J. Phys. Chem. B 2006, 110, 16633-16639
16633
In Situ Investigation of Acetonitrile Adsorption on Al2O3-Coated CaF2 Using Sum-Frequency Spectroscopy S. Beau Waldrup and Christopher T. Williams* Department of Chemical Engineering, UniVersity of South Carolina, Columbia, South Carolina 29208 ReceiVed: April 26, 2006; In Final Form: June 14, 2006
Sum-frequency spectroscopy (SFS) has been used to probe the interface between a model catalyst support (Al2O3-coated CaF2) and liquid consisting of neat acetonitrile and acetonitrile in ethanol. Vibrational features associated with both CtN stretching (∼2242 and 2281) and CsH stretching (∼2942 and 2989) are observed for adsorbed nitrile. The peak positions are only slightly shifted from the bulk values, indicating weak adsorption to the model support. In the case of the neat liquid, the C-H stretching vibrations have also been probed using various polarization combinations of the sum frequency, visible and infrared beams in order to determine the orientation of the nitrile with respect to the surface. The average molecular angle calculated (∼39°) shows the off-normal alignment of acetonitrile on the surface of the model support. Finally, an attempt has been made to obtain qualitative information about the adsorption isotherm of acetonitrile on Al2O3/CaF2. The lack of curvature in the isotherm indicates the lack of affinity for the surface by acetonitrile in ethanol.
Introduction Heterogeneous liquid-phase nitrile hydrogenation is an important industrial reaction for producing primary, secondary, and tertiary amines,1 which are important for the production of certain textiles, surfactants, and nylons.2 In 1905, Sabatier pioneered the first studies of production of amines by liquidphase nitrile hydrogenation where he proposed that primary amines were formed in a sequential fashion through the aldimine intermediate.3 It was later discovered that this reaction also leads to the formation of both secondary and tertiary amines.1 Von Braun4 and others,5-7 proposed a mechanism by which secondary amines were formed by following a stepwise reaction. First, nucleophilic addition of a primary amine to produce 1-aminodialkylamine occurs, followed by elimination of ammonia to yield the alkylidenealkylamine (secondary imine). Finally, the CdN bond is hydrogenated to yield the secondary amine. In this same way, tertiary amines are formed by the reaction of the aldimine with a secondary amine to form 1-aminotrialkylamine which undergoes elimination of ammonia to form the enamine. This enamine is then hydrogenated to form the tertiary amine. Sachtler and co-workers published a series of studies regarding the intermolecular hydrogen transfer mechanism during transition metal heterogeneous hydrogenation reactions.8-11 It was found that secondary and tertiary amines could form equally well using heterogeneous catalysts in both liquid- and gas-phase hydrogenation conditions. This suggested that the steps required to convert nitriles to primary, secondary, and tertiary amines were taking place on the catalyst surface.8 Despite such findings, the surface intermediates that are formed when nitriles adsorb on these transition metal catalysts have not been directly identified. Therefore, developing spectroscopic methods by which the surface species can be observed in situ is of crucial * To whom correspondence should be addressed. Department of Chemical Engineering, Swearingen Engineering Center, University of South Carolina, Columbia, SC 29208. Fax: (803) 777-8265. E-mail: willia84@ engr.sc.edu.
importance to being able to elucidate the mechanism for heterogeneous nitrile hydrogenation. However, industrial nitrile hydrogenation is typically performed through a liquid-phase reaction. This produces challenges for in situ spectroscopic characterization, since many conventional spectroscopic techniques used for studying gas-phase catalysis (e.g., transmission Fourier transform infrared or Raman) are not as effective due to bulk phase interference from the liquid. Nevertheless, in recent years there has been significant effort aimed at applying linear vibrational spectroscopic techniques such as infrared reflection-absorption spectroscopy (IRAS),12 attenuated total reflection infrared (ATR-IR)13 spectroscopy, and surface-enhanced Raman spectroscopy (SERS)14 to the study of solid-liquid catalytic interfaces. These approaches have been used for example by us15-17 and others18-22 to study the surface environment during enantioselective hydrogenation of R-ketoesters on cinchonidine-modified Pt. Recently, the nonlinear optical technique of sum-frequency spectroscopy (SFS) has been gaining acceptance as a method for catalyst studies.23,24 The advantage of SFS is that it allows for perfect interface selectivity to be achieved, thereby providing direct vibrational information about the surface species and intermediates. We have been attempting to apply several of these approaches to the study of nitrile adsorption and hydrogenation on catalytic supports and metals. ATR-IR studies of Al2O3 and Pt/Al2O3 in nitrile/hexane mixtures have revealed that acetonitrile adsorbs in two configurations on the catalysts surface: a weakly bound end-on species and an imine-type species.25 The latter was found to be reactive toward hydrogen, suggesting its importance in the nitrile hydrogenation mechanism. We have also examined the adsorption of acetonitrile on Al2O3 with SFS, confirming the presence of an end-on species on the surface.26 In that study, an optically flat sapphire prism was used to perform the experiment in a total internal reflection arrangement. This allowed for a determination of the orientation of the adsorbed acetonitrile. In the present study, we have examined a thin film of Al2O3 sputter-deposited on CaF2. This surface is more representative
10.1021/jp062562w CCC: $33.50 © 2006 American Chemical Society Published on Web 08/02/2006
16634 J. Phys. Chem. B, Vol. 110, No. 33, 2006
Waldrup and Williams
of an actual catalyst support, since the film dimensions and surface roughness are more similar to that of a particle. Adsorption and orientation have been studied both in neat acetonitrile and over a range of acetonitrile/ethanol mixture compositions. Theory The theory behind SFS has been developed and described extensively over the years through various contributions to the literature.26-31 When two laser beams are overlapped on the surface of a material with sufficient electric field strengths, mixing can occur at the surface to create a nonlinear response. The nonlinear response is shown by way of creating a new beam at the sum of the two frequencies. This can be explained by looking at the principles of how light interacts with matter. The equation below describes how the polarization induced in a material is affected by the electric fields of the incident beams.
