Vapor phase Beckmann rearrangement using high silica zeolite catalyst

Article abstract of DOI:10.1039/b316810g

Vapor phase Beckmann rearrangement of cyclohexanone oxime to ε-caprolactam has been studied using high silica zeolite catalysts. Catalysts with different crystal sizes and gel-ageing times have been activated by ionic exchange in different conditions by means of a highly basic solution and a nearly neutral solution both containing ammonium salts. Samples have been calcined at different temperatures in order modify the number of defective sites. We observed that samples exchanged by means of a highly basic solution (pH > 10)^1.2 and calcined at a relatively lower temperature (450°C) show the most interesting catalytic results. X-ray powder diffraction patterns of these samples show² retention of the unit cell symmetry (orthorhombic cell) if compared to the dried sample. NH ₃-TPD confirms the low acidity of high silica zeolites, however a higher amount of desorbed ammonia is observed for the samples exchanged at higher pH and calcined at 450°C. Due to silanol nests the IR spectra of the same samples show the formation of Si-NH₂ bonds which are absent in the same material exchanged by other methods. Such sites seem to promote the high stability of the high silica zeolite catalysts also to the regeneration which is needed to remove the heavy carbonaceous compounds from the catalyst surface.

Full text of DOI:10.1039/b316810g

View Article Online / Journal Homepage / Table of Contents for this issue
R E S E A R C H P A P E R
Vapor phase Beckmann rearrangement using high silica
zeolite catalyst
a
L. Forni, G. Fornasari, G. Giordano, C. Lucarelli, A. Katovic, F. Trifir o` ,
a
b
a
b
a
c
C. Perri and J. B. Nagy
c
a
Dep. Chemical Industry and Materials, University of Bologna, V.le Risorgimento 4,
I-40136 Bologna, Italy. E-mail: forni@ms.fci.unibo.it; fornasar@ms.fci.unibo.it;
c_lucarelli@libero.it; trifiro@ms.fci.unibo.it; Fax: þ39-051-2093680;
Tel: þ39-051-2093677
b
c
Dep. Chemical Engineering and Materials, University of Calabria, Via Bucci,
I-87030 Rende (CS), Italy. E-mail: ggiorda@unical.it; katovic@unical.it
NMR Laboratory, University of Namur, rue de Bruxelles 61, B-5000 Namur, Belgium.
E-mail: janos.bnagy@fundp.ac.be
Received 23rd December 2003, Accepted 4th February 2004
F|rst published as an AdvanceArticle on the web 9th March 2004
Vapor phase Beckmann rearrangement of cyclohexanone oxime to e-caprolactam has been studied using high
silica zeolite catalysts. Catalysts with different crystal sizes and gel-ageing times have been activated by ionic
exchange in different conditions by means of a highly basic solution and a nearly neutral solution both
containing ammonium salts. Samples have been calcined at different temperatures in order modify the number
1
,2
of defective sites. We observed that samples exchanged by means of a highly basic solution (pH > 10) and
ꢀ
calcined at a relatively lower temperature (450 C) show the most interesting catalytic results. X-ray powder
diffraction patterns of these samples show retention of the unit cell symmetry (orthorhombic cell) if compared
2
to the dried sample. NH
desorbed ammonia is observed for the samples exchanged at higher pH and calcined at 450 C. Due to silanol
3
-TPD confirms the low acidity of high silica zeolites, however a higher amount of
ꢀ
nests the IR spectra of the same samples show the formation of Si–NH bonds which are absent in the
2
same material exchanged by other methods. Such sites seem to promote the high stability of the high silica
zeolite catalysts also to the regeneration which is needed to remove the heavy carbonaceous compounds
from the catalyst surface.
on the understanding of the effect of the activation method
Introduction
on the structural arrangement of silicalite-1 and on its cata-
lytic performances. One of the methods we applied induces
a different unit cell geometry organization and a different
acid site distribution.
