\

w
^
M
w^^rw
^

Molecular Sieve Zeolites — Properties and

Applications in Organic Synthesis
 MOLECULAR SIEVE ZEOLITES

PROPERTIES AND APPLICATIONS IN

ORGANIC SYNTHESIS

^iinfflii
o <
ui o
o a-

BIBLIOTHEEK TU Delft
P 1914 5055

621756
 MOLECULAR SIEVE ZEOLITES

PROPERTIES AND APPLICATIONS IN

ORGANIC SYNTHESIS

PROEFSCHRIFT

TER VERKRIJGING VAN DE GRAAD VAN DOCTOR

IN DE lECHNISCHE WETENSCHAPPEN AAN DE
lECHNISCHE HOGESCHOOL DELFT, OP GEZAG
VAN DE RECrOR MAGNIUCUS IR H R \ AN
NAUTA LEMKE, HOOGLERAAR IN DE ALDELING
DER ELEKTROTECHNIEK, VOOR EEN COMMISSIE
UIT DE SENAAT TE VERDEDIGEN OP WOENSDAG
28 JUNI 1972 I F 16.00 UUR

DOOR

DIRK PIETER ROELOFSEN

scheikundig ingcnicur
sebori'n te Mcdan

DELFTSCHE UlTGEVERS MAATSCHAPPIJ N.V. DELFT 1972

/Q/^ SO^S-
 H

Dit proefschrift is goedgekeurd door de promotor

PROF. DR. IR. H. VAN BEKKUM

L
 CONTENTS
SUMMARY 7
CHAPTER I GENERAL INTRODUCTION 11
Introduction 11
Zeolitic molecular sieves 12
Applications of molecular sieves 15
Structure of molecular sieves used 16
Adsorption by molecular sieves 22
Shifting of chemical equilibria using
molecular sieves 28
Proton-exchanged type A molecular sieves 32
References 33
CHAPTER II SIMPLE TECHNIQUE TO DETERMINE VAPOR AD-
SORPTION CAPACITIES OF MOLECULAR SIEVES 35
CHAPTER m ESTER INTERCHANGE USING MOLECULAR SIEVES 41
Abstract 41
Introduction 41
Results and discussion 44
Conclusion 53
Experimental part 54
CHAPTER IV THE SYNTHESIS OF ENAMINES AND KETIMNES
USING MOLECULAR SIEVES 59
Abstract 59
Introduction 59
Results and discussion 60
Experimental part 64
CHAPTER V Part 1 SYNTHESIS OF ACETALS
USING MOLECULAR SIEVES 65
Abstract 65
Introduction 65
Results and discussion 66
Conclusions 72
Experimental part 72
Part 2 ACETALS OF SECONDARY ALCOHOLS 77
CHAPTER VI PROTON-EXCHANGED TYPE A
MOLECULAR SIEVES 81
Abstract 81
Introduction 81
Results and discussion 82
Experimental 91
SAMENVATTING 93

5
r

A Fellowship of the Netherlands Organization for the Advancement of

Pure Research (Z. W. O.) for a most stimulating stay of six months
at the Laboratory of Physical Chemistry of Prof. R. M. Barrer,
Imperial College, London, is acknowledged.

6
 SUMMARY

This thesis describes the application of molecular sieve zeol-'^es

to shift chemical equilibria (1) by the adsorptive removal of a r e a c -
tion product (D)-
K
A + B ^ ^ C + D (1)

° adsorbed ^^^
Two batch techniques have been used. In the extraction technique the
vapour of the boiling reaction mixture, entraining component D, is
condensed. The condensate is contacted with molecular sieve, and is
returned to the reaction mixture. In the suspension technique the
molecular sieve is suspended in the reaction mixture. In all r e a c -
tions investigated by us homogeneous or heterogeneous catalysis was
applied to accelerate the reaction of A and B. The interaction of the
homogeneous catalysts with the molecular sieves in the suspension
technique has been investigated, especially the exchange between
protons of acid catalyst and cations of type A sieve. The proton-
exchanged sieves thus formed in an organic medium appeared to be
quite stable.
Most of the worlc described has been published ^'^ or is in
press7. In fact, the chapters II-VI are reprints of publications 2-7.
In chapter I a general introduction is given. The structure of the
important molecular sieves is described and their adsorption pro-
perties are illustrated by several separations of organic compounds.
The application of molecular sieves in shifting chemical equilibria (1)
by adsorbing D is discussed.
In chapter 11^ the construction of a simple adsorption apparatus,
which enables a fast determination of the saturation capacities of
molecular sieves for vapours, is described.
In chapter III^ it is shown that aUcoxide-catalyzed ester inter-
change reactions proceed practically to completion when using type
A molecular sieves to adsorb the displaced alcohol. Examples are
given which cover a number of starting esters, displacing alcohols,
solvents, and ion-exchanged forms of type A sieve Several new t-
butyl esters were obtained.
Equilibrium constants were measured and rate studies made for
several representative ester interchange reactions.
The interaction of sodium and potassium isopropoxide catalyst
and sieve has been investigated. Interesting differences have been
found between type KA (3A), NaA (4A) and CaA (5A) sieves.

7
 The interfering dehydration of f-butyl alcohol by these sieves
and the influence of traces of water and triethylamine on this dehyd-
ration have also been studied
In chapter IV^ the synthesis of enamines and ketimines from the
corresponding carbonyl compounds and amines at room temperature
IS described using a commercial silica-alumina cracking catalyst to
catalyze the reaction and a suspended molecular sieve type A to ad-
sorb the water formed The mixture of sieve and silica-alumina can
be successfully regenerated Different solvents, catalysts and mole-
cular sieves have been tested. The combination of cyclohexane, silica-
alumina and molecular sieve CaA gives the best results, besides the
pure powder, pellets type CaA can also be used in combination with
silica-alumina catalyst. The moderate catalytic effect of pellets as
such, as reported in the literature, has been traced down to the bind-
ing agent present in pelletized material.
In chapter V, part P , the preparation of several acetals from
the corresponding aldehydes or ketones and alcohols using molecular
sieves to shift the equilibria by adsorption of the water formed, is
described. Two modifications of the suspension technique have been
applied with the mixture at 0°C or at room temperature.
The first modification involves the use of dual-function proton-
exchanged molecular sieves type Zeolon 500, AW 500, or mordenite.
The outer surface of the sieve catalyzes the acetal formation where-
as the inner surface adsorbs water
In the second modification molecular sieve and catalyst/>-
toluenesulphomc acid (/)TsOH) a r e added to the reaction mixture. In
this case proton exchange between catalyst and sieve occurs which
appeared to be fast with KA and NaA, much less pronounced with
CaA, Zeolon 500, and AW 500. All these sieves however can be used
because of the high rate of acetal formation. With KA and CaA essen-
tially quantitative conversions are obtained within a few minutes.
In chapter V, part 2'^, the synthesis of acetals of secondary alco-
hols and ketones or aldehydes, using the same technique, is des-
cribed. The sterically-hindered ketone acetals are difficult to p r e -
pare by conventional means
In chapter VI^ the interaction of NaA and CaA with/)TsOH in
dioxane has been investigated Proton-exchanged products were cha-
racterized by their crystallinity and their adsorption capacity for
water, hexane and cyclohexane. For comparison, proton-exchanged
samples prepared in aqueous solution of pH-> 5.0 were included in
this study
With NaA and anhydrous / J T S O H in dioxane a rapid proton in-
corporation up to 95^5 occurs in the presence of small amounts of
water or methanol. These proton-exchanged sieves have large ad-
sorption capacities. Quite unexpectedly the pores are widened com-
pared to those of NaA and CaA, even allowing for the adsorption of

8
cyclohexane in some cases. The stability after adsorption of water
is poor. However, after adsorption of hexane the sieves can be suc-
cessfully regenerated.
In the case of CaA in dioxane with anhydrous />TsOH and small
amounts of water or methanol, only residual sodium ions are ex-
changed for protons thereby limiting the exchange to about 20?J.
Therefore, these acid-treated CaA-sieves have properties similar
to those of untreated CaA.
Both NaA and CaA decompose in dioxane withpTsOH when the
amount of water exceeds the adsorption capacity of the sieve. Here
part of the sieve crystals break down completely.
Proton exchange in aqueous solution at pHi> 5.0 occurs rapidly
with NaA, only slowly with CaA. The products have poor adsorption
properties except at less than ca. 25% proton incorporation.

References:

1 D.P. Roelofsen, J. A. Hagendoormnid H. van Bekkiim, Chem. Ind.

(London) 1966, 1622.
2 D. P. Roelofsen, J. W. M. de Graaf, J. A. Hagendoorn, H. M. Ver-
schoor a.nd H. vanBekkvm, Rec. Trav. Chim. Pays-Has 89, 193
(1970).
3 D.P. Roelofsen, Anal. Chem. 43, 631 (1971).
4 D.P. Roelofsen. E.R.J.Wils dsiA H. vanBekkum, Rec. Trav. Chim.
Pays-Bas 90, 1141 (1971).
5 D. P. Roelofsen, E. R.J. Wils and H. vanBekkum, J. Inorg. Nucl.
Chem. 34, 1437 (1972).
6 D. P. Roelofsen and H. van Bekkiim, Rec. Trav. Chim. Pays-Bas 91,
605 (1972)."
7 D. P. Roelofsen and H. van Bekkiim, Synthesis, in press.

9
 CHAPTER I
GENERAL INTRODUCTION

Introduction
The use of zeolitic molecular sieves has been increased con-
siderably by the commercialisation of synthetic molecular sieves by
the Union Carbide Corporation in the early fifties. Present industrial
applications include their use as cracking catalysts, for the selective
adsorption.of »-paraffins to upgrade certain petrol fractions, and for
the exhaustive drying of gases and liquids.
While the term "molecular sieve" has been used more generally
for materials which possess small, not necessarily uniform, pores
such as certain carbons, porous glasses and Sephadex, we will use
it exclusively to describe natural and synthetic zeolites. These zeo-
lites are rigid crystalline aluminosilicates with high adsorption ca-
pacities. When adsorbed water is driven out, e. g. by a heat treat-
ment, only molecules small enough to enter the completely uniform
pores of 2-10 A in diameter (type KA, 3 A; type NaA, 4 A; type CaA,
5 A; type X and Y, 8-9 A: synthetic mordenite, 7 A) are adsorbed on
the extensive surface area (a typical value is 700 m^/g)/w the c r y s -
tals while larger molecules are excluded.
Synthetic molecular sieve is a powder-like substance consisting
of crystals of 0.1 to 10 micron. By adding a binding agent the powder
can be pelletized to simplify handling. Natural molecular sieves a r e
found as large deposits of chalk-lLke material.
In this thesis the application of synthetic and natural molecular
sieves to shift certain equilibria by the selective adsorption of one
of the reaction products (D) in the sieve is described:

K
A + B :^^ C + D

'— adsorbed
The components A, B and C of the reaction mixture are too large to
be adsorbed. Equilibria, in which D is water or a w-aliphatic alcohol,
have been studied using two different experimental techniques.
Catalysts were used to accelerate the reaction of A and B. Differen-
ces in the interactions of the various molecular sieves, when sus-
pended in the reaction mixture, with the catalysts have been studied.
As an interesting result of this study the preparation of quite stable
proton-exchanged type A sieves has been achieved. A promising way

11
to circumvent the problem of interaction of catalyst and sieve is the
use of solid catalysts, such as commercial silica-alumina cracking
catalyst for proton-catalyzed reactions.
An adsorption apparatus has been developed to allow a fast cha-
racterization of the properties of the molecular sieves.
In this introductional chapter a thorough description of structure
and properties of molecular sieves will be given followed by a dis-
cussion on the use of molecular sieves to shift chemical equilibria
and on the preparation of proton-exchanged sieves.

Zeolitic molecular sieves

Mineralogically silicate minerals are arranged in six classes.

The class of tectosilicates features a three-dimensional infinite frame-
work of SiO, - and AlO^-tetrahedra in which oxygen ions are shared by
neighbouring tetrahedra. The tectosilicate class may be subdivided in
several groups one of which is the zeolite group of minerals. A defini-
tion is given by Sinith^: "A zeolite is an aluminosilicate with a tetra-
hedral framework structure enclosing cavities occupied by large cat-
ions and water molecules, both of which have considerable freedom
of movement, permitting cation exchange and reversible dehydration".
The negative charge of the aluminosilicate structure, introduced by
incorporation of aluminium as AlO^-tetrahedra, is neutralized by
cations such as sodium, potassium or calcium. The alternation of
silicon and aluminium in the framework is governed by the rule of
Loewensteiir which states that two AlO^-tetrahedra never share the
same oxygen ion. The framework structure is schematically shown
in figure 1:

Si Al Si Si Al Si
y ^\ / ^\ y ^^ / ^\ / \ y\
0 0 0 0 0 0 0 0 0 0 0 0

Figure 1
Aluminosilicate framework of zeolites.

12
The radii of the constituing ions of the aluminosilicate framework
areSi**, 0.39 A; Al'% 0.57 A 0^', 1.35 A Both silicon and alumi-
nium fit neatly in the cavity formed by a dense tetrahedral arrange-
ment of the four surrounding oxygen ions. By a crystallographic
arrangement of the SiO.^- and A10.^-tetrahedra, a three-dimensional
aluminosilicate skeleton is formed containing cages. These cages
a r e connected by windows with a diameter, depending on the type
of molecular sieve, of 2-10 A Thus a pore system is formed which,
in the unactivated sieve, is filled with water molecules. After driving
out the water, the highly porous crystal will adsorb any guest mole-
cule which is small enough to enter the pore system. The number of
possible crystallographic arrangements of tetrahedra seems un-
limited^ 34 different types of natural zeolites have been identified,
while many synthetic zeolites with no natural counterparts have been
described.
In aqueous solution molecular sieves are stable in the pH range
of 5-12, zeolites with a low aluminium content, e g Si/Al larger
than 3, may withstand stronger acid solutions without structural
collapse Dehydration of molecular sieves is generally accomplished
by heating. All commercial zeolites may be heated to temperatures
of 600° C without structural collapse. Compounds adsorbed on mole-
cular sieves can be desorbed by a temperature increase, a lowering
of the pressure, or by displacement by another compound A recom-
mended, safe procedure for small-scale regeneration of molecular
sieve, which has been used in an organic medium, is as follows
the sieve is washed with methanol followed by water, to displace the
organic compounds and is subsequently dehydrated in an oven by step-
wise heating starting at room temperature and going up to 250-500^0.
The sieve is cooled in a desiccator without drying agent
General information on occurrence, structure, synthesis and
applications of molecular sieves is given by Hersli^, Gruhnei c s ^,
Eitel^ and Barrer^, and also in the proceedings of the two interna-
tional conferences on zeolites held in London in 1967" and in Wor-
cester in 1970^.

Natural molecular sieves

Large, beautiful crystals of natural molecular sieves, e g found
in basalts, have been reported in the literaturelO Because of their
size, in the order of several mm, and their purity they have been of
particular use in elucidating zeolite structures by single crystal X-
ray diffraction techniques. Also, work on ion exchange and on ad-
sorption kinetics has been successfully performed on these large
crystals. Unfortunately, because of their rareness they are not of
any commercial interest.
Extensive sedimentary deposits of micro-crystalline molecular
sieves, e.g. formed by alteration of volcanic tuff, have been found
in the United States and in New Zealand, Japan, East Africa and Ca-

13
nada. Of the 34 natural molecular sieves only six commonly oc-
c u r l l - 1 2 in the United States, namely analcime, chabazite (estim-
ated amount 105 tons), phiUipsite (5x10^ tons), mordenite (5x10^
tons), erionite (107 tons), and clinoptilolite (10^ tons). Very often
these deposits are easy to mine and prices of these natural zeolites
could become highly competitive!! with those of synthetic zeolites.
Most of these deposits, however, contain two or even more zeolites
and also non-zeolitic impurities.
Both the Union Carbide Corporation and the Norton Company
supply natural molecular sieves from stock. Union Carbide present-
ly offers AW 300 (in our opinion a mixture of clinoptilolite and
mordenite) and AW 500 (a mixture of chabazite and eriomte, see
chapter V) From the Norton Company, Zeolon 300 (natural m o r -
denite) and Zeolon 500 (a mixture of chabazite and eriomte) may be
obtained. Several mineral houses supply natural molecular sieves
on order

Synthetic molecular sieves

The synthesis of molecular sieves is generally accomplished
under hydrothermal conditions. For example, two of the more i m -
portant zeolites, type A and type X, are synthesized at temperatures
around 100 °C from an aqueous solution containing sodium aluminate,
sodium silicate and sodium hydroxide. After an induction period in
the order of 1 h the molecular sieve is formed as veiy small c r y s -
tals of 0.1 to 10 micron. The exact conditions such as concentration
of reactants, composition of basic materials, the temperature and
the reaction time, critically influence the quality of the product.
The crystal slurry is filtered and the sieve washed with water.
Ion-exchanged varieties are prepared by exchange in an aqueous
solution followed by washing with water
The molecular sieve is either dehydrated and sold as the pure
powder or mixed with a binding agent (typically 20 w f ) , extruded to
form pellets, and dehydrated by heating up to 600 °C. Molecular
sieve with binding agent is also available in small granules or in
beads. The binding agent is said to be an inert clay, however, we
noted that the binding agent showed a considerable catalytic activity
(chapters III and TV) for certain reactions.
Synthetic molecular sieves are manufactured iiile) alia by the
Union Carbide Corporation (Linde Division), the Norton Company
and the Davison Chemical Company (a division of W. R. Grace and
Company) in the United States, by the Farbenfabrik Bayer in West
Germany, by the Ceca Company and Pechiney in France, by Peter
Spence and Sons Ltd in England and by Uetikon in Switzerland
Synthetic molecular sieves are also produced in Eastern Europe
and in the U S.S.R
Prices of small (several kg) quantities of the more common
synthetic molecular sieves range between 10 and 60 Dutch guilders

14
per kg depending upon the manufacturer and the type of sieve. Prices
go down quickly when larger quantities are involved For instance,
ton lots of molecular sieve cost between 5 and 15 guilders per kg.

Applications of m o l e c u l a r sieves

A great variety of applications of molecular sieves a r e described

in the literature Some examples in the more important areas are
given below.
Adsorption. By the adsorption of small molecules which can selec-
tively enter the pore system of zeolites, excellent separations have
been accomplished, e g the industrial separation of branched and
unbranched paraffins using CaA sieve. The drying of liquid and gas-
eous streams by molecular sieves is another common application
both m industry and in laboratories Separations not based on the
size selectivity of zeolites have also been claimed, e g the separa-
tion of isomeric xylenes and the removal of hydrogen sulphide from
natural gas Molecular sieves are widely used as an adsorbent in
GSC to separate permanent gases Reactive, volatile, or toxic com-
pounds can be loaded on molecular sieve which acts as an inert
c a r r i e r . By heating or by displacement with another adsorbate, e g
water, the stored compound is released from the sieve.
Ion exchange. Zeolites may be used as high-capacity ion exchang-
e r s l 3 e g. for the recovery of radioactive ions from waste streams.
With very large ions an ion sieve effect may occur.
Catahsis. The crystalline character of the pore mouths and of the
internal surface of zeolites has attracted much attention The efforts
have been mainly concentrated on the large-pore zeolites type X and
type Y and mordenite. Much work has been done on the formation
and activity of hydroxyl groups. By incorporating a catalytically
active metal in the pore structure by ion exchange followed by reduc-
tion to the metal, a selective dual-function catalyst can be prepared.
Reviews on catalysis by zeolites have been given by Venuto'^^ and by
Voiiito and Landis ^^.
The high activity and high stability of certain large-pore zeolites,
after a suitable ion exchange and a stabilization-treatment, have
resulted in the widespread replacement of conventional cracking
catalyst by zeolite-containing catalyst. In the United States more
than 90^ of the refining capacity uses zeolite-based catalyst. In
Europe, due to a different market situation, the percentage is
roughly 10^. Zeolite-based metal catalysts play an important role
in paraffin isomerization.
Several types of selectivity may be distinguished when using
zeolite catalysts (0 Reactant selectivity. By using the internal s u r -
face of small-pore zeolites as the catalytic surface, highly selective
conversions are sometimes possiblel6. Thus the selective cracking
of w-paraffins from a mixture with branched paraffins is possible
with type A sieve which excludes the branched paraffins (;/) Product

15
selectivity. Here one of the products formed in the porous structure
is too large to escape from the pores. Thus in the alkylation of ben-
zene with /-butyl chloride^'^ using the commercial alkylation catalyst
SK500, a r a r e earth exchanged type Y sieve, l,3,5-tri-/-butylbenzene
was remarkably absent from the reaction mixture. Chemical break-
down of the sieve after the reaction showed that 1, 3, 5-tri-/-butyl-
benzene had been trapped in the cages (see also table II).(?//) Reaction
path selectivity. A reaction path selectivity, in which the crystalline
nature of the small zeolite-cages influences the reaction path is to be
expected but no clear-cut examples are known. The a.nii-Markownikoff
addition of hydrogen bromide to 1-alkenes using molecular sieves,
might form an examplelS.

Structure of m o l e c u l a r sieves used

Important properties'^ of the molecular sieves used in this

thesis have been collected in table I. The detailed structures of
these zeolites are discussed one by one.

Table I
Properties of molecular sieves used.
Type of Origin^ Typical No. of O-ions Apertures^ Water ads.
sieve Si/Al ratio forming in A cap. c in
windows g/100 g

A S 1 8 4.1 29
chabazited N 2 8 3.6x3.7 30
erionite^ N 3 8 3.6x5.2 21
mordenite S 5 12 6.7x7.0 15
X S 1.2 12 7.4 33
Y S 2.5 12 7.4 33

^ S-synthetic N-natural
^ As calculated from X-ray data and assuming an oxygen radius of
1.35 A. Minimum and maximum values of non-circular apertures
are given.
^ Water adsorption capacity of zeolite as mined or as synthesized in
g of water per 100 g of activated sieve.
'I Commercial Zeolon 500 and AW 500 a r e mixtures of chabazite and
erionite: see chapter V, part 1.

16
Type A molecular sieve
Type A molecular sieve is a synthetic zeolite without a natural
counterpart. The crystal structure was resolved by Reed and Breck^O
a.nd reiined by Broiissard sindShoe))iaker21^ Hon'ell22^ Smith and
Doirell23 and by Gramlich and Meier^^. It is synthesized in the so-
dium form (designation NaA or 4A) with an idealized unit cell com-
position of Na^j • (AlO^ )i2 • (SiCb )i2 • 27 H2O. By exchange of sodium L Ma'')
for potassiumfor about 45^, KA(3A)is obtained. Bij an 8O5S exchange
of sodium for calcium,' CaA (5A) is obtained.
A simplified model of the unit cell, showing only silicon and
aluminium ions in the vertices, is seen in figure 2 right; oxygen ions
are situated near the middle of each line. The structure is based on
truncated octahedra also called sodalite units (one such unit in figure
2 is shown, partly dashed). Each sodalite unit is linked (by the
square faces) to each of its six neighbours by four bridging oxygen
ions. Eight sodalite units surround a spherical a-cage with a dia-
meter of 11.4 A. The sodalite units themselves contain small ^-
cages of 6.6 A internal diameter. Each large a-cage may be entered
from six other a-cages through circular windows formed by eight
oxygt.i ions. Thus a three-dimensional pore system is formed.

Figure 2
Molecular sieve type A.
Right: unit cell build from sodalite units.
Left : line drawing of a-cage showing cation positions.

17
The j3-cages can be entered from the a-cages by very small molecu-
les only, e g water, through windows, formed by six oxygen ions,
with a crystallographic diameter of 2 2 A A schematic drawing of
the a-cage, again showing silicon and aluminium on the vertices, is
shown in figure 2 left In NaA, eight out of twelve sodium ions per
unit cell a r e situated in sites Si in the centers of the six-membered
rings. The four remaining cations are less localized. In the dehy-
drated sieve they a r e thought to be situated m sites Sn adjacent to
the eight-membered windows, controlling the access to the a-cages.
Because each eight-membered window is shared by two a-cages, four
out of SIX windows per a-cage have one Sjj cation each and the r e -
maining two windows have two Sjx cations each The Sxi cations p a r -
tially block the windows to give an effective pore diameter* of about
4 A in NaA When sodium is partly replaced by calcium the number
of cations decreases and the blocking by Sji cations is suspended,
thus widening the pores to about 5 A in CaA. In contrast, when s o -
dium IS partly replaced by potassium (ionic radii of 0 95 A and
1 33 A, respectively) the pores are reduced to about 3 A in KA.
In figure 3 a more detailed model of NaA, based on packing of
oxygen ions, shows one a-cage and the bridges connecting the soda-
lite units The white spheres are oxygen ions, the dark spheres
represent sodium ions The small spheres representing aluminium
and silicon ions are hardly visible, surrounded as they are by four
oxygen ions each Figure 3 illustrates the partial blocking of an
eight-membered window by an Sji cation Four Sj cations are also
shown

Chabazite
An idealized unit cell composition is Cag. (A102)-,2 (81^)24. 4OH2O,
but the sedimentary natural chabazitel2 generally contains sodium
and potassium in excess of calcium and magnesium while the Si/Al
ratio is higher than the idealized value of 2. A schematic line drawing
of the large ellipsoidal cage of chabazite is shown in figure 4 with
silicon and aluminium on the vertices Six distorted eight-membered
windows give access to six neighbouring cages, thus forming a three-
dimensional pore system.

