At solvent-dilution ratios of both 0.5 to 1 and 2 to 1, lowering the chill rate from 60’ F. per minute increased the crystal settling rate because of decreased supersaturation and nucleation and increased time allowed for the wax molecules to form large crystals (Figure 3, Tables I and 11). However, low chill rate gives low process capacity and the evaluation of an optimum chill rate should consider a balance between centrifugation efficiency and over-all dewaxing throughput.
dilution and chill rate. In the latter case, the use of crystal modifiers can change the crystal needles into aggregates for improving centrifugation efficiency. Acknowledgment
The author thanks G. Ciprios, R. B. Long, and E. N. Ziegler for their helpful review of the manuscript and J. Kostick for his assistance with the experimental work. literature Cited Chamberlin, N. F., Dinwiddie, J. A., Franklin, J. L., Ind. Eng.
Conclusions
Centrifugal separation of wax and dewaxed oil in a process for dewaxing by freezing is practical. Chill rates more than 100 times those used in the conventional dewaxing processes can he employed to achieve large capacity. The centrifugation efficiency was analyzed in terms of initial settling rate and wax compaction ratio. The wax crystals assume different shapes at different dewaxing process conditions. In general, crystal plates give higher settling rate and wax compaction during centrifugation than the crystal needles, aggregates, and other shapes. In the dewaxing of Solvent 100 Neutral, a middle distillate, the crystal plates form a t low solvent dilution and chill rate, whereas crystal needles form a t high solvent
Chem. 41, 566 (1949). Frolich, P. K., U. S. Patent 2,015,748 (Oct. 1, 1935). Li, N. N., IND.ENG.CHEM.PROCESS DESIGNDEVELOP. 7, 239 (1968). Li,’N. N,, Torobin, L. B., U. S. Patent 3,303,121 (Feb. 7, 1967). Moser, F. R., U. S. Patent 2,124,531 (July 26, 1938). Nelson, W. L., “Petroleum Refinery Engineering,” McGraw-Hill, New York, 1958. Salmon, R., Pullen, E. .4., U. S. Patent 3,130,143(April 21, 1964). Torobin, L. B., U. S. Patent 3,294,672(Dec. 27, 1966). Torobin, L. B., U. S. Patent 3,311,296(March 28, 1967). RECEIVED for review March 11, 1968 ACCEPTEDJune 27, 1968 Tripartite Chemical Engineering Conference, Montreal, September 24, 1968
ALKYLATION OF ISOBUTANE WITH VARIOUS OLEFINS IN T H E PRESENCE OF SULFURIC ACID R . J.
S H L E G E R I S ’ A N D
L.
F. A L B R I G H T
Purdue University, Lafajette, Ind. 47907 lsobutane was alkylated with
CS and C4
olefins in the presence of sulfuric acid in small reactors, one
of which allowed the observation of the change in product composition and yield with time. Average residence times of 5 to 60 seconds were used a t various rates of agitation. Both acid- and oil-continuous emulsions were studied. The hydrocarbon product was analyzed using gas chromatography. Numerous runs were made to determine quantitatively the effect of operating variables on the product quality and yields. Secondary and primary reactions were also investigated; considerable amounts of trimethylpentanes were formed during secondary reactions but dimethylhexanes were produced to an appreciable extent during the primary reactions.
the alkylation of isobutane (2-methylpropane) c3-C~ olefins is of considerable commercial importance for the production of gasolines with high octane numbers, numerous questions still exist concerning the reaction mechanism, which involves numerous consecutive and simultaneous chemical steps that yield a complex mixture of products. In addition, in all commercial applications, the reaction system consists of two liquid phases (the acid catalyst and the organic reactants and products), and mass transfer considerations are obviously of extreme importance (Jernigan et al., 1965), especially since the predominant reactions occur in the acid phase. The most generally accepted mechanism for the alkylation of butenes with isobutane to produce isooctanes is a chain reaction involving carbonium ions (Schmerling, 1945, 1953, 1955). LTHOUGH
A with
Present address, Chevron Research Co., Richmond, Calif. 92
I & E C PROCESS D E S I G N A N D DEVELOPMENT
The olefin molecule reacts with a tert-butyl cation to form a higher molecular weight cation. This cation abstracts a hydride ion from an isobutane molecule to produce a saturated isoparaffln and another tert-butyl cation which continues the reaction. Numerous secondary and side reactions are proposed to explain the complex mixture of products (Ipatieff and Schmerling, 1948 ; Schmerling, 1964). Although this mechanism explains many features of alkylation, other aspects of alkylation, including the production of trimethylpentanes and dimethylhexanes (Zimmerman et al., 1962), are still not definitely settled. Much of the early work on alkylation was primarily concerned with the chemistry of the process, but more attention has recently been paid to physical factors, such as mass transfer, solubilities, viscosity, and interfacial tensions (Albright, 1966; Cupit et al., 1961, 1962). The olefins are known to be highly soluble in the acid, but isobutane is only slightly soluble.
