INDUSTRIAL AND ENGINEERING CHEMISTRY
998
Vol. 21, No. 11
The Manufacture of Anhydrous Ethyl Alcohol‘ D. B. Keyes UNIVERSITY OF ILLINOIS,
NHYDROUS ethyl alcohol is known in this country as absolute alcohol. The manufacture of this chemical from 95 per cent (by volume) ethyl alcohol by modern distillation methods was first accomplished on a commercial scale in the United States. I n the last three or four years several foreign writers have seen fit to give all the credit of this interesting and valuable development to various French inventors. It is the object of this article to give a brief resume of the important processes developed in this country and in France, together with the date of the patent applications, The United States is entitled to the honor of this development, and the facts are presented in order t o justify this statement.
A
URBANA, ILL.
dehydrated with lime, was used again in place of the pure benzene in the original batch. It can be easily seen that the process must be made continuous before it would become commercially successful.
r
Figure 2-Kubierschky’s First M e t h o d ; G e r m a n P a t e n t 287,897
Kubierschky’s Process [German Patent 287,897; application date, October 24, 1914;
issued October 11, 1915.1
Figure 1-Young’s
Process: 142,502
German Patent
Starting with Young’s process, a brief abstract of the important new features of each invention will be given, together with the patent number, date of application, and date of issuance.
Kubierschky showed two different methods. The 6rst is indicated in Figure 2. Distillation and fractionation took place in the dehydrating column on the right. This was equipped with a total condenser, but the condensate was sent entirely to a separator. This condensate was a ternary mixture, and separated into two layers, as indicated. The benzene layer from this separator was returned as reflux to the top of the dehydrating column. The residue from the bottom of the dehydrating column was the product, absolute alcohol. The water layer from the separator was sent to a recovery column mhich operated on direct steam, as indicated. The
Young’s Process2 (Batch) [German Patent 142,502; a p p l i c a t i o n date, October 17,1901; issued June 25,1903. ]
No drawing was supplied with this patent, but Figure 1 illustrates the features of the invention. An ordinary still and fractionating column, together with reflux condenser, was used. A charge consisting of 95 per cent alcohol, together with benzene, was placed in the still and heated indirectly by steam. The first distillate was collected in the left-hand receiver and consisted of a ternary mixture of alcohol, benzene, and water, which separated into two liquid layers as indicated. When the distillate changed from two phases to one phase, consisting now of a mixture of benzene and alcohol, it was run into the middle receiver. When the benzene no longer appeared in the distillate, the distillate consisted of absolute alcohol, and was run into the right-hand receiver. The benzene and alcohol mixture, together with the benzene layer in the left-hand receiver, which had been 1 Received August 17, 1929. Presented before the Division of I n dustrial and Engineering Chemistry at the 78th Meeting of the American Chemical Society, Minneapolis, Minn.. September 9 to 13, 1929. * Sidney Young, Professor in University of Dublin, Ireland.
Figure 3-Kubierschky’s Second Method; G e r m a n P a t e n t 287,897
excess water left the bottom of this column, and the concentrated alcohol left the top as vapor. The original feed of 95 per cent (or lower) alcohol was sent in as reflux to the top of the recovery column. The combined feed and recovered alcohol entered the side of the dehydrating column t~ vapor. Kubierschky’s second method is indicated in Figure 3. In
INDUSTRIAL A-VD EYGINEERING CHEMISTRY
November, 1929
H
999
ti J
Figure 4-Steffens’
Process; U. S. P a t e n t s 1,586,717 a n d 1,586,718
this case the 95 per cent alcohol was fed directly in the form of liquid to the dehydrating column. The recovery column had its own reflux condenser, and the vapors from this condenser mere returned to the feed plate of the dehydrating column. The chief difficulty with the Kubierschky process, as indicated in its patent, would be the contamination of the water layer by benzene. This would prevent the existence of 95 per cent alcohol a t the top of the recovery column. I n other words, there would be a greater percentage of water than is indicated, and the capacity of the entire outfit would be reduced. This increase in the amount of water in the vapor from the recovery column is caused by the benzene, which tends to form the ternary constant-boiling mixture containing 7 per cent water and 74 per cent alcohol.
