Encapsulating Sub-Site Analogues of the [FeFe]- Hydrogenases in Micelles Enables Direct Water Interactions. Robby Fritzsch,1 Owen Brady,1 Elaine Adair,1 Joseph A. Wright,2 Christopher J Pickett,2 and Neil T. Hunt1* 1) Department of Physics, University of Strathclyde, SUPA, 107 Rottenrow East, Glasgow, UK 2) Energy Materials Laboratory, School of Chemistry, University of East Anglia, Norwich, NR4 7TJ, UK AUTHOR INFORMATION Corresponding Author *[email protected] 1 ABSTRACT Encapsulation of sub-site analogues of the [FeFe]-hydrogenase enzymes in supramolecular structures has been shown to dramatically increase their catalytic ability, but the molecular basis for this enhancement remains unclear. We report the results of experiments employing infrared absorption, ultrafast infrared pump-probe and 2D-IR spectroscopy to investigate the molecular environment of Fe (pdt)(CO) (pdt: propanedithiolate) [1] encapsulated in the dispersed alkane 2 6 phase of a heptane-dodecyltrimethylammonium bromide-water microemulsion. It is demonstrated that 1 is partitioned between two molecular environments, one that closely resembles bulk heptane solution and a second that features direct hydrogen-bonding interactions with water molecules that penetrate the surfactant shell. Our results demonstrate that the extent of water access to the normally water-insoluble sub-site analogue 1 can be tuned with micelle size, while IR spectroscopy provides a straightforward tool that can be used to measure and fine- tune the chemical environment of catalyst species in self-assembled structures. TOC GRAPHICS KEYWORDS [FeFe] hydrogenase; infrared spectroscopy, ultrafast infrared spectroscopy, micelle, water solubility 2 The ability of synthetic analogues of the active subsite of the [FeFe]-hydrogenase ([FeFe]-H ase) 2 enzymes to catalyze the production of molecular hydrogen has generated considerable interest, owing to the belief that they could offer an economically-viable alternative to the rare-metal catalysts that are the current technological state-of-the-art.1 A range of molecular architectures have been reported with varying catalytic abilities, but several challenges must be overcome before these systems can bridge the gap to technological applications.2-6 In particular, even the most effective synthetic [FeFe]-H ase analogues are yet to match the efficiency of the natural 2 enzyme, while the water insolubility and oxygen sensitivity of the analogues can also be prohibitive towards their development as robust fuel sources. Furthermore, in order to achieve catalytic turnover and the desirable light activated H production from water, it is necessary to 2 combine the sub-site analogue with a photosensitizer and sacrificial electron donor to create a functional system, although self-sensitizing systems have been reported recently. 7 Encapsulation of [FeFe]-H ase subsite analogues in self-assembled or supramolecular systems 2 has been shown to improve stability and lead to increases in catalytic efficiency.8-12 Examples include the use of polymeric micelles to form ‘microreactors’,8 encapsulation in gels or resins,1,13 oligosaccharides11 and dendrimers.1 The reasons cited for the increased efficiency range from co- localization of the components to second coordination sphere H-bonding interactions with the sub-site analogue that resemble those of the natural enzyme. A factor that is likely to be equally significant, but not widely considered, is the ability of such structures to place the catalytic species in an environment where it is stably in contact with aqueous solution and so more readily accessible by the protons needed for the H evolution reaction. This function is achieved by 2 hydrophobic channels in the natural enzyme and must be replicated in any supramolecular or self-assembled system. 3 Infrared (IR) spectroscopy is a sensitive probe of both the structure and chemical environment of [FeFe]-H ase sub-site analogues.14 Here, we employ IR absorption spectroscopy, ultrafast IR 2 pump-probe and 2D-IR spectroscopy to evaluate the effect of encapsulating the [FeFe]-H ase 2 subsite analogue Fe (pdt)(CO) (pdt = propanedithiolate), 1, in micelles formed by the surfactant 2 6 dodecyltrimethyl ammonium bromide (DTAB) in water. These oil-in-water microemulsions, in which the dispersed phase is a solution of 1 in heptane, provide a model system with which to gauge the ability of water to penetrate self-assembled structures and so to provide an insight into the likely role of this behavior in enhancing the catalytic function of 1. Our results show that, not only can micelles be used to enable water access to the catalytic centers, but that IR