victor a. veefkind · victor. contents chapter 1 general introduction 1 chapter 2 on the elementary...

SHAPE SELECTIVE SYNTHESIS OF ALKYLAMINES OVER

ACID CATALYSTS

Victor A. Veefkind

Leden van de promotiecommissie

Prof.dr. J.A. Lercher (promotor)Prof.dr. W.E. van der Linden (voorzitter/secretaris)Prof.dr. R.A. van Santen (TU Eindhoven)Dr. K. Seshan (Universiteit Twente)Prof.dr.ir. G. Versteeg (Universiteit Twente)Prof.dr.ir. H. Verweij (Universiteit Twente)Dr. F.H.M. Dekker (Akzo Nobel Arnhem)Dr. C. Williams (Shell Research and Technology Center Amsterdam)

The research for this thesis was carried out in connection with NIOK, the NetherlandsInstitute for Catalysis Research, and supported by the Department of Economic Affairs.Het onderzoek voor deze dissertatie is uitgevoerd in het kader van NIOK, het NederlandsInstituut voor Onderzoek van Katalyse, en met steun van het Ministerie van EconomischeZaken.

ISBN 90-365-1137-2

© Victor Veefkind, Enschede, The Netherlands

SHAPE SELECTIVE SYNTHESIS OF ALKYLAMINES OVER

ACID CATALYSTS

PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de Universiteit Twente,

op gezag van de rector magnificusprof.dr. F.A. van Vught

volgens het besluit van het College voor Promotiesin het openbaar te verdedigen

op donderdag 11 juni 1998 te 13:15 uur.

doorVictor Adriaan Veefkindgeboren op 26 juli 1970

te Amsterdam

Dit proefschrift is goedgekeurd door de promotorprof.dr. J.A. Lercher

For Sheila,for my parents and Ruben

Acknowledgments

Naturally the completion of a thesis involves more than just the PhD student. Over the

years a number of people have made smaller or larger contributions to this work, either

directly or indirectly.

Firstly I would like to express my gratitude to my promotor, Prof. Johannes A.

Lercher, not only for offering me the opportunity to perform research in the catalysis group,

but also for many fruitful and inspired discussions, critical comments, the supervision of this

work and the practical demonstrations of ‘how to get from A to B in the fastest possible way

by car’.

I would also like to thank Gabi and Christian for introducing me to zeolite catalysis

and for the many discussions and helpful suggestions. Also I would like to thank Martin

Smidt for his work on the partially exchanged mordenites.

I am grateful to the NIOK and the Dutch ministry of Economic affairs for financial

support and the members of the ‘begeleidingscommissie’ for critical questions and helpful

comments. Special thanks to Paul Kunkeler for scientific collaboration in a number of areas

and also to Prof. Herman van Bekkum for useful corrections for this thesis.

Then I want to thank the people in the catalysis group for the fact that I always could

drop into your rooms for a chat or a discussion. A special thanks for the Austrian delegation

who was here already when I first arrived, I really enjoyed your introduction into Austrian

culture and cuisine, and your uncanny ability to make a LIIT party a very memorable event.

Special thanks to Willie, Edu, Josemar, and of course to Axel and Saskia, for many

long nights with deep and sometimes also less deep conversations and the sharing of

unhealthy amounts of alcoholic beverages. It was always fun to have you around. Gerhard,

Pierre and Harry also for playing tennis and all football players for your boundless

enthusiasm and efforts, which finally got rewarded this year.

I would like to give special thanks to Karin, always there when I needed compounds

in a great hurry, to Cis for peptalk and whenever something needed to be arranged, to Bert

‘shepherd’ Geerdink for whenever I needed a beautiful digital picture or in fact anything

regarding computers. Cis and Bert, thanks also for the really enjoyable cooperation in

organizing symposia. Ipskamp Printpartners for the flexibility regarding the printing of this

thesis.

Then I would like to thank my parents who have always supported me in whatever I

did and without whom I would simply not be here! Also thanks to Rubencito, always full of

stories and master of the one-liners.

Last but certainly not least I would like to thank Sheila. Your support and love have

been invaluable. Thanks for being with me, I know it has not always been easy in the last

months of this thesis and thanks for staying up late with me on many a night to give your care

and support.

Victor

Contents

Chapter 1 General introduction 1

Chapter 2 On the elementary steps of acid zeolite catalyzed amination of light 15

alcohols

Chapter 3 Zeolite catalysts for the selective synthesis of mono- and diethylamines 37

Chapter 4 Role of strength and location of Brønsted acid sites for ethylamine 63

synthesis on mordenite catalysts

Chapter 5 On the potential to synthesize larger and mixed alkylamines with 91

zeolite catalysts

Chapter 6 Summary - Samenvatting 105

Curriculum Vitae 112

General Introduction

1

CHAPTER 1


Chapter 1

2

1.1 Amines

1.1.1 Historic perspective

In 1849, Wurtz prepared methylamines and ethylamines by hydrolysis of the

corresponding alkylisocyanates, trialkylcyanurates and alkylureas [1]. Substantial fundamental

work on synthesis, properties and structure of amines was done by Hofmann [2], who was also

the first to use the terms 'primary', 'secondary' and 'tertiary'. Industrially [3,4] the synthesis of

methylamines in batch mode from methanol and ammonia, using zinc chloride, was first reported

in 1884 [5]. The first report of amination of alcohols in the gas phase was in 1909 [6].

Methylamines were first made commercially in the 1920s for use in the tanning industry for the

dehairing of animal skins by Commercial Solvents Corporation in Terra Haute, Indiana [7]. The

process used at that time and the current processes [3,4,8] are essentially the contact of gaseous

methanol and ammonia over dehydrating catalysts (e.g. silica-alumina), followed by collection

and separation of the products. For higher aliphatic amines, catalysts having hydrogenating and

dehydrogenating properties have also become important.

1.1.2 Use of amines

Aliphatic amines are amongst the most important intermediates in the chemical industry.

The worldwide annual production of amines is estimated to be several hundreds of thousands of

tons. Methylamine production constitutes approximately half of this number. About 35-40% of

the remaining half are ethylamines [3]. Most attention will be focused on these two groups of

amines. Below the most important uses of methylamines and ethylamines are listed.

Monomethylamine (MMA) is mainly used as intermediate in the production of several

herbicides, pesticides and insecticides, for the production of Tovex (water gel explosive), the

solvent N-methyl-2-pyrrolidone and some pharmaceuticals.

Dimethylamine (DMA) is the compound with the highest demand on the global market.

It is used, amongst others, for the production of the solvents dimethylformamide and

dimethylacetamine, for water treating agents, surfactants, rubber processing compounds and a

range of agrochemicals.

Trimethylamine (TMA) is lesser in demand than the two other methylamines. It is used

as acid scavenger, and as intermediate in the production of choline chloride (animal food


3

supplement) and ion exchange resins.

Monoethylamine (MEA) is mainly used in the production of a wide range of herbicides

of the triazine type by reaction with cyanuric chloride, and also in the production of plasticizers.

Diethylamine (DEA) is used mainly for the production of vulcanization accelerators for

manufacturing of rubber, but also in the production of DEET (an insect repellent), and a number

of other specialty chemicals.

Triethylamine (TEA) is used as acid acceptor in organic synthesis, a salt former in

precipitation and purification operations (for example in the production of semisynthetic

penicillins), it is also used as polyurethane catalyst, anti-corrosion agent and in photographic

auxiliaries.

1.1.3 Current commercial practice

Most methylamines processes currently use amorphous silica-alumina catalysts for the

exothermic reaction which takes place at 663 - 703K and approximately 20 bar. Adiabatic fixed

bed tubular reactors are used for this purpose [9]. The product distribution obtained approaches

equilibrium, which gives a MMA/DMA/TMA molar ratio of 17/21/61. The demand for

methylamines, however, approximates a 30/60/10 MMA/DMA/TMA molar ratio, i.e., DMA is

in highest demand. Usually DMA production is enhanced by recycling TMA. A simplified

process diagram of a typical amine plant, including the purification section, is shown in Figure 1

[8].

Research efforts have been focussed on the use of shape selective catalysts, e.g. zeolites,

which could enhance DMA yields and decrease TMA yields in order to minimize recycling [4].

Nitto is currently using an alternative strategy in which two reactors, one with silica-alumina and

another with a modified mordenite, are used [10,11]. This modified mordenite catalyst will be

discussed in greater detail later in this chapter.

For ethylamine synthesis, apart from silica-alumina, also hydrogenation/ dehydrogenation

catalysts are used. These are mainly based on nickel, cobalt, iron or copper, and to a lesser extent

on platinum or palladium. In this process, similarly to the process based on silica-alumina, a

fixed bed is used. In addition to the alcohol and ammonia, hydrogen is added to the reaction

Ch

ap

ter 1

4

Figure 1. Typical amination reactor and separation train. (after [8])


5

Table 1. Major reactions during alkylation of ammonia with alcohols

Amination (=alkylation of ammonia)

1.2.3.

NH3

RNH2

R NH2

+++

ROHROHROH

��

RNH2

R NH2

R N3

+++

H O2

H O2

H O2

Disproportionation (=transalkylation)

4. R NHx y + R NHa b � R NH(x+1) (y-1) + R NH(a-1) (b+1) y=3-x, b=3-a

Side reactions

5. ROH + ROH � ROR + H O2

6. C H OHx (2x+1) � C Hx 2x + H O2 x>1

7. (C H ) NH Nx (2x+1) a b+ � C Hx 2x + (C H ) NH Nx (2x+1) (a-1) (b+1)

+ x>1

mixture. The reaction takes place at pressures between 5 and 200 bar and about 373-523K,

depending on the catalyst and whether a liquid phase or gas phase process is used. A twofold to

eightfold excess of ammonia is used; the hydrogen is not needed as a direct reactant but is used

to maintain the activity of the catalyst [3].

Synthesis of ethylamine and higher amines via the solid acid catalyzed route has the

advantage over the metal catalyzed route that no hydrogen, and generally lower pressures are

needed. The main drawback is that production of olefins is facile. This can cause deactivation

of the catalyst or an improperly functioning purification unit, caused by the oligomerization

products of these olefins. Here lies one of the more challenging tasks for research in solid acid

catalyzed ethylamine synthesis.

1.1.4 Reactions and thermodynamics

Table I shows the different classes of reactions that play a role in the amination of

alcohols over solid acid catalysts. The alkylation reactions are exothermic and are regarded as

irreversible due to their high equilibrium constant. They proceed in sequential order to yield the

mono-, di- and trisubstituted amine. The transalkylation reactions are regarded as being reversible

and are held responsible for the amine product distribtion at higher alcohol conversion.

0

20

40

60

80

100

0 2 4

NH3/MeOH ratio

Am

ine

se

lect

ivity

(m

ol%

)

0

10

20

30

40

50

550 600 650 700 750

T (K)A

min

e s

ele

ctiv

ity (

mo

l%)

Chapter 1

6

Figure 2. Dependence of the equilibrium amine distribution on (a) the NH /MeOH ratio at3

633 K and 20 bar and (b) the reaction temperature at NH /MeOH =1 and 20 bar. (4) MMA,3

(�) DMA, (�) TMA.

The formation of ether from two alcohol molecules is the most important side reaction

in methanol amination. In ethanol amination the production of ethene from either direct water

elimination or from a Hofmann type elimination (reaction 7 , Table 1) is the most important side

reaction. Reactions involving the formation of an olefin are irreversible under normal conditions

for amination (1-20 bar and 523- 673K). Note that the amination reactions can in principle also

proceed with dialkyl ether as alkyl source [12].

The equilibrium distribution of the reactants and products in methanol and ethanol

amination are given in Figures 2 and 3. These figures immediately show the necessity to shift the

reactions away from equilibrium. In methylamine synthesis this is required to increase DMA

production at the expense of TMA production and thus to obtain a selectivity to the amines

which is better in accordance with market demand. In ethylamine synthesis this need arises from

the necessity to avoid the formation of ethene and its oligomerization products and thus to

increase catalyst and equipment lifetime. It is obvious that the catalyst can play a key role in

shifting product selectivity to the desired products and avoiding to approach thermodynamic

equilibrium.

(a)

0

20

40

60

80

100

0 5 10 15

NH3/EtOH ratio

Am

ine

se

lect

ivity

(m

ol%

)

(d)

0

20

40

60

80

100

550 600 650 700 750

T (K)

Se

lect

ivity

(m

ol%

)

(c)

0

20

40

60

80

100

550 600 650 700 750

T (K)

Se

lect

ivity

(m

ol%

)

(b)

0

20

40

60

80

100

0 5 10 15

NH3/EtOH ratio

Se

lect

ivity

(m

ol%

)


7

Figure 3. Dependence of the equilibrium product distribution on: (a) and (b) NH3/EtOHratio at 573 K and p=1 bar, and (c) and (d) reaction temperature at NH /EtOH=4 and p=13

bar. (a) and (c) without ethene formation, (b) and (d) with ethene formation. (4) MEA, (�)DEA, (�) TEA, (z) ethene.

1.2 Zeolite catalysts for amination

1.2.1 General

Zeolites are currently applied industrially for a wide variety of heterogeneously catalyzed

processes [13,14,15]. Two unique properties of zeolite catalysts account for their wide range of

industrial applications, the same properties that make them very suitable for amination reactions:

(i) Zeolites have uniform pore systems of a size that is comparable to a number of organic

Chapter 1

8

molecules. They are crystalline materials that are composed of a three dimensional network of

metal oxygen tetrahedra with a one-, two- or three-dimensional channel system depending on the

way these tetrahedra are linked together [16,17]. Zeolites can be classified according to their

largest pore size. 'Small pore' zeolites are those containing 8-membered ring openings, 'medium'

containing 10-membered rings and 'large' containing 12-membered rings. Zeolites of all three

classes have been tested for amine synthesis. The pore openings of these zeolites range from 3

to 7.5Å and allow for exclusion of molecules based on their minimum kinetic diameter and

shape. This can be a very useful tool in catalysis to impose selectivity to a certain product (shape

selectivity). Several authors have described the effect of shape selectivity in a number of review

articles [18,19,20,21]. Mechanistically, different reasons for the occurrence of shape selectivity

are distinguished, i.e., (i) reactant shape selectivity (a consequence of one reactant being too large

to pass through the zeolite channel) and (ii) product selectivity (when only certain products are

of the proper size and shape are able to diffuse out of the channels. (iii) Transition state

selectivity occurs when the corresponding transition state of a certain reaction requires more

space than available in the framework of the zeolite. Zeolites, due to their pore size on molecular

level, are excellent catalysts to obtain high selectivities to lower alkylated amines.

(ii) Zeolites can contain high concentrations of localized acid sites. The acid sites of

zeolites are an integral part of the microporous structure resulting from an imbalance between

the metal oxygen stoichiometry an the formal charges of the cations. In zeolites the tetrahedra

are based on silicon and oxygen. In this network of tetrahedra a Si atom has a charge of +4, an

O atom of -2. As every O atom belongs to two tetrahedra, a purely silicious lattice is neutral and

possesses no acidity. Substituting part of the Si atoms by Al (+3), creates a negative charge at the

Al-O tetrahedra, which is balanced by a metal cation (Lewis acid site) or a proton (Brønsted acid

site) [16,22,23]. Thus these acid sites are localized and their concentration is proportional to the

aluminum concentration in the lattice. Due to their open structure, the accessibility of acid sites

is much larger than for amorphous materials of similar composition. High surface concentration

of reactants and longer residence times of reactants in the pores generally additionally enhance

the activity of zeolites [18,24]. For amination of methanol and ethanol the activity of silica-

alumina is compared with some zeolites in Table 2, (taken from ref.[25]).

Because the market for methylamines is by far the largest, most research into zeolite


9

Relative rates

Methanol amination Ethanol amination

Catalyst T=560 K T=707 K T=560 K T=707 K

Silica-alumina 1 1 1 1

H-Erionite 15 9.5 0 0.8

H-Mordenite 14 12 1 0.6

H-Y 19 15 11 2.5

All reactions carried out with 2:1 ammonia/alcohol, 20 bar, 9300 h GHSV at STP -1

Table 2. Relative rates of methanol amination and ethanol amination [25]

catalyzed amination of alcohols has been centered around catalysts for methanol amination. For

this reaction numerous zeolites have been tested as catalyst, such as ZK-5, rho, chabazite,

erionite, offretite, ZSM-5 and mordenite, ranging in size from 'small pore' to 'large pore' zeolites.

Corbin et al. have compiled a large number of reported results in methanol amination over

zeolites in an excellent review [4]. The zeolite catalyzed amination of ethanol is much less

described in literature [25,26,27]. In this thesis the attention will be focussed on the 'large pore'

zeolites mordenite (MOR), faujasite (FAU), mazzite (MAZ) and beta (BEA). The structures of

these zeolites are depicted in Figure 4 [28,29].

Considerable research efforts have been directed towards the use of mordenite in

methanol amination. It was found that this catalyst, especially after different post-synthesis

treatments, was one of the most effective amination catalysts and it is currently in use as catalyst

for methylamine production in the Nitto-process [30]. Apart from the commercial relevance of

mordenite as an amination catalyst, much of the fundamental knowledge in methylamine

synthesis has been obtained using mordenite as a model catalyst [31,32,33,34,35,36]. It is the

object of this thesis to explore further these fundamental aspects of amination, also directed to

ethylamine synthesis. Focusing on mordenite as amination catalyst seems a logical choice.

1.2.2 Mordenite in amination of methanol

Mordenite is one of the most successful zeolite catalysts for methanol amination. The

Chapter 1

10

Figure 4. Zeolite structures, emphasizing the diameter of the 12-ring; (a) BEA, (b) FAU,(c) MOR, and (d) MAZ.

catalysts with the highest DMA yield were mordenites which were subjected to a post-synthesis

treatment. Especially treatment with silylating agents was found to be very effective [35,37,38].

The most common methods are chemical vapor deposition of SiCl [35,39] or application of4

alkoxysilanes (viz., tetraethylorthosilicate (TEOS)) [38,40] in the slurry phase.

The chemical vapor deposition of SiCl onto mordenite is reported to result in a4

narrowing of the pores, but can easily lead to the formation of extraframework aluminum species.

Segawa et al. [37] claim a procedure in which the removal of aluminum from the framework and

deactivation of Brønsted acid sites is minimized. in this procedure the Na-form of the mordenite

is treated with SiCl and only after this treatment ion exchanged into its acidic form.4

H-MOR20

MMA

DMA

TMA

TEOS H-MOR20

MMA

DMA

TMA


11

Figure 5. Amine selectivities (C%) for H-MOR20 and TEOS H-MOR20 at T=633 K and 90%MeOH conversion (after [38]).

The treatment with TEOS is easier than that with SiCl . It involves suspending the zeolite4

in a suitable organic medium, such as n-hexane or toluene, and injecting an amount of silylating

agent sufficient to deposit 1-15 w% of silica on the zeolite [40] under controlled stirring. The size

of the TEOS molecules is too large to enter the pores of the mordenite. Consequently, it was

found that this treatment selectively covers only the outside of the mordenite crystallites with a

silica layer [37,38], decreasing the average diameter of the pore opening and leaving the inner

pore dimensions unaltered. This treatment imposes a diffusion controlled selectivity on the

higher alkylated amines, as was shown by sorption experiments. TMA, the unwanted product is

unable to leave the pores and is reacted with ammonia to MMA and the wanted product DMA.

Upon this treatment the selectivity to DMA was drastically increased as is shown in Figure 5

(After [38]). Additionally the selectivity to dimethyl ether decreased from 15 mol% to 2 mol%.

Using microcalorimetry [41] it was found that the heats of sorption of base molecules on

mordenite was on average higher than on any other tested zeolite. The strong acidity of

mordenite, compared to other zeolitic materials was confirmed by i.r. spectroscopy [42] using

H as probe molecule and observing the shifts in wavenumber for both the O-H stretching2

vibration of the acid site and the H-H stretching vibration of the sorbed probe molecule. Note in

this context that the strength of the interaction between the acid site and the base molecules can

be an important factor in amination of alcohols.

Chapter 1

12

1.3 Scope and structure of this thesis

Zeolite catalyzed amination of alcohols is becoming an increasingly important process. The

synthesis of methylamines over zeolite catalysts has been extensively studied and a commercial

process using a zeolite in combination with an amorphous catalyst is now practiced.

The elementary steps in the synthesis of methylamines, with respect to mechanism and cause of

shape selectivity, are slowly being understood over the last years, but are still cause for debate.

In order to be able to design a shape selective catalyst it is important to be aware of the

functioning of the catalyst on a molecular level and more clarity on this subject is desirable.

Thus the main goal of this thesis is to provide more clarity in the action of zeolite

catalysts in alcohol amination on a molecular level and correlating these to observed selectivities

and activities in the amination of alcohols. Since, contrary to zeolite catalyzed amination of

ethanol and higher alcohols, methylamine synthesis has already extensively been described in

literature, this will provide the basis for describing the mechanistic steps and the correlation to

observed selectivities and activities.

Thus, in Chapter 2 of this thesis, reported mechanistic views for methanol amination will

be shortly reviewed. Transient experiments using isotopic labeling, followed by GC-MS and in-

situ i.r. spectroscopy will be the primary means to address some of the open questions regarding

selectivity issues on a molecular level. Additionally, the mechanism of ethanol amination will

be compared to that of methanol amination, using coadsorption experiments followed by i.r.

spectroscopy under vacuum conditions and in-situ i.r. spectroscopy under reactive conditions.

The insight in the basic steps in ethanol amination will be used in Chapter 3, that will deal

specifically with ethanol amination over different large pore zeolites. The activities and

selectivities obtained using different catalysts are discussed in terms of the pore geometry and

acid site density. In order to suppress the formation of byproducts, either by influencing the

process parameters or by the design of the zeolite, mechanisms for formation of the most

important byproducts should be explored. The formation of ethene will discussed in this chapter,

based on kinetic experiments and temperature programmed desorption. Based on this

information, basic requirements for a stable operation with minimized ethene formation are

outlined.

In Chapter 4, the knowledge acquired in Chapters 2 and 3 is used in a detailed


13

1. A. Wurtz, C.R. Hebd. Seances Acad. Sci., 28, 223 (1849).

2. E. Fischer, Ber. Dtsch. Chem. Ges. Sonderheft 1902, 197.

3. G. Heilen, H.J. Mercker, D. Frank, R.A. Reck, and R. Jäckh, in W. Gerhertz (Ed.),Ullmann’s Encyclopedia of Industrial Chemistry, 5 ed., VCH, Weinheim, 1985, Vol.th

A2, p.1.

4. D.R. Corbin, S. Schwarz and G.C. Sonnichsen, Catal. Today, 37, 71 (1997).

5. V. Mertz, K. Gasiorowski, Ber. 1884, 623.

6. P. Sabatier, A. Mailhe, Compt. Rend., 148, 898 (1909).

7. R.S. Egly, E.F. Smith, Chem. Eng. Progr., 44, 387 (1948).

8. M.G. Turcotte and T.A. Johnson, in J.J. Kroschwitz (Ed.), Kirk Othmer Encyclopediaof Chemical Technology, 4 ed., John Wiley & Sons, New York, 1992, Vol 2, p. 369.th

9. R.N Cochran, M. Deeba, US 4 398 041 (1983).

10. W.S. Fong, PEP Review, 89-3-4 (1991).

11. Y. Ashina, M. Fukatsu, US 4 485 261 (1984).

12. A. Martin, B. Lücke, W. Wieker, and K. Becker, Catal. Lett., 9, 451 (1991).

13. I.E. Maxwell, W.H.J. Stork, Stud. Surf. Sci. Catal., 58, 571 (1991).

14. B.C. Gates, J.R. Katzer and G.C.A. Schuit, Chemistry of Catalytic Processes, McGraw-Hill, New York, 1979.

15. J.A. Moulijn et al. in: J.A. Moulijn, P.W.N.M. van Leeuwen and R.A. van Santen(Editors), Catalysis: An Integrated Approach to Homogeneous, Heterogeneous andIndustrial Catalysis, Elsevier, Amsterdam, 1993, p.33

16. K. Tanabe, Solid Acids and Bases, Kodansha, Tokio, 1970.

17. E.M. Flanigen, Stud. Surf. Sci. Catal., 58, 13 (1991).

18. N.Y. Chen, T.F. Degnan, Jr., and C.M. Smith, Molecular Transport and Reactions inZeolites, VCH Publishers, New York, 1994, p. 173.

investigation into the influence of steric factors in ethylamine synthesis over mordenite. A

microscopic model for ethylamine synthesis in mordenites is presented and discussed in terms

of accessibility of acid sites and the nature of the alkylating agent. This is combined with a study

into the acid site location in NaH-mordenite catalysts.

