Professional Documents
Culture Documents
A New Coated Catalyst
A New Coated Catalyst
A New Coated Catalyst
A New Coated Catalyst for the Production of Diacetone Alcohol via Catalytic
Distillation
Greg P. Dechaine† and Flora T. T. Ng*
Department of Chemical Engineering, UniVersity of Waterloo, 200 UniVersity AVenue West,
Waterloo, Ontario, Canada N2L 3G1
A method has been developed for the production of catalytically active coated ceramic distillation saddles for
use in catalytic distillation (CD) columns. A thin layer of magnesium acetate was applied to ceramic Norton
saddles using a sol-gel dip-coating method. The magnesium acetate coating was converted to the catalytically
active magnesium oxide (MgO) via a temperature-ramped calcination program. The kinetic performance of
the MgO-coated saddles for the aldol condensation of acetone was determined in a batch reactor and a batch
CD column. Although the coated saddles provided lower yields to DAA than Amberlite IRA-900 ion exchange
resins in a batch reactor, their improved mass transfer characteristics provided higher yields than the same
ion exchange resins held inside fiberglass bags in a CD column at low flow rates. The coated saddles also
showed significant improvements in selectivity to DAA compared to the resin catalysts for the aldol
condensation reaction carried out both in the batch reactor and in the CD column.
Figure 1. Photographs of the alligator clips used for suspending the saddles for dip-coating.
the performance of alkaline earth metal oxides (including MgO) moisture. This solution was stirred vigorously until all visible
and showed that these basic solids are active for the aldol solids dissolved and then refluxed for 24 h. The resulting
condensation reaction. Hydrotalcites and layered double hy- precursor solution is a colloidal suspension of magnesium
droxides in various forms have also been shown to be active acetate (Mg(OAc)2) particles in an ethanol solvent and is referred
for this reaction.13-17 MgO was chosen to be the catalytic to as a precursor sol.
material in this study since several methods for applying thin 2.2.2. Preparation of the Saddles. Norton saddles provide
films of magnesium oxide (MgO) to ceramic and/or glass the highest specific area (984 m2/m3) and the highest void
substrates18-22 were reported although none of these studies fraction (0.65) of the three commercial packings available in a
were of a catalytic nature. Ceramic distillation saddles should 1/4 in. nominal size and therefore were chosen for this project.
have similar surface properties as the glass and ceramic According to the manufacturer’s MSDS, Norton saddles are
substrates, and therefore these methods were chosen to coat the composed of 65%-75% silica, 20%-27% alumina, and the
saddles. balance other metal oxides. Prior to being coated, the saddles
This paper reports a new method for the preparation of a were cleaned and degreased using detergent and water, followed
coated ceramic saddle catalyst for use as a packing in a CD by toluene and then acetone. The cleaned saddles were
process. Particular attention is given to the methods necessary thoroughly rinsed with demineralized water, dried in an oven
for applying the precursor coating to the saddle as well as the at 80 °C, and stored in a sealed glass container to prevent any
successful conversion of this precursor coating to a catalytically contamination.
active form. The performance of these coated catalytic saddles 2.2.3. Coating the Saddles. A thin layer of the precursor
for the aldol condensation of acetone was determined in both a sol was deposited onto the ceramic distillation saddles using a
batch reactor and a CD column. The results obtained in a CD dip-coating process. According to Brinker and Scherer,25 the
column are significant since this is first report on the use of a thickness of the film deposited by dip-coating is directly
coated catalytic distillation packing in an actual CD column. proportional to the withdrawal speed of the substrate. Control
These data provide insight into the performance of such a coated of the withdrawal rate of the saddles was accomplished using
catalyst in a CD column for the aldol condensation of acetone. a moving platform lowered by a threaded shaft and an electric
motor. For each revolution of the threaded shaft, the platform
2. Experimental Section is lowered by a distance equivalent to the distance between the
threads of the shaft. Since the distance between the threads
2.1. Neutralization of Reagent Acetone. ACS grade acetone
remains constant, the linear speed of the platform is directly
contains up to 0.0003 mequiv/g of titratable acid,23 which is a
proportional to the rotational speed of the shaft and hence
combination of dissolved carbon dioxide (0.000 23 mequiv/g)
proportional to the speed of the drive motor. The precursor sol
and acetic acid (0.000 033 mequiv/g).24 Because the catalysts
was placed in a stainless steel dish supported on the moving
used in this study are basic, this acidity must be removed from
platform.
the reagent acetone to avoid neutralization of the catalyst sites.