* The effective pore diameter of a molecular sieve indicates the dia-

meter of the largest molecule which can still pass the restricting
windows This value is generally somewhat higher than the X-ray
determined value (table I) due to the non-rigidity of framework
oxygen ions and of sorbate molecules

18
 Figure 3
Space-filling model of molecular sieve type NaA.

Figure 4
Line drawings of the large cages of chabazite and erionite.
Left: chabazite. Right: erionite.

19
Erionite
The idealized unit cell composition is (Na,, Kj, Ca, Mg)^^. (A102)g.
(SiOg )2-7. 27 HjO. The large cage of erionite with its six distorted
eight-membered windows is shown in figure 4. A three-dimensional
pore system is formed by the connection of each cage to six others.

Mordenite
The idealized unit cell composition of the synthetic material is
Nag. (AIO2 )8 . (SiOg )^o • 24 H2 O. A model and a line drawing of the
pore system, showing silicon and aluminium only, are shown in
figure 5. The elliptical windows giving access to each pore section
are formed by twelve oxygen ions. The large channel-like pores all
run parallel and a r e interconnected by highly distorted eiglit-mem-
bered rings with a free diameter of 2.2 A. It is generally assumed
that diffusion between the pores is impossible except perhaps for
very small molecules such as water. Four side pockets, two on each
side of the pore, a r e shown in figure 5. They are formed by eight-
membered rings with a minimum free diameter of 3.9 A.
While the crystallographic structure of mordenite25 would seem
to allow the penetration of large molecules in the pores, natural
mordenite appears to have an effective pore diameter of only 4 A.
Explanations offered26-28 include the occurrence of stacking faults,

Figure 5
Pore structure of mordenite.
Left: Model based on skeletal tetrahedra with silicon and aluminium
in the centers.
Right: Line drawing showing wall of pore section and four side
pockets only.
20
 Figure 6
Structure of molecular sieves type X and Y.
Top: Model based on sodalite units with one a-cage.
Bottom: Line drawing of a-cage.

21
 the presence of amorphous material in the pore system and the
partial blocking of pores by cations. The pores of natural mordenite
can be effectively widened by chemical treatment Mordenite may
be synthesized as a large-port or a small-port variety. The synthetic
mordenite used by us, was of the large-port type.

Type X and Y molecular sieves

The synthetic molecular sieves type X and Y are isostructural
with the r a r e mineral faujasite. The Si/Al ratios are 1-1.5 (type X),
1 5-3 (type Y) and 2.3 (natural faujasite). Typical unit cell composi-
tions of NaX and NaY are Nagg . (AlO, )B6 . (SiOj) g^ . 264 H2O and
Na^g . (A102)5(, . (S1O2) ;,6 -264 H2O, respectively. The structure may
be visualized 29 with the same building blocks, the truncated octa-
hedra or sodalite units, as used for describing type A By connect-
ing each sodalite unit (containing a ^-cage) with four other sodalite
units by hexagonal prisms (containing y-cages) the structure shown
in figure 6 is formed. Each large cage or a-cage is surrounded by
ten sodalite units. Each a-cage can be entered from four other a-
cages through near-circular windows formed by twelve oxygen ions.
Thus a three-dimensional pore system is formed which can readily
adsorb even large molecules. The positions of cations in the fauja-
site framework, as determined mainly by X-ray diffraction, have
been reviewed recently 29.
Type X and Y are synthesized as the sodium forms, designated
NaX (13X) and NaY (Union Carbide designation SK 40), respectively.
-> The effective pore diameter may change when sodium is exchanged
for other cations. A 70 exchange of sodium in NaX for calcium
gives CaX (lOX) with a somewhat reduced pore diameter.

A d s o r p t i o n by m o l e c u l a r sieves

Sieving properties
Table n indicates the molecular sieve adsorption behaviour
around room temperature at atmospheric pressures of type A, X
and Y zeolites, especially towards compounds used in this thesis
A similar, more extensive, list is given by G) iibnei c s 30.
Of course each zeolite also adsorbs the molecules occluded by a
zeolite with a smaller effective pore diameter. In boundary cases
the adsorption of compounds which are not adsorbed by a certain
sieve according to table II, may prove to be so slow under normal"
conditions, as to be unrecognized. It must be realized, however,
that the temperature, the degree of ion exchange and the pressure
or concentiation of the sorbate have a critical influence on this
adsorption rate
At saturation vapour pressures and room temperature one unit
cell of CaA (with one a-cage) contains about 29 molecules of water
(32 g 100 g of CaA), 12 molecules of methanol (24 g^OO g of CaA),
5 3 molecules of "-propanol (21 g 100 g of CaA) or 3.7 molecules of

22
 ADSORBED BY NOT

KA NaA CaA CaX NaX NQY ADSORBED

y^3A i.^ A A i/^ 8 A ^9A v/'SA
CH3
CHj 38A CH3-CH3 i,.OA CHjCHjCHj i , 9 A C H ; C H - C H , 5 6A

n-C„ Hj„.j
CH3OH CH3-CHjOH CH3-CH2-CH2OH CH3-CH-OH
+ HOMOLOGUES CH3
CH3
H^O 30A CH3-C-OH
CH3
CH3 CH3
NH, 3.6 A CH3NH2 4 0A ;NH 4.6 A )N-CH3 5.4 A
CH CH3
CH3-CH2
V^ KA ;NH ;N-/>C*H9 ;N-nCiF9
CH3-CH2 n C-Hq flC^Fg
V NaA
V CaA O 0
Me (Bu

CaX
6L
NaX
NaY

Table II
Molecular sieve adsorption properties. Effective pore diameters of sieves and critical molecular diameters
CO (see note on p. 24) of sorbates given in A units, are estimated values.
diethyl ether (17 g/100 g of CaA). Similarly one a-cage of NaX con-
tains up to 33 molecules of water (33 g''100 g of NaX), 5.4 molecules
of benzene (24 g'lOO g of NaX), 3.8 molecules of «-hexane (ISg'lOOg
of NaX) or 1.2 molecules of 1, 3, 5-triethylbenzene (11 g/100 g of NaX).
The adsorption properties of chabazite31 and erionite32 are more
or less similar to those of type A sieve. Mordenite has not been in-
cluded in table II because relatively few separations on well-defined
materials have been described. However, cyclohexane, benzene and
cumene are readily adsorbed by the large-port variety as is expected
for a sieve with a pore size of 6.7 x 7.0 A.
The influence of the degree of cation exchange on the effective
pore diameter is illustrated in figure 7. When exchanging two sodium
ions for one calcium ion in NaA, the pores suddenly widen at 30 5^
exchange, allowing the adsorption of H-butane, due to the removal of
pore-blocking Sn cations.

14 -

H O 8 —
0. o

S "-
i, 2 -
y
1 1 I 1 ' 1 > 1
1 1 1 1
(D 40 60 80 1(DO
Hi lA 20 - ^ ION E X C H A N G E - / . CaA

Figure 7
Adsorption of "-butane at 0°C and 200 mm Hg by type A sieve^S.

The influence of temperature on the effective pore diameter is

illustrated by the adsorption of nitrogen on NaA: at -196°C no adsorp-
tion is detected while the adsorption is quite fast at -78°C. Similarly
1, 3. 5-tri-^-butylbenzene, with a critical molecular diameter* of

The critical molecular diameter of a sorbate is defined as the dia-

meter of the narrowest imaginary cylinder which can still accomo-
date the molecule; we estimated it by inspection of Coin-tauld space-
filling molecular models.

24
10 A, is not adsorbed by NaX at normal temperature according to
table II, while it is readily adsorbed34 at 180°C. Up to 1.2 mole-
cules per a-cage (18 g per 100 g of activated sieve) are adsorbed
when an excess of 1,3, 5-tri-/ -butylbenzene is heated with NaX
powder in a vacuum-sealed glass tube for three days at 180°C.
After cooling, the adsorbed compound can not be removed by wash-
ing at room temperature with cyclohexane. In the same way34 a
sorption complex of 1, 3, 5-triethyl-2, 4, 6-tribromobenzene (critical
molecular diameter 10.5 A) with NaX has been prepared. About two
of these huge molecules were adsorbed per o-cage (45 g/100 g of
activated NaX), causing some expansion of the unit cell. X-ray
structural analysis of this sorption complex seems interesting
because most adsorbed organic molecules are too mobile to be
properly detected35, 36.

Adsorption isotherms
Figure 8 shows two typical adsorption isotherms of molecular
sieves^'''.

30--
-WATER/CaA

0.25 0.50 0.75 1.0

>~ RELATIVE PRESSURE p/p,

Figure 8
Vapour-phase adsorption of water and methanol at 25 o c by CaA-
powder ( - 700 m^/g), a commercial silica sorbent (~ 600 m^/g) and
a commercial alumina sorbent (~ 250 m^/g).

25
The cages are already filled at a very low relative vapour pressure
p/pg. When p/ps approaches unity, capillary condensation occurs on
the outside surface of the sieve. The adsorption at p / p s = 0.5 gives
a good indication of the saturation capacity of the sieve. We construc-
ted a simple adsorption apparatus, described in chapter II, which
permits a fast characterization of the molecular sieve adsorption
properties using this one-point method. As shown in figure 8 mole-
cular sieves are superior to conventional adsorbents at low vapour
pressures. In this thesis the loading of adsorbate on the sieve, when
used for shifting equilibria, is quite modest (e.g. 10 g/100 g of acti-
vated sieve) so as to utilize fully the strong adsorption properties.
Adsorption isotherms at low vapour pressure37 are shown in figure
9 on a logarithmic scale.

METHANOL WATER

10 10 ' 10 ' 10 • 1
—>-VAPOR PRESSURE m m Hg

Figure 9
Vapour-phase adsorption isotherms for water and methanol on
pelletized type A sieves.

Separations with molecular sieves

Using size selectivity. Table II shows some possibilities in separat-
ing mixtures. Also, one can estimate the critical molecular diameters
of the components and compare these values with the effective pore
diameters of the different sieves. A few examples are given.
Probably the most common application of molecular sieves in

26
organic laboratory practice is in the drying of solvents. By choosing
a type A sieve which excludes the solvent, very low water contents
(as low as 1-10 ppm) are in reach. The residual water concentration
is determined by the nature of the solvent.
An industrial processes made possible by the size selectivity of
molecular sieves, is the selective adsorption by CaA of n -paraffins
(4.9 A) from the mixture with branched paraffins (5.6 A) to increase
the octane rating of gasoline and to purify n -paraffins used in the
manufacture of bio-degradable detergents. From a mixture of cis-
and/r(7»s-butene-2 the ^rw/.s-isomer is selectively adsorbed by CaA
sievelS; according to models t h e / r a » s - i s o m e r has a 0.3 A smaller
molecular diameter.
Similar separations using type X or Y sieve are less common
because all but the largest molecules are adsorbed (table 11). The
separation of 2, 6-dinitro-l, 4-di-/ -butylbenzene (I) and 2, 5-dinitro-
1, 4-di-/-butylbenzene (II) by CaX using cyclohexane as the solvent
has been described39. At reflux temperature component II (8 A) is
adsorbed by the sieve while I (9 A) is excluded. The purification of
1, 3, 5-tri-/-butylbenzene (III), as prepared by Friedel-Crafts alkyl-
ation of benzene with /-butyl chloride and aluminium chloride, is
possible34 by dissolving the crude compound contaminated with '" -
di-/-butylbenzene (IV) and/J-di-/-butylbenzene (V) in cyclohexane
and contacting it with a sufficient amount of NaX molecular sieve
{e.g. 100 g of activated sieve to 5 g of IV and V) at room temper-
ature. Components IV (7 A) and V (6 A) are adsorbed within a few
hours while III (10 A) remains in the solution.
Kinetic separations, based on a difference in r'ate of adsorption,
are also possible.
Using a difference in affinity. Even when all components of a mixture
can penetrate the sieve, a separation is possible when a difference in
affinity towards the zeolite (as expressed in the change of the free
energy G) exists. The experimentally determined adsorption heats of
benzene and "-hexane at low coverage and room temperature on NaX
a r e 18 and 14 kcal/mole, respectively40 and a selective adsorption
of benzene from a mixture with "-hexane is thus possible. Unfortun-
ately, the heat of adsorption is a function of the degree of filling of
the sieve and most adsorption heats have been determined at a low
degree of filling of sieve. Values as given by Kiselev and Lopatkin 40^
however, give a valuable indication of the adsorption heats.
Generally, the more unsaturated and more polar molecules are p r e -
ferred by molecular sieves. Thus propene is selectively adsorbed
from a mixture with propane by type A sieve at 2 5 ° c 4 1 . Similarly,
the concentration of o-xylene in a lithium-ethylenediaraine reduction
product containing 675? of unreacted o-xylene, 28^ of 1, 2-dimethyl-
cyclohexene and 55? of 2, 3-dimethylcyclohexene was reduced to 35?
by contacting a solution of the crude product in cyclohexane with a
portion of CaX sieve at room temperature42.
Even the separation of isomeric compounds is sometimes pos-

27
sible. Thus the preferential adsorption of p-xylene from mixtures
with /«-xylene and ethylbenzene using several ion-exchanged type X
and Y sieves has been described43 and a separation process, based
on this, has been developed. Higher separation factors* are found
with bulkier alkyl groups. When a mixture of m- and /?-di-/-butyl-
benzene in "-decane is contacted with type X or Y sieve, the m -
isomer is preferentially adsorbed. Separation factors of 3 on NaX,
8 on CaX and 20 on NaY have been obtained34. The difference in
affinity is tentatively ascribed to a better positioning of the aromatic
nucleus of the »?-isonier on the exposed cations in the centers of the
six-membered rings in the a-cages (figure 6). The two bulky / -butyl
groups of the w?-isomer can be accomodated in two twelve-membered
windows allowing a fair interaction of the aromatic ring with the
cations.

Shifting of c h e m i c a l equilibria using molecular

sieves

A chemical equilibrium, such as:

K
A + B —^ C + D (1)

can be shifted to the right by adsorptive removal of component D

from the solution by molecular sieve:

° adsorbed ^^^

The principle has been used in this thesis in the preparation of

esters by alcoholysis (chapter IE), when D is a linear alcohol, and
in the synthesis of enamines and ketimines (chapter IV) and of
acetals (chapter V), when D is water.
The position of equilibrium (1) is given by:

[C]x[D] ^
[A] X [B] "

so, the residual concentration of D in solution after selective ad-

sorption of D by the sieve governs the final conversion of A and B
into C. This residual concentration of D is determined by the

The separation factor is defined as the ratio of components A and B

in the zeolitic phase divided by the ratio of A and B in solution (or
in the vapour).

28
physical equilibrium (2) between D in solution and D adsorbed in the
sieve. The number of parameters determining equilibrium (2) is
very large indeed, especially when it is realized that the system be-
comes further complicated by the presence of a catalyst to accelerate
the reaction of A and B, and of a solvent. The influence of most of
these parameters is not easily quantified but some qualitative and
semi-quantitative indications will be given. - . . .

Experimental tecliniques
Three different techniques of contacting the reaction mixture
with the sieve have been investigated:
(/) Suspension technique. The molecular sieve is suspended in the
liquid reaction mixture. A fast removal of D is thus possible.
Unwanted interactions of catalyst or one of the reactants with the
sieve sometimes occur.
('/) Extraction technique. Condensed vapour of the boiling reaction
mixture flows through a compartment filled with molecular sieve
pellets and returns into the reaction mixture. Volatile component D
is retained in the sieve. An advantage is the simple work-up of the
reaction mixture. The heating may have an unfavourable influence on
the equilibrium constant K. Another disadvantage is the fact that D
must be transported to the sieve via the vapour phase. This process
generally requires a longer reaction time than necessary in the s u s -
pension technique. This is especially a problem when K is small
because then the maximum concentration of D in solution will be
very low during the reaction. The rates of adsorption of D by the
sieve in the extraction technique and in the suspension technique
may therefore be quite different as illustrated in figure 10^4.
(///) Flow technique. The reaction mixture flows through a column
filled with molecular sieve. In theory a higher conversion than
with the suspension technique can be obtained because the loading of
the sieve with D will be high at the beginning of the column and will
decrease along the column. In preliminary experiments45 several
difficulties were encountered: (a) the forward reaction of equilibrium
(1) and the adsorption of D by the sieve have to be fast to allow a
reasonable flow rate; (b) the interaction of acid catalyst, used in the
synthesis of acetals, with sieve is pronounced at the beginning of the
column due (see chapter VI) to the local high water loadings of the
sieve.
In this thesis the suspension technique and the extraction tech-
nique have been used.

Influence of type and amount of sieve on the adsorption of D

A primary consideration in selecting a sieve is the requirement
that its pores should be large enough to adsorb D without permitting
the co-adsorption of A, B, C or the solvent. This consideration has
been neglected by Brannock c. s.46 when NaX was used to remove

29
 1 — 1 ^ — j
»•—•—
1 a ' " " ^
^

1 /^

1
IVo 0 c

0 25 50 75 100
TIME hr

Figure 10
Alcoholysis of dimethyl terephthalate (25 mmole) with /-butyl alcohol
(125 mmole) in 125 ml of benzene a t 8 0 ° C , catalyzed by potassium
/-butoxide (2 mmole): (a) Suspension technique with 30 g of CaA pow-
der: (b) Extraction technique with 30 g of CaA pellets; (c) Blanc ex-
periment without sieve.

water formed by condensation of propionaldehyde and dimethylamine.

The unfavourable influence of such a co-adsorption is clearly illus-
trated by some experiments in chapter V when water was removed
with CaA sieve at 0°C from a reaction mixture also containing
ethanol. Both the adsorption of water and the conversion of ketone
to acetal went through a pronounced maximum, when plotted against
time, due to a slow co-adsorption of ethanol.
Sieves which meet this requirement may still give different ad-
sorption results as illustrated by figure 9 and by the conversions
obtained with KA. NaA and CaA in chapter IV.
An important parameter is the quantity of sieve used. Thus, a
reduction of the water loading from 10 to 5 g ot water per 100 g of
NaA reduces the residual vapour pressure in figure 9 with a factor
3. Although this figure applies to the adsorption of water and

30
methanol from the vapour phase, a similar influence of the degree of
loading of sieve on the concentration of D in a multi-component liquid
mixture is expected. Generally, 100 to 200 g of activated sieve was
used when 10 g of D was expected with equilibrium (1) shifted com-
pletely to the right.

Influence of composition of the reaction mixture

The composition of the liquid reaction mixture does influence
the residual concentration of D in the mixture. This concentration
will increase47 with the solvation power of the medium for D. There-
fore the solvent is an important variable. In chapter V it will be
shown that the replacement of dioxane by pentane did reduce the
residual water concentration under the particular conditions with a
factor 2. In the synthesis of an enamine in chapter IV the replace-
ment of ether by cyclohexane did reduce the percentage of unreacted
ketone from 11 to 45? in a particular case. More generally, with D
being a polar compound such as water or an alcohol, one should
preferably use a hydrophobic, apolar solvent such as hexane or
cyclohexane. Replacement of ether and methanol, as sometimes
used in literature48, 49^ by an apolar solvent such as cyclohexane,
is expected to result in higher yields of ketimines and enamines.
The amount and nature of the catalyst also influence the medium.
Thus/)-toluenesulphonic acid, used in the preparation of acetals, is
a hydrophilic compound. A comparison of the experiments 3 and 11
in chapter V part 1 shows that the replacement of/>-toluenesulphonic
acid by a solid catalyst, does result in a somewhat higher conver-
sion of ketone to acetal.
The influence of the composition of the reaction mixture on
equilibrium (1) is further complicated by a possible medium effect
on the equilibrium constant A'.

Influence of temperature
The influence of the temperature of the molecular sieve on the
adsorption of D is illustrated by figure 9; here a 25 °C temperature
increase results in an increase of the water vapour pressure by a
factor 5 to 10 at water loadings of 5 to 10 g per 100 g of sieve NaA
or CaA. Therefore, in the extraction technique cooling of the sieve-
containing compartment is essential. An extraction apparatus as
described in chapter III or in the literature by van der Vliigt c. s.50
and by Kramer^'- are quite satisfactory. In the suspension technique
one may wish to conduct the reaction at the lowest practical temp-
erature for the same reason. Furthermore, border-line case co-
adsorption of one of the other components and also any unwanted
reaction with the sieve a r e suppressed by a lowering of the temp-
erature.
The temperature of the reaction mixture does influence the
equilibrium constant/^. In the exothermic formation of acetals

31
described in chapter V, a lowering of the temperature resulted in a
substantial higher value of K.

Interaction of catalyst with sieve

In the suspension technique an interaction between catalyst and
sieve sometimes occurred leading to a reduction of the catalytic
activity. Such a deactivation was noted both with alkoxide catalyst
in chapter III (adsorption of alkali metal alkoxide) and with acid
catalyst in chapter V (ion exchange between protons of the catalyst
and cations of the sieve). These interactions were less important
with sieves containing divalent cations due to the lower mobility of
the ionic species. It is expected that the rate of deactivation will
also go down with an increasing silicon to aluminium ratio of the
zeolites due to a larger mean distance between anionic sites in the
aluminosilicate framework and therefore a higher diffusional b a r -
rier for ionic species.

Proton-exchanged type A molecular sieves

In the course of our work on the preparation of acetals

described in chapter V, it appeared that ion exchange occurred be-
tween protons of/'-toluenesulphonic acid in solution and cations
of the type A sieves. The partially proton-exchanged type A sieves
displayed good adsorption properties and were shown by X-ray
diffraction to be crystalline. In the literature32 it is generally
assumed that proton exchange of type A sieves leads to inferior
products. The results of our detailed investigation of the inter-
action of/)-toluenesulphonic acid with type A sieves, especially
in an organic solvent, and of the properties of these proton-
exchanged sieves are reported in chapter VI.
In literature, proton incorporation in zeolites to prepare
catalytically active forms has been effected by treatment with an
aqueous solution of acid, by heating of the ammonium-exchanged
forms and by reduction of heavy metal exchanged forms. When
applied to zeolites with a high aluminium content such as type A
sieve, break-down of the lattice was pronounced.
It is expected that our technique of proton exchange in an
organic medium to prepare thermally stable proton-exchanged type
A sieves, can also be applied to other molecular sieves.

32
References:
1 J.V.Smith. Mineral. Soc.Amer., Spec. P a p . i 281(1963).
2 W. Loewenstein, Amer. Mineral. 39, 92 (1954).
3 D.W.Breck, Advan. Chem. Ser. 101. 1(1971).
4 C.K. Hersh, "Molecular Sieves", Reinhold Publishing Corporation,
New York (1961).
5 O. Griihner, P. Jim and M. Ralek, "Molekularsiebe', VEB Deut-
scher Verlag der Wissenschaften, Berlin (1968),
6 W.Eitel, "Silicate Science", Vol.1 (1964) and IV (1966), Academic
P r e s s , New York, London.
7 R. M. Barrer in "Non-Stoichiometric Compounds', L. Mandelcorn
(ed.), Academic P r e s s , New York, London (1964), p. 310.
8 Papers Conference on "Molecular Sieves", 1967, Soc. Chem. Ind. ,
London (1968).
9 "Molecular Sieve Zeolites I and 11", Advan. Chem. Ser. Vol. 101 and
102, Amer. Chem. S o c , (1971).
\QR.M. Barrer, Endeavour 23, 122(1964).
11 K. S. Deffeyes, reference 8, p. 7.
12 R.A.Sheppard, Advan. Chem. Ser. 101, 279 (1971).
n L.V.C.Rees, Annu. Rep. Progr. Chem. , Sect. A 67, 191(1970).
14 P.B. Veniitound P. S. Landis, Advan. Catal. Relat. Subj. 18, 259
(1968).
15 P. B. Venuto, Advan. Chem. Ser. 102, 260 (1971).
16 iV. Y. Clien and P. B. Weisz^ Chem. Eng. Progr. , Symp. Ser. 63 (no.
73), 86 (1967).
n J. Gons, this laboratory, unpublished results.
18 L. C. Fetterly and K. F. Koetitz, reference 8, p. 102.
19 W. M. Meier and D. H. Olson, Advan. Chem. Ser. 101, 155 (1971).
20 T. B. Reed and D. \V. Break, J. Amer. Chem. Soc. 78, 5972 (1956).
21 L. Broussard and D. P. Shoemaker, J. Amer. Chem. Soc. 82, 1041
(1960).
22 p. A. Howell, Acta Crystallogr. 1^, 737(1960).
23 J- V. Smith and L. G. Dowell, Z. Kristallogr. 126. 135 (1968).
24 V. Gramlich sind W.M.Meier, Z. Kristallogr. 133, 134(1971).
25 W.M.Meier, Z. Kristallogr. 115, 439(1961).
26 L.B.Sand, r e f e r e n c e s , p. 71.
27 Y. Nishimiiraznd H. Takahashi, Kolloid-Z. Z. Polym. 245, 415
(1971).
28 B. W. Biirbidge, I. M. Keen and M. K. Eyles, Advan. Chem. Ser. 102,
400 (1971).
2% J. V.Smith, Advan. Chem. Ser. 101, 171(1971).
30 Reference 5, p. 90.
ZIR.M. Barrer, Advan. Chem. Ser. 102, 1(1971).
32 H. E. Rolison, G. P. Hamner and W. F. Arex, J r . , Advan. Chem.Ser.
1 ^ , 417 (1971).
33 F. Wolf, H. Filrtig and U. HOdicke, Chem. Tech. (Berlin) IJ, 524
(1966).