The last step of caprolactam commercial production is
obtained, nowadays, in liquid phase by means of Beckmann
rearrangement of cyclohexanone oxime, with oleum as pro-
moter. Such a process is energy consuming and co-produces
3
stoichiometric ammonium sulfate quantities. Thus driving
forces of new researches are the tuning of alternative paths
with a lower environmental impact (no ammonium sulfate
production and no sulfuric acid use) and energy saving
Experimental
High silica MFI zeolite materials were prepared according to
1
1
(
volume of solvents used).
lower number of steps of reaction and decrease of the
5
procedures reported in literature. Four samples with different
crystal sizes (2.5 mm, 10 mm, 30–50 mm and 70 mm) were
studied in detail. The samples were calcined at two different
–7
The first researches started in
the sixties but only recently Sumitomo Inc. found that high
siliceous silicalites, a peculiar MFI structure, are highly active
and selective towards Beckmann rearrangement in vapor
ꢀ
temperatures: lower temperature (about 450 C) and higher
ꢀ
temperature (about 550 C). Then they were activated, to get
4
,8
phase. Said catalysts show long-lasting tests and for this
4 4 3
the protonic form, using NH Cl or alkaline-NH NO solution
by usual procedure. Catalysts were characterized by means
of different techniques. Catalyst texture was determined by
ꢁ1
reason a 5000 t year -demonstration plant has been started,
however by the end of year 2003 the build-up of a produc-
tion plant, in Japan, has been announced. Main physico-
9
ꢀ
nitrogen adsorption at ꢁ176 C with a Micrometrics Porosi-
ꢀ
chemical characteristics of high-silica silicalites are silanol
nests which contribute, with a relatively low acidity, to the
meter ASAP 2010 after treatment of the samples at 200 C
under vacuum. X-ray powder diffraction (XRD) was carried
8
,10
ꢀ
good selectivity of the final product.
Even though recent
papers describe the possible reaction mechanism of cyclohexa-
out by a Philips X’pert Pro diffractometer between 4 and 40
2y angle. IR spectra were obtained on a Nicolet Nexus FT-
IR spectrometer equipped with a MCT (Hg/Cd/Tl) detector
being the IR source made of CSi. Solid state Magic Angle
Spinning (MAS-NMR) spectroscopy was used to characterize
the local Si environments in the samples. Spectra were
recorded on a Bruker MSL 400 spectrometer working at 79.4
8
none oxime on high siliceous silicalites little is still available
on the reaction mechanism. In the present work we study the
effect of physico-chemical parameters of silicalite-1 catalyst on
the vapor phase Beckmann rearrangement. We investigate
on the role of structural defective sites and crystal dimensions
on caprolactam selectivity and the effect of acid sites distribu-
tion on catalytic performances. This work focuses, especially,
2
9
MHz ( Si) a pulse length of 6 ms and a repetition time between
acquisition of 60 s, the reference at 0 ppm being taken as
1
842
P h y s . C h e m . C h e m . P h y s . , 2 0 0 4 , 6 , 1 8 4 2 – 1 8 4 7
T h i s j o u r n a l i s Q T h e O w n e r S o c i e t i e s 2 0 0 4

View Article Online
Si(CH
performed in the same MAS-NMR conditions with a contact
time of 5 ms. NH -Temperature programmed desorption
NH -TPD) was carried out on a 1100 Thermofinnigan
TPD/RO instrument, the catalyst was pre-treated from 20 to
3
)
4
(TMS). Cross-polarization NMR (CP-NMR) was
Results and discussion
Catalyst physical-chemical properties
3
(
3
By means of X-ray powder diffraction we observed the typical
silicalite-1 pattern. All dried samples show reflections attribu-
ted to the template molecules and a single reflection at 24.4
ꢀ
ꢀ
ꢀ
4
50 C with an isotherm at 450 C for 60 min and adsorption
ꢀ
occurred at 100 C followed by desorption step from 100 to
2y angle typical of the silicalite-1 orthorhombic unit cell. After
calcination we observe a double reflection, at the same 2y
angle, attributed to a change of the unit cell towards a mono-
clinic symmetry which is thermally more favored than the
ꢀ
ꢁ1
.
450 C at a rate of 10 deg min
Beckmann rearrangement of cyclohexanone-oxime to capro-
lactam was carried out in a glass fixed-bed microreactor,
packed with inert material (glass spheres). The reaction tem-
1
2
previous one as reported in Fig. 1.
