2D-IR pressure jump spectroscopy for the study of adsorbed species on zeolites (2D-PJAS-IR)

1730

Studies in Surface Science and Catalysis, volume 154 E. van Steen, L.H. Callanan and M. Claeys (Editors) © 2004 Elsevier B.V. All rights reserved.

2D-IR PRESSURE JUMP SPECTROSCOPY FOR THE STUDY OF ADSORBED SPECIES ON ZEOLITES (2D-PJAS-IR) Thibault-Starzyk, F.*, Chenevarin, S. and Fernandez, C. Laboratoire Catalyse et Spectrochimie, CNRS-ENSICAEN, 14050 Caen CEDEX, France. E-mail: [email protected] ABSTRACT A new infrared spectroscopic technique is presented for the characterisation of the interaction of probe molecules with zeolites. It uses the high time resolution of modem infrared spectrometers (in rapid or step scan mode), and allows to reach dynamic adsorption parameters on the |xs timescale. These dynamic properties are used to distinguish the various adsorption sites for probe molecules, the example of acetonitrile adsorption on mordenite is given. Analysis of the results by 2D correlation spectroscopy or by a 2"^ Fourier transform in the time domain leads to true 2D-IR spectroscopy. Keywords: infrared, probe molecule, adsorption, acid sites, mordenite, acetonitrile, confinement INTRODUCTION Infrared spectroscopy of adsorbed probe molecules is a well established technique for the characterization of surface sites on zeolites. Complex adsorption systems, with multiple adsorption sites and microporous environment of the probe, can lead to difficulties in understanding and interpreting the spectra. Modern spectrometers can achieve very high time resolution, allowing new approaches to the study of adsorbed molecules on solids and zeolites. We propose to perturb the probe by a pressure jump, and to record the relaxation by surface adsorption using |is time resolved IR spectroscopy. Adsorption can be monitored out of equilibrium, and incorporation of dynamic parameters in the spectroscopic data increases the resolution by separating the spectral components along this new axis. A new 2D technique is proposed which can help understanding complex spectral features typically obtained on zeolites. EXPERIMENTAL PART A Bruker IFS66s spectrometer was used in the step scan or rapid scan mode. The ir cell allowed in situ heating of the sample in the infrared beam, it was connected to a vacuum/gas line. In a typical experiment, the catalyst was activated by in situ heating up to 400°C under vacuum (lO'^Pa), and the probe molecule (perdeuterated acetonitrile 99.9% pure from Aldrich) was introduced in the cell (-100 Pa) at room temperature. A 30% pressure jump was applied every 9s in the cell, and the infrared spectrum was recorded at 0.2-30 ms time resolution and 64-8 cm"^ spectral resolution, at temperature between 300 and 525 K. RESULTS AND DISCUSSION Pressure jump spectroscopy of adsorbed species (PJAS) Time resolution in modern infrared spectrometers now allows the study of transient systems, and dynamics of adsorption can be explored by fast recording of adsorption out of equilibrium. Step scan Fourier transform interferometers can reach easily the microsecond timescale. The only condition is that the observation can be repeated a great number of times with sufficient reproducibility. The recording needs to be triggered to the observation with great accuracy. Rees [1] and Grenier [2] have shown that adsorbed molecules in a zeolite each have their eigen response frequency. By modulating the pressure of a probe gas molecule over a zeolite, they were able to explore the pressure modulation frequency. They pointed out the frequencies for given adsorbates on given zeolites, and extracted diffusion kinetic parameters out of their results. We thought we could use these eigen frequencies to extract spectral information out of complex mixtures on surfaces. Thus, several adsorption sites, or several adsorbed molecules in a mixture, would each lead to a specific eigen response frequency to a pressure modulation, and we could obtain the spectral signature for each individual species on each individual

1731 adsorption site. Instead of having to vary the modulation frequency progressively to explore the whole frequency domain, we have decided to generate a square pressure wave on the surface (or more exactly a pressure jump), which would be equivalent to the generation of an infinite number of pressure oscillations with different frequencies (Figure 1). Such a pressure jump can be described as a shift of the adsorption in an excited state, with a relaxation by diffusion and adsorption equilibria. The relaxation process can be followed by fast ir spectroscopy.

