New

Process for
Production of
Tetrahydrofuran
JUNlCHl KANEPAKA
TAISUKE ASANO
SHINOBU M A S A M U N E

A n e w one-stage, high-selectivity process

f o r manufacture of tetrahydrofuran
directly from maleic anhydride

etrahydrofuran (THF) is used mainly as a raw Another possible T H F process, using butadiene as a
Tmaterial for spandex fibers and polyurethan elas- raw material, was reported (2). The reaction scheme is
tomers, but also finds application as a solvent since represented as follows
T H F is a strong dissolving agent of both synthetic and
natural resins especially polyvinylchloride and vinyli-
denechloride copolymers. Recently, the demand of
T H F has grown rapidly with a n increased demand for
spandex fibers and polyvinylchloride.
At present, commercial production of T H F involves
either one of two methods: the furfural process or the
Reppe process. I n the former, furfural extracted from
corn husks is used as the raw material. And in the However, this process is not practical for obtaining high
latter process, acetylene and formaldehyde are used as yields of the product.
the raw materials. The new process of obtaining T H F from maleic
The major disadvantage of the furfural process is anhydride in one stage with high selectivity has been
the difficulty of ensuring a constant supply of furfural, developed. T h e reaction is expressed as follows
because the supply is always dependent on agricultural
conditions. Therefore, the trend has been to switch
from the furfural process to the Reppe process which is a + 5H2 -+ -t 2H20
fully synthetic method (7). 0

24 I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
 duced simultaneously in the wide range of T-BL to
T H F ratio.

New THF Process

I n this process, T H F can be produced continuously by
the direct hydrogenation of maleic anhydride. Before
the direct use of maleic anhydride, a number of catalysts
were investigated for this reaction ( 3 ) . They were
nickel molybdenum oxide, cobalt molybdenum oxide,
Raney nickel, etc. However, none of these catalysts
was adequate for industrial use. For example the
hydrogenation of maleic anhydride at 180 to 300 "C and
a t a pressure of 150 kg/cm2 for 2 to 10 hr using a nickel
catalyst leads to the formation of a considerable amount
of solid succinic acid, which gives the reaction mixture a
viscous, slurry-'like appearance. This means that the
yield of desired 7-BL and T H F is low, while operational
efficiency involved in the transportation of the reaction
products, separation of catalyst from reaction products
and reuse of catalyst suffers.
Moreover, in the direct hydrogenation of carboxylic
acid such as maleic anhydride, the reaction produces
dicarboxylic acids such as maleic acid or succinic acid
and lower organic acids which are by-produced con-
tinuously through the reaction along with water.
Maleic acid, especially, has a strong acidity. The pK,
values of these acids are shown in Table I. So, the
conventional hydrogenation catalyst is adulterated and
sometimes converts to the salt which dissolves into the
reactant solution. Thus, the catalytic activity is greatly
decreased in the course of reaction.
We have attempted to obtain a catalyst which yields
the reaction product in nonviscous liquid form, and
shows less deactivating tendency through the reaction
(4). Our newly developed catalyst proved to resist
the catalyst poisoning, and to keep continuously active
in the pilot plant operation.
Kinetic study in batch-type reaction. Because the
batch-type system is generally more convenient to in-
vestigate than the flow-type, the batch-type reaction
This new T H F process was developed on the basis of was applied for maleic anhydride, succinic anhydride,
the research carried out in Mitsubishi Petrochemical 7-BL, succinic acid and polyester. Except for maleic
Company Ltd. since 1963. The pilot plant designed to anhydride, all are the intermediate products, and the
take the engineering data has been operating successfully polyester is produced from succinic acid and 1,4-butan-
since January, 1968. The engineering facilities vital to diol through the condensation polymerization.
the process were developed jointly by the Mitsubishi Procedure. A conventional 300 ml autoclave pro-
Petrochemical Company Ltd. and Chiyoda Chemical vided with an electromagnetic stirrer was used. 100 g
Engineering and Construction Go. Ltd.
T h e main advantages of this process are :
Maleic anhydride is easily and inexpensively supplied
by oxidation of benzene or (&fraction.
Due to the simple process and single-stage reactor,
only a low capital investment is required.
TABLE I . pK, VALUES
High conversion and high selectivity : T h e selectivity
to THF is more than 90% and per-pass conversion of Organic acid Value of ljKa
maleic anhydride is 100%. Maleic acid 1.8
Succinic acid 4.1
High product purity of 99.95% or higher is easily
Propionic acid 4.9
assured.
Butyric acid 4.8
y-Butyrolactone (7-BL), which is the raw material of
2-pyrrolidone and n-methyl-2-pyrrolidone, can be pro-

