noted. The increased pressure did not give increased amounts of by-products as in the butadiene-acrolein reaction. T h e myrcene-acrolein reaction gave good yields a t all the pressure studied (runs 23 through 28). Because of t h e close boiling points of myrcene and myrcene-acrolein adduct (as compared to the reactants and product in the other reactions) this adduct could not be as cleanly separated by distillation. Unfortunately, the product could not be analyzed satisfactorily by gas chromatography and the results of this study were not as precise as those of the other ieactions. No deleterious effects because of higher pressures could be deduced from the physical properties of the product. T h e reaction temperature affected each of the Diels-Alder reactions studied in a different manner. The yields in the butadiene-acrolein study are somewhat scattered-eg., the yield increased slightly with increasing temperature (runs 4, 7, 8). Higher temperatures did not increase the amount of tar formation or the amount of by-products formed. T h e butadiene-methyl methacrylate reaction yields increased sharply at the higher temperatures and higher yields would very likely have been obtained a t temperatures beyond the limit of this coil reactor (runs 16 through 19). For example, Roberts, Jeydel, and Armstrong ( 5 ) obtained a n 847, yield in 15 hours a t 180' C. and Doucet and Rumf (3) obtained a 65% yield in 8.5 hours a t 220' C. Although it was realized that the reaction had not equilibrated i n the 13.2to 16.2-minute reaction time used in the temperature study, the use of a longer reaction time would have reduced the value of the comparison of the butadiene-methyl methacrylate reaction with the other reactions. Higher temperatures gave slightly more tar, but did not affect purity of the distilled product. T h e effect of temperature on the myrcene-acrolein reaction was quite different in that the yield decreased with increasing temperature (runs 26, 29 through 31), accompanied by rapidly increasing tar formation. Lt'hether the starting materials or the adduct \vas decomposing was not investigated.
The effects of reaction time on these. Diels-Alder reactions were extremely interesting. The butadiene-acrolein reaction had reached the point of maximum conversion in less than 18.9 minutes (runs 4, 9, IO). T h e short time required to complete this reaction had been previously suggested by Chayanov (7), who obtained a 91 plus yield of 3-cyclohexene-I-carboxaldehyde by raising the temperature of the reactants in a n autoclave to 130' C. in 7 to 10 minutes and then completing the exothermic reaction a t about 200' C. in 15 to 20 minutes. The butadiene-methyl methacrylate reaction had reached the maximum degree of conversion in less than 34.5 minutes (runs 19 through 22). By-products, however, were formed when the reaction was carried beyond the time required to attain equilibrium. T h e yields were about the same a t all the reaction times studied in the myrcene-acrolein reaction (runs 32 through 34). The reaction time of 17.0 minutes was the shortest possible because of the pump limitations. I n general, the yields obtained by using the coil reactor were not as good as those reported by previous investigators who used autoclaves or sealed tubes. However, the coil reactor easily and conveniently permitted investigation of some extremely interesting aspects of these Diels-Alder reactions. Acknowledgment T h e author is grateful to W. B. Trapp for suggestions in the course of this work and to D. E. Pletcher, J. D. Doedens, E. H. Rosenbrock, and C. W. Nielsen for suggestions on the design and construction of the coil reactor. literature Cited (1) Chayanov, N., J . Gen. Chem. U.S.S.R. 8 , 460 (1938); Chem. Abstr. 32, 7905 (1938). (2) Diels. O., Alder. K.. Ann. Chem. Liebies 470. 62 (1929). (3) Doucet, J., Rumf, P:, Bull. SOC. chim. prance 1 9 5 4 , ' ~ .60. (4) Mousseron-Canet, M., Mousseron, M., Zbid., 1956, p. 391. (5) Roberts, J. D., Jeydel, A. K., Armstrong, R., J . Am. Chem. Sac. 71, 3248 (1949). RECEIVED for review February 6, 1961 ACCEPTED J u n e 12, 1961
T H E R M A L D EA LKY LAT IO N =HY D ROC RACK1 N G
OF ALKYL P H E N O L S G. L. W E L L S ' A N D
RONALD LONG
Chemical Engineering Department, University of Birmingham, Birmingham, Entland
A study has been made of the thermal dealkylationhydrocracking of cresols and
of o mixture o f xylenols a t
about 30-atm. pressure, in the temperature range o f 600 to
700' C., i n contact with a coke bed in a flow reactor. Better yields of cresols and phenol (or phenol alone), have been obtained a t high conversions than from the catalytic or noncatalytic processes reported in the literature.