All catalysts retain monoclinic symmetry, but after activa-
ꢀ
ꢀ
perature ranged between 300 C and 400 C and weight hourly
ꢁ1
ꢀ
space velocity (WHSV) was 1.2 h . A solution of cyclohexa-
none-oxime, methanol and toluene, in a flow of nitrogen,
was utilized as the feed. The products were determined by
gas chromatography.
tion the samples calcined at 450 C and treated by an alkaline
NH NO solution modify again the unit cell towards an
4
3
orthorhombic symmetry. While the samples activated by a
NH Cl solution maintain the monoclinic symmetry. In Fig. 2
4
we display the unit cell evolution of the same sample after dif-
ferent thermal treatments and activation procedures. The same
behavior as Fig. 2 is observed for all samples (1792, 1793,
Catalyst preparation
1
795, 1796).
We suggest that thermal treatment at 450 C followed by
exchange with a NH NO alkaline solution promotes a struc-
The investigated catalysts were prepared according to the
composition reported in Table 1.
ꢀ
4
3
The samples were calcined in order to eliminate the tem-
plate. After calcination they were exchanged by aqueous solu-
tions in order to obtain the protonic form and finally calcined
again to stabilize the structure. The catalyst nomenclature is
arranged according to three numbers (xyx); the first number
tural rearrangement towards an orthorhombic unit cell sym-
metry which usually is less favored thermally, in silicalite-1
materials, than the monoclinic one. As a matter of fact by
FT-IR spectroscopy we verify that all samples calcined at
ꢀ
4
50 C and activated by the NH
4
NO
3
alkaline solution
ꢀ
ꢁ1
refers to the calcination temperature (1 ¼ 450 C, 2 ¼
(ꢁ111 samples) show a band at 953 cm typical of Si–N
ꢀ
13
stretching mode, as shown in Fig. 3.
5
50 C), the second number indicates the aqueous exchange
solution (1 ¼ NH NO , 2 ¼ NH Cl), the third number
4
3
4
The Si–NH2 bond is characterized by other vibrations at
540 and 3450 cm which are probably hidden by the wide
ꢁ
1
represents the last calcination temperature which is always
the same as the first one. Thus, for example, catalyst
3
band between 3600 and 3200 cm . The band at about 953
ꢁ
1
ꢀ
ꢁ1
1793-212 has been calcined at 550 C and exchanged by a
NH NO solution. The different activation solutions are char-
cm is typical of Si–O–M (where M ¼ heteroatom) vibra-
1
4
4
3
tion, and in silicalite-1 this vibration could be also attributed
to the Si–O–Na bond which yet is not present in activated sam-
ples as the ꢁ111 catalysts. Hence, FT-IR spectra of samples
acterized by a relevant pH difference, pH > 9 for NH
and pH ¼ 7 for NH Cl.
4 3
NO
4
exchanged by a NH
4
Cl solution (ꢁ121) do not show such a
vibration, see Fig. 3b, this result further confirms our findings.
The evaluation of OH stretching signals shows one broad
band with maximum at 3400 cm and a shoulder at 3617
Table 1 Silicalite-1 composition and crystal size. TPABr: tetrapropyl-
ammoniumbromide
ꢁ1
ꢁ
1
ꢁ1
5,15
cm and one IR signal at 3723 cm . These bands are attrib-
uted, according to the literature,
Crystallite
diameter
to nests of silanols,
bridged hydroxyl and terminal silanols, respectively. We also
investigated the Si chemical neighborhood by solid state
Catalyst
Starting gel composition
1
792
793
0.08Na
0.08Na
2
O–0.08TPABr–1SiO
O–0.08TPABr–1SiO
2
–20H
–20H
2
O
O.Gel
ageing for 5 days at room temperature.
10 mm
2.5 mm
29
NMR spectroscopy. By Si MAS-NMR we were able to dis-
1
2
2
2
tinguish the defective sites from the perfect tetrahedral SiO4 ,
These defective sites represent a small percentage of total
silicon atoms present in the lattice. As a matter of fact we
measured a maximum of 8 defective sites per unit cell with
1
795
796
0.04Na
0.02Na
2
O–0.08TPABr–1SiO
O–0.08TPABr–1SiO
2
–20H
–20H
2
O
O
30–50 mm
70 mm
1
2
2
2
Fig. 1 XRD pattern of 1793 catalyst: evolution upon thermal treatment and activation method to obtain silicalite-1 protonic form. + template,
orthorhombic, S monoclinic cell.
˙
T h i s j o u r n a l i s Q T h e O w n e r S o c i e t i e s 2 0 0 4
P h y s . C h e m . C h e m . P h y s . , 2 0 0 4 , 6 , 1 8 4 2 – 1 8 4 7
1843

View Article Online
Fig. 2 XRD patterns for catalyst 1792 after calcination and activation, ˙ orthorhombic, S monoclinic.
