Enantiospecific Desorption Triggered by Circularly Polarized Light

Abstract The interest in enantioseparation and enantiopurification of chiral molecules has been drastically increasing over the past decades, since these are important steps in various disciplines such as pharmaceutical industry, asymmetric catalysis, and chiral sensing. By exposing racemic samples of BINOL (1,1′‐bi‐2‐naphthol) coated onto achiral glass substrates to circularly polarized light, we unambiguously demonstrate that by controlling the handedness of circularly polarized light, preferential desorption of enantiomers can be achieved. There are currently no mechanisms known that would describe this phenomenon. Our observation together with a simplified phenomenological model suggests that the process of laser desorption needs to be further developed and the contribution of quantum mechanical processes should be revisited to account for these data. Asymmetric laser desorption provides us with a contamination‐free technique for the enantioenrichment of chiral compounds.

The separation of enantiomers is of great interest since more than 50 %o fp harmaceutically active ingredients are chiral, and nine of the top 10 drugs with respect to worldwide sales value have chiral active ingredients.A lthough they have the same atomic connectivity,e nantiomers of most chiral ingredients exhibit markedly different biological activities.T herefore,itisimportant to promote the enantiomeric enrichment of racemic drugs in order to reduce the amount or eliminate the inactive or even harmful enantiomer in the mixture. [1] Accordingly,the industrial need for affordable enantioenrichment methods [2] and the academic demand for high-resolution studies of chiral molecules in the gas phase [3] and at surfaces, [4] have led to ar apid growth in interest in laser desorption (LD). [5] Herein, we show that by controlling the handedness of circularly polarized light, and hence introducing agradient between the excitation of the enantiomers,p referential desorption of enantiomers from achiral surfaces can be achieved. This observation suggests that LD can be applied as an additive-free enantioenrichment method. In our report, racemic BINOL coated onto achiral optical borosilicate glass substrates (BK7) were chosen as am odel system for the enantioselective desorption by circularly polarized light.
Absorption and circular dichroism (CD) spectra of racemic,( R)-and (S)-BINOL in solution are presented in Figure 1a and b, respectively.T he strong UV absorption of BINOL extends up to 350 nm, and the enantiomers show optical activity throughout the depicted range.W ep repared the samples by evaporating racemic BINOL onto clean BK7 substrates in ah igh-vacuum chamber to form films with controlled thickness (for experimental details,s ee the Supporting Information). We irradiated the samples with circularly polarized sub-50 fs laser pulses with ac entral wavelength of 600 nm and 650 nm, that is,a bove the onset of the resonant two-photon absorption (TPA) of BINOL as extracted from second harmonic generation spectrum of BINOL films (see the Supporting Information), in order to desorb intact and neutral molecules, [6] as confirmed by timeof-flight mass spectrum of desorbed BINOL (see the Supporting Information). Theinset of Figure 2shows amicroscopy image of the sample region irradiated at 650 nm for 2hours.T he increased transparencyi ndicates the partially desorbed area of the BINOL film. Given the small chiroptical response that is expected from am icron-scale molecular film, we used second harmonic generation circular dichroism (SHG-CD) with ac hiroptical sensitivity of at least three orders of magnitude higher than its linear counterpart. [7] Irradiation of the samples as well as the SHG-CD measurements were performed under atmospheric pressure.D ata measured for films with various thicknesses show that the SHG intensity varies almost linearly with film thickness (see Figure 2). Thus,t he SHG intensity is ad irect measure for the residual molecular film thickness on the substrate. Figure 3i llustrates the results of the LD measurements performed with left circularly polarized (LCP; Figure 3a)and right circularly polarized (RCP;F igure 3b)l ight on racemic samples.A tf irst, the initial SHG intensities of the asprepared BINOL films were measured for both LCP and RCP light in order to obtain initial values of the SHG intensities I LCP and I RCP respectively,a nd the normalized values are shown in the graphs.The pulse energy and focusing were adjusted to suppress the nonlinear optical response from the BK7 substrate (see Figure S2 in the Supporting Information), while keeping the optical intensities well below the photo-damage threshold of BINOL. [8] After measuring the initial SHG intensities,w ed esorbed BINOL from the sample by irradiation with circularly polarized light for at ime t d .T he respective light exposure time t e is the actual light-matter interaction time and takes into account the pulse duration of 50 fs and the repetition rate of the laser system of 1kHz. Thecorresponding results of the LD of the racemic BINOL films during desorption by LCP and RCP are summarized in Figure 3c.T he SHG intensities depicted here are the average of the SHG signals generated by LCP and RCP light from the samples after each desorption step.R esults for LCP desorption are shown on the left-hand side and for RCP desorption on the right-hand side of the graph. Thel arge fluctuations of the data in Figure 3c are caused by variations of the thickness and homogeneity of samples and fluctuations in laser intensity.N onetheless,t he measurement accuracy and spot-to-spot reproducibility of the desorption behaviour is symmetric with regard to the vertical axis of Figure 3c.T he average data from multiple experiments are shown in Figure 3d.T he results indicate that the overall desorption, that is,t he sum of the desorption that occurs for (R)-and (S)-BINOL, of the racemic film does not depend on the polarization handedness of light. Thep artial desorption of (R)-and (S)-BINOL, however, is not necessarily identical.