Knowing in more detail the reaction steps for the synthesis of methylamines and

ethylamines and the parameters which influence these steps, this allows us to discuss these

implications for the synthesis of other methylamines, such as ethylmethylamines and

propylamines. This is done in chapter 5 of this thesis.

References

Chapter 1

14

19. P.B. Weisz, Pure Appl. Chem., 52, 2091 (1980).

20. E.G. Derouane, Stud. Surf. Sci. Catal., 5, 5 (1980).

21. J. Dwyer and A. Dyer, Chem. Ind., 265, 237 (1984).

22. J. Ward, J. Catal., 9, 231 (1967).

23. M.E. Davis, Ind. Eng. Chem. Res., 30, 1675 (1991).

24. J.N. Miale, N.Y. Chen and P.B. Weisz, J. Catal., 6, 278 (1966).

25. M. Deeba, M.E. Ford, T.A. Johnson, Amination with zeolites, Catalysis of organicreactions, Dekker, New York, 1990

26. W.W. Kaeding, US Patent 4,082,805 (1978).

27. J-P Shen, J Ma, D-Z Jiang and E-Z Min, Chinese Chem. Let., 5(4), 305 (1994).

28. W.M. Meier and D.H. Olson, Atlas of zeolite structure types, 3 edition, Butterworth-rd

Heinemann, Stoneham, 1992.

29. InsightII program, MSI Technologies Inc.

30. Jpn. Chem. Week, 33, 2 (1992).

31. Y. Ashina, T. Fujita, M. Fukatsu, K. Niwa and J. Yagi, Stud. Surf. Sci. Catal., 28, 779(1986).

32. Ch. Gründling, G. Eder-Mirth and J.A. Lercher, Res. Chem. Intermediates, 23 (1), 25(1997).

33. Ch. Gründling, V.A. Veefkind, G. Eder-Mirth and J.A. Lercher, Stud. Surf. Sci.Catal., 105, 591 (1997).

34. A. Kogelbauer, Ch. Gründling and J.A. Lercher, J. Phys. Chem., 100, 1852 (1996).

35. K. Segawa and H. Tachibana, in Guczi et al. (Eds.), Proc. 10 ICC, Elsevier,th

Amsterdam, 1993, p. 1273.

36. I. Mochida, A. Yasutake, H. Fujitsu and K Takeshita, J. Catal., 82, 313 (1983).

37. K. Segawa and H. Tachibana, J. Catal., 131, 482 (1991).

38. Ch. Gründling, G. Eder-Mirth and J.A. Lercher, J. Catal., 160, 299 (1996).

39. C.V. Hidalgo, M. Kato, T. Hattori, M. Niwa and Y. Murikami, Zeolites, 4, 175(1984).

40. T. Kiyoura and K. Terada, Eur. Patent Appl. 593.086, 1994

41. C. Lee, D.J. Parillo, R.J Gorte and W.E. Farneth, J. Am. Chem. Soc., 118, 3262(1996).

42. M.A. Makarova, V.L. Zholobenko, K.M. Al-Ghefaili, N.E. Thompson, J. Dewing andJ. Dwyer, J. Chem. Soc. Faraday Trans., 90(7), 1047 (1994).

On the Elementary Steps of Acid Zeolite Catalyzed Amination of Light Alcohols

15

CHAPTER 2

On the Elementary Steps of Acid Zeolite Catalyzed Aminationof Light Alcohols

Abstract

Potential elementary reaction steps in solid acid catalyzed amination of light alcohols are

critically compared using hydrogen mordenite as (model) catalyst. Transient kinetic experiments

combined with in-situ infrared spectroscopy and isotopic labeling techniques were the main experimental

means. While it was shown before that ammonium ions rapidly react with methanol to produce

methylammonium ions, these results indicate that also methoxy groups formed from methanol with the

strong acid sites react with ammonia to methyl ammonium ions. The rates of formation of amines via

methoxy amination or ammonium ion alkylation are similar. As the availability of free strong acid sites

under reaction conditions is very low it is concluded that the contribution of the methoxy group

amination to the amine formation is very small. The rate determining step for the overall reaction is the

ammonia mediated removal of the alkylammonium ions from the zeolite. It is shown that two

interconnected reaction steps govern this, i.e., alkyl scavenging and adsorption assisted desorption. The

combination of both steps determines observed rates and selectivities to the different amines. The main

pathway to ether formation is shown to be reaction of two alcohol molecules on weakly acidic sites such

as ammonium ions inside the pores. This reaction is rather susceptible to the available pore volume. A

high concentration of sorbed alkylammonium ions or a high substitution of these alkylammonium ions

decreases this available pore volume and thus ether formation. The formation of alkenes in the synthesis

of larger alkylamines over strongly acidic catalysts is largely attributed to Hofmann elimination

(decomposition of ethylamines) and not to elimination reactions of the alcohol reactant.

Chapter 2

16

2.1 Introduction

Shape selective zeolite catalyzed alkylation of ammonia by alcohols has attracted

significant attention over the past years [1,2,3,4]. This interest has its reason in the potential to

replace silica-alumina with catalysts able to produce preferentially mono- and di-alkylamines.

Most investigations were focused on methylamines, being the largest volume chemical and the

only amine that could be produced with very high yields [2,4]. For larger alkylamines the loss

of alkylating agent as alkene has prevented the acid catalyzed route to alkylamines to be

commercially of interest. Reductive amination of aldehydes using metal catalysts is the

commonly practiced route [3] in industry.

In order to overcome the limitations of the present generation of zeolite catalysts with

respect to activity and selectivity, it is necessary to understand the complex reaction sequences

occurring during acid catalyzed alkylation in the molecular sieve pores. The elementary reactions

involved can be divided into three groups: (i) sequential methylation of ammonia to mono-, di-

and trialkylamine, (ii) disproportionation, i.e., reactions between nitrogen containing species and

(iii) side reactions leading to unwanted products such as ethers or olefins. Mechanistic studies

have led to a relatively detailed picture of the possible elementary steps[5,6,7,8,9,10] and the

reaction models put forward using these studies have been critically compared in an excellent

review by Corbin et al.[11].

For methanol amination over acidic catalysts several mechanisms have been suggested

proceeding either over single acid sites or acid-base paired sites [5-11]. It has been speculated

that methoxy groups could be involved in the process [5], although it was noted that such species

could not be observed under reaction conditions and probably do not play a major role. Corbin

et al. [11]suggested that a weakly adsorbed ammonia reacts with a protonated methanol. This

methanol molecule may be protonated either by an adsorbed ammonium ion or by an acid site.

Disproportionation reactions occur by a reaction of gas phase amine with a protonated sorbed

amine. By contrast, Chen et al. [9] proposed that weakly adsorbed methanol reacts with an

ammonium ion or methylammonium ions. More specifically this was shown to involve

protonation of methanol by an ammonium ion and the subsequent interaction of the ammonia

nitrogen with the carbon atom of the protonated methanol [6,7,8]. Most likely because of the

good leaving group (H O) alkylammonium ions and water are rapidly formed. As boundary2


17

conditions it should be noted that the formation of a free acid site and a gas phase amine is an

unlikely elementary step, as free acid sites were undetectable under reaction condition for MOR

[5,8] and that the rate of amine desorption is several orders of magnitudes lower than the

observed reaction rate [9]. For larger alkylamines, desorption leads to decomposition and has

even been used for characterization of strong Brønsted acid sites. [12,13] .

Using in situ i.r. spectroscopy Gründling et al.[8,10] demonstrated that alkylation of

ammonium ions in zeolites by methanol occurs with high rates. The conditions of this experiment

(633 K and the absence of gas phase ammonia) preclude the presence of weakly adsorbed

ammonia or gas phase ammonia as suggested in the primary reaction step by Corbin et al. [11].

Upon alkylation of ammonium ions a mixture of methylammonium ions is observed covering

all acid sites of the parent material.

The methylammonium ions formed by alkylation of NH are stable at 633K and do not4+

leave the pores unaided. Only in the presence of ammonia, methylamines were observed in the

gas phase. Similar detailed observations were made recently for ethanol amination [14]

underlining the expected similarities between methanol and ethanol amination. For the ammonia

mediated release of amines into the gas phase two models were proposed, i.e., the “scavenging

mechanism” and the “adsorption assisted desorption (a.a.d.) mechanism”. Both mechanisms are

schematically depicted in Figure 1 (after [8]). The main difference is that in the scavenging

mechanism the sorbed alkylammonium ion remains at the active site and serves as a (rapidly

replenished) methyl source, while in the adsorption assisted desorption mechanism the sorbed

alkylammonium ion is replaced by another ammonia or amine molecule.

In that respect it was observed that the overall product selectivity did not match the

relative abundance of the different amines in the sorbed phase, which was enriched in higher

methylammonium ions. This was especially pronounced with surface modified mordenite to

slightly reduce the average pore mouth diameter [10,15]. The rate of monomethylamine (MMA)

formation was shown to correlate linearly with the total concentration of methyl groups in the

zeolite pores. This strongly suggested that the “scavenging mechanism” is predominant for amine

release. Also the fact that reforming of MMA or dimethylamine (DMA) to higher and lower

substituted amines is facile over acid catalysts [9,16] indicates the scavenging mechanism is

operative during amination. However, these arguments do not exclude an important contribution

Chapter 2

18

Figure 1. Mechanisms for ammonia mediated release of amines from the acidsites (after [8]).

of ammonia desorption assistance, either parallel to alkyl group scavenging or as a consecutive

step.

A technologically important concern has been the formation of ethers and olefins during

amination over acidic catalysts. Ether formation has been proposed to occur in a concerted way

i.e., via reaction between a protonated and a second coadsorbed methanol molecule [17] and via

reaction between an alkoxy species and a methanol molecule [9]. The reaction between a weakly

sorbed methanol and one attached to a Lewis acid site has also been suggested [7]. Additionally,

scavenging of methyl groups from sorbed methylammonium ions by methanol [18] and reaction

of two relatively weakly sorbed methanol species on top of ammonium ions[6,19] have been

proposed.

The second important byproduct, i.e., olefins, can be made either via direct elimination

of water over a free acid site or a Hofmann type elimination of the olefin from the sorbed

alkylammonium ion [12,13,20], i.e., the decomposition of the amine.

Based on this knowledge we report in this communication a series of mechanistic

experiments that should contribute to relating the elementary reaction steps to the overall reaction


19

Catalyst Specific Area

(m /g)2

Micropor. Vol.

(cm /g)3

Si/Al EFAL

(%)

Brønsted acid sites

(mol/g)

HMOR20 390 0.13 10 n.d. 1.3·10-3

HMOR15 350 0.15 7.5 11 1.7·10-3

n.d. = not detected (<5%)

Table 1. Physico-chemical properties of the investigated mordenite samples.

using a mordenite zeolite which we found to be a stable and promising catalyst for amine

synthesis. Additionally, the role of the pathways leading to ether and olefin formation is

addressed. Combined in situ i.r. spectroscopy and kinetic measurements using isotope labeling

and transient response techniques are the primary experimental means.

2.2 Experimental

2.2.1 Materials

The zeolite catalysts used in this study are Japanese reference catalysts (JRC) with a Si/Al

ratio of 10 (HMOR20) and 7.5 (HMOR15). The physicochemical properties are summarized in

Table 1 and have been described in detail previously[21,22].

Methanol and ethanol were obtained from Merck, p.a. grade. Mixtures of 5%

monomethylamine in He and 5% monoethylamine in He were obtained. as mixed gases of high

purity (99.999%) from Praxair, The Netherlands. Ammonia was used either as pure gas

(99.999%, Praxair) or as mixture of 5% in He (99.999%, Praxair). Additionally nitrogen labeled

ammonia was used from Isotech Inc.(99.9 atom% N).15

2.2.2 Measurement procedures

To record i.r. spectra under vacuum conditions, a Bruker IFS88 spectrometer was used,

equipped with a vacuum cell (base pressure 10 mbar), as described in [23]. The sample was-6

pressed into a self supported wafer and in situ evacuated under a dynamic vacuum of 10 mbar,-6

heated to 823K with 10K/ min., kept at 823K for 60 min. and cooled down to 323 K. The spectra

of the working catalyst under reaction conditions were taken on a Nicolet SXB20 spectrometer,

Chapter 2

20

using an i.r. reactor with a volume of 1.5 cm , which approximates a continuously stirred tank3

reactor [24]. The latter was integrated in a flow system equipped with a Hewlett Packard 5890

GC for simultaneous gas phase analysis. A sampling system with sample loops allowed to

measure transient responses with a resolution of approximately 5 seconds. For i.r. measurements

in both systems the sample was pressed into self supported wafers of approximately 3-5 mg/cm .2

Detailed descriptions of the reactor system and the procedures can be found in refs [8,24].

Experiments with isotopically labeled compounds were performed in a flow system with

a quartz plug flow reactor containing 30 mg HMOR15, which had been activated at 823K under

a He flow of 10 ml/min. Pressure transient experiments could be performed by switching from

a He stream to a stream containing 20% ammonia in He by means of a dead volume free

sampling valve. The products evolving from the reactor could be stored in a sampling valve,

allowing short sampling intervals. The contents of the sampling valves were analyzed on a

Hewlett Packard 5890 GC/MS, equipped with a Restek RTX-Amine capillary column and a

flame ionization detector. Details of the experimental procedures are described in ref. [25]. The

detailed experimental conditions are outlined per experiment in the results section.

2.3 Results and Discussion

2.3.1 Alkylammonium ions via methoxy groups as intermediates

So far most emphasis was given to the reactivity of an intact alcohol molecule with with

ammonia and most mechanisms assume that the reaction does not involve alkoxy groups at the

surface [7,8,9,11]. To explore the potential role of alkoxy type species in the formation of surface

bound alkylammonium ions, an experiment was designed in which methoxy groups were created

in the zeolite and subsequently exposed to ammonia at reaction temperature. After evacuation

of the self supporting wafer of MOR15 at 823 K for 1 hour, the wafer was exposed to 10 mbar-3

methanol at 323 K and i.r. spectra were recorded until equilibration. After this, the methanol

pressure was maintained at 10 mbar and the temperature was increased to 523 K with 10 K/min.-3

and kept there for 5 min. Then, the sample was evacuated for 1 hour and an i.r. spectrum was

recorded. Subsequently, the sample was heated to 633K in vacuum and then exposed to 10-3

010

2030

40 -.2

0

.2

.4

3500 3000 2500 2000 1500 Wavenumber (cm-1)

Absorbance

Time (m

in.)

NH3

MMA

MOR lattice vibr.

0

.02

.04

3100 3000 2900 2800

Wavenumber (cm-1)

Absorbance

2973

2865

(a) (b)


21

Figure 2.Infrared spectra; (a)C-H stretching region of i.r. spectrum of mordenite withchemisorbed methoxy species; (b)Difference spectra taken during reaction of ammonia withmethoxy species on HMOR15 at 633 K and p = 10 mbar.NH3

-3

mbar NH until changes in the spectra were not observed, followed by exposure to 10 mbar of3-2

NH . The spectra taken during these experiments are compiled in Figure 2.3

Fig. 2a shows the C-H stretching bands of MOR after treatment with methanol at 523

K. Fig. 2b compiles the spectra taken during the first 40 minutes of subsequent ammonia

addition. These spectra are shown in the form of difference spectra, i.e., the spectrum of the

activated sample is subtracted from the spectra after contacting with a adsorbing/reacting

substance, resulting in a negative and positive bands band depending wether the intensities of

bands decrease or increase with respect to the parent spectrum. Note that the first spectrum

clearly shows a negative band at 3610 cm as a result of Brønsted acidic hydroxyl groups being-1

transformed into methoxy groups. The positive band at 3565 cm and the vibrations in the region-1

indicated as ‘MOR lattice vibrations’ in Fig. 2b reflect changes in the mordenite due to the

temperature difference between the spectra taken during reaction at 633 K and of the activated

sample, taken at room temperature, and are not assigned to surface species.

The bands at 2973 and 2865 cm (Fig. 2a) are attributed to asymmetric and symmetric-1

CH stretching vibrations, the band at 1456 cm to the deformation vibrations of methoxy3-1

0%

4%

8%

12%

16%

20%

0 20 40 60 80 100

Time (min.)

Ac

id s

ite

co

ve

rag

e (

mo

l%)

0%

4%

8%

12%

16%

20%

0 200 400 600 800 1000

Time (min.)

Ac

id s

ite

co

ve

rag

e (

mo

l%)

Chapter 2

22

Figure 3. Acid site coverage with monomethylammonium ions, obtained byfitting of the i.r. spectra; (a) reaction of ammonia with methoxy species at 633K and p = 10 mbar; (b)scavenging of methyl groups by ammonia at T=NH3

-3

633K and p = 10 mbar.NH3-2

groups in agreement with literature [26,27]. These wavenumbers clearly differ from those of the

� and � of the methanol methyl group observed at 2957 and 2851 cm . The integral area ofas sym-1

the Brønsted acid site band at 3600 cm decreased 18% after the methanol treatment and-1


23

subsequent evacuation indicating a significant coverage of the acid sites by methoxy groups.

Oligomeric species resulting from reaction of methanol over acidic zeolites under MTG

conditions show bands at 2960, 2930 and 2870 cm [23].The absence of these bands (see Figure-1

2a) indicates that methoxy groups are the only sorbed species present in significant

concentration.

After heating to 633 K and exposing the sample to 10 mbar of ammonia, formation of-3

alkylammonium ions was observed as concluded from the appearance of the characteristic

asymmetric and symmetric deformation bands at 1605 and 1505 cm and the combination bands-1

at 2545 and 2455 cm . A multicomponent fit procedure as described in ref. [8] was applied to-1

these spectra and the result is depicted in Figure 3a. After 90 minutes changes between the i.r.

spectra were no longer observed and 18% of the strong Brønsted acid hydroxyl groups were

replaced by monomethylammonium ions was attained. This indicates that a quantitative

conversion of methoxy groups into methylammonium ions took place inside the mordenite

channels. Additional 44% of the hydroxyl groups were replaced by ammonium ions at that

pressure. Subsequently, the ammonia pressure was increased to 10 mbar. This resulted in a-2

sudden coverage of all acid sites with ammonia. In parallel, the gradually disappearing

methylammonium ions were monitored (see Figure 3b).

The rate constants for the reaction of methoxy groups with ammonia and the ammonia

mediated release of amines were calculated from a quantitative analysis of the infrared spectra

including a multicomponent fit. The rate constant was found to be 7 10 s mbar for the. -1 -1 -1

amination of methoxy groups and 3 10 s mbar for the scavenging reaction, i.e., two orders. -3 -1 -1

of magnitude faster. From earlier experiments reported by Gründling et al. [8] the rate of

methylation of ammonium ions was found to be 50 times higher and than the the release of

methylammonium ions induced by ammonia. This indicates that the formation of

methylammonium ions via amination of methoxy groups is slightly faster than the formation of

methylammonium ions by methylation of ammonium ions, but the difference is large for both

pathways .

Let us now address the question to what extent this reaction route can contribute to the

overall reaction. The rate constant of the ammonium addition to the methoxy group is only

moderately faster than the methylation of the ammonium ion [8]. Coadsorption of methanol and

3600 3200 2800 2400 2000 1600

0.2

Abs

orba

nce

(a.u

.)

Wavenumber (cm-1)

(a)

(b)

(c)

Chapter 2

24

Figure 4. Reaction of MEA with MEA at 558K; (a) after 3 min.; (b) after 100 min.;(c) ammonium ions on HMOR15.

ammonia, however, leads to preferential formation of ammonium ions and methanol

coordinating via hydrogen bonding to the ammonium ion [33,28]. Such an arrangement also is

well in line with the base strength of the two molecules. As the alkylammonium ions formed by

the surface reaction are even more stable than the ammonium ions, the concentration of methoxy

groups formed on the acid sites is concluded to be negligible during steady state operation over

strongly acidic zeolites. As the difference in the rates is at best a factor of 5 we conclude that

the pathway via amination of the methoxy groups is not important in the overall process.

The differences in the rate constants suggests that ammonia mediated desorption/reaction

is rate determining and the alkylammonium ions should consequently accumulate in the zeolite

pores. In order to explore the extent to which the possible elementary reactions (in the overall

reaction scheme described in Figure 1) contribute to ammonia mediated alkylamine release two

transient response experiments were performed.

For the first experiment 3 mg activated H-MOR15 were presaturated with


25

monoethylamine (MEA) in the in situ i.r. reactor. Then at 558 K, a feed of 6 ml/min. 5 vol%

MEA in He was passed over this MEA presaturated catalyst. The reaction of

monoethylammonium ions with MEA, was followed by GC and in situ i.r. spectroscopy. If the

incoming MEA displaces the monoethylammonium ions, MEA will be the only product detected

and methylammonium ions the only adsorbed species. If it reacts with the monoethylammonium

ions, diethylamine will be formed and the concentrations of ammonium ions will increase. The

ir. spectra at the start of the reaction and after 100 min. time on stream (see Figs. 4a and b) were

identical and typical for sorbed MEA. No indication of NH ions (Fig. 4c) was found,4+

supporting adsorption assisted desorption as major pathway. However, a steady state yield of

0.7% DEA was observed in the gas phase and after 100 min. the turn over number for DEA was

1.6, indicating that reaction between two MEAs must have taken place. This indicates that both

pathways, i.e., adsorption assisted desorption and alkyl scavenging play a role.

To gain more insight in the role of NH as scavenging molecule, experiments with N315

labeled ammonia were performed. For those experiments 30 mg HMOR15 were activated for

1 hr at 823K and then cooled to reaction temperature of 633 K. Subsequently, it was exposed to

a flow of 10 ml/min. of 20% NH in He for 15 minutes to quantitatively create labeled153

ammonium ions. Then, a 1:1 mixture of NH and methanol or ethanol was fed to the reactor153

for half an hour. The release of nitrogen labeled amines was monitored. Thereafter, NH was153

switched off and the alcohol was allowed to pass over the catalyst for 15 minutes to create high

degree of alkylation of the alkylamines. The material prepared so contained exclusively N15

labeled alkylammonium ions in the mordenite pores. After flushing the reactor with He for one

hour to remove excess NH in the system, 200 mbar NH in He were passed over the catalyst15 143 3

and the removal of the amines in the pores was monitored (see Figure 5 for the results for with

methylamines). In that graph the switch to non-labeled ammonia is tagged as t=0 min. In this

experiment a fixed pool of N-labeled amines exists with a fixed amount of methyl groups15

attached to them. Both may be removed by the stream of non-labeled ammonia. As a

consequence, all N-labeled amines analyzed must have originated from the zeolite pores and15

the non-labeled amines must have been formed by a scavenging type of reaction involving gas

phase ammonia.

It can be seen from these figures that the initial products are N-labeled amines15

MMA22%

15MMA45%

DMA2%

15DMA11%

TMA3%

15TMA17%

(b)

0 2 4 6 8

Time (min.)

0

100

200

300

400

Gas

pha

se a

bund

ance

(a.

u.)

(a) MMA

15MMA

DMA

15DMA

TMA

15TMA

Chapter 2

26

Figure 5. Reaction of 200 mbar NH with N-labeled methylammonium ions in HMOR15 at315

633K; (a) product distribution as function of time on stream; (b) product distribution integratedover duration of the experiment.

originating from chemisorbed species. The non-labeled species only slowly start to appear in the

gas phase. It can also be seen that the amount of non-labeled amine vs. labeled amine formed is

higher for MMA (primary product of scavenging by ammonia) than for DMA and TMA.