Transfer of the isobutane from the organic to the acid phase is probably the controlling step. Agitation may be the single most important operating variable, and it affects the type of emulsion formed between the acid and hydrocarbon which affects the mass transfer rate. I t is generally agreed that high local concentrations of olefin in the acid phase are to be avoided in order to suppress the undesired polymerization reactions (Goldsby and Gross, 1963; Mayer, 1963). In apparent direct conflict, three recent Texaco patents (Goldsby, 1966) describe methods in which propylene is brought in contact with spent alkylation acid to form dipropyl esters. These esters are separated from the other organic material in the acid and fed back to an alkylation reactor. When the esters are brought in contact Ivith isobutane, alkylate and regenerated sulfuric acid are obtained, I n the present investigation, isobutane was alkylated with Ca and C4 olefins and mixtures of olefins using sulfuric acid as the catalyst. Information was obtained on the importance of the various operating variable, character of the emulsion, and secondary reactions in continuous flow stirred-tank reactors.
P a c k i n g Nut N o t Shown
A o i t a t o r Shaft-
s a c k i n g Space C 1"o nO.D. t a i n s x f o1u/ 2r " React01
Bronze
Reactor P o r t s 1 / 8 ~ 10.0. t u b i n p
T e f l o n Gaskets Rsaction Vessel
Experimental
Sulfuric acid, isobutane, and olefin were metered to one of the two reactors used (Shlegeris, 1967). The single-stage reactor used was the same one employed earlier (Mosby, 1964; Phillips, 1965). The flow sheet, operating procedure, and product analysis were also identical, except for recovery of the product. I n this investigation, the product stream was collected in a separatory funnel at essentially atmospheric conditions, and the emulsion was allowed to separate into acid and hydrocarbon phases. The gaseous product from the separatory funnel was condensed in a dry ice trap. The second reactor (Figure 1) contained five separate cylindrical reaction chambers positioned one above the other. Each chamber had an inside diameter of 1.0 inch, a height of 0.75 inch, and a capacity of 0.45 + 0.1 cu. inch (7.7 ml.) when the agitators were in place. Each chamber contained a three-bladed impeller with a diameter of 0.95 inch, and the impellers were positioned on the agitator shaft which had a diameter of j/16 inch and was located a t the central axis of the reactor chambers. The holes in the partitions between the reaction chambers had a diameter l / 6 4 inch greater than the diameter of the shaft and hence gave close clearance to minimize backmixing between chambers. A packing gland and nut a t the top of the reactor provided a seal for the agitator shaft. Two openings in the side wall of each of the five reaction chambers were used for either adding reactants or removing product as desired. The reactor pressure was controlled by a Grove small volume, backpressure regulator of 15- to 150-p.s.i.g. range equipped with a Teflon diaphragm. For the fivestage reactor, two regulators were used. All major hydrocarbon components were identified chromatographically, and acid and hydrocarbon material balances were made for all runs. The yield, defined as the grams of product per 100 grams of olefin reacted, was also calculated for each run. The results are precise within about lo%, but the absolute accuracy is less. I n general, the values reported are higher than theoretical values. Yield results as reported here are, however, useful to indicate the relative importance of alkylation us. polymerization. I n polymerization reactions, the olefin reacts with itself, and no isobutane is incorporated into the product. Hence, the yield is about 100. I n alkylation, each molecule of olefin reacts with a molecule of isobutane, giving a theoretical yield of over 200. Hydrocarbon reactants were obtained from the Mathieson Co. The isobutane had a purity of 96% and contained 3.5% n-butane plus traces of ethane and propane. I-Butene was 98y0 pure with 1.5y0isobutane and traces of ethane and
1.0.
T e f l o n W-ring8
Reactor I n l s t 1/8" 0.0. t u b l n g T e f l o n Bearing
Figure 1.
Five-stage reactor
2-butene. 2-Butene was essentially a 1.3 to 1 mixture of cis and trans isomers of 2-butene with traces of isobutane, 1butene, ethane, and propane. The isobutylene(2-methylpropene) and propylene had minimum purities of 99%. Two concentrations of sulfuric acid were used. The 99% acid was a blend of 95 to 96% reagent-grade acid and fuming sulfuric acid. The 95.5% acid was alkylation acid furnished by the American Oil Co.; it was rather dark in color and contained conjunct polymers, or red oil. Comparison of Olefin Feeds in Single-Stage Reactor
A series of runs was made in the single-stage reactor to compare the products obtained when a pure butene was used as the olefin feed. The reactor was operated with the acid phase continuous a t the following range of conditions : Feed acid strength, 7c rZgitator speed, r.p.m. Temperature, O C. Isobutane-olefin volume ratio Acid-organic volume ratio Retention time, sec.
