The Feasibility of In Situ Vibrational Spectroscopy in Liquid-phase Heterogeneous Catalysis Research

(1)

(2)

(3)

The Feasibility of In Situ Vibrational Spectroscopy in

Liquid-phase Heterogeneous Catalysis Research

Proefschrift

ter verkrijging van de graad van doctor aan de Technische Universiteit Delft,

op gezag van de Rector Magnificus prof. dr. ir. J. T. Fokkema voorzitter van het College van Promoties,

in het openbaar te verdedigen op woensdag 22 maart 2006 om 13:00 uur

door

Gerben Marc HAMMINGA

(4)

Dit proefschrift is goedgekeurd door promotor: Prof. dr. J.A. Moulijn

Samenstelling promotiecommissie:

Rector Magnificus voorzitter

Prof. dr. J.A. Moulijn Technische Universiteit Delft, promotor Prof. dr. R.A. Sheldon Technische Universiteit Delft

Prof. dr. W. Buijs Technische Universiteit Delft Prof. dr. ir. B. M. Weckhuysen Universiteit Utrecht

Prof. dr. J. Pérez-Ramírez Catalan Institution for Research and Advanced Studies (ICREA), Spain

Dr. G. Mul Technische Universiteit Delft

Dr. K. Q. Almeida-Leñero Shell Global Solutions, Amsterdam

Dr. G. Mul heeft als begeleider in belangrijke mate aan de totstandkoming van het proefschrift bijgedragen.

The research reported in this thesis was financially supported by Avantium Technologies B.V., and the Netherlands Organisation for Scientific Research (NWO). All reported research was conducted in the Reactor & Catalysis Engineering group, DelftChemTech, Faculty of Applied Sciences, Delft University of Technology (Julianalaan 136, 2628 BL, Delft, The Netherlands).

Proefschrift, Technische Universiteit Delft met samenvatting in het Nederlands

ISBN-10: 90-9020453-9 ISBN-13: 978-90-9020453-6

(5)

(6)

(7)

Conventional analysis methods for performing kinetic and mechanistic studies of heterogeneously catalysed liquid-phase reactions are, in general, based on withdrawal of samples from the reaction medium, after which the samples are analysed by means of ‘off-line’ analytical techniques, such as GC, HPLC, NMR or MS. These techniques all have the disadvantage that withdrawal of samples might possibly influence the reaction kinetics. Moreover, these ‘off-line’ techniques are time consuming, sometimes require laborious sample treatment, and often, labile intermediates cannot be detected, while adsorbates on catalyst particles cannot be analyzed at all.

On-line monitoring tools have already proven their applicability in various disciplines of (fine/bio) chemical and chemical engineering research [1-50]. In heterogeneous catalysis, however especially ATR FT-IR spectroscopy is extensively used to study adsorbed surface species on thin film layers deposited on an internal reflection element (IRE) [51-54]. Only few papers on real-time monitoring have been dedicated to kinetic and mechanistic studies of heterogeneous liquid-phase catalytic reactions at relative low pressures (~ 1 bar) [11, 30, 31, 35, 38, 55]. These studies, however, did not describe the effect of reactants, reaction intermediates or products interacting with, or adsorbed on the catalyst surface. Only few of these studies apply on-line monitoring for mechanistic elucidation purposes, whereas none of them apply the generated data for setting-up kinetic models.

1.2. Process monitoring tools and principle

On-line monitoring of (chemical) processes by means of vibrational spectroscopic techniques has gained increased popularity in industry and research over the past few years, not in the least due to the enhanced sensitivity and capabilities of the equipment. As spectroscopic instruments are continuous and fast developing products this has not only led to a broadening of the application window for these techniques, but also to more robust, easy to use, and quicker systems. For on-line monitoring and remote sensing mostly attenuated total reflection (ATR) Fourier transform infrared spectroscopy (FT-IR), Raman and near-infrared (NIR) spectroscopy are applied. Although ATR FT-IR can be used for remote sensing, Raman and NIR are more suitable for this application as these techniques are more compatible with fiber optics. The application of fiber optics allows monitoring of (chemical) processes over more than hundreds of meters.

(11)

Chapter 1 3 ZnSe Support/focusing element ZnSe Support/focusing element Hastelloy housing Gold seal ATR Diamond crystal IR energy in IR energy out ATR-FTIR probe ZnSe Support/focusing element ZnSe Support/focusing element ZnSe Support/focusing element ZnSe Support/focusing element Hastelloy housing Gold seal ATR Diamond crystal IR energy in IR energy out ATR-FTIR probe Collection Fiber (200 µm) Excitation Fiber (100 µm) Long-pass Band-pass Mirror

Dichroic Extension sleeve

Probe housing Safire window

Laser path Scattering path Common path Collection Fiber (200 µm) Excitation Fiber (100 µm) Long-pass Band-pass Mirror

Dichroic Extension sleeve

Probe housing Safire window

Laser path Scattering path Common path

Fig. 1. (A) Construction of an ATR FT-IR probe. (B) Layout of a process Raman probe, adapted from [56].

The ATR FT-IR probe is shown in Fig. 1 A. The probe consists of a hollow tube of which in the end a ZnSe focussing element is placed that contacts a diamond crystal. The ZnSe focussing element not only acts as a kind of lens to focus the infrared light onto the diamond crystal and back into the tube, but also acts as a support for the diamond crystal.

Fig. 1 B depicts the construction of a process Raman probe, which is somewhat more complicated than the layout of the ATR FT-IR probe. In this figure the laser light used for excitation of the sample is guided to the process Raman probe by one 100 µm silica fiber. The laser light is collimated by a lens and passes a band-pass filter to reject any fluorescence or Raman scattering arising from the excitation fiber, after which it passes a dichroic that only passes the laser light. Finally the laser light is focussed onto the sample by a lens. Scattered Raman light by the sample is collimated by the lens and is reflected by the dichroic towards the mirror. The Raman light passes a long pass filter to reject light of short wave lengths, after which the light is focused onto the collection fiber (200 µm) by a lens.

As can be deduced from Fig. 1 both probes significantly differ by probe layout, and therefore also the way of sampling is different. The Raman probes works with a focal point in the sample medium whereas in the ATR technique the infrared light is reflected internally in the diamond crystal. Fig. 2 illustrates the ATR principle.

(A)

(12)

General Introduction Reaction medium IR energy out IR energy in Gas bubble Catalyst particle d_p(penetration depth)

IR energy in IR energy out

Evanescent wave θ n₁ n2 Reaction medium IR energy out IR energy in Gas bubble Catalyst particle d_p(penetration depth)

Evanescent wave Reaction medium IR energy out IR energy in Gas bubble Catalyst particle d_p(penetration depth) Reaction medium IR energy out IR energy in Gas bubble Catalyst particle d_p(penetration depth) d_p(penetration depth)

Evanescent wave

θ

n₁

n2

Fig. 2. Schematic presentation of the ATR-principle.

The ATR-principle (Fig. 2) is based on the condition of total internal reflection

(sin θ > n2/n1) of incident infrared light (with an angle of incidence θ) reflecting on the inner

surface of a crystal or internal reflection element (IRE) (with refractive index n1). The totally

reflected radiation induces an evanescent wave having the same frequency as the totally reflected light that penetrates a short distance into the sample medium (with refractive index n2) to interact with the molecules in the reaction medium. The amplitude of the evanescent

wave decays logarithmically and the penetration depth (dp) of this wave depends on the angle

of incidence (θ), the refractive index of both IRE and sample material and the wavelength (λ) of the applied radiation. The penetration depth (dp) can be calculated by the following

equation [54, 57, 58]: λ π θ = ⎡ _⎛ _⎞ ⎤ ⋅ ⋅ ⋅ ⎢ −_⎜ _⎟ ⎥ ⎢ ⎝ ⎠ ⎥ ⎣ ⎦ p 1 2 2 2 2 1 1 d n 4 n sin n (1)

The order of the penetration depth of the evanescent wave is typically that of the magnitude of the wavelength of the applied radiation.

1.3. Framework of this thesis

(13)

Chapter 1

5

The hydrogenation of γ-butyrolactone (GBL), which is of increasing industrial relevance, was chosen as model reaction to illustrate the important role, which on-line spectroscopic techniques can play when investigating heterogeneous catalysts in high pressure applications. This reaction was not only chosen to point out the power of on-line monitoring tools, but also to relate the structure of different copper containing catalysts to reactivity and selectivity towards the reactant and products in this hydrogenation reaction. This reaction also served as basis for both a mechanistic study in which the lactone ring size was varied, and a kinetic study in order to evaluate the on-line spectroscopy for obtaining kinetic models.