As the experiments described above, the isotope transients suggest that both reaction

pathways contribute to the overall reaction. Under the experimental conditions applied,

approximately 67 % of MMA was produced via adsorption assisted desorption, 33% via the

alkyl scavenging route. The fact that only small amounts of unlabeled (scavenged) di- and

trimethylamines are detected in the gas phase is explained with the fact that these products are

secondary and tertiary products for the alkyl scavenging route. The 15-20% of di- and

trimethylamines within the products produced via alkyl scavenging indicates that the substituted

amines are quite reactive. Also the fact that 67% of all formed amines are MMA, while the

surface concentrations of monomethylammonium ions at the start of the transient is much lower

(compare refs.[8,10]), indicates a significant role for scavenging.

Experimentally it was not possible to obtain in-situ information on the sorbed species

inside the working catalyst at high conversion. It has been shown [14] that at conversions

between 5 and 20%, all methylammonium ions are present in the pores, but changes in the

coverage with the different ammonium ions are very small. However, whereas at low

conversions MEA is the main observed product in the gas phase, at high conversions selectivity


27

to TEA is highest. Under these conditions the molecules in the pores, also those which are not

sorbed, will be higher substituted. Considering the higher base strength of the higher substituted

amines, it is likely that ammonia will be mainly engaged in scavenging type reactions under these

conditions. On the other hand formed TMA cannot scavenge and will therefore have to be

involved in adsorption assisted desorption. Thus it is expected that also at high alcohol

conversions, both mechanisms are operative.

2.3.2 The origin of ethers in amination

Dimethyl ether (DME) is by far the most abundant byproduct in the direct amination of

methanol. Its formation seems to be affected by several factors such as (i) the ammonia and

methanol partial pressures, (ii) the concentration of the acid sites (iii) the modification of the

outer zeolite surface by TEOS and subsequent hydrolysis (silylation) [10,14]. While it is

conceptually easy to understand that with increasing ammonia and decreasing methanol partial

pressure the rate and selectivity to form DME decreases, the other two factors are more subtle

to explain.

As it was observed that the silylation of the catalyst drastically reduced the DME

formation [14], one tends to speculate that the DME formation takes place at the outer surface

of the zeolite. Silylation is known to eliminate the (weak) acid sides on the outer surface [29,30].

If only these sites at the outer surface were active for DME production, one would expect for the

parent material a sympathetic variation of DME formation with acid site concentration (the

variations in the concentration of the acid sites should be reflected at the surface of the zeolite

crystals). In contrast, DME selectivity decreased with increasing concentration of acid sites [14].

Thus, let us tentatively conclude that the contribution of external acid sites to DME formation

is small (Note in that context that the estimated concentration of acid sites on the outside of the

parent material was approx. 0.1 mmol/g, based on the loss of acid sites during silylation, the

increase of percentage inactive material and EFAL).

If one assumes the DME to be formed inside the pores, three possible pathways can be

proposed: (i) Direct bimolecular reaction of two methanol over a Brønsted acid site, (ii)

scavenging of methyl groups by methanol, similar to amination, and (iii) methanol coordinated

to a weak sorption site (e.g., ammonium ion) reacts with another methanol molecule.

Chapter 2

28

The first of the three possibilities can be ruled out as only (alkyl)ammonium ions have

been observed in the materials studied and all sites have seen to be occupied. If

(spectroscopically undetected) trace amounts of methoxy groups or free Brønsted acid sites

would exist under reaction conditions, it is safe to assume that their concentration would be

proportional to the concentration of tetrahedrally coordinated aluminum in the zeolite lattice. It

was observed [14], however, that the rate of DME formation decreased as the concentration of

aluminum in the lattice increased.

The second route, i.e., the scavenging of methyl groups from methylammonium ions,

would require that the rate of DME formation would vary in parallel to their concentration. Thus,

a sympathetic variation of the rates of amination and DME formation should be observed. From

literature [14] we clearly see that both conditions are not fulfilled. Thus, we can conclude that

the formation of DME during amination does not proceed via (i) reaction of two methanol

molecules on an empty Brønsted site, (ii) reaction of methanol with methoxy groups formed by

elimination of water from a methanol on a Brønsted acid site or (iii) direct scavenging of methyl

groups from sorbed methylammonium ions by methanol.

The third route, i.e., the condensation of two methanol molecules to DME on weakly

acidic sites, such as the ammonium ions or defect sites remains as the most probable possibility.

Intuitively it is clear that a higher concentration of acid sites (leading to a higher concentration

of alkylammonium ions in the pores) and a higher degree of methylation of the ammonium ions

would diminish the space available for DME formation and reduce the rate.

However, one has also to consider that in a secondary reaction DME can also react with

ammonia to amines, although with lower reaction rate than methanol [31]. In that case the

decrease in DME selectivity upon silylation treatment could be caused by a lower concentration

of acid sites (reducing the rate of formation) or by a longer residence time of DME inside the

pores (favoring the rate of removal through amination). The concentration and degree of

substitution of the sorbed methylammonium ions are directly related to this. A high

concentration of highly substituted methylammonium ions will severely reduce the available

effective pore volume for diffusion. So one would expect a relatively high rate of DME

production for HMOR20 (Si/Al = 10), which has the highest available effective pore volume due

to a lower acid site concentration and relatively low degree of substitution of the

0

0.2

0.4

0.6

0.8

1

0 500 1000 1500 2000

TOS (s)

Re

lativ

e i

soto

pic

fra

ctio

n

0

1

0 50 100


29

Figure 6. Relative concentrations of CH OCH (4), CH OCD ()), CD OCD (2) and3 3 3 3 3 3

CH NH (�), CD NH (x) during isotopic transient experiment (p = 510 Pa, T = 633 K).3 2 3 2 (reactant). 3

methylammonium ions, and the lowest for HMOR10 (Si/Al = 5) and HMOR20-M (silylated

HMOR-20), in accordance with literature. Additionally one could speculate that a higher degree

of substitution of the ammonium ions might hinder favorable coordination of two methanol

molecules. To support the above conclusions, an isotope transient experiment was performed that

allows to relate the residence time of isotopically marked molecules and surface fragments to the

observed overall kinetic results.

3 mg HMOR20 were activated, saturated with ammonia (50 mbar at 360°C for 30

minutes) and exposed to 50 mbar of CH OH for 30 minutes. This treatment caused the acidic3

mordenite sample to be loaded with alkyl ammonium ions [8]. Then, the CH OH containing3

stream was switched to a CD OH/NH stream (50/50 mbar). The results are shown in Figure 6.3 3

In this figure, the relative abundance of CH OCH , CH OCD and CD OCD , as well as CH NH3 3 3 3 3 3 3 2

and CD NH in the effluent gas are plotted versus time on stream. CH NH was present long3 2 3 2

after the switch has been made, whereas CH OCH and CH OCD disappeared after 40 seconds.3 3 3 3

Chapter 2

30

The observation that CH NH was present in the effluent stream more than 30 minutes3 2

after the start consistent with earlier in situ i.r. experiments under comparable conditions [8] in

which methylammonium ions in HMOR were depleted by a stream of 50 mbar ammonia. DME,

however, was exclusively produced as CD OCD already after 40 seconds on stream suggesting3 3

that DME does not use the pool of methyl groups present in the form of methylammonium ions.

This proves that direct scavenging of methyl groups by methanol is not a significant route to

produce DME under the present reaction conditions. It is interesting to note that between 20 and

40 seconds on stream still some CH OCD is formed, although gas phase CH OH was not3 3 3

present. This effect tentatively attributed to the somewhat slower diffusion of DME out of the

mordenite pores of the mordenite, compared to methanol.

2.3.3 The origin of olefins in amination

Two experiments were performed to address the chemical routes to ethene formation, i.e.,

direct elimination of water from ethanol over acid sites and the Hofmann type elimination of

ethene from ethylammonium ions.

In experiment A, 10 mbar of methanol was reacted with 40 mbar MEA and in experiment B, 10

mbar of ethanol was reacted with 40 mbar of monomethylamine (MMA). For both experiments

150 mg HMOR15 was activated at 823 K for 1 h and cooled down to 573K and saturated with

ammonia. For experiment A, 10 mbar methanol and 40 mbar MEA, balanced with He to 1 bar

were passed over the catalyst with a flow of 15.2 ml/min. The methanol conversion was 100%

at 34% MEA conversion. The absence of ethanol as reactant precludes the direct dehydration of

ethanol as ethene source. As Figure 7 shows, a significant amount of ethene is still produced

under these conditions, the rate of ethene formation being 9 10 mol g s . This rate is close to. -8 . -1. -1

the value of 10 mol g s found in ethanol amination when 40 mbar ethanol and 160 mbar-7 . -1. -1

ammonia were reacted over H-MOR15 at 573 K [22].

In experiment B, 10 mbar ethanol and 40 mbar MMA, balanced with He were introduced

at a total flow of 13.4 ml/min. Here the rate of ethene formation was 7.5 10 mol g s , slightly. -8 . -1. -1

lower than in the previous experiment. Note that in this experiment the partial pressure of the

ethyl source was only 10 mbar, while it was 40 mbar in experiment A. The steady state

selectivities for both experiments are depicted in Figure 8.

Selectivity (mol%)

MMA

10%

DMA

6%

ethylene

14%

DEA

26%

EMA

28%

other

4%

DEMA

4%DMEA

8%

Selectivity (mol%)

MEA

12%DEA

2%ethylene

18%

DMA

33%

EMA

25%

DMEA

6%

DEMA

3%other

1%

(a) (b)

0%

20%

40%

60%

80%

100%

0 50 100 150 200 250

TOS (min)

Sel

ectiv

ity (

mol

%)


31

Figure 8. Steady state product distribution of Exp. A (a) and Exp. B (b) over HMOR15 at573K

Figure 7. Reaction of MeOH with MEA (10/40 mbar, 573 K); Selectivity to (�) DEA, (2)MMA, (q) DMA, (�) EMA, (×) DMEA, (+) DEMA and (z) ethene.

From the insensitivity of the system towards the ethyl source we conclude that most of

the ethene produced results from the decomposition of amines (Hofmann elimination) and not

from ethanol dehydration over the acid sites of mordenite. The relative insensitivity of the ethene

formation toward the partial pressure of the ethyl source also indicates that the decomposition

occurs via ethylammonium ions. Their concentration is determined by the Brønsted acid site

Chapter 2

32

concentration of mordenite.

It should be emphasized at this point that although over mordenite Hofmann elimination

is the dominating pathway to ethene formation, this may vary for other solid acids. Ethanol

amination over zeolite Y (Si/Al = 2.7), with a fraction of acid sites that are lower in strength than

those of MOR, showed a high rate of ethene formation. Since (i) the temperature necessary for

complete activation of NH -Y was more than 100 K lower than for mordenite and (ii) low heats4

of adsorption for ammonia and amines are reported for H-Y with a low Si/Al ratio [32,33,34]

one might expect that ammonium ions are less stabilized than in mordenite. As the rate of ethene

formation via direct elimination of water over acid sites is more than two orders of magnitude

higher than the rate of amine formation [34], a small fraction of free Brønsted acid sites can lead

to strong increase in alkene formation. In that context it is also emphasized that the relative

importance of both pathways to form ethene will change as a function of temperature and

reactant pressures.

2.4 Conclusions

It has been shown that the formation of methylammonium ions via reaction of methoxy

groups with ammonia is possible. The rate of formation of methylammonium ions via this route

is similar to the rate of formation via alkylation of ammonium ions. Taking into account the

minimal concentration of free acid sites under reaction conditions the importance of the

amination of metoxy groups for the overall reaction is seen low.

The various transient experiments show that the alkyl scavenging and the adsorption

assisted desorption reaction route both contribute to the overall reaction. The respective

contribution of the two pathways varies with the experimental conditions. In general, this

suggests that alkylamines formed will re-equilibrate during transport through the pores. For

catalyst design it implies that transport restrictions for bulky trialkylamines are necessary to

obtain shape selective catalysis (low substituted alkylamines) and that selective production of

mixed alkylamines (e.g.,ethylmethylamine) is only possible, if the re-equilibration can be

minimized.

The selectivity to DME was concluded to be affiliated with the total available micropore


33

1. A.B. van Geysel and W. Musin, in B. Elvers, S. Hawkins and G. Schultz (Eds.),Ullmann’s Encyclopedia of Industrial Chemistry, 5 ed., VCH, Weinheim, Vol. A16,th

p.535.


A2, p.1.


4. L.D. Pesce and W.R. Jenks, in J.A. Kent (Ed.), Riegel’s Handbook of IndustrialChemistry, 9 ed., Van Nostrand Reinhold, New York, 1992, p.1109.th

5. D.T. Chen, L. Zhang, Chen Yi, and J.A. Dumesic, J. Catal., 146, 257 (1994).

6. I. Mochida, A. Yasutake, H. Fujitsu and K. Takeshita, J. Catal., 82, 313 (1983).

7. F. Fetting and U. Dingerdissen, Chem. Eng. Technol., 15, 202 (1992).


volume for diffusion under reaction conditions. This volume is mainly determined by the

concentration and degree of substitution of the sorbed alkylammonium ions. The formation of

DME from methanol is concluded to be catalyzed by condensation over (alkyl)ammonium ions.

Scavenging of methyl groups from methylammonium ions by methanol is concluded not to be

a significant pathway. In addition to possible spacial problems, formation of DME, similar to the

higher substituted amines, can be reduced by diffusional constraints allowing DME to react to

form amines.

Ethene formation occurs primarily via Hofmann elimination and the influence of the acid

catalyzed direct water elimination from ethanol molecules is insignificant over strongly acidic

catalysts such as H-MOR. Catalysts with weaker acid sites than mordenite can probably also

produce ethene via the acid catalyzed dehydration and would be less suitable for amination of

higher alcohols. Consequently, lowering the reaction temperature will be the most effective

method to increase the overall selectivity to ethylamines for catalysts with strong Brønsted acid

sites.

References

Chapter 2

34

9. D.T. Chen, L. Zhang, J.M. Kobe, Chen Yi and J.A. Dumesic, J. Mol. Catal., 93, 337(1994).



12. A.I. Biaglow, C. Gittleman, R.J. Gorte and R.J. Madon, J. Catal., 129, 88 (1988).

13. J.G. Tittensor, R.J. Gorte and D.M. Chapman, J. Catal., 138, 1 (1992).

14. V.A. Veefkind, Ch. Gründling and J.A. Lercher, J. Mol. Catal., accepted forpublication (1998).

15. T. Kiyoura, K. Terada, EP 681 869 (1994).

16. J.W. Mitchell, K.S. Hayes and E.G. Lutz, Ind. Eng. Chem. Res., 33, 181 (1994).

17. S.R. Blaszkowski and R.A. van Santen, J. Am. Chem. Soc., 118, 5152 (1996).

18. Ch. Gründling, Ph.D. Thesis.


20. D.J. Parillo, A.T. Adamo, G.T. Kokotailo and R.J. Gorte, Appl. Catal., 67, 107(1990).


22. Chapter 3 of this thesis.

23. G.D. Pirngruber, G. Eder-Mirth and J.A. Lercher, J. Phys. Chem. B, 101, 561 (1997).

24. G. Mirth, F. Eder and J.A. Lercher, J. Appl. Spectrosc., 48, 194 (1994).

25. T.F. Narbeshuber, M. Stockenhuber, A. Brait, K. Seshan and J.A. lercher, J. Catal.,160, 183 (1996).

26. T.R. Forrester and R.F. Howe, J. Am. Chem. Soc., 109, 5076 (1987).

27. L. Kubelkova, J. Novakova and K. Nedomova, J. Catal., 124, 441 (1990).

28. A. Kogelbauer and J.A. Lercher, J. Chem. Soc. Faraday Trans., 88, 2283 (1992).

29. M. Niwa, S. Kato, T. Hattori, and Y. Murakami, J. Chem. Soc., Faraday Trans. I., 80,3135 (1984).

30. R.W. Weber, J.C.Q. Fletcher, K.P. Möller, and C.T. O’Connor, Microporous Mater.,7(1), 15 (1996).


32. B.E. Spiewak, B.E. Handy, B. Sharma and J.A. Dumesic, Catal. Lett., 23, 207 (1994).

33. D.J. Parillo and R.J. Gorte, J. Phys. Chem., 97, 8786 (1993).


35

34. D.T. Chen, S.B. Sharma, I. Filimonov and J.A. Dumesic, Catal. Lett., 12, 201 (1992).

Zeolite Catalysts for the Selective Synthesis of Mono- and Diethylamines

37

CHAPTER 3

Zeolite Catalysts for the Selective Synthesis of Mono- andDiethylamines

Abstract

The kinetics and mechanism of ethylamine synthesis from ammonia and ethanol over several acid

catalysts is described. Mordenite produced higher monoethylamine yields than the zeolites beta, Y,

mazzite and amorphous silica-alumina. The reaction proceeds via the initial formation of ethylammonium

ions and alkylamines desorb with the assistance of ammonia and equilibrate with other ethylammonium

ions before leaving the catalyst pores. The high yields of ethylamines with mordenite are related to the

high acid strength of the catalyst stabilizing (alkyl)ammonium ions and blocking so the dehydration of

ethanol. By choosing high ammonia partial pressures, reaction temperatures below 573 K (minimizing

ethene elimination from ethylammonium ions), and subtle modifications of the parent mordenite material

(EDTA leaching, silylation of the external surface) ethene make was further decreased. These measures

allowed to prepare a catalyst on the basis of a mordenite sample with Si/Al = 5 that showed 99 %

selectivity to ethyl amines at 60% conversion and was stable for long times on stream.

Chapter 3

38

3.1 Introduction

Alkylamines are widely used as intermediates in the synthesis of fine chemicals [1,2].

Conventionally alkylamines are produced via reductive amination of aldehydes or alkylation

of ammonia over acidic silica/alumina [1,2]. Zeolites have attracted increasing attention as

catalysts for the latter reaction route [3] as they display strong acidity and pronounced shape

selectivity favoring primary and secondary amines.

Consequently, zeolite catalyzed methylamine synthesis has been extensively studied

[4,5,6,7,8,9,10] including a recent exhaustive review on this subject [3]. One of the most

successful zeolites used for this reaction is mordenite, and a significant number of papers and

patents on the catalysis over mordenite has been published (see ref. [3] and references therein).

Treatment of mordenite with silylating agents has been shown to enhance the selectivity to

methylamines in general and to lower substituted amines (mono- and dimethylamine) in

particular. Such postsynthetic treatment is claimed to deactivate the outer surface of the

mordenite crystallites and to reduce the size of the pore mouth [6,11,12]. Following these lines,

the selectivity enhancement to lower substituted amines can be explained by the inability of the

trimethylamine (TMA) to leave the pores of the modified catalyst, and to enhancement of the

probability that TMA in the mordenite pores disproportionates with ammonia or

monomethylamine (MEA) to dimethylamine (DMA), the desired product [6,12]. A surprising

side effect is the significant decrease in formation of dimethylether (DME), the main side

product, after silylation [12,13].

The direct amination of ethanol to ethylamines has been much less discussed in the

literature, but reports indicate that zeolites Y (FAU), ZSM-5 (MFI), erionite (ERI), mordenite

(MOR) and beta (BEA) [14,15,16] are active catalysts. The biggest challenge in ethyl amine

synthesis is to achieve high selectivities with respect to ethanol use. The formation of

diethylether (DEE) and ethene should be avoided as ammonia and ethene are difficult to separate

(complicating the recirculation) and ethene may deactivate the zeolite via oligomerization.

Formation of ethene from ethanol is thermodynamically favored under most reaction

conditions applied as illustrated in Fig.1. In Fig. 1a the thermodynamic equilibrium between

monoethylamine (MEA), diethylamine (DEA), triethylamine (TEA), ammonia, water and ethanol

is shown starting from 1 mol ethanol and 4 mol ammonia. Formation of amines is favorable,

(b)

0

20

40

60

80

100

550 600 650 700 750

T (K)

Se

lect

ivity

(m

ol%

)

(a)

0

20

40

60

80

100

550 600 650 700 750

T (K)

Se

lect

ivity

(m

ol%

)Zeolite Catalysts for the Selective Synthesis of Mono- and Diethylamines

39

Figure 1. Dependence of the equilibrium product distribution on reaction temperature atNH /EtOH=4 and p=1 bar. (a) without ethene formation and (b) with ethene formation.(4) MEA,3

(�) DEA, (�) TEA, (z) ethene.

although it should be noted that thermodynamics limits the conversion of ethanol to 97 % at 573

K. Fig.1b also shows the thermodynamic equilibrium between the various molecules, but unlike

Fig. 1 permitting now the formation of ethene . One notes that ethanol can be almost completely

converted to ethene at temperatures above 550 K. Thus, ethene formation has to be kinetically

blocked in order to achieve a useful ethanol utilization.

The present contribution aims at providing evidence for the main elementary steps of the

alkylation of ammonia with ethanol using combined in situ i.r. spectroscopy and kinetic

measurements. This knowledge is used to tailor a mordenite based catalyst for achieving

maximum ethylamine yield.

3.2 Experimental

Brønsted acidic mordenites with a Si/Al ratio of 10 (HMOR20), 7.5 (HMOR15), and 5

(HMOR10) were obtained from the Japanese Catalysis Society [17]. HMOR10-E was obtained

by treating a Na-MOR (NaMOR10) with EDTA solution, followed by calcination and ion

exchange with NH NO . The reason for this treatment was the anomalously low pore volume of4 3

0.05 cm /g for HMOR10, which was restored to 0.10 cm /g for HMOR10-E [18]. HMOR20-M,3 3

Chapter 3

40

Catalyst Specific

Area

(m /g)2

Micropor. Vol.

(cm /g)3

Si/Al EFAL

(%)


(mol/g)

HMOR20-M 353 0.15 10 10 1.1·10-3

HMOR20 390 0.21 10 n.d. 1.3·10-3

HMOR15-M 390 0.17 7.5 ~10-15 1.5 10. -3

HMOR15 350 0.15 7.5 11 1.7·10-3

HMOR10-EM 360 0.12 5 ~10-15 1.9 10. -3

HMOR10-E 280 0.10 5 n.d. 2.0·10-3

HMOR10 130 0.05 5 8 2.1·10-3

HMAZ 330 0.12 10 n.d. ~1.2·10-3

HFAU 758 0.34 2.7 n.d. 3.0#10-3

HBEA 514 0.11 11 38 0.7#10-3

n.d. = not detected (<5%)

6CDNG��2J[UKEQ�EJGOKECN�RTQRGTVKGU�QH�VJG�KPXGUVKICVGF�$TÎPUVGF�CEKFKE�OQTFGPKVGU

HMOR15-M and HMOR10-EM were prepared by adding tetraethoxysilane to a suspension of

the activated MOR in n-hexane, followed by intense stirring at room temperature, removal of

the solvent and subsequent calcination in air [12]. BEA with a Si/Al ratio of 11 was obtained

from PQ Corporation [19]. FAU with Si/Al = 2.7 was supplied by Ventron. The silica-alumina

was LA-C25W from Akzo-Nobel (Si/Al =3.5 ).

The physicochemical data for the catalysts used are compiled in Table 1. The BET surface

area and the micropore volume of the catalysts were measured on a Micromeritics Accelerated

Surface Area and Porosimeter (ASAP 2400). The concentration of extra framework aluminum

(EFAL) was determined by Al MAS NMR. The concentration of strong Brønsted acid sites was27

calculated from the gravimetrically determined amount of irreversibly adsorbed ammonia at 373

K.

For adsorption/coadsorption measurements, a Bruker IFS88 FTIR spectrometer was

equipped with a vacuum cell. This high vacuum cell consisted of a stainless steel chamber


41

equipped with CaF windows and a resistance heated furnace in which a gold sample holder was2

placed. To analyze desorbing gasses the system was equipped with a Balzers QMG 420 Mass

spectrometer. The base pressure of the system was 10 mbar. Reactants were introduced to the-6

system with a pressure of 10 mbar, using a dosing valve. Spectra were taken at a resolution of-3

4 cm .-1

Under reactive conditions the i.r. spectra were recorded in situ using a Nicolet 20SXB

FTIR spectrometer in the transmission absorption mode with a resolution of 4 cm . An in situ-1

i.r. cell was used in combination with gas chromatography to allow simultaneous analysis of

sorbed species and products(for details see ref. [20]). Typically 3 mg of the catalyst were pressed

into a self supported wafer and activated in He flow for 1 hr at 823K and then cooled to reaction

temperature.