The reagent acetone was treated using Amberlite IRA-900 ion For most applications of dip-coating, the substrate has a
exchange resin in the hydroxide form. The dried resin was added relatively simple geometry (e.g., rectangular wafer) and is easily
to bottles of HPLC acetone (EMD Chemicals Inc. #AX0115- held in place while being withdrawn from the coating solution.
1) and agitated for 45-60 min. The treated acetone was decanted Ceramic distillation saddles, however, have an irregular shape
through a vacuum filter and distilled. The water content of the that makes it difficult to suspend vertically for immersion and
neutralized and distilled acetone was 0.295 wt % measured using withdrawal. This difficulty was overcome by using soldered
Karl Fischer titration (Mitsubishi model CA-06 coulometric KF alligator clips (RadioShack #270-0380). To accommodate a
titrator), and the DAA content was <0.02 wt %. Prior to use in larger number of saddles, the clips were first mounted in strips
reactor studies, the neutralized acetone was degassed using of 46 saddles (Figure 1a), and then five such strips were mounted
argon. together onto a master bracket for a total of 230 saddles (Figure
2.2. Production of Coated Catalytic Saddles. 2.2.1. Prepa- 1b). This master bracket was attached to a shaft for suspension
ration of the Precursor Sol. A combination of the methods of above the coating solution. With the saddles suspended above
Menon and Bullard21 and of Shukla et al.19 was used because the coating solution, the platform was raised until the saddles
of the relative simplicity of the method, and the resulting were immersed 3-4 mm below the surface of the liquid to
precursor solutions are stable for several weeks following their ensure that the withdrawal rate was constant during the coating
initial preparation. Magnesium ethoxide (Mg(OEt)2, Alfa Aesar deposition phase.
#14264) and glacial acetic acid (HOAc, Fisher Scientific #A38- Once the saddles were withdrawn and the coating deposited,
P212) were dissolved in ethanol (Commercial Alcohols Inc., the saddles were connected to a horizontally oriented motor and
maximum water content ) 0.1 vol %) in a 27:3:1 molar ratio rotated slowly at 45-75 rpm to prevent gravity drainage of the
of ethanol:HOAc:Mg(OEt)2. The Mg(OEt)2 was handled and coating while drying. A steady stream of ambient air was blown
dispensed in a glovebox to prevent reaction with atmospheric over the rotating saddles to enhance evaporation of any residual
9306 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008
Figure 3. Comparison of the conversion of acetone (XAc), the selectivity to DAA (SDAA), and the initial rate of production of DAA for the coated saddles
with varying number of coats applied to the saddle (46 saddles in a batch reactor at 54 °C with un-neutralized acetone and an air atmosphere).
(indicated in Figure 2 by the X on the column) was loaded next, reported to provide the optimum activity of the MgO catalyst
covered by another separator plate, and the remainder of the for the acetone condensation reaction.12 Preliminary attempts
column was filled with more Norton saddles. The thermocouples to calcine Mg(OAc)2 using a muffle furnace resulted in the
were equally spaced 6 in. apart along the height of the column. formation of significant amounts of coke on the surface of the
The reboiler consisted of a boiling flask and heating mantle precursor solids. Therefore, it was necessary to supply fresh
controlled using a variable autotransformer and is the same air to the solid surface during the calcination to ensure
reboiler setup used by Podrebarac et al.10 The heat input of the conversion of acetate to carbon dioxide rather than to coke.
reboiler as a function of the input voltage and the reboiler According to Menon and Bullard,21 two distinct transforma-
temperature was calibrated using pure water. Prior to starting tions occur when Mg(OAc)2 is heated to 475 °C. At 200 °C
the batch CD experiment, the acetone was degassed in a 500 the form of Mg(OAc)2 changes from R-Mg(OAc)2 to β-M-
mL separatory funnel. The water content of the neutralized, g(OAc)2, and any residual solvent is driven off. At 300 °C the
distilled acetone used in the CD experiments was 0.276 wt % majority of the Mg(OAc)2 decomposes pyrolytically to MgO.
as measured using Karl Fischer titration. Approximately 600 The majority of the decomposition occurs at ≈300 °C, although
mL of acetone was charged to the separatory funnel and it is not fully complete until ≈360 °C. In order to control the
degassed using an argon flow of 1.3 L (STP)/min for ap- decomposition and ensure adequate air supply, a means of
proximately 11/2 h. Prior to loading the catalyst, the column ramping the temperature slowly is necessary in order to properly
was swept with high flow rates of argon from the bottom inlet calcine Mg(OAc)2 to form MgO.
