacene Dimethyldinaphthylethylene Dimethyldinaphthylethane Trimethylpentacene Trimethyldinaphthylethvlene -
Secondary Secondary Secondary Primary Secondary Primary Secondary Primary Secondary
< -
Run No. Q
in which pentacenes are produced. A further study of the condensation process was made using chromatography and ultraviolet spectroscopy as analytical tools. The pyrolysis of l-methylnaphthalene at low conversion was chosen for this study, with the expectation that the single methyl substituent would give rise to fairly simple products. A 1.5gram sample of 1-methylnaphthalene was pyrolyzed in liquid phase for 1 hour at 475' C., the same conditions as in run 174, Table V. After pyrolysis, gas products were removed,. and the heavier products including unreacted l-methylnaphthalene were chromatographed on a 1.3 x 120 cm. column of Alorco F-20 alumina. The column was developed with n-pentane, to which increasing amounts of benzene were added until the development rate with pure benzene finally became too slow. A total of 135 fractions was obtained in this way. About a quarter of these fractions were evaporated to dryness and redissolved in pure 2,2,4-trimethylpentane (isooctane). The alumina was physically divided into 35 sections; each section was eluted with dioxane, evaporated to dryness, and redissolved in 10 ml. of dioxane. Ultraviolet spectra of all fractions were obtained on a Cary Model 11 recording spectrophotometer, and quantities of the identified compounds were calculated. Naphthalene was found in the product, and although it was not resolved chromatographically from 1 methyl-
-
179
191
475' C.,1 hour.
naphthalene, it was possible to determine both compounds quantitatively. The three types of binaphthyls were completely separated. There were indications that the binaphthyls were alkylated, but the number and position of the substituents (presumably methyl) could not be determined. The presence or absence of the unsubstituted binaphthyls eould not be established. In the mass spectrometric study reported above,
unsubstituted binaphthyls were not detected. A number of other well-resolved spectra were found in the 135 fractions, but the proper reference spectra for identification of these compounds were lacking. There was evidence of the possible presence of a dinaphthylethane, and Szwarc and Shaw reported the isolation of 1,2-di-l-naphthyl-ethane,lj2-di-2-naphthyl-ethane, and 1,2-di-6-methyl-2-
DO
W A V E L E N G T H , ANGSTROMS
Pyrolysis of 1 -methylnaphthalene. Spectrum of material desorbed from section of chromatographiccolumn VOL. 50, NO. 2
FEBRUARY 1958
245
naphthyl-ethane, from low pressure pyrolysis at 800' C. of l-methylnaphthalene, 2-methylnaphthalene, and 2,6dimethylnaphthalene, respectively (20). Spectra of most of the fractions obtained by eluting the column sections showed little structure. However, a series of six fractions had a number of well-resolved bands, and the relative absorbance of the main peaks remained constant over the series. The spectrum of one of these fractions (shown) is probably that of a single compound. A characteristic feature of this spectrum is the small range of the absorptivities of the peaks. In this respect it is quite unlike the vast majority of published spectra of polynuclear aromatics (3). The spectrum resembles that of dibenzo[a,l]pyrene, although there are several points of difference (2). However, dibenzo [a,l]pyrene (1) could
I not have been formed from l-methylnaphthalene without opening and rearrangement of rings. A more attractive possibility on chemical grounds would be the reaction of three molecules of 1methylnaphthalene to form a dinaphtho[a,Z]pyrene (11).
v\
Q I1
246
Table VIII. Pyrolysis of 1 -Methylnaphthalene: Ultraviolet Spectral Study Temperature, O C. Time, min. Products, wt. % charge Gas (through Cr) CS to ClO Naphthalene 1-Methylnaphthalene Alkyl 1 , l '-binaphthyls Alkyl 1,2'-binaphthyls Alkyl 2,2'-binaphthyls Unidentified Residue Total Gas, mole % Hydrogen Methane Weight of charge, grams Reaction tube vol., cc. Run No.