For measurements at higher conversion a quartz plug flow reactor was used in

combination with a GC equipped with FID and TCD. A Restek RTX amine column was used

for separation. Typical reaction conditions were 573K and 40 mbar ethanol and 160 mbar

ammonia or 100 mbar ethanol and 800 mbar of ammonia, balanced with He to atmospheric

pressure. The specific conditions in each experiment are reported in the Results section.

The chemicals used for these experiments were ethanol (p.a. grade) obtained from Merck,

ammonia gas (Praxair, 99.999% pure) and a mixed gas containing 19.5% NH in He (Praxair,3

99.999% pure).

3.3 Results

3.3.1 Coadsorption of ammonia and ethanol

The formation of surface species under non-reactive conditions was studied by co-

adsorption of ammonia and ethanol at ambient temperature. These coadsorption experiments

were carried out at partial pressures of 1 10 mbar for both components using a differentially. -3

pumped inlet system. First, one compound was adsorbed until full coverage was reached and then

the second was introduced, while maintaining the partial pressure of the first compound. During

coadsorption experiments the same equilibrium state was reached, regardless of the sequence of

adsorption of the two adsorbates. The spectra of ammonia sorbed on HMOR20, ethanol sorbed

3600 3200 2800 2400 2000 1600 Wavenumber (cm-1)

0.2

Abs

orba

nce

(a.u

.)

(a)

(b)

(c)

(d)

Chapter 3

42

Figure 2. I.r. spectra of (a) ethanol; (b) ammonia; and (c) ammonia and ethanol adsorbed onH-MOR20. (d) difference between (c) and (b). p , p = 10 mbar, T=300K.NH3 EtOH

-3

on HMOR20 and their coadsorption are shown in Figure 2. Also shown is the difference between

the spectrum of the coadsorbed species and the spectrum of ammonia sorbed on HMOR20. Upon

sorption of ethanol, bands appeared at 2984 and 2917 cm (attributed to asymmetric and-1

symmetric CH stretching vibrations) and bands at 2939 and 2880 cm (attributed to asymmetric3-1

and symmetric CH stretching vibrations). Additional broad bands at 3500, 3290,2920 and 24002

cm and an intense band between 1900-1400 cm were observed. These bands are attributed to-1 -1

O-H stretching and deformation vibrations from strongly bound methanol molecules

[21,22,23,24]. Sorption of ammonia gave rise to bands at 3378, 3300, 3200, 3050, 2964 and 2800

together with an intense band at 1465 cm , all characteristic for sorbed ammonium ions [25,26].-1

The multiple N-H stretching bands and the slightly asymmetric shape of the 1465 cm band-1

suggest different acid sites or more than one orientation of the ammonium ions with respect to

the zeolite lattice. The shoulder at 3378 cm is assigned to a free N-H stretching vibration from-1

ammonium ions. This shoulder disappears upon coadsorption with ethanol, which can also

0

0.2

0.4

0.6

0.8

1

1.2

1.4

300 400 500 600 700 800 900 1000

Temperature (K)

Sur

face

con

cent

raio

n (m

mol

/g)


43

Figure 3. Surface coverage of MOR20 with (�) ethanol, (4) ammonia and(z) Brønsted acid sites during TPD of coadsorbed species.

clearly be seen in the difference spectrum ( negative band at 3378 cm ). This indicates that there-1

is an interaction The additional band at 3600 cm is assigned to the O-H stretching vibration of-1

coadsorbed ethanol. Its position in the spectrum is similar to that of alcohol O-H bands sorbed

on alkali ion exchanged zeolites [27]. The ammonium N-H deformation band did not decrease

upon ethanol coadsorption indicating that ethanol adsorbs on top of or next to ammonium ions,

but does not replace the ammonia from Brønsted acid sites.

Upon coadsorbing ammonia onto a zeolite pre-equilibrated with ethanol the bands

attributed to ethanol interacting with the Brønsted acid sites of the zeolite disappeared. All bands

attributed to protonated ammonia appeared with exception of the shoulder at 3378 cm attributed-1

to the free N-H band. The spectrum of this coadsorption complex obtained was identical to the

spectrum obtained when starting with the ammonia pre-equilibrated zeolite shown in Fig.2.

Subsequently, the sample was evacuated and the temperature was increased by 10K.min-1

While monitoring the the adsorbates by i.r. spectroscopy and the desorbed products by mass

spectrometry. The results are shown in Figure 3. Below 300 K desorption of ethanol was

observed, followed by a release of ammonia at between 550 and 600 K which was accompanied

by the reappearance of the bands of the zeolite OH groups. This confirms that the ethanol

interacts with the ammonia but not directly with the Brønsted acid site.

(a)

0

0.2

0.4

0.6

0.8

1

550 650 750 850

T (K)

No

rma

lize

d M

S s

ign

al

(a.u

.) (b)

0

0.2

0.4

0.6

0.8

1

550 650 750 850

T (K)

No

rma

lize

d M

S s

ign

al

(a.u

.)

(c)

0

0.2

0.4

0.6

0.8

1

550 650 750 850

T (K)

No

rma

lize

d M

S s

igna

l (a

.u.) (d)

0

0.2

0.4

0.6

0.8

1

550 650 750 850

T (K)

No

rma

lize

d M

S s

igna

l (a

.u.)

Chapter 3

44

Figure 4. Temperature programmed desorption of (a) ammonia, (b) MEA,(c) DEA, and (d) TEA from H-MOR20. Dashed line= ammonia, solid line=ethene.

Temperature programmed desorption (TPD) of ethylamines from Brønsted acidic morde-

nites in all three cases showed a reactive desorption of the amine, i.e., the decomposition of the

amine into ethylene and ammonia at temperatures above 573K as it was also describe for other

alkylamines (see ref.[31]). The corresponding TPD traces are shown in Figure 4. The relative

intensity of the ethylene peak vs the ammonia peak correlated well to the substitution of the

ethylamine. With increasing alkyl substitution the onset of the ammonia desorption shifted to

higher temperatures(from 550 K for NH , 680 K for MEA, 700 K for DEA to 740 K for TEA)3

indicating that the amines first undergo decomposition (Hofmann elimination of ethylene from

the ethylamines) before ammonia desorbs. This is consistent with the higher base strength and

3600 3200 2800 2400 2000 1600 Wavenumber (cm-1)

0.1

Abs

orba

nce

(a.u

.)

(a)

(b)

1435

1608

1490

1400


45

Figure 5. In situ i.r. spectra of (a) ammonium mordenite and (b) catalyst after contacting for30 minutes with 40 mbar ethanol /He (T=573K), showing adsorbed ethylammonium ions.

hence the higher stability of the ammonium ions of the ethylamines compared to ammonia. It

should be emphasized at this point that the amine decomposition limits the potential reaction

temperature as it opens a pathway to ethene formation.

3.3.2 Reaction of ammonium ions with ethanol

In order to investigate the reaction pathway, NH -MOR was contacted with a stream of4

ethanol (40 mbar ethanol, 6 ml/min, 3 mg zeolite) at 573K, while analyzing the zeolite,

adsorbates and reaction products with in situ i.r. spectroscopy. Upon contact with ethanol the

ammonium ions rapidly disappeared and a mixture of ethylammonium ions appeared. Steady

state was reached in less than 20 minutes. Figure 5 shows the spectrum of the ammonia saturated

MOR and this MOR sample after contacting with 40 mbar EtOH for 30 minutes. The bands

attributed to the ethylammonium ions (3005-2875 cm , C-H stretching vibrations; 2850-2400-1

cm , combination bands typical for amines; 1610-1395 cm , N-H and C-H deformation-1 -1

0.0

1.0

2.0

3.0

4.0

373 423 473 523 573 623 673 723

Temperature (K)

Rat

e of

am

ine

form

atio

n (1

0-6 m

ol.g

-1.s

-1)

Chapter 3

46

Figure 6. Rate of MEA formation vs. temperature for different catalysts.(s) HMOR15, (z) BEA, (x) MAZ, (×) FAU.

vibrations) quickly developed. By gas chromatography only ethene and diethylether (DEE) were

detected as reaction product. At longer time on stream also some other hydrocarbons, but no

amines were observed.

After 90 minutes purging with He at the same temperature ammonia was passed over the

loaded catalyst. This resulted in the appearance of amines in the gas phase as detected by

gaschromatography, indicating that the presence of ammonia is indispensable for their catalytic

formation. The rate of amine release form the zeolite pores was 2-3 times lower than the rate of

formation of the methylammonium ions. This suggests that the rate of ammonia aided amine

release is rate determining. MEA was the favored product, TEA was hardly detected which could

be seen as a consequence of the excess of ammonia under these particular reaction conditions.

3.3.3 Amination of ethanol over acid zeolites

When introducing ammonia and ethanol simultaneously to a freshly activated H-MOR

sample, initially high rates of ethene and DEE formation were observed. This was attributed to

a chromatographic effect in which initially ammonia was retained at the entrance of the catalyst

0

0.5

1

1.5

2

2.5

3

3.5

Rat

e (1

0-6 m

ol.g

-1.s

-1)

FA

U

BE

A

MO

R

MA

Z

SiO

2/A

l2O

3

(a)

Amines DEE Ethene

0

20

40

60

80

100

Sel

ectiv

ity (

mol

%)

FA

U

BE

A

MO

R

MA

Z

SiO

2/A

l2O

3

(b)

MEA DEA TEA


47

Figure 7. (a) Rate of product formation and (b) amine selectivity over various catalysts at 15-21% conversion and 573K.

bed, giving the ethanol the chance to react with the free Brønsted acid sites present further in the

catalyst bed. This was largely prevented by presaturating the catalyst with ammonia. Therefore

all experiments reported here were performed after presaturation with ammonia.

Fig. 6, 7 and 8 compare the results for the amination of ethanol. In Fig. 6 the rate to

amines are plotted for each catalyst as a function of temperature. The experiments were

performed with 30 mg catalyst and a feed of 40 mbar ethanol and 160 mbar ammonia in He

(total flow 10 ml/min.). In Fig. 7, the rates to amines, DEE and ethylene at 573K are plotted and

in Figure 8 the selectivities to the different amines at 573K. These experiments were performed

with a feed of 40 mbar ethanol and 160 mbar ammonia, at a total flow of approx. 15 ml/min, but

the catalyst weight was varied to achieve a similar conversion over all catalysts of approx. 18%.

The overall rate of reaction was the highest over FAU, but the MOR sample showed the highest

rate to amines in the temperature region between 553K and 633K and the highest selectivity at

573K. The selectivity to ethylene was highest over FAU, while the selectivity to DEE was

highest over BEA and SiO /Al O . From the selectivities towards the different amines, it can be2 2 3

seen that FAU had the highest selectivity to TEA and the zeolites with a one-dimensional 12-ring

system (MOR and MAZ) the lowest, with BEA assuming an intermediate position. SiO /Al O2 2 3

exhibited a relatively low TEA selectivity. The MOR samples clearly showed the highest MEA

Chapter 3

48

Exp Catalyst T (K) reactantratio *

conv.(%) amineselect. (%)

amine distribution(%/%/%)

etherselect.(%)

1 HMOR20 573 1 E/ 2 A 5 65 89/10/1 10

2 HMOR20 623 1 E/ 2 A 25 40 83.16/1 5

3 HMOR15 573 1 E/ 2 A 8 74 84/13/3 8

4 HMOR10 573 1 E/ 2 A 9 61 87/12/1 5

5 HMOR10-E 573 1 E/ 2 A 15 58 81/16/3 10

6 HMOR15 573 1 E/ 4 A 11 84 82/15/3 3

7 HMOR15-M 573 1 E/ 4 A 9 93 85/15/0 0

8 HMOR15-M 573 1 E/ 4 A 32 92 78/21/1 0

*E = ethanol, A = ammonia

Table 2. Influence of various reaction parameters on conversion and selectivity in ethanolamination.

yield of all catalysts tested.

Table 2 compiles the results for ethanol amination over different mordenite catalysts. All

experiments were performed over 25-30 mg of catalyst. The experiments 1-5 were performed

with p = 25 mbar, p = 50 mbar at a total GHSV of 13000 h . Experiments 6-7 wereNH3 EtOH-1

performed at p =40 mbar, p = 160 mbar at a total GHSV of 8000 h and experiment 8 wasNH3 EtOH-1

performed with the same partial pressures of the reactants but at a GHSV of 2500 h . From-1

experiments 1 and 2 show that higher reaction temperatures (expectedly) increased catalyst

activity, but decreased the selectivity to amines, mainly due to higher ethene make. Experiments

1 and 3-5 show that for the mordenite based catalysts, the activities of the catalysts increased in

the order HMOR20< HMOR15< HMOR10< HMOR10-E, i.e., parallel to the concentration of

available acid sites. HMOR10, which had an anomalously low BET surface area and micropore

volume, was less active than the EDTA treated sample (HMOR10-E). From experiments 3 and

6 it can be seen that increasing the ammonia/ethanol (N/R) ratio from 2 to 4 increased ethanol

conversion and amine selectivity. The modified, i.e., silylated, catalyst showed slightly lower

activity, but amine selectivity increased from 84 to 93%. DEE formation was negligible when

using HMOR15-M and TEA formation was also severely reduced (compare experiments 6 and

7). Higher conversions, achieved by lowering the GHSV, influenced the amine distribution, but

0.0 0.6 1.2 1.8

Rate (10-6 mol.g-1.s-1)

HMOR20, 300°C,40/160 mbar E/A

HMOR15, 285°C,100/800 mbar E/A

SEHMOR10, 285°C,100/800 mbar E/A

Ca

taly

st +

con

ditio

ns

Ethylamines

Ethylene


49

Figure 8. Optimization of MEA yield by changing catalyst, partial pressures andtemperature.

do not influence the overall amine selectivity of the catalysts (compare experiments 7 and 8).

To explore the influence of the ammonia/ethanol ratio, two experiments were performed in

which the ammonia partial pressure was varied relative to a constant ethanol pressure of 50 mbar

at 573K, and one in which it was varied relative to an ethanol pressure of 100 mbar. From these

experiments an apparent reaction order of 0.94 in ammonia was obtained.

Optimization of the reaction temperature, the NH /ethanol ratio and the catalyst allowed3

to drastically increase the yield to ethylamines at the expense of ethene formation. The results

of this optimization are compiled in Figure 8. Decreasing the reaction temperature enhanced the

ethylamine selectivity via a drastic decrease in the rate of ethene formation. The consequential

loss in amine yield was compensated by increasing the ammonia partial pressure. As high acid

site concentrations and silylation the outer surface of the zeolites were seen to be beneficial to

the reaction rate and the amine selectivity (see also refs. [12,28]), a material with the maximum

acid site concentration (for the MOR materials this was the EDTA treated HMOR10) was used

for these modifications. With this strategy SE-HMOR10 was prepared showing a 99%

selectivity at 60% conversion (250 mg SE-HMOR10, 11 ml/min. 100/800 mbar EtOH/NH ,3

558K).

TOS (h)

0

20

40

60

80

100

Se

lect

ivity

(%

)

0 20 40 60 80

0

20

40

60

80

100

Co

nve

rsio

n (

%)

0 50 100 150 200 250 300

TOS (min)

0

20

40

60

80

100

Se

lect

ivity

(%

)

Chapter 3

50

Figure 9. Stability of HMOR10-EM (558K, 100/800 mbar EtOH/NH ). (4) MEA,3

(�) DEA, (�) TEA, (z) ethene, and (×) conversion.

Figure 10. Influence of water addition (at t=120 min., 558K, 100/800 mbarEtOH/NH ). (4) MEA, (�) DEA, (�) TEA, and (z) total amine selectivity.3


51

In general, the mordenite catalysts appeared white after reaction and within the time scale

of the experiments (typically 3-4 hours), an appreciable deactivation was not observed. In order

to test the long term stability of these catalysts, an amination experiment over SE-HMOR10

catalyst (100/800 mbar EtOH/NH , 558K, 0.3 kg amines.kg .h produced at 60% ethanol3 catalyst-1 -1

conversion) was performed for 75 hrs. The conversion and selectivity as a function of time on

stream are depicted in Figure 9. It can be seen that under these conditions the high selectivity was

sustained at stable conversion. The amount of amines produced in this period was approx. 25 kg

amines/ kg catalyst and the catalyst was still white after use. After 4 hours of time on stream (1.2

kg amines/ kg catalyst produced), the FAU and BEA sample typically showed a slight grey color.

As water is one of the reaction products, the influence on the catalyst performance was

tested. This was done by adding 33 mbar of water to the feed of a catalyst at stable operation. As

can be seen from Fig. 10, the influence of water was negligible.

3.4 Discussion

3.4.1 On the mechanism of ethanol amination

As described above, coadsorption of ethanol and ammonia leads to the same coadsorption

complex between the ammonium ion and ethanol regardless of the sequence of introduction of

the sorbate. The loss of the 3378 cm band, assigned to a free N-H stretching vibration [27], the-1

appearance of a band at 3600 cm , similar to alcohol adsorption on alkali ions, and the fact that-1

no ammonia is displaced upon introduction of ethanol strongly indicate a coadsorption complex

in which ethanol coordinates to NH as depicted in Figure 11. Additionally, TPD after4+

coadsorption (see Fig. 3) shows that the release of ethanol does not affect the acid sites, but

restitutes the free N-H stretching band. The bands of the acidic hydroxyl groups of the zeolite are

only restored after desorption of ammonia. Note that the proposed coadsorption complex is

analogous to the coadsorption complex observed with methanol and ammonia [29,30]. The

interaction between the molecules in that complex is rather weak. Consequently, it is not

detectable with i.r. spectroscopy under reaction conditions.

As ammonium ions and weakly adsorbed ethanol suffice to produce alkylammonium

ions, we have proposed that under reactive conditions the alkylation of the ammonium ion

Chapter 3

52

Figure 11. Coadsorption complex of ethanol and ammonia on acidic catalyst.

involves protonation of the alcohol by the ammonium ion, followed by an immediate release of

water and the simultaneous formation of a C-N bond (for details see ref. [29]). The

ethylammonium ions resulting from the alkylation of the ammonium ions are very stable, in line

with the high base strength of the substituted amines compared to ammonia [31,32]. During the

period that the catalyst was exposed to helium or ethanol/ helium, amines were not observed in

the gas phase. Additional ammonia was needed to remove amines from the surface. This second

step, the ammonia mediated release of amines into the gas phase, is slower than the their

formation, i.e., the initial alkylation of the ammonium ion is fast and the desorption of the

formed alkylamine is rate determining. Also the fact that the amination of ethanol is found to be

practically first order in ammonia strongly supports a mechanism in which the ammonia

mediated release of amines is rate determining.

The results of the transient experiments for ethanol amination described above are in

excellent agreement with those for methanol amination [27]. Also there the rate of release of the

amines from the surface was much lower than their initial rate of formation and reaction orders

of approx. 1 (ammonia) and 0 (methanol) have been found [9].

For the ammonia mediated release of the amines two types of mechanisms have been

proposed: (i) an alkyl scavenging type reaction in which a gas phase ammonia molecule removes

an alkyl group from a sorbed alkylammonium ion yielding a alkylamine in the gas phase and a


53

Scavenging: [R NH ] + NH (g) â [R NH ] + RNH (g)x y ads 3 x-1 y+1 ads 2+ +

A.a.d.: [R NH ] + NH (g) â [NH ] + R NH (g) where y=4-xx y ads 3 4 ads x y-1+ +

Scheme I. Mechanisms for ammonia mediated release of amines from acidic catalyst.

sorbed ammonium ion depleted of one alkyl group and (ii) adsorption assisted desorption (a.a.d.)

in which an incoming ammonia or alkylamine deprotonates the chemisorbed alkylamine which

desorbs. Both possible pathways are represented in Scheme I and are discussed in refs. [13,27].

3.4.2 Differences between the zeolites

Of the different zeolites investigated, mordenite gave the highest yield to amines and the

rates to ethene and diethylether were low compared to the other zeolites tested. This is attributed

to the high acidic strength of the mordenite at high acid site concentrations, compared to the

other zeolites, which helps to keep the acid sites covered with ammonium ions. This prevents

direct elimination of water from ethanol yielding ethene. The much lower rate of DEE formation

over MOR is explained by the small pores of MOR and low external acid site concentration

disallowing the spatially demanding reaction of two ethanol molecules over a weak acid site.

Both subjects will be discussed in detail below.

FAU showed the highest ethanol conversion, but also the by far highest yield to ethene.

As it contains the highest concentration of tetrahedrally coordinated aluminum (Brønsted acid

sites) and in consequence the highest concentration of Al in next nearest neighborhood to another

tetrahedrally coordinated Al, its acid site are among the weakest of the materials studied [33,34].

This lower acid strength is reflected in the fact that complete removal of ammonia is achieved

at temperatures lower than 673 K (compared to 823 K for mordenite). As a consequence

ammonia is less stabilized and some free acid sites may be available that catalyze direct ethanol

dehydration.

The other zeolites tested had a higher Si/Al ratio and, from the temperature needed for complete

ammonia removal (723- 823 K), a higher acid strength. The rate of formation of ethene was

rather similar for these zeolites. MOR, however, has the highest acid site concentration and,

therefore, the highest rate to amines.

Chapter 3

54

Figure 12. Zeolite structures, emphasizing the diameter of the 12-ring; (a) BEA,(b) FAU, (c) MOR, and (d) MAZ.

All the investigated zeolites have pores with a minimum opening of 12-membered rings.

FAU has a three-dimensional channel system connecting the supercages, BEA has a one-

dimensional 12-ring channel system and a two dimensional 12-ring channel system. MOR and

MAZ have a one dimensional 12 ring system and a one-dimensional 8-ring system too small for

molecules larger than MEA. For a schematic representation of the molecular sieve channels see

Fig 12. The amine product distribution seems to be related to the dimensionality of the 12-ring

system. Selectivity to TEA decreased from 23% over FAU, to 10% over BEA and 1-2% over

MAZ and MOR. It is speculated that due to the pore structure in FAU with a three dimensional

channel system with rings with an aperture of 0.74 nm and its even larger supercages, diffusion

of TEA out of the zeolite particle is relatively facile, i.e., the routes to the outside of the


55

crystallite should be short, numerous and with little obstructions. In BEA, due to the channel

structure outlined above, it is more difficult for TEA to exit, but transport will be most difficult

in MOR and MAZ since these zeolites have only a one-dimensional large pore system.

Interesting to note is the significantly lower rate of DEE formation over MOR, compared

to the other catalysts. Mordenite has the smallest 12-ring of the samples investigated. It has been

shown previously [13] that the formation of ethers proceeds under those circumstances in a

concerted reaction between two ethanol molecules over ammonium and alkylammonium ions

and depends, thus, on the available pore volume during reaction. Under reaction conditions, the

sorbed amines decrease this available volume drastically. The effect of this will be most

pronounced in mordenite, since it has the smallest pore diameter. A large external surface area

(pronounced meso and macroporosity) will consequently be unfavorable for low DEE selectivity

In that context we would like to point to the similar performance of amorphous silica/alumina

and BEA which both have a similarly large specific external surface area.

3.4.3 Options to enhance selectivity

Reaction conditions

In Table 2 it can be seen that upon increasing the temperature from 573 to 623 K the

selectivity did not change dramatically. Selectivity to MEA decreased from 89% to 83% at the

favor of DEA, which went from 10 to 16%, while TEA selectivity stayed low. Conversion

increased from 5 to 25 %. As DEA is a secondary product from MEA an increase in DEA

selectivity at the expense of MEA can be expected at increasing conversion. The thermodynamic

equilibrium composition (as shown in Fig. 1) is slowly shifted from DEA to MEA at increasing

temperatures, but this is not expected to have a significant influence since (i) changes over the

depicted temperature range (200 K) are not very large, (ii) conversions in the experiments shown

are not high enough to approach equilibrium and (iii) shape selectivity is imposed.