1.4. Solid acid catalysed reactions

The esterification of hexanoic acid with 1-octanol and the etherification 1-octanol was applied as a model reaction to study the applicability of ATR FT-IR spectroscopy in two-phase systems under atmospheric conditions (see Fig. 3). The catalyst used in both of the reactions was a Nafion resin on a silica support.

O OH + HO O O + H2O HO + OH O + H2O

Fig. 3. Esterification and etherification reaction analysed in the present study.

The use of a cation-exchange resin like Nafion in combination with a porous silica matrix is an excellent alternative for corrosive mineral acids used in acid catalysis. Not only from an environmental point of view, reducing the amount of corrosive mineral acid waste streams, but also economic aspects have driven research to develop solid acid catalysts. The advantages of using solid acids over mineral acids include reduced equipment corrosion, easy separation, less production of waste streams, and the possibility to recycle the catalyst. Cation-exchange resins have been used on a commercial basis as solid acid catalysts in a range of chemical processes among which are[59]:

- Acylation reactions [60, 61]. - Alkylation reactions [62].

- Benzylation of naphthalene [63].

- Dehydration of alcohols to olefins or ethers. - Esterification reactions [64, 65].

- Etherification reactions of olefins with alcohols.

(14)

General Introduction

(OCF2CF)mOCF2CF2SO3H CF3

[(CF2 CF2)n CF CF2]x

Fig. 4. General Nafion structure, where n = 6 or 7; x = 1000; m = 1, 2, or 3.

Nafion/silica composites are prepared by means of a sol-gel like synthesis, in which first silicon alkoxide, e.g. tetramethyl othosilicate (TMOS), is prehydrolysed by addition of hydrochloric acid. Next, the silicate solution is added to a base solution containing Nafion. Due to the presence of the base condensation of the solution occurs within a few seconds, leading to a gel like substance. The resulting composite material has a typical surface area of ± 350 m2_{/g and is highly porous.}

1.5. Copper based catalysts for the selective hydrogenation of γ-butyrolactone to 1,4-butanediol

The direct hydrogenation of maleic anhydride (MA) and its intermediates has been subject of catalysis research for several decades. MA was at first instance produced by oxidation of benzene, however currently n-butane is used for this purpose, as this is a much cheaper feedstock compared to benzene. Chemical processes based on MA include production of γ-butyrolactone (GBL), 1,4-butanediol (BDO) and tetrahydrofuran (THF). Starting from MA is economically more competitive than the conventional production routes like the Reppe- or Davy-Mckee processes [66, 67].

GBL is applied in the chemical industry either as a solvent, to replace chlorinated solvents, or as intermediate in the production of pyrrolidones. BDO is used as a monomer in the production of engineering plastics like polyurethanes, polybutylene terephtalates and spandex fibres (polytetrametylene ether glycol). THF finds its main application as solvent in polymer production processes like PVC, and polytetrametylene ether glycol (PTMEG). Hydrogenation products of maleic anhydride find their application in the pharmaceutical, textile and food industry [66, 68-74]. BASF one of the world’s largest BDO producers (140,000 tonne per year), uses both Reppe- and Davy-technology for their production purposes [75].

In the Reppe process acetylene and formaldehyde are reacted to form 2-butyne-1,4-diol. 2-Butyne-1,4-diol is hydrogenated to yield BDO. 1,4-Butanediol can be used in a dehydrocyclisation reaction to form THF. This process however suffers from several drawbacks [76, 77]:

(15)

Chapter 1

7

- Both acetylene and formaldehyde are facing increasing environmental restrictions. Acetylene because of explosion hazard, and formaldehyde because of possible carcinogenic effects.

- The process conditions employed in the Reppe process are moderately severe; 140 to 280 bar and 250 to 350°C and the reactions take place over precious metal catalysts.

The Davy-McKee process is based on the hydrogenation of esterificated maleic anhydride under mild non aggressive reaction conditions (50-100 bar, 150-220˚C), which includes three main steps [76]:

- Esterification of maleic anhydride with ethanol to diethyl maleate.

- Hydrogenation of diethyl maleate to produce a mixture of 1,4-butanediol, gamma-butyrolactone, tetrahydrofuran and ethanol.

- Separation and purification of product 1,4-butanediol and tetrahydrofuran.

- Recovery of ethanol for recycling to the esterification step. γ-butyrolactone is also recovered and recycled to the hydrogenation reaction system.

The competitiveness of the Davy process is greatly influenced by the transfer cost of the maleic anhydride feedstock. However, switching from benzene to n-butane as feedstock for the production of maleic anhydride has significantly contributed to the viability of this process route [76].

A process alternative to produce THF/BDO form n-butane was developed and commercialised by BP Chemicals together with Lurgi (under the name of: International Specialty Products) and also Invista (former DuPont). In these processes n-butane is converted to maleic acid, after which maleic acid is hydrogenated under mild conditions (175-225˚C, 20-170 bar) over a noble metal catalyst (e.g. Pd/Re on active carbon) [78].

The direct hydrogenation of MA is another process alternative to produce THF and BDO that is, however, not (yet) commercialised, but described in the patent literature. Fig. 5 depicts the reduction of maleic anhydride towards reaction intermediates and products. In this process MA is produced from n-butane, after which MA is dissolved in a liquid GBL stream. Next the MA containing GBL stream is hydrogenated over a noble metal catalyst under mild conditions (170-240 ˚C, 30-50 bar) to yield GBL, THF, and BDO [79-81].

O O O O O O H₂ - H₂O O O 2H₂ 2H₂ 2H₂ -OH HO - H₂O O

Maleic anhydride Succinic anhydride γ-Butyrolactone 1,4-Butanediol Tetrahydrofuran

+H₂O O O O O O O H₂ - H₂O O O 2H₂ 2H₂ 2H₂ -OH HO - H₂O O

Maleic anhydride Succinic anhydride γ-Butyrolactone 1,4-Butanediol Tetrahydrofuran

+H₂O

(16)

In the 1960’s the first patents were published on suitable catalysts for the production of cyclic ethers by hydrogenating γ-butyrolactone or γ-valerolactone. Catalysts consisting of transition metals like, cobalt, nickel and copper or platinum metals like ruthenium and platinum, were especially claimed to be suitable. The conversion levels and selectivities obtained with these catalysts when applied in liquid phase hydrogenations were quite low, 5-60% and 50 to 90%, respectively [82]. In general nickel, cobalt, noble metal based and copper/chromium based catalysts show high selectivity towards γ-butyrolactone and tetrahydrofuran, whereas copper zinc based catalysts are highly selective towards 1,4-butanediol [70, 82-100]. In the mid 90’s research on copper chromium based catalysts shifted towards copper zinc based catalysts due to the fact that the chromium (VI) based catalysts are considered to be carcinogenic and therefore need special handling and safety procedures when applied in chemical processes [70, 101-103].

Most of the binary copper-zinc oxide catalysts are prepared by means of a co-precipitation method, which is in fact a classical method for preparation of methanol synthesis catalysts. For this co-precipitation method an aqueous solution of metal nitrates is slowly added to a vessel containing de-ionised water. Temperature and pH are kept constant, normally in the range of 333-353 K and a pH of 6 to 8, respectively. A carbonate solution is commonly added in order to achieve a constant pH. Via this method, in general, precursor material is obtained that consists mainly of either a malachite (Cu2(CO3)(OH)2), or aurichalcite (Zn,Cu)5(CO3)2(OH)6 phase, or a mixture of these two phases, besides traces of other phases [103-115]. Binary copper zinc oxide catalysts, prepared by co-precipation of a metal nitrate solution, have been studied by Küksal et al. for the hydrogenation of succinic anhydride to γ-butyrolactone and 1,4-butanediol [73]. The applicability of these catalysts in the hydrogenation of γ-butyrolactone, the reaction kinetics of this reaction and its reaction mechanism has never been described in detail.

(17)

Chapter 1 9 Brucite-like layer Brucite-like layer CO₃2-_{, H} 2O Interlayer (An-/Water) = M2+ _{= M}3+ _{= OH} -Brucite-like layer Brucite-like layer CO₃2-_{, H} 2O Interlayer (An-/Water) Brucite-like layer Brucite-like layer CO₃2-_{, H} 2O Interlayer (An-/Water) = M2+ _{= M}3+ _{= OH} -= M2+ _{= M}3+ _{= OH}

-Fig. 6. Schematic representation of the hydrotalcite structure, adapted from [116].