95.5 and 9 9 . 0 600 and 2000 10 5.O:l
1.5:l 15-60
The results were correlated using a linear least-squares regression analysis, and statistical models were obtained for several variables. These models were written for coded data and, as shown in Table I, include interaction terms. Both agitation and retention time had strong effects in addition to a positive interaction between these variables. The results obtained a t the two acid strengths were often similar, although there was generally a n interaction between acid strength and one of the other variables. The alkylates obtained using isobutylene as the olefin were considerably inferior in quality (octane number) to the alkylates from the n-butenes. More light and heavy ends but less octanes, especially trimethylpentanes, were present in the alkylates produced using isobutylene. The n-butenes produced alkylates with similar octane numbers over the range of conditions used. VOL. 8
NO. 1
JANUARY 1 9 6 9
93
Table 1.
x2=
xq =
25
=
Predicted Responses over Range of Operating Conditions in SinglaStage Reactor
.Ifl .I .I
+ l Feed acid = 95.5% -1 Feed acid = 99% Agitator speed = 2000 r.p.m. -1 Agitator speed = 600 r.p.m. +1 Retention time = 60 sec. -1 Retention time = 15 sec. +1 2-butene -1 I-butene 0 isobutene $1 2-butene 0 1-butene -1 isobutene
Table IV.
Response Octane numberb
Prediction 87.42 0 . 8 1 x~ 0.30 x3 0.39X4 0.9OX5 0.34 ~ 1 . ~ 3 0 . 3 0 ~2x3$- 0 . 2 7
Yieldc
1 7 2 . 4 - 5 . 9 XI 3 1 . 4 ~2 1 4 . 8 x3 6 . 9 x4 9 . 2 x2x3 9 . 3 ~ 1 x 4 8 , 1 ~3x5 27.49 2.76 x z 1.74 xp 4 . 0 6 xg 0.88 ~ 1 x 3 1 . 7 9
+
+
++ + + + ++ + + 25.76 + 3 . 4 7 xs + 1 . 5 1 4 . 0 7 + 6 . 7 0 x5 + 1 . 2 1 +8 1 . 3 9 + 1 . 7 0 ~1.~ 46.31 - 5 . 4 4 xp - 1 . 7 6 xg + 1.94 - 2.05 - 2.29 - 1.66 + 1.42 10.17 + 1.48 + 0.86 1 . 7 4 + 2 . 3 4 x5 + 0 . 5 5 + 0 . 6 6 ~3x5 x3x5
Light ends,d
Standard Errora 0.50
10
XJ
Runa 102 106 Agitator speed, r.p.m. 600 2000 3000 Yield 102 169 176 Heptanes, Yo 24.0 38.7 46.1 Heavy ends, yo 54.2 34.3 28.3 a Operating conditions. 25' C.; 5.7: 1 ratio of isobutane topropylene; 60-second residence time; 99.3yo acid.
2.0 Table V.
2.5
~4
~2x3
Heavy ends,.
7c
X3X5
~1x3
2.7
~5
~4
70
XI
1.2
~3
~4
~ 2 x 3
a Standard error for prediction. Standard errors for coegicients not reported. Calculated based on chromatographic analysis. Grams of C& portion of product. C5 product per 100 grams of olefin. e C, portion of product.
++
Table II.
Alkylation Results in Single-Stage Reactor for Mixtures of C4 Olefins x1 = fraction 1-butene in olefin feed x 2 = fraction 2-butene in olefin feed
Standard Variable Octane number Yield Octanes, yc 2,2,4-TMP, 7 0 Heavy ends, T Light ends, %
87.38 231.6 22.35 10.14 41.56 34.88
Model 2 , 4 5 x12
+ + 1 ,88 x 2 ++ 614.48 .46 + 410.76 ,17 XI XI'
~2
~2
- 6 . 3 8 x1 - 6 . 9 0 xi - 8 . 3 4 xe
0.55 10 1.4 0.8 2.0 1.0
99.0 2000 10 5.0:l 1.5:l 60
The models for these runs (Table 11) indicate that there are no significant synergistic effects-Le., the response of a mixture was merely a weighted average of the responses for the pure olefins. Isobutylene alkylate again was of lower quality, although for these operating conditions the yields were comparable. The products from 1-butene and 2-butene were also of com94
Calculated octane number Yield
0
50
100
50
100
89.0 88.2 83.7 87.4 84.6 221 223 107 242 169
a Operating conditions. Ratio of isobutane to olefin of 5 : 7 to 5.7: 7; 60-second residence time; 99.370 acid; 2000 r.p.m.
parable quality, but 1-butene alkylate contained an increased amount of trimethylpentanes and total octanes and less heavy ends. Runs were also made using pure propylene as the olefin feed, and the following results were obtained :
E7707
Another series of runs was next made in which the acid phase was continuous and for which either a pure C4 olefin or a mixture of C4 olefins was used as the olefin feed stream. The operating conditions investigated were as follows: Feed acid strength, T0 Agitator speed, r.p.m. Temperature, C. Isobutane-olefin volume ratio Acid-organic volume ratio Retention time, sec.
