Computation of Distillat ion Apparatus for

Hydrocarbon Mixtures
AND K. L. GEDDES,Standard Oil Company (Indiana), Whiting, I n d
E. W. THIELE

-HE d e v e l o p m e n t or A method of cornpulatiori i.s presented for the PRlNCIPLE O F METHOD

. I
formance of a proposed colukn, for all components is assumed. The number of content of either of the con-
so t h a t the d e r i e n of euuiw
. . romnonenls
~~

1~
may be pither f i d e or itrfinite. s t l t n e n t s in the vapor rking
ment, so far as reflux ratio and p he method is &tended lo c&es ill which from a plate, provided the con-
number of plates are concerned, centration of that constituent
streams are druwn from the column, or the feed in l i q u i d on tile plate is
has depended on e x p e r i e n c e
Work been is admitted ut more than one point. k n o w n or vice v e r s a . Like-
duited in this laboratory for a wise a material balance makes
number of years with a view t,o supplying the deficiency, and ponsible t.he computation of the concentration of either con-
t,he metlrod diicli has been developed is described in this stituent on a plate from its concentration in the vapors
paper. It makes possible the solution of such problems rising from the plate below, provided that the liquid and vapor
with a minimum of simplifying assumptions and witliout the quantities, and the concentration of the constituent in tlie
expenditure of an unreasonable amount of labor. It has product, are known. Hence, when product compositions
actually been used with success to attack a number of different have been assumed, it is possible to pass from plate to plate,
column problems. computing liquid and vapor compositions successively. I n
That the need for a method has been felt byothorsisshown the case of mixtures containing more than three components,
by the recent appearance of a number of articles on the sub- tlie equations based on the material balance are still valid
ject (2, 9, 4,8, 12). These methods, as well as tlie present but two difficulties arise. On tlie one hand, the concentra-
one, all agree in assuming that Raoults law or some modifioa- tion of B constituent in a vapor is not fixed m.heu its eoncentra-
tion of it is sufficiently valid for hydrocarbon mixtures, and tion in the corresponding liquid is known; the concentrations
that the plate efficiency is the same for all constituents. The of all the other constituents of the liquid must also be k n o m .
present method is believed, however, to have several ad- On the other hand, tlie composition of the products cannot be
vantages over these various methods: (1) the computations arbitrarily chosen or assumed, because out of all the con-
are comparatively straightforward; (2) no further simplifying ceivable combinations of the constituents only a limited
sssumptions are made; (3) it is worked out for mixtures in number can actually occur as products of a separation in a
which individual constituents are not distinguishable; (4) single column.
it is worked out for columns of any complexity. The method here proposed to overcome these difficulties is
289
290 INDUSTRIAL AND ESGINEERING CHEMISTRY
to fix in the first instance both the number of theoretical plates
in the column, as well as the heat input, and to assume tenta-
tively the temperature on each plate, since the composition
of the products is the thing to be determined. This a t once
makes possible a calculation of the concentration of each
constituent in the vapor in equilibrium with a liquid when the
concentration of that constituent in the liquid is known,
or vice versa, provided that Raoults law can be assumed
throughout. Since the composition of the products is not
assumed but is to be found, it might appear that the material
1 equivalent hydrocarbons. The steps are drawn so as to
4 (I.
-2 13))1 PO3
15,700 GAL.
806 MOLES
53.44A.ll.
ro0,ooo La. 1- LJSS]
( I275 1079 ) 103
LO10 GAL.
196, M O L E S
28,579.000 8.T.U. 31 A.P.I.
FIGURE1. COLUMN
WITH REBOILER
balance equations could not be used. It turns out, however,
that they can still be used to find the ratios of the concentra-
tions of the constituents in the overhead and the bottoms,
and this, combined with the known concentration in the feed,