33
34 D. P. Roelofsen, unpublished results.
35 H.D.Simpson and H. Steinfink, J. Amer. Chem. Soc. 91, 6225,6229
(1969).
36J.V.Smith, r e f e r e n c e s , p.28.
37 Data of the Union Carbide Corporation. See also reference 5, p.
93, 106.
38 G.J. Griesmer, H.B.Rhodes and K.Kiyonaga, Petrol. Refiner 39
(June), 125 (1960).
39 A. J. Hoefnagel, J. H. A.J. Nunnink, A. van Veen, P. E. Verkade and
B.M.Wepster, Rec. Trav. Chim. Pays-Bas 88, 386(1969).
AQ A. V. Kiselev and A. A. Lopatkin, r e f e r e n c e s , p. 252.
^lA.Grossmann and W.Schirmer, Chem. Tech. (Berlin) 20, 34 (1968).
42 p. J. Heriveijer, this laboratory, unpublished results.
43 R. W. Neuzil, German patent 1, 931, 519.
44 D. P. Roelofsen, J. A. Hagendoorn and H. van Bekkitm, Chem. Ind.
(London) 1966, 1622.
45 R. Taal, this laboratory, unpublished results.
46 K.C Brannock, A.Bell, R.D.Burpitt andC. A. Kelly, J. Org. Chem.
29, 801 (1964).
47 K. Varga and //. Bever, Acta Chim. Acad. Sci. Hung. 52, 69 (1967).
48 E.P.Kyba, Org. Prep. Proced. 2, 149 (1970).
49 Kaziio Taguchi and F. H. Westheimer, J. Org. Chem. 36,1570(1971).
50 F.A. van der Vlngt, J. W. Verhoeven and U.K. Pandit, Rec. Trav.
Chim. Pays-Bas 89, 1258 (1970).
51 M. T. Kramer, Glas Instrum. Tech. Fachz. Lab. 14, 1265 (1970).

34
 CHAPTER II

Simple Technique to Determine Vapor

Adsorption Capacities of Molecular Sieves

D. P. Roelofsen

THE ADSORPTION ISOTHERM of a molecular sieve character-

istically consists of a steep section at low relative pressure
PIPS followed by a nearly horizontal section up to a relative
pressure of about 0.8, proving complete filling of the adsorp-
tion cavities; finally the adsorption isotherm increases
again because of capillary condensation on the outside surface
of the crystals. Therefore the saturation capacity of a
molecular sieve can be obtained satisfactorily using a one-
point method in the horizontal part of the isotherm.
This paper describes a vapor adsorption apparatus of
simple construction for the above purpose in which several
samples can be investigated at the same time. Thus this
technique offers distinct advantages over more conventional
volumetric or gravimetric adsorption methods (7) when
saturation capacities of molecular sieves are to be determined.
An adsorption apparatus with a similar aim but of a different
construction has been mentioned (2) but no technical details
have been published.

APPARATUS
The apparatus (Figure 1) consists of two glass-bells, a
coupling with Viton O-ring seals, and a holder with six
adsorption tubes. In the upper glass-bell the holder with
the glass adsorption tubes, each with a sealed bulb filled with
activated molecular sieve, is held at a temperature Ti. This

(1) D. M. Young and A. D. Crowell, "Physical Adsorption of

Gases," Butterworths, London, 1962.
(2) G. R. Landolt, Abstracts of Papers, TECH 007, 158th ACS
National Meeting, New York, September 1969. See also Anal.
Chew. 43, 613 (1971).
/ Reprinted from Analytical Chemistry, vol. 43. page 631,
April 1971, by permission of the copyright owner.
 UPPER GLASS-BELL

GROUND SURFACE

BAR MAGNET

SAMPLE IN
SEALED BULB

VITON 0-RINGS

GROUND SURFACE

LOWER GLASS-BELL

, ADSORBATE

CORK RING SUPPORT

Figure 1. Exploded view of adsorption apparatus

36

5C
 Figure 2. Top view of coupling
IS accomplished by forcing thermostated water through the
jacket of the glass-bell and in addition through a circular
groove in the brass coupling (Figure 2). In the lower glass-
bell liquid adsorbate is thermostatically controlled at a
temperature T, < Ti. Using published vapor pressure data
one can adjust p/ps above the samples by choosing Ti and To.
By connecting an oil-type vacuum pump to the Serto valve,
the apparatus is evacuated to remove interfering gases and
to increase the adsorption velocity On molecular sieve
samples only water, of the gases possibly present, will seri-
ously interfere with the adsorption of the vapor at the tem-
peratures and pressures used To remove water completely,
vacuum was applied during a few minutes, thus purging the
apparatus with the vapor of the liquid adsorbate. Any
last traces of water may be removed by activated 3A molecular
sieve pellets in a porcelain crucible 3A sieve has a very high
affinity and selectivity for water because of the small pores
which exclude nearly all compounds except water. Experi-
ments without 3A sieve, however, gave identical results when
due care was taken to purge the apparatus completely. The
sealed bulbs within the adsorption tubes are broken by sharp
bar magnets using a powerful horseshoe magnet Each
bar magnet moves smoothly in a glass tube provided with
a small hole near the top. The vapor contacts the adsorbent
through 4-mm holes bored m both parts of a 19/26 f grease-
less clear precision joint. Precision joints are preferred
although ordinary ground joints can also be used At the
end of each experiment, dry air is admitted through the
Serto valve.
The materials of construction are, unless stated otherwise,
borosilicate glass and tin-soldered brass which has been
nickel-plated. Apiezon grease was sparingly applied to the
Viton O-rings.
 EXPERIMENTAL PROCEDURE
The samples were activated as follows A borosihcate
glass tube of 8 mm o d , blown out at one end and with a
ground glass cone at the other end, was weighed, filled with
0 5-1 5 grams of molecular sieve, reweighed, a plug of glass
wool inserted, the tube weighed again and connected to a
high vacuum line together with five similar tubes. After
activation at 400 °C and 10^^ mm, the bulb was sealed off, the
two parts were weighed, and the amount of activated molec-
ular sieve was calculated
A sealed bulb and a bar magnet were inserted into a
carefully cleaned adsorption tube After a short evacuation
to remove traces of volatile material, the stoppers were
turned to exclude moisture. To minimize magnetic dis-
turbance, the assembled adsorption tubes were weighed
while suspended in a loop of nylon. The stoppers were
turned again so as to get a free passage through the socket
and cone of each joint. After pouring a small portion of
pure adsorbate in the lower glass-bell, the apparatus was
assembled (Figure 1), vacuum applied for a few minutes,
and after closing the Serto valve the sealed bulbs were broken
magnetically. After a convenient adsorption time (24
hours was chosen although a few hours will be sufficient in
most cases) dry air was admitted at such a low rate that no
condensation occurred. The upper glass-bell and the holder
were removed and, after cooling for a few minutes, the
stoppers were turned After equilibration in the balance
room, the tubes were reweighed By including a blank
with an empty sealed bulb in each experiment, a correction
for the dead volume was obtained. The adsorbed amount
was expressed in grams of adsorbate per 100 grams of activated
molecular sieve
When changing from one adsorbate to the other, the Viton
0-rings were degassed at 150 °C.
RESULTS AND DISCUSSION
This technique was used successfully to determine, for
instance, the adsorption capacities of type X and Y sieves
for cyclohexane and of type A and AW sieves for water
and different alcohols. The precision was checked as shown
by the examples on Linde sieve powders given in Table I.
The first measurement of Table I was repeated on the same
13X sample under identical conditions using an electro-
balance. A value of 18.3 g/100 g of sieve was found as
compared with a value of 18 9 g/100 g of sieve in Table I.

38
 Table I. Adsorption Capacities
Number of drams adsorbe 1/iuu grams
Mol. sieve Adsorbate "C PIP' expts Mean value Std dev
I3X Cyclohexane 45 0.50 17 18.90 0.08
13X Cyclohexane (2) 25 0.20 22 15.3 0.22
13X Cyclohexane (3) 25 0.46 1 19.6
4A Methanol 45 0.50 5 19,37 0 05
4A Methanol (4) 25 0.03 1 17

Although obtained under different experimental conditions

and on different batches of molecular sieves, literature data
are included in Table I for comparison.
The large difference in the mean adsorption capacity of
13X for cyclohexane, found in reference (2) and in this work,
is presumably due to a different amount of amorphous
material in the 13X samples.
An important advantage of the present method over that
described in reference (2) is that the activated samples were
contained in sealed bulbs which were broken magnetically,
thus largely eliminating any problem of moisture uptake.
In conclusion, the present technique allows a rapid and
reliable determination of adsorption capacities of molecular
sieves.

ACKNOWLEDGMENT
Thanks are due to H. van Bekkum, E, R. J. Wils, and
L. Maat for helpful discussions, to the general service of the
chemistry department for the construction of the apparatus,
and to Miss H. C. Niedeveld and J. Medema of the Chemical
Laboratory, National Defense Research Organization TNO,
for performing the electrobalance adsorption measurement.

(3) D. W. Breck and E. M, Flanigen, Papers Conference on "Molec-

ular Sieves," 1967 (1968), Soc. Chem. Ind., London, p 47.
(4) D. W. Breck, W. G. Eversole, R, M. Milton, T. B. Reed, and
T. L. Thomas.y, Amer. Chem. Soc, 78, 5963 (1956),

39
 CHAPTER III
ESTER INTERCHANGE USING MOLECULAR SIEVES i /
BY
D. P. ROELOFSEN, J. W. M. DE GRAAF *, J. A. HAGENDOORN **,
H. M. VERSCHOOR ***, and H. VAN BEKKUM

Ester interchange reactions using type A molecular sieves which selectively

adsorb the displaced alcohol, have been shown to proceed practically to com-
pletion. Examples are given which cover a number of starting esters, displacing
alcohols, experimental procedures, solvents, and cationic forms of the molecular
sieves. Several new /-butyl esters were obtained and characterized.
Equilibrium constants were measured and rate studies made for several re-
presentative reactions.
The interaction of sodium and potassium isopropoxide catalysts and sieves
has also been investigated. Both cation exchange with the sieves and adsorption
of alkoxide after the addition of methyl alcohol have been observed. Interesting
differences have been found between 3A, 4A, and 5A sieves.
The interfering dehydration of /-butyl alcohol has been studied with different
molecular sieves of type A. The influence of traces of water and triethylamine on
the dehydration has been investigated.

Introduction
Molecular sieves, crystalline aluminosilicates, are well-known as
adsorbents ^ and catalysts ^. Because of their high adsorptive capacities
and uniform pores they show unique properties in both applications •*.
Union Carbide molecular sieve type A, which has been used in this work,
IS a synthetic molecular sieve; three varieties, 3A, 4A, and 5A with estimated
pore diameters of 3, 4, and 5 A, respectively, are available.
Molecular sieves may be used in shifting equilibria by removing one of
the reaction products by selective adsorption. The successful use of type A

* Present address: Gulf Research Laboratoria N.V., Rozenburg, The Netherlands.

** Present address: Van der Grinten N.V., Venlo, The Netherlands.
*•* Present address: Central Laboratory, Dutch State Mines, Geleen, The Netherlands.
1 Preliminary communication: D. P. Roelofsen, J. A. Hagendoorn, and H. van Bekkum,
Chem. Ind. London 1966, 1622.
2 R. M. Barrer, Brit. Chem. Eng. 4, 267 (1959).
' P. B. Venuto and P. S. Landis, Advan. Catalysis 18, 259 (1968).
•* D. W. Breck, J. Chem. Educ. 41, 678 (1964).
/ Reprinted from Recueil Travaux Chimiques des Pays-Bas, vol. 89, page 193,
March 1970, by permission of the copyright owner.

41
 Tabl
Ester interchange usin,

Exp. Starting ester RCOORi Displacing alcohol Solvent Sieve » Technique

R20H

1 ethyl isobutyrate r-butyl alcohol benzene 4A E

2 methyl heptanoate /-butyl alcohol benzene 4A E
3 methyl octanoate isopropyl alcohol benzene 4A E
4 ethyl octanoate cyclohexanol benzene 5A E
5 ethyl octanoate phenol xylene 5A E
6 methyl hexadecanoate /-butyl alcohol benzene 4A E
7 polymethyl acrylate isopropyl alcohol benzene 5A E
8 polymethyl acrylate /-butyl alcohol benzene 4A E
9 cisltrans-iimethyl cyclohexane- /-butyl alcohol — 5A S-B
1,4-dicarboxylate
10 methyl methacrylate n-butyl alcohol benzene 4A E
11 methyl methacrylate n-butyl alcohol «-pentane 4A S-A
12 methyl benzoate /-butyl alcohol — 5A S-B
13 methyl benzoate benzyl alcohol benzene 4A S-A
14 methyl benzoate phenol /j-xylene 5A E
15 methyl 2-methylbenzoate /-butyl alcohol — 5A S-B
16 methyl 2-isopropylbenzoate /-butyl alcohol — 5A S-B
17 methyl 2-(3-pentyl)benzoate /-butyl alcohol toluene 4A E
18 methyl 2-(2,4-dimethyl-3-pentyl)- /-butyl alcohol toluene 4A E
benzoate
19 methyl 2-/-butylbenzoate /-butyl alcohol toluene 4A E
20 methyl 2,4,6-trimethylbenzoate /-butyl alcohol toluene 4A E
21 ethyl 4-aminobenzoate isopropyl alcohol benzene 4A E
22 dimethyl phthalate isopropyl alcohol benzene 5A E
23 dimethyl phthalate /-butyl alcohol benzene 5A E
24 dimethyl isophthalate /-butyl alcohol — 5A S-B
25 dimethyl terephthalate ethyl alcohol benzene 3A EJ
26 dimethyl terephthalate /i-propyl alcohol benzene 4A E
27 dimethyl terephthalate isopropyl alcohol benzene 5A E
28 dimethyl terephthalate /-butyl alcohol benzene 5A S-A
29 dimethyl terephthalate /-butyl alcohol benzene 5A E

• For every 10 g of methyl or ethyl alcohol to be adsorbed 100 g of activated sieve was
used. Pellets were used in all experiments, except in exp. 13 where powder vvas used.
" Very often a complete conversion may have been reached in a fraction of the time
stated.
" The potassium alkoxide (or sodium alkoxide in exp. 11 and 13) of the d isplacing
alcohol was used as the catalyst.
" In exp, 7 and 8 the conversion was measured by elemental analysis, and b y IR and
NMR measurements, in the other experiments by GLC.
' In most cases the crude product obtained after removal of sieve, cataly it, excess
alcohol, and solvent, appeared to be very pure. The yield however was de termined
after purification of the crude product by recrystallization or distillation; no correction
was applied for the loss of product due to sampling.

42
 I
molecular sieves

Reaction Reaction charge mraoles Con- Yield «

1 ri&l*fi 1 / ^ r\ Q.
Physical constants
time ", h version " % of new esters '
RCOORi R20H R20- •= %
I
64 65 190 5 96 1

1 83 72 192 3 100 1 83 b.p. 93720 mm; /ii» 1.4150

1 2^ 50 147 3 98 —'
100 41 192 8 100 1
110 65 190 10 99 74
131 65 192 3 98 77
72 58« 264 3 75 1
72 65 « 400 26 75 i
120 150 3100 40 99 62 m.p. 57- 58° (cis) "
m.p. \21-ni.y (trans) ^
24 75 212 18 100 —1
15 155 128 22 100 65
70 147 1400 30 100 83
2 210 620 30 100 70
46 53 158 11 100 -_i
21 94 1320 26 100 70
94 62 930 26 99 70 b.p. 7270,25 mm; < 1.4899
66 125 710 38 100 81 b.p. 7170.1 mm; ni" 1.4870
100 94 620 38 100 72 b.p. 8170.1 mm; ni" 1.4902

120 31 190 26 100 61 b.p. 7170.2 mm; «!,« 1.4922

70 62 380 38 100 70
16 90 290 13 1 77
15 52 275 5 100 93
21 52 270 5 100 85 m.p. 73.5-74°
21 52 1340 13 1 74 m.p. 84.5-85°
5 26 140 1 100 87
1 26 137 3 100 77
3 52 276 5 100 87
96 26 129 13 100 1
9 52 280 5 100 87

' Elemental C/H analysis wit hin0.2%oftheoretica 1 value. NMR spectra in accordance
with structure.
* Amount of polymeric ester in mg-equivs.
» Separation performed by selective inclusion of /ra/M-isomer in thiourea.
' Not determined.
1 See note Table VI.

43
sieves is described in the literature to remove water in esterifications ^"^
and in the synthesis of enamines ^"^i, phosphazines^^, acetals i^, poly-
ethers 1'', and polyamides ^^. Type A has also been used in ester inter-
change to remove glycol from polymerization mixtures i^, to prepare a
variety of esters i, and to prepare several methacrylates I'-is.
In the present article we describe in more detail the use of molecular
sieves in ester interchange, or more specifically in the alcoholysis of esters:
K
RCOORi + R20H z ^ RC00R2 + RiQH [1]
The synthesis of esters by ester interchange is attractive when conversion
approaches 100%. This object is difficult to attain solely by applying an
excess of R^GH. Complete conversion by removing one of the products
by distillation is only possible when a sufficient difference in volatility
exists. In contrast, the removal of the displaced alcohol R^OH by selective
adsorption on molecular sieves as described in the present article, is a very
simple and fairly general procedure.
We also studied the equilibrium constant and the rate of the ester inter-
change reaction, the interaction of metal alkoxide catalyst and sieves, and
the dehydration of the alcohol R^OH.

Results and discussion

Esters synthesized by ester interchange using molecular sieves
Both a suspension technique and an extraction technique have been
used. In the suspension technique, the sieve in pelletized form or as a powder
is suspended in the reaction mixture which consists of the starting ester,

5 Yasumasa Ito, Nippon Kagaku Zasshi 83, 195 (1962); Chem. Abstr. 58, 11264g
(1963).
6 R. L. Stern and E. N. Bolan, Chem. Ind. London 1967, 825.
' H. R. Harrison, W. M. Haynes, P. Arthur, and E. J. Eisenbraun, ibid. 1968, 1568.
* M. Coopersmith, A. J. Rutkowski, and S. J. Fusco, Ind. Eng. Chem. Prod. Res.
Develop. 5, 46 (1966).
» / . Szmuszkovicz, Advan. Org. Chem. 4, 11 (1963).
1° G. Stork, A. Brizzolara, H. Landesman, J. Szmuszkovicz, and R. Terrell, J. Am. Chem.
Soc. 85, 207 (1963).
11 R. Bonnett and T. R. Emerson, J. Chem. Soc. 1965, 4508.
12 C. C. Walker and H. Shechter, Tetrahedron Letters 20, 1447 (1965).
13 E. J. Hedgley and H. G. Fletcher, Jr., J. Org. Chem. 30, 1282 (1965).
" Dutch Patent Appl. 6,610,270; Chem. Abstr. 67, 109175v (1967).
15 Hiroshi Kachi and Hideo Sekiguchi, Japan. Patent 68 12,837; Chem. Abstr. 69, 77987J
(1968).
18 A. M. Schiller and W. F. Oliver, U.S. Patent 3,066,116; Chem. Abstr. 58, 10328b
(1963).
1' /. K. Haken, J. Appl. Chem. London 18, 17 (1968).
IS A. K. Misra, G. C. Misra, and L. M. Pande, Indian J. Technol. 6, 342 (1968).

44
the alcohol to be introduced, alkoxide catalyst, and solvent. We worked at
ambient temperature (technique S-A) or at the boiling point (technique
S-B). In the extraction method (technique E) the pelletized sieve is kept
above the boiling reaction mixture in a modified extraction apparatus
(Fig. 3). Condensed vapour flows back through the sieve pellets into the
reaction flask. Esters prepared by these methods are listed in Table I.
The reaction conditions are not necessarily optimal.
Table 1 shows that this ester interchange technique is generally applicable
as long as the displaced alcohol is selectively adsorbed from the reaction
mixture. This method is a powerful one, especially in the synthesis of
tertiary alkyl esters, for instance ?-butyl esters.
The alcoholysis of methacrylates, especially of methyl methacrylate with
«-butyl alcohol, has been studied independently by Haken ' ' who used
/7-toluenesulfonic acid as the catalyst and an unstated quantity of 3A or
4A pellets in a modified Soxhlet extractor. After 6 h a conversion of 95%
was obtained. In exp. 10 of Table I we reached a conversion of essentially
100% within 24 h, using potassium /(-butoxide as the catalyst, 4A pellets,
and the extraction technique. Haken mentions that the suspension technique
gave unsatisfactory results. In our opinion this may be due partly to the
exchange of hydrogen ions of the /j-toluenesulfonic acid catalyst with
cations of the sieve ^^. We found in exp. 11 of Table I that, when using
the much more active catalyst sodium «-butoxide, a conversion of 100%
was reached after 15 h using the suspension technique and 4A pellets at
room temperature.
Some factors which determine the choice of sieve, the reaction time, the
deactivation of the catalyst, etc., have been studied in more detail (see
below). Also a general guide is given when the synthesis of a certain ester
is required.

Position of the alcoholysis equilibrium

The position of the alcoholysis equilibrium [1] was determined in a few
cases (Table II) and compared with the relatively scarce Uterature data.
Table II
Alcoholysis equilibrium constant K, as defined in equation [1]

Starting ester K
RCOORI
R2 = n-Pr (25°) R2 = isoPr (25°) R2 = /-Bu (50°)

methyl heptanoate 0.70 0.41 0.008

methyl benzoate 0.76 0.51 0.008

19 D.P. Roelofsen, E.R.J. Wils andH. van Bekkum. J. Inorg. Nucl. Chem. 34,
1437 (1972). This thesis chapter VI.

45,
 Our values compare reasonably with literature 20 values of 0.6 and 0.3
for «-propyl and isopropyl alcohol reacting with methyl acetate at 200°.
With r-butyl alcohol the equilibrium [1] is extremely unfavourable. Thus
when preparing tertiary alkyl esters especially the use of molecular sieves
is to be recommended when a quantitative conversion is sought.

Rate of alcoholysis
Ester alcoholysis in the presence of an alkoxide catalyst is accepted to
proceed by equation [2]:
ki
RCOORI + R20- -^zL RC00R2 + RiQ- [2]

The catalyst is regenerated by a fast proton exchange:

Ki
RiQ- + R20H — 1 RiQH + R^O" [3]

The second-order rate constant /ci of the equilibrium [2] was determined
in cyclohexane for methyl heptanoate and methyl benzoate reacting with
«-propyl and isopropyl alcohol at 25°, and with r-butyl alcohol at 50°
(Table III).
Table III
Rate constant ^l, as defined in equation [2]

Starting ester ki, 1 • mol~i min-i

RCOORI R2 == /i-Pr (25°) R2 = isoPr (25°) R2 == /-Bu (50°)

methyl heptanoate 5.6 1.9 0,03

methyl benzoate 5.9 2.3 0.09

It follows from Table III that convenient reaction times are to be expected
for the alcoholysis of simple aliphatic or aromatic esters with primary or
secondary alcohols at room temperature. In these cases procedure S-A
may be used. Higher temperatures are needed to prepare tertiary alkyl
esters, and either technique E or technique S-B is convenient.
The influence of substituents on the alcoholysis of methyl benzoate
with r-butyl alcohol was studied by the competitive reaction of methyl
benzoate and substituted methyl benzoate with r-butyl alcohol and catalyst
at 50°. It was found that electron withdrawing substituents such as CI

P. R. Fehlandt and H. Adkiiis, J. Am. Chem. Soc. 57, 193 (1935).

46
or Br in the meta position increased the reaction rate, while an electron
donating methyl group in the meta position lowered the reactivity as would
be expected for a nucleophilic mechanism.
In a few cases alcoholysis using procedure E was followed with time
(Fig. 1).