VOL. 6 2 NO. 4 APRIL 1970 25

of each reactant was used and the amount of catalyst
was 0.05 wt% or 0.1 wtYG for maleic anhydride and
5 wt% for the other reactants. The reaction pressure 3,O
was 60 kg/cm2 for the former and 120 kg/cm2 for the
latter. The range of reaction temperature was from h
I
160" to 200°C for maleic anhydride, 240" to 270°C 4
I
for succinic anhydride, 240" to 280°C for 7-BL, and 2.0
260°C for succinic acid and polyester. To ascertain
the relationship between molecular weight and the
Y
U
I
reaction rate, the polyesters with mean molecular weight u
v

1000 and 2000 were examined. -

c 1.0
T o let the reaction proceed as slowly as possible before
reaching the prescribed reaction temperature, the auto-
clave with reactant and catalyst was kept a t 5 to 30
kg/cm2 for hydrogen pressure and agitated a t 50 rpm.
The reaction then proceeded at constant pressure and 0
1000 rprn through the prescribed reaction time, after Reaction Time (hr)
which the autoclave was immediately cooled off in
water to stop subsequent reaction. Figure 7. Test of Jut-order reaction mechanism for hydrogenation
from maleic anhydride to succinic anhydride
After separating out the catalyst, the several compo-
nents of reaction product were analyzed, THF, y B L , Reaction pressure, 60 kg/cm2; concentration of catalyst, 0.1 ~ t 7 ~
for 150°C, 0.05 wt7, for the others
n-propanol, n-butanol, propionic acid and butyric acid 0 15OoC, 0 160°C, 180°C, 2OOOC
by gas chromatography, while maleic anhydride, suc-
cinic anhydride, succinic acid and polyester were done
vza nuclear magnetic resonance spectra.

Result and Discussion 0.14

This reaction involved the following consecutive c
reaction mechanism -In
h
0
+
U
0 0.12

0. IO
_I
rn
I
?. 0.08
c
0
I!
(4) .-e,
>
0.06

+ HZ - C4H90H
(5)
0 I 2 3
Reaction Time (hr)
4 5
Hydrogenation of maleic anhydride to succinic
anhydride. As the reaction rates of Step 1 were too Figure 2. Yield of 7-BL us. reaction time in hydrogenation of succinic
fast to estimate, the data were gained using lower anhydride
reaction temperatures and a lower concentration of Reaction pressure, 120 kg/cm2, concentration of catalyst, 5 w t r o
catalyst than the other steps. Under these conditions, 0 240°C, 0 250°C, 260°C, rn 270°C
only the double bond was hydrogenated.
If the reaction proceeds in the first order for the con-
centration of maleic anhydride under the constant
hydrogen pressure, the rate equation may be written as

When we integrate the above equation, we note that,

a t t = 0, CMAH= C,wAHO
In ( C - w ~ H o I C , ~ A=d k~ t (7)
where CAWAH is concentration of maleic anhydride, k l is
rate constant, and t is reaction time.

26 INDUSTRIAL AND ENGINEERING CHEMISTRY

 0.10

0.08 4.0

0.06 3.0

0.04 2.0

0.02 1.0

0 I 2 3 4 0 I 2 3 4
Reaction Time (hr) t (hr)
Figure 3. Yield of T H F us. reaction time in hydrogenation of succinic Figure 4. Test of Equation 70
anhydride
Reaction pressure, 120 kg/cm2; concentration of catalyst, 5 wt%
Reaction pressure, 120 kg/cm2; concentration of catalyst, 5 wt% 0 240 O C , 0 250 " C , 260 'C, 270 "C
0 240°C, 0 25OoC, 0 26O0C, 270'C