is sufficient to fix values of the concentration of each con-
stituent in each liquid and vapor up and down the column.
The sum of the concentrations thus found for the constituents
of the various vapors and liquids will not, however, be equal to
unity unless the original tentative assumptions as to the plate
temperatures have been correct. The deviations from unity
which are found when the calculation has been completed
make possible a more correct assumption of temperatures for
another trial, and so on.
I n making heat balances to determine liquid and vapor
quantities, the gravities of the various liquids and vapors
have to be assumed, but accuracy in these assumptions is
relatively unimportant.
The composition of the feed is given by a true boiling point
analysis. Theoretically, a perfect fractionation should be
obtained, but for practical purposes a good column fractiona-
tion is sufficient. I n this investigation, the least efficient
equipment which has been used with satisfaction comprises a
94-cm. vacuum-jacketed column (16 mm. internal diameter),
packed with 5.5-mm. glass beads, a t a reflux ratio of 10 to 1,
when a charge of about one liter is used. It is obvious that
the labor of obtaining an exceedingly close fractionation of a
proposed feed to a column would be unwarranted in view
of the normal variation of composition of feed to a practical
column in the refinery. For the light ends of gasoline a
Podbielniak analysis is suitable. The resulting curve of
percentage off against true boiling point may be treated
either (I) as equivalent to a mixture of a moderate number
Vol. 25, No. Y
of pure hydrocarbons of definite boiling point, (2) as com-
posed of an infinite number of constituents, or (3) as composed
of certain definite components in the lower range of boiling
points, and of an infinite number in the upper range. The
first assumption will be used for the most part; the necessary
alterations by which the others may be used will be explained.
EXAMPLE OF COLUMN COMPUTATION METHOD
The method will be illustrated by application to a simple
reboiling column operated a t atmospheric pressure, without
steam. The column is shown diagrammatically in Figure 1;
it consists of two perfect plates above and one below the
feed, with a reboiling still, and reflux returned a t the boiling
point.
The composition of the feed, a treated pressure distillate,
is shown in Figure 2, the continuous line representing the
actual true boiling point analysis, and the steps the assumed
leave approximately equal areas on each side of the actual
line. The assumption of nineteen constituents is considered
sufficient in this case. The feed enters the column as a liquid
a t 100 F.
The feed of quantity as given in Figure 1 is assumed to go
to overhead and bottoms of the indicated gravities in the
0.2543
proportions shown. The heat input to the reboiler is 28,579,-
000 B. t. u. per 15,700 gallons (100,000 pounds) of feed.
The problem is to determine the composition of the two
0.8470
products under these conditions. I n column 1 of Figure 1
are given the plate temperatures assumed; column 2 gives
the assumed gravities of the liquids on the plates above the
feed, and the liquid gravities of the vapors below the feed.
0
0 10 20 30 4 0 5 0 60 70 80 90
WEIGHT PERCENT DISTILLED
FIGURE2. TRUEBOILING
ANALYSIS
OF TREATED
PRESSURE
DISTILLATE
These quantities, from which all the other quantities may be
derived, are encircled or underlined to distinguish them from
the others. (Sssumption of gravities and temperatures are
sufficient to permit the molal quantities within the column
to be calculated from heat balances, since the total heat of a
stock may be expressed in terms of the temperature and
gravity, and the molecular volume may be expressed as a
function of the gravity.) Study will show that these quanti-
ties may be differentiated into two classes, which will be
termed primary assumptions and trial values. The
primary assumptions, shown underlined in the figure, are
connected with variables, such as number of plates above and
below the feed, heat input (or reflux ratio), and split of the
feed in quantity between overhead and bottoms, which may
&larch, 1933 INDUSTRIAL AKD ENGINEERISG CHEMISTRY 291