Very little carbon deposition was observed and
comparatively
little isomerization and disproportionation
occurred; tar formation was small. The order of stability o f the cresols and xylenols t o hydrocracking has been evaluated.
Present address, National Coal Board, Hobart House, Grosvenor Place, London Silil, England.
RUDE TARS vary considerably in their properties and com'positions. Alkyl aromatics and alkyl phenols predominate in low-temperature tars whereas high-temperature tars are characterized by the presence of unsubstituted aromatics and relatively small amounts of low-boiling phenols. Tars from the low temperature carbonization of coal using either fixed or fluidized beds are rich in tar acids-up to 30% by weight. Only a small fraction of the higher boiling tar acids potentially available from high-temperature coal tar is a t present recovered. T h e employment of dealkylation processes in the treatment of the alkyl phenols, alkyl naphthalenes, and alkyl benzenes in tar oils appears to be a possible method of increasing the value of coal as a chemical raw material. T h e thermodynamics of the demethylation of cresols has been studied by Jelinek ( 4 )who has calculated the equilibrium constants for three primary reactions which are predominant in the demethylation of cresols in a hydrogen atmosphere.
VOL. 1
NO. 1
JANUARY 1962
73
rnoDucT
GAS
TO ATM.
Figure 1 .
GAS
w
C O N O E W SATE
P = T = H = S = F =
Layout of apparatus
Pressure gage Temperature point Hale Hamilton pressure controller Shut-off valve Fine control valve
COLD TRAPS
I
Q
LlPUID CONDENSATE.
Demethylation : C&(OH)CH*
+ Hz
+
CeHjOH
+ CH4
(1)
Dehydroxylation : CeH,(OH)CHs
+ H,
+
C$IS*CHI
+ HzO
(2)
Dehydroxylation/demethylation: CGHI(OH)CHI
+ 2H2
+
CeHe
+ CHI + HzO
(3)
The values of the equilibrium constants for the dehydroxylation and dehydroxylation/demethylation reactions are very large and consequently if equilibrium were reached, benzene would be expected to be the main aromatic product. I t is the kinetics rather than the thermodynamics of the reactions which will determine whether it is practically feasible selectively to dealkylate alkyl phenols. There are several possible techniques available for preliminary work on the problem: catalysis of the dealkylation relative to the dehydroxylation; rapid flow through a continuous flow reactor so that equilibrium is not attained (in the absence of an active catalyst) ; and provision of insufficient hydrogen for complete reaction. T h e thermal cracking of alkyl phenols has been studied by Senseman ( 7 7 ) and by Neuworth and Jones ( 8 ) . Comparatively little work has been carried out on the thermal hydrocracking of alkyl phenols (a),although unpublished work has been done by McNeil (6). I n view of previous experience of thermal hydrocracking in this laboratory ( 9 ) it was decided to study the thermal dealkylation-hydrocracking of alkyl phenols a t moderate pressures (up to 30 atm.) in the presence of a coke bed as contact medium.