This type of precursor material is generally prepared the same way as the binary copper zinc oxide precursor material at more or less equal temperature and pH range. However, not only pH and temperature are of influence on the crystalline phases present in the final precursor material, but also i) the trivalent metal ratio (x = M3+/(M2+ + M3+)), ii) the Cu/Zn ratio, iii) aging of the obtained precipitate, and iv) the concentration of the applied metallic salt solution. These also highly influence the characteristics of the final catalytic material.

1.6. Outline

Real-time spectroscopic monitoring techniques are upcoming tools in monitoring of chemical processes in different fields in industry and research. The functionality and suitability of these techniques for the different stages of studying heterogeneous catalysts in liquid-phase reactions was investigated. Chapter 2 of this thesis aims to elucidate the applicability of ATR FT-IR spectroscopy in solid acid catalysed systems. This is done by comparison of conventional data generated by GC-analysis with the on-line obtained data, to verify if with both techniques the same kinetic data are obtained. Another question to be answered in this chapter is whether reaction intermediates, reactants or products adsorbed on or interacting with the catalyst, could be observed in the on-line data.

(18)

the catalyst stability. In chapter 4 catalyst Cu,Zn,Al-hydrotalcite like materials are evaluated for the reduction of γ-butyrolactone. This study aims to investigate both the effect of Al content and the Cu2+/Zn2+ ratio on the synthesis of pure hydrotalcite-like Cu,Zn,Al precursors, and to link its properties to the copper surface area, activity, and selectivity of the catalyst obtained from these precursors. This chapter ends with a discussion on the effect of the residual sodium content in the synthesised precursor material on the activity of the final catalyst.

In chapter 5 the reaction kinetics of the hydrogenation of γ-butyrolactone is described. Above all, interest in this chapter went out whether real-time in situ ATR FT-IR spectroscopy is suitable for kinetic studies in heterogeneous catalysis at relatively high temperature and pressure. In order to use real-time in situ ATR FT-IR data to formulate a kinetic model, it is important that intrinsic kinetic data are used. To meet this criterion, mass transfer was investigated to ensure that no internal and external temperature and concentration gradients were present.

Chapter 6 describes the power of ATR FT-IR spectroscopy applied in mechanistic studies. A detailed description of the reaction mechanism for the reduction of γ-butyrolactone to the corresponding diol over copper-based catalysts is given on the basis of varying lactone ring size. As this study indicated the presence of a “short-lived” reaction intermediate for particular lactone ring sizes could only be observed with the in situ technique, which illustrates the great advantage over conventional ‘off-line’ analysis methods.

The applicability of fiber optic Raman spectroscopy as on-line reaction monitoring tool in heterogeneous liquid-phase catalysis research is evaluated in chapter 7. The impact of parameters like temperature, pressure, stirrer speed, catalyst and reactant concentration, are all of influence on the collected Raman signal. Besides the influence of these parameters on the obtained Raman spectra also the Raman sensitivity of the components to be monitored is of influence. These parameters were tested for various hydrogenation reactions over a Cu-ZnO catalyst. It is also discussed if the catalyst itself or interactions of reactant or product with the catalyst could be observed.

Finally in chapter 8 an evaluation of all preceding chapters will be given, and some future options for the use of the spectroscopic techniques in current emerging fields of combinatorial catalyst discovery micro reactor technology are given.

References

[1] Allmendinger, M., Eberhardt, R., Luinstra, G., and Rieger, B., J. Am. Chem. Soc., 124, (2002), 5646.

[2] Barbosa, C. J., Vaillancourt, F. H., Eltis, L. D., Blades, M. W., and Turner, F. B., J. Raman

Spectrosc., 33, (2002), 503.

[3] Berg, J. M., Rau, K. C., Kirk Veirs, D., Worl, L. A., McFarlan, J. T., and Hill, D. D., Appl.

Spectrosc., 56, (2002), 83.

[4] Boere, I. A., Bakker Schut, T. C., Boogert, van den J., Bruin, de R. W. F., and Puppels, G. J.,

(19)

Chapter 1

11 [5] Cooper, J. B., Chemom. Intell. Lab. Syst., 46, (1999), 231.

[6] Dadd, M. R., Sharp, D. C. A., Pettman, A. J., and Knowles, C. J., J. Microbio. Meth., 41, (2000), 69.

[7] Denmark, S. and Pham, S. M., Helv. Chim. Acta, 83, (2000), 1846. [8] Denmark, S. E. and Fu, J., J. Am. Chem. Soc., 122, (2000), 12021.

[9] Dozeman, G. J., Fiore, P. J., Puls, T. P., and Walker, J. C., Org. Proc. Res. & Dev., 1, (1997), 137.

[10] Dunuwila, D. D. and Berglund, K. A., J. Crystal Growth, 179, (1997), 185.

[11] Dyson, R. M., Hazenkamp, M., Kaufmann, K., Maeder, M., Studer, M., and Zilian, A.,

J. Chemometrics, 14, (2000), 737.

[12] Eikje, N. S., Ozaki, Y., Aizawa, K., and Arase, S., J. Biomed. Optics, 10, (2005), 1.

[13] Ellis, J. E., Davis, E. M., Dozeman, G. J., Lenoir, E. A., Belmont, D. T., and Brower, P. L.,

Org. Proc. Res. & Dev., 5, (2001), 226.

[14] Ewing, K. J., Nau, G., Bilodeau, T., Dagenais, D. M., Bucholtz, F., and Aggarwal, I. D., Anal.

Chim. Acta, 340, (1997), 227.

[15] Falcon, J. A. and Berglund, K. A., Cryst. Growth Des., 3, (2003), 947. [16] Févotte, G., Int. J. Pharm., 241, (2002), 263.

[17] Flecher, P. E., Cooper, J. B., Vess, T. M., and Welch, W. T., Spectrochim. Acta, Part A, 52, (1996), 1235.

[18] Ford, J. F., Mann, C. K., and Vickers, T. J., Appl. Spectrosc., 48, (1994), 592.

[19] Freitag, D., Götz, C., Luft, G., Pol, van der A., and Wierda, K., Appl. Spectrosc., 55, (2001), 136.

[20] Gaul, C., Arvidsson, P. I., Bauer, W., Gawley, R. E., and Seebach, D., Chem. Eur. J., 7, (2001), 4117.

[21] Grabarnick, M. and Zamir, S., Org. Proc. Res. & Dev., 7, (2003), 237.

[22] Hua, H. and Dubé, M. A., J. Polym. Sci. Part A: Polym. Chem., 39, (2001), 1860.

[23] Islam, M. T., Rodrígues-Hornedo, N., Ciotti, S., and Ackermann, C., Pharm. Res., 21, (2004), 1844.

[24] Josephsohn, N. S., Kuntz, K. W., Snapper, M. L., and Hoveyda, A. H., J. Am. Chem. Soc.,

123, (2001), 11594.

[25] Jovanovic, R. and Dubé, M. A., J. Appl. Poly. Science, 82, (2001), 2958.

[26] Kainz, S., Brinkmann, A., Leitner, W., and Pfaltz, A., J. Am. Chem. Soc., 121, (1999), 6421. [27] Kappler, B., Tuchbreiter, A., Faller, D., Liebetraut, P., Horbelt, W., Timmer, J.,

Honerkamp, J., and Mülhaupt, R., Polymer, 44, (2003), 6179.

[28] Kip, B. J., Berghmans, T., Palmen, P., Pol, van der André, Huys, M., Hartwig, H., Scheepers, M., and Wienke, D., Vib. Spectrosc., 24, (2000), 75.

[29] Konsler, R. G., Karl, J., and Jacobsen, E. N., J. Am. Chem. Soc., 120, (1998), 10780. [30] LeBlond, C., Wang, J., Larsen, R., Orella, C., and Sun, Y.-K., Top. Catal., 5, (1998), 149. [31] Lipshutz, B. H., Tasler, S., Chrisman, W., Spliethoff, B., and Tesche, B., J. Org. Chem., 68,

(2003), 1177.

[32] Maier, S., Loontjens, T., Scholtens, B., and Mülhaupt, R., Macromolecules, 36, (2003), 4727. [33] McConnell, J. R., Barton, K. P., LaPack, M. A., and DesJardin, M. A., Org. Proc. Res. &

Dev., 6, (2002), 700.

[34] McCreery, R. L., Horn, A. J., Spencer, J., and Jefferson, E., J. Pharm. Sci., 87, (1998), 1. [35] Mul, G., Hamminga, G. M., and Moulijn, J. A., Vib. Spectrosc., 34, (2004), 109.

(20)

[37] Pelletier, M. J., Appl. Spectrosc., 57, (2003), 20A.

[38] Pintar, A., Malacea, R., Pinel, C., Fogassy, G., and Besson, M., Appl. Catal. A - Gen., 264, (2004), 1.

