In Situ Studies of Butyronitrile Adsorption and Hydrogenation on Pt

Feb 13, 2007 - Francisco J. Ibañez and Francis P. Zamborini ... Liam McMillan , Lauren F. Gilpin , Justin Baker , Colin Brennan , Alan Hall , David T...
2 downloads 0 Views 176KB Size
Download PDF
3172

Langmuir 2007, 23, 3172-3178

In Situ Studies of Butyronitrile Adsorption and Hydrogenation on Pt/Al2O3 Using Attenuated Total Reflection Infrared Spectroscopy Ivelisse Ortiz-Hernandez and Christopher T. Williams* Department of Chemical Engineering, UniVersity of South Carolina, Columbia, South Carolina 29208 ReceiVed August 24, 2006. In Final Form: December 7, 2006 The adsorption and hydrogenation of butyronitrile (BN) in hexane on a 5% Pt/Al2O3 catalyst has been studied using in situ attenuated total reflection infrared (ATR-IR) spectroscopy. ATR-IR measurements were conducted on thin (∼10 µm) films of catalyst deposited on Ge wave guides. Multivariate analysis involving classical lease-squares (CLS) and partial least-squares (PLS) modeling was used to aid in the interpretation of the spectroscopic data. During the adsorption of BN over a concentration range from 4 to 40 mM in hexane, no clear evidence for adsorbed N-bound end-on species could be detected. However, a feature at ∼1635-1640 cm-1 indicated the presence of an adsorbed imine species, with the CdN group existing in a tilted configuration involving a strong degree of π interaction with the surface. This assignment is bolstered by the detection of N-H stretching bands that are consistent with imine vibrations. This imine-type intermediate is very prominent and shows transient behavior in the presence of solutionphase hydrogen, suggesting that, once formed, it can be converted into amine products that adsorb on the catalyst surface. Evidence for amine formation was observed in the form of N-H stretching and NH2 bending vibrations, with assignments confirmed through comparison studies of butylamine adsorption under identical conditions. Comparisons between Pt/Al2O3 and Al2O3 suggest that there may be some adsorption of these amines on the support surface. The mechanistic implications with regard to heterogeneous nitrile hydrogenation on transition metals under mild conditions are briefly discussed in light of these findings.

Introduction Nitrile hydrogenation is an important industrial reaction, since amines are used extensively for many industrial applications (e.g., textiles, carpets, and yarns) and have applications as fungicides and surfactants. Furthermore, they are employed in the manufacture of pharmaceuticals such as analgesics and antihistamines. Nitrile hydrogenations are also intriguing on a fundamental level from the standpoint of selectivity toward different amines. In perhaps the most extensive review on the subject, DeBellefon and Fouilloux traced the development of the reaction mechanisms for nitrile hydrogenation over the years.1 In the early stages of nitrile hydrogenation research, Sabatier and Senderens proposed that hydrogenation to a primary amine proceeds in a sequential fashion through an aldimine intermediate. However, it was later found that the reaction leads to the formation of both secondary amines and tertiary amines.2 In 1923, von Braun proposed a mechanism for the formation of secondary amines from the addition of a primary amine to an aldimine, leading to the formation of the 1-aminodialkylamine species.3 The elimination of ammonia results in the formation of a secondary amine. In the same fashion, tertiary amines are formed by the reaction of aldimine to form 1-aminotrialkylamine, which then releases ammonia, leading to an enamine intermediate. The enamine is hydrogenated, leading to the formation of a tertiary amine. A summary of these currently accepted reaction pathways is shown in Figure 1. It must be noted that this mechanism does not necessarily involve heterogeneous catalysts. Heterogeneous catalytic hydrogenations of nitriles have been widely studied, with Ni, Co, Ru, Cu, Rh, Pt, and Pd being the * To whom correspondence should be addressed. Fax: (803) 777-8265. E-mail: [email protected]. (1) DeBellefon, C.; Fouilloux, P. Catal. ReV.sSci. Eng. 1994, 36, 459. (2) Sabatier, P.; Senderens, J. B. C. R. Hebd. Seances Acad. Sci. 1905, 140, 482. (3) Von Braun, J.; Blessing, G.; Zobel, F. Ber. Dtsch. Chem. Ges. 1923, 36, 1988.

