Reaction kinetics for the catalytic dehydrogenation of sec-butyl alcohol to methyl ethyl
ketone has been investigated at atmospheric pressure and temperatures ranging from 650"
to 750F. in the presence of solid brass spheres, 1/8 in. in diameter. The nature of this EXPERIMENTAL EQUIPMENT
catalyst permitted a direct evaluation of the surface involved in this reaction and allowed the
definition of a surface-feed ratio to be expressed as S / F in place of the conventional weight- The experimental unit used in these .
feed ratio W/F commonly used in catalytic studies. Feed compositions ranged from sec- studies, which is presented diagrammatically
butyl alcohol to mixtures containing high percentages of methyl ethyl ketone and hydrogen. in Figure 1, consisted of a small continuous
In these studies mass transfer effects were found to be significant and, for a proper pilot plant having a stainless steel tubular
representation of conditions at the catalyst surface, must be taken into account. The reaction heated by an electric furnace.
effect of feed compositions on the initial rates of reaction showed that the rate-controlling Liquid mixtures of ser-butyl alcohol and
step was the desorption of hydrogen involving a single-site mechanism. methyl ethyl ketone were pumped into
In addition, the results of these studies have been used to produce values of height of the reactor from a calihrated feed tank.
reactor unit HRU which have been found to correlate with mass velocity and temperature. The introdurtion of hydrogen, nitrogen,
The HRU provides a simple means of calculating the depth of catalyst necessary to effect and air to the system was regulated by
a designated conversion. means of a needle valve, and the flow rates
were measured with a rotameter. The
reactor effluent was passed through a water-
The design of catalytic reactors accord- t o or from the interior of the catalyst may cooled condenser to a phase separator, from
which noncondensable gases could be
ing to fundamental principles requires be significant and presents a n additional rollected over salt water in the gas-sampling
an exacting background knowledge of variable i n t h e fundamental study of bottle or passed through the wet-test meter
the reaction mechanism taking place on reaction mechanisms. T h e use of solid to measure the rate of production.
the siirfacc of t h e catalyst. Although brass as catalyst for the dehydrogenation
empirical approaches have proved cx- of see-butyl alcohol, besides eliminating Reactor
pedient for the design of commerc+il this possible variable, also permits t h e Figure 2 shows the details of the reactor,
units, a more fundamental approach t o direct calculation of the surface taking which was fabricated from 1%-in., type-304
the solution of such problems becomes part i n the reaction. Since i t is possible stainless steel pipe and was 40 in. long.
increasingly significant. As a iesult of to account for the surface participating in Two stainless stcel sleeves, 1 and % in. I.D.,
t h e developments of the reaction-rate the reaction, t h e relationship of space provided a variation of rross-sectional area
mechanisms proposed by Hougen and velocity and conversion dxa taking place of the catalyst bed. I n this manner greater
Watson (2) and involving t h e participa- in an elementary section of reactor ranges of mass velocity through the catalyst
tion of active centers on t h e catalytic volume whose surface is dS may be werc ronveniently obtained with the equip-
ment available. Thermowells provided on
surfaces, i t is now possible to associate expressed as
the top flange and bottom plug permitted
the rate of reaction with the behavior the establishment of temperature profiles
of t h r reactants and products on these F dxA = rA dS over the full length of the reactor. The
active centers. Several possible mecha- entire volume of the reactor, with the
nisms have been produced, and t o date From Equation (1) the surface-feed exception of the catalyst zone, was filled
these constitute t h r fundamcntal back- ratio S / F , which is directly related t o with 3- to 5-mm. glass shot. The glass shot
ground for comprehension of the various space velocity, now can be defined i n above the catalyst bed provided adequate
steps taking place in the course of a terms of conversion and reaction rate heat transfer surface and insured a uniform
reaction. The possible rate mechanisms as follows: fluid flow pattern; whereas this packing
a r e numerous, and the development of below the catalyst bed arted as a support
and also decreased the void volume of the
the corresponding rate equations is reactor, causing the rapid removal of the
presented elsewhere (2). effluents.
The preponderance of catal>tic rcac-
tions has been carried out i n the presence The surface-feed ratio S / F is a more
of porous catalysts which provide an PRELIMINARY INVESTIGATIONS
direct relationship and is used in these
extensive network through which the
studies rather than t h e conventional T h e catalytic dehydrogenation of
reactants and products must diffuse i n weight-feed ratio W / F . For porous secondary alcohols t o ketones was first
order for the reaction to proceed. T h e catalysts t h e use of the weight-feed reported b y Ipatieff (4) with metals such
variables influencing t h e diffusional phe- ratio W / F proves more expedirnt a n d is as iron, zinc; brass, manganese, and
nomena occurring within t h e catalyst nickel. No significant dehydrogenation
found t o be specific t o the type of catalyst,
pores have been considered theoretically method of preparation, and size. On the was noted with aluminum, lead, tin,
by Thiele (12). These diffusional effects other hand, t h e ratio S / F is more bismuth, antimony, or magnesium. Bal-
have been shown t o decrease with the fundamental i n this application since i t andin and Liberman ( I ) and Neish (9)
size of the catalytic particles and t o deals only with the specific nature of the showed t h a t copper has initial dehydro-
become insignificant for fluidized beds.