[39] Puskas, J. E., Lanzendörfer, M. G., and Pattern, W. E., Polym. Bull., 40, (1998), 55.

[40] Senanayake, C. H., Smith, G. B., Ryan, K. M., Fredenburgh, L. E., Liu, J., Roberts, F. E., Hughes, D. L., Larsen, R. D., Verhoeven, T. R., and Reider, P. J., Tetrahedon Lett., 37, (1996), 3271.

[41] Skrifvars, M., Niemelä, P., Koskinen, R., and Hormi, O., J. Appl. Polym. Sci., 93, (2004), 1285.

[42] Storey, R. F. and Maggio, T. L., Macromolecules, 33, (2000), 681.

[43] Svensson, O., Josefson, M., and Langkilde, F. W., Chemom. Intell. Lab. Syst., 49, (1999), 49. [44] Tamagawa, R. E., Miranda, E. A., and Berglund, K. A., Cryst. Growth Des., 2, (2002), 511. [45] Tamura, R., Fujimoto, D., Lepp, Z., Misaki, K., Miura, H., Takahasi, H., Ushio, T., Nakai, T.,

and Hirotsu, K., J. Am. Chem. Soc., 124, (2002), 13139.

[46] Thompson, A., Corley, E. G., Huntington, M. F., Grabowski, E. J. J., Remenar, J. F., and Collum, D. B., J. Am. Chem. Soc., 120, (1998), 2028.

[47] Togkalidou, T., Tung, H.-H., Sun, Y., Andrews, A., and Braatz, R. D., Org. Proc. Res. &

Dev., 6, (2002), 317.

[48] Uibel, R. H. and Harris, J. M., Appl. Spectrosc., 54, (2000), 1868.

[49] Watson, T. J. N., Horgan, S. W., Shah, R. S., Farr, R. A., Schnettler, R. A., Nevill, C. R., Jr., Weiberth, F. J., Huber, E. W., Baron, B. M., Webster, M. E., Mishra, R. K., Harrison, B. L., Nyce, P. L., Rand, C. L., and Goralski, C. T., Org. Proc. Res. & Dev., 4, (2000), 477.

[50] Zoeller, J. R., Buchanan, N. L., Dickson, T. J., and Ramming, K. K., Catal. Today, 49, (1999), 431.

[51] Bürgi, T. and Baiker, A., J. Phys. Chem. B, 106, (2002), 10649. [52] Gisler, A., Bürgi, T., and Baiker, A., J. Catal., 222, (2004), 461.

[53] Poston, P. E., Rivera, D., Uibel, R. H., and Harris, J. M., Appl. Spectrosc., 52, (1998), 1391. [54] Rivera, D., Poston, P. E., Uibel, R. H., and Harris, J. M., Anal. Chem., 72, (2000), 1543. [55] Weitkamp, J., Hunger, M., and Rymsa, U., Micropor. Mesopor. Mater., 48, (2004), 255. [56] Carrabba, M. M. and Rauh, R. D., US 5,122,127 (1992) (EIC Laboratories, Inc)

[57] Colthup, N. B., Daly, L. H., and Wiberley, S. E., Introduction to Infrared and Raman Spectroscopy, Academic Press, San Diego, 1990, 358.

[58] Harrick, N. J., Phys. Rev. Lett., 4, (1960), 224.

[59] Harmer, M. A. and Sun, Q., Appl. Catal. A - Gen., 221, (2001), 45.

[60] Heidekum, A., Harmer, M. A., and Hoelderich, W. F., J. Catal., 188, (1999), 230. [61] Sun, Q., Harmer, M. A., and Farneth, W. E., Ind. Eng. Chem. Res., 36, (1997), 5541. [62] Botella, P., Corma, A., and López-Nieto, J. M., J. Catal., 185, (1999), 371.

[63] Beltrame, P. and Zuretti, G., Appl. Catal. A - Gen., 248, (2003), 75.

[64] Beers, A. E. W., Spruijt, R. A., Nijhuis, T. A., Kapteijn, F., and Moulijn, J. A., Catal. Today,

66, (2001), 175.

[65] Heidekum, A., Harmer, M. A., and Hoelderich, W. F., J. Catal., 181, (1999), 217. [66] Harris, N. and Tuck, M. W., Hydrocarbon Proc., (1990), 79.

[67] Jung, S. M., Godard, E., Jung, S. Y., Park, K-C., and Choi, J. U., Catal.Today, 87, (2003), 171.

[68] Castiglioni, G. L., Fumagalli, C., Lancia, R., Messori, M., and Vaccari, A., Chemistry &

(21)

Chapter 1

13 [69] Castiglioni, G. L., Guercio, A., Messori, M., and Vaccari, A., Erdöl und Kohle, Erdgas,

Petrochemie, 48, (1995), 174.

[70] Herrmann, U. and Emig, G., Ind. Eng. Chem. Res., 36, (1997), 2885.

[71] Jung, S. M., Godard, E., Jung, S. Y., Park, K-C., and Choi, J. U., J. of Mol. Cat. A: Chem.,

198, (2003), 297.

[72] Kanetaka, J., Asano, T., and Masamune, S., Ind. Eng. Chem., 62, (1970), 24. [73] Küksal, A., Klemm, E., and Emig, G, App. Cat. A: Gen., 228, (2002), 237.

[74] Zhu, Y-L., Xiang, H-W., Wu, G-S., Bai, L., and Li, Y-W., Chem. Commun., (2002), 254. [75] CEM, Chem. Mark. Rep., 8, (2005),

[76] Harris, N. and Tuck, M. W., Hydrocarbon Proc., 5, (1990), 79. [77] Tanabe, Y., Hydrocarbon Proc., 7, (1981), 187.

[78] Mabry, M. A., Prichard, W. W., and Ziemecki, S. B., US 4,550,185 (1989) (Du Pont de Nemours)

[79] Bertola, A., Constantinescu, T., Raucq, P., and Cassarino, S., US 6,433,193 (2002)

(BASF AG)

[80] Budge, J. R., Attig, T. G., and Pedersen, S. E., US 5,473,086 (1995) (The Standard Oil Co.) [81] Budge, J. R. and Pedersen, S. E., US 4,810,807 (1989) (The Standard Oil Co.)

[82] Johnson, O., US 3,370,067 (1968) (Shell Oil Company)

[83] Varadarajan, S. and Miller, D. J., Biotechnol. Prog., 15, (1999), 845.

[84] Asano, T., Kanetaka, J., Miyake, K., and Sugiura, H., US 3,492,314 (1970) (Mitsubishi Petrochemical Co.)

[85] Kanetaka, J., Shimodaira, T., and Mori, S., US 3,829,448 (1974) (Mitsubishi Petrochemical Co.)

[86] Yamaguchi, M. and Kageyama, Y., US 3,853,922 (1974) (Mitsubishi Petrochemical Co.) [87] Miya, B., US 3,894,054 (1975) (Koa Soap Co.)

[88] Michalczyk, G. and Gluzek, K-H., US 3,969,371 (1976) (Deutsche Texaco AG.) [89] Lange, E. and Hausen, M. Z., US 4,268,695 (1981) (Chemische Werke Huls AG.) [90] Drake, C. A., US 4,620,017 (1986) (Phillips Petroleum Company)

[91] Sharif, M. and Turner, K., US 4,584,419 (1986) (Davy McKee Ltd.)

[92] Cawse, J. N., Johnson, N. E., and Whitaker, M. T., US 4,652,685 (1987) (General Electric Company)

[93] Attig, T. G. and Budge, J. R., EP 0,322,140 (1988) (BP International Ltd.) [94] Budge, J. R., EP 0,404,408 (1990) (BP International Ltd.)

[95] Suzuki, S. and Ichiki, T., US 5,099,038 (1992) (Tonen Corporation)

[96] Dallons, J-L., Jacobs, P., Martens, J., Tastenhoye, P., Vander Eynde, I., and Van Gysel, A., US 5,118,821 (1992) (U C B, S. A.)

[97] Scarlett, J., US 5,395,990 (1995) (Eastman Chemical Company)

[98] Budge, J. R., Attig, T. G., and Pedersen, S. E., US 5,473,086 (1995) (The Standard Oil Co.) [99] Tuck, M. W. M., Wood, M. A., and Hiles, A. G., US 6,077,964 (2000) (BASF AG.)

[100] Bertola, A., Constantinescu, T., Raucq, P., and Cassarino, S., WO 00/27834 (2000) (Pantochim S.A.)

[101] Castiglioni, G. L., Ferrari, M., Guercio, A., Vaccari, A., Lancia, R., and Fumagalli, C., Catal.