T h e plot applied the data to Equation 7 shows the concentration of water. Therefor.;, Equation 8 becomes
linear dependence as shown in Figure 1.
Hydrogenation of succinic anhydride. The rela-
tion between the yield obtained over unit catalyst and
reaction time is shown for y-BL and T H F in Figures
2 and 3, respectively. The reaction rate of y-BL formed where CLAH = initial concentration of succinic an-
from succinic anhydride is extremely rapid in the early hydride
stage and then gradually slows after the reaction time a = nKp,H20 mKp,SA
elapses. T H F is obtained from the successive reaction n = coefficient to link the concentration of
of 7-BL (Figures 2, 3). Though by-products such as water with the conversion of succinic
n-propanol and n-butanol are not shown, their produc- anhydride: n = 1 for r - g L , n = 2
tion increases with increasing yields of THF. for T H F
T h e kinetic expression for this reaction is m = coefficient concerning the concentration
of succinic acid

If Kp' >> K, then the term [1 f K C s A H ] can be neglected

in comparison with a ( C E A H - C S A H ) : the integrated
where C,vAH = concentration of succinic anhydride
form of the resulting equation is given as follows
kz' = rate constant
K = adsorption coefficient of succinic an-
hydride
C, = concentration of inhibitor component
Kp' = adsorption coefficient of inhibitor compo-
T h e plot of the data according to Equation 10 is shown
nent
in Figure 4 and satisfies the above assumption.
Succinic acid and water may be considered as the chief From this result, it is difficult to distinguish which is
inhibitor components. Water is produced throughout the stronger inhibitor between succinic acid and water.
the reaction from succinic anhydride, y-BL, and THF. However it was confirmed that succinic acid is the
Furthermore the concentration of water is proportional strong inhibitor through the experiments in which
to the conversion of succinic anhydride. The con- succinic anhydride including succinic acid was used as
centration of succinic acid can be proportional to the the initial reactant.

VOL. 6 2 NO. 4 A P R I L 1 9 7 0 27
 Hydrogenation of y B L . T o eliminate the inhibi-
tion term by succinic acid, Reactions 3, 4, and 5, were
analyzed in the hydrogenation of r-RL, as a starting
material.
I n Figure 5 the conversion of 7-BL us. reaction time is
plotted. I t is clear that the rate of this step obeys zero-
order kinetics as follows

Integrating Equation 11 results in Equation 12

100

80 where C Y - ~ ~=Oinitial concentration of y B L

C Y - B L = concentration of y B L a t reaction time t
ka = k z ' / C y - B L o
60 The rate of side reactions 4 and 5 are considered as
first-order kinetics with respect to the concentration of
THF.
40

20
where CprOH= concentration of n-propanol
C B ~ O I=~ concentration of n-butanol

From the material balance, we get Equation 15

0 I 2 3 4 5 6 0
CTHF= k3CY-BLt - (CPrOH + CBUOH) (15)
Reaction time (hr) Reactions 4 and 5 are parallel reactions. From Equa-
Figure 5. Conversion of 7-BL us. reaction time i n hydrogenation of tions 13 and 14, we get Equation 16.
y-BL
Reaction pressure, 120 kg/cm2; concentration of catalyst, 5 ~ t 7 ~
CPrOIl
- kd
0 24OoC, 0 250°C, E 260°C, W 270°C, A 28OoC C B ~ H k:
Substituting Equations 15 and 16 into Equation 13

-
fl
15 When we integrate, we note that, at t = 0, C p r ~ l ~0
=i

-
Q) CPrOH =

E
-
c

0
C
0
IO
?a Similarly, for n-butanol
I C B ~ O=
H
K
2 5
c
.-0
v)
L
a
!=
9
0
0 0 I 2 3 4 5 6 I n Figures G and 7, the calculated curves and the
experimental points of conversion for n-propanol and
Reaction time (hr) n-butanol us. reaction time are shown. T h e calculated
Figure 6. Comparison between calculation curve and experimental curves were interpolated by the trial and error method
pointsfor conversion to n-popanol in hydrogenation of 7-BL giving several values to kd and k5 in Equations 18 and
Reaction pressure, 120 kg/cm2; Concentration of catalyst, 5 ~ t 7 19.
~ The values of k 4 and ka used in calculation are
0 240°C, 0 250°C, 0 260°C, 27OoC, A 28OOC shown in Table 11, and were estimated for per unit