Figure 1. Proposed mechanistic pathways for the formation of primary, secondary, and tertiary amine products (dashed box) from aliphatic nitriles (dashed circle).

most widely investigated.1-11 Raney Ni is one of the most active catalysts and is actually used in industrial applications for the production of primary amines. However, its skeletal structure and pyrophoric properties make it difficult to handle and the development of new catalysts is desirable.4 Most recently, Sachtler and co-workers5-7 published a series of studies involving the hydrogenation of nitriles on supported transition metals under mild conditions. Their key finding was in regard to the (4) Cerveny, L. Stud. Surf. Sci. Catal. 1986, 27, 105. (5) Huang, Y.; Adeeva, V.; Sachtler, M. H. Appl. Catal., A 2000, 196, 73. (6) Huang, Y.; Sachtler, W. M. H. Appl. Catal., A 1999, 182, 365. (7) (a) Huang, Y.; Sachtler, W. M. H. Stud. Surf. Sci. Catal. 2000, 130, 527. (b) Huang, Y.; Sachtler, W. M. H. J. Catal. 2000, 190, 69. (c) Huang, Y.; Sachtler, W. M. H. J. Catal. 1999, 188, 215. (8) (a) Ortiz-Hernandez, I.; Owens, J.; Strunk, M.; Williams, C. T. Langmuir 2006, 22, 2629. (b) Ortiz-Hernandez, I.; Williams, C. T. Langmuir 2003, 19 (7), 2956. (9) Arai, M.; Takada, Y.; Nishiyama, Y. J. Phys. Chem. B 1998, 102, 1968. (10) McMillan, S. T.; Agrawal, P. K. Ind. Eng. Chem. Res. 1988, 27, 243. (11) Hagiwara, K.; Yamazaki, T.; Katsurahara, T.; Ozawa, S. J. Colloid Interface Sci. 1996, 181, 306.

10.1021/la062502h CCC: $37.00 © 2007 American Chemical Society Published on Web 02/13/2007

ATR-IR Studies of BN Adsorption and Hydrogenation

Figure 2. Proposed reaction mechanism for nitrile hydrogenation on a Ru-supported catalyst as reported in ref 7b. See text for details.

intermolecular hydrogen transfer mechanism during the reaction, shown in Figure 2. Observations of H-D exchanges suggested that the amine is formed by a nucleophilic attack from the carbon of a solution-phase nitrile and an adsorbed intermediate. The hydrogen is then donated by a nitrile molecule itself and not from adsorbed hydrogen. This leaves a nitrile molecule adsorbed through the carbon to the metal site (e.g., Ru) and a partially hydrogenated nitrile that is an N-containing version of a butylidene species. Therefore, it was suggested that the availability of adsorbed hydrogen does not play a significant role on the hydrogenation of the CtN bond. While such studies have helped elucidate the heterogeneous nitrile hydrogenation mechanism under mild reaction conditions, there is still limited insight regarding the surface species involved in this reaction. For example, the MdNsCH2sCH3 species proposed in Figure 2 has not been readily detected in any surface studies of amine adsorption on metals. There is clearly a strong need to study surface species present under reaction conditions to further elucidate the reaction mechanism. In a previous work, we reported an attenuated total reflection infrared (ATR-IR) study of acetonitrile adsorption on 5% Pt/ Al2O3.8 Acetonitrile adsorption studies revealed the formation of an imine-type intermediate with a characteristic frequency of ∼1640 cm-1. In addition, a spectral feature was observed at 2275 cm-1 that was characteristic of linearly adsorbed acetonitrile on Pt. In the current work, the surface interactions of 5% Pt/ Al2O3 with butyronitrile under hydrogenation conditions have been investigated. As a comparison, butylamine adsorption on 5% Pt/Al2O3 was also tested to help elucidate the complex vibrational spectra in this system. The use of multivariate modeling by a combined classical least-squares (CLS) and partial least-squares (PLS) approach was critical for data analysis.8 Experimental Section Materials. The gases used for the experiments were ultrahigh purity (UHP) hydrogen, oxygen, helium, and carbon monoxide from AirStar. H2PtCl6 (99.5%, Premion) was obtained from Alfa Aesar. Hexanes (HPLC grade) were obtained from Fisher Scientific (Fisher brand, mixture of hexanes) or from VWR manufactured by EMD Chemicals (HX0290P-1) with 85% minimum n-hexane. Butyronitrile and butylamine (both HPLC grade) were obtained from Aldrich and used without further purification. Catalyst Preparation. The catalyst preparation has been discussed elsewhere,8 with some important aspects reviewed here. The catalyst samples consisted of 5 wt % Pt/γ-Al2O3 prepared using standard wet (aqueous) impregnation with H2PtCl6 as the precursor. The support was γ-Al2O3 powder from Alfa Aesar with a mean particle size of 37 nm and a surface area of 45 m2/g as given by the manufacturer (and measured in our laboratory using the Brunauer-EmmettTeller (BET) method). The resulting catalyst had ∼50% dispersion (obtained by H2 chemisorption) with a mean platinum particle size of 2 ( 1 nm, which was verified by high-resolution transmission electron microscopy. ATR-IR. All spectra were acquired using a Nicolet 670 FTIR spectrometer with a liquid nitrogen-cooled mercury cadmium telluride (MCT) detector. A horizontal ATR accessory (Spectra Tech) was used in conjunction with a home-built aluminum flow cell. The