ꢀ
respect to the 32 theoretical defective sites of silicalite-1 unit
cell. Si MAS-NMR shows a higher defective sites percentage
Since samples treated at 450 C and activated by a NH
4
NO
3
2
9
alkaline solution show a different crystallographic unit cell
(orthorhombic symmetry) and a higher defective sites and
H-bonded silanols distribution, we suggest that: (i) calcination
temperature has a determinant effect on the total number of
defective sites (higher calcinations temperatures promote sila-
ꢀ
for samples calcined at 450 C and activated by alkaline
2
9
4 3
NH NO solution. However using Si–CP–NMR we could
detect two different kinds of defective sites. The first one with
a chemical shift of ꢁ93 ppm is attributed to terminal silanols,
the second one at ꢁ103 ppm is related to nests of silanols or
bridegd silanols. While ꢁ113 ppm is the chemical shift of
nols condensation, with elimination of H
the pH of the NH NO solution is of basic importance for
the formation of H-bonded silanols for NH reacts with
groups (as demonstrated by
2
O molecules), (ii)
4
3
SiO
4
groups. Such sites are represented in Fig. 4.
3
2
9
By Si–CP–NMR analysis we detected, in general, a higher
amount of H- bonded silanols, with chemical shift at ꢁ103
ppm, than terminal silanols at ꢁ93 ppm. In addition, regard-
less of calcination temperatures, samples activated by the alka-
Si–O–Si bonds to give Si–NH
IR-spectroscopy).
2
5
,16
Terminal and H-bonded silanols, according to literature
determine the acidity of silicalite-1 materials whose peculiarity
is the presence of weak Brønsted acid sites. By NH -TPD we
determined the strength, the amount and distribution of
Brønsted acid sites (Figs. 5 and 6).
We observe three peaks. Two of them are observed during
line NH
percentage. Samples with the same active site amount, but
treated with NH Cl, are characterized by a lower H-bonded
silanol percentage than samples treated with NH NO . For
4
NO
3
solution show a higher H-bonded silanol
3
4
4
3
ꢀ
example, both 1792-212 and 1792-222 samples have three
defective sites per unit cell, however the ratio between
H-bonded silanols and terminal silanols is 10.6 and 1.52,
respectively.
In table 2 we report the percentage of defects per unit cell
and the ratio between H-bonded silanols and terminal silanols,
this ratio represents a relative value for it is referred to the
percentage of defective sites per unit cell.
temperature rate, in the region between 180–200 C and 250–
ꢀ
300 C, respectively. The third one is located in the isothermal
ꢀ
zone at 450 C and is caused by diffusion phenomena and can
ꢀ
be ascribed to the peak at 250–300 C. Hence we attribute the
peaks to terminal silanols (very weak acid sites) and H-bonded
29
Fig. 4 Scheme representing the silanols groups detected by SiCP-
NMR spectroscopy.
Table 2 Percentage of defects per unit cell and H-bonded vs. terminal
silanol ratio
Defective
sites per unit
cell I (%)
I
I
H-bonded
/
Sample
I
ꢁ103,H-bonded
I
ꢁ93,terminal
terminal
1
1
1
1
1
1
1
793-111
793-121
793-212
793-222
792-111
792-121
792-212
4
3
3
1
8
3
3
3
6
4
33
48
54
34
50
42
53
41
48
57
1
33.0
3.4
14
7
7.7
2.4
14
8
6.2
2.1
20
5
10.6
1.5
ꢀ
Fig. 3 FT-IR of silicalite-1 sample (a) after calcination at 450 C and
1792-222
27
2
activation by a NH
4
NO
3
solution (pH > 10) (b) after calcination at
Cl solution (pH ꢂ 8). The band at
bondings.