(see item 4 in Figure 2). The overhead assembly was removed For preliminary trials using Mg(OAc)2 · 4H2O powder, the
from the column, and the coated ceramic saddles were quickly same 0.1 °C/min ramping rate that Menon and Bullard21 used
added under argon flow. The remainder of the column was filled was employed until a temperature of 475 °C was reached,
with more Norton saddles and the overhead assembly reinstalled. whereupon it was held for 1 h. Unfortunately, this ramping rate
Throughout this process, argon gas was flowing from the bottom is very slow and requires a prohibitively long time to complete.
inlet to ensure that the coated ceramic saddles were contained Therefore, for the coated catalysts, the modified ramping
within an inert atmosphere. program shown in Table 1 was used. Since very little decom-
The column was further purged with argon and then the position occurs from ambient temperature to 250 °C, a higher
degassed acetone was quickly added to the reboiler. Again, the ramping rate of 1 °C/min was used through this region. In the
column was purged with argon from the bottom inlet, after range between 250 and 375 °C where decomposition does occur,
which the argon flow was switched to the top inlet (see item a slower ramping rate of 0.2 °C/min was employed. Finally,
10 in Figure 2). The flow of argon from this top inlet was 0.1 once the solids reach 375 °C and the Mg(OAc)2 has been
L (STP)/min. A 0.5 mL sample of the initial acetone in the converted to MgO, the temperature can be ramped up to the
reboiler was taken, after which the power to the heating mantle final temperature of 475 °C at a rate of 1 °C/min since coke
was turned on and the entire column was insulated. A stopwatch formation is no longer a concern. The final temperature is held
was started once the temperature of the reaction zone began to at 475 °C for an additional 6 h to ensure complete conversion
rise, indicating that the acetone vapor had reached the reaction to MgO.
zone. The initial temperatures of the reboiler and the column
were recorded once reflux had been established. 0.5 mL samples The coefficients of linear thermal expansion for pure mag-
of the reboiler liquid were taken at regular time intervals and nesium oxide and ceramic Norton Saddles are 6.01 × 10-6 °C-1
analyzed using the GC methods previously described. at 500 °C26 and 14.5 × 10-6 K-1 at 527 °C (800 K),
respectively. According to this data, at 500 °C, the MgO coating
will shrink by 141% more than the ceramic saddle substrate.
3. Results and Discussion
This effect will be magnified by the fact that the coating will
3.1. Calcination of Coated Ceramic Saddles. A final cool much more quickly than the substrate due to temperature
calcination temperature of 475 °C was chosen since this was gradients present throughout the saddle during the rapid cooling
9308 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008
process. Exposure of the coated saddles to extreme temperature The initial rate of reaction for these five catalysts is plotted
changes will cause the coating to crack and pull away from the as a function of the mass of catalyst coating in Figure 3b. The
substrate. Therefore, once the calcination program was complete, mass of coating was determined after the batch reactor test was
the catalyst saddles were cooled slowly at a rate of ≈2.5 °C/ completed. As a matter of reference, the mass of catalyst was
min using inert helium gas. The use of an inert gas for cooling estimated from the mass of precalcined coating assuming that
also minimized the exposure of the catalytically active MgO the initial coating was either pure Mg(OAc)2 or pure
layer to air, thus preventing the poisoning of the basic catalyst Mg(OAc)2 · 4H2O. This data is also included in Figure 3b. With
sites by atmospheric CO2. the exception of the value for 50 coats, the measured values
3.2. Effect of Coating Mass on Performance. At the outset correspond very closely to those for Mg(OAc)2 · 4H2O, which
of this work, it was unknown whether or not synthesizing a indicates that the original coating is reacting with atmospheric
thicker coating layer would result in increased yields of DAA. moisture leading to the hydrate form. Also worth noting is that
The external surface area of the coating will be roughly the the catalyst mass for the catalyst with 50 coats is higher than
same as that of the underlying support provided the thickness both estimates (which is not possible) and therefore is question-
of the coating is not large. If the coating layer has no porosity, able. The specific activity of the magnesium oxide coating was
depositing additional layers of coating should not result in obtained by regressing the data (excluding the data point for
improved catalytic performance. Therefore, varying the number 50 coats) to obtain a value of 178 mmol/(L min gMgO). The
of coats applied to the saddles and testing the resulting regression line was extrapolated to illustrate that the estimated
performance would provide information on whether the calcined mass of MgO (based on Mg(OAc)2 · 4H2O) for the catalyst with
coating layer has any porosity. 50 coats lies very close to the line obtained for the other data.