475 60
0.55 0.0 4.4 78.1 0.4
2.0 2.2 11.7
0.6 100.0 8
92 1.562 5.04 251
Spectra of dinaphtho[a,l]pyrenes have not been reported. However, the spectrum of I1 may resemble that of I. because cata-condensation of two benzo residues on I to form the skeleton of I1 would be expected to shift the spectrum a comparatively small amount toward longer wave lengths (7). With regard to the product of mass 252 found in the mass spectrometric study, perylene was definitely absent from the ultraviolet spectra. Spectra of bisnaphthylenes were not available. As for benzopyrenes and benzofluoranthenes, it was impossible to come to a definite conclusion, the reference spectra having been obtained in alcoholic solution. The results of the study are summarized in Table VIII. Although there was only 22% conversion of l-methylnaphthalene, over half of the converted material remained unidentified despite what appeared to be excellent resolution by the chromatographic technique. I n the present experiments, reactions in which naphthyl-H bonds were broken were prominent. Formation of alkyl binaphthyls may be evidence for the importance of liquid phase reactions in which radicals add to the aromatic nucleus. The resulting radicals could dimerize and subsequently dehydrogenate. Amyl- and Diamylnaphthalenes. The results for these compounds are shown in Table IX. The amylnaphthalene is believed to have contained over 90% 2-sec-amylnaphthalene. The diamylnaphthalene was presumably largely di-sec-amyl, but the ring positions of the predominant isomers are unknown. Amylnaphthalene cracked and condensed faster than the methylnaphthalenes or dimethylnaphthalenes studied. Diamylnaphthalene cracked and condensed faster than amylnaphthalene. Increasing the size and number of alkyl substituents in this fashion thus increases the
INDUSTRIAL AND ENGINEERING CHEMISTRY
reactivity of the alkylnaphthalenes. as in the case of the alkylbenzenes. Decreasing temperature decreased the production of heavy residue more than that of cracked products. Thus at the lowest temperature of the present tests, 425' C., cracking was accompanied by only a trace of heavy residue production, although condensed products of intermediate molecular weight might have been formed. The production of 6% heavy residue by dimethylnaphthalene in 0.5 hour at 450' C. places this compound among the most readily condensable compounds studied in this investigation. The gas analyses in Table IX furnish some clues to the mechanism of cracking of both the amyl- and diamylnaphthalenes. The disubstituted naphthalene gave more gas than the amylnaphthalene, but gas product distributions were almost identical with either starting material a t a given temperature. Ethane was the principal gaseous product, amounting to about 50 mole % of the gas in all the experiments. Propylene was produced to a large extent (13 mole 70) only at the lowest temperature, 425' C., while propane production assumed more importance as the temperature increased. The other principal product, methane, was about 17 mole yo at the lowest temperature and increased to about 34 mole 70 at the highest temperature. These results are compatible with a radical chain reaction in which a major factor is the probable weakness of the two aliphatic C-C bonds, 1-2 and 2-3, as indicated in the formula below: 1
2
3
4
5
H H H H H HC-C-C-C-CH H I H H H Naphthyl
In the alkylbenzene case discussed earlier. bonds of this type have a dissociation energy of about 63 kcal. per mole. Although the corresponding quantity for alkylnaphthalenes is not known, it is presumably not greatly different. Chain initiation may occur, then, by the breaking of one of the two weak C-C bonds, 2-3 or 2-1, giving rise to a propyl or a methyl radical, respectively. Either of these radicals is capable of propagating the chain reaction by attack on the amylnaphthalene molecule. Several modes of attack are possible, involving varying degrees of ease of removal of a hydrogen atom, and of ease of the subsequent @-fission of the naphthylamyl radical. A prominent path of reaction is the removal of the tertiary hydrogen atom attached to carbon atom 2, which by analogy with the alkyl benzene case is assumed to be the most weakly attached hydrogen atom
P Y R O L Y S I S OF A R O M A T I C S in the molecule. The ensuing 6-fission of the naphthylamyl radical, between carbon atoms 3 and 4, should be comparatively easy, as it results in the formation of 2-naphthylpropene, which should possess extra stability by conjugation of the double bond with the aromatic ring. The ethyl radical formed in the p-fission would continue the chain reaction by attacking another amylnaphthalene molecule. Such a reaction scheme would lead to the production of ethane as the principal gaseous product, followed in importance by methane and propane. The 2-naphthylpropene, by analogy with styrene, would be expected to polymerize and/or condense to heavy residue. These reactions would tend to be more prominent a t high conversion, where higher concentrations of 2-naphthylpropene would prevail. The data in Table IX bear o u t this interpretation, as in the experiments a t 425' C. cracking occurred almost to the exclusion of condensation, while a t higher temperature and higher conversion both reactions were comparable in extent. Propylene may arise from: attack of a radical upon a hydrogen atom on the 4-carbon atom of the alkyl chain of amylnaphthalene followed by 0-fission of the resulting naphthylamyl radical to give propylene and a naphthylethyl
Table X.