The reaction temperature has a more profound influence on the selectivity to amines than

on the amine distribution. Table 2 also shows clearly that the experiment performed at 623 K had

a significant lower selectivity to amines than the experiment at 573 K. With all acid sites covered

by alkylammonium ions under these conditions this suggests that the decomposition of the

ethylammonium ions, as found in the TPD experiments, poses a problem for ethanol amination

Chapter 3

56

as above 573K. The rates of this reaction seem to approach the rates for the ammonia mediated

desorption pathways and consequently decrease the selectivity to amines. It should be noted that

in the absence of ammonia this decomposition is quantitiative and has been been used to

determine acid site concentrations [31]. As a consequence all reactions discussed were therefore

carried out at or below 573K.

In order to maximize the selectivity to amines, not only formation of ethylene via

Hoffmann elimination has to be suppressed, but also the elimination of water from ethanol,

leading to ether or ethylene, has to be minimized. Experiments with ethanol as single feed

indicate that the rate of ethene formation is more than two orders of magnitude higher than the

rate of amine formation (>10 mol.g .s vs. 10 mol.g .s ). Thus, only a small fraction of free-4 -1 -1 -6 -1 -1

acid sites suffices to catalyze the conversion of a significant fraction of ethanol to ethene. This

can be minimized by (i) high ratios of ammonia to ethanol (compare exp. 3 and 6 in Table 2), (ii)

a strongly acidic zeolite (high stability of the ammonium ions, compare results for mordenite and

zeolite Y) and (iii) low reaction temperatures (compare experiments 1 and 2 in Table 2). In the

latter case three factors help to suppress ethene formation, i.e., the thermodynamic equilibrium

favors ethanol at lower temperatures (see Fig 1b), the relative adsorption constant of ammonia

is higher and formation of ethylene via Hoffmann elimination is minimized.

Another route to ethene could be the decomposition of DEE. If formation of ethene from

DEE was much faster than Hoffman elimination, i.e., a major pathway, then DEE and ethene

selectivities should go more or less in parallel. However, when comparing DEE and ethene

formation for example over MOR and BEA or over MOR in exp. 3 and 4 of Table 2, they show

to be NOT in parallel. This would indicate that formation of ethene from DEE is not a major

pathway in amination of ethanol.

Silylation of the outer surface of the mordenite crystallites

As outlined before [6,11,12,13], silylation of the outer surface of a molecular sieve

generates a thin silica layer blanketing the acid sites of the outer surface and decreasing the

diameter of the pore mouth opening. With mordenite this has been shown to be very effective

for enhancing the selectivity to methylamines in general and to monomethylamine, specifically.

Thus, one might expect that the treatment has a similar positive effect upon ethylamine synthesis.

0

1

2

3

4

5

6

7

8

0 0.5 1 1.5 2 2.5

Concentration acid sites (10-3 mol.g-1)

Ra

te to

ME

A (

10-7

mo

l.g-1.s

-1)

HMOR20

HMOR15

HMOR10-E

HMOR10

HMOR15-M

HMOR10-EM


57

Figure 13. Influence of catalyst treatment on rate of MEA formation.

Table 2 (comparing experiments 6 and 7) shows that silylation drastically reduces the

ether formation. The selectivity to amines in general and the monosubstituted product in specific

was significantly enhanced. Despite a slightly lower rate of total alcohol conversion, this results

in an increase of the rate of mono-alkylamine formation as illustrated in Fig. 13. There, it can be

seen that the formation of MEA (primary product) is approximately first order in the acid site

concentration for the non-silylated mordenite series. Silylation of the outer surface, however,

slightly enhances the rate of formation of MEA. The decreased pore mouth diameter hinders the

higher substituted amines to leave the pores generating so a higher degree of alkylation of the

alkylammonium ions. In methanol amination these were found to be more reactive than lower

substituted ammonium ions [12,13]. The smaller pore mouth could also enhance the

concentration of reactants in the zeolite and thus thereby enhancing the reaction rate.

Also the use of ethanol is more efficient since diethyl ether is not formed over the

silylated catalyst. As discussed above, formation of ethers under amination conditions has been

found to proceed mainly over the sorbed ammonium ions [13]. As silylation causes a decrease

in the size of the pore mouth rather and the elimination of external acid sites, it exerts two

Chapter 3

58

positive effects. The narrowing of the pore opening effectively increases the substitution of the

alkylammonium ions in the pores and decreases the internal available pore volume for the

concerted bimolecular DEE formation. On the other hand it also blocks strong acid sites on the

outer surface and suppresses so formation of alkylammonium ions that may also catalyze DEE

formation.

3.4.4 Deactivation

Two factors are important in the prevention of coking. (i) The suppression of olefin

formation and (ii) the prevention of olefin oligomerization to form stable species inside the

zeolite. The stability of the SE-HMOR10 catalyst is very good. No appreciable decrease in

activity and selectivity after 75 hours and no discoloration of the catalyst after use, were

observed. We observed some formation of olefinic species by i.r. spectroscopy, but no

pronounced coking or deactivation. The two reasons for the high stability of especially the

modified mordenite are proposed to be (i) the overall strength of the acid sites, and (ii) the pore

structure.

Coke precursor formation is greatly impeded by stabilization the ammonium ions due to the

strength of the acid sites. This effect is enhanced by the relatively large concentration of higher

ethylammonium ions present inside the pores of the silylated material and the one-dimensional

nature of the mordenite pores. These higher ethylammonium ions are even more strongly

adsorbed than ammonium ions, providing a very effective shielding of the acid sites from (i)

ethanol and (ii) the small amounts of ethylene produced, thereby inhibiting the formation of coke

precursors.

Mordenite has often been described as being very susceptible to coke formation [35],

mainly due to the one dimensional nature of the pore system in which a relatively small number

of obstructions can render a large portion of the pore system unaccessible. Our results, however,

indicate that modified mordenite is one of the most stable catalysts for amination reactions. The

formation of larger, stable coke species can be inhibited by the geometrical constraints due to the

pore structure. This space being even more confined in the case of the modified mordenite due

to the increased substitution of the ammonium ions at the acid sites.

FAU has much weaker acid sites and selectivity to olefins is high. This fulfills one of the


59

conditions for coke formation: the presence of precursors in the form of olefinic species. The

oligomerization process can also be facilitated by the higher availability of free acid sites. The

specific shape of the FAU structure with its large supercages facilitates the formation of larger,

more complex molecules which can easily get trapped within the zeolite.

3.5 Conclusions

Amination of ethanol over acidic catalysts has been shown to proceed in two steps. The

first step consists of the alkylation of ammonium ions by ethanol and the second, rate

determining, step is the ammonia mediated release of the formed amines (sorbed as

alkylammonium ions) into the gas phase. The precursor to this reaction is proposed to be a

coadsorption complex in which ethanol coordinates with the oxygen to the ammonium ion on

the acid site. After the formation of this complex, protonation of the ethanol from the ammonium

ion occurs, followed by release of water and the formation of a C-N bond. These steps are similar

to those reported for methanol amination [12,29,27].

As the ethyl group is a better leaving group than the methyl group, ethene formation from

ethanol, diethylether and the ethylammonium ammonium ion is facile. Formation of ethene is,

thus, the main problem in solid acid catalyzed ethanol amination. It can be suppressed by

applying high ammonia partial pressures, low reaction temperatures, and by choice of a proper

catalyst, i.e., a molecular sieve that has narrow pores and stabilizes (alkyl)ammonium ions well.

The influence of the ammonia concentration is twofold: (i) ethanol amination has a

positive order in p(NH ) of 0.9 for the formation of amines and (ii) high ammonia pressures will3

minimize direct exposure of ethanol to acid sites leading to ethene formation.

Low reaction temperatures are necessary as TPD of alkylamines shows that above 573K

ethylamines start to decompose at an appreciable rate to ethene and ammonia. For a given

ammonia partial pressure lower temperatures help to maintain the acid sites covered with

ammonium ions preventing direct access of ethanol or DEE.

The comparison of the acid catalysts studied shows conclusively, how the catalyst

properties influence the overall amine selectivity and the selectivity within the amines.With

respect to the first point the chemical composition and the pore geometry are found to be the

Chapter 3

60


A2, p.1.

crucial parameters. Zeolites with a low Si/Al ratio which are weaker solid acids and do not

stabilize ethylammonium ions well show a higher tendency to form ethene. This is attributed to

the presence of free acid sites which catalyze formation ethanol or DEE dehydration. It should

be noted at this point, however, that the rate of amination sympathetically varies with the

concentration of acid sites and that an optimum Si/Al ratio exist as a consequence. This is

exemplified clearly by the low amine yield found with FAU (Si/Al = 2.7)having the lowest yields

and rates in ethylamines and the high yields found with SE-HMOR10 (Si/Al = 5). In that context,

the temperature needed for complete ammonia removal during activation of the sample, as

observed with FTIR, seems to be a good indication to whether the chosen catalyst will show a

good amine selectivity. If complete ammonia removal takes place below 723 K, the catalyst will

likely show high ethene selectvities.

The pore structure of the zeolite has a definite influence on amine selectivity and amine

distribution. If the acid sites are in a relatively confined space, as in MOR which had the smallest

channels of the zeolites investigated, the available space for ether formation over weakly acidic

centers will be lower and consequently the rate to DEE will decrease. The rate to TEA is greatly

influenced by the diffusional constraints imposed on the products. FAU, with its open an 3-

dimensional pore system had the highest TEA selectivity (23%), whereas MOR and MAZ with

their one-dimensional pore system produced much less TMA (1-2%).

The most promising way of enhancing MEA yield is silylation of the outer surface of the

mordenite. Although the overall conversion drops slightly, selectivity to byproducts (especially

diethylether) decreased and the MEA yield increased. This is attributed to a decrease in available

pore volume for the bimolecular alcohol dehydration reaction caused by increasing substitution

of the alkylammonium ions and the elimination of accessible Brønsted acid sites or

alkylammonium ions on the outer surface of the molecular sieve crystals.

References


61



4. Y. Ashina, T. Fujita, M. Fukatsu, K. Niwa, and Y. Yagi, Stud. Surf. Sci. Catal., 84,28, 779 (1986).

5. R.D. Shannon, M. Keane, Jr, L. Abrams, R.H. Staley, T.E. Gier, D.R. Corbin, andG.C. Sonnichsen, J. Catal., 113, 367 (1988).

6. K. Segawa and H. Tachibana, J. Catal., 131, 482 (1991).

7. M.C. Ilao, H. Yamamoto, and K. Segawa, J. Catal., 161, 20 (1996).

8. U. Dingerdissen, E. Nagy, and F. Fetting, Chem.-Ing. Techn., 63, 625 (1991).

9. D.T. Chen, L. Zhang, J.M. Kobe, C. Yi, and J.A. Dumesic, J. Molec. Catal., 93, 337(1994).

10. Y. Ashina and M. Fukatsu, US 4485261 (1984).

11. M. Niwa, S. Kato, T. Hattori, and Y. Murakami, J. Chem. Soc., Faraday Trans I, 80,3135 (1984).

12. Ch. Gründling, G. Eder-Mirth and J. A. Lercher, J. Catal., 160(2), 299 (1996).

13. V.A. Veefkind, Ch. Gründling and J.A. Lercher, J. Mol. Catal. A, in press (1998);Chapter 2 of this thesis.


15. M. Deeba, M. E. Ford and T. A. Johnson, in D. W. Blackburn (Ed.), Catalysis ofOrganic Reactions, Marcel Dekker Inc. New York (1990).


17. M. Sawa, M. Niwa, and Y. Murikami, Zeolites, 10, 532 (1990).

18. G.D. Pirngruber, G. Eder-Mirth and J.A. Lercher, J. Phys. Chem., 101(4), 561 (1997).

19. G.S. Nivarthy, Y. He, K. Seshan, and J.A. Lercher, J. Catal., in press (1998).

20. G. Mirth, F. Eder, and J. A. Lercher, Appl. Spectrosc., 48, 194 (1994).

21. S.R. Blaszkowski and R.A. van Santen, J. Phys. Chem., 99, 11728 (1995).

22. A. Zecchina and C. Otero Arean, Catal. Rev., 35(2), 261 (1993).

23. A. Thursfield and M.W. Anderson, J. Phys. Chem., 100, 6698 (1996).

24. G. Mirth, J.A. Lercher, M.W. Anderson, J. Klinowski, J. Chem. Soc, Faraday Trans.,86, 3039 (1990).

25. L.H. Little, Infrared Spectra of Adsorbed Species, Academic press, London, 1966,p.180.

Chapter 3

62

26. A. Kogelbauer, J.A. Lercher, K.H. Steinberg, F. Roessner, A. Soellner, R.V.Dimitriev, J. Chem. Soc. Faraday Trans., 88, 2283 (1992).


28. K. Segawa and H. Tachibana, in Guczi et al. (Eds.), Proc. 10 ICC, Elsevier,th

Amsterdam, 1993, p. 1273.

29. A. Kogelbauer, Ch. Gründling and J.A. Lercher, J. Phys. Chem., 100 (5), 1852 (1996).


31. D.J. Parillo, A.T. Adamo, G.T. Kokatailo and R.J. Gorte, Appl. Catal., 67, 107(1990).

32. D.J. Parillo, R.J. Gorte and W.E. Farneth, J. Am. Chem. Soc., 115, 12441 (1993).


34. D.J. Parillo and R.J. Gorte, J. Phys. Chem., 97, 8786, (1993).

35. P. Magnoux, P. Cartraud, S. Mignard and M. Guisnet, J. Catal, 106, 242 (1987).

Role of Strength and Location of Brønsted Acid Sites for Ethylamine Synthesis on Mordenite

63

CHAPTER 4

Role of Strength and Location of Brønsted Acid Sites forEthylamine Synthesis on Mordenite Catalysts

Abstract

The elementary steps of alkylation of ammonia with ethanol and the role of the strength and location

of the acid sites of the mordenite catalysts were investigated. Both reactants have access to all Brønsted

acid sites in acid treated mordenites. However, only monoethylamine accesses all acid sites, di- and

triethylamine sorbed only 60 and 45 % of the sites, respectively.

Partial H / Na ion exchanged mordenites have Brønsted acid sites primarily in the main channel+ +

up to an ion exchange degree of 60 % H , only at higher H exchange degrees the sites in the side pockets+ +

also are exchanged. These latter acid sites contribute significantly to the overall rate of amination. At low

H exchange degree, the acid sites seem to weaker and producing more ethene (weaker stabilization of+

the ammonium ions). Alkylation/ de-alkylation experiments show that diethylether cannot enter the

mordenite channel side pockets to alkylate ammonium ions. With diethylether, alkylation did not proceed

to triethylammonium ions, but stopped at diethylammonium ions.

Chapter 4

64

4.1 Introduction

Alkylamines are widely used as intermediates and solvents in the production of fine

chemicals. Products that are derived from alkylamines include a wide range of agrochemicals,

pharmaceuticals, rubber accelerators, solvents, anti-corrosion agents and even explosives and pet

food supplements [1,2,3,4]. Their synthesis via the shape-selective zeolite catalyzed alkylation

of ammonia with alcohols has attracted significant attention [5]. This interest lies in the potential

to replace silica-alumina with catalysts able to produce preferentially mono- and di-alkylamines.

Most studies focus on methylamines being the largest volume chemical and the only amine that

could be produced with very high yields [2,4]. Recently, also shape selective synthesis of

ethylamines over zeolite catalysts has been reported in some detail [6].

Under reaction conditions the acid sites are covered with (alkyl)ammonium ions [7,8].

In situ i.r. studies show that the reaction proceeds in two well separated steps, i.e., (i) the alkylation

of the ammonium ion by the alcohol and (ii) the ammonia mediated release of the amines into

the gas phase. The first step is relatively fast and, hence, the second step determines the overall

[7]. Consequently, the pores are filled with chemisorbed alkylamines, i.e., alkylammonium ions.

Using in situ i.r. spectroscopy, it has been shown that the substitution of methylammonium ions

is influenced by the pore geometry. Silylation of the outside of mordenite crystallites, which leads

to a decrease in the average pore mouth diameter [9,10], significantly increases the methyl

substitution on the ammonium ions. This is related to the inability of trimethylamine to exit the

pores of the modified zeolite [10]. The rate of monomethylamine formation depends linearly on

the total concentration of methyl groups in the pores [11]. The high average substitution of the

methylammonium ions therefore increases the rate of methanol amination while the inability of

trimethylamine to exit the zeolite pores leads to high selectivity to lower substituted methylamines.

Amination of ethanol was found to be mechanistically similar to methanol amination [6].

The tendency of the ethyl group toward elimination leads to pronounced ethene formation. Ethene

can be produced via (i) direct water elimination from ethanol or diethylether over acidic sites or

(ii) Hofmann elimination of ethene from ethylammonium ions. Low reaction temperatures and

very strong sorption of ammonium ions were found to be key parameters to reduce ethene formation.

In that respect, mordenite was shown to be a very effective catalyst for monoethylamine synthesis

due to the very strong bonding of ammonia. Because of the larger size of the alkyl group, the


65

Figure 1. MOR structure and accessibility for ammonia (dark region).

amination of ethanol is expected to be more sensitive towards the steric environment of the acid

site than the methanol amination. The location of the acid sites may, thus, be of higher importance.

Among the zeolites seen suitable for shape selective alkylamine synthesis, mordenite is

industrially the most widespread applied [5,12,13]. Mordenite consists of a one dimensional system

of large channels (12-ring, 6.5 x 7.0 Å) lined with so-called side pockets, which have an aperture

of approx. 4.8 x 3.7 Å [14,15], these side pockets are separated by a restriction of 2.6 x 5.7 Å,

as shown in Fig.1 in which the accessibility of the MOR structure to ammonia is represented

(obtained using the Insight II program from Biosym/MSI). It has been shown [16] that even small

linear alkanes do not enter the side pockets and only sorb in the main channels. It is unclear at

present whether the reasons are entropic (the interaction with an acid site in the side pockets or

with the walls of the pocket do not compensate for the loss in entropy caused by severe restriction

of movement in the side pocket) or enthalpic (the secondary C atom cannot reach the hydroxyl

Chapter 4

66

group) in nature. More basic molecules, like ammonia, amines, and alcohols sorb on all acid sites,

provided their size allows it. To differentiate the role of the acid sites in these two location, we

will present a detailed qualitative and quantitative analysis.

To measure the concentration and strength of acid sites in zeolites the most common

techniques employed include thermogravimetric analysis, often combined with microcalorimetry

[17,18,19,20], temperature programmed desorption [21,22] and infrared spectroscopy [15,23,24,25,].

All techniques rely on the adsorption of base molecules. Intuitively, it is obvious to see that the

choice of probe molecule is very important. Combination of various probe molecules and techniques

has been seen to be a very effective strategy. Using that approach several groups have reported

that mordenite is one of the most acidic zeolites, and in relation to other frequently used molecular

sieves a sequence H-FAU < H-ZSM-5 < H-MOR [25,26].

Using ammonia as probe that is able to reach all acid sites and pyridine, cyclohexane and

benzene as probe that can only attach to sites in the main channels, the location of acid sites was

quantitatively evaluated [22,24,27,28,29,30]. It is interesting to note that the acid sites in mordenite

appear to be homogeneous when pyridine is used as probe molecules (TG-DSC measurements),

even with a combination of ammonia and pyridine differences in the acid site strength were not

apparent [17,18]. Recent measurements seem to show, however, differences in the acid strength

[31]. Using TPD, Karge et al. [22] showed that a significant fraction of the Brønsted acid sites

that can be reached by ammonia, is inaccessible for pyridine. The same authors [32,33] obtained

similar results using i.r. spectroscopy and since then several i.r. studies on the inhomogeneity of

Brønsted acid sites in mordenite have appeared [27,28,29,30].

The i.r. spectrum of mordenite shows typically a single band at approximately 3612 cm ,-1

attributed to strongly acidic hydroxyl groups. Unlike FAU which has two distinct bands for acid

sites in supercages and sodalite cages, one cannot differentiate between acid sites in the side pocket

and the main channel, except for a slight asymmetry of the SiOHAl band. Adsorption of benzene,

cyclohexane and pyridine shows that a large fraction of the acid site interacts with the probe

molecules, but always the same fraction of the OH band ( low frequency (LF) contribution at approx.

3590 cm ) remains unaffected[27,28,29,30,34]. In general, it is now well accepted the mordenite-1

hydroxyl band consists of a LF part at ca. 3590 cm corresponding to acid sites in the side pockets-1

and a HF band at ca. 3612 cm corresponding to acid sites in the main channels. While it is possible-1


67

to resolve the quantitative aspects by curve fit programs, the large number of fitting options lead

to controversial interpretations as well illustrated by Makarova et al. [29].

Based on ammonia TPD, Zholobenko et al. [27] concluded that the side pockets have

a higher “effective acidity”, i.e., that ammonia is stronger adsorbed on OH groups in the side pockets

than in the main channel. The authors are careful to point out that this not necessarily means a

higher ‘intrinsic acidity’. Maache et al. [28] compared adsorption of CO on samples with and

without poisoning of main channel sites by pyridine and showed that the HF band leads to a 70

cm larger redshift upon interaction with CO than the LF band. This indicates that the HF hydroxyl-1

groups have a higher acid strength than LF hydroxyl groups.

Recently, it has been shown [34] that upon exchange from NaMOR to HMOR (after

activation) the protons are introduced first in the main channels and only later in the side pockets.

The introduction of H exclusively in the main channels could provide a very useful tool to assess+

the influence of the acid site position on a catalytic reaction. It is the aim of this paper to rigorously

combine structural data regarding siting of protons in mordenite with mechanistic and kinetic data

for the amination of alcohols to obtain a microkinetic model for this type of alkylation reactions.

In situ i.r. spectroscopy and kinetic measurements are the primary experimental means.

4.2 Experimental

4.2.1 Material

The mordenites used in this study were the Japanese Reference Catalysts (JRC) MOR20

(Si/Al=10) and MOR15 (Si/Al=7.5). The concentration of acid sites for the 100% hydrogen forms

was determined with ammonia sorption (gravimetrically) to be 1.25 mmol/g and 1.65 mmol/g,

respectively. The percentage of Al in extraframework positions for MOR20 was found to be less

than 5%, for MOR15 this was 11% [35]. Additionally, a sample from Tosoh, MOR18.5 (Si/Al=9.3)

was used. The area of the Brønsted acid site band (between 3645 and 3500 cm ) of the H-MOR18.5-1

was found to be 106% of the area of the H-MOR20 Brønsted acid site band, i.e. within the error

margin of the predicted 7% increase of acid sites caused by the differences in Si/Al ratio for the

two mordenite samples.

All exchanged samples, regardless the starting material or exchanged ions, will be referred

Chapter 4

68

to as x % exchanged (unless specifically mentioned otherwise), in which x is the percentage of

Brønsted acid sites relative to the fully exchanged H-MOR parent, being either MOR20 or

MOR18.5. Unless otherwise mentioned, no distinction between samples originating from the two

parent materials is made, except for this normalization.

For Na Ú NH exchange, 2 g NaMOR were added to 100 ml ammonium nitrate solution,+ +4

stirred at 353K for 2 hours, filtrated and washed repeatedly with deionized water and dried at 393K

for 24 hours. The ammonium nitrate concentration was varied between 0 and 1.0 M (38 fold excess

of NH in solution, relative to amount of Na present in zeolite). Some samples were exchanged4+ +

for 24 hours instead of 2 hours to check if the distribution of exchanged ions was in equilibrium

after 2 hours of exchange.