100Q 100 Q

Idi-Me di-iPr di-Me di-tSu

•« o
• E ^ equil
equil ^Me-tBu without <^
without sieve
sieve

-—©-
10 20 30 AO 50 300 400 500 600
mm » mm »

Fig. 1. Alcoholysis of dimethyl terephthalate (0 026 mole), benzene (125 ml), and
4A pellets (20 g) using procedure E with a isoprop>l alcohol (0 140 mole) and
potassium isopropoxide (0 003 mole), b. /-butyl alcohol (0 140 mole) and
potassium /-butoxide (0 003 mole).
O, dimethyl terephthalate; 9 , methyl isopropyl (or methyl /-butyl) terephthalate;
• , diisopropyl (or di-c-butyl) terephthalate.

The difference between the rates in Fig. la and lb is striking, although

this was to be expected from the results in Table III. A comparison of the
experiments without molecular sieve in Fig. la and lb clearly indicates the
unfavourable position of the alcoholysis equifibrium when /-butyl groups
are introduced (cf. Table II).
The rate of alcoholysis in Fig. 1 is also governed by the transport of
methyl alcohol. Thus the suspension procedure is generally faster. For
instance dimethyl terephthalate reacting with r-butyl alcohol under the
conditions of Fig. lb (using 5A powder), gave a conversion of 100% after
only 60 min using the suspension technique i.

Alcohol I alkoxide equilibrium

The position of the equilibrium [3] between alkoxides and alcohols was
determined at room temperature for R^ = Me, and R^ = w-Pr, isoPr,
and r-Bu. We used conditions similar to those in the ester interchange
experiments: high R^OH/MeOH ratio, high R^QH/potassium alkoxide
ratio, and benzene as solvent. A technique was used in which a mixture

47
of MeOH, R20H, MeO-K+, R^O K+, and benzene was analysed by
GSC at 130°. Ki defined as Ki = [MeO^] [R20H] / [RSQ] [MeOH] was
calculated 21 (Table IV). Of course the position of the equilibrium may
change somewhat going from room temperature to the temperature of
analysis. For comparison, literature data 22 for sodium alkoxides in iso-
propyl alcohol as solvent are included.

Table IV
Equilibrium constant Ki, as defined above, for Ri = Mc

i^i
R20H
This work Lit. 22

EtOH 4
«-PrOH 5 8
isoPrOH 20 53
r-BuOH 50 > 20

An attempt to obtain Ki for Ri = Me and R^ = isoPr by determining

the chemical shift of the methine proton of the isopropyl group was un-
successful. A value of 3 was found, which seems to be too low.
Table IV shows that Ki is unfavourable, especially when R^OH is a
secondary or tertiary alcohol. When, however, methyl alcohol (or any
other alcohol R^OH) is adsorbed on molecular sieves, then the loss of
catalytically active R^O" will be reduced.

Ion exchange and adsorption of alkoxide on molecular sieves

Using the suspension technique it appeared that in some alcoholysis
experiments the concentration of alkoxide catalyst decreased, eventually
becoming zero. This phenomenon was investigated by contacting a solution
of potassium or sodium isopropoxide with different molecular sieves,
Methyl alcohol was added to simulate its formation during the alcoholysis
of methyl esters. Samples were taken to determine alkoxide and cation
concentrations. In all the experiments reactants were added in quantities
similar to those in the alcoholysis experiments (see experimental part).
The exp. I-IV below were started at room temperature by adding a
solution of potassium isopropoxide in isopropyl alcohol and benzene to

21 Potassium alkoxide molecules will exist mainly as ion pairs, owing to the low di-
electric constant of the benzene solvent. Cf. D. J. Cram, "Fundamentals of Carbanion
Chemistry", Acad. Press, New Y o r k (1965), p. 32.
22 J. Hine and M. Hine, J. A m . Chem. Soc. 74, 5266 (1952).

48
the sieves. After stirring for 10 min, the suspension was heated to reflux
temperature. Methyl alcohol was added after several hours.
I. 3A powder. In this case no adsorption of alkoxide occurred. About
10% of the potassium present in solution was exchanged for sodium (3A
is prepared by partial replacement of sodium in 4A by potassium). This
ion exchange took place at room temperature. An additional experiment
with potassium isopropoxide replaced by sodium isopropoxide confirmed
the fast cation exchange with sieve. Again no adsorption of alkoxide was
observed.
II. 4A powder. The result is shown in Fig. 2a.
III. 4A pellets. The result is shown in Fig. 2b. An X-ray powder photo-
graph (Guinier-de Wolff camera) of the sieve after the reaction, was com-
pletely identical with a photograph of the original sieve proving that the
crystaUinity of the sieve had been preserved.
IV. 5A powder. The result is shown in Fig. 2c. No calcium was detected
in solution.
V. To study the effect of methyl alcohol upon the adsorption of alkoxide
in exp. II, methyl alcohol, isopropyl alcohol, and benzene were refluxed
with 4A powder. Within 25 min the methyl alcohol was completely ad-
sorbed. After 3 h a solution of potassium isopropoxide in isopropyl alcohol
was added and the refluxing was continued. As in exp. II, very fast cation
exchange took place. The alkoxide concentration decreased only slightly
slower than in exp. II.
VI. Exp. II was repeated at 25°. Cation exchange was complete within
2.5 h. After addition of methyl alcohol the alkoxide concentration decreased
although much slower than in exp. II: 18 h after methyl alcohol addition,
one third of the alkoxide was stiU present.
i) Cation exchange
It appears that exchange between potassium in a solution of potassium
isopropoxide, isopropyl alcohol, and benzene, and the cations in the 3A
and 4A sieves takes place at room temperature. With 5A, no cation exchange
has been observed even at reflux temperature. After addition of methyl
alcohol a slow cation exchange occurs with the residual sodium cations
but not with the calcium cations mainly present in 5A (5A is prepared by
partial exchange of sodium in 4A for calcium).
This effect demonstrates the lower mobility of cations in 5A. In this
connection we mention the work of Ducros ^3 vvho found that the cations
in calcium chabazite (a natural molecular sieve) have a much lower mobility
than those in sodium chabazite. The accelerating action of methyl alcohol
must be due to an increased mobility of the cations in accordance with

23 P. Ducros, Bull. Soc. Franc. Mineral. Crist. 83, 85 (1960).

49
 MeOH

a A A powder

.0 1 2 3 A
-4 » > time of reflux, h
10 mm

b ^ A pellets

i & ^

-»• time of reflux, h

50 .alkoxide „ _^M.OH
c 5 A powder
^
- D -

'30

O
20 -A-
^•A-"

^
.--A- -A
NQ-" -A--A-
o4^ 0 1 2 3 7 8
< • • time of reflux, h

Fig. 2. Interaction of potassium isopropoxide with molecular sieves type A.

50
the work of Stamires ^*, who found that the conductivity of 4A is increased
by adsorbed water, ammonia, acetonitrile, and «-hexane.
ii) Alkoxide adsorption
The decrease in alkoxide concentration after addition of methyl alcohol
in exps. II, III, and VI with 4A sieve must be due to adsorption of sodium
methoxide. The molecular dimensions of sodium methoxide allow adsorp-
tion by the sieve. Because methyl alcohol is adsorbed much faster (within
a few minutes) than sodium methoxide, a gradual desorption of part of
the methanol is required. This mechanism is confirmed by exp. V in which
methyl alcohol is added to 4A sieve before the addition of potassium
isopropoxide.
Exp. I shows that 3A sieve does not adsorb potassium methoxide,
presumably because of the smaller pore diameter compared with 4A sieve.
Logically, 5A with a pore diameter larger than 4A, would be expected to
adsorb methoxide faster than 4A. In contrast it was found in exp. IV that
no measurable adsorption occurs. Several explanations seem possible:
1. It might be reasoned that the counter ion of methoxide has an influence
on the rate of adsorption. To check this possibility exp. IV with 5A powder
was repeated with potassium isopropoxide replaced by sodium isopropoxide.
Again no alkoxide adsorption was observed. Cation exchange with calcium
of the sieve was also unnoticeable.
2. Another possibility would be a stronger adsorption of methyl alcohol
by 5A compared with 4A, resulting in a lower concentration of methoxide
in solution and a consequent decrease in adsorption rate. To check this
possibility the methyl alcohol concentration in solution was measured after
refluxing a mixture of methyl alcohol, isopropyl alcohol, and benzene
with 4A and 5A powder. The equilibrium concentrations appeared to be
about the same.
3. It is noteworthy that in the case of 5A (Fig. 2c) the rate of cation
exchange is low while the rate of adsorption of alkoxide is even lower.
Thus it seems that the rate of alkoxide adsorption is governed by the
mobility of the counter ion which neutralizes the methoxide, in the sieve.
The decrease in alkoxide concentration in exp. Ill with 4A pellets before
methyl alcohol was added must be ascribed to a reaction with the binding
agent. Adsorption on the binder seems improbable considering the relative
slowness of the process.
Turning to the practical consequences in alcoholysis using the suspension
procedure, deactivation of the alkoxide catalyst may become a problem
when the rate of alcoholysis is low {e.g., when preparing r-butyl esters).
In these cases 3A or 5A powder should be used.

'* D. N. Stamires. J. Chem. Phys. 36, 3174 (1962).

51
Dehydration of t-butyl alcohol
During the synthesis of f-butyl esters it appeared that dehydration of
/-butyl alcohol sometimes occurred on the outer surface of molecular
sieves of type A, causing deactivation of the sieves by adsorption of water,
and also of the alkoxide catalysts as shown by a decrease in catalytic
activity and the precipitation of the sodium or potassium salt of the car-
boxylic acid.
We investigated this dehydration by contacting molecular sieves with
/-butyl alcohol at 80°. The influence of additional water and triethylamine
was also studied. From the decrease in /-butyl alcohol concentration
determined by GLC, the amount of water formed by dehydration was
calculated (Table V).

Table V

Dehydration of /-butyl alcohol by type A sieves *

Water formed by dehydration, g/100 g sieve

Added, g/100 g sieve Powders Pellets

3A 4A 5A 3A 4A 5A

— <0.2 <0.2 0.5 7.0 1.2 18.4

EtaN, 2 <0.2 <0.2 0.3 0.3 0.4 0.4

H2O, 2 0.3 0.3 6.0 5.1 1.0 16.1

H2O, 2; Et3N, 2 < 0.2 < 0.2 <0.2 0.3 <0.2 0.3
1

" Molecular sieve (1 g), thoroughly dried at 400° and lO""" mm, was contacted with
/-butyl alcohol (1 g), and benzene (4 g) for 4 days at 80°.

i) Powders
Any activity of the powders must be due to the outer surface, which
amounts to 2-10 m^/g. The activity of 3A and 4A powder is very low even
when water has been added. 5A powder is more active especially in the
presence of water. This is in accordance with the higher activity of bivalent
compared with monovalent ion-exchanged sieves for carbonium-ion type
reactions ^. The effect of water on the activity of 5A powder will be caused
by an increase in concentration of Bronsted acid sites formed by inter-
action of outer surface calcium-ions with water. Probably, Ca(OH)+-ions

52
and protons are formed. A similar activation by water has been observed
in the dehydration of /-butyl alcohol by CaX sieve ^s.

ii) Pellets
The pelletized sieve, according to the manufacturer, is prepared by
adding about 20 g of an inert clay binder to 100 g of the sieve. Clearly,
the binder is not inert towards the dehydration of /-butyl alcohol. When
comparing 3A, 4A, and 5A pellets, however, it appears that the binding
agents vary considerably in dehydration activity. Possible explanations are:
1. The clay may ion-exchange with the sieve, for instance during the
wet mixing in the manufacturing process, which may explain the excessive
activity of 5A pellets compared to 3A and 4A pellets,
2. The binding agent is not a well defined compound.
Water does not seem to have an accelerating effect on the activity of 3A
and 4A pellets. When water is added to 5A pellets, the sum of added and
formed water is almost constant, which means that dehydration stops as
soon as the sieve is filled completely with water (18 g/100 g of pellets).
An accelerating effect of water however was proven by two experiments
in which a contact time of only 8 h was allowed. The addition of 2 g of
water/100 g of sieve 5A enhances the dehydration from 4.6 to 9.4 g of
water/100 g of sieve.
The addition of triethylamine seems to neutralize the catalytically active
sites in all cases, also when water has been added.

Thus, if decomposition is a problem in alcoholysis, powder is preferred

instead of pellets and 3A and 4A instead of 5A. One should guard against
traces of water by careful drying of reactants, solvents, and glass ware, and
by complete activation of the sieve. When using the extraction technique
(Fig. 3) it is important to keep the sieve at a low temperature. Also, the
sieve can be pretreated with triethylamine (2 g/100 g of sieve). In the sus-
pension technique, decomposition is slowed down by the presence of
basic alkoxide. Overheating however should be prevented by stirring and
gentle heating.

Conclusion
In conclusion we advise the techniques and sieves mentioned in Table
VI, in the synthesis of "simple" esters:

« B. Gourisetti, J. Cosyns, and P. Leprince, Bull. Soc. Chim. France 33, 1085 (1966).

53
 Table VI
Advised conditions for ester interchange

Displacing alcohol Displaced alcohol

R20H Technique Molecular sieve R'OH

«-aliphatic alcohols
MeOH — —
EtOH E 3A pellets » MeOH
other S-A 4A powder or pellets MeOH, EtOH
E 3A pellets ^ MeOH
E 4A pellets MeOH, EtOH

secondary and branched- S A 4A powder or pellets MeOH, EtOH

primary alcohols S-A 5A powder or pellets /(-aliphatic alcohols
E 3A pellets » MeOH
E 4A pellets MeOH, EtOH
E 5A pellets 77-aliphatic alcohols

tertiary alcohols S-B 3A powder MeOH

S-B 5A powder n-aliphatic alcohols
E 3A pellets » MeOH
E 4A pellets MeOH, EtOH
E 5A pellets «-aliphatic alcohols

* 3A should be used in an extractor without cooling mantle, to bring the sieve to a

temperature of 50° or higher At room temperature the rate of adsorption of methyl
alcohol is low

Experimental Part
Materials
Molecular sieves were obtained from the Union Carb de Corporation Linde Division
and were activated in an oven at 400° and cooled under anhydrous conditions Different
batches were used Methyl 2-(3-pentyi)ben?oate and methyl 2-(2 4 dimethyl-3-pentyl)-
benzoate were prepared in this laboratory, and the other starting esters were obtained
commercially or prepared conventionally Catalvsts were prepared by dissolving sodium
or potassium in an excess of alcohol All reactants and solvents were thoroughly dried
with molecular sieves

Apparatus
The extraction apparatus used in ester interchange reactions is shown in Fig 3 The
main feature is a cooling mantle surrounding the molecular sieve section EffcLtive
cooling IS ensured when the pellets are under the liquid level in the extractor In this
way the adsorption of displaced alcohol becomes stronger while any undesirable reaction
of the sieve is slowed down Glass ware was dried at 100° GLC anahscs were performed
using different stationary phases Decomposition of I butyl esters during analysis was
inhibited by pretreating the column with Silyl-8

54
 Fig. 3. Extraction apparatus.

Ester interchange
Three examples are given, one of each of the three techniques.

Technique S-A. «-Butyl methacrylate (exp. 11).

To a 500 ml flask were added 15.5 g of methyl methacrylate (stabilized with hydro-
quinone), 100 ml of «-pcntane, a solution of 0.5 g of sodium in 11.1 g of n-butyl alcohol,
and 60 g of 4A pellets. The flask, which was equipped with a drying tube filled with 4A
pellets, was shaken during the night (15 h). A small sample was taken and analysed by
GLC: n-butyl alcohol appeared to be completely converted to n-butyl ester. The reaction
mixture was centrifuged and the sieve residue was thoroughly washed with /i-pentane.
The combined organic extracts were washed with water and dried with magnesium
sulfate. After the addition of hydroquinone, the product was distilled giving 13.7 g
(65%) of n-butyl methacrylate, b.p. 55-57°/14 mm, nl° 1.4238.

Technique S-B. /-Butyl benzoate (exp. 12).

To a 250 ml flask, which was equipped with a reflux condenser and drying tube, were
added 20 g of methyl benzoate, a solution of 1.2 g of potassium in 160 g of /-butyl alcohol,
and 60 g of 5A pellets. The mixture was boiled under reflux during a weekend (70 h)
by heating in an oil bath at 100°. GLC analysis showed a 100% conversion. The reaction
mixture was centrifuged and the sieve washed with n-hexane. The combined organic
extracts were washed with water, dried with magnesium sulfate, and /i-hexane and /-
butyl alcohol were distilled off. The residue (23.7 g, 90%) was distilled giving 21.7 g
(83%) of /-butyl benzoate, b.p. 99-101712 mm, n!,» 1.4883.

55
Technique E Diisopropyl terephthalate (exp 27)
To a 500 ml flask, which was equipped with the extractor (Fig 3) containing 60 g of
5A pellets, were added 10 g of dimethyl terephthalate, 250 ml of benzene, and a solution
of 0 2 g of potassium in 16 8 g of isopropyl alcohol A reflux condenser with a drying
tube was attached to the extractor. The mixture was boiled under reflux for 3 h when
GLC analysis showed complete conversion The mixture was washed with water, dried,
and the solvent and isopropyl alcohol were distilled off The residue (11 7 g, 90%, m p
55-56°) was recrystaflized from ethanol/water to give 11 3 g (87%) of diisopropyl tereph-
thalate, m p 55-56°.

Position of the alcoholysis equilibrium and rate of alcoholysis

The alcoholysis of methyl heptanoate with isopropyl alcohol will serve as an example
To a 150 ml flask thermostatted at 25 were added 0 33 g (3 4 mmoles) of potassium
isopropoxide in 47 2 g (787 mmoles) ot isopropyl alcohol (concentration of potassium
isopropoxide determined by titration of a sample with 0 005 A'-hydrochlonc acid using
a pH-meter), and 50 ml of cyclohexane A reflux condenser with drying tube, which
contained 3A pellets, was attached to the flask A mixture ot 5 45 g (38 3 mmoles) of
methyl heptanoate and 1 54 g of /i-decane (ineit marker) was added Samples of 0 5 ml
were withdrawn through a rubber septum and shaken with a saturated aqueous solution
of sodium chloride The organic layer was dried with magnesium sulfate and analysed
by GLC The plot of log [methyl heptanoate] against time showed a linear decrease
during the first 7 mm Then the curve flattened because of the decrease in the concentration
of potassium isopropoxide due to the equilibrium [3], and because of the reversed reaction
of the equilibrium [2] becoming important From the slope of the first part of the plot
a A: 1 of 1 9 1 • mol ' mm ^ was calculated The samples taken after 26 and 52 h had the
same composition An equilibrium constant K of 0 41 was calculated No correction
for alkoxide was apphed
The other experiments were conducted similarly using n-decane, /i-undecane, or n-
dodecane as the marker

Alcohol/alkoxide equilibrium
Gas chromatographic method
Analyses were made at 130° using a 1 m copper column, 4 mm internal diam , filled
with Poropak Q (column packing manufactured by Waters Associates, Inc) After
each analysis, 0 5 ml of water was injected to destroy any alkoxide remaining in the
column One experiment will be described m detail The other two experiments were
carried out similarly
To a solution A of 39 6 g ol benzene and 9 74 g (162 mmoles) of isopropyl alcohol,
small portions of methyl alcohol were added After each addition, the mixture was
analysed (Table VII) A plot of theraUo of the peak heights of methyl alcohol and benzene
against mmoles of methyl alcohol added, was a straight line To a solution B of 39 6 g
of benzene, 9 74 g (162 mmoles) of isopropyl alcohol, and 0 80 g (20 4 mg-atom) of
potassium, small portions of methyl alcohol were added Again the mixture was analysed
after each addition (Table VII) Methyl alcohol in solution B was determined using the
plot obtained with solution A By subtracting this value from the amount of methyl
alcohol added, the amount of potassium methoxide was calculated The amount of
potassium isopropoxide and isopropyl alcohol were obtained by subtraction and Ki
was calculated (Table VII) From the data an average value of 20 is obtained.

56
 Table VII

Equilibrium constant for MeOH + isoPrO^ ^ MeO~ + isoPrOH

MeOH added, mmoles 2.48 4.96 9.92 19.8

A: MeOH/CeHe » 0.077 0.14 0.31 0.55

isoPrOH/CeHc " 1.23 1.22 1.23 1.22 1.23

B: MeOH/CsHe " 0.025 0.043 0.10 0.25

isoPrOH/CoHe " 1.09 1.14 1.12 1.18 1.20
MeOH, mmoles — 0.85 1.45 3.5 8.6
MeO K.+ , mmoles — 1.63 3.51 6.4 11.2
isoPrO K+, mmoles 20.4 18.8 16.9 14.0 9.2
' isoPrOH, mmoles 142 143 145 148 153
A:, — 15 21 19 22

" Ratio of gas chromatographic peak heights.

NMR method
A solution A of 510 mmoles of benzene and 169 mmoles of isopropyl alcohol, and a
solution B of 510 mmoles of benzene, 149 mmoles of isopropyl alcohol, and 19.7 mmoles
of potassium isopropoxide were made, and mixtures of A and B were investigated by
NMR using a Varian A-60 spectrometer. The NMR spectra of these solutions show
a heptet belonging to the methinc proton of the isopropyl group. The chemical shift of
this proton in Hz from TMS (60 MHz) was measured by the conventional side band
technique (Table VIII).

Table VIII
Chemical shift of the methine proton

isoPrO'/isoPrOH 0.000 (sol. A) 0.040 0.062 0.084 0.132 (sol. B)

Chemical shift, Hz 234.3 236.8 237.7 238.9 240.4

The addition of 22.3 mmoles of methyl alcohol to solution A did not alter the chemical
shift (234.3 Hz). Upon addition of 22.3 mmoles of methyl alcohol to solution B a chemical
shift of 239.2 Hz was observed. Using the relation between chemical shift and composition
of the solution (Table VIll), the solution was calculated to contain 15.0 mg-ions of iso-
propoxide, 154 mmoles of isopropyl alcohol, 4.7 mg-ions of methoxide, and 17.6 mmoles
of methyl alcohol. An equilibrium constant ^i of 3 is obtained from these data.

57
Ion exchange and adsorption of alkoxide on molecular sieves
A thermostatted 400 ml flask fitted with a reflux condenser and drying tube, a rubber
septum for sampling, and a magnetic stirrer, was used. In the experiments, about 2 g
of potassium or 1.2 g of sodium dissolved in 20 g of isopropyl alcohol, and 80 g of benzene,
5 g of methyl alcohol, and 50 g of sieve were used. Samples were withdrawn with a syringe,
put into 1 ml tubes, and kept at 60° to prevent possible precipitation of sodium iso-
propoxide. After centrifuging, a sample (0.20 ml) was put into a 100 ml beaker with 40 ml
of water, and titrated with 0.005 A'-hydrochlonc acid using a pH-meter. The amount
of alkoxide could then be calculated. The neutralized aqueous solution was diluted to
100 ml and analysed for sodium, potassium, and calcium, using a Kipp flame photo-
meter.
The following batches of molecular sieves were used: 3A powder, 3139; 4A powder,
450337; 4A pellets Vie', 451184; 5A powder, 540033.

Dehydration of /-butyl alcohol

A pyrex glass tube blown out at one end and having a ground glass joint at the other
end was weighed, filled with molecular sieve (pellets had to be ground), and connected
to a high vacuum line. After activation for two days at 400° and lO-* mm, the bulb was
sealed off while still under vacuum. The weight of activated sieve was determined. The
sealed bulb was placed in a pyrex glass tube, a 20% solution of /-butyl alcohol in benzene
(5 g of solution to 1 g of sieve) and water and triethylamine (Table V) were added, and
the tube was cooled down in liquid nitrogen and sealed off under vacuum. By shaking
the sealed tube, the thin-walled glass bulb was broken. The tube was rotated in a water
bath at 80° for 4 days. After ccntrifugmg, the tube was opened and the reaction mixture
analysed by GLC From the ratios of the peak areas of /-butyl alcohol and benzene
before and after contact with sieve, the amount of water formed was calculated.
The following batches of molecular sieves were used: 3A powder, 3139; 3A pellets
Vie", 348048; 4A powder, 450337; 4A peflets Vie", 46578; 5A powder, 540033; 5A
pellets Vs", 53793.