Effect of Propylene in Olefin Feed for Alkylating lsobutane Runa 98 111 103 110 102 C. 10 10 10 25 25
Temperature, O Propylene in olefin feed,
%
~ 2 . ~ 3
x2x4
2,2,4-TMP,
Effect of Agitator Speed When Propylene Used to Alkylate Isobutane 107
x2x5
Octane, %
Effect of Temperature When Propylene Used to Alkylate lsobutane Run0 103 102 704
Temperature, ' C. 10 25 40 Yield 107 169 215 Heptanes, % 32.6 38.7 34.3 Heavy ends, 7o 49.9 34.3 32.8 a Operating conditions. 2000 r.p.m.; 5.7: 1 ratio of isobutane to propylene; 60-second residence time; 99.3% acid.
i{
+ + +
Table 111.
I L E C PROCESS D E S I G N A N D DEVELOPMENT
Increasing the operating temperature from IO' to 40' C. increased the yields and quality of the product (see Table 111). Yields of product were much higher with 99.3y0 acid, compared to 95.5%. Increased agitation increased both the yield and quality of the product (see Table IV). U p to perhaps 10% of the propylene was converted to propane. Propane formed was difficult to measure, since some was present in the propylene feed. Preliminary evidence indicated that more propane was produced with 99% (fresh) acid than with 95.5% (used) acid. Small amounts of all trimethylpentanes and dimethylhexanes that were produced during butene alkylation were also produced here. Several runs were made using an olefin mixture containing 50% propylene and 50% C4 olefins. The three Ca olefins in this mixture were present in equal parts. As indicated in Table V, increasing the temperature from 10' to 25' C. only slightly changed the yield and decreased the calculated octane number. The iodine number was measured for the alkylate product in many cases and generally was less than 2, indicating little unsaturation. Much higher degrees of unsaturation were found earlier (Mosby and Albright, 1966) for alkylates produced in the same reactor but with a different recovery technique. In the present technique, there was a greater chance for secondary reactions in the recovery flask, since the emulsion
Table VI. Effect of Emulsion Type on Alkylation Product Continuous Phase Oil Acid
Calculated octane number 88.2 91.5 Yield 134 187 21.2 23.9 Light ends, FG 34.2 51.5 Octanes. 97 Heavy ends, % 44.6 24.6 10' C., 3000r.p.m., 60-second residence time, 5: 7 isobutene to 1-butene.
generally required several minutes for separation of the two phases. Possibly the olefins originally present were saturated as a result of these secondary reactions. Emulsion Characteristics in Single-Stage Reactor
The method of starting the reactant streams to the reactor was important in determining the type of emulsion produced. When acid flow only was initially fed to the reactor until the reactor was completely filled with acid and then the hydrocarbon flow was started, the emulsion produced was acidcontinuous a t volumetric ratios of acid to hydrocarbons from about 1 : 1 to 2: 1. When, however, the hydrocarbon flow was started first and later the acid flow was started, the emulsion generally was hydrocarbon continuous even a t volumetric ratios greater than 1 to 1. The two types of emulsions were different in appearance as they left the reactor. The acidcontinuous emulsion was more frothy and separated more slowly than the oil-continuous emulsion. Several runs were made for various olefin feeds in the singlestage reactor to compare oil-continuous and acid-continuous emulsions. I n all cases, the yields and quality of the alkylates were significantly better when the emulsion was acid-continuous. Table V I indicates such a comparison for runs using I-butene as the olefin a t 10' C., 3000 r.p.m., 95.5% used acid, 60-second residence time, and a 5 to 1 ratio of isobutane to I-butene. The quality of the alkylate changed less for the different olefins used when the emulsions were oil-continuous as compared to the acid-continuous emulsions. Since acid-continuous emulsions give higher quality alkylate, they were used in most runs. In general, the volumetric ratio employed was about 1.5 to 1, since there was less chance a t such a ratio for a phase inversion. Secondary Reactions in the Five-Stage Reactor
Secondary reactions were important in the five-stage reactor (Figure 1) when either I-butene or 2-butene was used in the olefin feed stream. Even for runs with retention times as low as 5 seconds per reactor stage, all C4 olefins reacted in the stage
in which they were added, indicating that the primary reactions were very fast. However, the yield and quality of the alkylate often varied as the emulsion moved from stage to stage, indicating that slower, secondary reactions were also important. The octane number increased from 86.7 to 90.9 when 2-butene was used as the olefin and from 88.4 to 89.8 for 1-butene. Furthermore, the yield of total alkylate increased for each of the above olefins. When a mixture of 50% isobutylene and 50% 2-butene was employed, the octane number increased from 87.2 to 89.4, with an increased yield. Table VI1 shows how the yield of select compounds varied in the different stages for the above mixture of olefins and also for 2butene. When isobutylene was used as the olefin feed, the product composition and yield changed a t most only slightly as the emulsion moved to the later stages of the reactor. Changes in the product between the different stages in the five-stage reactor indicated which groups of the product ivere from the initial reactions and which were from secondary reactions. The initial reactions occurred in the stage, generally stage 2, in which the hydrocarbon reactants and the sulfuric acid were first in contact. Secondary reactions occurred in the later stages of the reactor, generally stages 3, 4, and 5. The yields of heavy ends were generally constant from stage to stage, implying that the heavy ends were products of the initial reactions, presumably forming through olefin polymerization. The yields of dimethylhexanes (DMH) increased slightly from stage to stage. but the largest portions were formed initially; hence, most of the dimethylhexanes were formed in the initial reactions. The trimethylpentanes were the principal products formed during the secondary reactions. Results for 2,2,4trimethylpentane (2,2,4-TMP), the single most important member of this family of compounds, are also shown in Table V I I . The amounts of light ends generally increased somewhat in later stages of the reactor, indicating that they were produced to some extent during secondary reactions. The changes of alkylate quality in the five-stage reactor when 2-butene was the olefin were large, but the composition of the product leveled off after stage 4 (or about 45- to 60-second total retention time). With 1-butene as the feed olefin, the product was still improving in quality after a total retention time of 60 seconds, indicating that the secondary reactions were slower for 1-butene. The quality of 1-butene alkylate from the last stage of the reactor was not as high as 2-butene alkylate; but the secondary reactions, which were beneficial to the product, may not have been allowed to go to completion for 1butene. In one run, the olefin (a 50: 50 mixture of 2-butene and isobutylene) and acid were in contact in stage 1, followed by isobutane in stage 2. With this contacting procedure the domi-
Yields of Various Components of Alkylate at Various Stages of Five-Staged Reactor Yield ____-Component Stage 2 Stape 3 Stage 4 Stage 5 Olejn Feed
Table VII.