absorption spectroscopy is a simple and reliable tool to evaluate water access during the materials design process. The IR spectrum of 1 in heptane solution displays five peaks attributable to CO ligand stretching vibrations at 2075, 2036, 2006, 1991 and 1982 cm-1, in agreement with previous studies (Fig.1(a)).15 This heptane solution formed the dispersed phase of heptane-DTAB-water microemulsions, as specified in Table 1, that have been shown to produce near spherical micelles with little or no inter-micelle interactions.16-18 In all cases, there is estimated to be less than one molecule of 1 per micelle (Table 1). Micelle encapsulation results in the formation of a shoulder on the low frequency side of the peaks at 2075 and 2036 cm-1 and a broad, featureless component in the region of the three closely-spaced peaks near 2006-1982 cm-1 (see arrows, Fig. 1(a)). It is noteworthy that a similar 4 Figure 1: (a) FT-IR spectra of 1 in heptane solution (blue) and encapsulated in microemulsion ‘6’ (black, see Table 1 for sample details). Arrows mark the additional shoulders observed in the spectra of microemulsion samples. (b) Example of fit to microemulsion ‘6’ FT-IR spectrum using sum of five Lorentzian (blue) and five Gaussian (red) lineshape functions. The fit residual is shown in light grey, offset slightly for visibility. Similar results were obtained for the other five microemulsion samples studied. A heptane solution of 1 required only the five Lorentzian functions to produce an accurate fit. (c) FT-IR spectra showing effect of increasing dispersed phase content of the microemulsion on the shoulder amplitude. This is also shown in (d) using the area ratio of the Gaussian (shoulder) to the Lorentzian (main peak) function for the 2036 cm- 1 peak. 5 Sample V wt % R n n disp.phase disp. phase micelle, est. 1 heptane (µl) (nm) (per micelle) (per micelle) 1 8 0.55 1.64 0.03 4 2 16 1.10 1.67 0.07 8 3 24 1.64 1.70 0.10 12 4 32 2.17 1.73 0.13 17 5 40 2.70 1.75 0.17 21 6 48 3.23 1.78 0.20 25 Table 1: Details of microemulsion samples discussed in text. Samples were made up by adding the required volume of a 55 mM stock solution of 1 in heptane (V ) to a 20wt% disp. phase DTAB/water-solution and allowing the mixture to equilibrate over 24h. This is also shown as a wt% of the dispersed phase. Estimates of the micellar radius (R ) and the population micelle, est numbers of 1 and heptane (n and n respectively) are based on an assumed surfactant 1 heptane aggregation number per micelle of 50 as per Ref 18. effect was also apparent in IR spectra of samples of 1 encapsulated in polymeric micelles that demonstrated enhanced catalytic activity, though this was not discussed in detail by the authors.8 Control experiments showed that 1 was insoluble in water in the absence of surfactant (as determined by the sensitivity limit of IR absorption spectroscopy), while inclusion of surfactant into heptane solutions of 1 had no impact upon the IR spectrum. Adding 1 to samples featuring a 50:50 v/v mixture of water and heptane produced a biphasic system in which the IR spectrum of 1 could only be detected in the heptane layer. This spectrum exactly matched that observed in pure heptane solution. These experiments confirm that the new features apparent in the IR 6 spectrum of 1 are due to micelle encapsulation within an aqueous continuous phase and are not due to water saturation of the heptane phase. Lineshape fitting of the IR absorption spectra showed that the spectrum of 1 in heptane solution is well-represented by five Lorentzian lineshapes corresponding to each of the CO stretching vibrational modes. To obtain the same quality of fit to the IR spectra of the microemulsion samples these Lorentzian peaks had to be augmented by five additional Gaussian lineshapes, each broader (~7 vs ~2.5 cm-1 FWHM) and shifted to the low frequency relative to the corresponding Lorentzian (Fig 1(b)). The quality of fit obtained using this approach could not be improved by the use of more complex Voigt profiles. Increasing the quantity of dispersed phase added to the microemulsion to swell the micelle size (Table 1) resulted in a progressive increase in the amplitude of the shoulder relative to the main peak. This continued up to a dispersed phase content of 2 wt% where a plateau was reached (Fig.1(c)). By contrast, increasing the concentration of 1 for constant volume of dispersed phase had no effect on the spectrum of 1. Ultrafast IR pump-probe experiments were carried out on the heptane solution and two representative microemulsion samples (Samples ‘3’ and ‘6’ from Table 1), chosen to provide a change in micelle size for a given dispersed phase concentration of 1. The pump-probe spectra of the heptane solution exhibited peaks assignable to bleaches (negative) and transient absorptions (positive) originating from the v=0-1 and v=1-2 transitions of the CO stretching vibrations of 1 respectively (Fig. 2(a)), in good agreement with prior publications.15 The spectra of both microemulsion samples were broadly similar, though with noticeable broadening to the low frequency side of the peaks consistent with the shoulders observed in the absorption spectra. 