UV Circular Dichroism of Molecular Films on Plasmonic Nanostructures: A Comparative Case Study with On-Resonant Aluminum and Off-Resonant Silver Nanoparticles

Abstract

Electronic absorption bands of important biomolecules and pharmaceutical compounds lie in the ultraviolet (UV) between 180 nm and 280 nm and thus in a spectral range that does not overlap with the localized surface plasmon resonances (LSPRs) of conventional gold (Au) or silver (Ag) nanoantennas. Aluminum (Al) nanostructures support resonances in the UV, and there is significant interest in utilizing UV-resonant nanostructured Al substrates for enhancing the sensitivity of chiroptical spectroscopies, such as circular dichroism (CD). In this study, we compare the CD of a chiral molecular film on Al and Ag substrates and evaluate the role of inherent and induced absorptive CD as well as of scattering CD in the UV. The CD signal of the test molecule 2,2’-Bis(di-p-tolylphosphino)-1,1’-binaphthyl (Tol-BINAP) was measured on large random arrays of Al and Ag nanoparticles (NPs) generated by hole-mask colloidal lithography (HCL). Al NPs provide size-tunable quadrupole and dipole resonances overlapping with molecular absorption bands in the UV, while Ag NPs lack plasmon resonances in this range. The CD signal on the Al NP substrate was increased by up to 154% compared with Ag NP controls. Using Poynting’s theorem generalized to chiral media, we decompose the CD into scattering and absorptive components. The chiral film induces an asymmetric scattering response in resonant achiral Al nanocylinders, dominating the total CD. In contrast, CD in off-resonant Ag nanocylinders arises from asymmetric absorption. The induced and inherent absorptive CD components have opposite signs and partially cancel each other.

Graphical Abstract

Circular dichroism measurements on UV-resonant aluminum and off-resonant silver nanostructures show that on-resonance scattering in aluminum leads to stronger chiral signal enhancement. The results demonstrate that scattering-dominant and UV-accessible plasmonic resonances are effective for enhancing chiral optical responses.

INTRODUCTION

Chirality is a fundamental concept in stereochemistry that applies to molecules that cannot be superimposed on their mirror image. Since the two distinct mirror images of a chiral molecule, the so-called enantiomers, can trigger different physiological effects, characterizing the chirality of pharmaceutical compounds and biological molecules is critical for understanding their structure, function, and interactions with other molecules.¹ The two enantiomers of a chiral molecule interact differently with circularly polarized light (CPL) of opposite handedness, and the difference in the absorption of left-CPL and right-CPL defines the circular dichroism (CD).² CD spectrometers are widely available, easy to use, and provide detailed stereochemical information for an uncomplicated distinction between enantiomers.³ Natural CD signals of molecules are, however, weak, and require high sample concentrations, which has motivated the development of CD enhancement strategies.^4,5

As noble metal nanoantennas enhance electromagnetic fields and control optical phases, they are of interest as platforms for surface-enhanced chiral sensing. Refs. 4,6, and 7 provide excellent reviews of the strategies explored so far. One key finding of prior studies has been that the CD signal obtained with plasmonic antennas can have contributions from inherent as well as induced CD.⁸ The inherent CD of a chiral molecule depends on its interactions with chiral light fields and can be enhanced if it is located in an electromagnetic field with enhanced optical chirality density, defined as⁹

$$C=-\frac{\omega }{2c^2}\mathrm{I}\mathrm{m}\left({\tilde{\mathbf{E}}}^{\mathrm{*}}\cdot \tilde{\mathbf{H}}\right)$$ (1)

In this expression, $\omega$ and $c$ are the angular frequency and speed of light in vacuum, and $\tilde{\mathbf{E}}$ and $\tilde{\mathbf{H}}$ are the complex electric and magnetic field intensities. Induced CD results from asymmetric attenuation of the incident left- and right-CPL by a nanoantenna embedded in a chiral matrix, such as a film of chiral molecules.^6–12 The relevance of induced CD for signal enhancement by nanoantennas was corroborated by CD measurements of a molecular film on an array of achiral Si NPs with electric and magnetic dipole resonances in the visible.¹³ The authors found that the detected CD enhancement was determined by an asymmetric absorption in the Si NPs embedded in the chiral molecular matrix rather than a significant contribution from optical chirality.