Another set of ion exchanged samples was prepared by NH Ú Na exchange, in which4+ +

2 g NH MOR were added to 100 ml sodium hydroxide solution, stirred at 373K for 2 hours, filtrated4

and washed repeatedly with deionized water, after which the sample was dried at 393K for 24

hours. The NaOH concentration was varied between 0 and 0.1 M (3.8 fold excess). A number

of NH MOR samples were exchanged with NaNO instead of NaOH.4 3

4.2.2 Characterization

To follow activation and exchange degree, a Bruker IFS88 spectrometer was used, equipped

with a vacuum cell (base pressure 10 mbar), as described in [36]. All samples were pressed into-6

self supported wafers and in situ evacuated under a dynamic vacuum of 10 mbar, heated to 823K-6

with 10K/ min., kept at 823K for 60 min. and cooled down to 373K. Spectra were typically taken

during heating at 533K, after 1 hour at 823K and at 373K. In Fig. 2 some typical spectra are

represented, with the most common sorbed species. Typical bands are attributed in Table 1. All

spectra were normalized to the 1860 cm lattice vibration overtone. The area of the Brønsted acid-1

site band of the normalized spectra was used to calculate exchange degrees. The same approach

was used to follow the sorption of the reactants and products in HMOR20. After the activation,

as described above, a partial pressure of 10 mbar monoethylamine (MEA), diethylamine (DEA),-3

triethylamine (TEA), ethanol or ammonia was applied to the system and the uptake of these

compound was followed by i.r. spectroscopy. After reaching equilibrium, the partial pressure of

3600 3200 2800 2400 2000 1600 Wavenumber (cm-1)

0.1

Abs

orba

nce

(a.u

.)

(a)

(b)

(c)

(d)


69

Figure 2. I.r. spectra of MOR in activated and non-activated state. (a) NH (100%),3

(b) NH3 + H O (23%), (c) H O (0%) and (d) H-MOR (100%).2 2

Feature wavenumber (cm )-1 See Fig.2

MOR lattice vibration overtonesMOR silanol groupsMOR � of acid sitesOH

1970, 1860, 165537403610

All curvesAll curvesH-MOR

NH �3 NH

NH 3 NH

3375, 3260, 3040, 28151450

NH , NH + H O3 3 2

NH , NH + H O3 3 2

H O �2 OH

H O 2 OH

3638, 3585, 3455, 32701635

H O, NH + H O2 3 2

H O, NH + H O2 3 2

Table 1. I.r. bands and assignments typical of MOR in activated and non-activated state.

the adsorbate was increased to 10 , 10 and 1 mbar until equilibrium was reached at these pressures.-2 -1

The spectra of these sorbed amines at 673 K were used as references for a multicomponent

fit routine. In this multicomponent fit, spectra obtained with in-situ i.r. spectroscopy under reaction

conditions (673 K, 40 mbar EtOH/ 160 mbar NH in flowing He, 10 ml/min. total flow), could3

3600 3200 2800 2400 2000 1600 Wavenumber (cm-1)

0.1

Abs

orba

nce

(a.u

.)

(a)

(b)

(c)

Chapter 4

70

Figure 3. I.r. spectra of sorbed ethylamines on MOR; (a) triethylamine; (b) diethylamine; and(c) monoethylamine.

be fitted with reference spectra in the 1700 -1350 cm range to obtain quantitative information-1

on the surface concentrations of the sorbed species. For the spectra of the working catalyst under

reaction conditions an i.r. reactor with a volume of 1.5 cm was used, which approximates a3

continuously stirred tank reactor [37]. The effluent gas stream was stored in 16 loops of a multi

loop valve and subsequently analyzed by a HP 5890 GC, equipped with FID. This allows us to

simultaneously analyze the composition of the adsorbed phase and the gas phase.

All reaction experiments with the partially exchanged samples were performed in a quartz

plug flow reactor with an internal diameter of 4 mm. The volume of liquid ethanol was controlled

by a Cole Parmer syringe pump with Hamilton 250µl syringe, this ethanol was vaporized at 373K

into the gas stream. Products were analyzed on-line on a HP 5890 GC, equipped with FID, using

a Restek Rtx-Amine capillary column.

100 mg of catalyst were activated in situ under He flow (10K/min. to 823K, dwell 1 hour)


71

Compound Wavenumber (cm )-1 Assignment

NH3 3378, 3285, 3200, 3050, 2960, 281516201460

N-H stretch NH4+

NH at Lewis center3

N-H deformation NH4+

MEA 3353, 3250, 31752987-28892799, 2705, 2568, 24671600, 1551, 14891475 (shoulder)1456, 1398

N-H stretch -NH3+

C-H stretchovertones/ combination bandsN-H deformation -NH3

+

C-H deformation CH 2

C-H deformation CH3

DEA 31922984 (as), 29462912 (as), 28462789, 2481-2281160714701450, 1392

N-H stretch NH2+

C-H stretchC-H stretchovertones/ combination bandsN-H deformation -NH2

+

C-H deformation CH2as, sym C-H deformation CH3

TEA 3170, 30522985, 2948, 28892810, 2755, 2697(2520, 2488)14761459, 1393

N-H stretch -NH +

C-H stretchovertones/combination bandsovertones/combination bandsC-H deformation CH 2

as, sym C-H deformation CH3

Ethanol 2984, 29172939, 28803500, 3290, 2920, 240014501395

as, sym C-H stretch CH3as, sym C-H stretch CH2OH-stretch + resonance bands [6]CH deformation2

CH deformation3

Table 2. I.r. bands and assignments for the sorption of ethanol, ammonia and ethylamines.

and cooled to reaction temperature (558K). Then the catalyst was saturated with 800 mbar ammonia

in He (10 ml/min.) and the reaction mixture was passed over the reactor (100 mbar EtOH, 800

mbar NH , 300 mbar He at 10 ml/min. total flow).3

TPD experiments were performed to examine the stability of the NH MOR sites. For these4

experiments 100 mg MOR was evacuated to 10 mbar in a dynamic vacuum, heated to 423K with-6

10K/min. and degassed at this temperature for 1 hour. Then the sample was heated to 1053K with

Chapter 4

72

10K/min. under dynamic vacuum and the gas phase was sampled with a Balzers QMG 420 MS

detector.

4.2.3 Chemicals

Ammonia was supplied by Praxair, The Netherlands, as a pure gas (99.999%) and as 19.5%

NH in helium (99.999%). monoethylamine was also supplied by Praxair as 99.99% pure gas in3

a lecture bottle. Diethylamine, triethylamine and ethanol were obtained p.a. grade from Merck.

Sodium nitrate, sodium hydroxide and ammonium nitrate were supplied p.a. grade by Merck.

4.3 Results

4.3.1 Sorption of reactants and products

The accessibility of the acid sites in HMOR20 for reactants and products was probed via

adsorption at room temperature, monitored by i.r. spectroscopy. The i.r. spectra during the

equilibration of HMOR20 with 10 mbar of MEA, DEA and TEA at 308K are depicted in Figure 3.-3

Note that these are difference spectra, i.e., the spectrum of the parent zeolite has been subtracted

from those obtained after sorption of a particular compound. The main spectral features are

summarized in Table 2. As this table provides a detailed assignment for the observed bands, only

the most important features of these spectra will be discussed here. Sorption of ammonia [6,7]

and ethanol on HMOR20 has been described in detail elsewhere [6] and the spectra are not shown

in Fig. 3.

All amines are sorbed in the protonated form, as can be seen from the N-H stretch vibrations

(3400- 2800 cm ), the typical combination bands (2800- 2300 cm ) and the deformation bands-1 -1

(1620- 1500 cm ). These bands are also observed in salts of these amines [38,39]. For DEA and-1

TEA complete coverage of the Brønsted acid sites was not reached, as the band at 3610 cm was-1

still visible although with strongly diminished intensity. The fractional coverages of the strongly

acidic hydroxyl groups with MEA, DEA and TEA at 1 mbar were 1, 0.6 and 0.45, respectively.

This indicates that the larger amines cannot access part of the MOR hydroxyl groups.

0%

20%

40%

60%

80%

100%

0 1 2 3 4

NH4/Na ratio

Exc

hang

e de

gree

0%

20%

40%

60%

80%

100%

0 1 2 3 4 5

Na/NH4 ratio

Exc

ha

ng

e d

eg

ree

0%

50%

100%

0 25 50

0%

50%

100%

0 25 50


73

Figure 4. Exchange degree as function of initial cation ratio (solution/zeolite) for(a) Na Ú NH exchange and (b) NH Ú Na exchange.+ + + +

4 4

4.3.2 Na/H exchange

In Figure 4a the exchange degree is plotted versus the initial ratio NH /Na4 (solution) (zeolite)+ +

for the Na Ú NH exchange of the mordenite and in Fig. 4b against the ratio Na /NH+ + + +4 (solution) 4 (zeolite)

3580

3590

3600

3610

3620

3630

0% 20% 40% 60% 80% 100%

Exchange degree

Wav

enum

ber

(cm

-1)Chapter 4

74

for the NH Ú Na exchange of the zeolite. Here, the exchange efficiency is defined as the fraction4+ +

of cations originating from the solution that is exchanged into the zeolite to obtain a certain exchange

degree for this cation. Normally this value corresponds to the thermodynamic equilibrium. For

the Na Ú NH exchange this exchange efficiency approximates unity until 65-70%, then the+ +4

efficiency drops markedly. Even with a 38 fold excess of ammonium ions, a 100% exchanged

sample could not be reached (see inset in Fig. 4a). This illustrates the necessity of multiple ion

exchanges to reach a fully exchanged zeolite. For the NH Ú Na exchange, initially NaNO4 3+ +

was used but the equilibrium is not favorable (points at Na /NH = 38, see inset in Fig. 4b). The+ +4

other exchanges were performed with NaOH. The efficiency is much higher, presumably because

ammonium ions from the zeolite in the basic conditions can react with OH and remove NH from- +4

the solution and keep the equilibrium favorable for exchange. The danger is that sodium silicates

can be formed, which could explain why the exchange efficiency is not equal to 1. Some of the

samples in Fig. 4 were exchanged for 24 hours instead of 2 hours, but this had no influence on

exchange degree indicating that equilibrium had been reached.

The wavenumber of the band of the Brønsted acid sites is plotted against the exchange

degree in Fig. 5a. The open symbols correspond to wavenumbers at 823K (after 1 hour of activation)

and the solid symbols correspond to the band positions at 373K (after this activation). Results

from the Na Ú NH exchange (squares) and the NH Ú Na exchange (triangles) are depicted.+ + + +4 4

373K

823 K

Figure 5. (a) Position of Brønsted acid site band as function of exchange degree; (b) Spectra of

hydroxyl region of various exchanged mordenites.

Nor

mal

ized

MS

sig

nal (

a.u.

)

0

0.002

0.004

0.006

0.008

0.01

0.012

0.014

350 550 750 950 1150

T (K)

25%

50%

100%

0

20

40

60

80

100

0% 20% 40% 60% 80% 100%

Exchange degree

Are

a of

143

0 cm

-1 b

and

(% o

f fre

sh

NH

4MO

R)


75

Note that this figure shows the need to compare i.r. data of spectra obtained at equal temperature,

as the differences between spectra taken at 373 K and 823 K were approximately 15 cm . The-1

position of the maximum of the Brønsted acid site band varies clearly with the exchange degree.

At low H content the HF contribution dominates, at higher degrees there is also a LF contribution+

to the acid site band. The figure also demonstrates that the wavenumber of this band is independent

of the exchange route, as points obtained from Na ÚNH and from NH Ú Na are on the same+ + + +4 4

line. The variation of the wavenumber of the acid site band as a function of exchange degree is

demonstrated in Fig. 5b. Again, at low H exchange degrees the protons give rise to a HF � and+OH

only at higher exchange degrees a LF contribution appears.

Temperature programmed desorption (TPD) of ammonia from some partially exchanged

samples are compiled in Fig. 6a. It can be seen that the maximum in the desorption rate shifts

from 773 K for the 100% exchanged samples to 713 K for both the 50% and the 25% exchanged

samples. The fact that with the 25% and 50% exchanged samples the TPD maximum appears at

the same (lower) temperature indicates that the stability of the ammonium ions is constant and

lower for samples up to 50 % exchange than the stability of the ammonium ions in MOR with

Figure 6. (a) TPD curves for MOR20 of different exchange degree. (b) Area of 1430 cm band-1

at 573K, as percentage of area of 1430 cm band 100% NH4MOR at 373K, vs. exchange degree.-1

0

20

40

60

80

100

Sel

ectiv

ity (

mol

%)

0 5 10 15 20 25 30 35 40

Time (min)

Chapter 4

76

Figure 7. Alkylation of ammonium ions by 40 mbar EtOH, 573K; (z) NH , (4) MEA,3

(�) DEA, (�) TEA.

higher H exchange degrees. In another set of experiments the desorption of ammonia (from the+

freshly exchanged catalyst) at 573 K was compared for a number of exchanged MOR20 samples.

The decrease in the 1430 cm band was taken as measure of the fraction ammonia desorbed. The-1

results are shown in Figure 6b. In this figure the area of the remaining 1430 cm band, as a-1

percentage of the area of the 1430 cm band of a 100% NH MOR20 before heating, is plotted-14

vs. the exchange degree. A sample that would retain all ammonium ions at 573 K would fall on

the diagonal of this plot. As can be seen there are two regions in this plot, (i) a region from 0 to

ca. 60% exchange, with a slope < 1 indicating these samples lose a fraction of their ammonium

ions at 573 K, and (ii) a region from ca. 60 to 100% with a slope of 1, indicating that the ammonium

ions that are introduced at higher exchange degree do not desorb at 573 K.

4.3.3 Alkylation of ammonium ions in MOR

First, HMOR15 was exposed to a 10 ml/min. flow of 16% NH in He at 573 K to convert3

all Brønsted acidic hydroxyl groups to ammonium ions. Then, the flow was switched to a 10 ml/min.

flow of 4% EtOH in He and the alkylation of the ammonium ions was followed by i.r. spectroscopy.

The resulting spectra were quantitatively evaluated using a multicomponent fit and the variations

0

20

40

60

80

100

Sel

ectiv

ity (

mol

%)

0 10 20 30 40 50 60

Time (min)


77

Figure 8. Reaction of ethylammonium ions (from EtOH) with 160 mbar ammonia, 573K;(z) NH , (4) MEA, (�) DEA, (�) TEA.3

of reactants and products in the pores is displayed in Figure 7.

The fraction of ammonium ions rapidly decreased to 15 mol%. Ethylammonium ions were

formed in sequential order according to their degree of substitution. The concentration of MEA

passed a maximum after 6 minutes on stream, DEA reached this maximum after 8 minutes. This

pattern is typical for a series of sequential reactions proceeding from ammonia, through MEA

and DEA to TEA. The reaction did, however, not proceed until completion, i.e., to 100% TEA.

In a consecutive experiment, 10 ml/min. 16% NH in He was passed over the catalyst loaded3

with the alkylammmonium ions. The quantitative evaluation of the varying concentrations of sorbate

is compiled in in Figure 8. MEA reacted in approximately 10 minutes, but the rate of removal

of tri- and especially diethylammonium ions was considerably lower.

A similar set of experiments was done with diethylether as a sterically more hindered

alkylating agent. H-MOR was transformed into the ammonium form in situ. Subsequently, a stream

of 10 ml/min. 4% diethylether (DEE) in He was passed over the zeolite. The quantitative evaluation

of the i.r. spectra recorded during this procedure is shown in Fig. 9. The concentration of ammonium

ions in the pores decreased, but slower than with ethanol as alkylating agent. The concentration

ammonium ions reached a steady state of approximately 44% of the original concentration. MMA

0

20

40

60

80

100

Sel

ectiv

ity (

mol

%)

0 50 100 150 200

Time (min)

0%

50%

0 20 40

0

20

40

60

80

100

0 10 20 30 40 50 60 70

Time (min)

Sel

ectiv

ity (

mol

%)

Chapter 4

78

Figure 10. Reaction of ethylammonium ions (from DEE) with 160 mbar ammonia,573K; (z) NH , (4) MEA, (�) DEA, (�) TEA.3

Figure 9. Alkylation of ammonium ions by 40 mbar DEE, 573 K; (z) NH , (4) MEA,3

(�) DEA, (�) TEA.

0

2

4

6

8

10

0% 20% 40% 60% 80% 100%

Exchange degree

Rat

e (1

0-7 m

ol.g

-1.s

-1)


79

Figure 11. Rate of amination reactions as function of exchange degree (100 mbar EtOH,800 mbar ammonia, 573 K); (4) MEA, (�) DEA, (�) TEA, (z) ethene, and (�) amines.

was initially rapidly formed, but after 10 minutes began to be consumed for the make of

diethylammonium ions. Interestingly, triethylammonium ions were not formed inside the pores.

When passing 10 ml/min. of 16% NH in He over the catalyst, the concentration of MEA decreased3

rapidly, DEA significantly slower (see Fig. 10.)

4.3.4 Acitivity and selectivity of partially H exhanged MOR+

The rates of alkylamine synthesis as function of the exchange degree of the catalyst, under

the conditions mentioned in the Experimental section, are shown in Figure 11. It is obvious that

upon introduction of Brønsted acid sites into the side pockets, the reaction rate increases. This

indicates that the accessibility of the ammonium ions, and the formed monomethylammonium

ions does not limit amine formation. The selectivity to ethene decreased, the DEA selectivity

increased (see Figure 12 ) with the H exchange degree. Note, however, that all reactions were+

performed with the same space velocity and that, therefore, the conversion increases from left

to right in this plot.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

35% 100%

Exchange degree

Con

cent

ratio

n (m

mol

/g)

NH3 MEA DEA TEA

0

20

40

60

80

100S

elec

tivity

(m

ol%

)

0% 20% 40% 60% 80% 100%

Exchange degree

Chapter 4

80

Figure 13. Sorbed species in amination of ethanolover 35% exchanged and 100% exchanged MOR (40mbar EtOH, 160 mbar ammonia, 573K).

Figure 12. Selectivity of amination reactions as function of exchange degree (100mbar EtOH, 800 mbar ammonia, 573 K); (4) MEA, (�) DEA, (�) TEA, (z) ethene.

Finally, the composition of the

sorbed amines in the pores was

determined under reaction conditions

(40 mbar EtOH, 160 mbar NH , T= 5733

K) using in situ i.r. spectroscopy. This

was done for a 35% exchanged catalyst

and for a 100% HMOR18.5. The steady

state concentrations of the adsorbed

species in the pores of these two samples

is given in Fig. 13. Since the acid site

concentration of the 35% is much lower

than that of the 100% exchanged

sample, it will consequently contain

much less sorbed (alkyl)ammonium

ions. The 35% exchanged sample


81

contains TEA, DEA and ammonium ions in approximately equal concentrations. The concentration

MEA in the pores is negligible. In the 100% exchanged sample, 60% of the sorbed species are

ammonium ions, MEA, DEA and TEA constitute 6%, 9% and 24% of the sorbed species. Note

that the proportion of NH ions in the 100% exchanged sample is much higher, but that the fraction4+

sorbed MEA remains small.

4.4 Discussion

4.4.1 Accessibility of acid sites in mordenite

All amines sorbed on H-MOR are protonated. Ethanol, ammonia and MEA are able to

cover all acid sites at a partial pressure of 10 mbar, while DEA and TEA accessed progressively-3

less acid sites. This difference agrees well with the size of these molecules and was also reported

elsewhere for sorption of different substances in mordenite, which yielded 33-50% inaccessible

acid sites for bulky molecules[27,28,30]. Note that such observation also is in agreement with

structural calculations ([40], and references therein) which revealed that between 40-45% of the

Al atoms in MOR are located deep in the side pockets of the mordenite.

This inaccessibility of part of the acid sites for larger molecules is well reflected in the

alkylation experiments. In the alkylation of ammonium ions with diethylether (DEE), ca. 40%

of the ammonium ions were not alkylated. Note that this percentage of unreacted ammonium ions

corresponds to the 40% acid sites not accessible for DEA and to the 40-45% of Al atoms calculated

to be located in the side pockets [40]. Therefore, we conclude that DEE cannot alkylate ammonium

ions located in the side pockets, either because of its own size or the size of the transition state

of the reaction.

Approximately 15% of the sorbed ammonium ions were not alkylated by ethanol. We

attribute this to the fact that ammonium ions deep inside the side pockets are blocked by formation

of mono- or diethylammonium ions in the main channel (at the entrance of these side pockets)

and cannot be accessed by the alkylating ethanol

4.4.2 Na/H exchange

The exchange between Na and H in MOR proceeded with an exchange efficiency very+ +

NH�

4 ; NH3 � H � logpNH3

(NH�

4 ) 11.02�pH

NH4OH ; NH3 � H2O logpNH3

(NH4OH) 1.75

NH�

4 � H2O : NH4OH � H � log(NH4OH)

(NH�

4 ) 9.27 � pH

Chapter 4

82

(I)

(II)

(III)

close to unity, until an exchange degree of 60-70%. The last 30-40% of the sodium ions are more

difficult to exchange. These are the sodium ions that after activation of the mordenite are found

to populate the side pockets. It is likely that these sodium ions are more stabilized in the side pockets

due to the close vicinity of the negatively charged framework oxygens. The equilibrium for

exchange from ammonium to sodium using NaNO is not favorable. However, by using NaOH,3

this equilibrium could be shifted in favor of the intended exchange. The concentration ammonium

ions in basic aqueous solutions is determined by the solubility of ammonia according to equation I

[41].

The pH of the NaOH solutions was in the range 11-13. Therefore, the equilibrium of this reaction

lies to the right. Also the formation of undissociated NH OH has to be taken into account. This4

is defined according to equation II [41].

Substituting both equations we obtain equation III [41], which indicates that under basic conditions

the concentration ammonium ions in solution is kept low and, therefore, the equilibrium for ion

exchange shifted in the desired direction.

However, also when using NaOH the exchange efficiency was not equal to unity. This may be

due to the formation of some sodium silicates.

The exchange degree influences the position of the Brønsted acid site band. At low Na+

exchange degrees, the band appeared at higher wavenumber and as the exchange degree increased

it shifted to lower wavenumbers indicating heterogeneity of the acid sites. In accordance with

the literature, we attribute the higher wavenumber band (3620 cm ) to Brønsted acid hydroxyl-1

groups in the main channels, the lower wavenumber band (3590 cm ) to Brønsted acid sites in-1


83

the side pockets [27,28,29,30]. As outlined in the introduction, different ratios for acid sites in

the main channels (mc) vs. acid sites in the side pockets (sp) have been reported. Most authors

claim a 1:1 or 2:1 mc/sp ratio. Fitting of the acid site band has been attempted using various

methods and band shapes for the two components of this band (see ref. [29], and references therein),

but consensus has not been reached yet about the exact shape of the bands. Sorption of ethylamines

mentioned earlier indicate a 60:40 ratio mc/sp, which is close to the values reported by others

[27,28,29,34,40]. The 40/60 ratio reported by Eder et al. [16] has been derived from hydrocarbon

sorption on the same material and this could suggest that the more polar amines protrude further

into the side pockets than the apolar linear alkanes. Note that this would suggest Brønsted acid

sites at the entrance of the side pockets, which agrees nicely with the model proposed for blocking

the further remote sites in the side pockets during alkylamine synthesis as discussed above.

The exchange route, e.g., from sodium to ammonium or from ammonium to sodium, did

not affect the position of the Brønsted acid site band of the activated mordenite. This indicates

that either the sodium ions preferentially occupy the side pockets during exchange, or a

redistribution takes place upon activation at 823 K.

TPD of ammonia for the 25% and the 50% exchanged samples, i.e., with ammonium ions in the

mc, showed maxima at 713 K, for the 100% exchanged samples these are shifted to 773K, a

difference of 60K. This indicates that the ammonium ions in the main channels are less stabilized

than those in the side pockets. This was confirmed in the experiments in which the decrease of

the 1430 cm deformation band of NH at 573 K was measured for MOR20 samples of varying-1 +4

exchange degree. Again, the samples with an exchange degree <60% could not retain all of their

ammonium ions, while the ammonium ions introduced at higher exchange degrees were stable.

This clearly confirmed that the ammonium ions that are introduced first, and are located in the

mc, are thermally less stabe than those introduced at higher exchange degree, which are located

in the sp.