The coating apparatus described earlier is capable of coating Therefore, any further analysis will use this estimated value
230 saddles simultaneously in groups of 46 saddles. To test the rather than the (apparently erroneous) measured value for the
effect of additional coating layers on the performance, a batch mass of MgO on the catalyst with 50 coats.
of 230 coated catalyst saddles was made using a single batch The drawback of a thicker catalyst layer is the potential for
of coating solution. The saddles were withdrawn at a speed of increased diffusional resistance as a result of longer pores. The
1.25 mm/s and dried for 10 min on the rotating dryer between impact of pore diffusion on the rate of reaction is usually
each dip. Following the first 10 coats, 46 of the saddles were quantified using the effectiveness factor, η:
removed. Following the next 10 coats (20 total coats) another
46 saddles were removed. This was repeated after every 10 rate of reaction with pore diffusion limitations
η) (5)
coats, resulting in 5 batches of 46 saddles, each with 10 more intrinsic rate of reaction
layers than the previous group. These 5 batches of catalyst were
calcined and tested in a batch reactor with the results shown in In general, η for an nth-order irreversible reaction is a function
Figure 3a. The conversion of acetone, XAc, and the selectivity of the dimensionless Thiele modulus, φ:29
to DAA, SDAA, plotted in Figure 3a follow the standard IUPAC
nomenclature27 for conversion (X), yield (Y), and selectivity (S): V n + 1 (kmcat)CA
n-1
φ) (6)
moles of acetone converted to product i S 2 De
Yi ) (2)
initial moles of acetone
where V ) volume of catalyst (m3), S ) external surface area
total moles of acetone reacted of catalyst (m2), n ) the order of the reaction (2 in this case),
Xi ) (3) k ) the intrinsic rate constant for the reaction in the absence of
initial moles of acetone
diffusion (mol1-n/m3(1-n) min gcat), mcat ) mass of catalyst (gcat),
Yi CA ) external concentration of reactant A (mol/m3), and De )
Si ) (4) the effective diffusivity of the reactant/product in the pores of
XAc
the catalyst (m2/min). Although the functional relationship
between η and φ is complex, particularly for the complex shape
From these results, it is evident that increasing the number of
of the distillation saddles used in the present work, it is still
coats does indeed lead to increased yield of DAA. This suggests
possible to examine the rate data to determine whether or not
that there is some porosity in the coated layer following
pore diffusion is significant. In the case where pore diffusional
calcination and that there is an incentive to deposit as much
resistance is negligible, η f 1 and the observed rate of reaction
coating as possible.
should remain constant:29
In an attempt to quantify the difference in performance
between the five different catalysts, the method of initial rates28 η2 (robs)2 k2
was used since basic catalysts applied to this reaction system ) ) )1 (7)
are susceptible to deactivation.8,9 A quadratic polynomial with η1 (robs)1 k1
0 intercept was fit to the first 10 min of concentration vs time
data (5 data points), providing a smoothed functional relationship On the other hand, when there is strong pore diffusion resistance,
for CDAA(t). This polynomial was then differentiated (to give η f 1/φ and the ratio of observed reaction rates is inversely
dCDAA/dt ) rDAA) and solved at t ) 0 to give the initial rate of proportional to the ratio of the Thiele moduli:29
reaction before any potential deactivation and other detrimental
effects have occurred. Although the first data point, usually taken η2 φ1
) (8)
after 2 min, could have been used to determine the initial rate η1 φ2
(i.e., dC/dt ≈ ∆C/∆t), using regression to smooth a curve over
multiple data points is preferred since inaccuracies in the first The volume of MgO on the surface of the saddle is given by
data point are less likely to skew the results. Also, data points mcat/FMgO, where mcat is the mass of MgO coating on the saddles.
at >10 min were not used in the regression analysis to avoid For the thin coatings present on the saddles, the external surface
incorporation of data with potential catalyst deactivation. area, S, will remain relatively constant. Finally, since CA and
Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 9309
De are constant, then, for the case of strong pore diffusional Table 2. Effect of Coating Mass on Catalyst Effectiveness Factor
resistance, eqs 6 and 8 combine to give no. of MgO mass k1,0 × 1000 η/η20 ) φ/φ20
coats (g) (L/(mol min gcat)) k/k20 (eq 9)
η2
( )()
mcat,1 k1
φ1 3/2 1/2
10 0.0190 0.273 0.32 1.61
) ) (9)
η1 φ2 mcat,2 k2 20 0.0179 0.844 1.00 1.00
30 0.0318 0.803 0.95 0.43
40 0.0667 1.206 1.43 0.12
Therefore, if there is significant pore diffusion resistance, the 50 0.1907a 1.066 1.26 0.03
ratio of effectiveness factors should be inversely proportional a
This is the catalyst mass estimated from the original mass of
to the mass of catalyst coating on the saddles. coating, assuming that the coating was Mg(OAc)2 · 4H2O.