Induced CD has been introduced as an asymmetric absorption but in the case of noble metal NPs with large scattering cross-sections embedded in a chiral medium, asymmetric scattering can also contribute to the total CD.¹⁴ In this case, the total CD is given as the difference in extinction of left- and right-CPL. The Poynting theorem for chiral media has provided a general framework for characterizing asymmetric absorption⁸ and scattering¹⁴ by achiral metal NPs embedded in a chiral medium, as well as for dissecting contributions from inherent and induced CD.⁸ The mechanisms underlying surface-enhanced chiroptical effects remain a topic of ongoing research. Radiative coupling and chirality transfer through dipole interactions have been investigated in detail,^8,15–17 and more intricate interactions between the surrounding molecular matrix and antenna beyond the dipole approximation and Coulomb interactions are conceivable and have been investigated.^17–19 Noble metal (Ag, Au) nanostructures provide plasmon resonances in the visible range of the electromagnetic spectrum, whereas the molecular absorption bands of biomolecules lie in the UV. This energetic mismatch poses an inherent challenge for the amplification of molecular CD signals.^8,20–25 A chiral medium can still induce CD in the metal nanostructure at its plasmon resonance if the tail of the medium’s optical rotatory dispersion (ORD) and the plasmon resonance overlap, as ORD and CD are connected via the Kramers-Kronig transform.^26,27 Unlike Ag and Au nanoantennas, Al nanostructures provide strong resonances in the UV,^28,29 and are therefore of interest as alternative substrates for surface-enhanced chiroptical spectroscopies under resonant molecule – plasmon coupling conditions. Chiral Al nanoantennas have previously been applied to detect chiral molecules through changes in the CD spectrum of the antennas in the presence of chiral analytes.^20,30 Since chiral antennas exhibit a strong CD signature, they require the isolation of a molecular signal of interest from a typically much stronger antenna signal. The use of racemic nanoplasmonic arrays is one potential strategy to minimize the CD background from plasmonic antennas, but it requires sophisticated nanofabrication with limited scalability.³¹

In this work, we investigate the CD of a chiral molecular film deposited on achiral Al and Ag NPs and examine how the energetic overlap between plasmonic and molecular resonances affects scattering and absorptive CD, as well as the contribution from inherent and induced CD. Specifically, we investigate the CD of thin films of (R)-(+)-2,2’-Bis(di-p-tolylphosphino)-1,1’-binaphthyl ((R)-Tol-BINAP) and (S)-(−)-2,2’-Bis(di-p-tolylphosphino)-1,1’-binaphthyl ((S)-Tol-BINAP) in the spectral range between 180 and 350 nm on large-scale random arrays of achiral Al and Ag NPs generated by hole-mask colloidal lithography (HCL). The Al NPs are found to provide stronger peak CD signal intensities than the Ag NPs that were included as non-resonant plasmonic controls. Electromagnetic simulations of the relative contributions from absorptive and scattering CD for Al and Ag NPs embedded in a molecular film reveal that the scattering contribution dominates for the UV-resonant Al nanostructures, while absorptive CD is higher for Ag NPs. Inherent and induced absorptive CD compensate each other to a significant degree in both metals due to a depolarization effect.^32,33 Consequently, the higher scattering CD for resonant Al NPs accounts for the increase in CD compared to the non-resonant Ag NPs.

RESULTS AND DISCUSSION

CD of Molecular Films on Plasmonic Antennas: Absorptive and Scattering CD Contributions.

The experimental system under investigation in this work is schematically outlined in Figure 1a. Al or Ag NPs are embedded in a film of chiral molecules that shape the electromagnetic field incident on the NPs when irradiated with CPL to record CD spectra. In the case where the NPs provide a plasmonic resonance that overlaps with a molecular absorption band (Figure 1b, “on resonance”), the chiral matrix affects the induced electromagnetic field in the NPs, resulting in an asymmetric dissipation in and scattering by the NPs. These two processes give rise to induced dissipative CD in the metal and scattering CD, respectively. Furthermore, differences in the near-field surrounding the NPs under left- and right-CPL translate into differences in absorption within the molecular film. This process generates induced CD in the molecular film. Inherent CD, on the other hand, arises from the optical chirality density in the vicinity of the NPs. The contributions from scattering CD are strongly wavelength dependent and expected to be greatly reduced if the NP lacks a resonance in the range of molecular absorption (Figure 1c, “off resonance”). Induced absorptive and inherent CD can contribute both under on- and off-resonance conditions and their relative magnitude depends on the details of the system under investigation.