Langmuir, Vol. 23, No. 6, 2007 3173 design of the flow cell has been previously reported.8b Two to four separate reservoirs equipped with glass frit-capped gas inlets allowed saturation or purging of liquids with gases. The flow system was automated and controlled using a Labview (National Instruments) interface, which allows the running of experiments with identical concentration-time profiles.8a The automated system can handle up to four reservoirs without mixing between solutions. The desired liquid was pumped through the flow cell at a flow rate of 36 cm3/min using a poly(tetrafluoroethylene) (PTFE)-lined gear pump (Cole Parmer). Teflon tubing was used in this study based on its resistance to the solvents and reactants employed. The thin catalyst films for ATR-IR studies were prepared using a suspension of the catalyst in ethanol. The suspension consisted of 50 mg of catalyst in 20 mL of solvent. A germanium wave guide was coated six times and dried under a lamp in air. The resulting film thickness was ∼9 µm. Ethanol was chosen as the solvent of choice to minimize the presence of water in the film. For each experiment, the ATR accessory optics were aligned and optimized, and the sample was left under a flow of solvent until the system reached equilibrium (usually at least 2 h, as determined by achieving a consistent, unchanging absorption spectrum). The catalyst films were found to be stable under the flow conditions utilized, showing no visible signs of deterioration from peeling or the removal of particles. Infrared data were collected continuously every 2 min during all stages of the experiment, with data collection consisting of 128 scans per spectrum using a resolution of 4 cm-1 and a collection time of 69 s. A catalyst pretreatment was then performed by flowing N2-saturated solvent followed by a reduction treatment with a H2saturated solvent for 30 min. This treatment was repeated three times to remove contamination from the film. The gas flow rates were set to 100 cm3/min, and a glass frit was used to saturate the solution with the desired gas. After treating the film, a nitrile or an amine was introduced into the system to examine the adsorption and hydrogenation properties. The concentrations used in this study were 0.004, 0.010, and 0.040 M in hexanes. The adsorption steps were repeated multiple times (i.e., two or more times as indicated in the discussion of the results). As has been reported previously,8 under the flow conditions used here, the liquid species are completely replaced within these alumina/catalyst films within 5 min. As such, transient spectral variations over longer time scales can be attributed to changes occurring on the catalyst surface. After adsorbing the nitrile or amine, the system was purged first with nitrogen followed by hydrogen to test for the reactivity of any remaining surface species. The procedure for data analysis has been discussed in detail elsewhere.8a The analysis starts with a baseline correction with a shape generated with CLS. This allows for focusing on the variations that occur during the BN adsorption/hydrogenation and final reduction treatment steps. After a baseline subtraction, the data is truncated to the desired region of interest. This data can be mean centered if it improves the PLS results observed. The corrected data is introduced into the PLS modeling program that defines the data matrix (A) as a linear combination of the product of the loading vectors and scores plus the matrix containing the residuals. The number of vectors presented in this report was determined by the evaluation of the precision error sum of squares (PRESS), which indicates whether the data is being overestimated or underestimated. In the present case, two vectors were sufficient in all cases to describe the main sources of variations, although the first loading vector typically contained most of the variations. As described previously,8a the use of premixed, very stable, and experimentally designed chemical perturbations to the flow cell system ensures that the independent variables are uncorrelated, allowing for separation of the spectral signatures of unknown chemical species (i.e., surface adsorbates). As described above, in the present case, such perturbations take the forms of nitrile, amine, and hydrogen concentration step profiles in the system.

Results and Discussion As a first step to studying butyronitrile hydrogenation, the adsorption of this reactant on the 5% Pt/Al2O3 catalyst was

3174 Langmuir, Vol. 23, No. 6, 2007

Ortiz-Hernandez and Williams

Figure 3. Raw spectral data for butyronitrile adsorption on 5% Pt/Al2O3 in the 2220-2300 cm-1 region before (dashed) and during exposure to 4 mM (solid line), 10 mM (triangles), and 40 mM (circles) butyronitrile in hexane.