lf-1

0:' A/

fM: '''"

WaveniimlK^r fciir*) Figure 1. Principle of the PJAS-IR method. A pressure jump is generated, leading to a relaxation process by adsorption and diffusion on the surface. IR spectra can be recorded for each pressure jump in step or rapid scan modes. The infrared spectra collected can be analysed by Fourier transform to extract the relaxation processes eigen frequencies, with correlations with the infrared frequencies.

bm?rgy» aili.y

Wiwattttttwf< f Cfir*

Figure 2. IR spectra recorded in the gas phase pressure modulation experiment with acetic acid. The spectrum at time 0 was subtracted for clarity, and only energy changes are displayed.

1732 Gas phase acetic acid: monomer and dimer in equilibrium The response to a pressure jump in the gas phase is free from diffusion and adsorption limitations. It is much faster, and the eigen frequency is unique if only one relaxation process exists in the gas phase. We checked the feasibility of our method by studying acetic acid dimerisation in the gas phase. Gas phase acetic acid is in equilibrium with its dimer, and the monomer/dimer equilibrium is easily influenced by pressure variations. This equilibrium is thus a system of choice for the study of a pressure perturbation. Pressure modulation with a 25 Hz sine wave was monitored by step scan spectroscopy at 200|LIS resolution and 4 cm"' spectral resolution, and a complex infrared variation was obtained (figure 2). Although the raw infrared data collected seemed hardly useable, data treatment by 2D correlation spectroscopy (2D-COS) [3] allowed a very clear identification of bands from the two interconverting species, the monomer and the dimer. Their complex and intricate spectral features could easily be separated, and band assignment was surprisingly clear (figure 3). The experiment was reproduced with a pressure jump, and less intense but similar results were obtained. This showed that our experimental setup could be used to monitor fast transient processes, even under strong vibration conditions.

H PM-

•m^-

W<**tM nt t-t t.

Figure 3. Left: 2D-C0S analysis of the pressure modulation experiment on gas phase acetic acid. Right: positive part of the correlation map, showing the two observed species, monomeric and dimeric acetic acid. Study of adsorbed acetonitrile on mordenite Mordenite is an important zeolite due to its many industrial applications and to it specific pore structure. Its pore system is characterized by large channels formed by 12-membered rings with an elliptical shape of ca. 6.7x7.0 A, running along the c crystallographic axis. In the b direction, the large channels have side pockets formed by 8-membered rings with ca. 3.9 A free diameter. Due to such different possible locations for the acid sites, confinement effects are important in mordenites [4, 5], and have been used to tune catalytic activity [6, 7]. The location and strength of Bronsted acid sites in mordenites have been studied since long by infrared spectroscopy. Bronsted acid sites are due to H+ ions compensating the charge defect in the framework due to the presence of aluminium atoms, and they lead to the presence of very clear v(OH) vibration bands in the infrared spectrum. The v(OH) band in mordenite has a fine structure which has been shown to be due to OH groups in different confined locations (figure 4). Two main v(OH) vibration bands are usually identified, for OH groups in the main channels (V(OHMC) at 3610 cm"') and in the side pockets of the structure (V(OHSP) at 3583 cm"^) [8]. Using mainly crystallographic data, Alberti proposed that three types of oxygen atoms can bear acidic protons in the structure, at positions 0 7 in the main channels, 0 2 at the opening between large channels and side pockets, and 0 9 at the end of the side pockets [9]. The third v(OH) vibration band for OH groups at the opening of the side pocket (02H, denoted here as OHSR) has been identified recently [10].