1
1
795-111
795-212
24.0
19.0
ꢀ
4
9
50 C and activation by a NH
53 cm represents the stretching of Si–NH
4
3
ꢁ1
2
1
844
P h y s . C h e m . C h e m . P h y s . , 2 0 0 4 , 6 , 1 8 4 2 – 1 8 4 7
T h i s j o u r n a l i s Q T h e O w n e r S o c i e t i e s 2 0 0 4

View Article Online
Table 3 Total catalyst acidity and acid site distribution, very weak
(
terminal silanols) and weak (H-bonded) silanols
ꢁ
5
Acidity/10
Very weak
acid sites (%)
Weak acid
sites (%)
ꢁ1
ꢃ mol NH
3
g
1
1
1
1
1
1
1
1
1
1
1
1
1
793-111
793-121
793-212
793-222
792-111
792-121
792-212
792-222
795-111
795-212
796-111
796-121
796-222
7.15
5.53
5.07
5.30
7.43
7.00
5.63
4.02
7.80
5.86
4.92
4.55
2.19
25
35
30
39
20
30
27
32
28
35
22
27
30
75
65
70
61
80
70
73
68
72
65
78
73
70
ꢀ
Fig. 5 NH
4
with NH NO
3
-TPD profiles for silicalite-1 catalyst treated at 450 C
alkaline solution.
3
silanols (weak acid sites), respectively, in agreement with the
NMR results.
In Table 3 we report the total amount of desorbed ammonia
per gram of catalyst and an evaluation of acid site distribution.
We observe low total acidity for all catalysts, however the
ꢀ
samples treated at 450 C and activated by a NH
4
NO
3
alkaline
solution (-111 samples) are characterized by a higher total
acidity than the same samples treated in the other conditions.
In addition, the evaluation of acid site distributions, on the
basis of the areas of the peaks distinguished by TPD profiles
shows, for all samples, a higher amount of weak acid sites
(H-bonded silanols) with respect to very weak acid sites (term-
inal silanols). Said distribution is especially remarkable for the
111 catalysts, this result is in agreement with the NMR data.
In particular, sample 1793-111 shows a relative higher acid-
ity, when compared to other 1793 catalysts, the first TPD peak
Cyclohexanone oxime conversion reaches a maximum (90%)
for sample 1793-111 with the lowest crystal size (2.5 mm) and
undergoes a substantial decrease with increasing crystal dimen-
sion. A different behavior is observed for caprolactam selectiv-
ity, the minimum value (90%) is observed for catalyst with
crystals size 2.5 mm (1793-111), then the selectivity increases
progressively to achieve 100% with sample 1796-111. These
results are attributed to the different distribution of silanols
on the internal and external catalyst surface. Beckmann rear-
rangement to caprolactam is promoted by H-bonded silanols
-
ꢀ
(
terminal silanols) has its maximum at 180 C, while a shoulder
ꢀ
attributed to H-bonded silanols is detected at 290 C. 1793-212
is characterized by the same behavior as sample 1793-111,
however it shows a lower acidity which, according to Zecchina
et al., may be attributed to the decrease of stronger vicinal
silanols because of the higher number of hydroxyl-group con-
5
,15
which, according to Zecchina and Dalloro,
are located in
1
5
the internal surface of the crystals, while terminal silanols
which are mainly distributed in the external surface affect the
non-selective conversion of cyclohexanone oxime. As a matter
of fact by increasing crystal size dimensions, the ratio between
internal/external surface increases, thus a higher number of H-
bonded sites are available for the selective conversion of cyclo-
hexanone oxime. Hence sample with the lowest crystal size
dimension (2.5 mm) displays the higher conversion but the
lower caprolactam selectivity because of the relative higher
amount of terminal silanols with respect to the amount of
H-bonded silanols. We observed the formation of the follow-
ing by-products: cyclohexanone, cyclohexen-1-one, trans pen-
ten-nitrile, aniline whose selectivities are rather low and have
a value, for each component, set around 1–2.5%.
ꢀ
densations during calcination at 550 C. Catalysts 1793-222
has a higher percentage of very weak silanols, such a result
is related to the higher calcinations temperature which is
responsible for the decrease of the number of defective sites,
and by the method of catalyst activation. This hypothesis is
supported by the decrease of total number of silanol groups
and by the higher percentage of terminal–OH groups in sample
1
793-121 than in 1793-111. These results are confirmed also by
the characterization of the other samples. Sample 1796 shows
the same silanol distributions observed for the other materials,
but its acidity is consistently lower. This is caused by the low
number of defective sites which are affected by the low miner-
The same cyclohexanone oxime conversion behavior is
observed also for catalysts -222 (1792-222, 1793-222, 1796-
2
alizing agent quantity (Na O) used during its preparation.
2
22), its values are always lower than 35% as illustrated in
Fig. 8.