examined. As discussed in the Experimental Section, ATR-IR flow experiments were performed using a combination of single and multiple concentration steps conducted with a variety of time intervals over several hours. Such experiments aid in providing a large amount of spectroscopic data for multivariate analysis, allowing the liquid-phase and adsorption peaks to be distinguished. Figure 3 shows the raw spectral data obtained in the CtN stretching region (∼2200-2300 cm-1) for an experiment that involved alternating the liquid flow over the catalyst between three reservoirs containing 4, 10, and 40 mM butyronitrile (BN) in hexane (see Figure 4d, dashed line, for the nitrile concentration step profile). The spectrum obtained in pure hexane just prior to nitrile exposure (dashed line) reveals a featureless background. However, after switching to 4 mM BN (solid line), a very weak peak is observed at 2256 cm-1 that is associated with a νCtN stretch. This peak varies in size depending on the concentration, as shown in the spectra that were subsequently obtained during the 10 mM (triangles) and 40 mM (circles) BN steps. In examining other areas of the raw spectra, we were unable to distinguish clear peaks associated with butyronitrile. The primary reason for this is that a large number of vibrational peaks associated with BN overlap with the peaks from hexane. The hexane solvent (∼7.6 M) has spectroscopic signals that are 2-3 orders of magnitude greater than those of nitrile (0.004-0.04 M) in this system. A 1% variation in the concentration of hexane at the interface during flow, therefore, results in IR absorption changes that are considerably larger than the spectral variations associated with liquid-phase butyronitrile. As a result, its spectroscopic signature always serves to obscure the νCsH and δCsH vibrations of other species in solution. The exceptions are vibrations whose frequencies are influenced by the presence of the BN cyano group.30a However, these vibrations have roughly half to a quarter of the absorption of the νCtN vibration, which, as Figure 3 shows, is very small, even at the highest concentration. As a result, these peaks are not readily observed in the raw data. Multivariate analysis of the spectral data was performed across the entire spectral range of the experiment (i.e., 1100-4000 cm-1). This analysis revealed the presence of both bulk butyronitrile peaks and surface peaks that were present as variations within the data. The first spectral window to consider is the CtN stretching region corresponding to the raw data in Figure 3. Figure 4a shows the first loading vector obtained in

Figure 4. Multivariate analysis of butyronitrile adsorption on 5% Pt/Al2O3. (a) First loading vector obtained for the 2200-2320 cm-1 range (solid line), (b) first loading vector obtained for the 1480-1750 cm-1 range (diamonds), and (c) first loading vector obtained in the 1300-1350 cm-1 range (triangles). (d) Associated scores (solid line, diamonds, triangles) and experimental liquidphase nitrile profile (dashed line).

ATR-IR Studies of BN Adsorption and Hydrogenation

this region, revealing a 2256 cm-1 peak that corresponds directly to the νCtN stretch12-16 of BN. There is also a small satellite peak at 2296 cm-1 that is associated with the well-known νC-C + δC-H combination band typically observed in the vibrational spectra of liquid nitriles.14,16a The associated score of this loading vector is shown in Figure 4d (solid line), revealing small but discernible steps that correspond directly with the liquid-phase BN concentration profile (Figure 4d, dashed line). Similar behavior was observed in both the 1300-1350 cm-1 and 13801420 cm-1 regions, where δCH2 wagging and δCH3 deformation vibrations occur, respectively. For example, Figure 4c shows the first loading vector obtained for the 1300-1350 cm-1 region, revealing a shape that contains two peaks at 1328 and 1342 cm-1. These vibrations are assigned to the δCH2 wags of BN, with the lower frequency band arising from the methylene group next to the CtN.32 The associated score of this vector (Figure 4d, triangles) also follows the liquid-phase BN concentration profile. We were unable to extract any meaningful vectors from BN contributions in the νC-H stretching region (i.e., 2800-3000 cm-1) due to interference from hexane IR absorption. In the above cases, the second loading vector was found to not contribute significantly to improving the model for the system. This contrasts with our previous results for acetonitrile adsorption on Pt/Al2O3 in hexane,8a where a second loading vector in the CtN stretching region suggested the presence of adsorbed endon acetonitrile. The contributions took the form of blue-shifted spectral shoulders that appeared on the νCtN and νCsC + νCsH combination bands and which had been observed previously in vibrational studies of acetonitrile on Pt.12-17 While we cannot rule out that some of this species may be present for butyronitrile, it clearly is not as prominent a surface species. Multivariate analysis in the 1500-1750 cm-1 spectral region revealed significant variations that appear to be associated with an adsorbed species. The first loading vector (Figure 4b) shows a peak centered at 1635 cm-1 with a satellite at ∼1520 cm-1. The associated score shown in Figure 4d (diamonds) shows a response that clearly correlates strongly with the presence of BN in solution. However, in contrast to the squarelike score profiles for the other two regions, the score progression clearly suggests an adsorption and desorption process over the course of the experiment. In our previous study of acetonitrile adsorption on Pt/Al2O3 in hexane, we observed a similar feature at 1641 cm-1 that was assigned to an adsorbed nitrile species on platinum that exhibited both end-on and π-bonded characteristics. The assignment was based largely on an array of ultrahigh vacuum (UHV)13-15 and electrochemical16 vibrational studies of acetonitrile adsorption on platinum single crystals. This species was similar in structure to an imine species that has been proposed as an intermediate during nitrile hydrogenation. Therefore, the present feature at 1635 cm-1 is assigned to a similar imine-type species adsorbed on the catalyst surface, formed from residual adsorbed hydrogen from the pretreatment steps. The origin of the feature that appears in the 1520-1545 cm-1 range will be addressed after consideration of the butyronitrile hydrogenation data. It should be noted that water has infrared absorption located at ∼1630-1650 cm-1 and at 3500-3600 cm-1. However, several pieces of evidence argue (12) Sexton, B. A.; Avery, N. R. Surf. Sci. 1983, 129, 21. (13) Ou, E. C.; Young, P. A.; Norton, P. R. Surf. Sci. 1992, 277, 123. (14) Krtil, P.; Kavan, L.; Nova´k, P. J. Electrochem. Soc. 1993, 140, 3390. (15) Hubbard, A. T.; Cao, E. Y.; Stern, D. A. Electrochim. Acta 1994, 39 (8/9), 1007. (16) (a) Marinkovic´, N. S.; Hecht, M.; Loring, J. S.; Fawcett, W. R. Electrochim. Acta 1996, 41 (5), 641. (b) Morin, S.; Conway, B. E.; Edens, G. J.; Weaver, M. J. J. Electroanal. Chem. 1997, 421, 213. (17) Brunner, E. J. Chem. Eng. Data 1985, 30, 269.