1-Methylanthracene Temperature, C. Time, min. Products, wt. % charge Gas [through Ca] O
CS to c11
CIO+distillatea Residue Total Moles gas/mole charge Gas, mole % Hydrogen Methane Ethylene Ethane Propylene Propane Butenes Butanes
co
COe
475 30
3.1 0.2 68.4 (28.3)b
...
0.42 12 88
... ... ... ... ...
... ...
I
Table IX. Temperature, O C. Time, min. Products, wt. %, charge Gas [through C,] cs to ClO CUI+distillate' Residue Total Gas, mole % Hydrogen Methane Ethylene Ethane Propylene Propane Butenes Butanes Pentenes Pentanes
Amyl- and Diamylnaphthalenes Amylnaphthalene Diamylnaphthalene
425 30
450 30
2.8 0.3 97.0
6.4 ... 92.4
Trace
0.4 99.2
100.1
12.7 1.2 78.7 7.5 100.1
4.0 0.7 95.2 0.08 100.0
11.3 1.8
2.1 34.5 0.1 43.0 0.9 12.8 0.9 3.2 0.3 1.7 0.4 0.2
2.6 16.7 0.8 48.2 13.9 9.0 2.5 3.1 0.7 2.5 0.0 0.2
1.6 24.6 0.4 49.6 3.0 11.1 0.0 4.1 0.4 3.3
0.7 11.2 1.5 3.7 0.4 1.9 0.3 0.2
0.7 2.6 0.5 0.2
coz
425 30
1.9 24.6 0.0 53.7
2.1 17.3 1.8 50.9 13.1 6.4 2.1 2.2
co
476 30
451 30
... 5.6 ...
1.7 0.2
Weight of charge, mg. 351.0 258.1 358.2 374.5 0.92 Reaction tube vol., cc. 1.02 0.97 0.95 Run No. 245 240 249 250 Distilled at 300° to 350° C., 1 micron of mercury. See experimental section.
radical, or cleavage of the 2-amyl group from the naphthalene ring to give a 2-amyl radical which could undergo p-fission to give propylene and an ethyl radical. Apparently a t a higher tem-
197.9 0.65 233
perature, radicals tended to react with propylene and remove it from the gas fraction. Very few of the propyl radicals cracked, as cracking of such a radical would give rise to the production of a
Alkyl Tricyclic Aromatics and Tetracyclic Aromatics
2-Methylanthracene 475 30
I
1.0 0.2 84.8 14.3 100.3 0.17
36 64
... ... ... ... ... ... ... ...
7,12-Dimethylbens[a]anthracene
Retene
60 3.7 0.0 57.0 39.8 100.5 0.53
7
25 66 0 4 0 3
0
....2
451 60
475 60
0.6 0.3 99.2 0.3 100.4 0.07
6.3 0.4 79.6 12.4 98.7 0.73
13 61 1 4 1
14 0 1 4 1 222.8 0.75 221
Weight of charge, mg. 288.3 120.5 108.1 Reaction tube vol., CC. 0.83 0.69 0.50 Run No. 194 199 178 a Distilled at 300' to 350" C., 1 micron of mercury. See experimental section. * By difference. Part of residuelost.
476 30
I
6.5
2.2
Trace
~
4 73 0 13 0 9 0 0 1 0 168.1 0.59 173
4 91 0
5 0
0 0 0 0 0
136.0 0.76 260
VOL.
476 30
Trace 51.6 41.8 99.9 1.03
-
3-Methylcholanthrene
50, NO. 2
39.8 57.6 99.6 0.69
11 73 0 13 0 3 0 0 0 0
144.7 0.91 257
FEBRUARY 1958
247
Table XI.
Diphenylmethane Temperature, O C. Time, min. Products, wt. yocharge Gas [through Cr]
cs to c11
Cm' distillate' Residue Total Moles gas/mole charge Gas, mole % Hydrogen Methane Ethylene Ethane Propylene Propane Butenes Butanes
co
coz
476 60
Trace 1.3 98.1 1.5 100.9 0.016 92 2
*..