Acknowledgements
We wish to thank Mrs. G. van Minnen-Pathuis and Messrs. K. Breebaart,
H. van Doesburg, A. Scholten, A. B. Smaal, J. L. M. Syrier, and C. G.
Voermans for experimental assistance, Mr. M. van Leeuwen for carrying
out the elementary analysis, and Mr. A. Sinnema for recording and inter-
preting the NMR spectra.
Two of the authors (J. W. M. de G. and J. A. H.) gratefully acknowledge
financial support from the "Delfts Hogeschool Fonds".

58
 CHAPTER IV
THE SYNTHESIS OF ENAMINES AND KETIMINES USING
MOLECULAR SIEVES^
BY

D P ROELOFSEN and H VAN BEKKUM

The synthesis ol enamines and kLlimines lioin the. coiicipondmg carbonyl

compound and amine in high yield at room temperature is described using
a commercial silica-alumina cracking catalyst to catalyze the reaction and
a suspended molecular sieve type A to shitt the equilibrium The mixture
ot molecular sieve and silita-alumina can be successtully regenerated
Dillcrcni solvents catalysts and molecular sieves have been tested The
combination ol cyclohexane silica alumina and molecular sieve type 5A
gives the best results besides the pure powder pellets type 5A can also be
used m combinalion vviih silica-alumina catalyst
The moderate calalvtic ellecl of pellets as such as reported in the lilera
turc has been traced down to the binding agent present m pelleli/td material

Introduction
A general method' - ol preparing enamines and ketimines consists ol re-
Huxing d solution ot the corresponding carbonyl compound and amine,
and removing the water from the condensate e g by using molecular
sieve-' " According to a second synthesis' - a drying agent is suspended
in the reaction mixture This technique has the advantage ot being simple
and rapid while it can be operated at d lower temperature The use ol

'PAS Snulh The Chemistrv of Open-Cham Oiganic Nitrogen Compounds Vol I

W A Benjamin Inc New York (1965) p •?27
/ W Hcnius m Enamines Synthesis Structure and Reactions A O Coaly. (Ed),
Marcel Dekker New York (1969) p 55
T P/iunsidndB Tiiiuh J Org Chem 27 1041 (1962)
•*i S mm-koMc- Advan Org Chem 4 II (1961)
^G Stoik A Bir-oUiia H Luiulc snuiil I Szmu\:ko\u: AnA R Poull J Amer Chem
Soc 85 207 (1%^)
" F A Mm lid Miif't J W K o / m i u n a n d t A Puinlii Rec Trav Chim 89 1258(1970)

// Reprinted from Recueil Travaux Chimiques des P a y s - B a s , vol. 9 1 ,

page 605, May 1972, by p e r m i s s i o n of the copyright owner.

59
molecular sieve has been ddvocdtcd toi this purpose^" '" Molecular sieves
torce these equilibrium reactions to highei conversions as do conventiondl
drying dgents The molecular sieve should admit wdter while d!l other com-
ponents ot the leaction mixture should be excluded so as to minimi/e the
residudl wdter conLentration in the solution
Recently Taguchi dnd Wcsiluumi'" icpoiicd that then moicculdr sieve
type 5A mateiial suspended in the reaction mixtuie served not only as d
Wdter adsorbent but diso as a cdtdlyst Foi scvcrdl rcdsons d o the smdii
external surlace died ot the sieve crystals we considered it doubttui thdt
the sieve itselt would be the source ot this catdl>tic activity
The present pdper contdins d study ot the cdtalysis ot the leaction be-
tween Ldrbonyl compounds dnd dmines An impioved technique ot pie-
pdring endmines dnd ketimines has been developed

Results and discussion

R( cic lion of c 1 cloln \anonc and du th\ lammc
The rdte ot reaction of cyclohexdnone and diethylamine yielding l-diethyl-
dminocyclohexene hds been studied in ether using dillcrcni molecukir sieves
and cdtdlysts (see Tdble I)
The experiments 1 - 5 clcdrly prove that d large dillerence in activity exists
between pelletized molecular sieve 5A and pure moleculdr sieve powder 5A,
the same holds for 3A and 4A This difference must be due to a catalytic
activity ot the 20 wt"„ huulm^ af^cnt used in the mdnutdctuiing of pellets
Catalytic activity ol the binding agent hds been described earlier'' Ap-
pdrently Taguchi dnd Westheimer^^ have used a moleculdr sieve matenal
which contdined d binding agent, their stdtement thdt the molecular sieve
catalyzes the reaction is strictly speaking not true From the ddtd in Tdble 1
we conclude that the 5A mdtenal of Taguchi and Westhemwi'" hdd d Idrger
catalytic dctivity than all three bdtches ot 5A pellets tested by us (exp 1,
3 dnd 4), dIso when duplicdling the prepdrdtion ot three other compounds
with our 5A pellets 1/8 , we tound d 5— 10 times lower rate than found by
Taguchi and Wcsilwimei^'^ The scatter in activity ot different bdtches ot
the same molecular sieve might be due to the use ot different binding dgents

" R Bonnett and T R Emeison J Chem Soc 1965 4508

^ K C Brannock A Bill R D Buipiii dnd C A Kilh J Org Chem 29 801 (1964)
" E P K\hii Org Prep Proced 2 149(1970)
'" Ka-iio Taguchi dnd F H Wisihcinni J Org Chem 36 1570(1971)
''DP Roclo/itn J W M ck Giaal J A Hagendoorn H M Vcrschooi and // \cm Bikkuni
Rec Trav Chim 89 191(1970) This thesis chapter UI.

60
 Table I

The rate ot reaction between cyclohexanone and diethylamine in ether

Exp Moleculdr sieve Added catalyst UoX""!

1 5A pellets 1/8 _ 15
-> 5A pellets 1 8 powdered - 15
3 5A pellets 1 16 batch 1 - 65
4 5A pellets 1 16 batch 11 - 30
5 5A powder - >100
6 5A powder K2CO3 anhydrous >100
7 5A powder CaClj anhydrous >100
8 5A powder pTsOH H^O >100
9 5A powder Amberlyst 15 H* torm >100
10 5A powder lOX mol sieve powder >100
11 5A powder SK 500 mol sieve powdered 30
12 5A powder H mordenite pow der >100
n 5A powder S1O2' 20
14 5A powder A1,0/ 47
15 5A powder SiOj A 1 , 0 , ' 09
16 5A pellets 1/8 S1O2 A l ^ O j ' 29
17 ^^ pellets 1/16 — 15
18 3A powder - >100
19 3A powder SiO, AljOj' 00
20 4A pellets 1/16 - 75
21 4A powder - >100
22 4A powder S1O2-AI2O3' 09

0 10 mole of cyclohexanone 0 12 mole ol diethylamine and 80 ml of ether were shaken

at room temperature with 40 g of molecular sieve pellets (containing about 8 g of binding
agent) or with 32 g of molecular sieve powder and 8 g of added catalyst
*• Time required to obtain a 50''„ conversion ol cyclohexanone to 1 dicthylaminoeyclohexene
Aerosil with a BET surface of 190 m^ g
•^ Aeroal with a BET surface of 100 m^ g this was the most active alumina sample out of three
' Commercial cracking catalyst see Experimental Pari lor specifications Three other silica-
alumina cracking citalysts with a lower ilumina content and or with a lower pore volume
were less active but not by more than a factor 2

by the manufacturer The Union Carbide Corporation* confines itself to

saying that the binding dgent is "d low surface area clay which becomes
essentially amorphous on firing" We hdvc determined the BET surface
(nitrogen at —178 C**) of pure molecular sieve type 3A powder and of
3A pellets 1/16 , values ot 10 and 35 m^/g, respectively were obtained

'Private communication o{ R L Maw Union Carbide Corporation Linde Division

Tonawdnda (N Y )
** At ihis temperature the inner surface area of the molecular sieve crystals is not determined

61
Assuming the presence of 20 wt"„ ot binding dgent in the pellets the sur-
face drcd ot the binding dgent is estimdted to be 135 m' g
Table 1 compdres the catalytic dctivity ot severdi compounds Commei-
cial silicd-dlumina crdcking catalyst appcdrs to be the most dctive When
crdcking cdtdlyst is ddded to 5A pellets (exp 16). the ledction is accelerated
but seems limited by the rdte ot water ddsorption bv the pellets
The dctivit> of silicd-dlumind is ascribed to the presence ot acid sites''
which will enhance the reactivity ot adsorbed cdrbonyl compound It is
the Idrge surface ared the high porosity dnd the high concentration ot dcid
sites which seem to mdke the commercidl silicd-dlumina cracking cdtdlvst
so well suited for the rcdction studied According to Gokhum^- dlumina
hds d Idrge concentration ol dcid sites while silica hds not in dccoiddnce
with their cdtdlytic activities in Table I The binding dgent being d clay
dlso contains dlumind. this could explain the moderate catalytic activity
of pellets Stiong Bronsted acids such ds/^-lolucnesulfonic acid and Amber-
lyst 15 (d mdcroporous sulfondted-polystyrenc cdtion exchanger) dre less
active probably because the protonation of amine predominates
The inHuence ot severdi parameters on the dctivity of the silicd-dlumind
Cdtdlyst has been investigated The reaction appeared to be roughlv tiist
order in the quantity of catalyst Paiticle si/c influenced the activiiv as
follows d sieve traction ot 20 — 60 \x was 1 5 times as active as a traction
ot 90—150 n An incieasc in temperature ol calcination ot the catalyst
from 200 to 600 C did increase the activity by a factor 2 Regeneration of
the mixture of silica-alumina and 5A powder by preheating at 120 C
followed by calcination at 400 C was tound possible without reducing
Its dctivity
Silicd-dluinind Cdtdlyst has been tested in several solvents and with dit-
lerent moleculai sieves. Table II gives the results Invariably the highest
conversion ot cyclohexanone was obtained when using cyclohexane Be-
cduse ot the apolar chdutctei ot cyclohexane and its size (6 I A) the selec-
tivit) lor water adsorption from this medium is high Type 5A sieve ap-
pears to be supciior to "^A and 4A Appaientlv at this low degree ot loading
(about ^ g ot water on 100 g ot sieve) the specific inteiaction ot water with
the bivalent calciuip cations in 5A is stiongei than the interaction with
potassium or sodium cdtions in 3A oi 4A In tdct when using hdit the
dmount ot sieve in a separate experiment the ditlercnces nedrlv disdppcdrcd
becduse this interaction becomes relativelv less impoitant The good results
with 5A sieve also show that the adsoiption ol dicthvlamine bv SA is very
slow at room tempcratuie

\1 S (lolcl'.nin m Experiment il Methods in C il ilvlie Rese ireh R B tihlii\(>ii (fd )

Academic Press New York (19681 p 361

62
 Table II

The reaction ol cyclohexanone and dielhylamme using siliea-dlunima catalyst

Exp Molecular sieve Solvent Uo% 'f h Final con-

version in "„

23 3A powder ether 09 80 -
24 3A powder cyclohexane 1 2 88
25 4A powder ether 09 75
26 4A powder cyclohexane 1 2 82
27 5'\ powder ether 09 89*
28 5A powder cyclohexane 1 2 96"
29 5A powder benzene 23 S?"
30 5A pellets 1/8 ether 19 95
5A pellets 1 8 cyclohexane 2 1 97
3-' 5A pellets 1/8 benzene 62 93

0 10 mole of cyclohexanone and 0 12 mole of diethylamine were contacted at room tem-

perature with 32 g of molecular sieve powder 8 g of sihed-dlumind and 80 ml of solvent
or with 40 g of molecular sieve 5A pellets 18 8 g of sihca-alumina and 40 ml of solvent
The conversion decreased somewhat after this maximum probably due to a very slow
adsorption of diethylamine by the sieve

Table 111

The synthesis of several enamines and ketimines"

Exp Reactants t,oy m h tioo% *" m h Yield' m '

33 cyclohexanone dielhylamme 1 2 24 82
34 cyclohexanone dielhylamme 1 1" 24" 91<i
•(5 aeetophenonc aniline 0 1 1 85
^6 cyclooctanone morpholme 26 120 89
37 camphor aniline 12 70 95

•' 0 10 mole of ketone and 0 12 mole of amine were contacted al room lemperature with 32 g
of moleeOlar sieve powder lype 5A 8 g of silica-alumina calalysl and 80 ml of cyclohexane
'' Reaction lime needed to get a virtually complete conversion
Yield after purification by dislillalion or crystallization (acelophenone anil)
The physical constants and IR and N M R spectra of the c o m p o u n d s synthesized agreed
well with l i l e r a t u r e ' "
'' Here 0 20 mole instead of 0 12 mole of diethylaniine was used

By working with a larger excess of diethylamine higher conversions were

obtained which might be useful for preparative purposes Thus, by using
0 20 mole instead of 0 12 mole ot diethylamine in the experiments 2 7 - 2 9 ,
the conversion rose from 89 (exp 27) to 97%, from 96 (exp 28) to 99 4%
dnd from 87 (exp 29) to 97%.

63
Picpaialion of sewial cnammcs and kclimmcs
The procedure using cyclohexane as the solvent, 5A powder as the ad-
sorbent, and silica-alumina catalyst was selected for some preparative ex-
periments Table III gives reaction rates and yields
A comparison with the corresponding data of Taguchi and Westheimei'"
shows that the present method gives higher reaction rates as well as higher
yields
Experimental part
Mall I lah
H mordenite is a svnthelie molecular sieve powder manufactured bv the Norton Companv
under ihe trade name Zeolon lOOH The other molecular sieves were obtained from the Union
Carbide Corporation Lmde Division All sieves were calcined in an oven dt 400 C
The siliea-alumina catalyst trade name Ketjencat was obtained from the Koninkli)ke
Zwavelzuurlabneken v h Ketjen N V Amsterdam It is a fluid cracking catalyst grade HA-
-HPV (High Alumina High Pore Volume) with the following specifications particle size
between 20 and 150 n, alumina content of the cdlcined cdtdlyst 26 0 wt°(, surface dred and
pore volume alter steaming lor Ih at 600 C 517 m^ g and 0 88 ml g respectively A corre
sponding fluid cracking catalyst of the American Cyanamid Compan> (Aerocal Triple A
High Alumina with 25 wt , alumina and a pore volume of 0 89 ml g) gave similar results
v\hen replacing Ketjen silica-alumina in exp |5 of Table I a t,o%,of 1 3 h was found Silica
-alumina catalysts were calcined belore use at 400 C unless stated otherwise
Pure grades of dr> reactdnls dnd dry solvents were used Of the catalvsts used in Table 1
the mdcroporous rcsin Amberlyst 15 (particle si/e 0 1-1 0 mm) uds diied ai I2U C while
silica and alumina were calcined at 400 C
Anal\ si<i
The reaction of cyclohexanone with diethylamine was followed b) GLC analysis of the re-
action mixture A 2 m copper column internal diameter 4 mm filled with 10'„ SE 30 on
Chromosorb W was used at 110 C with hvdrogen as the earner gds dnd d hedl conductivity
detector Corrected peak areas of cyclohexanone and l-dieth>laminocyclohexenc were used
to calculate the conversions
The other reactions in Table 111 were followed by IR dUdlysis After settling of sieve and
catalvst d sample of the solution was examined in a cell ol 0 018 mm thickness The carbon) 1
absorption band and an absorption band of the enamine or ketimine (exp 35 8 30|i exp 36
6 12n exp 37 5 94n) were used to tollow the reaction progress
S\nlhc lie pi oc I dun
One example on the preparation of ( + )-camphor anil (exp 37) is given 15 2 g (0 10 mole)
ol ( + )-cdmphor 11 2 g (0 12 mole) of aniline 80 ml of cyclohexane 32 g of molecular sieve
5A powder and 8 g of siliea-alumma ealalvsl were shaken at room temperature After 70 h
the carbonvl-absorption band was no longer detected in the IR spectrum The suspension
was littered and the sieve was carefullv washed with several portions ot cjclohexane totalling
250 ml The solvent was removed from the solution with a flash evdporator and the residue
was distilled giving 21 5 g (95''„) of ( + )-cdmphor dnil b p 106-108 C 0 6 mm n^^ 1 5401
m p 15-16 C

Ackno\> ledgements
We w ish to thank the Ketjen Corporation for samples of cracking catalysts
Mr C \«n/s('/;/)«i? for checking the manuscript and Ir /• \cin dci Heijdcn
for the determination of BET surfaces

64
 CHAPTER V Part 1
SYNTHESIS OF ACETALS USING MOLECULAR SIEVES '^
BY

D P ROELOFSEN, E R J WILS, and H VAN BEKKUM

Acetdls cdn be prepared in high yield fioin the corresponding aldehydes or

ketones dnd alcohols using moleculdr sieves lo shift the equilibria by adsoiption
of the water formed Examples include acetals ol acetone cyclohexanone iso
butyraldehyde and benzaldehyde
Two batch techniques have been applied using the molecular sieve suspended
in the reaction mixture at 0° oi room temperature
The fiist method involves the use of dual function piolon exchanged molec
ular sieves type Zeolon 500 AW 500 oi mordenite I he outer surface ol the
sieve catalyzes the acetal toi malion where is the inner surface selectively adsorbs
the water formed
According to the second pioeedure moleculai sieve and catalyst /> toluene
sulfonic acid are added to the reaction mixtuie In this case proton exchange
between catalyst and sieve occuis which dppedred to be fast with the moleculdr
sieves types KA and NaA much less pronounced with the types CaA Zeolon
500 and AW 500 All these sieves however can be used because of the high rate
of acetal formation With KA and CaA essentially quantitative conversions aie
obtained within a tew minutes

Introduction
Fhe dcid-catalyzed formation of acetals from aldehydes or ketones and
dicohols IS represented by the overall equilibrium (1)
^ ^ OR
C=0 -I- 2 ROH i?^: ^C -I- HjO (1)
OR

For aldehyde acetals' and a few ketone acetals'"^ the equilibrium con-
stant K IS sufficiently large to permit then preparation by using an excess
ot alcohol FOI most ketone acetdls however, K is smdil and a reasonable

' £ Schmiu dnd I Luhhorn m The Chemistry of the Fiber I inkage S Palaiied)
Interscience Publishers. London (1967) p 314
' N B Lorelte W L Howard dnd J H Bnnin Jr J Org Chem 24 1731 (1959)
'' R E MiCm.A W Baker dnd R S Oohlkc J Org Chem 22 1175(1957)
" C Djeiassi L A MiluherdndB J MiUclur J Am Chem SCK 81 947(1959)
J M Bill D G KuhUi P SuilnclldndR G /cpp J Org Chem 30 4284(1965)

/ Reprinted from Recueil Travaux Chimiques des Pays-Bas. vol 90, page 1141,
September. October 1971. by permission of the copyright owner.

65
conversion is only possible by adding a substance such as tetraalkyl
oithosilicate, dialkyi sulfite, acetone dimethyl acetal, or trialkyl ortho-
formate. which reacts chemically with water'. The azeotropic removal of
water is in general inadequate, one of the reasons being the temperature
dependency^-of the position of equilibrium (1).
The formation of acetals is catalyzed inter cilia by acids such as hydro-
chloric acid and />-toluenesulfonic acid (/)TsOH), by molecular sieves
with acid properties''', and by acid ion-exchange resins-.
We have investigated the use of molecular sieves (zeolites) suspended
in the reaction mixture, to remove the water selectively at low tempera-
ture. The use of molecular sieves to shift equilibria has been reviewed
shortly in an earlier paper**. Recently another three applications have
been published""".

Results and discussions

Two batch techniques have been considered by us. The first method
involves the use of a dual-function proton-exchanged molecular sieve with
the outer surface catalyzing the reaction and the inner surface shifting
equilibrium (1) bv adsorbing the water formed. According to the second
technique molecular sieve and acid catalyst (/)TsOH) are added to the
reaction mixture. An inherent problem is thiit proton exchange between
acid and sieve occuis which leads to a decrease in the concentration of
the catalyst,
The ideal moleculai sieve should adsorb water and exclude all other
components of equilibrium (1). Of the commercially available sieves
only type KA (estimated pore diameter 3 A) meets these requirements
(methanol forms an exception being adsorbed slowly at room tempera-
ture) Also larger-pore zeolites have been tested including NaA (4 A).
CaA (5 A), the chaba/ite (5 A) containing commercial materials Zeolon

.An allemalive pu-ecduie mav be visualized in which .i mixture of reagents is forced

through a column hlled with molecular sieve.
' L A Haiiiilioii and ,\ J Pilman. U. S patent V454.596. Chem. Abstr. 72. 428 lOf
(1970)
' P B.Vc'iiito dnd P S L(m<//^. Advan Catalysis 18. 259 11968).
" D. P. Rc'clojsen. J. W M dc Graaj. J .A Havendoorn. H W I crst/ii^or and H. luii
Bekkum. Rec. Trav. Chim. 89. 193 (1970) T h i s t h e s i s chapter III.
'' Hii"ie, ) liki. Koichi Hatada. Kaziihiko ,\(H;iila and Ktn-uhi Kap\ama. Bull Chem
Soc. Japan 42. 3546(1969)
'" W. Oppol:erdnd K A'c/Zcr. Tetiahedion Led I l l 7 ( l 9 7 0 ) .
" £ P K\ha.Org Prep. Proced 2. 149(1970).

66
500 and AW 500". and synthetic H-mordenite (7 A). Coinmeiciai AW
300 (pore diameter 3 A according to information of the Union Carbide
Corporation) has not been included since adsorption measurements
showed a i datively low adsorption capacity and poor selectivity for
water.
First we investigated the selectivity ot diffeient sieves for water
adsorption and the relative activity of various catalysts for the formation
of acetal.

Selectivity Jor water adsorption

The adsorption of water by diflTerent sieves from a mixtuie of ethanol
and pentane-"* was followed with time by GSC using reactants in quan-
tities comparable to those in the actual preparation of the diethyl acetals.
f-or analytical leasons we preferred pentane to hexane which was used
in all other experiments. It must be noted that the amount of sieve was
always large enough to accommodate all the water. Table 1 gives the
amount of water in solution when equilibiium is reached between the
components in the sieve and in the solution.
Table I

Adsoiption ot water tiom <i watei-ethanol-pentane mixtuie hv molce-

ulai sieves'

lype of sieve 'A of watei not adsoibed Lquilibiation time

at equilibiiuni in min

KA 03 1
NaA 04 1
CaA 3 (0 5)" 360(1)"
Zeolon 500 4 30
AW 500 pellets 5 180
AW 500 pow del 5 60
H-AW 500 6 60
H-moidenite 8(6)" "i60|5)''

' 0 2 mole ot ethanol. 0 05 mole ot watei and 7S ml of peiilane weie

stiircd with lOg of sieve (20 i; ol 11 moidenite) at 0" Samples weie
taken attei 1 mm and then aftei suitable inteivals
" A minimum in the watei eoneenlralion vMth lime oeeuis Contei text

Zeolon ^OO and AW 500 appealed lo have veiy similai compositions and piopeities
For details see the Fxpenmental Pail
As solvents we chose alkanes because ot then non polaiitv Thus when using the pol.ii
solvent dioxane. the amount of watei in solution attei eontael with the sieve .ippeaied
to be loughb twice as Luge as when using pentane

67
 Clearly only KA and NaA adsorb water exclusively. The other sieves
adsorb ethanol and a minor quantity of pentane also. On CaA the adsorp-
tion of water is fast compared with the adsorption of ethanol causing the
watei concentration in the solution to pass through a minimum. A similar
kinetic effect, although less pronounced, is observed with H-mordenite.
I he slow adsorption of water by Zeolon 500 and AW 500 is noteworthy.

Ac lit itv of different c atalxsts

The formation of cyclohexanone diethyl acetal was used as a test reac-
tion foi the comparison of different catalysts. The time necessaiy to
reach half the equilibrium conversion of cyclohexanone to the acetal.
1,12. was determined The reaction was followed by GLC. The activities
of diffeient catalysts are shown in Fable 11. Details about the prepaiation
and pioperties of the sieves are given in the Experimental Part.
Table II
The catalytic aetivitv for the formation of cyclohexanone diethyl acetal'

Cilalvsl" HCl /)TsOH H-NaA H-CaA H-Zeolon H-AW H-mordenite

500 500

H"-exchange '"•? — — SO' 20" 22' 27" 92"

t,, m m <2 --2 -> 1000 400 5 is 4s

' 0 1 mole of cyclohexanone, 0 2 mole of elhdnol. 75 ml of hexdne dnd Cdlalvsl were

stilled at 20° 0 5 mgeq o f / J T S O H 01 concentiated hydioehlorie acid, and 0 5 g of the
sieves weie used as the catalysis The lon-exchangc capacities ot the sieves lange be-
tween 7 mgeq/g tor NaA and 2 mgeq/g foi moidenite
" Activation of the sieves at 400° in an oven except foi H-NaA and H-CaA which weie
activated at 200° under vacuum
' Calculated from the acid uptake duiing piepaiation
'' ( aleiilated fiom the elemental analysis and assuming that the cations H+ Na*. K", Cd'*.
and NTg neutiali/e the AlOj and LeO. groups

Homogeneous catalysis by HCl and /)TsOH appeared to be very fast.