Run A'o. 122
125
2-Butene
Light ends Total octanes Heavy ends 2,2,4-TMP 2,4- and 2,5-DMH
38 44 73 20 11
55 73 83 34 14
57 111 74 55 15
50 106 76 52 15
50:50 Mixture of 2-butene and isobutylene
Light ends Total octanes Heavy ends 2,2,4-TMP 2,4- and 2,5-DMH
65 44 79 19 12
79 79 81 39 15
76
78 85 79 44 16
85
86 41 16
Runs at 10' C., 5 to 7 ratio of isobutane to olejn, 2000 r.p.m.
-
VOL. 8
NO. 1
JANUARY 1969
95
Table VIII. Hydrocarbon Product Obtained When 50: 5 0 Mixture of Isobutylene and 2-Butene Contacted with 99% HzS04
Isopentane 2,3-Dimethylbutane 3-Methylpentane 2,4-Dimethylpentane 2,3,4-Trimethylbutane 2-Methylhexane 2,3-Dimethylpentane and 3-methylhexane 2,2,4-Trimethylpentane 2,4- and 2,5-dimethylhexane 2,2,3-Trimethylpentane 2,3,4-Trimethylpentane and 2,3-dimethylhexane 2,3,3-Trimethylpentane 2,2,5-Trimethylhexane Residue Calculated octane number
5.0
4.6 1.4 1.9 0.8 2.1 2.3 1.9 4.2 0.4 1.8 1.4 8.0
64.1 82.4
nant reaction path was polymerization, as indicated by low yields and large percentage of heavy ends in the product. I n another run, only acid and the above olefin mixture were fed to the first stage of the reactor, and a sample was collected from stage 2. The product shown in Table VI11 was mainly heavy ends (72%), but all of the usual alkylate components were present in appreciable quantities. Measurement of the iodine number indicated that little unsaturation was present in the product. Thus, even though no isobutane was present, the collected hydrocarbon product was predominantly paraffins. There was obviously a significant buildup of soluble organic materials in the acid. A comparison of the products taken with and without isobutane flowing indicated that some of the components in the product were not greatly affected by the absence of isobutane. In particular, the light ends were present in comparable amounts. Also, the amounts of the dimethylhexanes as well as 2,2,5-trimethylhexane were only slightly less when isobutane was absent, indicating that the main source of these components may be through reactions of the olefin alone. Methods of Adding Reactants
Several methods of adding the isobutane and olefin feeds were tested in the five-stage reactor in runs a t IOo C. and with agitator speeds of 2000 r.p.m. Fresh (99%) acid was added in all runs to the first stage. Half of the olefin feed (a 50: 50 mixture of 2-butene and isobutylene) was premixed in three runs with the isobutane feed, and introduced to the stage 2 of the reactor. The second half was added to the second, third, or fourth stage. The total over-all ratio of isobutane to olefins was 5 to 1. Only small differences were noted in the quality of the exit alkylate for these runs, with perhaps slightly poorer quality in the run in which all the olefins were added to the second stage. The importance of premixing the olefin and isobutane feed was next investigated. Runs were made in which all, half, and none of the olefin feed was premixed with the isobutane feed. No significant differences were noted in the composition or amount of alkylate produced in these runs. In one run, 2-butene: premixed with isobutane, was added to the second stage; and isobutylene was introduced in the fourth stage. The order of introducing the olefins was reversed in another run. In both runs, a product quality and yield increased in the stage following the one in which 2-butene was added; but there was essentially no change in the stage following the one in which the isobutylene was added. The 96
I&EC PROCESS D E S I G N A N D DEVELOPMENT
quality and compositions of the alkylate products from the last stage of the reactor were similar in both runs. Two runs were also made a t 10' C. to test the preferred order of adding propylene and a 50: 50 mixture of isobutylene and 2-butene. Very low yields of alkylate were obtained in stages 2 and 3 when propylene was added as the only olefin to stage 2. The addition of mixed Cd olefins to stage 4 resulted in the alkylation of the propylene, so that good over-all yields of alkylate were obtained in stage 4 and especially stage 5. LVhen the C 4 olefin mixture was added first, yields of alkylate were good, even when the propylene was added later. Alkylate quality and yields were similar in these two runs. The strength of the acid (obtained after the emulsion had separated) was measured for various stages of the reactor. Essentially all of the drop in acid strength occurred in the stage to which the first olefin was added. The possible exception was the run in which pure propylene was the second olefin added. When the yield and quality of the alkylate improved as it flowed through the reactor, the strength of the acid also simultaneously increased. Part of the apparent acid consumption in the initial stages of the reactor is then temporary, if additional residence time is allowed in the reactor. Mechanism of Alkylation