Today, 27, (1996), 181.

[102] Horn, G. and Frohning, C. D., US 5,302,569 (1992) (Hoechst AG.)

(22)

[104] Agrell, J., Boutonnet, M., Melián-Cabrera, I., and Fierro, J. L. G., App. Cat. A: Gen., 253, (2003), 201.

[105] Alejo, L., Lago, R., Peña, M. A., and Fierro, J. L. G., App. Cat. A: Gen., 162, (1997), 281. [106] Fujita, S-i., Moribe, S., Kanamori, Y., Kakudate, M., and Takezawa, N., App. Cat. A: Gen.,

207, (2001), 121.

[107] Fujitani, T. and Nakamura, J., Catal. Let., 56, (1998), 119.

[108] Grunwaldt, J.-D., Molenbroek, A. M., Topsøe, N-Y, Topsøe, H., and Clausen, B. S., J. Catal.,

194, (2000), 452.

[109] Günter, M. M., Ressler, T., Bems, B., Büscher, C., Genger, T., Hinrichsen, O., Muhler, M., and Schlögl, R., Catal. Let., 71, (2001), 37.

[110] Klenov, D. O., Kryukova, G. N., and Plyasova, L. M., J. Mater. Chem., 8, (1998), 1665. [111] Pollard, A. M., Spencer, M. S., Thomas, R. G., Williams, P. A., Holt, J., and Jennings, J. R.,

App. Cat. A: Gen., 85, (1992), 1.

[112] Porta, P., De Rossi, S., Ferraris, G., and Pompa, F., Sol. State Ion., 45, (1991), 35.

[113] Sengupta, G., Das, D. P., Kundu, M. L., Dutta, S., Roy, S. K., Sahay, R. N., Mishra, K. K., and Ketchik, S. V., App. Cat., 55, (1989), 165.

[114] Spencer, M. S., Top. Catal., 8, (1999), 259.

[115] Topsøe, N-Y and Topsøe, H., Top. Catal., 8, (1999), 267.

(23)

Chapter 1

(24)

(25)

Part of this chapter has been published in Vib. Spectrosc., 34, (2004), 109.

2

Real-time in situ ATR FT-IR Spectroscopy Applied in Solid Acid Catalysed

Liquid-phase Hydrolysis Reactions

Though real-time in situ ATR FT-IR spectroscopic probes have been applied in various facets of chemical/biochemical (engineering) and pharmaceutical research, only few papers have been dedicated to the use of this technique in heterogeneous catalysis research. To evaluate its applicability in heterogeneous catalysis, two hydrolysis reactions were studied at atmospheric pressure and elevated temperatures. The reactions monitored in the liquid-phase were hydrolysis of 1-octanol and hexanoic acid (esterification) and 1-octanol etherification over a Nafion on silica catalyst.

(26)

ATR FT-IR Spectroscopy in Hydrolysis Reactions

1. Introduction

Attenuated total reflection (ATR) Fourier transform infrared spectroscopy (FT-IR) as real-time in situ monitoring tool, has proven its applicability in various disciplines of (fine) chemical and chemical engineering research. For the development of new synthesis routes and processes for pharmaceutical products, ATR FT-IR was successfully to elucidate reaction mechanisms and kinetics [1, 2]. Also for the development of pharmaceutical crystallisation processes, for which the kinetics of crystal growth and nucleation are of great importance, this technique has proven to be most valuable for the determination of supersaturation concentrations [3-6].

In polymer research this real-time in situ technique also showed to be an excellent monitoring tool, as reactions can be analysed without complicated reactor modifications or expensive deuterated monomers in order to reveal the reaction kinetics and mechanism [7-15]. In catalysis research, e.g. homogeneous catalysis, with this technique not only kinetic and mechanistic information was obtained, but also information about the internal co-ordination of the applied metal complex to the substrate [16-30]. Although in polymer and catalysis research mostly organic phases are studied, also in aqueous phases e.g. in biocatalysis research, ATR spectroscopy has shown considerable potential as novel method for on-line measurements of biocatalytic conversions [31].

In heterogeneous catalysis, this technique is extensively used to study adsorbed surface species on thin film layers deposited on an internal reflection element (IRE) [32-35]. Only few papers on real-time in situ ATR FT-IR spectroscopy have been dedicated to kinetic and mechanistic studies of heterogeneous liquid-phase catalytic reactions, and mostly at relative low pressures [36-41]. In these studies, however, no evidence was found of reactants, reaction intermediates or products interacting with or adsorbed on the catalyst surface. E.g., Pintar et al. report that, within the applied operating window (298K or 308K, 1 or 10 bara), no influence of catalyst particles dispersed in the liquid-phase was found on the collected ATR FT-IR spectra [40].

In order to further evaluate real time in situ ATR FT-IR spectroscopy as analysis tool in heterogeneous catalysis a liquid-phase esterification and etherification reaction were monitored. The usage of a cation-exchange resin like Nafion in combination with a porous silica matrix is an excellent alternative for corrosive mineral acids used in acid catalysis. The Nafion resin has a backbone structure, which is similar to that of Teflon, with the Nafion perfluorinated cation-exchange polymer having pending sulfonic acid groups [42]. The structure of Nafion is given in Fig. 1.

(OCF2CF)mOCF2CF2SO3H CF3

[(CF2 CF2)n CF CF2]x

Fig. 1. General Nafion structure, where n = 6 or 7; x = 1000; m = 1, 2, or 3.

(27)

Chapter 2

19

also whether the catalyst, and species interacting with, or adsorbed on, the catalyst surface could be detected. The esterification of hexanoic acid with 1-octanol was used as model reaction, whereas in the etherification 1-octanol was applied (see Fig. 2).

Further information on the applied catalysts can be summarised as follows. Nafion/silica composites are prepared by means of a sol-gel like synthesis, in which first silicon alkoxide, e.g. tetramethyl othosilicate (TMOS), is prehydrolysed by addition of hydrochloric acid. Next, the silicate solution is added to a base solution containing Nafion. Due to the presence of the base condensation of the solution occurs within a few seconds, leading to a gel like substance. The resulting composite material has a typical surface area of ± 350 m2_{/g and is highly porous.}

O OH + HO O O + H2O HO + OH O + H2O

Fig. 2. Esterification and etherification reaction analysed in the present study.

Not only from an environmental point of view, reducing the amount of corrosive mineral acid waste streams, but also economic aspects have driven research to develop solid acid catalysts. The advantages of using solid acids over mineral acids include reduced equipment corrosion, easy separation, less production of waste streams, and the possibility to recycle the catalyst. Cation-exchange resins have been used on a commercial basis as solid acid catalysts in a range of chemical processes [42]:

- Acylation reactions [43, 44]. - Alkylation reactions [45].

- Benzylation of naphthalene [46].

- Dehydration of alcohols to olefins or ethers.

- Condensation reactions like the manufacture of bisphenol-A from phenol, and acetone.

- Phenol stream purification after the decomposition of cumene hydroperoxide to phenol and acetone.

- Olefin hydration to from alcohols. - Esterification reactions [47, 48].

(28)

2. Experimental 2.1. Materials

The esterification reaction of hexanoic acid (95+%, Merck) and 1-octanol (95+%, Baker) was performed in a reflux slurry configuration shown in Fig 3. The reaction kinetics were either studied at 427 K in cumene (98%, Acros), or at 447 K in n-decane (95+%, Merck). The pressure was atmospheric, the total reaction volume was 200 ml containing 0.4 mol/l acid and alcohol, and 1.0 g of a Nafion on silica composite consisting of 13% w/w Nafion, respectively. A sieve fraction of 35 µm to 75 µm of the Nafion composite catalyst was applied, which was kindly provided by Dupont. In experiments where silica was added, Silica gel Davisil™ grade 643 (99+% Aldrich), was used, in a somewhat smaller sieve fraction (35-53 µm, N2-BET area of 300 m2/g). Reaction medium IR Energy out IR Energy in Catalyst particle ATR Diamond Crystal Thermometer Drying tube IR-probe Heater Reflux cooler Magnetic stirrer IR energy out IR energy in Reaction medium IR Energy out IR Energy in Catalyst particle ATR Diamond Crystal Thermometer Drying tube IR-probe Heater Reflux cooler Magnetic stirrer IR energy out IR energy in

Fig. 3. Reflux set-up and ATR-principle. 2.2. Esterification and etherification experiments

A typical experiment was performed as follows. First, the glass vessel was filled with solvent after which hexanoic acid and catalyst were sequentially added. The reaction mixture was heated to reflux temperature. Subsequently, 1-octanol was added, which is the starting point of the reaction and collection of FTIR spectra. In the case of studying the etherification a mixture of solvent and catalyst was heated to reflux temperature, followed by 1-octanol addition.