Langmuir, Vol. 23, No. 6, 2007 3175

against the assignment of the 1635 cm-1 feature to interfacial water from surface oxide hydrogenation. First, this feature is not present during the analysis of the pretreatment periods where only hydrogen is flowing. Second, the feature is present during BN adsorption in the absence of H2. Finally, in the case of bare Al2O3 films, this feature is not present during either BN adsorption or hydrogenation experiments. Nevertheless, our catalyst has extensive exposure to air prior to the start of the experiment, and it is likely that some residual oxygen remains on the catalyst surface even after H2 pretreatment. Thus, the possibility of this surface oxygen playing a role in the adsorption of BN cannot be ruled out in the present experiments. To probe the effects of hydrogen on this system, experiments were performed that involved the simultaneous introduction of H2 and BN over the catalyst surface. Based on the results obtained for BN adsorption, a concentration of 40 mM was chosen for such hydrogenation experiments. After the standard reduction treatment, two nitrile/H2 steps were used followed by a final H2 step to study the reactivity of any persistent surface species. The concentration of hydrogen in hexane at room temperature can be estimated at ∼5 mM based on solubility data available in the literature.17 Given the 8:1 BN/H2 ratio, low temperature, and small amount (∼10 mg) of catalyst present in the system, the conversion in the reactor is very low. For example, based on kinetic studies of liquid-phase BN hydrogenation on Pt/NaY at 110 °C and 24 bar conducted by Huang et al., we would estimate the conversion over the entire course of our longest experiment (i.e., 6 h) to be less than 1%.5 Figure 5a shows the first loading vector obtained in the CtN stretching region, revealing a shape very similar to that observed during the adsorption of BN (cf. Figure 4a). Again, this peak is representative of the νCtN stretch of BN, and its score versus time profile (Figure 5c, squares) suggests that it is largely due to the liquid-phase nitrile (Figure 5c, dashed line). The upward sloping behavior of the score during the step may suggest that there is some adsorption under these conditions as indicated by the decaying profile after switching back to pure hexane. However, the adsorbed species appears to be weakly bound. This may explain why the second loading vector obtained from this region (not shown) does not contribute significantly to the variations or show any evidence for an adsorption profile. Analysis of the 1500-1750 cm-1 region shows qualitatively different behavior in the presence of hydrogen when compared with straight BN adsorption. The first loading vector is shown in Figure 5b, revealing a band at 1641 cm-1 that correlates with the presence of the H2-saturated nitrile solution. This band is upshifted slightly from what was observed in the absence of H2 (Figure 4c), and the satellite peak is also upshifted to 1544 cm-1. The associated score profile shown in Figure 5c (solid line) reveals that these peaks also exhibit a different time dependence. The main difference is that the formation of this surface species goes quickly through a maximum within the first 10 min of each nitrile/H2 step, followed by a gradual decline over the remaining 20 min. In addition, there appears to be substantially more accumulated species on the catalyst surface, as indicated by the fact that the score does not return to the very flat baseline. Finally, a comparison of the loading vectors and scores between BN adsorption (cf. Figure 4b and d) and BN hydrogenation (Figure 5b and c) indicates that the coverage of this species is larger in the latter case. This species also appears to be reactive toward hydrogen, since it is readily removed upon exposure in the absence of BN in solution (Figure 5c, arrow). As in the case of adsorption, the second loading vector does not contribute significantly to the modeling of the variations in this spectral region.