Experimental T h e reactor for the flow apparatus (Figure 1) was a n Inconel tube about 30 inches long, having inside and outside diameters of 7/* inch and l 3 / , inches, respectively. The flow apparatus and procedure were essentially similar to those described in an earlier article ( 9 ) . I n carrying out an experiment, the hydrogen flow rate was adjusted to that required for a given temperature and pressure. The pump was then started and the flow rate set by timing the fall of liquid level in the metering device. After 74
I&EC PROCESS DESIGN A N D DEVELOPMENT
about 30 minutes, in which period reactor conditions settled down, the condensate collection vessel was drained and the cold traps were switched into circuit. -4check was kept on flow rates during a run which normally lasted for 30 minutes during which time temperatures and pressures were recorded. The collection vessel was drained, and the time taken for the run was noted; the pump was then switched to water, and about 5 minutes later, the hydrogen supply was turned off and the apparatus purged with nitrogen. Liquid samples were analyzed by gas chromatography. The method employed was essentially that described by Karr et a / . (5). For the analysis of the products from the hydrocracking of cresols and xylenols, column 1 was used. However, poor peaks were obtained after 2 : 5 xylenol was eluted in the analysis of the products of hydrocracking of a high-boiling xylenol fraction. Consequently, the low boiling alkyl phenols were analyzed by column 1 using an internal standard, and the A higher boiling products were analyzed on column 2 . katharometer was used as detector and a simple method of estimating peak areas was adopted. As water gave a poor
Different Chromatographic Columns Were Used for lower and Higher Boiling Alkyl Phenols Internal Column Column Length, Diameter, Carrier Temp., NO. Material Meters Mm. Gas ' C. 1 17% wt. dinonyl phthalate on Johns-Manville (2.22 firebrick 3.5 8 Hz 160 2 20% wt. silicone elastomer E301 1.7 8 Hz 185 on Celite 545 (Sample-O.025 ml. injected by hypodermic syringe through a serum cap) Table I.
Table II. Fractions Varied in Preliminary Study Factor Levels 10 and 30 atm. Pressure 650 and 700' C. Temperature 2 : l and4:l Ratio Hz/o-cresol feed Nominal residence time 2.5 and 5 sec.
peak o n the chromatograms, it was analyzed by the Dean and Stark (72) procedure on the liquid products. Little attention was paid to gas analysis in the present work. Gas chromatography, using a n alumina column slightly deactivated with liquid paraffin and using isobutane as a n internal standard, indicated that only methane with traces of ethane and ethylene were present in the hydrogen issuing from the apparatus.
Figure 2. Effect of nominal residence time on conversion of ocresol to various products
Results
See Figure 3 for symbols and conditions
o-Cresol. A preliminary four-factor study a t two levels was made to assess the effects of varying temperature, pressure, nominal residence time, and hydrogen-cresol ratio in the feed upon the phenol content of the products and the percentage conversion of o-cresol. T h e nominal residence time was calculated by dividing the free volume of the coke bed in the constant temperature zone of the reactor by the flow rate a t the temperature and pressure in this zone. The results indicated that temperature and residence time were of prime importance and that over the range studied the effects of pressure and feed ratio were negligible. Following the preliminary work a series of experiments was carried out in which the pressure and Hz-o-cresol feed ratio were kept constant a t 30 atm. and 2 to 1, respectively, and the nominal residence time was varied from 0 to 40 sec. and the temperature from 600” to 750” C. I n plotting the results of these experiments. the “conversion” and “yield” have been defined as follows:
tent-of that portion of the limiting reactant in the feed which is converted and disappears during the course of the reaction. Therefore, in the hydrocracking of o-cresol, the yield of phenol is expressed as phenol in reactor effluent in terms of weight carbon in the phenol divided by the total o-cresol converted in terms of weight carbon in the cresol. I n Figure 2, the percentage conversion to individual products is plotted against the nominal residence time in the reactor and, in Figure 3, the percentage yield is plotted against the percentage conversion of the o-cresol feed. Most of the experiments were carried out a t temperatures of 650” and 700’ C. Only one run was carried out a t 750’ C. Figure 3 shows that the primary products were phenol and toluene and that benzene was a secondary product. There is considerable scatter in the results for “gas and losses” as these were obtained by difference. T h e gas produced was mainly methane with only traces of ethane and ethylene. Diphenyl was not detected in the products.