P h ) 0(χ(1):E h + χ(2):E hE h + χ(3):E hE hE h)
(1)
Here E h is the magnitude of the electric field of the incident light, 0 is the permittivity of free space in a vacuum, P h denotes the induced polarization of a material, and χ(n) are the susceptibilities that describe the directional response of a material to an electric field by way of a tensor quantity. The dynamics of eq 1 are such that the first term describes the response that governs such techniques as Raman and IR spectroscopies, so-called linear techniques. The higher order susceptibilities χ(2),χ(3), etc. describe the nonlinear response of the material and will give significant contributions to P h at high electric field strengths. Knowing that the intensity of the incident beam is related to the induced polarization and the electric field within that medium, we can write a new relationship relating the intensity of the SF light to the incident beams as shown in eq 2.
ISFG ∝
|χI(2)|2UIRUvis Aτ
(2)
In this equation χI(2) is the interfacial nonlinear susceptibility, Uvis and UIR are the visible and infrared pulse energies, respectively, A is the overlapping spot size, and τ is the pulse width. The interfacial nonlinear susceptibility can be broken down into two parts n
χI(2) ) χNR(2) +
∑1 χR(2)
(3)
where χNR(2) is the nonresonant part and χR(2) is the resonant part. The resonant part arises from molecules at the interface, which necessitates the summation. The resonant susceptibility can be shown to be proportional to the hyperpolarizability
N〈β〉 χR(2) ) 0
(4)
where N is the number of SF-active molecules and 〈β〉 is the hyperpolarizability tensor averaged over all of the molecular orientations. Symmetry considerations show that 〈β〉 is only nonzero when a molecular vibration is both Raman and IR active. This result tells us that materials which are centrosymmectric (i.e., the material has inversion symmetry) within the coherence length of the laser are SF inactive. Conversely, regions of a material where inversion symmetry is broken, such
Figure 1. Schematic representation of the experimental setup used to conduct the SFS experiments described.
as an interface, are SF active. This effect gives rise to determining the orientation of molecules relative to a surface. By sampling a small subset of χR(2) with orthogonal polarization states, the elements of 〈β〉 can be measured and thus molecular orientation can be determined. Looking back at the ISFG given in eq 2, we see that ISFG then becomes proportional to the resonances given by the hyperpolarizabilities. n
|χNR(2) + ISFG ∝
∑1
N〈β〉 2 | UIRUvis 0 (5)
Aτ
In the equation above there are a few things to note. The first is that the SF intensity is enhanced when the IR frequency is in resonance with the SF-active vibrational modes. So in effect, scanning the frequency of the IR beam produces a vibrational SF spectrum. The second is that the intensity of the SF light is dependent upon the square of the sum of the resonant and nonresonant susceptibilities. In the case when there is a small nonresonant signal, the spectra will look similar to a typical vibrational spectrum with positive-going peaks. However, in the case when there is a large nonresonant signal (e.g., gold), there will be some derivative peaks and unusual line shapes present (e.g., see Figure S3). With regard to the SFG beam that is created at the point of overlap of the two pump beams, the angle at which this beam exits the medium can be calculated by considering the conservation of momentum at the surface. The resulting equation for the SF angle (θSF) is
(
θSF ) sin-1
)
ωvisnvis sin(θvis) ( ωIRnIR sin(θIR) ωSFnSF
(6)
where ωi is the frequency of the ith beam, ni is the frequencydependent refractive index of the ith beam, and θi is the angle at which the ith beam is incident/reflected to/from the surface with respect to the normal. In this equation the ( refers to choice between a counterpropagating (-) or a copropagating (+) geometrical setup. Experimental Section SFS Laser System. The optical setup for performing SFS has been significantly modified since our previous study.26 We therefore provide a detailed description of our new arrangement shown in Figure 1. The SFS experiments were conducted by