Figure 1.

In surface-enhanced CD spectroscopy (a), a nanoantenna is embedded in a chiral molecular film that induces absorptive and scattering CD for the nanoantenna. If a plasmonic NP provides a resonance that overlaps with an absorption band of the molecule, inherent and induced absorptive CD as well as scattering CD contribute to the experimentally measured CD as indicated in (b) for the “on-resonance” case. If molecular and plasmon resonances do not overlap in the “off-resonance” case (c), scattering is reduced, and the absorptive CD is determined by the asymmetric dissipation of the incident field in the molecule and NP.

Fabrication and Characterization of UV-resonant Al NPs.

To experimentally characterize the CD spectroscopy of test molecules on plasmonic nanostructures under on-resonance or off-resonance conditions, it is necessary to generate plasmonic substrates with resonances in different spectral ranges. We chose hole-mask colloidal lithography (HCL) as a low-cost and large-scale fabrication method to generate Al and Ag NPs. This strategy has previously been established for the fabrication of substrate supported UV-resonant Al NPs.^28,34–38 Depending on the size and aspect ratios of the NPs, the plasmon resonances of Al NPs can be tuned throughout the UV, visible and infra-red range of the electromagnetic spectrum. To achieve resonances in the UV, we designed relatively small nanostructures with average diameters of approximately 40, 70, and 100 nm, while maintaining a constant Al deposition thickness of 80 nm.

Figure S1 outlines the fabrication process of Al NPs by HCL as applied in this work (see Sample Fabrication in Experimental Section for details). A particular advantage of the HCL fabrication strategy is that it allows the generation of large areas of random NP arrays on quartz cover slips (Figure 2e), which can be analyzed using conventional CD spectrometers. The diameter of the metal NPs in the array can be controlled through choice of the diameter of the polystyrene (PS) nanobeads bound to the PMMA film as these define the feature sizes of the hole mask used in the HCL process (Figure S2). The number of metal NPs on the quartz substrate is determined by the concentration of the PS nanobeads bound to the substrate, which depends on the diameter of the PS nanobeads and the salt concentration of the PS nanobead solutions used to assemble the nanobead mask. Salt is added to improve the binding of smaller PS nanobeads and is instrumental in achieving high densities of surface-bound nanobeads. A high number density of Al NPs in the arrays is desirable to maximize the contribution from any NP-enhanced signal, but high salt concentrations also risk the formation of NP aggregates. As aggregation results in a spectral red-shift, the HCL process needs to be optimized for applications in the UV to achieve NP arrays with high area number densities of monomers and low aggregation levels. The focus of this work is on the properties of individual Al NPs. The random distribution of Al NPs achieved by HCL naturally avoids systematic far-field coupling between NPs and is well-suited for the spectroscopic characterization of average NP properties.

Figure 2.

SEM images of Al NPs with diameters of 40 nm (a), 70 nm (b), and 100 nm (c), generated using masks created from polystyrene (PS) nanobeads of 40 nm, 60 nm, and 100 nm, respectively, with a constant Al deposition thickness of 80 nm. The insets are magnified images of NP monomers in the random arrays, with all inset scale bars representing 50 nm. (d) Experimentally measured extinction spectra of the Al NP arrays corresponding to the diameters shown in the (a)–(c), with dipole and quadrupole resonance modes labeled as “D” and “Q”, respectively. (e) Photographs of Al NP-coated quartz substrates used for the extinction measurement in (d). The left, center, and right samples correspond to the SEM images in (a), (b), and (c), respectively. (f) Resonance peak wavelengths of dipole and quadrupole modes plotted as a function of NP diameter.