... ... ... ...
... 6 ...
Fluorene 475 60 2.0 1.4 96.5 0.2 100.1 0.133
14 32 0 25 0 16 3 9 1 0 209.5 0.41 203
Methylene-Bridged Aromatics
11-H-benzo[blfluorene 476 30
methyl radical and ethylene, and essentially no ethylene was found in the product gas. Apparently, hydrogen abstraction reactions by propyl radicals are facilitated by the high concentrations in the liquid phase and by the relatively low temperature, which increases the life of the radicals. Alkyl Tricyclics and Tetracyclics. The results with a few compounds of these types (Table X) reinforce the findings presented above for alkylbenzenes and alkylnaphthalenes. I n a 0.5-hour test at 475' C., 1- and 2-methylanthracene condensed faster than the parent compound, which was only slightly reactive under these conditions (Table 111). Here again, as with 1- and 2-methylnaphthalenes, the I-methyl isomer was more reactive than the 2-methyl isomer. Phenanthrene, the unsubstituted parent of retene, was completely unreactive in a prolonged test at 500' C, (Table 11). O n the other hand, retene (1-methyl-7-isopropylphenanthrene)was very reactive even at 475' C., showing once more the effect of alkyl substituents on condensation. The presence of propane in the gas product shows that isopropyl radicals are dissociated from the phenanthrene nucleus. Two alkyl benz [alanthracenes were tested, 7,12-dimethylbenz [alanthracene and 3-methylcholanthrene. Both compounds underwent rapid condensation, forming methane as the main gas product. The parent compound, benz[a]-
248
Bibenzyl
*..
-
...
0.018 30 70
... ... ... ... ...
...
...
... 2 ...
... ...
* I .
...
197.1 0.95 247
anthracene, was far less reactive (Table 111). Methylene-Bridged Aromatics The substitution of a phenyl group for a hydrogen on methane lowers the bond dissociation energy of the C-H bond on the aliphatic carbon atom from 101 to 77.5 kcal. (76). Although data are not available for ( C ~ H S ) ~ C H the~ , substitution of an additional phenyl apparently has little effect, judging from the fact that the ( C ~ H SC-H )~ bond dissociation energy is reported to be about 75 kcal. (76). If a dimethylene group is interposed between two aryl groups, the aryl CH2-CH2 aryl bond acquires a very low dissociation energy, which is known in the case of bibenzyl to be 48 kcal. (4). Thus a compound with one or two methylene groups between two aryls possesses a number of relatively weak bonds which would give rise to increased reactivity. Table XI shows the results obtained with six methylene-bridged aromatics. Diphenylmethane and fluorene, pyrolyzed at 475' C. for 1 hour, formed less heavy residue than did the other four compounds in 0.5 hour at 475' C. The higher reactivity of 11-H-benzo [ b ] fluorene as compared with fluorene appears to arise from the greater activating effect of the naphthylene group relative to the phenylene group (76, 20). Bibenzyl and 9,1O-dihydroanthracene, each having two methylene groups, produced about the same amount of res-
INDUSTRIAL AND ENGINEERING CHEMISTRY
476 30 1.0
Trace
...
-
3.6
90 8
Acenaphthene
0.9 0.2 95.6 3.4 100.1
0.1 52.2
0.3 94.8 4.7 99.8 0.035
CHz 9,lO-Dihydroanthracene 476 30
475 30
Trace
Weight of charge, mg. 204.7 308.7 0.92 Reaction tube vol., cc. 0.60 206 Run No. 204 See experimental section. a Distilled at 300' to 350° C., 1 micron of mercury.
CHs---CHz
15.3 . I .
0.128
0.129
46 34 0 16 1 2
58 21 0 11 8 0 1 0 0 546.1 1.14 207
... ... ...