The alkylation reactions of this investigation included initial reactions involving primarily the olefins followed by relatively slow secondary reactions of some of the initial products with isobutane. Olefins were undoubtedly the principal reactants in the initial reactions, for the following reasons: Olefins react rapidly with sulfuric acid-e.g., at least 45% of 1-butene reacts with sulfuric acid in a second (Naworski, 1966). Isobutane was present in the acid phase of the emulsion to only a very small extent, especially during the initial phases of the reaction. The solubility of isobutane is low, about 0.1 weight yo in fresh acid, but probably somewhat higher in used acid (Albright, 1966). Olefins, however, are much more soluble in sulfuric acid. The low solubility of isobutane coupled with a rapid rate of olefin reactions would cause mass transfer limitations for isobutane, as has been suggested (Albright, 1966; Jernigan et a/., 1965). The yields of individual components in the reactor showed that the products of the initial reactions with olefins were primarily the more undesired materials, including the heavy ends, light ends, and dimethylhexanes. Acid-soluble products, probably including esters, some of Ivhich react further in secondary reactions, were also produced in the initial reactions. Under the correction conditions, the esters react further to produce high quality alkylate. The Texaco patents indicate that at least one set of conditions does this (Goldsby, 1966). Reversible esterification reactions can also occur (Deno, 1964). The importance of secondary reactions in the present investigation was undoubtedly due to some extent a t least to low initial concentrations of isobutane in the acid phase which resulted in the rather low quality alkylate produced. Concentrations were low because of the low residence times of the emulsion in the reactor. Furthermore, the quality of agitation in the small reactors used here was not so good as desired. The secondary reactions probably occurred as isobutane was dissolved in the acid and then reacted with the esters or other hydrocarbon materials dissolved in the acid. The quality of alkylate produced thus depends to a large extent on the initial method of bringing the reactants in con-
tact with the acid. Good agitation and high ratios of isobutane to olefin are obvious ways of increasing the dissolved ratios of isobutane to olefin in the acid phase. At least in the early stages of the reaction, the quality of the alkylate often improved with time. Such an increase would, however, not go on indefinitely, since there are other secondary reactions such as the degradation of trimethylpentanes (Hofmann, 1964). There must then be an optimum time of contact for the alkylate and the acid, depending on the choice of olefin as well as the operating conditions. Isobutylene was found in the staged reactor to react most rapidly of the three butenrs; hence, its optimum contact time was shortest. 1Butene reacted most slowly. The relative rates of solution of the butenes may be the important factor in explaining these observed differences. Isobutylene is dissolved most rapidly and 1-butene most slowly (Davis, 1928). Because of the high rates of isobutylene solubility, it tends to have higher dissolved concentrations in the acid compared to the other two C1 olefins. Hence, isobutylene is more apt to react in undesired primary reactions. The source of dimethylhexanes, undesirable products because of their low octane numbers, has been a subject of much debate (Phillips, 1965). I n the present investigation, the dimethylhexanes were formed largely in the primary reactions when 2-butene and probably isobutylene were used. Thus, the mechanism for the formation may be the one suggested earlier (Hofmann and Schriesheim, 1962) and shown below for isobutylene :
R+
+ C=C-C 1
C
+ R-H
+ (C=C-C+ I
j
C
C
-
C+-C=C)
C