4.4.3 The role of the alkylating agent

Via a pseudo-first-order rate law, rate constants for the formation of methylammonium

ions were found of 6.410 s mbar for alkylation by ethanol and 2.110 s mbar for alkylation. -5 -1 -1 . -5 -1 -1

by DEE, indicating that ethanol is a better alkylation agent than DEE. The formation of an

Chapter 4

84

alkylammonium ion at the acid sites in the zeolite pores involves protonation of an alcohol molecule

by an ammonium ion [7,42,43], followed by a nucleophilic attack at the carbon atom by the

ammonia and the release of water. Intuitively, it can be seen that the transition state complex will

be much larger with DEE than the transition state complex with ethanol. Indeed, with DEE as

alkylation agent TEA was not formed inside the pores. Thus, we propose that DEE, with the

reactive oxygen atom in the middle of the molecule, has steric difficulties to approach the protons

on the sorbed diethylammonium ions in the main channels of the mordenite.

It can be argued that the alkylation of ammonium ions by DEE produces ethanol which,

unlike DEE, is able to alkylate ammonium ions in the side pockets and produce triethylammonium

ions in the main channels. However, the conversion of DEE is below 1% and since the rate of

alkylation by ethanol is only 3 times faster than the alkylation by DEE, the amount of acid sites

being alkylated by ethanol formed in the course of this reaction, is very small. Additionally, the

rate constant for the alkylation of MEA was calculated to be 1.8 10 s mbar , which is a factor. -6 -1 -1

10 slower than the alkylation of ammonium ions. This means that after the formation of MEA

in the main channels, the rate of ethanol formation decreases even further and its concentration

in the pores will be negligible compared to the DEE concentration. Therefore, it was concluded

that the influence of this secondary alkylation by ethanol was negligible compared to the alkylation

by DEE under the present experimental conditions.

4.4.4 Implications for ethanol amination under reactive conditions

Most strikingly, we observed that the rate of alkylamine synthesis increased not strictly

proportional to the concentration of Brønsted acid site in the partial ion axchanged MOR (see

Figure 12).We note a higher specific contribution of the Brønsted acid sites located in the side

pockets to the overall rate of alkylamine formation. We think that this is not primarily related to

the higher acid strength of these hydroxyl groups, but rather an effect of the varying composition

and reactivity of the alkylammonium ions in the MOR pores.

To support this suggestions let us consider the individual reaction steps. The rate constant

for the removal of MEA by ammonia was 2.4 10 s mbar . Under reactive conditions with an. -5 -1 -1

excess of ammonia, this implies that the rate of formation of monomethylammonium ions (6.410. -5

s mbar ) is of the same order of magnitude than the rate of removal by ammonia. As a consequence-1 -1


85

the concentration of sorbed MEA in the pores should be quite low. Indeed, this was observed in

the experiment in the in situ reactor. With the 35% exchanged sample MEA was not observed,

with the 100% hydrogen form of MOR, we observed approximately 6% MEA. The rate of removal

of DEA and TEA were significantly lower than that of MEA. Therefore, to have a rather low

concentrations of the multiple substituted alkylamonium ions should be beneficial for the overall

rate of amine formation. As molecular modeling suggests that the side pockets do not allow the

formation of amines higher than monoethylamine, the contribution of the side pockets to the overall

rate is significant. At exchange degrees higher than 60%, the side pockets get populated with acid

sites and the rate of amination increases from 410 mol g s to 8 10 mol g s , i.e., it doubles.. -7 -1 -1 . -7 -1 -1

The turn over frequency (TOF) for MEA formation for the lower exchanged samples was calculated

to be 510 s , for the additional sites this was ca. 50% higher at 7.4 10 s , indicating the latter. -4 -1 . -4 -1

sites are more reactive in amination of ethanol.

Both in the 35% exchanged sample and in the 100% exchanged sample there is a fraction

of sites covered with ammonium ions. This could, in part, be due to blocking of active sites,

especially in the 100% exchanged sample. However, in the 35% exchanged sample the acid sites

are mainly located in the main channels and, therefore, accessible to ethanol. It is more likely that

the large contribution of ammonium ions is a consequence of the fact that the alkylation of NH4+

ions and the removal of monoethylammonium ions proceed at comparable rate. This also explains

why MEA was not observed in the 35% sample and only 6% in the 100% exchanged sample. The

large increase in the percentage NH in the 100% exchanged sample can be explained by the fact4+

that the on the sites in the site pockets DEA cannot be formed. As mentioned earlier, the MEA

that is formed reacts quite rapidly, leaving a large percentage of the sites covered with NH .4+

Therefore, the large concentration of ammonium ions in the 100% exchanged sample reflects the

increase of the concentration of acid sites in the side pockets.

When we compare the selectivity to ethene over the various exchanged samples, we observe

that the selectivity to ethene is highest at low exchange degrees. We also observe in the temperature

programmed desorption of ammonia that the acid sites introduced into the mordenite at low

exchange degrees retain ammonia to a lesser extent than the sites introduced at higher exchange

degrees. This is in agreement with the results of Zholobenko et al. [27] who claim a higher ‘effective

acidity’ for acid sites in the side pockets. Thus, we conclude that the formation of ethanol is easier

Chapter 4

86

over sites that provide less stabilization of ammonium ions. This agrees very well with earlier

reports [6], which claim that strongly acidic sites are necessary in ethanol amination to keep

ammonium ions from desorbing spontaneously at reaction temperature, thus preventing the

formation of free acid sites available for the acid catalyzed direct elimination of water from ethanol.

This direct dehydration reaction is orders of magnitudes faster than the amination reactions,

implying that only a very small fraction of acid sites suffices to explain the levels of ethene

4.5 Conclusions

Upon sorption of ammonia and amines, the zeolite proton attaches to the sorbate leading

to the formation of ammonium ions. These ions can be formed on the acid sites only, if the amines

reach at least the vicinity of the acid sites, i.e., the proton cannot be spatially be separated from

the oxygen of the hydroxyl group via remote interactions. Though accessible to MEA, not all acid

sites were found to be accessible to DEA and TEA. The acid site coverage was 60% for DEA

and 45% for TEA. The differences in accessibility are attributed to the location of the acid sites

in either main channels or side pockets of the mordenite.

The unperturbed acidic hydroxyl groups located in the main channel and in the side pockets

of mordenite were shown to be distinguishable with i.r. spectroscopy, in agreement with results

reported [27,28,29,34,40]. Partially exchanged NaHMOR samples of an exchange degree up to

60% were found to have the protons located in the main channel, regardless of exchange route

or duration of exchange. This indicates that the siting of the H and Na is thermodynamically+ +

determined. The acid sites that are introduced at low exchange degree desorb ammonia at lower

temperature than those added at higher exchange degree.

Alkylation of ammonium ions over acid catalysts is possible with ethanol and DEE,

although alkylation with DEE is three times slower. With DEE only mono and diethylammmonium

ions are formed, which is attributed to a larger transition state than in the case of the alcohol. Since

DEE cannot reach Brønsted acid sites in the side pockets, acid sites located there are not utilized.

Ethanol is able to alkylate most of the ammonium ions in the side pockets. Due to the higher rate

of alkylation and the availability of more sites, ethanol amination is much more efficient than

DEE amination.


87

1. A.B. van Geysel and W. Musin, in B. Elvers, S. Hawkins and G. Schultz (Eds.),Ullmann’s Encyclopedia of Industrial Chemistry, 5 ed., VCH, Weinheim, Vol. A16,th

p.535.


A2, p.1.


4. L.D. Pesce and W.R. Jenks, in J.A. Kent (Ed.), Riegel’s Handbook of IndustrialChemistry, 9 ed., Van Nostrand Reinhold, New York, 1992, p.1109.th

5. D.R. Corbin, S. Schwarz, G.C. Sonnichsen, Catal. Today, 37, p.71 (1997).



8. V.A. Veefkind and J.A. Lercher, Appl. Catal. A, submitted 1998; Chapter 2 of thisthesis.

9. K. Segawa and h. Tachibana, J. Catal., 131, 482 (1991).

The important role of the side pockets in ethanol amination is also evident in the variation

of the rate of amination with varying exchange degree. The TOF over the acid sites in the side

pockets is 1.5 times higher than that over the acid sites in the main channel.

It was shown in separate alkylation/ de-alkylation experiments that the rate of MEA removal

by ammonia is in the same range as the rate of formation (2.4 10 to 6.4 10 s mbar ). This is. -5 . -5 -1 -1

confirmed by the very low MEA concentration in the pores and the relatively high concentration

of ammonium ions under reactive conditions, showing that the monoethylammonium ions have

relatively short lifetime under these conditions. This leads to elevated concentrations of ammonium

ions inside the pores, especially in the highly exchanged sample. The confined space in these side

pockets does not allow the formation of DEA, which reacts much slower than MEA. This inability

for DEA formation could also be the main factor explaining the apparent higher reactivity of the

sites in the side pockets.

References

Chapter 4

88


11. V.A. Veefkind, Ch. Gründling and J.A. Lercher, J. Mol. Catal., in press (1998).

12. Jpn. Chem. Week., 33, p. 2 (1992).


14. W.M. Meier and D.H. Olson, Atlas of zeolite structure types, 3 edition, Butterworth-rd

Heinemann, 1992

15. F. Geobaldo, C. Lambert, G. Ricchiardi, S. Bordiga, A. Zecchina, G. Turnes Palominoand C. Otero Areán, J. Phys. Chem., 99, 11167 (1995).

16. F. Eder, M. Stockenhuber and J.A. Lercher in: Zeolites: A Refined Tool for DesigningCatalytic Sites, L. Bonneviot and S. Kaliaguine (editors), Elsevier, Amsterdam, 1995,p. 495.

17. D.T. Chen, S.B. Sharma, I. Filimonov and J.A. Dumesic, Catal. Letters, 12, 201(1992).

18. D.J. Parillo and R.J. Gorte, J. Phys. Chem., 97, 8786 (1993).

19. B.E. Spiewak, B.E. Hardy, S.B. Sharma and J.A. Dumesic, Catal. Letters, 23, 207(1994).

20. N. Cardona-Martinez and J.A. Dumesic, Adv. Catal., 38, 149 (1992).

21. H.A. Benesi and B.H. Winquist, Adv. Catal., 27, 98 (1978).

22. H.G. Karge and V. Dondur, J. Phys. Chem., 94, 765 (1990).

23. B.H. Ha and D. Barthomeuf, J. Chem. Soc., Faraday Trans 1, 75, 2366 (1979).

24. H.G. Karge, Z. Phys. Chem., 122, 103 (1980).


26. C. Lee, D.J. Parillo, R.J Gorte and W.E. Farneth, J. Am. Chem. Soc., 118, 3262(1996).

27. V.L. Zholobenko, M.A. Makarova and J. Dwyer, J. Phys. Chem., 97, 5962 (1993).

28. M. Maache, A. Janin, J.C. Lavalley and E. Benazzi, Zeolites, 15, 507 (1995).

29. M.A. Makarova, A.E. Wilson, B.J. van Liemt, C.M.A.M. Mesters, A.W. de Winterand C. Williams, J. Catal., 172, 170 (1997).

30. J. Datka, B. Gil and A. Kubacka, Zeolites, 17, 428 (1996).

31. A. Auroux, J. Datka, Appl. Catal. A, 165, 473 (1997).

32. H.G. Karge, Z. Phys. Chem., 95, 241 (1975).


89

33. H.G. Karge, In Mol. Sieves 2, Int. Conf., 4, Katzer, J.R., Ed., ACS Symp. Ser., 40,th

Washington D.C., 1977, p. 584.

34. J. Datka, B. Gil, A. Kubacka, Zeolites, 18, 245 (1997).



37. G. Mirth, F. Eder and J.A. Lercher, J. Appl. Spectrosc., 48, 194 (1994).

38. N.B. Colthup, L.H. Daly and S.E. Wiberley, Introduction to Infrared and RamanSpectroscopy, Academic Press, San Diego, 1990, p.343

39. R.M. Silverstein, C.G. Bassler and T.C. Morrill, Spectrometric Identification ofOrganic Compounds, 3 edition, John Wiliey & Sons, New York, 1974, p.109.rd

40. A. Alberti, Zeolites, 19, 411 (1997).

41. M. Pourbaix, Atlas of electrochemical equilibria in aqueous solutions, PergamomPress, London, 1966, p.494.



On the Potential to Synthesize Larger and Mixed Alkylamines with zeolite Catalysts

91

CHAPTER 5

On the Potential to Synthesize Larger and Mixed Alkylamineswith Zeolite Catalysts

Abstract

Ethylmethylamines were synthesized over mordenite using two different routes. Reaction of

methanol with monoethylamine and reaction of ethanol with monomethylamine yielded

ethylmethylamines. Scavenging and adsorption assisted desorption mechanisms were shown to operate

in parallel. The primary products of the scavenging mechanism, ethylmethylamine and the disubstituted

reactant amine, were the most abundant products. The formation of ethene occurred exclusively via

Hofmann elimination from alkylammonium ions.

n-Propanol amination yields 73% monopropylamine and 10% dipropylamine over HBEA. Mordenite

catalysts show a higher propene selectivity and seem less suitable for propylamine synthesis. Isopropanol

amination yields mainly propene (>78%) over the investigated beta and mordenite catalysts. The results

suggest that the propylamines formed, experience difficulties to leave the zeolite pores, especially in

silylated mordenite, and, thus, undergo preferentially Hofmann elimination to propene and lower

(alkyl)ammonium ions.

Chapter 5

92

5.1 Introduction

The production of alkylamines via reaction of ammonia and an alcohol over solid acid

catalysts has been traditionally restricted to methylamines and to a lesser extent, to ethylamines.

Most other amines are produced via reductive amination, in which an aldehyde, ketone or

alcohol is reacted with ammonia over a supported metal catalyst in the presence of hydrogen

[1,2]. The use of solid acids in these cases has been regarded as less attractive due to the high

reactivity of the alcohol, yielding olefins via acid catalyzed water elimination.

Traditionally, in methylamine synthesis amorphous silica-alumina catalysts have been

used, but the use of zeolite catalysts can provide a major advantage due to the shape selectivity

to lower alkylated amines as shown by several groups [3,4,5,6,7]. Research into zeolite catalyzed

amination of methanol has been quite extensive ([8], and references therein) and has resulted in

the commercial use of zeolites for this reaction [9,10]. Research efforts with regard to the acid

catalyzed amination of larger alcohols have been scarce [11,12,13,14].

Recently, the amination of ethanol over zeolite catalysts has been subject to more detailed

investigations [15,16]. Primary reaction pathways have been discussed, together with various

routes to byproduct (viz. ethene and diethyl ether) formation. The selectivity toward these

byproducts and toward the different amines has been seen to depend upon the pore geometry and

acid site strength.

The elementary reaction steps of methanol amination and ethanol amination were found

to be very similar. In both cases alkylation of sorbed ammonium ions is the first step, followed

by removal of the amines from the zeolite pores by ammonia. Both scavenging, in which an alkyl

group is transferred from a sorbed species to a gas phase ammonia or amine molecule as

adsorption assisted desorption (a.a.d.) were shown to take place in the zeolite [15,17]. As a

result, a mix of different alkylammonium ions is present inside the zeolite pores and the

selectivity to the different amines is mainly determined by diffusion controlled selectivity [7].

In the present contribution we use the information previously obtained for methanol and

ethanol amination to explain and predict the behavior of zeolite based catalysts for the synthesis

of ethylmethylamines and propylamines. For the synthesis of ethylmethylamines the emphasis

will be more on the shape selectivity whereas the synthesis of propylamines will be used to

address olefin formation in the amination of higher alcohols.


93

Catalyst Specific Area

(m /g)2

Micropor. Vol.

(cm /g)3

Si/Al EFAL

(%)


(mol/g)

HMOR15 350 0.15 7.5 11 1.7·10-3

HMOR10-EM 360 0.12 5 ~10-15 1.9 10. -3

HBEA 514 0.11 11 38 0.7#10-3

L-HBEA 500 - 11 < 10 ~1.2 10. -3

6CDNG��2J[UKEQ�EJGOKECN�RTQRGTVKGU�QH�VJG�KPXGUVKICVGF�$TÎPUVGF�CEKFKE�OQTFGPKVGU

5.2 Experimental

The catalysts used were HMOR15 (mordenite, Japanese Reference Catalyst (JRC) with

a Si/Al ratio of 7.5), HMOR10-EM (EDTA treated silylated HMOR10, JRC, Si/Al=5) [16],

HBEA (zeolite beta, PQ Zeolites, Si/Al=11) [18] and L-BEA (macrocrystalline beta, Si/Al=12)

[19]. The BET surface and micropore volume (determined by the Harkins-Jura method from the

N adsorption data) were determined on a Micromeritics ASAP 2400 porosimeter. The2

percentage of Al atoms in extraframework positions was determined by Al MAS-NMR. The27

concentration of Brønsted acid sites was calculated from the amount of irreversibly adsorbed

ammonia at 373 K. The physicochemical characteristics of these catalysts are described in

Chapter 3 and compiled in Table 1. The standard activation procedure for the kinetic

measurements was heating with 10 K/min. to 823 K for the mordenite based catalysts and 723

K for the beta based catalysts under 10-15 ml flowing He keeping the catalyst at this temperature

for one hour and subsequent cooling to reaction temperature (usually 573 K). Then, the catalyst

was exposed to a 15 ml/min. flow of 20%NH in He for 30 minutes to cover all acid sites with3

ammonium ions. Simultaneously, the reaction mixture containing the alcohol and ammonia or

an amine in He, was led over the bypass line. The reaction was started by switching the reaction

mixture from the bypass to the reactor by means of a dead volume free switching valve.

To record i.r. spectra under vacuum conditions, a Bruker IFS88 spectrometer was used,

equipped with a vacuum cell (base pressure 10 mbar), as described in [20]. The sample was-6

pressed into a self supported wafer and in situ evacuated under a dynamic vacuum of 10 mbar,-6

heated to 823K with 10K/ min., kept at 823K for 60 min. and cooled down to 523 K. The sorbate

0

20

40

60

80

Sel

ectiv

ities

(m

ol%

)

0 50 100 150 200

TOS (min)

0

20

40

60

80

100

EtO

H c

onve

rsio

n (%

)

Chapter 5

94

Figure 1. Reaction of EtOH with MMA (10/40 mbar, 573 K); (4) MEA,(�) DEA, (z) ethene, ()) DMA, (9) EMA, (9) DMEA, (+) DEMA, and(0) EtOH conversion.

was introduced via a dosing valve.

5% monomethylamine (MMA) in He, 5% monoethylamine (MEA) in He, NH , and 20% NH3 3

in He were supplied by Praxair, the Netherlands, as high purity (99.999%) gasses. Methanol,

ethanol, 1-propanol and 2-propanol were obtained p.a. grade from Merck.

5.3 Results

5.3.1 Ethylmethylamine synthesis

The synthesis of ethylmethylamines was performed over 150 mg H-MOR15, that was

activated and presaturated with NH at 573 K, as described in the experimental section. 3

An experiment was performed with 10 mbar of EtOH and 40 mbar of MMA in He at a

total flow of 13 ml/min, corresponding to a WHSV for EtOH of 0.08 h . The results are depicted-1

in Fig. 1. Due to the low space velocity equilibrium was reached only after ca. 120 minutes. The

steady state ethanol conversion was 60%. Dimethylamine (DMA), resulting from methyl

scavenging by MMA, was the most abundant product, followed by ethylmethylamine (EMA),

the product of ethyl scavenging by MMA, and ethene. Selectivity to MEA is initially high, but

decreases to ca. 16% at steady state.

0 50 100 150 200 250

TOS (min)

0

20

40

60

80

100S

ele

ctiv

ity (

mo

l%)


95

Figure 2. Reaction of MeOH with MEA (10/40 mbar, 573 K); (�) DEA,(z) ethene, (2) MMA, ()) DMA, (9) EMA, (9) DMEA, (+) DEMA, and(0) EtOH conversion.

Synthesis of ethylmethylamines was also performed with 10 mbar MeOH and 40 mbar

MEA in He at a total flow of 15 ml/min. The conversion of methanol was 100%. The selectivity

as a function of time is depicted in Fig. 2. Also in this experiment the products of scavenging by

MEA, diethylamine (DEA) and EMA, are the most abundant products, followed by ethene. In

this second experiment ethene is initially formed at high selectivity, whereas in the fist

experiment the selectivity to ethene increased to its maximum value at steady state, in parallel

with EMA and diethylmethylamine (DEMA). Regardless the reactants, the selctivity to DEMA

was always lower than to dimethylethylamine (DMEA).

As this could be the consequence of diffusion imposed shape selectivity, diffusion constants for

the different amines were derived from their uptake as function of time, as measured at 10 mbar-2

at 523 K, using i.r. spectroscopy. A typical uptake is depicted in Figure 3. The spectra presented

are difference spectra, i.e., the spectrum of the surface has been subtracted from the original

spectra. The figure shows clearly the decrease of the Brønsted acid sites at 3612 cm , which-1

shows as a negative band, and the simultaneous increase of bands typical of protonated amines.

The integrated area of the Brønsted acid site band was used to calculate the uptake of these

amines The diffusion coefficients for the different amines under these conditions were obtained

Chapter 5

96

Figure 3. Uptake of 10-2 mbar EMA in HMOR15 at523 K.

Amine D (cm s )2 -1

MEA 4.0 10. -12

EMA 3.6 10. -12

DEA 1.1 10. -12

DMEA 1.0 10. -12

DEMA 3.6 10. -13

Table 1. Diffusion coefficients fordifferent amines.

by fitting of the uptake curves after Crank [21], for a plane sheet (1 µm crystallite size).

5.3.2 Propylamine synthesis

The synthesis of 1-propylamine and isopropylamine was performed starting from 1-

propanol and 2-propanol, respectively. As it had been shown for ethylamine synthesis that at high

NH /alcohol ratios and temperatures below 573 K olefin production decreased [16], experiments3

were performed at 558 K and partial pressures for propanol and ammonia of 100 and 800 mbar,

respectively. The catalysts used in these experiments were HMOR15, HMOR10-EM, HBEA and

L-HBEA. 100 mg catalyst were activated and pretreated as described in the Experimental section.

The conversion, yield and selctivity of 1-propanol amination are shown in Figure 4. The highest

yield to monopropylamine (MPA) is achieved over BEA and HMOR10-EM. Lower yields to

MPA were obtained over L-HBEA and HMOR15. Whereas the HBEA catalyst was the most

selective to MPA (73%), the HMOR10-EM catalyst showed high propene selectivity (37%). Also

the HMOR15 gave higher propene selectivity than the two BEA samples. In general the

dipropylamine (DPA) selectivity was slightly higher over the BEA samples (10-11%) than over

the mordenite samples (8%). The results for 2-propanol amination are presented in Fig. 5.The

conversion in the case of 2-propanol amination was much higher than in the case of 1-propanol

amination. However, mainly propene was produced over the three catalysts tested (HMOR10-

EM, HMOR15 and HBEA). Only over HBEA, the yield to mono-isopropylamine (MIPA) was

HBEA

propene

8%

MPA

73%

DPA

10%

other

9%

L-HBEA

propene

9%

MPA

68%

DPA

11%

other

12%

HMOR10-EM

propene

37%

MPA

48%

DPA

8%

other

7%

HMOR15

propene

28%

MPA

58%

DPA

8%

other

6%

0

5

10

15

20

25

30

HBEA L-HBEA HMOR10-EM HMOR15

Catalyst

Con

vers

ion/

Yie

ld (

mol

%)

Conversion

MPA Yield


97

Figure 4b. Selectivity in reaction of 1-propanol with ammonia (100/800 mbar, T= 553 K).

Figure 4a. Conversion and yield in reaction of 1-propanol with ammonia (100/800 mbar, T=553 K).

0

10

20

30

40

50

60

70

HBEA HMOR10-EM HMOR15

Catalyst

Con

vers

ion/

Yie

ld (

mol

%)

Conversion

MIPA Yield

HBEA

propene

78%

MIPA

20%

DIPA

0%

Other

2%

HMOR10-EM

propene

89%

MIPA

10%

DIPA

0%

Other

1%

HMOR15

propene

80%

MIPA

17%

DIPA

1%

Other

2%

Chapter 5

98

Figure 5a. Conversion and yield in reaction of 2-propanol with ammonia (100/800 mbar, T=553 K).