3176 Langmuir, Vol. 23, No. 6, 2007

Ortiz-Hernandez and Williams

Figure 6. Multivariate analysis of butyronitrile hydrogenation on 5% Pt/Al2O3. (a) First loading vector obtained for the 3000-3500 cm-1 range (line) and second loading vector (squares). (b) Associated scores and experimental liquid-phase nitrile profile.

Figure 5. Multivariate analysis of butyronitrile hydrogenation on 5% Pt/Al2O3. (a) First loading vector obtained for the 2200-2320 cm-1 range (squares) and (b) first loading vector obtained for the 1500-1750 cm-1 range (line). (c) Associated score profiles and experimental liquid-phase nitrile profile. The arrow represents the introduction of hydrogen after removing BN from the system.

The enhancement of the 1641 cm-1 peak in the presence of H2-saturated BN lends support to its assignment to an iminetype intermediate. The assignment is bolstered further still by an analysis of the N-H stretching region. Figure 6a shows the first (solid line) and second (squares) loading vectors obtained in the 3000-3500 cm-1 range. These two vectors are complex, with both vectors contributing significantly to the variations observed in this region. The first loading vector has broad features at 3390 and 3247 cm-1, which are attributed to the antisymmetric and symmetric νN-H stretches, respectively, of the primary amine.30 The peak at 3247 cm-1 is also similar in frequency to the characteristic N-H stretches for methyleneimine33 and ethyleneimine,35 and thus it is assigned to the N-H stretch of the adsorbed imine species. The second loading vector shows bands

at 3200 and 3066 cm-1 and also exhibits a derivative line shape that, in multivariate analysis, serves to model changes in peak positions.36 The peak at 3066 cm-1 is consistent with a combination band consisting of νCdN + δCH2, while the 3200 cm-1 feature likely arises from the overtone of the νCdN, consistent with previous vibrational studies of alkylenimines.34 The scores for these vectors (Figure 6b) show considerable correlation with the introduction of H2-saturated nitrile, with the first loading vector score being the most similar to that of the imine-type species. However, there are fluctuations in the scores that suggest the complexities within this spectral region. In this case, we have loading vectors with very broad peaks that are overlapping, as well as some apparent shift in frequency indicated by the derivative shape. Nevertheless, the analysis in this region appears to suggest the presence of amine and imine species on the catalyst surface during hydrogenation. Furthermore, the size of the variations in the imine-related bands is very small, providing an explanation for the absence of this feature in the experiments performed under adsorption conditions (cf. Figure 4). The transient behavior of the imine species was probed by performing experiments with longer steps. For example, Figure 7 shows multivariate analysis results in the 1550-1750 cm-1 region for an experiment involving a BN/H2 step of 3 h, an H2 step of 1 h, a BN/H2 step of 2 h, and then a hydrogen step of 1 h. The results show the transient behavior, with a steady decrease over the entire course of the longer step that fails to reach a steady value. However, the intermediate hydrogen treatment removes the species from the surface (as described previously)

ATR-IR Studies of BN Adsorption and Hydrogenation

Figure 7. Multivariate analysis of butyronitrile hydrogenation on 5% Pt/Al2O3. (a) First loading vector obtained for the 1550-1750 cm-1 range (squares). (b) Associated score and experimental liquidphase nitrile profile. The two downward arrows indicate where hydrogen-saturated hexane was flushed through the system.

and allows for the restoration of the original transient behavior. Given the presence of an amine species that is indicated by the analysis of the N-H stretching region, it is perhaps possible that this product is being formed by the gradual decrease of the imine over time. The presence of amine on the surface may also be supported by the small peak that appears between 1520 and 1545 cm-1 during both BN adsorption and hydrogenation experiments. There are several possible assignments for this feature that could be consistent with those from available adsorption studies under UHV conditions. Trenary and co-workers18-27 have investigated the adsorption and decomposition of cyanides using reflection absorption infrared spectroscopy (RAIRS). These RAIRS studies show vibrational bands at 1520 (C-N) and 1565 (NH2 deformation) cm-1, which where assigned to the formation of aminocarbynes from various molecules (e.g., methylamine, dimethylamine, and trimethylamine). To detect the NH2 deformation band, the molecule must be adsorbed in a tilted or flat configuration based on IR surface selection rules for metals. In a similar fashion, Erley et al.28 studied the decomposition of (18) Jentz, D.; Celio, H.; Mills, P.; Trenary, M. Surf. Sci. 1995, 335, 1. (19) Jentz, D.; Mills, P.; Celio, H.; Belyansky, M.; Trenary, M. J. Phys. Chem. 1996, 105, 3250. (20) Jentz, D.; Mills, P.; Celio, H.; Trenary, M. Surf. Sci. 1996, 368, 354. (21) Mills, P.; Jentz, D.; Trenary, M. Surf. Sci. 1996, 368, 348. (22) Mills, P.; Jentz, D.; Trenary, M. J. Am. Chem. Soc. 1997, 119, 9002. (23) Kang, D. H.; Trenary, M. Surf. Sci. 2000, 470, L13. (24) Kang, D. H.; Trenary, M. J. Am. Chem. Soc. 2001, 123, 8432. (25) Kang, D. H.; Trenary, M. Surf. Sci. 2002, 519, 40. (26) Herceg, E.; Trenary, M. J. Phys. Chem. B 2005, 109, 17560. (27) Herceg, E.; Mudiyanselage, K.; Trenary, M. J. Phys. Chem. B 2005, 109, 2828.