be given any desired arbitrary values (within limits). The

trial values, shown encircled, are connected with variables,
such as temperatures and gravities, which acquire fixed al-
though as yet unknown values as soon as the primary as-
sumptions are settled. Thus, any column problem reduces
itself to the fixing of the primary assumptions and then ad-
justment by trial and error of the trial values until the
correct values of the latter have been determined, the final
result being to establish the product composition and the true
values of the quantities for which trial values are required,
for the given values of the primary assumptions.
Relationships between molecular weights, boiling pointy,
gravities, vapor pressures, total heatq, etc., are taken largely
from published sources (1,i,6,7',13). The results of the heat
balances are given in Figure 1: within the column are shon n
the calculated liquid and vapor molal flows from plate to platc
The drawing up of heat balances to determine the molal
quantities within the column is made easier by considering
the liquid and vapor streams within the column in the follon-
ing JTay, which is most conveniently explained by referenre
to Figure 3. I n that part of the column above the feed plate,
the vapor rising from a plate is conceived of a5 consisting
of two parts, one equal in quantity and composition (but not
in temperature) to the net overhead product, the other
equal in quantity and composition to the liquid overflow from
the next plate above. This latter is a quantity which,
so far as the heat balance is concerned, may be considered a.
rising from a plate as vapor to the next plate above where
it is condensed and returned as liquid reflux, "circulating,"
so to speak, between the two plates. Below the feed plate,
in a similar manner, the liquid overflow from a plate is con-
ceived of as consisting of two parts, one of the quantity
and composition of the bottoms, the other of quantity and
composition of the vapors rising t o the plate from the plate
below (here the reboiler).
An example of the heat balance for determining the molal
quantities between plates 2 and 3 is as follows: Taking that
part of the column below a plane drawn between plates 2 and 3
(all total heats are per gallon with 0" F. as base temperature),
we h R ve :
Heat entering:
Feed (53.4" A. P. I. and 100' F.) =
15,700 X 303 4,755,000 B. t. u.
Reboiling heat 28,579,000
33,334,000
Heat leaving:
Bottoms (37" A. P. I. and 435" F.) =
5010 X 1676 8,395,000
Product (62.5" A. P. I. and 355' F.) =
10.690 X 1913 20.460.000
I ,

28,855,000
Heat leaving section due to vaporization
of overflow from plate 3: 4,479,000
. (42.5"
Each gallon vaDor . A. P.I. and
355' F.) = 2060 B. t. u.
Each gallon liquid (42.5' A. P. I. and
322" F.) = 1149
Net heat leaving per gallon of reflux = 911
Gallons of overflow = 4,479,000
____
R11
= 4920
Moles of overflow = 4920-2-8. 33
= 212
193
Here 193 is the molecular volume corresponding to the as-
sumed gravity of the liquid on plate 3, and 8.33 is the i\eight
in pounds of one gallon of water a t 60" F. Hence the total
moles of vapor rising to plate 3 will be 610 +
212 = 822.
The details of the first trial computation are given in Table
I; the nomenclature used in the subsequent discussion is as
folloTvs:
292 I N D U S T R I A L A N D E:N G I N E E R I N G C H E M I S T R Y Vol. 25, No. 3
n = subscripts for plates, numbered from bottom upward
(reboiler = 0)
Ff = Aa + Bb
v, = vapor risin from late n, moles Bb = Ff
0, = liquid overlowing From plate n to plate (n - I), moles
2, = mole fraction of constituent in liquid on plate n
Yn = mole fraction of constituent in vapors above plate n
Yn' = mole fraction of constituent in va or above plate n t o
F is determined by the time unit, and f is based upon the
which feed vapors have been adzed feed analysis (Fffor each constituent is given in column 13,
P, = vapor pressure of pure constituent a t temperature of Table I). Hence Bb, and Aa by difference from Ff,may be
plate n found (columns 12 and 14). By multiplying yo/b by Bb, By,
Pn = total pressure of all hydrocarbon vapors above plate n
F = total feed, moles is found (column 4). If the solution to the computation is
f = mole fraction of a constituent in total feed correct, obviously the calculated Zyo = 1, and ZByo = B .
F, = total vapor in feed, moles In other words, the sum of the values of 1 9 6 ~ 0(column 4)
Y = mole fraction of a constituent in feed vapors should equal 196 for the correct solution. The same applies
iL = total liquid in feed, moles
to 196b and 190y1, and similarly the summation of 6102,
2: mole fraction of a constituent in feed liquid
total overhead product (C, D . . . for side streams, etc.), and 6 1 0 should
~~ be 610.
moles
a = mole fraction of a constituent in overhead product
(c, d . . . for side streams, etc.)
B = total bottoms, moles
b = mole fraction of a constituent in bottoms
w = weight of a constituent or fraction
v = volume of a constituent or fraction