Theoptical activity of the sample is determined from the measured SHG intensities and expressed by the anisotropy factor g = 2(I LCP ÀI RCP )/(I LCP + I RCP ). It should be mentioned here that the initial values of I LCP and I RCP are not necessarily identical and thus the g value of the as-prepared samples (g 0 ) might deviate from zero.A lthough g 0 varies from sample to sample,itstays unchanged without LD or when desorbed with linearly polarized light as is shown below.Since our aim in this communication is to look at the optical activity generated as the result of interaction with circularly polarized light, in  order to make comparison between data from various measurements more reliable,weintroduce Dg = gÀg 0 .
Before and after each desorption step we measured the optical activity of the film. We repeated the cycle of measurement-desorption for at least 2hours for all the samples.I nf act, as is shown in Figure 4, the anisotropy factor Dg of the remaining BINOL film fully correlates with the handedness of the light that has been used for LD. Figure 4a demonstrates that despite the fluctuations in the laser intensity and variations in sample properties,the striking correlation between the handedness of the circularly polarized light used for LD and the emerging optical activity is consistent for every single measurement.
We observed symmetric intensity decrease (see Figure 3c) and antisymmetric Dg for LD with LCP and RCP light for all samples and at both wavelengths (600 and 650 nm) used in this study (see Figure 4a). Negative test experiments on achiral 2-naphthol samples show no Dg generation, as expected (see Figure 4a,g reen symbols). Our observations clearly show that desorption with circularly polarized light leads to preferential desorption of enantiomers of BINOL from an achiral surface,aswitnessed by the resulting dependence of Dg on handedness of light. This effect can be confidently assigned to the polarization handedness of the light since no Dg is observed in the case of desorption with linearly polarized light (see black symbols in Figure 4a). This also unambiguously demonstrates that the desorption channel of photoexcitation without subsequent thermal equilibration is opened up at the expense of thermal desorption. [9] Accordingly,i nt he present case,L Dc annot simply be considered to be equivalent to at hermal desorption at ah igh heating rate.T he interaction of circularly polarized light with the enantiomers clearly plays anon-negligible if not adominant role in the process of LD.
Although in early days of LD electron-and photonstimulated processes,s uch as those suggested by Menzel, Gomer, and Redhead, [10] and Antionewicz [11] (quantum mechanical mechanisms), were considered to play am ajor role, [12] currently the most commonly accepted mechanisms for LD are based on thermal desorption, where the photon energy is efficiently coupled with vibrational modes and phonons in molecular and solid state system, respectively.I n this thermal picture,the polarization of the photons plays no role.T he inclusion of quantum mechanical (QM) processes, however, is ac ore necessity for explanation of our observation, since only in this way the emergence of such al arge anisotropy factor can be understood.
We have constructed as implified phenomenological model, which could describe our observation (for details, see the Supporting Information). Briefly,i nt his model, we allow (R)-and (S)-BINOL molecules to have different desorption rates caused by TPAw hen interacting with circularly polarized light. Thed esorption rates include thermal as well as QM contributions as mentioned above. Thed esorption process occurs at the surface layers.I n contrast, the TPAo fb ulk BINOL molecules ends with af ast energy relaxation and subsequent local heating according to the model. Bulk heating occurs since excited molecules cannot evaporate from the film before collisional energy exchange if excited deep within the molecular film. These processes facilitate diffusion within the film, which might be derived from the compositional gradient between the bulk and the surface,ifenantiospecific desorption occurs. This way,the depletion of the strongly desorbing enantiomer caused by LD is partially compensated from lower layers.As aresult, the enantiomeric excess (ee)that is generated at the surface is transferred through the bulk of the film and the strongly desorbing enantiomer is continuously delivered to the surface for further LD.