The “conversion” to any product is the percentage of that product expressed in terms of the feed (measured in terms of carbon content)-e.g., phenol in reactor effluent in terms of weight of carbon in the phenol divided by the total o-cresol feed expressed in terms of the weight of carbon in the cresol. T h e results are expressed in terms of carbon content wherever possible so that no allowance need be made for added hydrogen. I n some cases, where the molecular weight of the feed is unknown, the results are expressed in terms of percentage weight of the substance. T h e “yield” of any product in the reactor effluent is the percentage-in terms of carbon con-
Table 111.
Results on Hydrocracking Cresol Mixtures
All runs carried out at 650’ C. and under 30 atm. pressure; 0-
Feed o-Cresol m-Cresol p-Cresol Products ( yG wt. as carbon)
56
16
16
20.0 80.0
50.0 50.0
Conversion,
Yield,
Conversion,
%
%
7%
Phenol Benzene Toluene Xylene Gas
26.3 10.9 14.1 0.2 13.2
40.7 16.8 21.8 0.3 20.4
29.2 8.9 12.6 0.7 18.6
Unconverted Feed o-Cresol m-Cresol p-Cresol
Wt., % 3.3 32.0
Convd., 83.5 60.0
p- ond o-Cresol
Run Number 57 58 Nominal Residence Tim?, Secs. 16 18 Per Cent WeiPht
55
%
the Hz-feed ratio was 2 to 1
and rn-Cresol
80.0 20.0
60
27
33
53.3
50.0
50.0
46.7
50.0
50.0
Yield,
Conversion,
Yield,
Conversion,
Yield,
Conversion,
Yield,
Conversion,
%
%
%
%
%
70
%
%
70
41.7 12.7 18.0 1 .o 26.6
33.0 15.8 13.9 1.2 14.1
42.3 20.3 17.8 1.3 18.1
26.4 5.4 12.6 0.1 9.5
48.9 10.0 23.4 0.2 17.2
30.3 18.3 13.9 1.9 17.9
36.8 16.9 22.2 2.3 21.8
27.9 22.0 14.6
33.9 17.8 26.8
Convd., Wt., % 8.8 21.1
59
% 82.4 57.8
Convd., Wt., % 13.2 8.8
% 83.6 56.0
Convd., Wt., % 24.5
54.0
Wt., % 6.6
21.5
54.0
11.1
%
VOL.
1
NO.
Convd.,
Yield,
Convd.,
86.8
Wt., % 7.4
85.2
77.8
10.4
79.2
%
1
%
JANUARY 1962
75
The yield of phenol remained virtually constant a t high conversions of o-cresol. Initial experiments were a t first not reproducible but the percentage conversion of feed, a t any given nominal residence time and temperature, gradually fell until reproducible results were obtained. This effect was considered to be due to the formation of a carbon film on the reactor surface.
present work did not separate m- and p-cresols; consequently, experiments were planned using mixtures of o- and p-cresols and o- and rn-cresols, respectively. Hydrocracking was then carried out a t 650' C. and 30 atm. with a hydrogen-cresols feed ratio of 2 to 1. T h e results of these experiments are summarized in Tables I11 and IV. Table I V shows that
&,
Relative Stabilities of Alkyl Phenols to Hydrocracking
km.,re,",
I n studying the relative stabilities of alkyl phenols to hydrocracking, it was easier to use a mixture for comparative purposes as this avoided having to ensure strict reproducibility of experimental conditions. Consequently, the variables-nominal residence time, volume change, and hydrogen concentration-could be eliminated. I t is assumed that the rate of hydrocracking is proportional to the first power of the alkyl phenol concentration and to some unknown power of the hydrogen concentration. The hydrocracking of o-cresol has been found experimentally to be first order with respect to o-cresol, but it is not proposed to discuss the work on the kinetics of this reaction until more accurate confirmatory results have been obtained.