1733 However, this infrared band remains unclear and can not be seen directly on the spectrum of the fully acidic sample. 3810

1 '%H .-^"^^

'k...^!Jfe-«-^.-^'''

.4 I I L I I I I I I I I ^ I I I I I I I I L t I I 1 I I ^ L I I I I

3630

3610

3590

3570

3550

Figure 4. Left: infrared room temperature spectra of Na,H-Mor samples with various sodium contents showing the fine structure in the v(OH) vibration region, and Right: the three possible OH groups in H-MOR that were assigned to the three corresponding v(OH) vibration bands (in ref 10). Acetonitrile is an important probe molecule for acid sites in zeolites. In mordenites, several types of adsorption can take place. Temperature, pressure and confinement all play a role. At room temperature and low coverage, a hydrogen bond is formed between acetonitrile and OH groups. At high coverage, a very strong H-bond is formed between probe molecules (probably dimers) and OH groups in the side pockets. If the pressure is reduced [3], or if the temperature is risen [11], this particular H-bond disappears, and at very high temperature, acetonitrile is protonated and a nitrilium is formed [11]. The formation of this nitrilium ion seems very much influenced by confinement [12], and probably takes place in the side pockets, but this has so far not been established. Acetonitrile adsorption on H-mordenite was studied by pressure jump 2D spectroscopy, allowing the measurement of spectroscopic and dynamic features of adsorption in the micropores. The time trace at a given infrared frequency is shown on the left of figure 5. The time trace was obtained for each wavenumber, on the whole of the mid-infrared (by the Fourier transform interferometer). The pressure jump was visible between 0 and 1 s, and the decrease of the baseline from 5 to 8 s was due to the modification of the volume of the cell to prepare for the next jump, at 9 s. The relaxation process on the surface was clearly visible after the pressure pulse: some oscillations and envelopes were visible on the time trace, with intensities rapidly decreasing. The corresponding frequencies were obtained by the 7^^ Fourier transform in the time domain (for the first 128 points after the pressure maximum, 3.58s), shown on the right of figure 5. The 2D-IR is shown on figure 6. The Fourier transform in the time domain leads to 4 main peaks, indicating 4 oscillation frequencies due to the pressure pulse. The two very intense and narrow peaks at 5 and 12.5 Hz on figure 5 (right side) were also observed in the absence of a sample. They were not damped quickly, and were not affected by surface dynamics on the sample. They were eigen vibration frequencies of the cell itself The two broad bands at 2.5 and MHz could be observed only in the presence of a solid sample in the cell. Their broadness indicates a rapid relaxation, due to diffusion and adsorption kinetics on the surface. The experiment was performed at various temperatures: 300 K, 423 and 523 K. These four peaks in the pressure response frequency scale were observed on all three experiments. They were at lower frequencies at lower temperatures, because of diffusion and adsorption activation at higher temperature. The two narrow and intense peaks at 5 and 12.5 Hz at 523 K affected all wavenumbers the same, but the two surface responses at 2.5 and 14Hz presented a more complex correlation with wavenumber values.

1734

-0 270

0 275

-0 280

2

3 4 5 Time (s)

Figure 5. PJAS-IR experiment on mordenite at 525 K under 78 Pa of acetonitrile (8 cm"' and 28 ms resolution). The data were averaged on 100 pressure jumps. Right: time trace of the infrared intensity at 2295 cm''; left: corresponding Fourier transform on the first 128 points (3.58 s) after the maximum pressure.

3600

3400

3200

3000

2800

2600

2400

2200

Figure 6. 2D-PJAS-IR spectrum of acetonitrile on mordenite at 525 K. The low frequency response was particularly interesting since it presented a fine structure in the v(OH) and in the v(CN) vibration regions (figure 7). In the v(OH) vibration region, the three v(OH) vibration bands of the OH groups in the three possible locations (OHMC 3605 c m \ OHgR at 5395 c m ' and OHSP at 3585 cm'^) were identified (at 2.25 Hz, 2.73 and 2.5 Hz respectively). This is the first direct infrared detection of the three OH groups in H-mordenite. In the v(CN) vibration region, at 523 K, two v(CN) vibration bands were expected: one around 2290 cm"' for the H-bond, and a second one above 2315 cm'^ for protonated acetonitrile. They could indeed be observed, at 2.73 and 2.5 Hz, respectively. This was a clear indication that protonation of acetonitrile was linked to the OH group in the side pockets, OHSP, and was favoured by confinement in this constrained environment [4]. Interestingly, the perturbation of the H-bond by the pressure jump was only detected at the opening of the side pockets (OHSR), and the OH groups in the main channels at 3605 cm"' were not linked to the change in the H-bond. This is probably due to the concentration of acetonitrile remaining mostly constant