reaction and n o t t h e method of prepara- genating properties which gradually
For larger particles in fixed beds, diffusion
tion and size of t h e catalyst. Therefore, diminish owing t o irreversible poisoning
whenever t h e surface of t h e catalyst can effects. Ivannikov, Tatarskaya, and
be conveniently obtained, t h e surface- Gavrilova (6) showed t h a t uranium acts
Jcseph J Prrona is a t present u l t h the Oak
Ridge National Laboratoiy, Oak Ridge. Tennessee. feed ratio S / F may be applied. as a promoter t o copper, which did not
GLASS REACTOR
Feed Rate = 120 grams/hr
e 0.05
Conditions: 750F - I atm
LL
m o.04F
- \ Catalyst = 170 grams
I .
inch brass spheres
a
-5 000, I I , 1 , 1 , , , , , , , ,
Fig. 1. Flow diagram of experimental equipment. Fig. 3. Activity of catalyst in glass reactor vs. time.
attached to the use of brass for the h typical analysis of the effluent gases
liquid inlet, I/thermowe"
,gas inlet commercial production of methyl ethyl was found by mass spectrometer meas-
ketone. Thc technical information in the urements t o have the following com-
literature involving the use of brass for position :
the production of ketones is essentially Mole
nonexistent despite the current com- frdction
niercial interest and activity 011 this Hydrogen 0 959
subject. Therefore, it was found nccessary Methane 0 033
to conduct a series of exploratory investi- Ethane 0 001
gations in order to study the behavior of Ethvlenv 0 007
this catalyst under varying conditions of
operation. 1 000
Brass ball bearings %-in. in dianietcr
having a tolerance of fO.OO1 in. and This indicates that the reaction in the
composed of 65% copper and 35y0 zinc absence of catalyst is basically of a
were used throughout this study. cracking nature with a preponderance
x
Brass spheres in a glass reactor packed of hydrogen resulting in the effluent gases
.-c while no ketone was noted in the cffluent
I
with glass shot showed a continued
derrcase in activity with time as indi- liquid. In all cases the alcohol involved in
cated in Figuie 3 . Inspection slioned a this decomposition reaction consumed
less than 1% of the alcohol introduced
into the reactor.
Experimentation with the stainless
steel reactor revealed that blank runs
following a regeneration period with air
produced relatively high rates of cracked
gases which decreased rapidly with time.
Average ratcs of gas production in the
course of a run were measured and are
expressd as the zigzag lines presented
in Figure 5 . Instantaneous rates of gas
thermowell-
II Fig. 4. Appearance of new and used brass
production can be established from the
straight-line relationship drawn through
Fig. 2. Experimental reactor. ball bearings as catalyst. these average rates and piesented in
Figure 5 . These observations strongly
suggest that the cracking reactions
might have been catalyzed to some extent
show poisoning effects after 60 hr. of use. black carbonaceous deposit on both the by the nietallic oxides produced during
Several dehydrogenating catalysts of the brass catalyst and the glass spheres the regeneration period and that the
zinc oxide type are reported in the which was completely removed by the decrease in gas rate was due to the
literature (7, 8, 10). Iron, manganese, passage of air a t approxiniately 800F. reduction of the metallic oxides by the
and nickel, although capable of dehydro- The appearance of the oxidized catalyst product hydrogen. In order t o test this
genating secondary alcohols to ketones, is essentially black and is compared with theory, a regeneration period with air
have the added catalytic ability to a new catalyst in Figure 4. was followed by a nitrogen purge and
decompose the ketones produced into Several blank runs were made with then hydrogen was passed through the
olefins and oxides of carbon (4).I n the glass reactor without a catalyst. reactor at 800F. for 10 hr. before the
addition, Ipatieff (4) noted that zinc With a temperature of 750F. in the start of a run with see-butyl alcohol and
tended to oxidize in the course of the zone ordinarily occupied by the catalyst, without a catalyst. The rate of cracked
reaction involving the dehydrogenation an effluent gas rate of approximately gases produced under these conditions
of alcohols. 300 cc./hr. was observed for see-butyl was found t o be only about 25 cc./hr.