Figure 5-Rodebush
Process; U. S. P a t e n t s 1,487,182 a n d 1,583,314
U. S. Patent 1,487,182; application date, April 14, 1921; issued March 18, 1924.1
The new features of this process, as shown in Figure 5 , are as follows: (1) The reflux to the dehydrating column is the ternary mixture (later used by Steffens). (2) Water is used in scrubbing the water layer, but no separate scrubber is used. (Ethyl acetate, the preferred “third” liquid, would not separate from the constant boiling ternary mixture without the addition of water.) (3) Alcohol liquid or vapor (95 per cent alcohol) was returned to the feed plate in the dehydrating column (later used by Steffens).
Steffens’ Process3 [Date of invention, September 29, 1919. U. S. Patent 1,586,717 (Process) ; U. S. Patent 1,586,718 (Apparatus) ; application date for both patents, April 5, 1921; issued June 1, 1926.1
The Rodebush process was invented prior to the Steffens’ process but is more complicated, so the Steffens’ process will be described first. The process, as shown in Figure 4,is essentially a modification of the Kubierschky process in order to make the latter a commercial success. There are four main features: (1) An additional separator is used to which water is continually added in order to scrub the water layer and give a second oil layer, the object to remove benzene from the water layer. (2) Liquid, 95 per cent alcohol, is returned to the feed plate of the dehydrating column instead of vapor. This is also a feature of the Rodebush process. (3) Part of the condensed ternary mixture is used as reflux for the dehydrating column. This is also a feature of the Rodebush process. (4) The oil layer from both separators is returned to the feed plate of the dehydrating column.
This process is of great commercial importance, as indicated by plant performance. Rodebush Process‘ [Date of invention, June 10, 1929. U. S. Patent 1,583,314 (Process); application date, April 14, 1921; issued May 4, 1926. 8
Foreign patents covering this process and apparatus appeared and
were issued about 1922.
‘Foreign patents covering this invention appeared and were issued about 1922.
n,!
t
Figure 6-Keyes
Process, Second Method
(4) The oil or benzene layer was fractionated in a separate column. The condensed ternary mixture was used partly as a reflux, and the rest returned to the separator. The “third” liquid was returned to the feed plate of the dehydrating column. (5) The water layer was fractionated in a recovery column; the ternary mixture condensed, part used for reflux, and part returned to the separator; 95 per cent alcohol was taken off near the top of the column and returned to the feed plate of the dehydrating column.
INDUSTRIAL A N D ENGINEERING CHEMISTRY
1000
Y
n
L
Vol. 21, No. 11
The second method is indicated in Figure 6. This is a modification of the Rodebush process. The new feaJ tures are as follows: (1) Two columns are used for the fractionation of the oil layer-the first to remove the ternary mixture, and t h e second to obtain a binary mixture consisting of 95 per cent alcohol. It is pointed out that carbon tetrachloride used as. the(21third” liquid will reverse the layers in the separators, and thereby change somewhat the diagram as indicated above. (3) A separate water scrubber was used, as indicated by Steffens.
This process is quite capable of handling a great variety of dehydrations, using many different substances for t h e “third” liquid. Although the original cost of the equipment will be higher than the Steffens or Rodebush process, the control is better and the capacities and efficiencies somewhat higher. Ricard-Allenet5 Process 4
[Cases l a , l b , and IC: British Patent 211,454 (Process); convention date (Belgium), February 15, 1923.]
Figure 7-Ricard-Allenet Process: British P a t e n t 211,454; Case la
The new features of case l a are as follows: Figure
(1) The water layer, as indicated by Figure 7, goes to the
8-Ricard-Allenet
Process; British Patent 211,- bottom of a column containing a solid dehydrating agentThis was the original United 454: Case IC States process that proved to for example, sodium carbonate. (2) Water and sodium carbonate come out of the bottom of be commercially siccessful. Absolute alcohol has been and the solid is recovered by distillation. manufactured on a commercial scale in this country since 1919, this column ( 3 ) The “third” liquid (butyl chloride) and alcohol, plus a without interruption, by the process whose basic features are little water, are returned t o the dehydrating column below the described above. feed plate.
Keyes Process [First Method: U. S. Patent 1,676,735 (Process) ; application date, December 27,1922; issued June 10, 1928. Second Method: application date, March 30, 1926.]