(29)

Chapter 2

21

The spectra were automatically collected every two minutes. After reaction, the acquired real-time successive spectra were subject to a post-processing algorithm called ConcIRT™ Opus 1. This algorithm searches the acquired spectra for changes as a function of time. This way the algorithm is capable to extract absorption bands that change at the same rate, assuming that these bands belong to the same component. The algorithm constructs an individual spectrum for each extracted component and simultaneously displays the resulting profiles. Unless otherwise stated, the spectra shown in the figures were solvent and acid subtracted, and not corrected for 1-octanol or the catalyst.

2.3. ATR FT-IR spectra of solids

To evaluate the absorptions of the solid catalyst (Nafion/silica, and silica, respectively), powdered samples (without solvent!) were put on the probe such that the diamond sampling surface was completely covered. This was followed by carefully pressing the solid with a spatula on the diamond crystal. Spectra of the powdered samples were also recorded at 128 scans and a spectral resolution of 4 cm-1.

2.4. Off-line analysis

Off-line GC-analysis was conducted using a Chrompack CP 9001 gas chromatograph equipped with a CP-Sil8 CB column (length 50 m, internal diameter 0.25 mm and film thickness of 0.12 µm) and Chrompack liquid sampler CP 9050. In these experiments, n-hexadecane (99+%, Merck) was added as internal standard during the reaction.

2.5. On-line particle size measurements

For the investigation of particle dimensions and population a Lasentec FBRM D600R in-process probe was used, which was obtained from Mettler Toledo. This probe has a tip diameter of 25 mm and a sapphire window that can withstand pressures up to 10 bar and temperatures between -90°C and 300°C. Particle sizes in the range of 0.5 µm to 2.5 mm can be detected in scan intervals of at least 2 seconds. The applied conditions depend mainly on the solvent used.

(30)

Fig. 4 FBRM probe and chord length measurement, adapted from [49]. 2.6. TGA and TPD-MS investigations

To investigate the nature of chemisorbed species at the catalyst surface both Thermogravimetric Analysis (TGA) and TPD-MS were applied. For the TGA analysis a TGA/SDTA851e thermobalance, equipped with a TSO 801RO sample robot and a TSO 800GC1 gas control unit was used (All obtained from Mettler Toledo). Helium was applied as a carrier gas in all experiments, and fed with a flow rate of 100 ml/min. Samples of approximately 25 mg were heated at a constant rate of 10 K per minute, to a final temperature of 1073 K. The desorption of chemisorbed reaction species present on the catalyst carrier was studied by TPD-MS using an atmospheric plug-flow reactor. Approximately 35.2 mg of sample was weighed into a quartz TPD reactor. After assembly of the reactor in the set-up, a helium flow of 50 ml/min (maximal flow rate) was applied over the reactor bed. When a stable MS-signal was obtained (after half an hour), the reactor was heated to 1073 K at a heating rate of 10 K per min.

3. Results

3.1. Reaction kinetics

A first impression of the applicability of the ReactIR system in the analysis of an esterification reaction in cumene is shown in Fig. 5. Although all spectra were collected in the spectral range of 4000-650 cm-1, only the 1780-1680 cm-1 region is shown that contains the absorption frequencies of the ester and acid. The decrease of the asymmetric C=O stretching frequency of hexanoic acid at 1715 cm-1 (at 427 K) is accompanied by an increasing absorbance of the asymmetric C=O stretching frequency of the ester at 1740 cm-1 (at 427 K), respectively. Similar results are obtained for the esterification in n-decane (at 447 K), however due to the more hydrophobic character of n-decane compared to cumene, the asymmetric carbonyl stretching frequencies of both acid and ester are shifted to somewhat higher wavenumbers, 1720 cm-1 and 1745 cm-1, respectively.

Fibre optic Beam splitter Laser beam Sapphire window Detector Laser

diode Laser beam _{Scan direction}

Typical particle

Chord length

(31)

Chapter 2 23 Acid Conversion 1715 cm-1 Ester Formation 1740 cm-1

Fig. 5. Real-time FTIR waterfall plot of the 1780-1680 cm-1_{region collected in cumene. Dynamics of} asymmetric C=O stretching frequencies of hexanoic acid and ester, respectively. Conditions: 200 ml, 0.4 mol/l 1-octanol, 0.4 mol/l hexanoic acid, 1.0 g Nafion/Silica at 427 K. Spectra were not corrected for solvent, 1-octanol, and catalyst.

In Fig. 6A and B the corresponding concentration profiles are shown for esterification reactions monitored in either cumene or n-decane.

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0 30 60 90 120 150 180 Time [min] Concentration [mol/l]

Hexanoic acid (GC) Hexanoic acid (IR)

Ester (GC) Ester (IR)

(A) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0 30 60 90 120 150 180 Time [min] Concentration [mol/l]

(B) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0 30 60 90 120 150 180 Time [min] Concentration [mol/l]

(B)

Fig. 6. (A) ConcIRT™ concentration profiles and concentration profiles by GC-analysis for the

(32)

The concentration profiles obtained by FTIR analysis (Fig. 6A and B) are constructed by the ConcIRT™ Opus 1 post-processing algorithm and are indicated by an open marker. The solid points indicate the results obtained by ‘off-line’ GC-analysis. Obviously, the data generated by the on-line FTIR spectroscopic technique are in excellent agreement with the ‘off-line’ GC-analysis, for both cases. These concentration profiles can be used to construct a first order dependency of hexanoic acid against time (see Fig. 7A and B). The obtained first order plot is very useful when comparing kinetic experiments, as described by Beers et al [47].

Both Fig. 7A and B show that the esterification reaction is first order dependent in hexanoic acid at both reaction temperatures and the reaction rate is positively influenced by increased reaction temperature. The selectivity (at approximately 180 minutes) is negatively influenced by the increasing temperature, as the selectivity towards the ester drops from 97% in cumene to 92% in n-decane, respectively. This is caused by a positive effect on the rate of the etherification of 1-octanol, as will be discussed later.

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0 20 40 60 Time [min] Ln(C 0 /Ci ) [-] (GC) (IR) (A) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0 20 40 60 Time [min] Ln(C 0 /Ci ) [-] (GC) (IR) (A) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 10 20 30 40 50 Time [min] Ln(C 0 /Ci ) [-] (GC) (IR) (B) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 10 20 30 40 50 Time [min] Ln(C 0 /Ci ) [-] (GC) (IR) (B)

Fig. 7. (A) First order dependency of hexanoic acid against time for both GC- and ConcIRT data. Esterification

reaction conducted at 427 K in cumene. Conditions as described in figure 5. (B) First order dependency of hexanoic acid against time for both GC- and ConcIRT data. Esterification reaction conducted at 447 K in n-decane. Conditions: 200 ml, 0.4 mol/l 1-octanol , 0.4 mol/l hexanoic acid, 1.0 g Nafion/Silica at 447 K.

(33)

Chapter 2

25

Internal mass transfer limitations were not present in the experiments conducted at 427 K in cumene. The experiments conducted at 447 K in n-decane show a slight decrease in catalyst activity upon increasing particle size, which could indicate the reaction is slightly internally mass transfer limited. External mass transfer limitations could be neglected as no influence of stirring speed was found on the reaction rate.

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 35-75 75-125 Particle size [µ m] Activity [mmol/g cat /min] FTIR GC (A) 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 35-75 75-125 Particle size [µ m] Activity [mmol/g cat /min] FTIR GC (B)

Fig. 8. Influence of particle size, ‘off-line’ GC-analysis versus on-line FTIR analysis: (A) Esterification in cumene (427 K). (B) Esterification conducted in n-decane (447 K).

Fig. 9 illustrates a typical result of the second region of interest, the 1250 cm-1 to 1000 cm-1 region, where 1-octanol shows IR absorptions. When performing the reaction at 427 K in cumene, the asymmetric C-O stretching frequency of 1-octanol only appears as a shoulder on the asymmetric C-O stretching vibration band of hexanoic acid. Therefore, the acquired spectra were subjected to a post-processing algorithm, which subtracts the influence of solvent and hexanoic acid from the original successive spectra. As can be deduced from Fig. 9, after this procedure the asymmetric C-O stretching frequency of 1-octanol can be detected (1045 cm-1 at 427 K).