Hole masks were generated using fixed concentrations of 0.1% w/v of PS nanobeads with diameters of 40, 60, and 100 nm and variable salt concentrations. Figure S3 shows SEM images of Al NP arrays obtained with 60 nm PS nanobeads and salt concentrations of 0, 2, 5 mM NaCl. The area number density of Al NP arrays obtained with 60 nm PS nanobeads doubles from 15.9 particles/μm² to 31.1 particles/μm² when the salt concentration is increased from 0 mM to 2 mM. A further increase in the salt concentration from 2 mM to 5 mM results in only a moderate increase of the area number density to 33.2 particles/μm². The polydispersity index (PDI) of the Al NPs was 0.02 for 0 mM NaCl indicating a monodisperse distribution, 0.09 (monodisperse) for 2 mM NaCl, and 0.12 ( polydisperse distribution) for 5 mM NaCl.³⁹ Based on these data, we chose a salt concentration of 2 mM for all experiments with 60 nm PS nanobeads to compromise between high NP number density and low levels of aggregation. For 40 nm PS nanobeads a NaCl concentration of 5 mM was used to maximize the number of NP monomers in the arrays. In the case of the 100 nm PS nanobeads, salt addition yielded little change in the number of bound PS monomers but increased the agglomeration (Figure S3). Consequently, no NaCl was added in this case.

Figure 2a–c shows SEM images of Al NP arrays generated using PS nanobeads of 40 nm, 60 nm, or 100 nm diameter (0.1% w/v) and the optimized salt concentrations described above. The area number density and surface coverage of the arrays are listed in order of increasing Al NP diameter: 63.4 μm⁻² (8.60%), 31.1 μm⁻² (15.7%), 2.14 μm⁻² (1.58%). The number density decreases with increasing NP diameter, from 40 nm to 100 nm nanobeads, while the filling fraction (surface coverage) peaks for the NPs generated from 60 nm nanobeads. The average sizes of the Al NPs, which were determined by Gaussian fits to the NP size histograms, were 38.9 ± 12.8 nm, 73.6 ± 23.1 nm, and 96.5 ± 8.89 nm (Figure S4). We refer to these arrays in the following as 40, 70, and 100 nm Al NP arrays.

The magnified SEM images (see insets in Figure 2a–c and Figure S5) of the fabricated Al NPs reveal a broad range of different NP morphologies. In the case of the 40 and 70 nm Al NPs, some NPs have crystal-like morphologies with distinct edges and facets. A high level of crystallinity in NPs is advantageous for optical applications, as it increases the quality factor and reduces plasmonic losses.^40,41 100 nm Al NPs lack these features, indicating that the feature size of the hole-mask affects the formation of these crystal-like shapes. Al grain sizes of up to 40 nm have been reported for Al films when deposited at a deposition rate of 2 Å/s,⁴² as applied in this work. This grain size is comparable to the diameter of the holes generated by 40 nm or 60 nm PS nanobeads but smaller than the holes generated by 100 nm PS nanobeads. Masks with holes smaller than the grain size favor the formation of morphologically defined NPs with distinct crystal-like structures, while larger holes result in the formation of less-defined, amorphous structures with smaller grains on the NP surface. SEM images shown in Figure 2a–c provide 2D projections of 3D NPs. We also performed AFM imaging experiments to characterize the height of the NPs (Figure S6). The height of the 40 nm NPs was 28.5 nm, the 70 nm NPs had a height of 66.0 nm, and the 100 nm NPs had a height of 85.3 nm.

Figure 2d presents the experimental extinction spectra of 40 nm, 70 nm, and 100 nm Al NP arrays measured in the wavelength range between 180 – 500 nm for the large-scale substrate samples shown in Figure 2e. The spectra represent the raw experimental data, with the observed systematic differences in intensity resulting from differences in both the cross-sectional area of the individual NPs and the area number density of the arrays. The 70 nm Al NP array exhibits the highest extinction in the UV region of interest in Figure 2d. This is attributed to a significantly higher NP area number density than for the 100 nm NP array and larger optical cross-sections per NP than for the 40 nm NPs. Extinction efficiencies ($Q_{\mathrm{e}\mathrm{x}\mathrm{t}}$) for individual nanoparticles with diameters of 40 nm, 70 nm, and 100 nm (Figure S7) were estimated from the measured extinction spectra (Ext) using the equation:⁴³

$$Q_{\mathrm{e}\mathrm{x}\mathrm{t}}=\frac{\mathrm{E}\mathrm{x}\mathrm{t}}{N_A\cdot A_p}$$ (2)

where $N_A$ is the area number density (particles/μm²) and $A_p$ is the geometric cross-sectional area of a single particle (μm²).