The basic step is the abstraction of a hydride ion from the allylic position of an olefin molecule in the absence of isobutane (as is the case in the primary reactions). The resulting unsaturated cation then adds to another olefin, yielding the skeleton of a dimethylhexane molecule which can then become saturated through hydrogen transfer reactions. This unsaturated Cs cation may also be a source of acidsoluble hydrocarbons which dilute the acid. Ring closure to form a C j ring may occur between the positive ion on one end of the CS ion and the unsaturated group on the other end. In fact, cyclopentenyl cations have been reported to be important in the acid-soluble portion of the hydrocarbons produced when C4 olefins are in contact with strong sulfuric acid (Deno, 1964; Den0 et al., 1964). These latter cations are supposedly stable. Of interest, conjunct polymers (red oil) present in used sulfuric acid contain large numbers of Cs ring compounds (Miron and Lee, 1963). The results of this investigation clearly indicate that a t least some alkylate compounds are produced by the following two routes: From olefins in the absence of isobutane as indicated in Table V I I I . Bv secondarv reactions involving isobutane and acid-soluble organic materials in the absence ofuolefins. The above methods of forming trimethylpentanes are not clearly explained by the generally accepted mechanisms which indicate that the main reactions involve both isobutane and olefins. A mechanism has been suggested for the production of trimethylpentanes in the early stages of the reaction from only
olefins (Hofmann and Schriesheim, 1962). Such a process requires hydrogenation-dehydrogenation steps to saturate part of the hydrocarbons and dehydrogenate the remainder. I t is undoubtedly this latter portion of hydrocarbons which is acidsoluble and which in part a t least enters into the secondary reactions as the isobutane dissolves into the acid phase. The products, primarily trimethylpentanes, formed during secondary reactions, are not produced in any significant amounts by destructive alkylation. If this mechanism were important, there would be a net decrease in the amounts of lighi ends or especially heavy ends with time in the reactor. No such decrease, however, occurred (Table V I I ) . The secondary reactions appear instead to be related to esters or red oil complexes which are formed in part by reactions of the olefins with the acid. Reactions involving isobutane and acid and including hydride and proton transfer steps also may lead to acid-soluble hydrocarbon compounds. I n summary, the alkylation mechanisms proposed to date do not explain important aspects of the reactions noted in this investigation. Propylene as Olefin Feed
Higher alkylation temperatures are the key to successful alkylation of pure propylene. However. propylene can also be alkylated a t the lower temperatures normally employed for butenes if it is mixed with butenes. How much butene is required is not yet known, hoxvever. Lower reaction temperatures are desirable for olefin mixtures containing 50% butenes or above in order to produce high quality product. The C4 olefins in a mixture evidently act as an initiator for the alkylation of the propylene. There are a t least two ways in which the C4 olefins may affect the chemistry. Perhaps the propyl cation is not able to abstract a hydride ion rapidly from a molecule of isobutane or other hydrocarbon in order to initiate the reaction and to form tert-butyl cations. Butenes, however, form these cations easily. Another possible explanation may be that the propylene forms a n ester which is fairly stable, breaking down only slowly to allow alkylation. The butenes may promote the breakdown of propyl esters. Acknowledgment
The authors are grateful for the financial support provided by the American Oil Co., National Science Foundation, and Purdue University. R . M. Phillips, Procter and Gamble Co., offered valuable advice. literature Cited
Albright, L. F., Chem. Eng., 73, N o . 14, 119, No. 17, 143, No. 19, 205, No. 21, 209 (1966). Cupit, C. R., Gwyn, J. E., Jernigan, E. C., PetrolChem. Engr. 33,47 (1961); 34,49 (1962). Davis. H. S.. J . A m . Chem. SOC. 50,2780 (1928). Deno,’N. C.; Chem. Eng. .Terns 42,‘88 (Oct. 5 , 1964). Deno, N. C., Boyd, D. B., Hodge, J. D., Pittman, C. U., Turner, J. O., J.A m . Chem. SOC. 86, 1745 (1964). Goldsby, A. R., U.S. Patent 3,227,774, 3,227,775, 3,234,301 (1966). Goldsby, A. R., Gross, H. H., U. S. Patent 3,083,247 (1963). Hofmann, J. E., J . Org. Chem. 29, 3627 (1964). Hofmann, J . E., Schriesheim, A,, J . A m . Chem. SOC.84, 953, 957 (1962). Ipatieff, U. N., Schmerling, L., Aduan. Catabsis 1, Chap. 2 (1948). Jernigan, E. C., Gwyn, J. E., Claridge, E. L., Chem. Eng. Progr. 61,No. 11, 94 (1965). Mayer, I., U.S. Patent 3,109,042 (1963). Miron, S., Lee, R. J., J . Chem. Eng. Data 8 , 150 (1963). Mojby, J. F., Ph.D. thesis, Purdue University, 1964. Mosby, J. F., Albright, L. F., IND.END.CHEM.PROD.RES. DEVELOP. 5 , 183 (1966). VOL. 8
NO. 1
JANUARY 1969
97
Naworski, J. S., Jr., Ph.D. thesis, Cornel1 University, 1966. Phillips, R. M., M.S. thesis, Purdue University, 1965. Schmerling, L., in “Chemistry of Petroleum Hydrocarbons,” Vol. 111, p. 363, B. T. Brooks et al., eds., Reinhold, New York, 4
Shlegeris, R. J., Ph.D. thesis, Purdue University, 1967. Zimmerman, C. A., Kelly, J. T., Dean, J. C., IND.END. CHEM. PROD.RES.DEVELOP. 1, 124 (1962).
n cc
1723.