1 34.7 1.20 248
idue. The lower boiling product from bibenzyl was largely toluene, thus indicating that scission of the weak CH2-CH2 bond is an important reaction. Acenaphthene produced the largest amount of residue of the six compounds in this category. Decacyclene, a possible condensation product from acenaphthene, was shown previously (Table 11) to be extremely involatile. Tilicheev tested bibenzyl, fluorene, and diphenylmethane and found their reactivity to decrease in that order (27). Detailed information about the condensation products was obtained in the case of the simplest compound, diphenylmethane. The higher boiling fractions were analyzed in the mass spectrometer with the results shown in Table XII. Masses 166 and 334 were the main products and were probably fluorene,,
and tetraphenylethane, (C6Hs)*CHC H ( C ~ H S ) ~About . half the products in Table X I 1 can be accounted for by coupling of radicals formed by dissociation of one or more of the bonds indicated in the formula below: H
I I
H-CsH p C - C s H p H H Many of the remaining masses found
P Y R O L Y S I S OF A R O M A T I C S in the mass spectrometer, such as 128, 202, 226, 228, 252, 254, 256, require more drastic molecular rearrangements. Although the conversion of diphenylmethane was less than 5y0,the product was very complex.
Temperature, O C. Time,min. Products, wt. % charge Gas [through C,] cs to ClO CIO' distillatee Residue Total Gas, mole % Hydrogen Methane Ethylene Ethane Propylene Propane Butenes Butanes
co coz
475 30
502 90
0.01
0.32
Trace
Trace
0.0
4.9
...
91 9
... ... ...
... ... ... ... 9 . .
...
49 47
...
... *.. ... ...
... 4 ...
VOl. %
CioHs CiaHio CiaHiz Cl4HlZ CisHio Cl7Hle CisHia CisHiz CiBHiz C19H14 CioHie CzoHio CzoHiz CZOHl4 CioHlb CzoHis CziHiz CziHis Cz&s CZEHZO CzeHzz CSZHZE
128 166 168 180 202 216 226 228 240 242 244 250 252 254 256 258 264 270 330 332 334 410
Unsubstituted Heterocyclic Aromatics. Eight compounds in this category were tested, with the results shown in Table XIII. Practically no reaction of quinoline occurred in 0.5-hour a t 475' C. However, in a more severe test at 500' C., condensation to heavy residue occurred. The gas formed contained an equal quantity of hydrogen and methane, showing extensive decomposition of the quinoline nucleus. Under the same test conditions there was no reaction of its hydrocarbon analog, naphthalene (Table 11). Apparently, the replacement of a CH group in naphthalene by a nitrogen atom leads to a decrease in the dissociation energy of one or more of the C-H bonds, facilitating condensation. The position of the nitrogen atom in the nucleus also has an effect on condensation rate, as shown by the fact that isoquinoline reacted more rapidly than quinoline. Acridine condensed far more rapidly than quinoline or isoquinoline and actually showed measurable condensaflon even a t 450' C. Fusion of a benzo group onto quinoline to form acridine creates a structure similar to that of
Pyrolysis of Diphenylmethane". Mass Spectrometric Analysis of Products above C ~ O
Formula
Mass
Heterocyclic Aromatics
Quinoline
Xll.
Table
0.22 1.35 95.3 0.01' 0.06 0.01 0.05 0.09 0.03 0.13 0.44 0.02 0.06 0.04 0.05 0.06 0.01 0.01 0.01 0.20 1.82 0.03
Masses Expected from Simple Radical Reactions as Primary or Secondary Products
Possible Compounds Corresponding to Formula Naphthalene Fluorene Feed Stilbene Pyrene
Secondary Secondary
Triphenylene Phenylbiphenylene Methylterphenylene Phenylfluorene Triphenylmethane
Secondary Primary
Perylene Binaphthyl Methylphenylfluorene Methyltriphenylmethane
Secondary Primary
Dimethylphenylfluorene
Secondary
Tetraphenylethylene Tetraphenylethane Pentaphenylethane
Secondary Primary Secondary
475O C., 1 hour.
anthracene. I t is therefore not surprising that acridine is much more reactive than quinoline, just as anthracene is more reactive than naphthalene. A comparison of the data in Tables XI11 and I11 shows that acridine is much more reactive than anthracene, forming 56% heavy residue in 15 minutes a t 475' C., while anthracene formed 1.4%.
Indole condensed more rapidly than quinoline or isoquinoline. Carbazole, in which a benzo group is fused onto indole, proved very unreactive even in a prolonged test a t 500' C. Thianaphthene, the sulfur-containing analog of indole, pyrolyzed more slowly than indole a t 475' C. Moreover, dibenzothiophene was practically un-
Table XIII.