Langmuir, Vol. 23, No. 6, 2007 3177

Figure 8. Multivariate analysis of butylamine adsorption in the presence of H2 on 5% Pt/Al2O3. (a) First loading vector obtained for the 1500-1750 cm-1 range for 5% Pt/Al2O3 (solid line) and Al2O3 (squares). (b) Associated score profiles and experimental liquidphase butylamine profile.

trimethylamine on Pt(111) using high-resolution electron energy loss spectroscopy (HREELS). A band at 1510-1520 cm-1 was reported and assigned to H2CdN adsorbed to Pt through the nitrogen. The assignment was verified through isotopic labeling with deuterium. Rasko´ et al. studied the adsorption and reaction of acetonitrile on Al2O3-supported catalysts.29 Evidence for acetonitrile hydrogenation in the form of methylamine was observed from temperature-programmed desorption (TPD) and temperature-programmed reduction (TPR) experiments. This amine formation at elevated temperature was accompanied by the appearance of vibrational bands at 1587-1585 and 15121505 cm-1 likely associated with NH2 deformations, as has been suggested in a RAIRS study of propylamine on Pt(111) by Gleason et al.37 To determine whether the 1544 cm-1 feature detected during BN hydrogenation could be associated with an adsorbed amine species, an experiment involving two step pulses of H2-saturated 40 mM butylamine (BA) solution was performed. Figure 8a shows the first loading vector obtained in the 1500-1750 cm-1 region for such an experiment. The vector reveals a clear band at 1535 cm-1, which is not present in the bulk liquid-phase spectra (28) Erley, W.; Xu, R.; Hemminger, J. C. Surf. Sci. 1997, 389, 272. (29) Rasko´, J.; Kiss, J. Appl. Catal., A. 2006, 298, 115. (30) (a) Lin-Vien, D.; Colthup, N. B.; Fateley, W. G.; Grasselli, J. G. The Handbook of Infrared and Raman Characteristic Frequencies of Organic Molecules; Academic Press Limited: San Diego, CA, 1991; Chapters 2 and 10 and references within. (b) Bell, F. K. Absorption Spectra of Alkyl Amines, 1927, 1837.

3178 Langmuir, Vol. 23, No. 6, 2007

Ortiz-Hernandez and Williams

of BA. At the same time, a second band appears as a shoulder at 1619 cm-1, which is assigned to the δNH2 scissoring vibration of the amine. The score versus time profile (Figure 8b) shows that these bands correlate with the presence of BA in solution, and together they exhibit some degree of absorptive behavior. For comparison purposes, the adsorption on an Al2O3 film was also tested. The results reveal two bands at 1548 and 1619 cm-1. The bands correlate with the presence of amine in the system, and the score exhibits a clear liquid-phase behavior. It is wellknown that n-butylamine strongly adsorbs to Al2O3 through a Lewis-type configuration.38 The time-dependent profile clearly shows that butylamine is not reactive under room temperature and pressure and that full desorption is obtained after removing the amine from solution. Given these results, and the available literature discussed above, we tentatively assign the small feature at 1544 cm-1 during BN hydrogenation to the δNH2 band of an adsorbed amine. The presence of a band at 1535 cm-1 in 5% Pt/Al2O3 and at 1549 cm-1 on the Al2O3 support suggests that the amine can interact with the catalyst through a Lewis or Brønsted acid-base interaction. A Brønsted interaction of the amine through the NH2 with the Al2O3 surface to give an RNH3+ bonding is also possible. The acidity of Al2O3 has been widely studied using various probe molecules, with NH3 being one of the most widely used. For example, Peri et al.39 looked at the adsorption of NH3 on dried Al2O3. According to that work, adsorbed NH3 produces IR vibrations at 3400, 3355, 3100, and 1620 cm-1 with some additional bands at 1560 and 1510 cm-1. The peak at 1510 cm-1 was assigned to the δNH2 vibration. The bands for NH3+ deformation are expected in the range of 1485-1550 cm-1, and the NH3+ stretching appears in the 3030-3130 cm-1 range.39 The presence of the variations obtained for butyronitrile hydrogenation in the 1544 cm-1 region and the presence of the bands observed in the 3000 cm-1 region, therefore, suggest the formation of amine adsorbed on the Al2O3 support through Brønsted-type interactions. In the present case, the exposure of the catalyst to humid air during the growth of the film may lead to the presence of hydroxyl groups in the support, resulting in increased availability of Brønsted sites at the catalytic surface. Finally, the formation of secondary and/or tertiary amines was considered. In a previous ATR-IR study of the hydrogenation of butyronitrile on Raney metals, Chojecki31 detected the formation of butylidene-butylamine on Raney-Ni, with bands appearing between 1670 and 1660 cm-1 in the liquid phase. This