Of the proton-exchanged sieves, H-Zeolon 500, H-AW 500. and H-
mordenite are reasonably active with the first two having the highest
cictivit\ although only Ihe oiitsRJc suifaces of these sieves are available
lot calahsis The nigh aetivitv ol H-Zeolon 500 and H-AW 500 com-
paied 1(1 H-NaA and H-( ,iA may be explained by the huge silica/
alumina ratio and the presence of multivalent cations in the first two
sieves (see Experimental Part). Both factors are known'' to enhance
the acidity of hydioxyl groups . Also the catalytic surfaces of H-Zeolon

' 'Ihe acidity ol the hydroxyl groups on the inside surface ot the moleculai sieves has
been studied". It seems permissible to apply the same reasoning to the aciditv ot the
hv diovvl groups on the outside surlace of the sieve crystals

68
500 and H-AW 500 are relatively large according to the cyclohexane-
adsorption data in Table V^ . The unexchanged sieves KA, NaA, CaA,
Zeolon 500, and AW 500 were omitted from Table II because then-
activity was found to be negligible.

The preparation of cyclohexanone diethyl acetal

The use of molecular sieves in the preparation of cyclohexanone
diethyl acetal was investigated in detail (Table HI). We used two tern
peratures and the two procedures already mentioned. For comparison
equilibrium experiments without sieve are included.
Exp. 1 and 2 show the large influence of temperature on this equili-
brium** which is of the same order as the recorded^ temperature effect
in the case of cyclohexanone dimethyl acetal. We also noted, in agree-
ment with the experience'^ with cyclohexanone dimethyl acetal, that the
addition of solvent to an equilibrated mixture of cyclohexanone and
ethanol has only a slight effect on the equilibrium conversion in apparent
contradiction with the stoichiometry of equilibrium (1).
Proton-exchanged sieves. H-AW 500 (exp. 3) is a very tthcient sieve.
The conversion of 86% may be further raised by removing the sieve and
adding a second portion erf 10 g of sieve which brings the coi ersion to
99.5% within an additional 60 min. After removing this sei^oiJ portion
of sieve, isolation of the reaction product gave pure cyclohexanone
diethyl acetal in 88% yield in a preparative example. With H-moidenite
(exp. 4) a maximum in the conversion shows up. Water is adsoibed
quite fast (cf Table I) by this sieve; subsequently part of it is desorbed
slowly by ethanol which has a lower adsorption rate. An experiment
with H-NaA (not included in Table III) gave unsatisfactory results as
was to be expected (Table II). By adding a small amount of H-mordenite
as a catalyst the efficiency of H-NaA was improved but the leaction
stopped far from the expected value. Presumably, protons of H n lO; den-
ite exchange with sodium ions of H-NaA, thus rendering the catalyst
inactive.

" We wondeied, howevei. whether the activity of H-Zeolon 500 and H A W 500 could
be ascribed to impurities So we measured the activity of/)TsOH-lreated (as in Ihe
prepaiation of H-AW 500) kaolin (surface area 4 m-'/g), silica (190 m-'/g), alumina
(100 m^g), maghemite. and magnetite, but found no conclusive evidence However,
chabazite crystals from Ziegelberg, West-Germany, which were pure according lo the
X-ray diffrdction pattern, did show moderate activity after grinding and a siiiiil u-
/'-TsOH treatment.
•"' Because of the higher temperature needed we were unsuccessful in preparing evcii>-
hexanone diethyl acetal using an extraction technique" with molecular sieve KA pellets
in the extraction chamber
'' D.G.Kublei and L. £. iMce/icv, J. Org. Chem. 25, 1437(1960).

69
 Table 111

The piepaiation ol cyclohexanone diethyl acetal'

Exp Sieve Catalyst Temp " Time Conveision

mmoles X mm ',
1 — HCIO s 23 — ->")
2 — H( 1 0 5 0 — '1

3 H \WSOO 0 60 86
4 H moidenite — 0 60(180)' 79 (76)"

5 KA / ; T s O H 2 x K)' 0 2 x IS' 98
6 NaA /-fsOH 10 0 15 max 94'
7 CaA /iTsOH 10 21 1 (270)' 78 ( 6 7 ) '
8 CaA /)TsOH 10 0 1 (270)" 97 (86)"
9 Z e o l o n 500 /-TsOH 10 0 60 84
10 AW 500 pellets /;TsOH 10 0 380 78'-
11 A W 500 p o w d e r /JTSOH 10 0 190 84

0 05 mole of cyclohexanone 0 2 mole of ethanol and 75 ml of hexane weie

stilled with catalvst and sieve 20 g ot H moidenite or 10 g of the other
sieves weie used Samples weie taken after 1 mm and then attei suitable
intervals
Generally the tempeiatuie was consideiably higher (up to 15°) duiing the
first few minutes after the addition of the sieve due mainly to the I iige he.it
of adsorption of water
Conversion of cyclohexanone to acetal was determined by Ol C
A maximum in the conveision with time was observed See text
An extia amount of/iTsOH was added attei 1 5 mm Contei text
The ohlam.ible conveision depends eiilieall> upon the mannei in which the
reagents aie idded See text
Piobably equilibiium was not yet leached

Unexchanged sieies When using unexchanged sieves,/)TsOH was

ddded as a catalyst We equilibidted cyclohexdnone, ethanol, hexane.
and /iTsOH at the desiied temperatuie befoie sieve was added In some
pieliminaiy expeiiments we tound that this pioceduie gave the best
results The conversion is detei mined kinetically by the i ates of both the
foiwaid and the backwaid reaction of equilibiium (I), by the lates of
wdtei dnd ethdnol ddsorption, dnd ot exchdnge of catalytic piotons in
solution with the sieve Ultimately the selectivitv ot the sieve toi watei
(cf Table I) detei mines the thermodynamic value
With Zeolon 500 and AW 500 (exp 9, 10, I i) no inhibition ot ihe leac-
tion by proton exchange is expected since the pioton-exchanged sieves
themselves are active catalvsts (Table II) The leaction time seems to he

70
governed by the rate of Wdter ddsorption (cl Tables I and II) which ex-
plains the diffeience in leaction rate ot AW 500 pellets (exp 10) and the
same matenal aftei giinding (exp 1 I) The conversions with AW 500
aie somewhat lowei than with H AW 500 (exp 3) because /;TsOH
increases the equilibiium concentiation of watei in solution due to its
hydrated water and its polarity
In contrast the proton exchanged forms of KA NaA and CaA are
inactive as cdtdlysts ( Fable 11) and any proton exchange with these sieves
will lead to a decrease in the reaction rate
With KA the late of pioton exchdnge is very fdst When the sieve was
added to the equilibiated ledction mixtuie the rcdction stopped within
a few minutes due to the disappearance of the catalyst Depending upon
the way the sieve was added conveisions of 80 to 96% weie observed It
appeared that the highest conveision was obtdined by idpidly ddding the
KA-powder to the non-stiried ledction mixtuie By adding anothei
poition of/)TsOH cdtdlyst a conveision ot 989^ (exp 5) could always
be obtdined In d piepdidtive exdmple using the fourfold dmounts of
reactants pure acetal was isolated in 91% yield
NdA (exp 6) behaves similai to KA with the leaction stopping some-
where between 76 and 94% As with KA the highest conversion was ob-
tained by rapidly adding the sieve to the non stirred leaction mixture
Contrary to the experience with KA another portion of /;TsOH brings the
conversion to d final value of only 87% Doubtless this is due to the
partial pioton exchange of NaA which enlaiges the poie diametei of the
sieve'' Consequently the innei suiface is made accessible to ethdnol and
the selectivity ot the sieve foi watei decie.ises Unexchanged NaA
(cf Table 1) does not adsoib ethanol at 0° with a measurable rate
With CdA (exp 7 8) proton exchange is slow and limited to 2 0 % " A
distinct kinetic effect is shown which is explained by a tast adsorption of
water followed by a much slower adsorption of ethanol cdusing a partial
desorption of watei (ct Table I) Ihe chdnge ot the conveision with time
in exp 8 is given dSdn example 97 0%(1 mm) 96 7(2) 96 4(5) 96 2(10),
95 7 ( 1 5 ) 9 3 8(30) 9 1 2 ( 6 0 ) 87 0(180) 86 5(270) By neutializingthe
catalyst after 1 mm pure cyclohexanone diethyl acetal was obtained in
88?? yield in a piepaiative example using the fourfold dmounts of leac-
tants
The relative lates of pioton exchange with KA NaA and CaA aie in
dgreement with the ion exchdnge piopeities ol zeolites in diffeient aico
hols'^ and with oui work"* on the ion exchange of sodium and potassium
tons in an organic medium with KA NaA dnd CdA
' D.P. Roelofsen, E.R.J. Wils a n d / / i an Rekkum J Inorg Nucl Chem 34
1437(1972) T h i s t h e s i s c h a p t e r VI
* A D\ir dnd R B Gcttins Procds Inteinit C ont on Ion Exchange London July 1969
Soc Chem Ind london<1970) p 357

71
The preparation of other ac etcds
The preparation of several other acetals has been investigated using
KA or CaA with / J T S O H , OI H-AW 500 as such. The results are shown
in Table IV together with equilibrium values. In all experiments with KA
extra / J T S O H was added as in the preparation of cyclohexanone diethyl
acetal. With CaA kinetic effects are involved, the cause of which has
already been discussed.
Cyclohexanone. methanol. The equilibrium conversion (exp. 12. 13) is
favorable compared with that of cyclohexanone diethyl acetal (Table
HI). However, the reaction rate is quite low while the affinity of the
sieves for water is reduced considerably by the presence of the polar
methanol. Therefore a kinetic effect with CaA (exp. 17) is hardly visible.
In exp. 14 and I 5 a second portion of 10 g of H-AW 500, after removal of
the first portion, raises the conversion to 95 and 99.5% respectively, in
60 min.
Cyclohexanone, n-propanol. H-AW 500 (exp. 20) is a surprisingly
efficient sieve to shift this equilibrium. Probably the adsorption of n-
propanol is very sluggish under these conditions. On the contrary, with
CaA (exp. 22) the change of conversion with time proves a gradual
adsorption of/;-propanol.
Acetone, ethanol. The low equilibrium conversion of acetone with
ethanol (exp. 23) should be representative for that of other aliphatic
ketones with ethanol. Also in this case a reasonable conversion (exp. 24,
25) has been obtained. Higher conversions are possible by removing the
sieve and adding a fresh portion of sieve (and / J T S O H in the case of KA).
Isobutyraldehyde and henzaldehyde. ethanol. Very high conversions
of aldehydes to acetals have been obtained using molecular sieves.

Conclusions
From the results presented in this article it follows that molecular
sieves are well suited to prepare both aldehyde and ketone acetals in
high yield starting fi om the aldehyde or ketone and the alcohol.
In general a temperature of 0° will combine a favoiable equilibrium
conversion with a reasonable reaction rate and water-adsorption r<,ic
With methanol KA and H-AW 500 are best suited With ethanoi aiiu
higher alcohols also CaA can be used; for an optimal conversion the
reaction should be quenched a few minutes after the addition of C a.A.

Experimental part
liid<j;c'nl\
Pure giades of leagents ,ind solvents weic used I iqiiids woie Jiicd u>.ii. lo. i.,ui ii
sieves /I-1 oluenesultonic aeid monohydi.ite w.is obtained tiom B.ikoi <ind I's.^l v, j ,

72
 Table IV

The preparation of acetals'

Exp. Reactants Sieve Catalyst Temp. Time Conversion"

mmoles °C min 7c

12 cyclohexanone. — HCl 0.5 23 — 39

methanol
13 — HCI0.5 0 — 54
14 H-AW 500 — 21 200 85
15 H-AW 500 — 0 400 'M
16 KA / ) T s O H 2 x 10 0 2 x 15' ys.h
17 CaA /;TsOH 10 0 5 (400)'' 95.4(S*5.(I)"

18 cyclohexanone. — HCl 0.5 22 — T>

Ai-propanol
19 — HCl 0.5 0 — 26
20 H-AW 500 — 0 60 98
21 KA / ) r s O H 2 x 10' 0 2 x 15' 98
22 CaA /;TsOH 10 0 1 (400)" 98 (40)"

23 acetone, ethanol — //I sOH 0.5 0 — 10

24 KA / / r s O H 2 x 10 0 2x 15' 93
25 CaA /)TsOH 10 0 :(2ooi" 81 (50)"

26 isobutyraldehyde. — /) IsOH 0.5 0 — 68'-

ethanol
27 KA / i l s O H 3x 10 0 3x15' 4y.6
28 CaA //IsOH 10 0 1 (400)" W.7 (98.4)"

29 henzaldehyde. — //fsOH 0.5 0 — 52

ethanol
30 KA / ; T s O H 2 x 10' 0 2 x 15' 99.4
31 CaA /;TsOH 10 0 1 (400)" 98.9(90)"

0.05 mole of aldehyde or ketone. 0.2 mole of alcohol, and 75 ml eif hexane were stirred
with catalyst and 10 g of sieve. In the equilibrium experiment', I 2 and 13 no hexane was
added to prevent the formation of two liquid phase- Sampies were taken after I min
and then after suitable intervals.
Conversion of aldehyde or ketone ti< acetul was dctci mined In f.l C.
Another amount of/)TsOH was added after 1 5 min. See text.
A maximum in the conversion wilh lime was observed. See lex;
This value refers to the upper of the iwo layers which formed on equilibration.

73
Molec iilai sicic
All sieves were activated at 400° in an oven unless stated otherwise
KA NaA CiiA These sieves weie obtained as powders from the Union Caibide Cor-
poration
H inoidcniic Zeolon lOOH obtained as a powdei from the Norton Companv isdeseiibed
as a synthetic large-pore moidenite in which sodium ions have been leplaeed by protons
Oui analysis gave atomic latio of Na Al Si = 0 08 1 7 The satuiation capacity foi water
IS 20 g/100 got activated sieve
Zeolon '^00 AW 500 Zeolon 500 was obtained as a powdei tiom the Noiton Company It
is desciibed by the manulaetuiei as a natuial product containing loughly 60% chabazite
.ind 40*^ eiionite AW 500 was obtained in the foim of pellets from the Union Caibidc
Coipoiation
the X-ray diffiaetion patterns (Guinier de Wolfl cameia. C uK„ ladiation) ot Zeolon 500
and AW 5()0 were identical By eompaiison with the patterns of puie zeolites it was shown
that both materials contained chabazite and eiionite Ihe chemical compositions and
.idsorplion propeilies aie piesented in Table V
Erom the lesults in l.ible I II III and V it is concluded that Zeolon 500 and AW 500
have veiy similai piopeities Iheietore these eommeieial pioduets and then pioton-
exchanged foi ms can lepl.ice each othei in the applications deseiibed m this aitiele

Table V

Adsorption capacities and composition of dehydrated sieves

Sieve Ads capacity g/IOOg' Composition wt9f"

watei cyclohexane SiO Al,0, Ee O, CaO MgO (NaK),0

Zeolon 500 22 6 34 69 6 15 5 48 5 3 15 33
AW 500 19 2 27 69 4 1^ 8 47 47 15 t 9
H AW 500 17 0 18 72 7 156 47 42 14 1 5
CaA 25 6 <-0 2 — — — — — —
' Vapoi adsoiption capacities (24 h 45° p/p^ = 0 5) determined using the technique
deseiibed eailiei'
'' Analysis by X lay fluoiescenee as well as bv atomic absoiplion Tiaee amounts ol
many nioie elements have been detected
' Pait ot fhe non will be incoipoiated in the fiamewoik Howevei an non eontaining
matenal was obtained fiom Zeolon 500 using a magnet The X-rav diffractiein pdttein
was in accordance with the ASTM dilTiaction file p.itteinsof magnetite and maghemite
// Zeolon "lOO H-AW 500 The piepaiation of H-AW 500 is given H-Zeolon 500 was
piepaied the same way
A quantity ot 200 g ot powdeted AW 500 pellets was stiiied with 75 g ot /iIsOH
12 g ot water and 750 ml ot dioxane After 24 h the sieve was filtered washed fiist with
2 I ot DME to remove dioxane insoluble salts of pTsOH, followed by 200 ml of benzene
and activated by slowly incieasing the temperatuie up to 400° Proton exchange was
deteimined by tiUating a sample of the pTsOH/dioxanc solution The value obtained
agreed satisfactorilv with the analy sis in Table V
We also prepared a pioton-exchanged AW 500 sample by suspending AW 500 in water
and keeping the pH at a value of 3 foi 4 days bv .idding .iqueous hvdioehloiic aeid A

'•* D P Roelojsen, \nd\ Chem 43 631 (1971) T h i s t h e s i s chapter II.

74
proton exchange of only 22% was observed Therefore this sample showed a lower catalytic
activity and was found to be less suited for the preparation of acetals than H AW 500
prepared with / J T S O H in dioxane
H NaA H CaA These sieves were prepired in a similar way as H AW 500 See Tdble II
for the extent of proton exchange
Thermal slahilit) The crystalhnities of the proton exchdnged sieves were checked by
X rdy diffraction using a Guinier-de WolfT camera H Zeolon 500 and H AW 500 after
activation at 400° gave diffraction patterns identical with those of the unexchanged mater
lals Because of their thermal stability these proton exchanged sieves can very well be
reused Eor instance H AW 500 used in exp 3 after legeneration by gradual heating up to
400° gave results identical with those of the original material H NaA dnd H CaA after
activation at 200° in vacuum showed some loss in crystdllinity compared with NdA and
CaA Eull details about the preparation and properties of proton exchanged type A sieves
will be given in a future paper"

Gas chromatographic anahses

Gas chiomatographic analyses were performed using hydrogen as the carrier gas and a
heat conductivity detector Quantitative analyses were generally based on corrected peak
areas Low water concentrations however were calculated from peak heights
Acetal formation was followed using a 2 m copper column internal diameter 4 mm
filled with 109? SE 30 (silicone gum) on Chromosorb W at 80-100° or on a similar 3 m
column filled with I 5% PEG on Chromosorb W at 60-90° After stopping the stirrer and
allowing the sieve to settle a 0 5 ml sample of the reaction mixture was put into a small tube
containing 0 05 ml of triethylamine centrifuged and a 5 /xl portion was analyzed Decom
position of acetdls during dUdlyses was further inhibited by daily pietreatment of the
columns with triethylamine
Water analyses (Table I) were performed with a 1 m copper column internal diameter
4 mm filed with Poropak Q at 120°

Preparation of acetals
Three examples all concerning the synthesis of cyclohexanone diethyl acetal on a pre
pardtive scale are given using H AW 500 KA and CaA (cf exp 3 5 8)
H AW 500 0 I mole of cyclohexanone 0 4 mole of ethanol I 50 ml ot hexane and 20 g
of H AW 500 were stirred at 0° for 1 h giving a conversion of 869f (Gl C) The sieve was
removed by filtration and washed carefully with 1 50 ml ot hexane 1 o the filtrate 20 g of
tresh H AW 500 was added and the suspension stirred agdin for 1 h at 0° giving a final
conversion of 99 5'% The suspension was filtered and the sieve was washed with 1 50 ml of
hexane The combined oiganic solutions were washed with two 50 ml portions of an
aqueous I 57c sodium hydroxide solution and with two 50 ml portions of water After drying
with potassium carbonate solvent and ethanol were removed with a flash evaporator and
the residue was distilled giving I 5 0 g (889f) of cyclohexanone diethyl acetal b p 75-77°/
15 mm «o" 14347 purity (GLC) 99 8%
KA 0 2 mole of cyclohexanone 0 8 mole of ethanol 300 ml of hexane and 0 04 mole of
/'T?OH were stirred at 0° The stirrer was stopped and 40 g of KA powder was added rdpid
ly After 1 min the stirrer Wds restarted After 15 mm GLC analysis showed a 967r con
version 0 04 mole of /iTsOH was added and after another I 5 mm a 989?^ conversion was
reached The working up procedure was the same as described for H AW 500 31 3 g
(9\7r) of cyclohexanone diethyl acetal b p 75-78°/15 mm «u" I 4347 purity (GLC)
99 27c was obtdined

75
 CaA. 0.2 mole of cyclohexanone, 0.8 mole of ethanol, 300 ml of hexane, and 0.04 mole
of pTsOH were stirred vigorously at 0° while 40 g of CaA powder was added rapidly. In
order to quench the reaction a solution of sodium ethoxide, prepared from 2.2 g of sodium
and 40 g of ethanol, was added after I min to neutralize the catalyst. The working-up
procedure was the same as described for H-AW 500. 30.2 g (88%) of cyclohexanone
diethyl acetal, b.p. 75-78°/16 mm, «D'^ 1 4346. purity (GLC) 98.4'% was obtained.

Acknowledgements
We wish to thank P. T.Alderliesten.J. Gons. A. F. Heller, and R. Taal
for experimental assistance and Ir. A. C. W. C. Bat and coworkers of the
Mineralogical Department for the analyses of the molecular sieves. We
are much obliged to Professor P. M. de Wolffoiihe Physics Department
for the use of the X-ray equipment and to Dr. E. F. Bohmfalk for critic-
ally reading the manuscript. Ir. L. P. van Reeuwijk of the Agricultural
University, Wageningen, and Dr. J. C. Platteeuw of the "Koninklijke/
Shell-Laboratorium", Amsterdam, were so kind to send us samples of
pure erionite and chabazite. Generous gifts of molecular sieve samples
from the Norton Company and the Union Carbide Corporation are
acknowledged.

76
 CHAPTER V Part 2

ACETALS OF SECONDARY ALCOHOLS'^

by
D. P. Roelofsen and H. van Bekkum

Whereas the literature provides many examples on the synthesis

of ketone acetals of primary alcohols, relatively few exaiTiplesl-3 of
the preparation of ketone acetals of secondary alcohols are given.
Generally, indirect methods have been applied while yields are often
low and procedures rather laboriousl.
We wish to report the direct synthesis of these acetals (III) from
the corresponding carbonyl compounds (I) and secondary alcohols (II)
using molecular sieves4 to adsorb selectively the water formed:

\ H* \ /Of'
C=0 + 2 ROH =?^ C + H2O

I II III
Without molecular sieve the conversion of I into III was less than 3f>
(the estimated lower limit of detection) under the experimental con-
ditions. With molecular sieve the conversion was always better than
80^ resulting in good yields of III. Some examples are given in Table
1.
The cyclohexanone acetals in Table 1 took part in a second equili-
brium under the reaction conditions:

. o<!: - o OR + ROH

III IV II
Care had to be taken during work-up as well as during GLC analysis
to prevent a further decomposition of III into IV.

/ Accepted for publication in Synthesis. Permission to reprint the

manuscript has been granted by the copyright owner.

77
 Table 1

Synthesis of acetals (III) from carbonyl compounds (I) and alcohols (II)

I II Conversion^ III
Yieldb b.p. n25
°/mm "D

acetone isopropanol 82 62 76-79/65 1.3985

cyclohexanol 80 60 87-92/0.8 1.4670
cyclohexanone isopropanol 90C 82 62-65/2 1.4410
sec-butanol 80d 75 65-68/0.7 1.4482
isobutyraldehyde isopropanol 100 73 75-76/45 1.3980
sec-butanol 100 81 55-58/4 1.4109
cyclohexanol 100 88 90-93 '0.5 1.4655
benzaldehyde isopropanol 99 94 59-61/0.7 1.4700 1

^ Conversion of I into III as determined by GLC.

^ Yield after distillation; purity better than 98^ as determined by
GLC and NMR.
^ 5^ conversion of I into IV.
d 15^ conversion of I into IV.

Preparation of III by alcoholysis^ of dimethyl acetals, now using

molecular sieves to adsorb methanol formed, seemed worth trying;
however, in a tentative experiment 15^^ of the dimethyl acetal of
cyclohexanone was converted into the methyl isopropyl acetal and
only 75^ into the diisopropyl acetal under conditions similar to those
in the direct synthesis (see below).
Type 5A sieve was preferred to 3A and 4A because of its relative-
ly slow interaction with the catalyst^, while its pore diameter of 5 A
is still small enough to exclude all components of the reaction mix-
ture other than water. Cyclohexane was the preferred solvent because
of its size (6.1 A) and its apolarity.
When the present technique was applied to the synthesis of alde-
hyde acetals of secondary alcohols, essentially quantitative conver-
sions were obtained. Examples are included in Table 1.