Considering any one component (for concreteness, the con-

stituent whose normal boiling point is 207" F.), then by
definition the mole fraction in the bottoms (or in the re- FEED
boiler) is b. By Raoult's and Dalton's laws:
gob = P ~ P , (1)
A material balance over the part of the still below plate 1
gives
01x1 = VOVO Bb +
\ /
'\ /

xl/b = vo (y_ - 1) + I
- 7- /'
'-
H EIA T
01 b
FIGURE3. FLOWOF PRODUCTS
AND
so that xl/b may be calculated from yo/b. The last form, REFLUX
Equation 2, is most convenient for slide rule computation.
By a repetition of Equation 1 with changed subscripts: The deviations of the summations from the correct values
show the errors in the assumptions as t o plate temperatures.
P P
y1 = -'
P
x1 or yl/b = f (xl/b)
P
(3) The gravity assumptions may also be changed. Since a
change in any one quantity affects all the others to some
extent, corrections require much judgment, but one or two
so that yl/b may be found when xl/b is known. repetitions of the calculation will generally give a fairly
I n Table I, columns 3, 5, and 8 give the results of these correct solution to one familiar with the process. Even if
computations; p is in all cases 760 mm. Starting a t the top the first trial is quite inaccurate, it will give a fairly good
of the column, and assuming the vapor above the top plate idea of the sharpness of fractionation obtainable with a
to be of the same composition (y3) as the overhead product, given heat input and number of plates, and will often show
a: that, in order to obtain the desired fractionation, it would
be better to change the primary assumptions (assumed num-
(4) ber of plates, heat, input, or proportion of overhead), rather
than to attempt a more accurate solution for the product
and by a material balance of the vapor and liquid passing composition, on the basis of the original assumptions, by
through a plane just below plate 3: altering the trial values.
If open steam is used in the column, it must be considered
VZYZ= A a + OSXS in the heat balances, and the value of p , which includes only
the hydrocarbon vapors, will be different in different parts
yda =
Os ( x d a - 1)
vz +1 (5) of the column, depending upon the moles of hydrocarbon
vapors present.
so that y,/a may be found from x3/a. The last columns of Table I show how the gravity of the
By an equation similar to (l), x,/a may obviously be bottoms may be computed from the known molecular weights
found from y2/a; and by an equation similar to (5), yl/a and molecular volumes (liquid volume of one mole a t 60" F.)
may be found from a / a (columns 19, 18, 15, and 9 of of the constituents. Multiplication of column 12 by columns
Table I). 22 and 24 gives columns 23 and 25, respectively; the summa-
Division of yl/b by yl/a gives a/b (column 10). tion of column 23 divided by the summation of column 25
nives the sDecific a a v i t v a t 60" F.
I For the lightest constituents the plate temperature is above the critical.
Experience indicates that the vapor pressure curves may be extrapolated - The true boil& poini analysesof the products will usually
for a considerable distance above the critical for in calculations in- have to be interpreted in terms Of practical tests* Dew points
volving Raoult's law. and bubble points are obtained very easily; flash points de-
 March, 1933 I N D U S T R I A 1, A N D E N G I N E E R I N G C H E M I S T R Y 293