Furthermore, Fig. 9 shows the production of the ester by the asymmetric C-O stretching frequency at 1170 cm-1. Besides the bands assigned to the ester and alcohol, an unexpected very strong IR band was observed at 1100 cm-1. Although the asymmetric C-O stretching frequency of 1-octanol is observed in several spectra at the beginning of the esterification reaction, the interference with the 1100 cm-1 band of the unknown component is too strong for constructing useful concentration profiles for either the 1-octanol reactant or the dioctyl ether by-product (1115 cm-1). Therefore, GC-analysis was used for calculating the reaction selectivity.

(34)

in Fig. 10 did not show any peaks that could not be assigned to either the solvent (n-decane), reactants (alcohol, 1050 cm-1) or product (ester, 1170 cm-1).

Ester Formation 1170 cm-1 1-Octanol Conversion 1045 cm-1 Compound X 1100 cm-1

Fig. 9 Real-time FTIR waterfall plot of the 1250-1000cm-1_{region collected in cumene. Dynamics of} absorption frequencies of 1-octanol, ester, and unknown compound X, respectively. Conditions: 200 ml, 0.4 mol/l 1-octanol , 0.4 mol/l hexanoic acid, 1.0 g Nafion/Silica at 427 K.

Ester Formation

1170 cm-1

1-Octanol Conversion

1050 cm-1

(35)

Chapter 2

27

To investigate whether the development of the unknown band shown in Fig. 9 is solvent related, or a temperature effect, the experiment was repeated in n-decane, only now at lower temperature (427 K). The result of this experiment is shown in Fig. 11, which clearly demonstrates that the absorption band at around 1100 cm-1 is again appearing, although to a lesser extent. This suggests that besides temperature, also the solvent affects the formation of this band to some extent. It should be noted that the solvent did not affect the esterification rate, as a comparable rate was found compared to the experiment in cumene (427 K).

Ester Formation 1170 cm-1 1-Octanol Conversion 1050 cm-1 Compound X 1100 cm-1

Fig. 11. Real-time FTIR waterfall plot of the 1250-1000 cm-1_{region collected in n-decane. Dynamics} absorption frequencies of 1-octanol, ester, and unknown compound X, respectively. Conditions: 200 ml, 0.4 mol/l 1-octanol, 0.4 mol/l hexanoic acid, 1.0 g Nafion/Silica at 427 K.

3.2. Observation of catalyst interactions

To further investigate the origin of the unknown compound, the etherification of 1-octanol over the Nafion/silica catalyst was investigated. The etherification at 427 K in cumene is depicted in Fig. 12, in which the development of the bands in the 1250-1000 cm-1 region is shown. By comparison of the spectral development in Figs. 9 and 12, the unknown band is even more rapidly formed than in the esterification reaction. This indicates that the presence of hexanoic acid in cumene decreases the rate of formation of the unknown compound significantly.

The GC-analysis of this experiment however, indicated that a considerable amount of dioctyl ether was formed. This was not directly clear from the spectra shown in Fig. 12, where the frequency at 1100 cm-1 of the unknown compound strongly interferes with the asymmetric C-O-C stretching frequency of dioctyl ether at 1115 cm-1. In this experiment additional bands

(36)

bands emerge with the same rate as the 1100 cm-1 band of the unknown compound, these bands also belong to this compound.

1-Octanol Conversion 1050 cm-1 800 cm-1 Compound X 1100 cm-1 1260 cm-1

Fig. 12. Real-time FTIR waterfall plot of the 1250-1000 cm-1_{region collected in cumene. Dynamics} absorption frequencies of 1-octanol, and unknown compound X, respectively. Conditions: 200 ml, 0.4 mol/l 1-octanol, 1.0 g Nafion/Silica at 427 K.

1-Octanol Conversion 1050 cm-1 Compound X 1100 cm-1 800 cm-1 1260 cm-1

(37)

Chapter 2

29

In Fig. 13 the band dynamics of the unknown compound in the 1250 to 1000 cm-1 region for the etherification of 1-octanol at 447 K in n-decane over Nafion/silica is shown. Although the unknown compound in n-decane can be formed, in comparison to the etherification in cumene (427 K) the rate of formation of the unknown compound in n-decane at 447 K is much less. This indicates that a relatively low temperature favours the formation of the unknown compound.

Since, the 1100 cm-1 band strongly interferes with the asymmetric C-O stretching vibration of 1-octanol, the ConcIRT™ post processing algorithm was only able to construct a good concentration profile out of the obtained IR data for 1-octanol (see Fig. 14). However, the interference of the 1100 cm-1 with the asymmetric C-O-C stretching vibration of the dioctyl ether was too strong to be able to construct a useful concentration profile for this product. Therefore, the concentration profile for dioctyl ether was calculated from the mass balance. From this figure, it can be deduced that high temperature is favourable for the rate of etherification and that the profiles constructed form the IR data nicely coincide with the profiles obtained from GC analysis.

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0 30 60 90 120 150 180 Time [min] Concentration [mol/l]

1-Octanol (GC) 1-Octanol (IR)

Ether(GC) Ether (IR)

(A) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0 30 60 90 120 150 180 Time [min] Concentration [mol/l]

1-Octanol(GC) 1-Octanol (IR)

Ether (GC) Ether (IR)

(B)

Fig. 14. (A) ConcIRT™ concentration profiles and concentration profiles by GC-analysis for the etherification at 427 K in cumene. Profiles constructed out of the 1115 cm-1_{and 1050 cm}-1_{absorption bands. Conditions as} described in figure 11. (B) ConcIRT™ concentration profiles and concentration profiles by GC-analysis for the etherification at 447 K in n-decane. Profiles constructed out of the 1115 cm-1_{and 1050 cm}-1_{absorption bands.} Conditions as described in figure 12.

(38)

ATR FT-IR Spectroscopy in Hydrolysis Reactions 1-Octanol Conversion 1050 cm-1 800 cm-1 Compound X 1100 cm-1 1260 cm-1

Fig. 15. Real-time FTIR waterfall plot of the 1250-1000 cm-1_{region collected in n-decane. Dynamics} absorption frequencies of 1-octanol, and unknown compound X, respectively. Conditions: 200 ml, 0.4 mol/l 1-octanol, 1.0 g Silica at 447 K.

In this experiment, additional bands at 1260 cm-1 and 800 cm-1 are also clearly emerging in the successive reaction spectra, while from GC-analysis it followed that ethers were not formed at all.

(39)

Chapter 2 31 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0 30 60 90 120 150 180 210 240 Time [m in] C o n cen tr at io n [ m o l/l ]

1-Octanol (GC) 1-Octanol (IR)

Fig. 16. ConcIRT™ and GC-concentration profiles for

1-octanol at 447 K in n-decane. Profiles constructed out of the 1100 cm-1_{, 800 cm}-1_{and 1050 cm}-1_{absorption bands.} Conditions as described in Fig. 15.

To illustrate the effect of the difference in reaction temperature and reaction mixtures on the dynamics of the band at 1100 cm-1, the relative absorbance at this wavenumber is plotted versus time in Fig. 17.

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0 10 20 30 40 Time [min] Absorbance [A.U.] (A) (B) (C) (D) (E) (F)

(40)

As illustrated by Fig. 17, during the esterification reaction in n-decane this band was not observed at all (curve A (●) in Fig. 17). In the experiment in decane, where no acid was present in the reaction medium, only a relative low intensity of the 1100 cm-1 band was obtained. In cumene, even in the presence of the esterification reaction, the increase in the absorption band was observed, which is significantly enhanced in the absence of acid (curve E). Only without Nafion on the catalyst carrier, a more rapid formation of the band at 1100 cm-1_{was observed in n-decane (curve F).}

To investigate whether the emerging 1100 cm-1 absorbance band is associated with catalyst particles, ATR-spectra were taken from both silica particles and Nafion on silica particles (both without solvent), as shown in Fig. 18A.

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 750 850 950 1050 1150 1250 1350 Wavenumber [cm-1_] Absorbance [A.U.] (I) (II) 1075 cm-1 800 cm-1 975 cm-1 (A) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 750 850 950 1050 1150 1250 1350 Wavenumber [cm-1_] Absorbance [A.U.] 1100 cm-1 1050 cm-1 (I) (II) (B)

Fig. 18. (A) ATR-spectrum (I) of silica (35-53 µm) and ATR-spectrum (II) of Nafion/silica particles (35-75 µm). Spectra of powdered samples on the probe, in absence of solvent. (B) ATR-spectrum (I) of 1-octanol and silica particles in reaction mixture at 447 K in n-decane. ATR-spectrum (II) of filtrated reaction mixture.