Synthesis of acetals (III)

Ketone acetals. 0.10 mole of carbonyl compound (I), 0.40 mole of
secondary alcohol (II), 150 ml of cyclohexane and 3.8 g of /)-toluene-

78
 sulfonic acid monohydrate were stirred at 0°. 40 g of molecular
sieve powder type 5A, obtained from the Union Carbide Corporation,
was added while stirring. After 2 h 6 ml of triethylamine was added
to quench the reaction. The suspension was filtered and the sieve
was washed carefully with five 50 ml portions of pentane. The com-
bined organic solutions were washed with two 50 ml portions of an
aqueous 15^ sodium hydroxide solution and with two 50 ml portions
of water. After drying with potassium carbonate, volatile com-
ponents were removed using a flash evaporator and the residue was
distilled in vacuum. A few grains of potassium carbonate prevented
the formation of IV during distillation.
Aldehvde acetals. The procedure was the same as for ketone acetals
except that the reaction time and the amounts of /)-toluenesulfonic
acid monohydrate, molecular sieve and triethylamine had been
halved.

GLC analysis:
Preferably, we used an alkali-containing column (2 m copper
column filled with lOfcSE 30 and 2fo potassium hydroxide on Chro-
mosorb W-AW) to prevent the decomposition of III into IV.
Samples of the reaction mixture had been quenched with triethyl-
amine or di-H -propylamine, shaken with a 2N potassium hydroxide
solution, and dried with potassium carbonate before injection on the
column.

The authors wish to thank J. van der Waal for experimental assist-
ance.

References

1 W.L.Howard, N.B. Lorelle, J. Org. Chem. 25, 525(1960).

2 W.T Reichle, Dissertation, Ohio State University, 1958;
Diss. Abstr. 19, 3134 (1959).
3 C. L. Stevens, R. L. McLean, A.J. Weinheimer, J. Amer. Chem.
Soc. 80, 2276 (1958).
4 D.P. Roelofsen, E.R.J Wils, H. van Bekkum, Rec. Trav. Chim.
Pays-Bas 90, 1141 (1971). This thesis chapter V part 1.

79
 CHAPTER VI
PROTON-EXCHANGED TYPE A MOLECULAR SIEVES^''
D P ROElOf-SEN F R J WII^S' and H VAN BEKKUM

Abstract —The inler.ittion of molecular sieves type NaA (4Al and CaA (5A) with /' toluenesulphomc
acid l/>TsOH) in dioxane has been investigated in detail Proton exchanged products were character-
ized hv then crysiallmity and their adsorption capacity for water hexane and cyclohexane For
comparison proton exchanged samples prepared in .iqueous solution of pH - s () were included in
this study
With NaA in dioxane and /iTsOH a lapid proton mcoiporalion up lo 9S per cent occurs in the
presence of small .imounts of water or methanol The products have large .idsorption capacities
Quite unexpectedly the pores are widened compared to those of NaA and CaA even allowing for
the adsorption ot cyclohexane in some cases The stability .iftei the adsorption of w.iter is pool How
ever after adsorplion of hexane the sieves can be successfully legenerated
In the case ol CaA in dioxane with />TsOH and small amounts of water or methanol only residual
sodium ions are exchanged tor protons thereby limiting the exchange to 20 percent Therefore these
acid-treated CaA sieves have properties similar to those of untreated t aA
Both NaA and CaA decompose in dioxane with /iTsOH when the amount ol waler added exceeds
the adsorption capacity of the sieve Heie pan of the sieve crystals bieaks down completely
Proton exchange in aqueous solution at pH 2= S () occurs lapidly with NaA but slowly with CaA
The products have poor idsoiption properties except al less than ca 2S per cent ot proton
incorporation

I N T R O D U C I ION
IN THE course of our work[l] on the preparation of acetdls from ketones 01 alde-
hydes and alcohols in the piesence of/;-toluenesulphonic acid catalyst and type A
molecular sieves it appeared that ion exchange occurred of protons of the catalyst
and cations of the sieves The partial proton-exchanged sieves displayed good
adsorption properties and. at the same time, were shown by X-ray diffraction to
be crystalline In the literature it is generally assumed that proton exchange of
type A sieves leads to inferior pioduets[21 We theiefore undertook a detailed
investigation ol the interaction ol p-toltienesulphonic acid with type A sieves.
especially in an organic solvent and of the properties of these proton exchanged
sieves
Pioton exchange of type A sieves in aqueous solution has been studied for
NdA in dilute hydrochloric acidl"?], for NaA and CaA using substituted benzoic
*Piesenl address Chemical Laboratory National Defence Research O r g a n i / i l i o n T N O Rijswijk
The Netherlands

1 D P Roelofsen F R J Wils a n d H vanBekkum Reel Trav Chim 90 1141(1971 This thesis

chapter V
2 H E Robson G P Hamnei and W F Arey Ir Proc 2nd Inl Conf on Molecular Sieie
Zeolites Adian Chem Ser 102 417(1971)
3 F A Belinskaya S P Zhdanov E A Materova and M A Shubaeva Tear lonnogo Ohmena
Khromaloi;r Tr Vse9 Neuieh Tekh Konf I96S ijd Nauka Moscow (I96X) p 17 Chem
Abstr 71 S40l2q( 19691
// R e p r i n t e d from J o u r n a l of Inorgaiui and N u c l e a r C h e m i s t r y vol 34 page 1437
A p r i l 1972 by p e r m i s s i o n of the copyright owner

81
acids[4] and for NaA by electrodialvsis['>l Pioton exchanged type A sieves are
also termed when NaA is treated with hvdrogen chloride at 200°C[6] or when
ammonium-exchanged NaA is heated|7] X-ray diffraction showed that materials
with d pioton incorporation of 2s-sO pei cent dlreddy possessed a ieduced cnstal
linity No adsoiption measuiements weie made In a ehromatogiaphic study [8]
anhydrous silver-exchdnged NaA was reduced at 7^° in hydiogen to metallic
silver and proton-exchanged NdA Above 2") per cent proton incoipordtion a
poie widening was observed allowing foi the adsorption of heptane at 20°C
However isobutane was not adsoibed At 75 pei cent pioton incoiporation a
change in stiucture presumably a breakdown ot the lattice was noted
The ion exchange ot zeolites in organic media has leceived only limited atten-
tion In geneial the use ot an oiganic solvent instead of watei decreases the late
ol ion exchange while the selectivitv changes consideiablv Barter ei al studied
the cdtion exchdnge ol dndlcite[9] dnd type Y|101 in methanol and in ethanol
The Cdtion exchdnge of NdX hds been studied in diflfercnt dicohols [I 1| Dyei
ct al \\2 n ] described the exchdnge ot several zeolites with diffeient cdtions
in wdtei in methanol in ethanol and in mixed media Barrett cl al \\A \'>\
investigated the cation exchdnge of type A sieve in methdnol in ethdnol in glycol
dnd in mixtuies of these with udtei Rddak and Susic[l6| deseiibed the equilibria
of NaA with the alkali metal ions in mixtures ot wdtei and methanol Two
patents! 17 18] described the use ot oiganic solvents toi piepaiing ion-exchanged
forms of zeolites especially ot fdujdsite moidenite and chabd/ite pioton ex-
change IS mentioned hut no examples ate given

RESULfS AND DISCUSSION

The inteidction ol NaA and CaA with /^-toluenesulphonic acid in dioxane'
was investigated (Figs I dnd 2) A number ot proton-exchanged samples were
subsequently made by interaction ot /'-toluenesulphomc acid in dioxane with
Dioxane w is chosen as the oiganiL soKcnt bei. luse of the high solubililv ol /> toluenesulphonii-
acid ind Ihe \eiy low solubility I 1 mg/IOO ml) ol the sodium and calcium s ill ol ;> toluenesulphonic
acid in this solvent
4 N F Limolenko \ D HoiiseMLh ind I P Shiiinski\a DoU Ikucl \iink Be laiinsk SSK 10
S72I1966I
5 \ 1 Rislienenko V d II n null F NLimaik L ki Kliim /li 16 S l l l t r r i ) )
ft R \1 BaiiLiand \ Ci Kancllopoiilos / Che in Soe \ ""6^(19^0)
7 D VV Bieck W C, tveisole R \1 Mihon T B Reed md V I Thomas / •)», elum i<i(
78 S961il9s6)
S F Wolt ind H I 111 lie \UiUciiUu Suic p P ^ SOL ( hi.m Ind l o r d (|9(S)
9 R VI B a i r e i a n d D ( Sammon / (/ii HI Soi ft7S(|9s6)
III R M B l u e r J \ DaMes n u l l \ C Rees / uun^ ninl Clum 30 r,11ii9ft8)
li P ( Huang A Mizany ind I t Piultv / ;>/m Cluin 68 2S7si|964)
I"" A Dyci ind R B Cictlins I'loe Inl (ani on laii F \e luiiii.e p IS" | ondi n |4(,9 SOL ( liLm
Ind London(1970)
M -X D v e i a n d R B Cjeltins / inoi.. mid Chem 12 2401(1970)
14 R B Baiiell 1 A M . i i m s k \ a n d P P.i\eliLli I'IIH 2nd tin Cmil on Xlnlee iileii Sieie Zeoliles
Aehiin Chem Ser 101 414(1971)
15 R B B i i i e t t a n d J A Maiinsky J /;/i\v Chem 75 8 s ( I 9 7 1 )
Ift V \ l Radik ind M V SUSIL / iimu miel Chem 3.1 I927(19-'ii
17 S Tuckei i S /ir//<«/1 ^2X •'68 SeptLnibei | s | | 9 7 ( ) )
18 R \ Hildebrandt (7 i /><(;<•«/1492 (188 Januarx 27(19701

82
 Addition o( 25 mmole H;0 or MeOH, no

Fig I Influence ol water and methanol on ihe interaction of anhydrous/>! sOH (105
mmole) with NaA (I 0 g) and with CaA (lOg) in dioxane at 20°C.

100
y*^ IZ

e rpg*q of X
/ 75 „ Na + Ca in CoA y"
o A
Co in CaA / ni

E
Na in CaA /
h _•-*—"-^•"'^
/ l

10 15 20 25
Time, hr

Fig 2 The inteiaclion of/iTsOH H.O (lOs mmole) with CaA (10 g) in dioxane al 20 C.

NaA and CaA. For comparison several samples were prepared with a diluted
(pH s 5) aqueous solution of/?-toluenesulphonic acid. The samples were charac-
terized by elemental analysis, crystallinity. adsorption capacity for water, hexane
and cyclohexane. and thermal stability (Tables I and 2).

Interuc tion ofp-loliienesiilphonic acid in dioxane uiV/i type A sieves

Figure I shows the interaction ofNaA and of CaA with a solution of anhydrous
/'-toluenesulphonic acid (/)TsOH) in dioxane as influenced by systematic addi-
tions of water or of methanol (MeOH). Without these polar compounds proton
exchange is negligible. Cations in mgeq. present in the 10 g portions of the starting
sieves, are indicated on the vertical axis of Fig. 1.
\'iiA.CciA-p7 sOH-MeOH-dioxane. With NaA each of the first two additions
of 2.5 mmole of methanol causes a pioton incorporation of about 25 mgeq. A
small exchange occurs initially, possibly due to residual amounts of water in the
solution and in the sieve. The proton exchange is limited to about 95 per cent of
the exchange capacity of NaA. This is confirmed by the analysis (Table I) of
sample 14 which was prepared under similar conditions. It may be remarked that

83
 Table I Piopeilies ot pioton

Sample Type Predicted Found composition" Civ si ~

atom ratio belore
H/AI Na Ca Al Si O aLliv

Starting sieves
1 NaA 0 01 0 97 0 99 198 G
2 CaA 0 21 0 42 0 98 1 98 O

NaA / J T S O H H O dioxane
1 HNaA 0 27 0 2S 0 7S 0 97 1 94 O
4 HNaA 0S| 0 S4 0 46 0 97 1 94 M
5 HNaA 0 77 0 66 014 0 96 1 92 W
6 HNaA I OS 0 91 0 07 0 98 196 A

CaA-/iTsOH H,0-dioxane
7 HCaA 0 19 0 IS 0 01 0 41 1 0 97 1 94 G
8 HCaA 0 41 0 17 0 01 0 40 i 0 99 1 98 G
9 HCaA 0 70 0 17 0 01 0 40 I 0 99 198 G
10 HCaA I 20 0 18 0 04 0 19 1 1 OS 4 10 W

NaA-pl sOH-MeOH-dioxane
11 HNaA 0 28 0 28 0 72 0 98 1 96 G
12 HN.1A 0S| 0S2 0 48 0 98 196 G
13 HNaA 0 77 0 7S 0 2S 0 98 196 G
14 HNaA 0 94 0 94 0 06 0 99 1 98 W

CaA p T s O H M e O H - dioxane
15 HCaA 0 07 0 05 0 11 0 42 0 98

NaA-/)TsOH H 0 H , 0
16 HNaA 0 27 0 27 0 71 - 1 099 1 98 G
17 HNaA 0 S2 0 SI 0 49 — 1 0 9S 190 M
18 HNaA 0 78 0 79 021 - 1 0 99 198 W

CaA p FsOH H 0 H.O

19 HCaA 0 20 0 20 0 10 0 IS 1 0 98 1 96 G
20 HCaA 0 4S 0 46 0 06 0 24 1 0 99 1 98 G
21 HCaA 0S8 064 0 02 0 17 1 091 1 82 M

Calculated Irom the decrease in acid concentiation ot the solution containing />TsOH and the
composition ot the starting sieve
tPhysicallv adsoibed watei omitted H/Al calculated assuming a balancing ol \1() gioups bv
H* Na a n d C a * O/Al Lakulated assuming O = 2(A1-Si)
t G - g o o d M - m e d i u m VV —weak A —absent Samples equilibrated above a saturated aqueous
solution of ammonium chloride

84
exchanged type A sieves

Cryst^ Ads water (g/lOOg) Ads hexane'ijlg/100g) Ads cyclohexane §(g/100g)

.after alter activation at after activation at after activation at
•icliv 200 C 100°C 400°C 200 C 100°C 400°C 200°C 300°C 40O°C

10 26 I 27 2 27 2 16 17 18 09
10 23 9 24 1 24 8 14 1 14 1 14 1 1-4

10 27 1 31 I 2 1 2
06 27 2 27 0 26 2 120 11 9 11 8 1 0
0 IS 16 9 10 8 48 27
01 04 0 I

07 24 9 13 4
07 24 9 25 0 27 4 139 13 I 150
06 24 9 2S9 26 I 13 9 18 27 32
OSS 23 6 24 7 26 1 136 2 5 28 38

08 27 6 12 7 12 4 55 09
OSS 29 7 29 4 27 3 13 S I 0 1 7 90
0 IS 26 8 25 2 24 8 140 14 6 54 0 8 10 I n 1
0 1 26 0 17 1 37 14 1 06 05 06 02

0 65 210 13 1 2-0

06 21 4 22 3 22 2 II 06 10
0 1 15 02
,0 1 12 0 1

0 55 25 6 26 2 28 4 14 3 18 1-5
0 15 14 6 119 60 16
<-0 1 49 09

§Fxpressed as a fraction of the crystallinity of the starting sieve (NaA or CaA) Samples had been
activated at 200 C and saturated with water vapor at 45°C
Adsorption conditions 20 hr 45°C plp^ = 0 5 Adsorption expressed in g of soibale per 100 g of
activated sieve
fiBecausc an is admitted at the end of each adsorption expeiiment an adsoiption of nitrogen of
0 5 I 5 g/100 g sieve is observed when the sorbate is excluded from the inside adsorption area of the
sieve

85
 Table 2 Regeneration of proton-exchanged sieves

Adsorption+(g/100g.ict sieve)
Experimental conditions* HNaA HNaA HCaA
(sample4) (sample H) (sample9)

Activation 400°C: adsorption of hexane 11-8 1S4

Regeneration 400°C adsorption ot hexane II 9 IS s
Regeneration 400 C .idsoiption ot hexane 11 7 15 2

Activation 20()°C adsorption of water 27-4 27 0 24 8

Regeneration 200 C .idsoiption of water 16 6 1 9 24 S
Regeneration 20(rC adsorpuon ol water 12 1 1 s 24 0

Activ.uion 400 C adsorption ot water 26-3 24 2 25 8

Regeneration 400 C adsoi ption ol waler 5-7 08 21 S
Regeneration 400"C . adsorplion of water 4-4 21 8

Activation 400°C. adsorption of water 26 5 24 4

Desoiption- 20^C. residual water 12-4 14 8
Regeneration 400 C adsorption of w aler 5-3 0s

'Activation and regener.ition .it the specihed lempei.ituielor 2 days .it 10 ' mm Hg
t Adsorption at 45 C lor 20 hr and pip. = 0 s
tDesorption during 5 days at 10 ^ m m H g

both NaA and CaA are saturated with methanol after the third addition. With
CaA exchange is very difficult and the maximum degree ot exchange is 5-10 per
cent. Of the Ca-'* and Na* present in CaA only part of the Na* is exchangeable
as shown by the analysis of sample 15 The agreement between H/Al ratios as
found by analysis and as predicted from the acid uptake during preparation and
also the constancy of the Si/Al ratio (Table I. samples I 1-15) indicate that no
partial dissolution of the NaA or CaA sieve occurs.
Three alternatives were considered for the mechanistic role of methanol in
facilitating proton exchange- (a) MeOH and H* enter the <'/)(/)n sieve as M e O H j -
species lb) MeOH is adsorbed in the sieve and incieases the mobility of Na and
H* in the sieve by allowing Na* and H to jump between adjacent MeOH-
molecules. (c) Adsorbed MeOH serves as a carrier of Na* and H ions, fo
decide among these possibilities we compared the rates of proton exchange be-
tween /)TsOH (105 mmole) and NaA (lOg) in dioxane (200ml) in two experi-
ments (i) when methanol (35 mmole) was adsorbed betoie the addition o f / J T S O H .
and (ii) when methanol (35 mmole) was adsoibed aftei the addition ot /'TsOH.
No difference was observed and this rules out explanation (a), fhe possibilities
(b) and (c) are two extremes and the actual mechanism will presumably lie some-
where in between. However, explanation (c) is suggested as the more likely
alternative. We propose that methanol has a reasonable mobility in the sieve and
exchanges quite rapidly with the small (< 2 per cent of total methanol in the
expenment described above according to GLC) equilibrium amount of methanol
in solution Because methanol in solution will be partially protonated. H* and
Na* are exchanged by a desorption (MeOH with solvated Na )-adsorption

86
( M e O H / ) mechanism. The mobility of methanol in the sieve will diminish as
the H*/MeOH ratio approaches unity due to the strong localization of MeOH.,+
by the anionic framework of the sieve thus inhibiting further proton exchange.
This kinetic explanation of the final H'/MeOH ratio is supported by the large
temperatuie effect, where it is found that a ratio of 2 is reached at 75°C.
The inability to exchange Ca-' in CaA may be ascribed to a low mobility of
divalent cations in methanol-filled molecular sieve[12. 19]. Also the thermody-
namic selectivity of the sieve for Ca-* overH* might be quite high in this medium.
NaA,CaA-pT.sOH-H.,0-dio.xanc. After the first two additions of water to
NaA and / J T S O H in dioxane a fast proton incorporation occurs until the ratio of
H*/H20 in the sieve is about unity; a similar behaviour was observed with KA.
Here water appears to play a similar mechanistic role as methanol. The limited
extent of proton exchange with CaA after the second and third addition of
water corresponds to the 209? Na' in CaA sieve as shown by the analysis of
samples 7-10 (Table I). According to the adsorption data in Table I. NaA
(sample 1) and CaA (sample 2) are saturated with water after the sixth and fifth
addition of water, respectively. A breakdown of part of the sieve crystals, as
indicated in Fig. I by the sudden rise of acid removed from solution, coincides
with the saturation of the sieves.
This breakdown of the sieve has been studied in more detail in the case of CaA
(Fig. 2). The starting formulation is similar to that in Fig. I after the fifth addition
of water.* Four regions may be distinguished in Fig. 2. In region 1 exchange of
most of the Na* in the sieve and H ' in the solution occurs leading to a precipita-
tion of sodium /'-toluenesulphonate. Fig. 2 suggests that no major changes occur
in region II. However, a sample of the suspension in region 11 taken after 3 hr
and freed from solid material by centrifuging, does form a white precipitate after
standing overnight. A heavier precipitate is formed from a sample of the suspen-
sion taken later in region 11. e.g. after 7 hr. The neutralization curves of these
samples show the buffer action at pH 4-5 of an aluminium salt, while a gel-like
precipitate is also formed. We conclude that in region II the breakdown of the
sieve crystals has already started, but the precipitation of the products is retarded
until region III. In region 111 the suspension becomes very viscous due to a
heavy precipitate, presumably /'-toluenesulphonic acid salts of calcium, silicon
and aluminium. In region IV the acid in the solution has been consumed. After
removing the solids by filtration and washing with DMF to dissolve salts of
/i-toluenesulphonic acid, a crystalline sieve (6-5 g on a calcined basis) results.
Properties and composition are similar to those of sample 10 in Table 1. In the
dioxane solution appreciable amounts of dissolved silica and alumina are found.
The DMF extract does contain large amounts of sodium and calcium/'-toluene-
sulphonate (analysed by flame emission spectrometry) and of aluminium and
silicon (a sample was neutralized in water to pH 5 and a precipitate was formed
with Al:Si = I ;0'85). Summarizing a rapid exchange of Na' and H* takes place
"When inlerpreling Figs. 1 and 2 it must be realized that the titration tsf a sample in aqueous solu-
tion up to pH 7 means that not only protons but also any dissolved salts of a weak base (e.g. an
aluminium salt) and /'-toluenesulphonic acid arc determined.
19. D. P. Roelofsen. .1. W. .\I. dc tiiaaf. J. A. Hagendoorn, H. M. Veischoor and H. van Bekkum.
«(-(•/. Trot. C/i/m. 89. 193(1970). This thesis chapter III.

87
together with a much slower breakdown of the sieve. During this breakdown
about one third (in our formulation) of the sieve crystals are broken down com-
pletely with only a slight preference (see the analysis of samples 7-10 in Table 1)
for the removal of aluminium over silicon.
A breakdown similar to that of CaA shown in Fig. 2 may occur with NaA or
KA. After one hour virtually all N a ' or K* has been replaced by H . Region II
and 11 i also occur but more water is necessary than with CaA (see also Fig. I) to
exceed the adsorption capacity. The analyses (Table I) of samples 5 and 6,
especially a comparison of the predicted and the found H/Al ratios, prove that a
complete breakdown of part of the NaA crystals has occurred.