As shown in Figure 3d,t he desorption curves indicate afast initial desorption rate manifested by arapid drop in the SHG intensity of the samples,w hich becomes slower for longer desorption times.N ote that the SHG intensity is linearly proportional to the film thickness and thus to the number of BINOL molecules (see Figure 2). Consequently, the intensity change of the SHG signal directly monitors the change in the number of molecules,t hat is,t otal desorption rate.T he correlation between the SHG intensity decay and material removal by desorption is supported by confocal microscopy images and profilometric data (see the Supporting Information). Abiexponential fit in Figure 3dshows afast initial desorption rate (ca. 0.3 min À1 ), which becomes almost two orders of magnitude slower for longer desorption times (ca. 0.004 min À1 ). Although these rates depend on the used experimental conditions,t heir ratio (ca. 75) indicates an upper limit for the significant difference in the desorption rates of both enantiomers induced by circularly polarized light. We emphasize that cascaded second order nonlinearity, which is responsible for TPAp rocess, [13] possesses an even higher asymmetry than the SHG process.
Furthermore,t he SHG intensity on Figure 3d drops to 55 %o fi ts initial value with the higher desorption rate. Looking at Figure 4a,wesee that in each case the anisotropy factor of the film almost saturates at the same time as the slow desorption rate sets in. Within this oversimplified model, this result suggests that the film has been almost purified enantiomerically very quickly by desorbing one enantiomer while largely leaving the other enantiomer in the film. Such ap rocess would only be feasible if the molecules in the film are highly mobile,p ossessing ah igh diffusion coefficient. Apparently the TPAi nt he bulk accelerates the diffusion process so that the enantiomeric excess that is generated at the surface is quickly transferred through the bulk of the film. It should be noted that this oversimplified model, although self-consistent, very likely overestimates the achieved ee.
In order to verify the absolute value of the achieved ee in the desorbed samples experimentally,k nowledge of the anisotropy factors of the pure enantiomers g p determined by the same method are required. However, as was demonstrated very recently,u nlike racemic films,e nantiopure films of BINOL crystallize in optically active superstructures (see the Supporting Information) wherein the anisotropy factor of the film varies strongly and is very different from that of the constituent molecules.C onsequently,m easurements of g p with our setup are not reliable.W ep repared samples with 10 % ee to calibrate the Dg scale,with the assumption that at this enantiomeric ratio,t he film properties would be close enough to the racemate rather than the pure enantiomer films.A lthough the 10 % ee samples do not resolve the calibration problem completely,they provide amore reliable value due to their much smaller distribution width. Taking the average of the values (g = 0.069, for ee = 10 %) as ap oint on the hypothetical calibration curve,w ec onclude that the generated g of 0.13 for the laser-desorbed spots corresponds to an ee of 20 %. Earlier studies on nonlinear anisotropy factor of BINOL monolayers measured by SHG-CD also report avalue of 0.7 that might be used as g p . [7b] This value also leads to an averaged ee of only 20 %, which represents alower limit of enantioenrichment.
Thep ossibility of enantiospecific LD provides new and unprecedented opportunities.I nt his work, we report enantiospecific desorption by circularly polarized light. The achieved average ee lies between 20 %and > 90 %according to the used estimations.W eb elieve that by optimizing the process,f or example,t emperature,p hoton energy and time profile,a nd thickness of the film, and by introducing desorption-adsorption cycles,w em ay achieve as eparation close to 100 %for other systems too.Asaresult of this work, an easy enantiomeric enrichment of racemic samples is now less remote.I nt he pharmaceutical industry,e nantiospecific LD may become attractive because the risk of contamination through additional chemicals typically used for the separation of enantiomers is of no relevance.I nh eterogeneous asymmetric catalysis,e nantiospecific LD may also become an important tool for the separation of nonvolatile reaction products.