On the contrary, caprolactam selectivity shows a different
Catalytic activity
Crystal dimensions play an important role on the Beckmann
rearrangement catalytic activity of silicalite-1. For this reason
four catalysts 1792-111, 1793-111, 1795-111, 1796-111 with dif-
ferent crystal sizes were tested in the reaction at the conditions
behavior, with increasing crystal size a decrease of caprolactam
selectivity is observed and the same behavior is shown by -121
and -222 samples. Usually caprolactam selectivity values are
ꢀ
reported in the experimental section at T ¼ 300 C. Cyclohexa-
none oxime conversion and caprolactam selectivity are
reported in Fig. 7 as a function of crystal size.
Fig. 7 Cyclohexanone oxime conversion and caprolactam selectivity
values as a function of crystal size for catalysts prepared according to
-111 method. Values are referred to the fourth hour of reaction.
ꢀ
Fig. 6 NH -TPD profiles for silicalite-1 catalyst treated at 550 C
3
4
with NH Cl alkaline solution.
T h i s j o u r n a l i s Q T h e O w n e r S o c i e t i e s 2 0 0 4
P h y s . C h e m . C h e m . P h y s . , 2 0 0 4 , 6 , 1 8 4 2 – 1 8 4 7
1845

View Article Online
Fig. 8 Cyclohexanone oxime conversion and caprolactam selectivity
values as a function of crystals size for catalysts prepared according
to -222 method. Values are referred to the fourth hour of reaction.
Fig. 10 Catalytic performances of catalyst 1793 as a function of
weak H-bonded silanols distribution. Values are referred to the fourth
hour of reaction.
much lower than those obtained on catalysts -111, as a matter
of fact, the highest value for -222 samples (80%) is observed for
catalyst 1793-222, the other catalysts show remarkably low
caprolactam selectivity values (about 58% for 1792-222 and
improves catalysts physico-chemical properties and leads to
better catalytic performances. On the contrary, the increase
of calcinations temperature decreases the total acidity and also
catalytic performances.
3
0% for 1796-222). The results for the -222 series are in good
In Fig. 10 we report caprolactam selectivity as a function of
the amount of H-bonded silanol groups.
1
7
agreement with those of Weitkamp et al. and are attributed
to the high calcination temperatures, which cause the conden-
sation of a higher number of silanols by elimination of water
molecules. Hence the number of defective sites decreases but
the catalyst is also less activated by the use of an almost neu-
By increasing H-bonded silanol percentage we observe an
increase of caprolactam selectivity from a value of 80%
(1793-222) to 90% (1793-111). These results confirm that active
sites towards caprolactam formation are H-bonded silanols,
which are located in the internal surface of the catalyst. The
tral NH Cl solution. The different behavior of selectivity may
4
ꢀ
be explained by a different acid site distribution between the
external and internal surfaces, due to the different treatments
of samples. The good catalytic activity observed for catalysts
catalytic tests were carried out at 300 C to discriminate the
performances among the different samples.
The effect of reaction temperature on catalyst life and
catalytic performances are reported in Table 4 for catalyst
1792-212.
-111 has to be ascribed mainly to the activation with a
NH NO solution for it is able to modify silicalite-1 structure
4
3
and promotes a higher H-bonded silanol amount than terminal
silanols, as reported in table 3.
Cyclohexanone oxime conversion increases with increasing
ꢀ
temperature of reaction and at 400 C it is complete, however
caprolactam selectivity increases from 85 to 90% when tem-
Silanols distribution plays an important role in catalyst acid-
ity thus we carried out some tests in order to relate the catalytic
behavior to samples acidity. We compared the catalytic results
obtained on the 1793 samples series. In Fig. 9 we report cyclo-
hexanone oxime conversion and the caprolactam selectivity as
a function of catalyst total acidity.
Apparently no linear correlation occurs between sample
total acidity and catalytic performance, however it is possible
to point out two couples of data which are better reciprocally
correlated.
ꢀ
ꢀ
perature is raised from 300 C to 350 C, then it decreases to
ꢀ
80% (400 C). This phenomenon can be ascribed to (i) a better
catalyst activation and (ii) an easier tars desorption from the
catalyst surface, nevertheless by increasing the temperature,
the thermodynamics of by-product formation is promoted
ꢀ
so, at 400 C, we observe a decrease of caprolactam yield
and verify the formation of higher quantities of heavy volatile
ꢀ
products. In addition, when temperature is raised from 300 C
we observe a relevant longer catalyst lifetime. This is attributed
to the improvement of catalytic performances and more
efficient heavy products desorption.