--d(creso1,) dt
= ka(cresol,)(H2)n
for a n individual cresol ( u ) . I n hydrocracking a mixture of isomers, the term ( H z ) will ~ be the same for each one. Therefore dx/dt a k , ( n a o - x ) for cresol ( a ) dv/dt akb(nb0
-y)
for cresol ( b )
where nao = number of moles of cresol ( u ) initially 7Lbo
= number of moles of cresol ( b ) initially
and y are number of moles of (a) and ( b ) , respectively, which have reacted after time t.
x
dx/dt dy/dt
- dx - k,(nao - x ) dy kb(nbo - Y )
dx ka(noO - x ) =
dY kb(nb"
- y)
&2EE!
= 1.2
kp.cresol
and that E,-,,,,,,
- E,.,,,,,, = 1.4 kcal./mole - Eo.o,e,ol= 0.3 kcal./mole
These values agreed reasonably well with those obtained in a separate series of experiments on the hydrocracking of cresols individually. T h e latter gave 1.1 and 0.4 kcal./mole, respectively. T h e order of stability of the cresols with regard to hydrocracking was thus: m-cresol > p-cresol > o-cresol. Xylenol s
Hydrocracking experiments were also carried out upon mixtures of xylenols a t a temperature of GOO0 C. and 30 atm. of pressure with a nominal residence time of 20 seconds and a hydrogen-feed ratio of 2 to 1. The main reaction was the conversion of the xylenols to cresols. T h e yield of phenol was very small under the conditions used, and the production of benzene was negligible. The only isomerization reactions detected were 2,4-xylenol to ?,3-xylenol and 2,6-xylenol to 3,5-xylenol. Although the gas chromatographic technique used was not capable of detecting all possible isomerizations, the extent of isomerization was believed to be small on the basis of the analytical results obtained. Relative Stabilities of Xylenols to Hydrocracking The procedure described above was used to calculate the differences in over-all activation energies for the hydrocracking of xylenols. By combining the results in Table V, the order of stability of the xylenols to hydrocracking was :
3,5-xylenol
Integrating, with x = o, y = o
2.1 and
>
2,5-xylenol > 3,4-xylenol and 2,6-xylenol 2,4-xylenol > 2,3-xylenol
>
The Hydrocracking of a High-Boiling Xylenol Fraction
using the Arrhenius equation k = A.e-"IRT and assuming the A factor to be the same when hydrocracking isomers. The gas chromatographic technique employed in the
Table IV.
Calculation of Differences in Over-all Activation Energies for Pairs of Isomers Using the relationship in Equation 4
Run No.
nao
xo
nbo
E.
nb ka
Hydrocracking of m-Cresol ( a ) and 55 80 32.0 20 3.3 8.8 56 50 21.1 50 80 13.2 8 . 8 57 20
- Eb Kcal./Mole
0-Cresol ( b ) in a Mixiure
1.98 2.01 2.20 Average 2.1
1.3 1.3 1.5 1.4
Hydrocracking of p-Cresol ( a ) and o-Cresol ( 6 ) in a Mixture 0 58 53.3 24.5 46.7 21.5 1 . 0 0.6 59 50.0 6 . 6 50.0 11.1 1 . 3 5 0.4 60 50.0 7 . 4 50.0 10.4 1.22 Average 1 . 2 0.3
76
I&EC
PROCESS DESIGN A N D DEVELOPMENT
A series of runs was carried out on a commercial "highboiling xylenol" fraction (Table VI). I n the gas chromatogram of this, a number of broad bands were obtained which were designated xylenol I, xylenol 11, and high-boiling alkyl Table V. Order of Stabilities of Isomeric Xylenols I s Evaluated from Difference in Over-all Activation Energies of Hydrocracking Mixture A Order of stability 3,4-xylenol > 2,4-xylenol > 2,3-xylenol Difference in over-all activation energies, kcal./mole ~ 0.1 E1,4- E Z ,=~ 0.5 E z , ~ E z , =
-
Mixture B 3,5-xylenol > 2,5-xylenol
Order of stability Difference in over-all activation energies, kcal./mole E s ,~ E2,6 0.5 E2,6 Mixture C 2,6-xylenol > 2,4-xylenol Order of stability Difference in over-all activation energies, kcal./mole E z ,~ Ez,c = 0.4
>
E2,6