Preparation and properties ofproton-excliani;ed type A sieves

Preparative methods. Three different methods of proton exchange have been
applied in Table I. Details will be found under the experimental section.
Samples 3-6 were prepared by shaking a suspension of NaA in dioxane with
/;1 sOH.H.O. In the case of sample 6 a small excess of/'TsOH.H^O was used and
some water was added to obtain the high degree of exchange. Samples 7-10 were
prepared similarly from CaA with increasing amounts of /'TsOH.HjO while
some water was added to saturate the sieve.
Samples 11-15 were prepared by suspending NaA or CaA in a solution of
anhydrous / J T S O H in dioxane while sufficient methanol was added to saturate
the sieve. With samples 14 and 15 an excess of acid was used.
Samples 16-21 were prepared by suspending NaA or CaA in water and adding
a diluted aqueous solution o f / J T S O H . H ^ O . The suspension was carefully kept at
pH s 5 0 during the addition, since a 5 « pH =s 12 is believed to leave the sieve
intact.
Crystallinity of siimples. In some cases the intensity and width of the X-ray
dift'iaction lines had changed considerably after the protonation of NaA or CaA
and after the subsequent activation. However the diffraction patterns could always
be ascribed to a type A crystal structure. The proton-exchanged CaA samples
after hydration gave two sets of diffraction lines in most cases, proving the
existence of two zeolitic phases with a difference of about I percent in their cell
constants,
Adsorption capacities. Adsorption capacities have been determined for water.
hexane and cyclohexane after different activation procedures (Table I),
The adsorption data in Table I may be used to estimate the lattice breakdown,
Let the adsorption of a sorbate 5 by NaA be 100 arbitrary units (wt, of adsorbed
5 per wt, of activated sieve). The adsorption of S by a proton-exchanged NaA
sieve will deviate from 100, However, the difference cannot be attributed exclu-
sively to the lattice breakdown: (a) A linear contraction of 2 per cent (hydrated
sieves) as a maximum has been observed by X-ray diffraction when part of the
Na* is exchanged for H ' , This may result in a reduction of the adsorption to 94
units, (b) The replacement of Na* by H* increases the available adsorption volume
per unit cell while the weight of a unit cell decreases. During dehydroxylation
of the sieve, in which process protons liberate oxygen from the framework form-
ing water, a similar increase in adsorption capacity is expected. Thus, in passing
from NaA (idealized formula NaAlSiOj via a 1009? proton-exchanged sieve

88
(HAlSiOj) to a completely dehydroxylated product (AlSiOjJ the adsorption
value should increase from 100 via 123 to 138 units "• (c) The density ot the
sorbate in the sieve will be influenced by the protonation and dehydroxylation ot
the sieve especially with a polar sorbate such as water Such an effect will be less
pronounced with the apolar sorbates hexane and cyclohexane (d)The adsoiption
data in Table I especially for hexane and cyclohexane may not always represent
equilibrium values because the critical molecular diameters ot the sorbates are
close to the pore diameters Thus the decrease in the adsoi ption of hexane by the
samples 3 and 11 when the activation temperature is increased from 200 to 300
and to 400°C will presumably be due to a slight contraction of the pores and
certainly not to a decrease in the internal tiee volume The contraction of the
lattice IS probably caused by dehydroxylation of the sieve
Stahibtx of proton-exchcmi>ed samples after adsoiption The influence of
adsorbed hexane and water on the stability of three samples was investigated
(Table 2)
HCaA (sample 9) with its low proton content is quite insensitive to adsorbed
water The instability of watei-saturated HNaA samples 4 and 13 even after a
preliminary removal of part of the water at 20°C, was also clearly shown by X-ray
diffraction using a high-temperature Guinier-Lenne camera When slowly heated
in vacuum the diff"raction lines vanished at about 50°C in contrast HCaA samples
9 and 10 after water adsorption decomposed aiound 850°C
HNaA sample 17 prepared in water was subjected to a vacuum of 10"'
mm Hg during 5 days at 20°C The water content decreased from 37 8 to 17 9 g
per too g of activated sieve After subsequent activation at 400°C this sample
adsoibed very little watei (I 5 g / | 0 0 g sieve) Thus even a caietul activation ot
samples piepared in watei leads to amoiphous pioduets
HNaA samples 4 and 13 and piesumably H t a A sample 9 aie stable towards
apolar sorbates such as hexane
Pioton-excham^ed CuA When working in dioxane the maximum degree of
proton incorporation is limited by the Na content The removal ot Ca * is virt
ually impossible The HCaA samples showed good thermal stability and could
be regenerated after adsorption ot water These facts accord with the low degree
of pioton incorporation The adsorption pioperties are similar to those ot CaA
The high crystallinity of the samples is in accord with the adsorption data
The high stability of CaA towards /'TsOH in an organic medium allows the
regenerative use of CaA used in the prepai ation ot acetals [ I ]
Several attempts were made to reach highei degrees of proton incorporation
in CaA By replacing dioxane by DMF in the formulation of sample 10 a H/Al
ratio of 0 13 was i cached as with dioxane part of the sieve ciystals had been
broken down Methanol was replaced by acetonitrile and nitromethane in the
formulation of sample 15 Although both compounds are readily adsorbed no
proton incorpoiation was observed
In aqueous solution at pH 5 0 a more substantial pioton incorpoiation is pos
sible (see Table I) although long reaction times are required However the prod-
ucts are thermally unstable except at a low degree of proton exchange (sample 19)
'The combined volume of Ihe large cavity and the smallei sod ilite c ivity has been used However
the sodalite cavity is only accessible lot w,ilerand not foi hexane orcyclohexane

89
Node alumination was found instead a slight de silication was observed in sample
21 It must be noted however that at pH 5 0 the solubility of alumina in water is
very low indeed |20] The solubility of silica is somewhat higher|20] which could
account tor dissolution of a small percentage of the silicon during the preparation
of sample 21
Ermolenkocr^/141 deseiibed Ihe inteiaction of Na'K dnd Ld\ with substituted
benzoic acids in aqueous solution The starting solution was at pH 3-4 In accord
with the present work they found that the replacement of Nd by H is preferred
over the repldcement of Cd The ddsorption properties of the proton exchdnged
mdteridls were not determined
Pioton exclianued NaA Proton exchdnge in dqueous solution vvith NaA at
pH s 5 0 dppears to be quite rapid The adsorption properties of the products
are infenor to those of samples prepared in dioxane however Thus a comparison
of the water adsorption by samples I 3 I 1 dnd 16 aftei activation at 200°C proves
that a partial breakdown in aqueous solution has already occurred aftei 27 per
cent proton incorporation At 51 pei cent exchange (sample 17) both the adsorp
tion of water and the X ray crystallinity aftei activation at 200°C are very low
Breck ct cd [7| claimed that NaA is unstable in an aqueous acid solution but
no details were given Russian woikers|31 have used a titration method in an
extensive study ot the proton exchange in water of NaA NaX and (Na K)
erionite The titration curve showed that proton exchange with NaA took place
at ca pH 7 The irreversibility ot the titration curve and the decreased intensities
of the X ray diffraction pattern of a sample with about 60 per cent proton incor
poration proved in dgreement with oui lesults that a breakdown of the sieve
structure had occuired The adsorption pioperties ot the proton exchanged type
A sieve were not investigated
Fhe present work shows that proton exchange ot NaA in dioxane with /'TsOH
IS possible in the presence of an equivalent amount of water or methanol The
thermal stability ot these pioduets especially when using methanol is very satis-
factory according to the adsorption data ( fable I) It can be concluded that by
avoiding the use ot excessive water proton exchange ot type A sieve leads to
relatively stable pioduets A pore widening possibly by removal ot Na ions
allows the adsorption of hexane after activation at 200°C at a proton exchange of
ca 30 pel cent A similar phenomena had been obseived in AgA reduced with
hydrogen|81 X lay dififiaction shows that a shiinkage ot the sieve structure
(dehydioxyldtion) occuis at higher dctivdtion tempeiatuies This leads to a lower
rate of hexane adsorption in the case ot samples 3 dnd 11 in Tdble I The same
dehydroxylation leads to an effective widening of the pores in samples with high
proton contents such as samples 12 and 13 even allowing for the adsorption of
cyclohexane (critical moleculai diametei 6 \ ) The adsoiption ot cyclohexane
by zeolites with pores formed bv eight oxygen atoms has never been observed
To check the molecular geometry of the sieve structure we used 1 3 5 triethvl
benzene (critical molecular diameter 9 A) as a sorbate After 20 hr at 4s C and
Fhe presence of in idsoibcd pol ir compound is requucd Icnlitive expeiiments showed ihii
elh inol iLelonitiile ind nitiometh inc ilso pcrtorm this lunLtion
''() W F 1 inkc Sohihilities ell In neinie and XteUil Oruanic Cempeiinds 4th ed \m Lhem SOL
I 194(19S8) 2 14s2(l96s)

90
/'//', = 0-5 the adsorption of this compound was found to be only 1-0, 0-6, 0-8 and
0-7 g/100 g of activated sieve for the samples 1.4. 12 and 13 respectively after
activation at 400°C. These data clearly prove that the sieve character has been
preserved. For comparison, NaX molecular sieve adsorbed 21-9 g/100 g NaX
under identical conditions. The unexpected widening of the pores as shown by
the adsorpdon of cyclohexane, must be ascribed to a change from the circular to
a less regular shape of the pores, or to the removal of some of the oxygen atoms
which form the eight-membered pores, or to a combination of these effects. Com-
plete dehydroxylation of a fully proton-exchanged type A sieve will remove one
oxygen atom from every eight-membered ring of oxygen atoms assuming a random
attack. To estimate the degree of dehydroxylation of samples 4 and 13. the loss in
weight was determined when a sample, previously activated in vacuum at 400°C,
was calcined at IOO()°C. About 40 and 70 per cent of the protons were removed
at 400°C from samples 4 and 13 respectively. This means that on average about
one oxygen atom out of every two eight-membered rings is removed after activat-
ing sample 13 at 400°C.
The HNaA samples cannot be regenerated after the adsorption of water.
Reactivation after the adsorption of an apolar compound such as hexane is pos-
sible however, without a loss in adsorpUon capacity (Table 2)
The low crystallinity of the HNaA samples after activation and water adsorp-
tion (Table I) is somewhat surprising in view of the high adsorption capacities.
Possibly the crystallinity decreases when water is adsorbed. Therefore measure-
ments were made of the crystallinity of sample 13 after activation at 400°C and
adsorption of cyclohexane. A relative crystallinity of 0-3'' was found compared
to a value < 0 I after activation at 400°C and adsorption of water. Another reason
for the apparently low crystalhnities may be a change in the intensity of the refer-
ence reflection of the sieve due to proton exchange, and independent of any loss
of crystallinity. In spite of the low crystallinities the molecular sieve character
of the proton-exchanged sieves is maintained as shown by the high adsorption
of small-sized molecules such as water, hexane and, in a few cases, cyclohexane,
and the low adsorption of the large 1.3,5-triethylbenzene,

E X P F R I M E N I AL
Maleruils
/'-Toluenesulphonic acid monohydrate (relerred to as /'TsOH,HjO in the present article) was
obtained from J T Baker Chemicals N V (BAR-quality) Karl-Fischer water determination and
alkalimetric titration showed the real composition to b e / ' F s O H i 1 - 1 2 H ; 0 A solution of anhydrous
pTsOH (leleiied lo .is /'TsOH) in dioxane was obtained by refluxing 50 g ol /'FsOH H ; 0 and ISO ml
of dioxane for 48 hr using an extraction apparatus [19| filled with 50 g of molecular sieve pellets type
3 A The water content was now found to be < 0 1 mg/ml
Molecular sieve powders type KA (3A) NaA (4A) and C.iA (5AI were obtained from the Union
Carbide Corporation The sieves were calcined in an oven al 400"C and cooled under anhydrous
conditions
Methanol hexane cyclohexane I 1. 5-tnethylhenzene ben/ene, dioxane DMF acetonitrile and
nitromethane were dried with pelleti/ed moleculai sieves type 1A or 4A

hileracUon e>jpTsOH m dioxane with l\pe' A sictcs

In the experiments shown in Fig 1 10 g of activiited sieve (NaA or CaA) in 200 ml ot dioxane

*This sample showed a linear contraction ot about 4 per cent compared to hydrated NaA

91
was stirred at 2()°C with 105 mmole of anhy drous /'TsOH Every 24 hr 25 mmole of water or methanol
was added In the experiment shown in Fig 2 10 g of activaled CaA in 200 ml ot dioxane was stirred
at 20% with 1 OS mmole ol/'TsOH H O The acid removed fiom Ihe solution leterred to in Figs 1 and
2 was determined by titration ol 0 5 ml samples heed fiom solid matenal by centrifuging in 50 ml
ol water wilh 0 05 N NaOH to pH 7 0

Pnpcinilion ot prolan exchanged l\pe A sieies

The samples in Table I were prepared bv shaking NaA or CaA wilh the solution containing p-
toluenesulphoniL acid tor 1-2 davs except the samples 14 (4 days) 15 (7 days) 20 (8 days) and 21
(17 day s) The lesidual concentration of acid in the solution was determined bv titiation
Samples 1-10 were prepared by suspending SOgof .ictivatedsievc (NaAorCa,'\) in 1 I of dioxane
and adding /'TsOH H.O and water as follows (sample no g of /'TsOH H O g of w.iter resp I
s 1 17, 0 i 4. 13 s.O s 5,51,0 i 6.7S 1 9 i 7 17 5.8 1 i 8,15, 6 6 i 9. 52 5, 5 0 i 10, 100 10
Samples I I-l 5 were piepared by suspending 50 g ot activaled NaA oi CaA in 1 1 of dioxane contain
inu anhvdious/iTsOH and methanol as follows (sample no mgeq o f / ' F s O H g ot methanol resp)
s 11 9 0 6 II 1 12. 176. 11 1 II 264, II s 14 378, 12 s 15, 189. 12 Samples 16-21 were pie
paied by suspending hydiated NaA oi CaA (50 g on an anhydrous basis) m 2 1 of watei and slowly
adding .i solution ol ihe lequiied amounl ot /'TsOH in 1 1 of walei Ihe suspension was kepi above
pH SO
After Ihe a n d treatment the suspensions were filtered and the sieves washed either with DMF
(samples 1-1 S) to dissolve salts ot /'-loluenesulphL'nic .icid (solubility al 20''C of Ihe sodium .md
calcium salt about 8 g/lOO ml ol DMF) or wilh water (samples 16-21) SubsequenlK samples 1-15
were washed cither with benzene (samples 1-10) oi with methanol (samples 11-15) to remove D M F
from the sieve crystals I he samples were finallv stored overnight in a vacuum desiccaloi above activ-
,ited I OX pellets (samples 1 I (land 16-21 lor above methanol (samples 1 1-IS)

C heireie le i izalion oj su>ie\

Samples after equilibration above a saturated aqueous solution ol ammonium chloride were
analvsed by X ray fluorescence for Ca Al Si and by alomic absorplion tor Na
X-ray powder diflfiaction patterns using a Gumier dc Wolff camera with C u K , ladiation weie
used to estimate the crysl.illinily ot the samples The crystallinity ol the samples before .ictivation
(hut aller hydration at room tempeiature and R H 7" per cent) was estimated by using three proton
exchanged samples and the sl.uling sieve (NaA oi C aA) in the sample holder and visualK comparing
thp intensities ol the X rav diflTi action patterns Proton exchanged samples after activation lor two
davs lit 200°C and 10 ' mm Hg followed by water adsorption at 45°C for 20 hr at pip. = 0 5 were
mixed V ith 17 per cenl by wt ot calcite as an internal standard By companng the intensity of the 121
reflection of the sieve and the 104 reflection ol calcite using a photodensilometer and a polar plani-
metei the crvslallinily of Ihe proton exchanged sieves could be expressed |6] as a fraetion of the
civstallinity ot the starting sieves (NaA or CaA) With most ot the acid treated C a ^ samples two sets
ot diffiaLlion lines were observed Here the conliibutions from both phases to the civstallinity were
added The thermal stability ot a few samples was checked using a high temper.iture Giiinier-Lenne
camera with C iiK, radiation al a vacuum ol 10 ' m m H g
The vapoi adsorption capacities tor water hexane cyclohexane and 1 1 5 tnelhylbenzene at45°C
tor 20 hi dl pip — 0 5 weic delermiiied using Ihe technique deseiibed earlier|21] The samples (0 5-
1 s g) in glass bulbs were connecled to a vai uum line and held in a small oven at room temper.iture
The oven leached the desired lemperature of 200 300 or 400°C m about 4 hr and was kept at this
lemperature tor two days The final vacuum was better than 10 ' m m H g

-l(A'(im/( i/i,(Hi( i / v - W e wish to thank Ir A C W C Bot and hei co workers of the Mineralogical
Department for the analyses also Professor P M de Wolflf of the Physics Department and Protessor
L L van Reiien ot the Chcmisliv Department for the use ot the X-ray equipment We aie obliged to
Dr D I lones forcriticalK re.idinglhe manuscnpt
21 D P Roelotsen 4««/\( Chem 43 611 (1971) T h i s t h e s i s c h a p t e r II

92
 SAMENVATTING

De zeolitische moleculaire zeven vormen een groep van poreuze,

kristallijne adsorbentia. Zij zijn opgebouwd uit silicium. aluminium,
zuurstof en een metaal als natrium, kalium of calcium. In tegenstel-
ling tot amorfe adsorbentia, zoals actieve kool, silica en alumina,
bezitten de moleculaire zeven porien die van moleculaire afmetingen
zijn en die bovendien uniforin zijn. De diameter van de porien v a r i -
eert met het type moleculaire zeef van 2 tot 10 A. ,.i<a
Moleculen die zo klein zijn dat ze de porien kunnen binnendringen -^
worden geadsorbeerd op het grote inwendige oppervlak (in de orde
van 700 m^/g) van de microkristallieten.
In dit proefschrift worden moleculaire zeven gebruikt om chemi-
sche evenwichten (1) te verschuiven door een reactieproduct (D) s e -
lectief te adsorberen:

' K
B - ^ C + D (1)

D ^ u ^ (2)
geadsorbeer(d

Het type moleculaire zeef dient zo gekozen te worden dat A, B en C

en het oplosmiddel de porien niet kunnen binnendringen.
Twee technieken werden toegepast. Bij de extractietechniek wordt
de damp van het kokende reactiemengsel gecondenseerd en het con-
densaat geleid over moleculaire zeef. D wordt door de moleculaire
zeef geadsorbeerd en de rest van het condensaat vloeit terug in het
reactiemengsel. Bij de suspensietechniek wordt het reactiemengsel
met moleculaire zeef geroerd waarbij de temperatuur nog vrij geko-
zen kan worden. Bij alle onderzochte evenwichten werd de reactie
van A en B versneld door homogene of heterogene katalyse. De in-
teractie van de homogene katalysatoren met de moleculaire zeven
in de suspensietechniek werd onderzocht. Onder meer werd de ionen-
wisseling tussen de protonen van een zuur en de kationen van de type
A zeven in organisch milieu bestudeerd. De gevormde proton-uitge-
wisselde moleculaire zeven blijken een redelijke stabiliteit te bezit-
ten.
Het grootste gedeelte van het in dit proefschrift beschreven werk
is reeds gepubliceerd of wordt binnenkort gepubliceerd. De hoofd-
stukken II-VI zijn overdrukken van publikaties.
In hoofdstuk I wordt een algemene inleiding gegeven. De structuur
van de gebruikte moleculaire zeven wordt beschreven en de adsorptie-

93
eigenschappen worden toegelicht aan de hand van enkele scheidingen
van organische verbindingen. Enkele facetten van het gebruik van
moleculaire zeven voor het verschuiven van chemische evenwichten
(1) worden besproken.
In hoofdstuk II wordt een eenvoudig adsorptie-apparaat beschreven.
Hiermee is een snelle bepaling van de verzadigingscapaciteiten van
moleculaire zeven voor organische dampen mogelijk.
In hoofdstuk III wordt aangetoond dat door alkoxide gekatalyseerde
omesteringsreacties aflopend kunnen worden gemaakt door met mole-
culaire zeef van het type A het verdrongen alkohol te verwijderen.
Een groot aantal voorbeelden wordt gegeven waarbij zowel de uit-
gangsester, het verdringende alkohol, het oplosmiddel en de mole-
culaire zeef worden gevarieerd. Ook enige niet eerder beschreven
^-butylesters konden op deze wijze worden bereid.
De kinetiek en de evenwichtsligging van enige omesteringsreacties
werden bestudeerd.
De interactie van een natrium- of kaliumalkoxide katalysator met
de zeef werd onderzocht. Interessante verschillen tussen de typen
KA, NaA en CaA moleculaire zeef werden aangetoond.
De invloed van sporen water en triethylamine op de storende ont-
leding van ^-butylalkohol door de moleculaire zeven werd bestudeerd.
In hoofdstuk IV wordt de synthese van enaminen en ketiminenbij
kamertemperatuur uit de overeenkomstige carbonylverbindingen en
aminen besproken. Een commerciele silica-alumina-kraakkatalysa-
tor werd gebruikt om de reactie te versnellen terwijl de gesuspen-
deerde moleculaire zeef van het type A het gevornide water adsor-
beerde. Het mengsel van zeef en silica-alumina kan met succes wor-
den geregenereerd. Verschillende oplosmiddelen, katalysatoren en
moleculaire zeven werden geprobeerd. De combinatie van cyclo-
hexaan, silica-alumina en moleculaire zeef CaA geeft de beste r e -
sultaten; behalve het zuivere CaA poeder, kan ook geextrudeerd CaA
gebruikt worden in combinatie met silica-alumina-katalysator. De
katalytische activiteit van geextrudeerd CaA zelf, zoals vermeld in
de literatuur, bleek veroorzaakt te worden door het bindmiddel dat
gebruikt wordt in geextrudeerd niateriaal.
In hoofdsttik V, deel 1, wordt de synthese van verscheidene ace-
talen uit de overeenkomstige aldehyden of ketonen en alkoholen onder-
zocht. Gevormd water wordt geadsorbeerd door moleculaire zeef.
Twee modificaties van de suspensietechniek werden gebruikt, zowel
bij 0 °C als bij kamertemperatuur.
In enige experimenten werden proton-uitgewisselde moleculaire
zeven type Zeolon 500, AW 500 of mordeniet gebruikt. Het uitwendi-
ge oppervlak katalyseert de acetaalvorming terwijl het inwendige
oppervlak het gevormde water adsorb eert.
In de overige experimenten werd /)-tolueensulfonzuur (pTsOH) als
katalysator toegevoegd. In dit geval treedt protonuitwisseling op

94
tussen de katalysator en de zeef. Deze uitwisseling is snel bij KA en
NaA, maar veel minder snel bij CaA, Zeolon 500 en AW 500. Toch
kunnen deze zeven alle gebruikt worden omdat de acetaliseringsreac-
tie zeer snel verloopt. Met KA en CaA worden kwantitatieve omzet-
tingenbereikt binnen een paar minuten.
In hoofdstuk V, deel 2, wordt de analoge bereiding van acetalen
van secundaire alkoholen en ketonen of aldehyden beschreven. De
sterisch gehinderde ketonacetalen zijn moeilijk op conventionele wij-
ze te bereiden.
In hoofdsttik VI wordt de interactie van NaA en CaA met/)TsOH in
dioxaan onderzocht. Proton-uitgewisselde producten werden gekarak-
teriseerd door hun kristalliniteit en door bun adsorptiecapaciteit
voor water, hexaan en cyclohexaan. Ter vergelijking werden ook de
eigenschappen onderzocht van enige proton-uitgewisselde zeven, die
bereid waren in waterig inilieu bij p H > 5.0.
Met NaA en watervrij /?TsOH in dioxaan, in de aanwezigheid van
kleine hoeveelheden water of methanol, treedt een snelle protonuit-
wisseling tot 95^5 op. De gevormde producten hebben een grote ad-
sorptiecapaciteit. De porien zijn verwijd vergeleken met die van
NaA en CaA waardoor zelfs de adsorptie van cyclohexaan in sommi-
ge gevallen mogelijk is. De thermische stabiliteit na de adsorptie
van water isgering. Daarentegen kunnen de zeven met succes gere-
genereerd worden na de adsorptie van hexaan.
Met CaA en watervrij /)TsOH in dioxaan, in de aanwezigheid van
kleine hoeveelheden water of methanol, worden slechts de nog aan-
wezige natriumionen vervangen door protonen zodat de uitwisseling
beperkt blijft tot ongeveer 20-^ Hierdoor zijn de eigenschappen van
deze, met zuur behandelde, CaA zeven practisch gelijk aan die van
onbehandeld CaA.
Zowel NaA als CaA worden afgebroken door pTsOR in dioxaan als
de toegevoegde hoeveelheid water de adsorptiecapaciteit van de zeef
overschrijdt. Hierbij ontleedt een gedeelte van de moleculaire zeef
kristallieten volledig.
Protonuitwisseling in waterig milieu bij p H > 5.0 verloopt snel
met NaA, echter langzaam met CaA. De kristalstructuur van de
moleculaire zeven gaat evenwel verloren als deze protonuitwisseling
meer dan 25^ bedraagt.

95
 STELLINGEN

Door onvoldoende experimentele gegevens is een belangrijk gedeelte

van de resultatenvan Zalkow c.s.betreffende devorming van acetalen
van steroidketonen moeilijk interpreteerbaar.
L. H. Zalkow, R. Hale, K. French en P. Crabbe, Tetrahedron 26, 4947
(1970).

n
Hedge veronderstelt dat bij de door hem gevonden scheiding van 2,6-
en 2,7-dimethylnaftaleen met behulp van moleculaire zeef type X, Y
en L adsorptie-evenwichtbereikt is; er zijn evenwel aanwijzingen dat
kinetische factoren een rol spelen.
J. A. Hedge, Advan. Chem. Ser. 102, 238 (1971).

Ill

Zowel Bayer en Hagenmaier als Morozova en Zevail menen ten on-

rechte dat de door hen uitgevoerdechloride-bepaling bij devaste-fase
peptidesynthese een maat is voor het rendement van een koppelings-
stap.
E. Bayer en H. Hagemnaier, Tetrahedron Lett. 1968, 2037.
E. A. Morozova en M. A. Zevail, J. Gen. Chem. USSR 40, 1365 (1970).

. IV
Harrison e. s. tonen het nut van het gebruik van moleculaire zeven
bij hun veresteringstechniek niet aan.
H. R. Harrison, W. M. Haynes, P. Arthur en E. J. Eisenbraun, Chem.
Ind. (London) 1968, 1568.

De bereiding van di-tert. butylacetalen uit aldehyden en tert.butyl-

alkohol is niet zo gemakkelijk als diverse leerboeken en overzichts-
werken, die zich baseren op een onvolledig onderzoek van Adkins,
Adams en Minne, willen doen geloven.
H. Adkins en E. W. Adams, J. Amer. Chem. Soc. 47, 1368 (1925).
N. Minne en H. Adkins, J. Amer. Chem. Soc. 55, 299(1933).