Figure 5b. Selectivity in reaction of 2-propanol with ammonia (100/800 mbar, T=553K)


99

Amine Origins

MMA Scavenging of methyl group by NH (formed by a.a.d. of NH (scav. prod.))3 4+

A.a.d. of MMA+ (formed by methylation of NH (scavenging product from4+

MEA ))+

DMA Scavenging of methyl group by MMAA.a.d. of DMA (formed by methylation of MMA or ethyl scav. from DMEA )+ + +

DEA Scavenging of ethyl group by MEAA.a.d. of DEA (formed by methyl scav. from DEMA )+ +

EMA Scavenging of methyl group by MEAA.a.d of EMA+

DMEA Scavenging of methyl group by EMAA.a.d. of DMEA+

DEMA Scavenging of methyl group by DEAScavenging of ethyl group by EMAA.a.d. of DEMA (formed by methylation of DEA )+ +

Table 3. Overview of reactions leading to the different amines in reaction of MeOH with MEA.

similar to the MPA yield in the 1-propanol amination. The selectivity to MIPA was 20, 17 and

10% over HBEA, HMOR15 and HMOR10-EM respectively.

5.4 Discussion

5.4.1 Ethylmethylamine synthesis

For clarity, in Figure 6 the molecular representations of methylamines, ethylamines and

ethylmethylamines are shown, together with their abbreviations and some molecular dimensions.

In Table 3 the most simple pathways for the formation of the different amines in the

reaction of MeOH with monoethylamine (MEA), are compiled. For the reaction of EtOH with

MMA an analogous table can be easily constructed by interchanging ‘ethyl’ and ‘methyl’.

The directly formed alkylammonium ions are MEA and its methylated products EMA and+ +

DMEA . All other ammonium ions inside the pores have to originate from scavenging reactions,+

in combination with adsorption assisted desorption. The most important scavenger will be MEA

Chapter 5

100

Figure 6. Structures of methyl-, ethyl-, ethylmethyl- and propylamines

since it is present in large excess.

In both experiments the combined selectivities to EMA and DMA (when using MMA as nitrogen

source) or DEA (when using MEA as nitrogen source) are ca. 55%. These amines are the primary

products of methyl- and ethyl-scavenging by the reactants MMA and MEA in the respective

reactions. This indicates the importance of this reaction path. However, also products which

0 50 100 150 200 250

TOS (min)

0

2

4

6

8

10

Eth

ene

rat

e (1

0-8 m

ol.g

-1.s

-1)


101

Figure 7. Rate of ethene formation as function of time on stream;(z) MeOH + MEA, (c) EtOH + MMA.

could only be the result of adsorption assisted desorption processes, i.e., MMA in methanol

amination and MEA in ethanol amination, were observed. This strongly supports the conclusion

from Chapter 2 that both scavenging and adsorption assisted desorption occur in parallel.

Both reactions gave a lower DEMA selctivity than DMEA selectivity, although DEMA

in the first experiment is produced via the same pathway than DMEA in the second experiment.

This is attributed to the influence of product selectivity. The diffusion coefficients obtained from

the sorption experiments at 523 K indicate indeed that DMEA can diffuse quicker through the

zeolite channels than DEMA. The latter compound will then have a higher chance to react

further.

Based on the observation that the rate of ethene formation for reaction of methanol with

MEA is similar to the rate of ethene formation in ethanol amination (approx. 1 10 mol g s ). -7 . -1. -1

and relatively independent of the compound supplying the ethyl group (i.e., ethanol or MEA),

it was concluded in Chapter 2 that the Hofmann elimination from ethylammonium ions is the

dominating pathway in ethene formation. This implies a dependence of the rate of ethene

formation on the concentration of ethylammonium ions in the pores. The rate of formation of

ethene as a function of time on stream, is shown in Figure 7. It can be seen that the rate to ethene

increases with time to a steady state value for both experiments. This indeed the expected

Chapter 5

102

behavior, if Hofmann elimination dominates. With increasing time on stream an increasing

concentration of ethylammonium ions is formed inside the pores. The steady state concentration

is expected to be reached earlier when 40 mbar MEA is used then when 10 mbar EtOH is used

as ethyl source since it is fed to the catalyst in higher concentration. This correlates with the

observations for the rate of ethene formation given in figure 7, where the steady state using MEA

is reached after approx. 60 min and using ethanol after approx. 120 min.

5.4.2 Propylamine synthesis

The synthesis of propylamines is more selective over BEA than over MOR, which

produces a significant amount of propene. Especially HMOR10-EM, a very good catalyst in

ethanol amination [16], shows poor propylamine selectivity. The selectivity to propylamines in

isopropanol amination was significantly lower than in n-propanol amination over all catalysts.

The yield was similar for the two reactions over H-BEA, but lower over the mordenite catalysts.

Note that the isopropyl group, apart from being bulkier than the n-propyl group, is also more

reactive [22]. It is proposed that both factors contribute to the high propene selectivity in 2-

propanol amination. Note that BEA, as discussed in Chapter 3, has larger pores than MOR. The

isopropylamines do not readily leave the pores of the catalysts, especially in the case of

mordenite and silylated mordenite (reduced pore openings), and are easily reacted to propene via

Hofmann elimination.

Fetting et al.[14] reported that the protonic form of zeolite Y was unsuitable for the

amination of n-propanol or isopropanol due to the formation of large amounts of side products

(mainly olefins). The propylamine synthesis experiments reported here, indicate that solid acid

catalysts might be successfully applied in n-propanol amination after further optimization of the

catalyst and reaction parameters. For isopropanol amination, in agreement with [14], solid acids

might be less suitable due to extensive propene formation.

5.5 Conclusions

The synthesis of ethylmethylamines can be performed both with MeOH and MEA as with

EtOH and MMA and will yield methylamines and ethylamines as well the desired


103



A2, p.1.

3. K. Segawa and h. Tachibana, J. Catal., 131, 482 (1991).



6. D.T. Chen, L. Zhang, J.M. Kobe, Chen Yi and J.A. Dumesic, J. Mol. Catal., 93, 337(1994).


ethylmethylamines. Both scavenging reactions and adsorption assisted desorption were

concluded to occur. The primary scavenging products EMA and the disubstituted reactant amine

were the main products in ethylmethylamine synthesis. Commercially, the choice of reactant

system will probably depend largely on the cost of the reactants and the demand of the

disubstituted reactant amine (DEA or DMA).

Ethene is produced regardless the type of reactants, due to the Hofmann elimination reaction

from sorbed ethyl(methyl)ammonium ions. The rate of ethene formation is, therefore, largely

dependent on the concentration ethyl(methyl)ammonium ions and on the temperature.

The synthesis of n-propylamines over zeolite catalysts seems feasible. The BEA based

catalysts perform better than those based on MOR (higher propene selectivity). This is attributed

to the slower diffusion of the formed propylamines out of the mordenite pores, especially in the

case of the silylated mordenite, enhancing so the amine residence time and giving them a higher

chance to eliminate propene via Hofmann elimination.

This is in agreement with the observation that the amination of isopropanol produces

propene in high selectivity (>78%) over all catalysts. Selectivity to monoisopropylamine is

highest over the BEA catalyst, but with 20% still quite low.

References

Chapter 5

104

8. D.R. Corbin, S. Schwarz, G.C. Sonnichsen, Catal. Today, 37, p.71 (1997).

9. Jpn. Chem. Week., 33, p. 2 (1992).



12. M. Deeba, M. E. Ford and T. A. Johnson, in D. W. Blackburn (Ed.), Catalysis ofOrganic Reactions, Marcel Dekker Inc. New York (1990).


14. F. Fetting, T. Petry and U. Dingerdissen, Chem.-Ing.-Tech., 63, 492 (1991)

15. V.A. Veefkind and J.A. Lercher, J. Mol. Catal., in press (1998).



18. G.S. Nivarthy, Y. He, K. Seshan, and J.A. Lercher, J. Catal., in press (1998).

19. P.J. Kunkeler, D. Moeskops, and H. van Bekkum, Microp. Mater., 11, 313 (1997).


21. J. Crank, The Mathematics of Diffusion, 2nd edition, Clarendon Press, Oxford, 1975,p. 48.

22. J. McMurry, Organic Chemistry, Brooks/Cole publishing Company, Pacific Grove,1988, p.597.

Summary

105

CHAPTER 6

Summary - Samenvatting

Chapter 6

106

6.1 Summary

Aliphatic amines are amongst the most important chemical intermediates. The worldwide

annual production of these amines is estimated to be several hundreds of thousands of tons. The

most practiced production route is the reaction of ammonia with alcohols to produce alkylamines

of different substitution. For the amination of light alcohols, and especially methanol, an

amorphous silica-alumina catalyst is commonly used. Under the reaction conditions normally

applied in the industrial production of the alkylamines, the product distribution approaches

thermodynamical equilibrium over these catalysts. This product distribution often does not match

the market demand for the alkylamines, which in general is higher for lower substituted amines.

This mismatch in product distribution and market demand has led to extensive research and

screening efforts to use zeolites in methylamine synthesis. As a result, a zeolite based catalyst

for methylamine synthesis is now commercially available. The elementary steps in the synthesis

of methylamines, with respect to mechanism and shape selectivity, have also become more clear

over the last years, but are still cause for debate. A deeper insight in these elementary steps will

provide a better fundamental basis for further catalyst development and improvement.

In Chapter 2 reported mechanistic views on alcohol amination and byproduct formation

were critically compared. Mechanistic experiments were performed using in-situ i.r.

spectroscopy, isotopic labeling in combination with GC-MS and kinetic experiments. It was

shown that methoxy groups, formed by reaction of methanol with Brønsted acid sites react

rapidly with ammonia to form sorbed ammonium ions. The rate of this reaction and the very

limited availability of free Brønsted acid sites under reaction conditions prevent this reaction

pathway from playing an important role. The reaction of methanol with sorbed ammonium ions,

though slightly slower than methoxy amination, is therefore concluded to be the most important

route to sorbed methylammonium ions. The release of methylamines from the zeolite via reaction

with ammonia is rate determining. It was shown that two mechanisms, i.e., (i) the scavenging

mechanism, in which ammonia or an amine receives a methyl group from a sorbed

methylammonium ion, and (ii) the adsorption assisted desorption mechanism, in which ammonia

or an amine assists a methylammonium ion to desorb as methylamine by replacing it as sorbed

ammonium ion, operate in parallel to release the methylamines into the gas phase. The

Summary

107

combination of both steps determines rates and selectivities to the different amines.

The main pathway to ether formation is shown to be the reaction of two alcohol molecules on

weakly acidic sites such as ammonium ions in the pores. Scavenging of methyl groups from

methylammonium ions by methanol was concluded not to be a major pathway. Being a

bimolecular reaction on top or next to sorbed (alkyl)ammonium ions makes ether formation

susceptible to the available pore volume. This available pore volume is mainly determined by

the pore dimensions and the concentration and substitution of the alkylammonium ions.

The formation of alkenes poses a problem in the amination of higher alcohols than methanol. It

was shown that ethene formation over mordenite catalysts can be largely attributed to Hofmann

elimination (decomposition of the sorbed ethylammonium ions) rather than to direct elimination

of water from ethanol over mordenite catalysts. Consequently the reaction temperature becomes

one of the most important parameters to control ethene formation.

In Chapter 3 attention was focused on the kinetics and mechanism of ethylamine

synthesis. Ethanol amination was performed over various sold acid catalysts. It was shown that

mordenite produced higher yields to monoethylamine than the zeolites beta, Y, mazzite and

amorphous silica-alumina. The same mechanisms as discussed for methanol amination in

Chapter 2 were shown to be operative in ethanol amination. The high yield to ethylamines over

mordenite were concluded to be the result of high acid site concentration, giving high rates to

ethylamines, in combination with high acidic strength and a low effective available pore volume,

keeping ethene and ether formation low. Based on the knowledge obtained in Chapter 2 and this

chapter a catalytic system was designed, consisting of an EDTA treated silylated mordenite with

Si/Al = 5 and a reactant mixture with ratio ammonia/ethanol =8 at 558K, which showed a

sustained 99% amine selectivity at 60% conversion for longer times on stream.

Chapter 4 deals with the role of strength and location of Brønsted acid sites in mordenite

for the elementary steps in ethylamine synthesis. Mordenite has one-dimensional 12-ring

channels, giving access to 8-ring ‘side pockets’ as shown on the cover of this thesis. It was found

that approximately 40% of the acid sites were located in the side pockets of the mordenite, where

they are accessible to ammonia, ethanol and monoethylamine but not for diethylamine and

triethylamine.

Partially exchanged NaHMOR samples were found to have the Na ions preferentially located+

Chapter 6

108

in the side pockets, regardless of exchange route or duration, indicating the siting of the H and+

Na ions is thermodynamically determined. Consequently NaHMOR samples with a H content+ +

of 60% or lower, have acid sites located almost exclusively in the main channels of the

mordenite. The acid sites introduced at low exchange degrees were found to desorb ammonia at

lower temperature than those added at higher exchange degree. These lower exchanged samples

show higher ethene selectivity than the higher exchanged sample, probably due to the more

weakly sorbed ammonia ions present.

Alkylation of ammonium ions is possible with ethanol as well as diethyl ether. The latter

compound, however, is able to alkylate only 60% of the ammonium ions which agrees well with

the 60% acid sites present in the main channels of mordenite. By comparing the rates of amine

formation over various partially exchanged samples, it was shown that the sites available in the

side pockets contribute significantly to the overall rate of amination.

In Chapter 5 the potential of zeolite catalysts in the synthesis of ethylmethylamines and

propylamines is discussed. Ethylmethylamines can be produced via two different routes, i.e.,

from methanol and monoethylamine or from ethanol and monomethylamine. Both routes

produced ethylmethylamines as well as ethyl- and methylamines, as a result of a combination of

scavenging and adsorption assisted desorption steps. The primary scavenging products

ethylmethylamine and the disubstituted reactant amine, were the main products. Ethene is

produced regardless the combination of reactant via Hofmann elimination and its rate of

formation is, therefore, mainly determined by the concentration of ethyl groups of sorbed

ammonium ions and by the temperature.

The synthesis of propylamines from n-propanol was much more selective than the synthesis of

isopropylamines from isopropanol which yielded mainly propene. BEA showed in all cases a

higher amine selectivity than MOR. The results suggest that the formed propylamines easily

undergo Hofmann elimination and that their retention in the pores should be minimized. BEA,

due to pore size and structure, imposes less diffusional constraints upon the amines than MOR,

giving less chance to the amines to undergo Hofmann elimination.

Summary

109

6.2 Samenvatting

Alifatische amines behoren tot de belangrijkste intermediairen voor de chemische

industrie. De jaarlijkse wereldproduktie van deze amines wordt geschat op enige

honderdduizenden tonnen. De meest gebruikte produktieroute is de reactie van ammonia met

alcoholen. Voor de aminering van lichte alcoholen, met name methanol, wordt meestal amorf

silica-alumina gebruikt. Onder typische reactiecondities voor de industriële productie van

alkylamines benadert de produktdistributie het thermodynamische evenwicht over deze

katalysatoren. Deze produktdistributie komt meestal niet overeen met de vraag naar de amines,

welke meestal hoger is voor de lager gesubstitueerde amines. Het verschil tussen de vraag en de

bereikte produktdistributie over de amorfe katalysatoren heeft geleid tot uitgebreide onderzoeks

inspanningen en katalysatorscreenings om zeoliet katalysatoren voor methylamine synthese te

gebruiken. Dit heeft o.a. geleid tot een commercieel toegepaste methylamine katalysator op

zeoliet basis. Ook de elementaire stappen zijn de laatste jaren duidelijker geworden, maar geven

nog wel aanleiding tot discussie. Een dieper inzicht in deze elementaire stappen kan een betere

fundamentele basis verschaffen voor verdere katalysator ontwikkeling en verbetering.

In Hoofdstuk 2 worden de gerapporteerde mechanistische inzichten in alcohol aminering

en bijprodukt vorming kritisch vergeleken. Mechanistische experimenten werden uitgevoerd,

gebruikmakend van in-situ i.r. spectroscopie, isotoop labeling gecombineerd met GC-MS en

kinetische experimenten. Er werd aangetoond dat methoxy groepen, gevormd door reactie van

methanol met brønsted zure plaatsen snel reageren met ammonia tot methylammonium ionen.

De reactiesnelheid en de zeer geringe beschikbaarheid van vrije Brønsted zure plaatsen onder

reactiecondities verhinderen dat deze reactieroute een belangrijke rol speelt. De reactie van

methanol met gesorbeerde ammonium ionen, hoewel iets langzamer dan methoxy aminering, is,

door de grote hoeveelheid aanwezige ammonium ionen, de belangrijkste route voor de vorming

van gesorbeerde methylammonium ionen. Het vrijkomen van de methylamines uit de zeoliet via

reactie met ammonia is snelheidsbepalend. Twee mechanismes, namelijk (i) het scavenging

mechanisme waarbij ammonia of een amine een methylgroup van een gesorbeerd ammonium ion

overneemt, en (ii) het adsorptie geassisteerde desorptie mechanisme waarbij ammonia of een

amine een methylammonium ion helpt te desorberen door het als ammonium ion te vervangen,

Chapter 6

110

vinden parallel plaats. De combinatie van deze twee stappen bepaalt de reactiesnelheden en

selectiviteiten naar de verschillende amines.

De belangrijkste route voor ether vorming is de reactie van twee alcoholmoleculen over zwak

zure plaatsen zoals ammonium ionen in de poriën. Er werd geconcludeerd dat scavenging van

methylgroepen van methylammonium ionen door methanol geen belangrijke bijdrage levert.

Omdat ether vorming een bimoleculaire reactie is die op of naast ammonium ionen plaatsvindt,

is deze reactie gevoelig voor het beschikbare porie volume. Dit volume wordt voornamelijk

bepaald door de poriedimensies en de concentratie en substitutie van de alkylammonium ionen.

De vorming van alkenen is een probleem in de aminering van hogere alcoholen dan methanol.

Etheen vorming over mordeniet katalysatoren kan voornamelijk aan Hofmann eliminatie

(decompositie van gesorbeerde ethylammonium ionen) worden toegeschreven. Hierdoor is de

reactietemperatuur één van de belangrijkste parameters om etheen vorming te onderdrukken.

Direkte water eliminatie van de ethanol speelt over mordeniet een veel kleinere rol bij de

gebruikte temperaturen.

In Hoofdstuk 3 wordt meer aandacht besteed aan de kinetiek en het mechanisme van de

ethylamine synthese. Ethanol aminering werd uitgevoerd over verschillende vast zure

katalysatoren. Mordeniet gaf hogere monoethylamine yields dan de zeolieten beta, Y, mazziet

of amorf silica-alumina. Er werd getoond dat voor ethanol aminering dezelfde mechanismen

gelden als beschreven in Hoofdstuk 2 voor methanol aminering. De hoge monoethylamine yield

over mordeniet was het gevolg van de hoge concentratie zure plaatsen, die een hoge snelheid

voor ethylamines oplevert, in combinatie met een hoge zuursterkte en een laag beschikbaar

porievolume, waardoor de etheen- en ethervorming laag gehouden werd. Op basis van de

opgedane kennis in Hoofdstuk 2 en dit hoofdstuk werd een katalytisch systeem ontworpen

bestaande uit een EDTA behandelde gesilyleerde mordeniet met Si/Al=5 en een reactiemengsel

met een ammonia/ethanol ratio van 8 bij 558K, welke voor langere tijd een 99% amine

selectiviteit bij 60% conversie vertoonde.

Hoofstuk 4 behandelt de rol van de sterkte en locatie van Brønsted zure plaatsen in

mordeniet voor de elementaire stappen in ethylamine synthese. Mordeniet heeft ééndimensionale

12-ring kanalen die toegang geven tot 8-ring ‘side pockets’ zoals ook op de voorkant van dit

proefschrift geïllustreerd. Ongeveer 40% van de zure plaatsen bevinden zich in de side pockets

Summary

111

van het modeniet, waar zij toegankelijk zijn voor ammonia, ethanol en monoethylamine, maar

niet voor diethylamine en triethylamine. In gedeeltelijk uitgewisselde NaHMOR bevinden de Na+

ionen zich preferentieel in de side pockets, onafhankelijk van de uitwisselingsroute of -duur. Dit

geeft aan dat de locatie van de Na en H ionen thermodynamisch bepaald is. Derhalve zijn in+ +

NaHMOR met een H+ gehalte van minder dan 60% de zure plaatsen vrijwel alleen in de grote

kanalen aanwezig. De zure plaatsen die bij lage uitwisseling zijn gecreeerd, desorberen ammonia

bij een lagere temperatuur dan de zure plaatsen die bij hogere uitwisselingsgraad zijn ingebracht.

Deze lager uitgewisselde monsters vertonen een hogere etheen selectiviteit, vermoedelijk door

de aanwezigheid van minder sterk gesorbeerde ammonium ionen.

Alkylering van ammonium ionen is mogelijk met ethanol zowel als met diethyl ether. Deze

laatste stof kan echter slechts 60% van de ammonium ionen alkyleren, hetgeen goed overeenkomt

met de 60% zure plaatsen die aanwezig zijn in de grote kanalen van mordeniet. Door de snelheid

van aminevorming over verschillende uitgewisselde monsters te vergelijken werd aangetoond

dat de zure plaatsen in de side pockets een significante bijdrage leveren aan de totale

reactiesnelheid van ethanol aminering over mordeniet katalysatoren.

In Hoofdstuk 5 wordt het potentieel van zeoliet katalysatoren voor de synthese van

ethylmethylamines en propylamines behandeld. Ethylmethylamines kunnen via verschillende

routes geproduceerd worden, namelijk uit methanol en monoethylamine of uit ethanol en

monomethylamine. Beide routes geven ethylmethylamines, zowel als ethyl- en methylamines als

gevolg van een combinatie van scavenging en adsorptie geassisteerde desorptie stappen. De

primaire scavenging producten, ethylmethylamine en de digesubstitueerde reactant amine, waren

de voornaamste producten. Etheen wordt onafhankelijk van de gebruikte reactanten geproduceerd

via Hofmann eliminatie en de vormingssnelheid hiervan wordt voornamelijk bepaald door de

concentratie ethyl groepen in de poriën en door de temperatuur.

De synthese van propylamines uit n-propanol was veel selectiever dan de synthese van

isopropylamines uit isopropanol, wat voornamelijk propeen opleverde. BEA gaf in alle gevallen

een hogere amine selectiviteit dan MOR. De resultaten geven aan dat de gevormde propylamines

gemakkelijk Hofmann eliminatie ondergaan en dat hun retentie in de poriën minimaal moet zijn.

BEA, dankzij poriestructuur en -grootte geeft minder diffusie belemmeringen voor de

propylamines dan MOR, waardoor ze minder de kans krijgen Hofmann eliminatie te ondergaan.

Chapter 6

112

Curriculum Vitae

Victor Veefkind werd geboren op 26 juli 1970 te Amsterdam. In 1988 werd hetVWO

diploma behaald in acht eindexamenvakken aan het Christelijk College Groevenbeek te Ermelo.

In datzelfde jaar begon hij met de studie Scheikunde aan de Universiteit Utrecht alwaar in 1989

het propadeutisch examen werd afgelegd. In 1994 werd het doctoraal examen behaald met als

hoodvak Anorganische Chemie. Hiervoor deed hij in de groep van prof.dr.ir. J.W. Geus

onderzoek naar hydrolyse reacties over titania katalysatoren. Tevens werden keuzevakken

Chemische Informatica en Anorganische Chemie gevolgd. In het kader van dit laatste vak werd

een stage gelopen bij Alberta Sulfur Research Ltd. in Calgary, Canada. Gedurende zijn studie

was hij student-assistent op het chemie-praktikum voor eerstejaars biologiestudenten.

Vanaf 1994 was hij vier jaar als Assistent in Opleiding verbonden aan de vakgroep Katalytische

Processen en Materialen van de faculteit Chemische Technologie aan de Universiteit Twente.

Gedurende deze periode werd onder leiding van Prof.dr. J.A. Lercher het in dit proefschrift

beschreven onderzoek verricht.

victor a. veefkind · victor. contents chapter 1 general introduction 1 chapter 2 on the elementary...

Documents