7 Figure 2: (a) Representative IR pump-probe spectra of 1 encapsulated in heptane at a range of pump-probe delay times, as shown in the figure legend. (b) Temporal decay of peak amplitudes taken from (a) with fits to exponential functions shown in grey. The legend identifies specific peaks in the spectrum by frequency. (c) Example of fitting IR pump-probe spectrum of 1 in heptane solution to a single exponential function. Experimental data (black dotted line) is compared with the fit result showing the amplitude of a single exponential decay function with a decay timescale of 120 ps (blue line). The FT-IR spectrum is shown for comparison (dk grey). The fit residual is shown (lt grey) but offset for clarity. (d) Results of fitting the IR pump-probe spectrum of microemulsion ‘6’ to a bi-exponential function, showing the amplitudes of the 44 ps (red) and 120 ps (blue) components. The fit residual is shown (lt grey) but offset for clarity. Similar results were obtained for microemulsion ‘3’. 8 Global fitting of the temporal dynamics of the pump-probe data to both single and bi- exponential functions was used to quantify the vibrational relaxation dynamics of the modes of 1. In both heptane solution and the microemulsion samples, a very fast decay component was observed on sub-ps timescales that was attributed to the instrument response and so fitting was started at a pump-probe delay of 2 ps (Fig 2(b)). For the heptane solution, the data were well- represented by a single decay timescale of 120 ps, the amplitude of which closely matched the peaks in the pump-probe spectra (Fig.2(c)). This agrees well with previous reports of the vibrational lifetime of the carbonyl modes of 1 in alkane solution showing that all modes relax with a similar vibrational lifetime. 15 By contrast, a bi-exponential function was required to adequately represent the microemulsion data, returning decay timescales of 44 and 120 ps irrespective of micelle size (Fig. 2(d)). Applying a bi-exponential fit to the heptane solution data led to a collapse of this function to a single exponential result. Examination of the amplitudes of the two components in the microemulsion data showed that the 120 ps decay timescale corresponds to the peaks observed in the heptane spectrum, while the 44 ps lifetime is primarily associated with the shoulders and the broad feature that are exclusive to the encapsulated system (Fig 2(d)). The relative amplitude of the 44 ps component was found to increase slightly with increased micelle size, but pump-probe experiments are less sensitive than IR absorption and the change in amplitude of the 44 ps component was comparable to the anticipated error in the measurement. 9 Figure 3: 2D-IR spectra of (a) 1 in heptane solution at a waiting time of 5 ps. The region between 1980 and 2050 cm-1 is shown for clarity. Blue lines show position of peaks in IR absorption spectra. (b&c) 2D-IR spectra of 1 encapsulated in microemulsion ‘6’ at waiting times of (b) 5 ps and (c) 20 ps. The colour scale runs from negative (red) to positive (blue). Black arrows show the additional diagonal features arising from microemulsion encapsulation. Red arrows show the absence of off-diagonal peaks that would indicate coupling or chemical exchange between the main diagonal lineshape and shoulder. 2D-IR spectroscopy of the heptane solution revealed diagonal peaks corresponding to each of the bands in the IR absorption spectrum of 1 that were linked by off-diagonal peaks (Fig. 3). This is consistent with observations that the CO-stretching modes of 1 are vibrationally coupled. 15 Spectra obtained on microemulsion ‘6’ were similar, but the circular diagonal peak shapes of the heptane spectra are augmented by diagonally-elongated features, which correspond exactly to the shoulders observed in the IR absorption spectrum (see black arrows, Fig. 3). Experiments on other microemulsion samples produced similar results, but returned a lower signal to noise ratio. Experiments as a function of 2D-IR waiting time showed that these shoulders evolve slightly over the first 20 ps of the waiting time to a somewhat more circular shape. This can be seen in 10