pend on vapor pressure, and viscosities can doubtless be 15 per cent off by weight). The total will have the same
worked out as a special case of blending oils, though not significance as the sums in Table I.
much has been done along these lines as yet. The A. S. T. M. It is not convenient to use the continuous method a t the
distillation, however, is not determinable a t present from the lower end of the true boiling scale, as the humps in the slope
true boiling point curve, and there is need for information curve make it difficult to compute and draw properly derived
along this line (see, however, Lewis and Robinson, 9). I n curves for integration.
some studies made in this laboratory, gasoline has been
defined by its dew point; that is, the cut between gasoline CORRECTIONS TO RAOULTS Lam
, and the next higher boiling product has been so made that The ease with which corrections to Raoults law could be
incorporated in the method obviously depends on the char-
acter of the correction. The only definite proposal for such a
correction with which the authors are acquainted is that of
040 ,6020 - Souders, Selheimer, Smith, and Brown (11) to use fugacities
of the pure components a t the given temperature and pres-
sure instead of the vapor pressures. I n effect, this is equiva-
lent to assuming a different set of vapor pressures for each
030 Mi5 - value of the total pressure on the column; once the necessary
tables or charts to carry this out are available, there should
q$
_I be no increase in difficulty of calculation.
020 .0010 - FEEDPARTLY OR TOTALLY VAPORIZED
If the feed enters between two plates partially vaporized,
>< =a it is reasonable to assume as a close approximation that the
liquid portion mingles, without boiling, with the reflux from
010 .ooo,-
- the plate above, whereas the vapor portion of the feed mingles
without condensation with the vapors rising to the plate
above. Suppose the feed enters between plates 1 and 2, then
the computation may be carried down the column t o give
On On L-
2;o 300 400 500 gl/a, the composition of the vapors rising to plate 2, and
up the column from the bottom to give yl/b, the composition
of the vapors rising from plate 1. However, yII and yl are
not the same. The derivation of the equation for Bb in this
case is as follows: A material balance over the whole still
gives
Aa + B6 = Ff (7)
By definition,

Substituting from Equation 7 ,

Rearranging and collecting terms,

This equation seems complex, but several of the quantities are

the same for all constituents, and it has been found practical
to evaluate the equation. If the feed is all vapor, F , = F and
yj = f. In case the feed is all liquid, Equation 8 reduces to
Equation 6.
The use of Equation 8 may be avoided (without any great
inaccuracy) in many cases by assuming, in setting u p the heat
balances, that the feed enters the feed plate as a liquid, but
possesses total heat equal to that actually possessed by the
partially vaporized feed. I n this case Equation 6 is used for
evaluation of a and b,
A method of computing pipe still curves, such as are
needed in determining the feed vaporization, has been pre-
sented by Murray (10). He assumes FL and sums, for all
constituents,
Ff
F PP + F L ( l -;)
2 94 INDUSTRIAL AND ENGINEERING CHEMISTRY
This summation should equal unity for a correct solution.
It has been found shorter for slide rule work to assume
F./FL and find the summation for all constituents of
which equals F,. When Fw/(F - F,) from the summation
turns out the same as F ~ / F assumed,
L the solution is correct.
[Lewis and Matheson (8) give still another method.] Where
many constituents are assumed, it is important to arrange all
work so that the least effort is involved.
STRIPPERS
SIDE-STREAM
The discussion of this case refers to Figure 5. The plate
Vol. 25, No. 3
Hence ys/d may be found. From this point on, the calcula-
tion is just the same as for a simple column. When the feed
plate is reached, d and b are evaluated; since d / a and c / a
are known, a and c can be found and the solution completed.
It is evident that any number of side-stream strippers can
be taken care of.in the same way. The solution of a column
with side-stream strippers is somewhat more complicated
than that of a single column having as many plates as the
main column and strippers combined.
FEEDAT Two POINTS
Reference is made to Figure 6 for the arrangement under
discussion. Working up from the bottom, yl/b may be
evaluated in the usual way, or y1 = klb. (In the following