Unsubstituted Heterocyclic Aromatics
Isoquinoline
Acridine
475 30 0.02 1.3
...
0.2
450 15
Trace 0.2 99.1 0.7 100.0
475 30 0.7 1.1 42.3 56.2 100.3
-
Indole 477 30 0.07 0.05
...
3.3
$ 7
93
7
... ... ... ..* ... ... ... ...
320.9 433 * 2 Weight of charge, mg. 351.8 0.88 1.02 0.93 Reaction tube vol., cc. 266 216 285 Run No. Distilled at 300' t o 350° C., 1 micron of mercury.
...
31.5 50.4 0.2 11.1 0.0 5.0 0.3 0.3 1.0 0.0
168.0 0.74 223
304.7 1.09 193
66 34
... ... ... 9 . .
... ... ...
43 57
... ... ... ...
... ... ... .... 245.2 1.03 256
Carbazole 474 60
0.0 0.1 100.1 0.0 100 * 2
93 7
... ... ... ... ... ... ... ...
217.1 0.69 200
502 90
Thianaphthene 476 60
0.06
0.01
Trace
Trace
... 0.0
85 12
... ... ... ... ... ... 3
...
218.3 0.72 271
... 0.2
70 30
... ... ... ... ...
... ... ... 300.4 0.90 265
Dibenzothiophene 474 60
0.0 0.2 98.8 0.0 99.0
82 18
... ... ... ... ...
... ... ...
428.7 1.01 201
502 90 0.02
Trace
... 0.0
87 13
... ... ... ... ... ... ... ...
376.5 0.95 270
Dibenzofuran 475 60
0.0 0.2 99.9 0.0 100.1
84 16
... ... ... ... ... ... ...
...
318.9 1.01 220
See experimental section.
VOL.
so,
NO. 2
FEBRUARY 1958
249
Table XIV.
Substituted Quinolines
Quinoline
Temperature, C. Time, min. Products, wt. % charge Gas [through C4 J cs to ClO CIO+distillate” Residue Total Gas, mole % ’ Hydrogen Methane Ethylene Ethane Propylene . Propane Butenes Butanes
co con
2-Phenylquinoline
476 30 0.01
...
...
0.0
91 9
476 30
0.0 0.4 99.6 0.0 100.0
0.2 0.4 99.4 Trace 100.0
87 13
... ... ... ... ...
...
... ... ... ... ... ... ...
... ...
...
Quinaldine
475 60
28 63
...
0.2
... ... ... ... 8 ...
Weight of charge, mg. Reaction tube vol., cc.
a
351.8 213.3 371.3 1.02 0.81 0.87 Run No. 285 202 246 Distilled at 300’ to 350’ C., 1 micron of mercury. See experimental section
changed in a test at 500’ C. As in the case of indole, the fusion of a benzo group onto the original heterocyclic caused a decrease in reactivity. Dibenzofuran, like its analogs carbazole and dibenzothiophene, was hardly affected by pyrolysis at 475’ C. This compound was not tested at 500’ C., but from the results obtained with dibenzothiophene it seems likely that dibenzofuran would also be unreactive a t the higher temperature. The gas product from the unsubstituted heterocyclics was predominantly hydrogen except in the case of indole and in the severe pyrolysis of acridine. In this respect, acridine is similar to its analog anthracene, producing mostly hydrogen in mild pyrolysis but in severe pyrolysis giving methane and other saturates-evidence of disruption of the aromatic nucleus. Substituted Quinolines. Pyrolysis of two 2-substituted quinolines (Z-phenylquinoline and quinaldine) at 475’ C. was carried out, with the results shown in Table X I V . Quinoline did not condense to a measurable degree under the test conditions; introduction of a phenyl substituent in the 2-position does not facilitate condensation, as no bonds appreciably weaker than those in the parent compound are present. A methyl substituent in the 2-position introduces weaker C-H bonds and a weaker C-C bond than those in quinoline and gives rise to some degree of reactivity. Summary Preliminary studies of the effect of concentration on the condensation of
250