Samples calcined at the same temperature (1793-212, 1793-
ꢀ
ꢀ
2
22 at 550 C and 1793-111, 1793-121 at 450 C) and activated
by NH NO alkaline solution and NH Cl neutral solution,
In order to verify the stability of silicalite-1 samples, we per-
ꢀ
4
3
4
respectively, show different results. Catalyst 1793-212 is char-
acterized by higher catalytic performances than 1793-222.
The same behavior is observed also for catalysts 1793-111
and 1793-121 but the catalytic results are much improved.
formed some catalyst regeneration cycles at 450 C in air. Such
operation is required to eliminate tars adsorbed on catalyst
surface, since tars formation occurs mainly on silicalite-1 exter-
nal surface. The dimension of silicalite-1 channels prevents
heavy product formation through a shape selective action, thus
catalyst deactivation occurs mainly through pores blocking
and cavities inaccessibility. This operating condition allows a
good catalyst regeneration and remove the organic residues.
4 3
We verify that the activation by a NH NO alkaline solution
Table 4 Effect of reaction temperature on the catalytic activity for
1
792-212 sample. Values reported to the fourth hour of reaction
Cyclohexanone
oxime conversion selectivity
Caprolactam
Reaction
Caprolactam Life
ꢀ
temperature/ C (%)
(%)
yield (%)
time/h
a
b
c
300
350
400
59
98
85
90
80
50
88
80
7
40
100
40
a
b
Cyclohexanone oxime conversion 38%. Cyclohexanone oxime conversion
c
5%. Cyclohexanone oxime conversion 90%.
Fig. 9 Catalytic performances of catalyst 1793 as a function of total
acidity. Values are referred to the fourth hour of reaction.
7
1
846
P h y s . C h e m . C h e m . P h y s . , 2 0 0 4 , 6 , 1 8 4 2 – 1 8 4 7
T h i s j o u r n a l i s Q T h e O w n e r S o c i e t i e s 2 0 0 4

View Article Online
Crystals size deeply affects catalytic performances for it
Table 5 Catalytic performances for 1793-111 catalyst after one to
ꢀ
three regeneration cycles. Catalytic tests are carried out at 350 C,
values are referred to the fourth hour of reaction
modifies the ratio between internal and external surface.
Caprolactam selectivity increases with increasing crystal size
for -111 samples, whereas cyclohexanone oxime conversion
decreases. On the contrary, other samples (-121,-212,-222)
show a decrease of both cyclohexanone oxime conversion
and caprolactam selectivity and a deactivation rate consis-
tently higher. Such a behavior has been explained by the differ-
ent acid sites distribution between internal and external
surface. As a matter of fact silanol nests which are located in
the channels of silicalite-1 undergo a slower deactivation for
the small channels dimensions, hinder tars formation through
a shape selective activity. Terminal silanols are mainly located
at the external surface of the catalysts, promote cyclohexanone
oxime non-selective conversion thus they quickly deactivate.
After regeneration upon 1793-111 catalyst, the catalytic activ-
ity is unchanged for thermal treatment in air removes tars and
leaves the silicalite-1 structure unchanged, we propose that
Si–NH2 bondings give to the silicalite-1 structure a higher
thermal stability.
Fresh
99.9
I cycle
99.9
II cycle
99.9
III cycle
99.9
Cyclohexanone oxime
conversion (%)
Caprolactam
94.0
91.8
94.0
91.3
selectivity (%)
Catalytic results for sample 1793-111 are rather high and we
do not observe any relevant difference between catalytic per-
formances. In table 5 we report the results for catalyst 1793-
1
3
11 after three regeneration cycles with operating temperature
ꢀ
50 C.
The stability of catalytic activity is attributed to the Si–NH2
bondings which are co-ordinated to hydroxyl groups and leads
to a higher stability of silanol-nests since they are unaltered
ꢀ
after thermal treatment in air at 450 C for 10 h.
We verified the effect of temperature and we found that for
ꢀ
silicalite-1 at T ¼ 350 C the best equilibrium between capro-
lactam reaction activation energy and minimization of by-
products formation is reached, at this temperature the higher
caprolactam yield is obtained and the longer catalyst lifetime
achieved.