The aldol condensation of acetone (eq 1) is second order in
3.3. Catalytic Performance of Coated Saddles. Once the
acetone concentration (CAc) while the reverse reaction is first
procedure for producing the coated saddles (i.e., 30 coats) was
order in DAA concentration (CDAA), yielding the following rate
finalized, the catalytic performance of the saddles was studied
expression:6
further. The performance of the coated saddles in a batch reactor
(92 saddles, 30 coats) is shown in Figure 4. The number of
1 dCDAA saddles loaded to the batch reactor was doubled to 92 saddles
rDAA ) ) k1CAc2 - k2CDAA (10)
mcat dt (vs 46 saddles in the previous section) in order to reduce the
time required for completing the experiment. Also, these
where k1 ) the rate constant for the forward reaction (L/(mol experiments were performed using neutralized acetone (see
min gcat)), k2 ) rate constant for the reverse reaction (min-1 section 2.1) to reduce the effects of deactivation. The data from
gcat-1), and mcat ) mass of catalyst (gcat). At the start of a batch one of the batch reactor experiments of Podrebarac et al.6,30
run (t ) 0), the concentration of DAA is zero, and therefore for the strongly basic macroporous anion exchange resin
the second term in eq 10 disappears. Therefore, using the initial Amberlite IRA-900 is included in Figure 4 for comparison.
rate of reaction allows eq 6 to be used even though the aldol The resin catalyst used by Podrebarac et al.6,30 exhibits a
condensation reaction is reversible. The initial rate and the mass higher conversion of acetone, although the majority of the
of catalyst can be used together with the initial acetone increased conversion arises from the production of byproduct
concentration (12.91 mol/L at 54 °C) to calculate the initial rate MO rather than DAA. Since the dehydration of DAA to MO is
constant (k1,0) for the forward reaction. The values of k1,0, as acid catalyzed, the use of un-neutralized acetone (i.e., acetone
well values of η/η20 and φ20/φ (i.e., the values for 20 coats were containing trace acetic acid) could neutralize some of the basic
used as the baseline) are summarized in Table 2. Both η and φ sites on the resin, and the trace acetic acid could potentially
were referenced to the values for the catalyst with 20 coats since increase the conversion of DAA to MO. In any case, the resin
the initial reaction rate for the catalyst with 10 coats seems catalyst does reach the equilibrium conversion to DAA much
artificially low. faster than the coated saddles, implying a higher activity. In an
If there were significant diffusional resistances within the effort to compare the activities of these two catalysts, the initial
pores of the MgO coating, then eq 9 would apply and the ratio rates of the reaction are summarized in Table 3. The turnover
of the effectiveness factors would decline rapidly according to frequency (TOF) for each catalyst is also included in Table 3.
the values calculated in the last column in Table 2. However, The number of basic sites present on the coated saddles was
the ratio of the effectiveness factors remains relatively constant estimated assuming a basic site density of 0.15 mequiv/g as
with a value around one, indicating that eq 7 applies and that measured by Matsuda et al.31 for MgO derived from Mg(OAc)2.
pore diffusion has very little effect on the rate of reaction with This method of calculating TOF assumes that all sites present
the coated catalyst. This analysis was performed assuming an on the catalyst are of equal activity for the aldol condensation
irreversible reaction (eq 6) when it was stated earlier that the reaction. The TOF for the coated saddles are 2 orders of
aldol condensation reaction is reversible (eq 1). However, in magnitude higher than those for the resin catalyst, suggesting
the case of a reversible reaction, diffusional resistances within stronger basic sites on the MgO coating than on the ion exchange
the pores would result in lower effectiveness factors than the resins. The higher initial reaction rate of the resin catalysts is
irreversible case, and therefore the effects would be even more therefore likely due to a much larger number of basic sites
pronounced. present in the reactor. Since the ion exchange resins are not
The analysis to this point does not give any indication about stable beyond 60 °C, it is not possible to use temperature-
the durability of the coating. Although the catalyst produced programmed desorption of CO2 to compare the distribution and
with 50 coats had the highest yield, its integrity was question- strength of the basic sites of the coated saddles and the resins.