anthracene showed that condensation occurred to a significant extent on1)- in liquid phase. Unsubstituted aromatics having the anthracene configuration as part of their structure tend to form heavy residue. In this class, naphthacene was the most reactive of the compounds tested. Unsubstituted aromatics without the anthracene configuration were very resistant to pyrolysis. Alkyl aromatics underwent condensation more rapidly than the corresponding unsubstituted parent compounds. Condensation increased with the number of alkyl substituents. Furthermore, the reactivities correlate with the number of C-C bonds in the position with respect to the aromatic ring; bonds of this type have a comparatively low dissociation energy. An effect of position of the alkyl substituent on the aromatic nucleus was found, 1-methylnaphthalene and 1-methylanthracene being more reactive than the corresponding 2isomers. Mass spectrometric analyses of products above Clo from pyrolysis of the two methylnaphthalenes, two of the dimethylnaphthalenes, and diphenylmethane provided information on possible modes of condensation. Chromatography and ultraviolet spectroscopy showed that important products from 1-methylnaphthalene were naphthalene and alkylbinaphthyls. Aromatics with one or more methylene groups bridging two aromatic nuclei are subject to condensation. The most rapid reaction in the methylene-bridged aromatics was found with acenaphthene. Quinoline and acridine produced more residue than did their hydrocarbon
INDUSTRIAL AND ENGINEERING CHEMISTRY
analogs, naphthalene and anthracene. Indole and thianaphthene underwent slight condensation, while carbazole and dibenzothiophene were relatively inert. Alkyl substitution on quinoline increased the production of heavy residue. Acknowledgment The authors wish to thank B. S. Greensfelder for helpful discussions. R. L. Way, E. R. Coltrin, and C. A. Siven gave experimental assistance. Many of the compounds tested were purified by the Spectroscopic Department of these laboratories. M. J. O’Neal, Jr., and the mass spectrometric group a t the Houston Research Laboratory of Shell Oil Co. provided the mass spectrometric analyses shown in Tables VI, VII? and X I I . D. D. Tunnicliff of these laboratories carried out chromatography and spectroscopy on the product of pyrolysis of l-methylnaphthalene. literature Cited (1 1 Clar, E., “Aromatische Kohlenwasserstoffe,” p. 42, Springer-Verlag, Berlin. 1941. (2) Clar, E.’, Stewart, D. G., J . Chem. SOC.1951, p. 687. (3) Friedel, R . A., Orchin, M., “Ultraviolet Spectra of Aromatic Compounds,” Wiley, New York, 1951. (4) Horrex, C., Miles, S. E., Discussions Faraday Sac. No. 10, 187 (1751). (5) Jaquiss, M. T., Szwarc, M., ilhture 170, 312 (1952). ( 6 ) Kooyman, E. C., Farenhorst, E., Trans. Faraday SOC.49, 58 (1953). (7) Kooyman, E. C., Strang, A , , Rec. trao. chim. 72, 329 (1953). (8) Levy, M., Szwarc, M., J . Am. Chem. Sod. 77, 1949 (1955). (9) McClure, D. S.,J . Chem. Phys. 17, 905 (1949). (10) Madison: J. J., Roberts, R. M., Shell Development Co., Emeryville, Calif., unpublished results. (11) Martin, E. L., J . Am. Cnem. SOC.58, 1441 (1936). (12) Nauman, R. V., Ph.D. thesis, University of California, Berkeley, Calif.. 1947. Norton. J. A,. Chem. Rcvr. 31, 441, 448 (1942). O’Neal, M. J., Jr.. Wier, T. P., Jr., Anal. Chem. 23, 830 (1951). Roberts, R. M.. Madison, J. J., Zbid., 29,1555 (1957). Szwarc, M.: Chem. Revs. 47, 172 (1950). Szwarc, M., J . Chem. Phys. 17, 431 (1 949’1. -,. \ - -
(18) Zbid., 16, 128 (1948). (19) Zbid., 23, 204 (1955). (20) Szwarc, M., Shaw, A,, J . A m . Chem. SOC.73, 1379 (1951). 121) Tilicheev. M. D.. J . Abbl. Chem. (U.S.S.R.) 12, 741 (193jj; Foreign Petroleum Technol. 7. 343 11939). (22) Tunnicliff, D. D., Stone,‘H., Anal. Chem. 27, 73 (1955). RECEIVED for review October 19, 1956 ACCEPTEDMay 27, 1957 Divisions of Petroleum and Organic Chemistry, Symposium on Polycyclic Hydrocarbons, 130th Meeting, ACS, Atlantic City, N. J., September 1956.