28 I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
time (hr), unit pressure (atm) and unit amount of
catalyst (g). TABLE 11. VALUES O F k4 AND ks ESTIMATED FOR
T h e activation energy and frequency factor of each SEVERAL REACTION TEMPERATURES
reaction step are shown in Table 111. The frequency Reaction ka, k 5,
factor is estimated for unit time, unit pressure, and unit temp., O C I / h r atm g l / h r atm g
amount of catalyst as well as in Table 11. 240 1.43 x 10-4 0.67 x 10-4
Hydrogenation of succinic acid and polyester. 250 2.14 x 10-4 1 . 2 0 x 10-4
260 3.16 x 10-4 1.85 x 10-4
I n Equation 9, succinic acid is considered as the chief
270 3.77 x 10-4 2.94 x 10-4
inhibitor component. Therefore, it was examined to
280 6 . 6 6 X 10-4 4.16 x 10-4
see whether this catalyst is capable of hydrogenating
even succinic acid to y-BL and T H F . Moreover, poly-
ester was detected in the flow-type reaction; its content TABLE 1 1 1 . K I N E T I C DATA FOR EACH
tended to increase according to the decrease of water REACTION STEP
content in the reactor, though it was not detected in the Activation
Species of rate energy E, Frequency factor,
batch-type reaction. Accordingly, the examination for constant kcal/mol 7/hr atm g
polyester was also carried out. ki 12.5 3.04 X 106
In Figure 8, the yields of T H F and 7-BL us. reaction kz 18.2 2.25 X 106
time are plotted for succinic acid and polyester. It may ka 17.9 6.64 X 106
be concluded from Figure 8 that succinic acid is ad- k4 22.0 7.98 X 106
sorbed more readily than y B L , thus the rate of reaction ks 26.5 1.45 X 108
should adhere to zero-order kinetics for the concentra-
tion of succinic acid. In Figure 9 the conversion of
succinic acid us. reaction time is plotted, and it appears
that zero-order kinetics fits the data for the conversion
under 70 mol %. Moreover it may be concluded that
the hydrogenation rate of polyester is faster than the
hydrolysis rate of polyester. Because, if hydrogenolysis
and hydrolysis occur as the same grade, the amounts of
y B L and T H F should be almost the same, but the
amounts of both products are very different in Figure 8.
With respect to the effect of molecular weight of poly-
ester, it may be said that the reaction rate is diminished a
little. 0.12
Reaction rate of Step 1 is more rapid than in Step 2

0.IO

0.08
-I
m
h I
a\o h
0.06
-
Q) L

-0
E
0
LL
I
I- 0.04
rc
0

I
c 0.02
0
c
E
.-0
v)
L
0)
>
c 0 2 4 6 8 10
0
0 Reaction time (hr)
Figure 8. Yield of THF or I/-BL us. reaction time in hydrogenation
0 I 2 3 4 5 6 of succinic acid and polyester
Reaction time (hr)
Reaction temperature, 260 OC; reaction pressure, 120 kg/cm2;
Figure 7. Cvmparison between calculation curve and experimental concentration of catalyst, 5 wt%
points for conversion to 7-butanol in hydrogenation of y-BL 0 THF (succinic acid), 0 THF [polyester, mean molecular weight
ofpolyester (MW) = 20001, 0 THF (polyester, M W = lOOO), 7-
Reaction pressure, 120 kg/cm2; concentration of catalyst, 5 wt% BL (succinic acid), A y-BL (polyester M W = 2000), A y B L (poly-
0 24OoC, 0 25OoC, 0 260°C, 270°C, A 280°C ester, M W = 1000)