6Aa
numbering includes the side-stream stripper. The arbi-
trarily assumed moles of side product (a primary assumption)
are C, and the concentration of a given constituent in this
side product, c (as yet unknown).
Working down the column in the usual way, xd/u is found;
working u p the stripper, ys/c is found. By a material balance
around the stripper,
cc + V8V8 = (C + VS)XC
c + vs + -
x,/c = - C- V8 ,
or
c +va
c / a = 54la
-
x4lc
Defining two new quantities, D and d, by the equations:
D = A + C (11)
Dd = Aa +
Cc (12)

Bb
FIGURE
6. COLUMNWITH DOUBLE
FEED

derivation k denotes a quantity, the numerical value of which

I -51
has been evaluated a t a stage where i t is introduced.) Taking
that part of the column below a plane between plates 1 and 2,

or V
x2 = -'yl + -Bb - -
FLX/ v + B- b - FLQ
--'klb -
0 2 0 2 0 2 020 2Oa

Making a material balance on that part of the still below a

plane between plates 2 and 3 gives:

HEAT or 2 3
v 2
= - y2
Os
+ -oBs b - -
FLX~
0 3
FIGURE
5. C O L U M N WITH S I D E - S T R E A M
STRIPPER
Substituting from Equation 16 and collecting terms,
That is, the overhead and side product are treated as one
product. From Equation 12:
or
Dd = A a + Ca.
Working down from the top of the column, in the usual way,
d/a =
A
-
D
+--.
C
D (:)
there results:
y3' = kaa
Making a material balance in the section of the column
(including stripper) just above plate 3, By a material balance at the upper feed point,
March, 1933 INDUSTRIAL AND ENGINEERIXG CHEMISTRY 295

By ,Raoult's law, and substituting from Equation 18, LITERATURE

CITED
z3 = 23 y 3 = kna
Pa
k12 + (19)
(1) Bahlke and Kay, ISD.ENQ.CHEM.,21, 942 (1929).
(2) Brown, Souders, and Syland, Ibid., 24, 522 (1932).
(3) Cope and Lewis, Ibid., 24, 498 (1932).
Equat,ing (17) and (19), (4) Fenske, ISD. EXG.CHEM.,24, 482 (1932).

++
ksb +
k7 = kiia kiz
Ff = Aa Bb
Hence, there are two linear equations which can be solved
for a and b, and the whole solution of the column follows.
The labor of computation per plate between the two feed
points is a t least twice as great as in ordinary cases.
Taking these methods together, it appears that the com-
plete solution of any column, no matter how complex, is
possible to a degree of accuracy which is limited only by the
available knowledge of the equilibrium relationships and of
the plate efficiencies.
(20)
(5) FitzSimons and Bahlke, Am. Pet. Inst., Bull. 9, S o . 1, Sect.
111, 70-2 (1930).
( 6 ) Fortsch and Whitman, Ibid., 18, 796 (1920).
(7) Leslie, Geniesse, Legatski, and Jagrowski, Ibid.,18, 45 (1926).
(8) Lewis and Matheson, Ibid., 24, 494 (1932).
(9) Lewis and Robinson, Ibid., 14, 481 (1922).
(10) Murray, Ibid., 21, 917 (1929).
(11) Selheimer, Souders, Smith, and Brown, Ibid., 24, 515 (1932).
(12) Souders and Brown, Ibid., 24, 519 (1932).
(13) TTilson and Bahlke, Ibid., 16, 115 (1924).
RECEIVEDAugust 16, 1932. Presented before the Division of Petroleum
Chemistry a t the 84th Meeting of the -4merican Chemical Society, Denver.
Colo., August 22 t o ?6, 1932.

Sulfur Compounds Derived from Petroleum

P. J. WIEZEVICH, L. B. TURNER,AND PERK. FROLICH

Standard Oil Development Co., Elizabeth, X. J.