Considerable importance has been alcohol feed rates of 200 to 600 g./hr. in comparison with a rate of about 400
1
0
\
following regeneration1 0
- I - I
C
g 200
.-
c
0
0 STAINLESS STEEL REACTOR
500
s loo
0
( No Cotolyst Present)
a
0 In Feed Rote = 200 g I h r
s
4-
2
00 I ; I, ' I A A 6 y.
0 Hydrogen Reduction for 10 h n
Time, hours
P
Fig. 5. Cracking activity of blank reactor after regenera-
tion (no hydrogen reduction).
B
750 000 850 900
Temperature, O F
v
Alcohol Ketone Hydrogen
0 1000
A 0594 0.406
+ 0.196 0.804
0529 0.471
O . W 0 ~ ' '
10
"
20
"
30 ' I
40 '
50
$, Surface-Feed Ratio
-I
T,OR
A ' ' I ; ' ' " 20
0.00~" "
s, "
Surface-Feed Rotio
I ' " "
25
Fig. 8. Relationships of rate constants and temperature.
Fig. 9. Effects of mass transfer associated
in the catalytic dehydrogenationof sec-butyl
alcohol in the presence of brass at plex to form methyl ethyl ketone in the the reaction. I n Equation (4)K A is the
750F. and 1 atm. vapor phase and adsorbed molecular adsorption equilibrium constant for step
hydrogen, (3) desorption of the hydrogen 1, and K A K is the product of K A and the
in the vapor phase. The first two steps equilibrium constant for step 2.
were assumed to be a t equilibrium, with The constants G, K A , and K A R are
and Burwell (6), who approached equi- the third step controlling the rate of summarized in Table 3 and are correlated
librium for this reaction from both
directions. Altogether, ten possible reac-
tion model mechanisms were assumed to
apply, and the corresponding rate equa- TABLE4. COMPARISON
OF CALCULATED REACTION
AND EXPERIMENTAT, RATESAT 700F.
tions were derived for them. The resulting Average reaction
equations are presented in Table 3*. Average partial pressures rate, 1b.-moles/
along with the corresponding calculated In main gas stream, atm. A t interface, atm. (hr.)(sq. ft.)
constants. The solution of simultaneous (ZAL", SIF
equations involving the feed compositions Con- at Alcohol Ketone Hydro- Alcohol Ketone Hydro- Calcu- Experi-
and initial rates of reaction produced the Run version (zA)~", gen gen lated mental
rate constants summarized in Table 3. M-1 0.0476 1.9 0.879 0.042 0.042 0.859 0.056 0.056 0.0236 0.0246
A review of these constants shows that M-4 0.097 6.4 0.490 0.419 0.049 0.472 0.426 0.052 0.0137 0.0150
the desorption of hydrogen from a single M-8 0.089 3.7 0.453 0.042 0.489 0.445 0.052 0.491 0.0230 0.0235
site is the only mechanism for which the M-13 0.068 11.3 0.177 0.791 0.013 0.172 0.792 0.014 0.0057 0.0061
constants mere all positive a t 650, 700,
and 750F. Therefore, the rate of reaction
for the dehydrogenation of see-butyl
alcohol to methyl ethyl ketone in the TABLE6. PERTINENT
DATAFOR THE CALCULATION
OF REACTION
RATESO F
presence of brass becomes ILLUSTRATIVEEXAMPLE
Mass transfer Reaction
Partial pressure of coefficients, Partial pressures of rate,
components in main gas 1b.-moles/(hr.) components at catalyst 1b.-moles/
Con- stream, atm. (sq. ft.)(atm.) surface, atm. (hr.)(sq. ft.)
version
XA PA PK PH k,, k,, k9Ei PAi PKi PHi TA
J
1.41
Figure 8. The positive trmperature co- tion terms in the denominator of the
HRU =*.=rnG
efficient exhibited T\ ith K,, the adsorp- 1.2 In +. rate equation arc negligible, Hurt (3)
defines.the height of a catalytic reactor
tion equilibrium constant of thr alcohol,
may appear inronsistmt with present 1.1 0" through the expression
concepts which point out that this
variation should be negative. The major- x = (HCU + N'I'CjN, = (HRU)N,
ity of catalytic rractions found in the (9)
literature shon thcsc constants t o have a where I f G Cr represents the height of
negative tempcraturr coefficient; how- catalytic unit resulting from the surface
ever, a few excrptions havr hcen reported. reaction; JITU, the height of transfer
In the hydrogenation studies of Sussinan unit due t o mass transfer effects; and
and Potter (11) over a copper-magnesia H R U , the height of reactor unit, which
catalyst a positive trmperature coefficient represents the contribution of both mass
for the adsorption equilibrium constant transfer and surface-reaction effects.
of propylene is similarly reported. These Feed =100%Alcohol
Although Hurt in the development of
constants can be expressed in equation these concepts has postulated a uni-
form with temperature as follovc s: molecular reaction, these units have been
assumed to apply in thc dehydrogenation
10,735
l o g C = -~
T
+ 7.776 G, Mass Velocity, lblhr ft'
Alcohol
Ketone
fraction
0 0026
0.4987
0 0.2 0.4 0.6 0.0
J
1.0
interface, atm.
