Reports: ND551503-ND5: Correlating Molecular Environment with Performance in Immobilized Green Catalysts Using Nonlinear Optical Techniques

Julianne M. Gibbs-Davis, PhD, University of Alberta

Immobilized catalysts have great promise in green chemistry as they can be easily removed from the reaction mixture and reused owing to their surface immobilization.  Combinations of types of catalysts also allow for cooperative catalysis whereby both the nucleophilic reactant and electrophilic reactant are simultaneously activated enhancing the reaction rate and enantioselectivity.  In the first year of our New Directions grant, we utilized second harmonic generation (SHG) spectroscopy to explore components of a bifunctional immobilized organocatalytic system consisting of a Lewis basic amine (for activating nucleophiles) and a Lewis acidic urea (for activating electrophiles).  Using SHG we found that a model reactant ketone (4-nitroacetophenone) binds to both Lewis basic and Lewis acidic monolayers. By varying the concentration of the ketone, the binding constant for each monolayer was determined.  In the second year of our grant, we have refined our analysis of the SHG data to better explain ketone binding to the Lewis acidic urea monolayers.  We have also utilized another second-order nonlinear optical technique, vibrational sum frequency generation or SFG, to shed light on why the urea monolayer exhibited cooperative binding of the ketone.  Finally, we have employed SFG to explore the reactivity of another important system to materials chemistry: the copper-catalyzed azide-alkyne cycloaddition, or the so-called click reaction.  This work reveals that the surface sensitivity and molecular specificity of these techniques are exceptionally well suited to characterize functionalized surfaces relevant to catalysis.

Unlike the Lewis basic 3-aminopropylsilane monolayer (APS) and 4-aminophenylsilane (APhS) monolayer, the Lewis acidic urea monolayer (ureidopropylsilane or UDPS) exhibited an s-shaped binding isotherm with the electrophilic ketone (Figure 1A). The Frumkin-Fowler-Guggenheim (FFG) model was hence fit to the UDPS monolayer data as the FFG model incorporates a surface coverage dependent term that stems from cooperative interactions between neighboring adsorbates.  The resulting binding affinity obtained from the fit was far lower for the UDPS than the Lewis basic monolayers when the noncooperative binding constant for UDPS (surface coverage 0%) was compared with the binding constants obtained using the Langmuir model for the other monolayers (Figure 1B). This low binding affinity was surprising as the UDPS was expected to form a stronger interaction with the ketone through multiple hydrogen bonds. However, the binding affinity of the UDPS monolayer was much greater as the surface coverage of ketone approached saturation (100%).  From these results, we proposed that the urea groups in the monolayer tended to form a hydrogen-bonded network with one another, which the ketone molecule had to break to bind a urea group. This requirement made binding energetically costly at low surface coverage, but became increasingly easier as the network was disrupted resulting in an increased binding affinity.

Figure 1.

To gain molecular insight into the interaction between the urea group and the ketone, broadband vibrational SFG was employed to study the N-H stretch of the former. Figure 2A illustrates the SFG spectra in the N-H stretching region at the UDPS-functionalized silica/d3-ACN interface in the absence of ketone. The spectral fit identified three peaks with frequencies of 3282, 3344, and 3429 cm-1, which were assigned to the N-H stretch for the disubstituted nitrogen in the urea and the symmetric and asymmetric stretch of the terminal NH2 group. Upon adding ketone, however, the amplitudes of the peaks and the overall shape of the spectrum changed significantly.  Specifically, figure 2B and 2C illustrates the SFG spectra with the presence of 40 mM and 200 mM ketone, respectively. From the fit, two new peaks became apparent: one at 3373 cm-1, the other at 3472 cm-1, which were assigned to the symmetric and the asymmetric stretch of the NH2 bound to the ketone, respectively.  The presence of new peaks attributed to the ketone-bound urea supports that binding significantly changed the local environment of the urea. Moreover, the shift to higher wavenumbers indicated that weaker hydrogen bonds formed with the ketone than with neighboring urea sites, as increased hydrogen-bonding corresponds to lower vibrational frequencies.  This shift supports our hypothesis that the urea groups form a stable hydrogen-bonded network that the ketone disrupts upon binding.  We also observed that the peak amplitude changed sign for the bound versus unbound NH2 groups, resulting in deconstructive interference in the spectra, which supports that binding to the ketone led to significant reorientation of the urea's terminal NH2 group.  This investigation using SFG and SHG represents the first of its kind for studying the interactions of well-defined immobilized organocatalysts to better understand how the intrinsic affinity of the immobilized catalyst for the reactants influences the overall activity.  A manuscript on this work is currently being prepared.

Figure 2.

We have also begun exploring the immobilization step using SFG spectroscopy.  One of the most common methods for immobilizing functional groups like catalysts to surfaces involves using the high-yielding copper-catalyzed 1,3-dipolar azide-alkyne cycloaddition, one of the "click reactions" (Figure 3A).  Upon measuring SFG from the azide monolayer on silica, which is the precursor for immobilization by this reaction, we found that the azide functional group was very SFG active, which had not been established before.  Indeed, the benzyl azide monolayer exhibited a prominent peak at 2096 cm-1 that was nine times stronger than the CN group often used as an SFG label.

Figure 3.

The click reaction with a cyano-substituted alkyne was then probed based on the decrease of the azide peak and the increase of the cyano peak (2234 cm-1) (Figure 3B). Both the disappearance of the azide and appearance of the cyanide fit well to a first order rate law, indicating that the alkyne in solution was in great excess compared with the azide on the surface resulting in pseudo-first order kinetics. From varying the copper catalyst concentration, we found that the reaction order for the copper ion was 2.1. This value agreed with what has been observed in solution at comparable copper concentrations, which suggests that the mechanism on the surface is similar to that in solution, despite the confined environment of the interface (Li, Weeraman, Gibbs-Davis, ChemPhysChem, 2014).

(Word Count = 993)