(CgHsS)&(CH8)2 = CzHlSH + C*H6SC(CHa): CH,

C OXSIDERBBLE quantities of mercaptans and other
sulfur compounds are being produced a t various refin-
eries during the recovery of spent caustic obtained by
treating sour crudes. The recovery operation generally
distilling over a t 105-115" C. Oxidation of acetone diethyl
mercaptole (6)results in the formation of the well-known
compound, sulfonal (CH&C(S02C2H&.
consists in steaming the spent caustic, condensing out the
steam, and burning the mercaptans and other compounds
liberated (6). THIOETHERS
Table I (4) shows some of the physical properties of the
aliphatic mercaptans available in the literature. Although Sabatier and Mailhe (8) mention the use of
cadmium sulfide as a catalyst for the conversion of mercaptans
TABLE
I. PROPERTIES
OF ALIPHATICMERCAPTANS to thioethers, tests in this laboratory showed that the activity
MEBCAFTAN BOILINQPOINT PR~SEURE
DENSITY25'/4' of this compound alone was not sufficiently high for com-
c. Mm. mercial purposes. It was found that in presence of proper
Methyl 5.8-6.2 756 0.85991 catalysts, ethyl mercaptan can be converted into ethyl thio-
Ethyl 34.4-34,6 752 0.83147
n-Propyl 67.4-67.6 763 0.83572 ether with good yields, according to the following equation :
Isopropyl 52.3-52.5 753 0.80851
n-Butvl 98.6-99.0 768 0.83651
sec-B&yl 84.4-84.6 754 0.82459 2C2H6SH = CZH~SCYH)
f HzS
n-Amyl 126.3-126.5 759.5 0.83750
sec-Amyl ll?.5-112.6 753 0.82815
Using a cadmium sulfide-zinc sulfide-aluminum oxide catalyst
The caustic treatment removes mostly the lighter com- a t 350" C., a conversion of 45-55 per cent to ethyl thioether
pounds, the heavier ones usually being converted to disul- is obtained per pass, about 5 per cent of the mercaptan being
fides by means of doctor solution. A typical analysis of a lost in the formation of side products such as ethylene,
mercaptan mixture available a t one refinery is as follows: hydrogen, etc.
When the ethyl mercaptan fraction from petroleum is used,
hf ERCAPTaN a rapid poisoning of the catalyst results, so that it is necessary
%
' by weight to carry out a preliminary purification by contacting the
Methyl 30
Ethyl 53 vapors with active charcoal, followed by a simple distillation,
Propyl and higher 17 for removing the undesirable heavy ends. If an aluminum
reactor is used with such an arrangement, a relatively long
These compounds can be separated fairly well by fractiona- charcoal and catalyst life is obtained.
tion, and may be used as odorants, denaturants, pickling The boiling points of the various aliphatic thioethers are
inhibitors, and raw materials for the preparation of other shown in Table 11. Methyl thioether and ethyl thioether
products described below. Moreover, the heavier mer- dissolve cellulose nitrate in presence of ethanol. The solu-
captans form oil-soluble lead salts. tions compare favorably in various properties, such as vis-
Ethyl mercaptan reacts with acetone in the cold and in cosity, color, odor, and drying behavior, with such solvents
presence of anhydrous zinc chloride or hydrogen chloride to for cellulose nitrate as are now in actual use in industry.
produce acetone diethyl mercaptole (3) : Propyl thioether dissolves cellulose nitrate when mixed with
2CzHsSH + (CHa)zCO (CzHsS)zC(CHa)z + HzO methanol, as shown by the solubility chart in Table 111.
Another possible use for the thioethers is as rubber solvents.
When distilled a t atmospheric pressure, decomposition of this The three thioethers which have been examined (methyl,
product occurs at about SO" C., the main compound formed ethyl, and n-propyl) dissolve rubber very satisfactorily.
probably being propylene ethyl thioether : The boiling points of the ethyl and n-propyl ethers are higher