VOL. 6 2 NO. 4 APRIL 1970 29'

 or 3, and proceeds almost instantaneously. The rate of
Reaction 2 is next to Step 1. Reaction 3 is a rate-
determining one. Therefore, an appropriate choice of
100 reaction conditions makes it possible to control the
-
d
product distribution and the ratio of THF to y-BI,.
It is also possible to produce only THF by means of
-
0)
0
80 recycling y-BLto the reactor.
E
Y Reaction scheme. Figure 10 represents the overall
E
0
reaction scheme, “via y B L ” means that its reaction
U 60 proceeds mainly through the following steps.
I 4 I1 4 I11
c 40 or
0
c
.-
0 I + 11 + VI 4 VI1
E
> 20 “uza succinic acid” is
c
0
0
IV --f VI11 4 IX 4 VI1
0 2 4 6 8 IO or
R e a c t i o n time (hr) I 4 v + IX 4 VI1
Figure 9. Conversion of succinic acid us. reaction time “oia polyester” is
Reaction temperature, 260OC; reaction pressure, 120 kg/cm2;
concentration of catalyst, 5 wt% x +XI 4 VI1

Figure 10. Reaction scheme

30 INDUSTRIAL A N D E N G I N E E R I N G CHEMISTRY
 Figure 7 1. Processflow scheme

I t can be seen in Figure 10 that the following several The heat of formation of T H F and water by hydrogen-
reactions occur simultaneously : ating maleic anhydride in the liquid phase is so high
that heat removal may be an important problem.
1. hydrogenation for double bond: (I), (VIII) In addition to the usual method-i.e., the internal
2 . hydrogenative dehydration: (11), (III), (IX) heat exchange, the external circulating heat exchange,
3. equilibrium reaction between hydration and de-
and so on, a special method of heat removal is applied
hydration: (IV), (V)
in this process by which either heat removal or selec-
4. hydrogenative cyclization : (VI)
tivity rise-up of T H F can be achieved effectively.
5. dehydrative cyclization : (VII)
Process flow sheet. A schematic flow diagram is
6. equilibrium reaction between dehydrative con-
illustrated in Figure 11. Maleic anhydride, recycled
densation polymerization and hydrolysis : (X)
y B L from the distillation section, and hydrogen gas are
7. hydrocracking of polyester: (XI)
all fed to the reactor after being preheated (the tempera-
The order of the reaction rates to T H F of the above ture of the feed is always kept lower than that of the
three paths is reaction system).
The reactor effluent is cooled down to about 40 to
“via y-BL” > “via polyester” >> “via succinic acid”
60OC and comes to the flash drum, where gas can be
The plan of this process should be scheduled as the partially separated from the liquid product. The
above reactions are balanced mutually.
T h e side reactions produce a small quantity of lower
organic acids, such as propionic and butylic acid, and
lower alcohols, such as n-propanol and n-butanol. The
lower organic acids may be produced from the thermal
cracking of dicarboxylic acids and y-BL. The lower
alcohols are produced from the successive hydrogena- TABLE IV. HEAT OF FORMATION FROM MALEIC
tion of T H F as previously mentioned. ANHYDRIDE TO EACH PRODUCT A T 260°C, I 2 0 KG/CM2
Thermodynamic consideration. The heat of forma- Heat of
tion included in this process is shown in Table IV. formation,
Product Reactant By-Froduct kcallmol
Succinic anhydride Hz -30.57
y-Butyrolactone 3Hz H2 0 -50.42
AUTHORS Junichi Kanetaka, Taisuke Asano, and Shznobu Tetrahydrofuran 5Hz 2H20 -75.46
Masamune are employed by the Mitsubishi Petrochemical Co., n-Propanol 7Hz 2H20, CHI -112.90
Ltd., in Tokyo, Japan. This paper was presented as part of n-Butanol 6Hz 2Hz0 -100 87
the Symposium on Novel Processes and Technology of the European Propionic acid (via y B L ) 5Hz HzO, CH4 -93.20 ,
and Japanese Chemical Industry at the 158th National Meeting Propionic acid (via suc-
of the American Chemical Society in Nem York, N . Y., Septem- cinic acid) HzO
HP, con -29.22
ber 7-12, 1969.