Clearly, spectra with relative high intensity can be obtained without severe pressing of the solid on the ATR probe. The broad band of SiO2 between 1120 and 1020 cm-1 is typically

assigned to Si-O-Si stretching vibrations, and the weak band at around 975 cm-1 to Si-O-H [50-53]. The higher intensity of the silica spectrum, compared to that of the Nafion/silica composite, is most likely related to the smaller particle size (35-53 µm) of the Davisil silica used, compared to the particles of the analysed Nafion/silica composite (35-75 µm). Also the coverage of the silica by the Nafion is likely to affect the frequency and the intensities of the various bands, as was indicated by Pálinkó et al [54]. Unfortunately, Pálinkó did not describe the 1275 to 775 cm-1 region. The IR bands associated with the SiO2 particles are close to the

(41)

Chapter 2

33

1100 cm-1 ((C-O)M stretching) and 800 cm-1. The 1260 cm-1 absorbance band is assigned to the CH2-wag of the Si-O-R surface group [50-53]. To check whether the 1100 cm-1

absorbance band is really caused by Si-O-R surface groups associated with SiO2 particles,

rather than with dissolved alkoxy species, the following experiment was conducted. After the reaction of 1-octanol with silica in n-decane, the silica particles were filtered off and ATR spectra were taken from the liquid phase.

In Fig. 18B, spectra (I) and (II) display the reaction medium before and after filtration, respectively. Fig. 18B shows the presence of the 1100 and 800 cm-1 bands caused by the alkoxy linkages, in contrast to Fig. 18B (II), where these absorptions have disappeared. Given our filtration efficiency of over 95%, only the asymmetric C-O stretching frequency of 1-octanol is clearly visible at 1050 cm-1_{. Fig. 18B thus strongly suggests that the high intensity}

bands at 1100 cm-1_{in the spectra are vibrations related to SiO}

2 particles.

3.3. Investigation of the origin of vibrations assigned to Alkoxy-linkages 3.3.1. Effect of catalyst attrition

The possibility of the formation of small silica particles during reaction (i.e. catalyst attrition) was investigated by means of on-line particle size measurements. The most severe absorbance in the on-line ATR-FT-IR measurement is expected from particle sizes of, at least the size of the penetration depth of the evanescent wave into the sample. The penetration depth of the evanescent wave can be calculated with the following formula [34]:

λ π θ = ⎡ _⎛ _⎞ ⎤ ⋅ ⋅ ⋅ ⎢ −_⎜ _⎟ ⎥ ⎢ ⎝ ⎠ ⎥ ⎣ ⎦ p 1 2 2 2 2 1 1 d n 4 n sin n (1)

In which dp is the penetration depth in nm, λ is the wavelength in nm, n1 is the

refractive index of diamond, n2 is the refractive index of the reaction mixture (in this case the

refractive index of the solvent is taken; cumene: 1.491, n-decane 1.411), and θ is the angle of incidence of the infrared light (in this case 45°). As the refractive index of diamond is depending on the wavelength of the infrared light, a Herzberger-type dispersion equation was used to calculate the refractive index of diamond [55].

λ λ = + ⋅ + ⋅ + ⋅2 2+ ⋅ 4 2 n A B L C L D E (2) In which L=1/(λ2-0.028) λ in µm, and A = 2.3755, B = 3.3644·10-2, C = -8.8752·10-2, D = -2.4046·10-6, and E = 2.2139·10-9.

(42)

cumene and n-decane were approximately 0.904 µm, and 0.852 µm, respectively. Thus, the on-line particle size measurements give information on the presence, or formation of particle sizes of below 1 µm during the reaction. For these measurements an insertion probe was used. To investigate whether catalyst attrition could be inducing the 1100 cm-1_absorbance

band, chord length distributions were measured of catalyst particles applied in the esterification at 427 K in cumene and at 447 K in n-decane, respectively. The ATR-spectra obtained were consistent with the results shown in Fig. 9 for cumene at 427 K and Fig. 10 for n-decane at 447 K. The resulting chord length distributions are depicted in Fig. 19A and B. The chord length distribution for the esterification reaction at 427 K in cumene is given in Fig. 19A. At the end of the reaction a small but significant shift in the chord length distribution has occurred indicating that de-agglomeration of catalyst particles has occurred. The end curve shows no formation of fines indicating that attrition of catalyst particles is limited. 0 20 40 60 80 100 120 1 10 100 1000 Chord length [µm] [Chords/s] (A) Start reaction End reaction 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 1 10 100 1000 Chord length [µm] [Chords/s] (B) Start reaction End reaction

Fig. 19. (A) Chord length distributions of Nafion/silica catalyst particles at the start and end of the esterification reaction in cumene at 427 K. Conditions as described in fig. 4. (B) Chord length distributions of Nafion/silica catalyst particles at the start and end of the esterification reaction in n-decane at 447 K. Conditions as described in fig. 5B.

(43)

Chapter 2

35 3.3.2 TGA and TPD-MS

In order to validate whether physisorbed or chemisorbed 1-octanol is causing the very strong 1100 cm-1 absorbance band in the ATR-spectra, TGA analysis was applied on the Nafion/silica catalyst (as received), the spent Nafion/silica catalyst after 3 hours of reaction in a mixture containing cumene, hexanoic acid and 1-octanol at 427 K (case D in Fig. 17) and silica particles applied in a reaction mixture of n-decane and 1-octanol at 447 K after 3 hours of reaction (case F. in Fig. 17).

Fig. 20 A, B and C depicts the derivative profiles of the TGA results obtained for the three cases. 0 0 0 0 0 0 0 300 400 500 600 700 800 900 1000 1100 TSample [K] mg/s [A.U.] (A) (B) (C) HF/SiF4 SO2 SO2 SO2 H2O H2O Cumene (427 K) Si-O-R?/HF/SiF4 Si-O-R? n-Decane (447 K) H2O SO2 SO2 0. 01

Fig. 20. (A) TGA derivative profile of Nafion/Silica as received (19.7 mg). TGA conditions: 100

ml/min He, heating rate of 10 K/min. (B) TGA derivative profile of spent Nafion/Silica (23.6 mg) reacted for 3 h in a cumene, hexanoic acid, 1-octanol mixture under conditions as described in Fig. 5. TGA conditions: 100 ml/min He, heating rate of 10 K/min. (C) TGA derivative profile of silica (34.4 mg) reacted for 3 h in a n-decane, 1-octanol mixture under conditions as described in Fig. 15. TGA conditions: 100 ml/min He, heating rate of 10 K/min.

In the case of the Nafion/silica catalyst (as received), Fig. 20 A., between 323 and 398 K physisorbed residual water is evaporated. In the interval of 523 to 723 K SO2 evolution

is observed, which is caused by decomposition of the SO3- group of the Nafion resin. At

around 773 K several decomposition products like, HF, and SiF4 are observed. The obtained

TGA results are in good agreement with the results described in literature [56, 57].

In the derivative profile of Nafion/silica catalyst used in the esterification reaction in cumene at 427 K (Fig. 20 B.), besides the evolution of water in the interval between 323 and 398 K also the evaporation of cumene is observed at around 427 K. Like in the case of the unused Nafion/silica catalyst the production of SO2 is detected in the interval from 523 to 723

(44)

assigned to HF and SiF4 at around 773 K has broadened, which indicates the presence of an

additional decomposing species.

The broadening of the decomposition band at around 773 K, in the case of the spent Nafion/silica catalyst, was compared to a TGA profile of silica particles reacted in n-decane at 447 K in the presence of 1-octanol. The resulting derivative profile is depicted in Fig. 20 C, which shows the presence of some water and solvent. However, at around 573 K the evolution is detected of a species believed to be related to a Si-O-C8H17 like chemisorbed

compound that maximises at around 773 K. This confirms the result found for the spent catalyst, indicating the presence of an additional chemisorbed species (besides HF and SiF4)

on the catalyst carrier surface at around 773 K.

The desorption experiment for silica particles reacted in n-decane at 447 K in the presence of 1-octanol was repeated in a set-up combining TPD and mass spectrometry (TPD-MS). 0 20 40 60 80 100 20 40 60 80 100 120 m/e Rel. Intensity [%] 27 29 28 39 41 42 43 55 56 70 69 83 84 112 0 20 40 60 80 100 20 40 60 80 100 120 m/e Rel. Intensity [%] 27 29 28 39 41 42 43 55 56 70 69 83 84 112

Fig. 21. Mass spectrum of chemisorbed evolved from silica particles reacted in n-decane at 447 K in the

presence of 1-octanol, conditions as described in Fig. 14. TPD conditions: 35.2 mg of sample, 50 ml/min He, heating rate of 10 K/min. Mass spectrum obtained at TPD temperature of 773 K.

The Feasibility of In Situ Vibrational Spectroscopy in Liquid-phase Heterogeneous Catalysis Research