able. A large portion of the coating layer flaked off upon addition Therefore, further discussions on the effect of basic site strength
to the reactor. A similar effect was seen for the catalyst produced and density on the relative rates of these two different types of
with 40 coats, although to a lesser degree. This effect was likely catalysts are not warranted.
due to aging of the coating solution as the coating process In terms of the selectivity of the catalyst toward DAA, Figure
proceeded. Toward the end of the coating process, the coating 4b indicates that the coated saddles are significantly more
solution became much thicker and more viscous due to selective to DAA than the resin catalysts. The most likely reason
evaporation and general consumption of the coating solution. for this difference in selectivity is the difference in the porosity
This caused more coating to be deposited per dip, which in turn of the two catalysts, although the acidity in the un-neutralized
led to a large excess of coating; particularly for the last batch acetone used by Podrebarac et al.6 may result in increased MO
(i.e., 50 coats). The catalysts produced with 30 coats or less production. According to Podrebarac et al.,6 diffusional limita-
did not exhibit this flaking behavior and remained intact even tions cause the concentration of DAA within the pores of the
after 7 h of use in the batch reactor. Therefore, the optimum resin catalyst to remain at the equilibrium value even if the bulk
number of coats in the batch reactor was 30 coats at a concentration is not. This leads to increased production of MO
withdrawal speed of 1.25 mm/s. This recipe was used for the since the active sites within the pore are exposed to a higher
remainder of this work to produce the coated saddles. concentration of DAA. The coated saddles, on the other hand,
9310 Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008
Figure 4. Comparison of the acetone conversion and DAA selectivity of the coated saddles (92 coated saddles, 30 coats, neutralized acetone, Ar atmosphere)
and the resin catalyst of Podrebarac et al.6,30 (Amberlite IRA 900 macroporous resin, 2 mL resin, 0.49 mmol H+/mL Resin, un-neutralized acetone, air
atmosphere) at 54 °C.
Figure 7. DAA and MO yield with coated saddles after 24 h in a batch CD column.
the productivity of DAA, the run at 55 V remained constant sol-gel dip coating method followed by calcination to produce
while the run at 85 V showed a slight decline after 24 h. a catalytically active MgO layer. The initial rate of production
Therefore, it appears slight deactivation is occurring in the CD of DAA increases proportionally with the mass of coating, and
column based on the very limited data. Podrebarac et al.10 also hence there is incentive to deposit as much coating mass as
observed a similar decline in MO productivity with time, and possible. However, the durability of the coating degenerates with
they attribute this to the formation of higher molecular weight an increase in the mass of coating, resulting in a tradeoff
products due to further condensation reactions involving MO between yield and durability. The durability of the coating is
on the catalyst surface.9 The fact that MO is a precursor to the also significantly affected by the calcination method used and
deactivation could explain why the effects of deactivation are requires slow temperature ramping during both heating and
more prevalent in the formation of MO than DAA. However, cooling to ensure adequate structural integrity of the catalytic
more experiments at much longer reaction time are required coating.
before any conclusions could be made regarding the apparent The initial rate of DAA production of the coated catalyst
deactivation. saddles in a batch reactor was significantly lower than that for
Another major concern when using the coated saddles is the the ion-exchange resin catalyst used by Podrebarac et al.6 but
potential for the MgO coating to erode and leach from the saddle with a much higher selectivity to DAA. The coated saddles are
substrate leading to catalyst in the reboiler. As mentioned much more active than the ion exchange resins based on
previously, the aldol condensation of acetone is an equilibrium- calculated TOF. The higher rates exhibited by the resins likely
limited reaction. At 54 °C, the equilibrium concentration of resulted from the much higher concentrations of basic sites on
DAA is 4.3 wt %.6 During the batch CD experiments, the the resins. Despite having much lower concentrations of basic
temperature in the reboiler ranged between 55.2 and 63.5 °C. sites, the coated saddles performed very well in a CD column.