VOL. 6 2 NO. 4 A P R I L 1 9 7 0 31
 separated gas is fed to the water-scrubber to recover
TABLE V. QUALITY O F T H F PRODUCT T H F accompanied with the unconverted gas.
The liquid product is diverted to the distillation
Analytical Design oj
data ofpilot commercial section, composed of four columns. I n the first coIumn,
plant product plant product y B L is recovered from the bottom and recycled to the
Boiling point ("C) 64.6-66,2 . I .
reactor to undergo the further reaction to T H F . How-
Color (APHA) 2 ... ever, if desired, it can be withdrawn as a final product.
Specific gravity (20/4'C) 0.887 * . e
The overhead product of the first column is purified
Refractive index ( n ~ " ) 1.407
to THF through successive columns by removing by-
Purity (wt70) 99.99 99.95
0.0002 0.0002
products such as water, n-propanol, and n-butanol.
Water content (wt%)
Peroxide (wt%) 0 0 Finally, purified THF is withdrawn from the top of the
Carbonyl compound last column. Purity of final T H F is 99.95 wt% or
(wt%) 0 0 higher.
Hydroxyl compound Raw materials for THF production. Maleic an-
(wt%) trace 0.01 hydride, produced by the oxidation of benzene or
(&fraction, is usually used as a raw material. Succinic
anhydride and y-IBL could also be used, if desired.
TABLE VI. ECONOMIC DATA BASED ON 5000 M E T R I C THF can be produced directly from benzene or Cq-
TONS/YEAR (105.7 BPSD) O F TETRAHYDROFURAN
W I T H PRODUCTION 333 STREAM DAYS/YEAR fraction by a two-stage oxidation-reduction process
Item Unit Consumption which involvcs effectively combining this T H F pro-
Raw materials cess with the existing maleic anh>-drideprocess.
Feed stock (MAH) IO3 ton/year 8 .O Impurities contained in hydrogen come into question
(1 03 lb/day) (52.8) for this process. T h e catalyst is permanently adulterated
Net hydrogen (as 10070 by sulfur compounds and teniporarily by carbon mon-
purity) 108 NM3/year 9.7
(as including 5y0 me- (million cu. oxide. Thus, hydrogen sulfide and carbon monoxide
chanical loss) ft.jday) (1.03) must be removed by conventional methods. Methane,
Products ethane, and nitrogen have little effect on the catalyst
Tetrahydrofuran 103 tonsjyear 5 .0 activity .
(barrel/stream (105.7)
day 1 Quality of THF product. Pilot plant data indicate
Utilities that the purity of T H F product is 99.95 to 99.99 wt7$.
Steam (180h) tons/year 2.12 I t easily passes the strict performance te.st for the raw
( l o 3 lb/hr) (4.7) material used for spandex fiber and urethan elastomer
Cooling water tons/year 249
(At = 9 . 5 ' c ) !gpm) (1100) and easily qualifies as a solvent for polyvinylchloride
Boiler feed water T / H tonsjyear 2.28 (PVC).
(gpm) (10.1) Analytical data of this pilot plant product are listed in
Fuel Gas 103 kcal/hr 1.25
(MM Btu/I-I) (4.96)
Table V together with the design of the first commercial
Electric power kWh/hr 366 plant.
Catalyst and chemicals lo3 USA $/Year 84.6 Economics. Economic data for the process are
Plant costs 106 USA $ P .38 illustrated in Table VI. 5000 metric tons per year
(106 barrelsjstream day) of tetrahydrofuran is manufac-
tured in 333 annual stream days. Plant costs listed in
Table V I include engineering fee and all the investmrnt
inclusive of facilities without battery limit. (Investment
without battery limit is taken as 15yc of investment
within battery limit.)
Although the cost of maleic anhydride is approxi-
mately 12p!/lb and the plant is operated a t full capacity
of 5000 tons per year, the production cost of tetrah>-dro-
furan is about 29.1#/lb with this proccss. Thcreforc,
payout time will be 2.2 years for 38.0d/lb of the THF
selling price.

REFERENCES
(1) European Chemical News, 34 (Dec. 16, 1966).
(2) U S Patent 3,238,225.
(3) US Patent 2,772,291, 2,772,292, and 2,772:293.
(4) Chemical Week, 63 (Mar. 5 , 1969).
Patents
Japanese Patent Publication No. 43-6946, 43-6947, 44-7934.
Netherland Patent 6,611,852.
France Patent 1,519,062.
Italy Patent 791,831.
(12 additional patents have been applied i n J a p a n a n d other foreign countries)

32 INDUSTRIAL A N D E N G I N E E R I N G CHEMISTRY