1735 in the main channels during the pressure jump. It had been shown by molecular dynamics simulations that a pressure increase in the main channels allowed a higher proportion of acetonitrile to enter in the side pockets: the main channels act as a pipe leading acetonitrile to the side pockets, in which the loading variation is the most important. When the experiment was performed at room temperature, no difference could be detected between the three OH groups, which only lead to a broad signal in the v(OH) vibration region. At intermediate temperature, 423 K, the pressure jump lead to the formation of the acetonitrile dimer and the particularly strong H-bond at 2315 cm"^ This also lead to a correlation with the OHSP on the 2D map, confirming the formation of this specific H-bond in the side pockets only [5]. v(CN)

v(OH) 4

'

—-"^'"111=^^=^^''^"^"'^ z:z:^^z-'=^~~-'^'^zz^'i ^^.Z"""" 311^^^^^^^^^^^^^^^ l^;^;"'""---—"-—~~—•••:zi^zzzzzzr~~'~'''~''"' ^^-'"•'i:^^'-^^^^^^^

id r E - - : - ^ ; : : - - - ' : ; :

'•••'•••"'•'

"

•"••"•'••.-;""•"::""•"•

: : : : j

^4

•

]

i-i

•

Hz

^•':

.

!

i

2.73HZ/3595 c m - ^

2.73Hz/2285cm-i

^s

[

i6

+

+ 'iA

2.5Hz/3585crr

"

-'

.

J

2.5Hz2320cm-i

1

2.25Hz/3605cm-i ;—

i^\

E j HD

iisiQ

i

iidiu

j...

£«

)

1

1

1

ssi

i&ii

^lU

^a

:

-

1

lis

1

uu

1-

1

1

liia

im

-n«)

^fflQ

-»»

ctn"^

Figure 7. Enlargement on the v(OH) and v(CN) regions around 2.5 Hz from figure 6. CONCLUSION A new spectroscopic 2D IR technique is presented for the study of adsorption in micro and mesoporous systems. It is based on |LIS time resolved spectroscopy and sudden pressure changes in the infrared cell containing the solid sample and the gas probe molecule. The 2D spectrum is obtained by two Fourier transforms, in the infrared wavelength domain and in the time domain. The potential of the technique was shown by studying probe molecule adsorption on mordenites, and confirmed previous studies performed in classical adsorption cells. 2D-PJAS-IR allowed the first direct detection of three distinct OH vibration bands in the infrared spectrum of mordenite, and revealed three different interactions with acetonitrile, depending on the adsorption site. Protonation of acetonitrile at high temperature was shown to take place only in the side pockets. ACKNOWLEDGEMENTS We are grateful to Bruker Optics for technical and financial support. REFERENCES 1. 2. 3. 4. 5. 6.

Onyestak G., Valyuon J., Rees L.V.C., Phys. Chem. Chem. Phys., 2 (2000), 3077. Bourdin V., Grenier Ph., Meunier F., Sun L.M., AIChE J., 42 (1996), 700. Thibault-Starzyk F., Vimont A., Gilson J.P., Catal Today, 70 (2001), 229. Smirnov K. S., Thibault-Starzyk F., J. Phys. Chem. B, 103 (1999), 8595. Marie O., Thibault-Starzyk F., Lavalley J. C , Phys. Chem. Chem. Phys., 2 (2000), 5341 Terlouw T., Gilson J. P., Shell Int. Res. Mij. B. V., EP n° 458 378 (1991).

1736 7. Marie O., Thibault-Starzyk F., Massiani P., Lavalley J. C, Stud. Surf. Sci. Catal., 135 (2001), 220 12-P-14 8. M. Maache, A. Janin, J.C. Lavalley and E. Benazzi; Zeolites, 15, 507 (1995) 9. Alberti A., Zeolites, 19 (1997), 411. 10. Marie O., Thibault-Starzyk F., Massiani, P. under press. 11. Czyzniewska J., Chenevarin S., Thibault-Starzyk F., Stud. Surf. Sci. Catal., 142 (2002), 342. 12. Thibault-StarzykF., Travert A., Saussey J., Lavalley J.-C, Topics in Catalysis, 6 (1998), 111.