At these temperatures, the equilibrium concentration of DAA The coated saddles do not experience the severe mass transfer
should be less than 4.3 wt %. However, the concentration of limitations that were encountered by the resin catalyst held in
DAA in the reboiler exceeded this equilibrium value after only fiberglass bags. At lower flow rates in the CD column, the coated
2 h for experiments at 85 V and after 3 h at 55 V. At this point, catalysts produce more DAA than the resin catalysts held in
any possible MgO catalyst in the reboiler will catalyze the the fiberglass bags. At higher column flow rates, however, the
reverse reaction (i.e., the decomposition of DAA to form resin catalysts held within fiberglass bags benefit from improved
acetone). As the concentration of DAA in the reboiler increases, mass transfer, resulting in a higher production of DAA than
the rate of this reverse reaction would increase. If there were the coated saddles. The coated saddles however provided a
significant amounts of catalyst in the reboiler due to leaching higher selectivity to DAA than the resin catalysts in the CD
from the coated saddles, the decomposition of DAA would result experiments due to the reduced external mass transfer resistance
in a decrease in DAA productivity rather than the observed and the intraparticle diffusional limitations of the resins.
increase in DAA concentration. The fact that the productivity Therefore, coated catalytic saddles would be particularly
of DAA continued for 24 h indicates that very little (if any) of advantageous for reaction systems where high selectivity is
the coating layer was removed from the saddles over the course required.
of the experiment. As well, no solids were observed in the Slight deactivation of the coated saddles apparently occurred
reboiler when the experiment was terminated, further supporting in the CD experiments when operated for extended periods (i.e.,
the claim that the coating did not leach off during the course of 24 h) while the productivity of both DAA and MO over a 12 h
the experiment. Whether or not the coating could remain intact period remained constant. Erosion of the coated catalyst did
under our CD experimental conditions for periods significantly not occur over the 24 h of CD experiments.
longer than 24 h is unknown.
Acknowledgment
4. Conclusions
The authors acknowledge the financial support of the Natural
A method has been developed for the production of catalyti- Sciences and Engineering Research Council of Canada (NSERC)
cally active coated ceramic distillation saddles for use in CD. in the form of research grants and a scholarship for Greg
Mg(OAc)2 was deposited onto ceramic Norton saddles using a Dechaine.
Ind. Eng. Chem. Res., Vol. 47, No. 23, 2008 9313
Literature Cited (17) Di Cosimo, J. I.; Diez, V. K.; Apesteguia, C. R. Synthesis of a,β-
unsaturated ketones over thermally activated Mg-Al hydrotalcites. Appl.
(1) Ng, F. T. T.; Rempel, G. L. Catalytic Distillation. Encyclopedia of Clay Sci. 1998, 13, 433.
Catalysis; Wiley Interscience: New York, 2003; Vol. 2, p 477. (18) Chakrabarti, S.; Ganguli, D.; Chaudhuri, S.; Pal, A. K. Crystalline
(2) Taylor, R.; Krishna, R. Modeling Reactive Distillation. Chem. Eng. Magnesium Oxide Films on Soda Lime Glass by Sol-Gel Processing. Mater.
Sci. 2000, 55, 5183. Lett. 2002, 54, 120.
(3) Oudshoorn, O.; Janissen, M.; van Kooten, W.; Jansen, J.; van
Bekkum, H.; van den Bleek, C.; Calis, H. A Novel Structured Catalyst (19) Shukla, S. K.; Parashar, G. K.; Mishra, P.; Yadav, B. C.; Shukla,
Packing for Catalytic Distillation of ETBE. Chem. Eng. Sci. 1999, 54, 1413. R. K.; Bali, L. M.; Dubey, G. C. “SOL-GEL” Derived Thin Film of
(4) Mehrabani, A.; Akbarnejad, M. M.; Hosseini, H. Structural Catalytic Magnesium Oxide and its Use in Humidity Sensor. Physics of semiconductor
Packing for Reaction-Distillation Columns. Ind. Eng. Chem. Res. 2002, 41, devices: proceedings of the Eleventh International Workshop on the Physics
5842. of Semiconductor Devices; Dec 11-15, 2001, The International Society
(5) Beers, A. E.; Nijhuis, T.; Aalders, N.; Kapteijn, F.; Moulijn, J. BEA for Optical Engineering (SPIE), 2001, p 1203.
Coating of Structured Supports - Performance in Acylation. Appl. Catal., (20) Choi, H.; Hwang, S. Sol-gel-derived Magnesium Oxide Precursor
A 2003, 243, 237. for Thin-Film Fabrication. J. Mater. Res. 2000, 15, 842.
(6) Podrebarac, G. G.; Ng, F. T. T.; Rempel, G. L. A Kinetic study of (21) Menon, M.; Bullard, J. W. Constrained Phase Evolution in Gel-
the aldol condensation of acetone using an anion exchange resin catalyst. Derived Thin Films of Magnesium Oxide. J. Mater. Chem. 1999, 9, 949.
Chem. Eng. Sci. 1997, 52, 2991.
(22) Ho, I.-C.; Xu, Y.; Mackenzie, J. D. Electrical and Optical Properties
(7) Craven, E. C. The alkaline condensation of acetone. J. Appl. Chem.
of MgO Thin Film Prepared by Sol-Gel Technique. J. Sol-Gel Sci. Technol.