2,6-xylenol = 0.2
phenols I and 11, respectively. T h e xylenol I contained mainly 2,4 and 2,5-xylenols while xylenol I1 comprised mainly 2,3, 3,4, and 3,5-xylenols with possibly some ethyl phenols. T h e high-boiling alkyl phenols comprised mainly alkyl phenols boiling higher than the xylenols-e.g., trimethyl phenols and phenols substituted with larger alkyl groups. During this series of experiments the pressure was kept a t 30 atm. (gage) and the hydrogen-feed ratio a t 4 to 1. I n Figure 4, the yields of individual products are plotted against the conversion of the feed material. For convenience, the 4.3y0 by weight of rn- and p-cresols in the feed originally was considered as a product of reaction. I n Figure 5, the conversions of the various constituents of the feed and of products are plotted against nominal residence time. Cresols were primary products whereas phenol, toluene, and benzene were secondary products-the latter probably arising from both phenol and toluene. T h e high-boiling alkyl phenols were hydrocracked more readily than xylenols. At high conversions of the feed, the phenol yield ceased to increase although the yield of cresols fell as they were converted to phenol. The reason is probably that the phenol was itself being dehydroxylated to benzene. There would thus be advantages in operating with a relatively short residence time if a high yield of cresols were required. If 100 grams of feed material were used, then a t a conversion of 90% by weight of high-boiling xylenols, the yields of cresols and phenol were both 20% by weight. Therefore, 90 grams of high-boiling xylenols gave 18 grams of phenol and 18 grams of cresols. If the yield of phenol from cresols is assumed to be 40% under these conditions, then the 18 grams of phenol came from 45 grams of cresols as intermediates. Consequently, 90 grams of feed could give rise initially to 63 grams cresols viz. a yield of -70% by weight. If, however, phenol were required as the main product a mixture of xylenols and cresols might be subjected to hydrocracking, the xylenols forming cresols (and then phenol) and the cresols forming phenol. After removal of phenol and aromatics, the residue might be recycled.
f
10
. 1 100
20 10
20
30
40
60
SO
' / r C O N V C R S t O N Of
IO
SO
7 0
PO
CRESOL
-
I
30+
w
10
10
30
20
40
50
Figure 3. o-cresol
10
70
60
3.wr A S
01. CONVERSION O F C R E S O L
PO
100
c*aBoQ
Yield of products plotted against conversion of
0
Reactor temperature:
750' C. C. 65OOC. 600'C.
0 700' X
Discussion T h e primary products of the thermal hydrocracking of o-cresol were phenol, toluene, water, and methane, while benzene was a secondary product. An interesting feature is that there appeared to be little disproportionation and isomerization. Only traces of tarry material and a little xylene were produced. T h e thermal hydrocracking of xylenols and of a high-boiling xylenol fraction indicated that the primary aromatic products were cresols and xylenes with toluene and phenol as secondary products. Further hydrocracking led to the formation of
Table VI.
Properties of High-Boiling Xylenol Fraction of Coal Tar Specific gravity 1.02 Boiling range 5% 219OC. 220° c. 90% 228' C. 95% ' 230OC. m- and &-cresols 4 . 3 7 0 by weight Xylenol I 2 3 . 9 % by weight Xylenol I1 4 4 . 2 70 by weight High-boiling alkyl phenol I 16.97, by weight High-boiling alkyl phenol I1 10 . 3 7 , by weight
A
Pressure: 30 atm. (gage) Hs-Feed = 2 : l
60
so 40; 10
-
20;
!
10 ?
c Figure 4. Yield of products plotted against conversion of high boiling xylenols
7
g
40-
*
10-
20 10
40
10
60
70
10
PO
100
CIISOL
. )