The new features of the first method are as follows: (1) Two separators were used, as indicated by Steffens, and the oil layer distilled, as indicated by Rodebush, but the pure “third” liquid was returned as reflux to the dehydrating column. (2) Some of the water coming off from the bottom of the recovery column was used for scrubbing purposes in the second separator.
Case l b is the same except that carbon tetrachloride is the “third” liquid. The oil layer, therefore, is on the bottom of both separators. Case ICis slightly different, as indicated by Figure 8. The new features are as follows: (1) The “third” liquid (in this case, ethyl acetate) does not separate, so the condensate passes directly to the bottom of the sodium carbonate column. Separation takes place as a result of a salting out effect in this column. The oil layer is returned to the dehydrating column below the feed plate. (2) .The 95 per cent alcohol feed is used as a reflux on the dehydrating column. (3) The inventors mention the distillation of each layer, but do not give any details.
These three cases are evidently attempts to improve, by the use of a solid dehydrating agent, the well-known Kubierschky process. The use of a solid dehydrating agent makes the process impractical from a commercial standpoint, owing to the high handling cost of this solid dehydrating agent. [Cases 2a, 2b, 26, and 2d: British Patent 214,581; convention date (Belgium), April 16, 1923.1
Cases 2a and 2c cover a process as indicated in Figure 9. The features are as follows: (1) The oil layer is returned as reflux to the dehydrating column. This feature was first used by Kubierschky. (2) The water layer is sent to a separate column and the ternary mixture (as vapor) is returned to the condenser of the dehydrating column. This feature was used both by Rodebush and Keyes. (3) The residue from the bottom of this recovery column is sent to a second recovery column. This was one of the features of the Keyes process. (4) Alcohol (95 per cent) is returned as vapor t o the feed plate of the recovery column. This feature was common to all previous processes mentioned.
Cases 2b and 2d cover a process which is indicated in Figure 10. The features of this process are as follows: Figure 9-Ricard-hllenet
Process: British P a t e n t 214,581 ; Cases 20 a n d 2c
5
Rirard, Allenet & Cie., France.
November, 1929
toor
INDUSTRIAL A N D ENGINEERING CHEMISTRY
Y
Figure 11-Ricard-Allenet
Figure 10-Ricard-Allenet Process; British P a t e n t 214,581 ; Case* 2 b a n d 2 d
(1) The process is very similar to case 2r The condensate, however, from the first recovery column is sent to an auxiliary separator. (2) The water layer from this separator, and also part of the oil layer, is returned as reflux to the first recovery column (3) The rest of the oil layer is added to the regular reflux from the dehydrating column.
These particular features are of no fundamental importance and their value on a commercial hook-up is questionable. [Cases 3a and 3b: British Patent 215,716; convention date (Belgium), May 9, 1923 1
i
Process; British P a t e n t 243,368; Case 46
regular reflux condenser and returned his oil layer to the middle OK base of the column.
[Cases 4a and 4b: British Patent 243,368; convention date (Belgium), November 20, 1924.1
Case 4a covers the dehydrat,ion and purification of impure pthyl alcohol-for example, alcohol denatured with methanol. The water is rectified and then the ternary mixture separated and scrubbed with water as in the Steffens process. The “heads” are further rectified and the %ils” are also rectified. Gase 4b is indicated in Figure 11. The features are as follows:
Case 3a is the process case of the apparatus cases 2a and 2b. Propyl alcohol is dehydrated and the “third” liquid is trichloret hylene. Case 3b is a special case for the dehydration of glycerol. Extra heating elements are used in the dehydrating column. Toluene is the “third” liquid. The water layer is thrown away and the oil layer is returned as reflux (as indicated by Ku bierschky).
(1) A separate separator is used which is apparently a water scrubber, and was first used by Rodebush. ( 2 ) The oil layer, as also part of the condensed ternary mixture, is returned as reflux to the dehydrating column. (3) The water layer has two recovery columns, a feature which has already been indicated by both Rodebush and Keyes. (4) The ternary mixture from the first recovery column is not returned to the separators, but is discarded, owing to the impurities, as indicated in the statement concerning case 4a.
Note-This case is very similar t o the Stevens U S Patent 1,304,232, i