Acetonitrile Adsorption overlapping two beams, one a nontunable 532 nm beam and the other a tunable (0.8-5 µm) infrared beam. These two beams were created using an optical parametric generator/amplifier (OPG/OPA) system (LaserVision) that was pumped by a 20 ps, 1064 nm mode-locked Nd:YAG laser (Continuum). The laser pulses at 20 Hz and can produce pulse energies up to 50 mJ/ pulse. Inside the OPG/OPA system, a combination of sumfrequency and difference-frequency operation is used to create the tunable infrared beam wherein the spectral resolution changes throughout the infrared frequency from 5 to 16 cm-1 based on the efficiency of the nonlinear conversion. The OPG/ OPA system outputs a beam that has a combination of polarizations, and a polarizer (LaserVision) is used to select the midinfrared polarized light. Directly following the OPG/OPA system, the midinfrared beam first contacts a piece of silicon that is used to overlap a HeNe laser with the infrared to assist in alignment. It then enters a polarization periscope that allows for geometric rotation of the polarization. This beam is then focused using a 100 mm focal length lens before contacting the sample surface. Typical incident IR pulse energies at the surface are 50-80 µJ/pulse in the 2200-2300 cm-1 region and 200-250 µJ/pulse in the 2800-3100 cm-1 region, with a spot size roughly 0.25-0.5 mm in diameter. The reflected beam from the surface is then steered through another 100 mm focal length lens again focusing the beam onto the reference. The reference sample is made of barium titanate fused into glass. Upon exiting the OPG/OPA system, the 532 nm nontunable visible beam is first sent through a series of high density filters in order to adjust the power of the beam and to avoid sample damage. The s-polarized beam (with respect to the sample surface) is then sent through a half-wave plate that can be used to rotate its polarization. While the polarization of the 532 nm beam is essentially pure, it is filtered using a beam-splitter polarizer (CVI Laser) to ensure correct polarization. Next, the beam passes through a 50-50 beam splitter that splits half of the beam off to be used for the reference channel and allows half of the beam to pass through continuing on to the sample channel. The beam then passes through a time delay that allows for adjustment of the beam path length. This adjustment of the beam path length is essential to guaranteeing temporal overlap of the beams on the sample surface. The 532 nm beam proceeds to pass through a 1000 mm focusing lens that gently focuses the beam to a spot size (ca. 1 mm) larger than that of the IR beam on the sample surface. Typical 532 nm pulse energy at the surface was 20-30 µJ/pulse. The other half of the visible beam is used for the reference channel. It first passes through a time delay to control the path length of the beam and then through a 100 mm focal length lens before being focused onto the reference sample surface. The resulting SF beam from the sample (reflected) and the reference (transmitted) surface is steered using a series of mirrors, and the SF signal is detected using photomultiplier tubes (PMT) (Electron Tubes Limited). Before entering the PMTs, each of the beams passes through a band-pass filter (Semrock) that allows >90% of the light at wavelengths of 472 ( 15 nm through and cuts out >99.99999% of the light in all other regions. In addition the sample SF beam passes through a polarization selecting cube (CVI Laser) as well as a monochromator (Acton Inc.) that further prevents stray light and spurious signals from being detected. The SF signal is processed using a boxcar integrator/amplifier that allows for gating of the pulsed signal thereby increasing the signal-to-noise ratio. Instrument control and data acquisition
J. Phys. Chem. B, Vol. 110, No. 33, 2006 16635 were completed by using a LabView program (National Instruments). Infrared and visible beam intensities were measured using a pyroelectric detector (Molectron) and were used to correct the spectra. The procedure for optical alignment and calibration is described in the Supporting Information. Flow Cell and Materials. A 0.5-mL-volume Teflon flow cell is used for each of the SFS experiments conducted in this work and was reported previously.26 The cell (Figure S1) is designed with a 180° clear optical path on the incident side and allows for the detection of both transmitted and reflected SF beams from the surface of the prism. The prism that was used as the model support was an Al2O3 (deposited by magnetron sputtering) coated CaF2 (ISP Optics) equilateral prism (60°, 0.5 in. × 0.5 in. face). The prism was polished on three faces to 60/40 (scratch/dig) and 1/4λ at 632 nm. Before each experiment, the prism was cleaned using sequential reduction/oxidation/reduction for 15 min each in a flow tube reactor at 350 °C. Neat liquid anhydrous acetonitrile (99.9+%) was purchased from Alpha Aesar, and ethanol (99.5+%) was purchased from SigmaAldrich. During the SFS experiments, the cell with the prism is mounted on its edge. A three-way micrometer-driven stage is used to finely adjust the prism for proper alignment. A magnetically driven pump facilitates the flow of liquids through the cell during spectral acquisition, which occurred after the spectral response reached a steady state (typically