species reacted to form di-n-butylamine, with a characteristic vibrational band at 1153 cm-1 corresponding to C-N-C bending. In the present study, variations were observed in the 1100-1200 cm-1 region, including what appeared to be a shoulder at 1150 cm-1 (data not shown). However, due to interference with a larger feature at 1155 cm-1 associated with the Al2O3 support, the clear assignment of this variation to a secondary or tertiary amine was not possible.

(31) Chojecki, A. Selective Hydrogenation of Butyronitrile over Raney-Metals. Ph.D. Dissertation, Technischen Universita¨t Mu¨nchen, 2004. (32) Durig, J. H.; Drew, B. R.; Koomer, A.; Bell, S. Phys. Chem. Chem. Phys. 2001, 3, 766. (33) Halonen, L.; Deeley, C. M.; Mills, I. M. J. Chem. Phys. 1986, 85 (2), 692. (34) Pouchan, C.; Zaki, K. J. Chem. Phys. 1997, 107, 342. (35) Hashiguchi, K.; Hamada, Y.; Koga, T. Y. M.; Kondo, S. J. Mol. Spectrosc. 1984, 105, 81. (36) (a) Otto, M. Chemometrics; Wiley-VCH: New York, 1999. (b) Beebe, K. R.; Pell, R. J.; Seasholtz, M. B. Chemometrics: A Practical Guide; John Wiley and Sons: New York, 1998. (c) Haaland, D. M.; Melgaard, D. K. Appl. Spectrosc. 2001, 55, 1. (d) Wold, S.; Sjo¨stro¨m, M.; Erikson, L. Chemom. Intell. Lab. Syst. 2001, 58, 109.

Acknowledgment. The authors would like to thank the National Science Foundation for its generous funding of this project through a CAREER award (CTS-0093695). I.O. thanks the Sloan Foundation for a graduate fellowship.

Concluding Remarks The adsorption and hydrogenation of butyronitrile (BN) in hexane on a 5% Pt/Al2O3 catalyst has been studied using in situ attenuated total reflection infrared (ATR-IR) spectroscopy. The presence of an adsorbed imine intermediate was suggested by the appearance of bands in the CdN and NsH stretching regions. In addition, this species correlates well with the formation of liquid-phase and adsorbed primary amine species. Using this information and that for acetonitrile reported previously,8a it is interesting to speculate on some important steps that may be present in the reaction mechanism. Upon adsorption onto the surface, a nitrile can undergo partial hydrogenation to a surface imine species. As shown here, the imine is reactive toward H2, likely toward the formation of adsorbed amine with the NH2 pendent from the surface, allowing for observation by infrared spectroscopy. The remaining surface species is then attached through the carbon atom, which is hydrogenated to give rise to the primary amine. Once the amine forms, it can adsorb to the Al2O3 support where it could, under the right conditions, undergo further reaction with the adjacent adsorbed imine species on the platinum active sites to form secondary amines. Further mechanistic insight may be gained through the expansion of the in situ vibrational studies presented here. The first step would be to perform experiments at higher H2 pressures and temperatures, where we would expect the behavior of the adsorbed species to be affected. Second, expanding the hydrogenation studies to other nitriles such as acetonitrile, propionitrile, and benzonitrile and the use of other solvents (e.g., ethanol and water) will provide several points of comparison for vibrational studies. Finally, simultaneous ATR-IR spectroscopy with liquidphase kinetic measurements (i.e., operando spectroscopy) will allow for direct correlations of surface species with product formation to be drawn. Such studies are currently planned in our laboratory.

LA062502H (37) Gleason, N. R.; Jenks, C. J.; French, C. R.; Bent, B. E.; Zaera, F. Surf. Sci. 1998, 405, 238. (38) Sokoll, R.; Hobert, H. J. Catal. 1990, 125, 285. (39) Peri, J. B. J. Phys. Chem. 1965, 69, 231.