Schmerling, L., Znd. Eng. Chem. 45, 1447 (1953). Schmerlinc. L.. J . A m . Chem. Sac. 67. 1778 (1945’3. Schmerling; L.; in “Friedel-Crafts and Related’Reactions. Vol. 11. Alkylation and Related Reactions,” G. A. Olah, ed., Interscience, New York, 1964.
RECEIVED for review March 21, 1968 ACCEPTED August 29, 1968 Division of Petroleum Chemistry, 156th Meeting, ACS, Atlantic City, N. J., September 1968.
HYDROCRACKING OF GAS OIL S. A. Q A D E R
A N D G.
R. H I L L
De$artment of Fuels Engineering, University of Utah, Salt Lake City, Utah 84112
Gas oil boiling in the range 300” to 430’ C. was hydrocracked in a conventional continuous fixed-bed tubular flow reactor over a dual-functional catalyst. Gasoline of almost the same composition was produced in yields of 60, 82, and 58% in the single-pass, double-pass, and recycle operations, respectively, with diesel oil of 50 diesel index as a by-product. Hydrocracking of gas oil proceeds through a mechanism involving a combination of simultaneous and consecutive bond-breaking reactions followed by isomerization and hydrogenation of the products. The over-all kinetics observed indicated that gas oil hydrocracking, desulfurization, and denitrogenation reactions are all first-order and the rate constants can be represented hr.-’, k, = 0.681 4 X 1 O5 e-16’a00/RT hr.-I, and k, = 0.8253 X 1 O5 e-17,400’RT by: k, = 1 X 10’ e--21J00/RT hr.-I The cracking reactions involving the breakage of C-C, C-S, and C-N bonds on the acidic sites of the catalyst are rate-determining.
is being developed and practiced in the petroleum industry for converting different types of charge stocks to middle distillates, gasoline, or liquefied petroleum gases. The incentives of getting a variety of valuable Products from inferior grade feed stocks led to the development of a number of large-scale processes (Craig and Forster, 1966; Duir, 1967; Prescott, 1966) in the recent past. However, detailed experimentation was not done to explore the mechanisms and product distributions involved in the hydrocracking of various petroleum fractions, although some work was reported on the reaction mechanisms of some pure hydrocarbons. Flinn et al. (1960) and Archibald et al. (1960) studied the product distribution and reaction mechanisms in the hydrocracking of some pure hydrocarbons over dual-functional catalysts. Products rich in isoparaffins were formed because of a mechanism combining rapid isomerization with cracking and subsequent hydrogenation of the olefinic fragments. Sullivan et ai. (1964) found considerable cyclization of the side chains \\-hen some aromatic hydrocarbons were hydrocracked over a nickel sulfide catalyst. The products were found to be predominantly Tetralins and indanes of lower molecular lzeight than that of the reactant. lVyers et al. (1962) and Larson et al. (1962) investigated the performance of several catalysts for hydrocracking gas oil and furnace oil fractions and reported that the acid sites crack and isomerize olefin intermediates and the products get saturated at the hydrogenation sites. Kozlowski et al. (1962) hydrocracked gas oils boiling in the range 300’ to 450” C. and obtained kerosine-type jet fuel with freezing points below -15.5’ C. in yields of about 20 to 40%. I n the present communication, the results of the hydrocracking of a gas oil in a continuous fixed-bed bench scale YDROCRACKING
98
I&EC PROCESS DESIGN A N D DEVELOPMENT
reactor over a dual-functional catalyst carried out as a part of the program on the processing of mixtures of coal-derived liquids and petroleum fractions are reported. The influence of process conditions on product distribution is discussed and a kinetic evaluation of the data is presented. Experimental
Materials. A gas oil fraction (Table I) was used as the feed. The catalyst (commercial) contained 6% nickel and 19% tungsten, both as sulfides, supported on silica-alumina in the form of pellets of 0.083-inch diameter and 0.125-inch height, with a surface area of 212 sq. meters per gram. 5-A molecular sieves were of chromatographic grade. Active carbon was used as adsorbent. Equipment. T h e hydrotreating unit (Figure 1) consisted of a vertical tubular stainless steel reactor of 0.75-inch inside diameter and 40-inch length with extensive means for controlling temperature, pressure, and gas and liquid flow rates. T h e reactor was heated uniformly by a tubular ceramic furnace of 1.5-inch inside diameter and 38-inch length, which was well insulated. The first 20-inch length of the reactor from the top was packed with ceramic beads of 0.17-inch diameter, the next 6.5 inches with the catalyst (60 cc.), and the following 12 inches again with ceramic beads. The top bed of ceramic beads acts as the preheating zone. The temperature a t the center of the catalyst bed was measured with a thermocouple placed between the reactor and furnace walls. Temperature measurements at several points along the reactor tube indicated that the difference in temperature inside the reactor and the space between the reactor and furnace walls a t each point was less than 1” C. when the temperature was controlled for one hour or longer. The temperature of the catalyst bed was maintained constant throughout. The hydrogen supply was taken from a hydrogen cylinder with a maximum pressure of 2300 p.s.i. Procedure for Hydrocracking Experiments. T h e equip ment was first flushed with hydrogen to remove air, pressurized,