Conclusions
High silica MFI zeolite lattice is formed by tetrahedral silicon
oxides and are characterized by defective sites namely H-
bonded silanols (vicinal or nests of silanols) whose amount
and distribution is modified by the different thermal treatment
References
3
and activation methods. By NH -TPD analysis it is possible to
quantify catalysts total acidity and distinguish and quantify
different kinds of acid sites. Our silicalite-1 samples show a
1
2
3
M. Kitamura, H. Hichihashi and I. Tojima, US Pat. 5 212 302,
993.
1
G. P. Heitmann, G. Dahlhoff and W. F. Holderich, J. Catal.,
¨
ꢁ
5
ꢁ1
very weak acidity in the range 7.0–5.0 ꢃ mol NH g and
1999, 186, 12–19.
J. N. Armor, Catalysis of Organic Reactions, Chemical Industries,
3
ꢀ
weak and very weak acid sites. Samples calcined at 450 C
and activated by an alkaline NH NO solution are character-
1
8, Kosak J.R. Editor.
4
3
ꢁ
5
ꢁ1
4
5
H. Ichiashi and M. Kitamura, Catal Today, 2002, 73, 23–28.
J. Ritz, H. Fuchs, H. Kiezka and W. C. Moran, Ullmann’s
Encyclopedia of Industrial Chemistry, ed. F. T. Campbell et al.,
VCH, Weinheim, 1986, 5, 31–49.
J. N. Armor, Appl. Catal. A, 1999, 189, 153–162.
R. A. Sheldon, Appl. Catal. A, 1999, 189, 163–183.
C. Flego and L. Dalloro, Microporous Mesoporous Mater., 2003,
ized by a relatively higher acidity (7.15 ꢃ
mol g ) and a
higher ratio between weak acid sites and very weak acid sites.
Said catalysts are very active in vapor phase Beckmann rear-
rangement and complete cyclohexanone oxime conversion
and caprolactam yield higher than 95% are achieved. It is
6
7
8
ꢀ
also demonstrated that a calcination temperature of 550 C
6
http://www.sumitomo-chem.co.jp/cgi-bin/search/release_2e.cgi?
0, 263–271.
promotes H-bonded silanols condensation, hence a decrease
of the amount of defective sites and a decrease of catalytic
activity.
9
year ¼ 2000.
10 W. F. Holderich and G. Dahlhoff, Chem. Innov., 2001, 2, 29–40.
¨
11 Verified Synthesis of Zeolitic Materials, ed. H. Robson, Elsevier,
Amsterdam, 2001.
Catalysts activation with an alkaline solution of NH NO
4
3
modifies the unit cell symmetry from monoclinic to ortho-
rhombic through Si–NH bondings formation and allows the
1
2
3
Y. S. Ko and W. S. Ahn, Microporous Mesoporous Mater., 1999,
0, 283–291.
B. A. Morrow, I. A Cody and L. S. M. Lee, J. Phys. Chem., 1975,
9, 761–762.
2
3
retention of a higher H-bonded silanol to terminal silanol
ratio. On the basis of structural features and catalytic results
we confirmed that Beckmann rearrangement reaction is pro-
moted by H-bonded silanols and especially by silanol nests
which are mainly located in the catalyst cavities. Caprolactam
selectivity shows a dependence with amounts of silanol nests.
Sample 1796 which has been prepared with a low quantity of
1
7
14 S. P. Zhdanov, L. S. Kosheleva and T. I. Titola, Langmuir, 1987,
3, 960–967.
15 S. Bordiga, P. Ugliengo, A. Damin, C. Lamberti, G. Spoto, A.
Zecchina, G. Span o` , R. Buzzoni, L. Dalloro and F. Rigetti,
Top. Catal., 2001, 15, 43–52.
1
6
G. Centi, S. Pheratoner and F. Trifir o` , J. Phys. Chem, 1992, 96,
617–2629.
2
9
mineralizing agent shows a Si-MAS-NMR profile which is
not typical of silicalite-1 structures, in addition its total acidity
is lower than that of other catalysts.
2
17 H. Kath, R. Glaser and J. Weitkamp, Chem. Eng. Technol., 2001,
24, 150–153.
¨
T h i s j o u r n a l i s Q T h e O w n e r S o c i e t i e s 2 0 0 4
P h y s . C h e m . C h e m . P h y s . , 2 0 0 4 , 6 , 1 8 4 2 – 1 8 4 7
1847