1963, 13, 71.
1997, 9, 295.
(8) Bakker, R.; Hangx, G.; Kwant, G.; Maessen, H.; Markusse, P.
(23) Reagent Chemicals: American Chemical Society Specifications, 9th
Reaction and catalyst deactivation kinetics of the aldol condensation of
ed.; Oxford University Press: New York, 2000.
acetone. Intelligent Column Internals for Reactive Separations (INTINT) ,
2001. (24) Pemberton, E.; Card, S.; Craven, E. Observations on the Testing
(9) Di Cosimo, J. I.; Apesteguia, C. R. Study of the catalyst deactivation of Acetone. J. Soc. Chem. Ind., London 1935, 54, 163T.
in the base-catalyzed oligomerization of acetone. J. Mol. Catal., A: Chem. (25) Brinker, J. C.; Scherer, G. W. Sol-Gel Science: The Physics and
1998, 130, 177. Chemistry of Sol-Gel Processing; Academic Press: San Diego, CA, 1990.
(10) Podrebarac, G. G.; Ng, F. T. T.; Rempel, G. L. The production of (26) Touloukian, Y. S.; Kirby, R. K.; Taylor, R. E.; Lee, T. Y. R.
diacetone alcohol with catalytic distillation Part I: Catalytic distillation Thermal Expansion: Nonmetallic Solids; IFI-Plenum: New York, 1977.
experiments. Chem. Eng. Sci. 1998, 53, 1067. (27) Haber, J. Manual on Catalyst Characterization. Pure Appl. Chem.
(11) Podrebarac, G. G.; Ng, F. T. T.; Rempel, G. L. The production of 1991, 63, 1227.
diacetone alcohol with catalytic distillation Part II: A rate-based catalytic (28) Missen, R. W.; Mims, C. A.; Saville, B. A. Chemical Reaction
distillation model for the reaction zone. Chem. Eng. Sci. 1998, 53, 1077. Engineering and Kinetics; John Wiley & Sons; New York, 1999.
(12) Zhang, G.; Hattori, H.; Tanabe, K. Aldol addition of acetone, (29) Froment, G. F.; Bischoff, K. B. Chemical Reactor Analysis and
catalyzed by solid base catalysts: magnesium oxide, calcium oxide, strontium Design, 2nd ed.; John Wiley & Sons: New York, 1990.
oxide, barium oxide, lanthanum (III) oxide and zirconium oxide. Appl. Catal. (30) Podrebarac, G. G. DeVelopment and Modeling of a New Catalytic
1988, 36, 189. Distillation Process; University of Waterloo: Waterloo, Ontario, Canada,
(13) Kustrowski, P.; Sulkowska, D.; Chmielarz, L.; Dziembaj, R. Aldol 1996.
condensation of citral and acetone over mesoporous catalysts obtained by (31) Matsuda, T.; Tanabe, J.; Hayashi, N.; Sasaki, Y.; Miura, H.;
thermal and chemical activation of magnesium-aluminum hydrotalcite-like Sugiyama, K. Properties of magnesium oxides prepared from various salts
precursors. Appl. Catal., A 2006, 302, 317. and their catalytic activity in 1-Butene isomerization. B. Chem. Soc. Jpn.
(14) Aramendia, M. A.; Borau, V.; Jimenez, C.; Marinas, A.; Marinas, 1982, 55, 990.
J. M.; Ruiz, J. R.; Urbano, F. J. Magnesium-containing mixed oxides as
basic catalysts: Base characterization by carbon dioxide TPD-MS and test (32) Porter, K. E.; Templeman, J. J. Liquid Flow in Packed Columns
reactions. J. Mol. Catal., A: Chem. 2004, 218, 81. III: Wall Flow. Trans. Inst. Chem. Eng. 1968, 46, T86.
(15) Prinetto, F.; Ghiotti, G.; Durand, R.; Tichit, D. Investigation of
Acid-Base Properties of Catalysts Obtained from Layered Double Hydrox- ReceiVed for reView January 2, 2008
ides. J. Phys. Chem. B 2000, 104, 11117. ReVised manuscript receiVed September 14, 2008
(16) Tichit, D.; Naciri Bennani, M.; Figueras, F.; Tessier, R.; Kervennal, Accepted September 23, 2008
J. Aldol condensation of acetone over layered double hydroxides of the
meixnerite type. Appl. Clay Sci. 1998, 13, 401. IE800009U