= partial pressure of ketone at
x, , Conversion interface, atm.
Hydrogen 0 4987 = partial pressure of hydrogen a t
Fig. 12. Reciprocal rate-conversion rela-
1,0000 interface, a t m .
tionship for illustrative example (rigorous
approach involving reaction rate and mass = reaction rate, 1b.-moles of alco-
The height of the reactor required when hol converted/(hr.) (sq. ft.)
sec-butyl alcohol is charged to the reactor transfer effects).
becomes = surface area of catalyst, sq. ft.
= absolute temperature, OK. or OR.
rates taking place a t the catalyst surface = weight of catalyst, lb.
are presented in Table 6 along with their
corresponding conversions ZA and other = conversion, 1b.-moles of alcohol
1.000 - 0.0026 pertinent data. converted/lb.-mole of alcohol in
= 0.147 In A plot of reciprocal rates and conversions feed
p a , - 0.0026 = height of catalyst bed, ft.
is presented in Figure 12, which upon
Assumed values of z produced the follow- integration between the limits of conversion, = viscosity of gas, lb./(hr.)(ft.)
ing partial pressures of alcohol leaving the ZA = 0 to XA = 0.90, yields a value of = density of gas, lb./cu. ft.
reactor PA,, corresponding to conversions xA.
Z, ft. PA,, atm. XA
LITERATURE CITED
0 00 1 000 0 000
0 05 0 713 0 168 The height of the reactor required to effect 1. Balandin, A. A., and A. L. Liberman,
0 10 0 508 0 326 a 90% conversion therefore becomes Compt. rend. acad. sci. (U.S.S.R.),
0 20 0 259 0 588 28, 794 (1940).
0 40 0 0682 0 872 2. Hongen, 0. A., and K. M. Watson,
I 00 0 0037 0 993 Chemical Process Principles, pp. 902-
926 and 982-990, John Wiley and
The relationship of reactor height z vs. h = 0.51 ft. Sons, New York (1947).
conversion ZA is presented in Figure 11. 3. Hurt, D. M., Ind. Eng. Chem., 35, 522
From this figure a reactor height of z = 0.47 The results produced by both methods
are in fair agreement. The method involving (1943).
ft. is required to effect a 90% conversion. 4. Ipatieff, V. N., Catalytic Reactions a t
PARTB. A rigorous solution to this the use of the rate equation and mass
transfer contribution is rigorous and accu- High Pressures and Temperatures,
problem involving the use of the reaction- pp. 16-18, The Macmillan Company,
rate equation and mass transfer contribu- rate but is quite demanding from a calcu-
lation point of view. On the other hand, the New York (1936).
tions necessitates a trial-and-error proce- 5 . Ivannikov, P. Y., M. G. Tatarskaya,
dure, in which the composition of alcohol, approach involving the use of the height of
reactor unit proves somewhat less exact, and E. Y. Gavrilova, Sintet. Kauchuk,
ketone, and hydrogen must be established 9, 16 (1936).
a t the catalyst surface in order to produce but far less time consuming for a reasonable
conclusion. This inexactness is inherently 6. Kolb, H. J., and R. L. Burwell, Jr.,
the proper rate. For a designated conversion J . Am. Chem. Soc., 67,1084 (1945).
xA a preliminary rate of reaction was
associated in the proper number of reactor
7. Mottern, H. O., U. S. Pat. 2,586,694
calculated by using the component com- units Nn as reflected in the differences
involved in the logarithmic expression at (February 19, 1952).
positions of the main gas stream. Mass 8. McNeil, Donald, and F. R. Charlton,
transfer calculations involving the use of high conversions.
British Pat. 636,743 (May 3, 1950).
j d factors for granular parkings ( 2 ) permitted 9. Neish, A. C., Can. J . Research, 23B,
the evaluation of mass transfer coefficients NOTATION 49 (1945).
k, for alcohol, ketone, and hydrogen from C = constant in rate equation 10. Padovani, Carlo, and Giuseppe Salvi,
which the partial-pressure differences of Riv. combustibili, 5, 81 (1951).
these components across the gas film were
D, = diffusivity of transferable com-
11. Sussman, M. V., and Charles Potter,
calculated. With the composition of these ponent, sq. ft./hr.
Ind. Eng. Chem., 46, 457 (1954).
components defined a t the interface, a new F = feed rate, 1b.-moles of alcohol/ 12. Thiele, E. W., Ind. Eng. Chem., 31,
reaction rate was calculated and the mass hr. 916 (1939).
transfer calculations repeated until the G = mass velocity of gas stream,
reaction rate no longer changed. The final Ib./(hr.)(sq. ft.) Presented at A.I.Ch.E. Sezttle meeting