The 40 nm Al NP array spectrum contains one prominent peak at 265 nm. This fundamental resonance is assigned to the dipole mode ($D^{40}$). For the 70 nm NP array, the dipole mode ($D^{70}$) is located at 310 nm and a second higher energy band, assigned to the quadrupole mode ($Q^{70}$), peaks at 209 nm. For the 100 nm Al NP arrays, dipole ($D^{100}$) and quadrupole ($Q^{100}$) modes shift to 432 nm and 223 nm, respectively, and both modes broaden. Figure 2f shows the measured peak wavelengths as a function of Al NP size for both dipole and quadrupole modes.

Comparative CD Analysis of Tol-BINAP on Resonant Al and Off-Resonant Ag Nanoparticles.

The chemical structures of the two enantiomers of the target molecule Tol-BINAP, which were deposited as a thin film onto the Al NP or Ag NP arrays by spin coating, are shown in Figure 3a. We characterized the CD signal of this test molecule on 70 nm Al NPs (Figure 3b) and Ag NPs fabricated using the same hole-mask array parameters and metal deposition thickness (Figure 3c). The Ag NPs have an average diameter of 74.0±19.2 nm, determined by a Gaussian fit of the SEM NP size histogram (Figure S8), and the area number density and surface coverage area in the array are 30.2 μm⁻² and 13.9%, respectively. These numbers are very close to the corresponding values for the 70 nm Al NP array (73.6 nm, 31.1 μm⁻², 15.7%), and we refer to these samples in the following as 70 nm Ag NP arrays.

Figure 3.

(a) Chemical structures of the left- (S or (−)) and right- (R or (+)) handed enantiomers of Tol-BINAP investigated in this study. SEM images of 70 nm Al NPs (b) and Ag NPs (c) fabricated using masks created with 60 nm PS nanobeads and a constant metal deposition thickness of 80 nm. All scale bars are 200 nm. (d-f) Measured extinction spectra of Tol-BINAP films (d) on quartz, and random arrays of 70 nm diameter Al NPs (e) and Ag NPs (f) on quartz. (g,h) CD spectra of (S)-Tol-BINAP, (R)-Tol-BINAP, and their racemic films (R,S) on Ag and Al nanostructures, measured before (g) and after (h) annealing, with insets (black) showing the percentage of CD increase in the wavelength range between 180 nm and 200 nm. All dashed lines indicate the absorption peak wavelengths of Tol-BINAP films. Q and D denote the electric quadrupole and dipole resonances of Al NPs, respectively.

The spectra of Tol-BINAP films on quartz show strong absorption for wavelengths < 250 nm (Figure 3d). The extinction spectra of right- or left-handed Tol-BINAP films are identical. A comparison of the Tol-BINAP spectra with the 70 nm Al NP spectrum (Figure 3e) reveals that the peaks in the molecular spectra overlap with the quadrupole mode of the NP array. In contrast, the 70 nm Ag NP array does not sustain any plasmon resonances in the wavelength range between 180 – 350 nm where the spectrum is instead determined by broad Ag interband transitions (Figure 3f).^44,45 The dipole and quadrupole modes of the 70 nm Ag NP array lie outside of the range of interest at 455 nm and 355 nm, respectively (Figure S9).

After confirming that the achiral Ag and Al NP arrays exhibit minimal CD signals between 180 nm and 350 nm (Figure S10), a comparison of the Tol-BINAP CD spectra recorded on these arrays offers insights into the mechanisms determining the CD of molecular films on plasmonic substrates. The strong morphological similarity between the Ag and Al NP arrays, along with the lack of UV resonances in the Ag NP arrays, make them a useful off-resonance benchmark to characterize the effect of the on-resonance UV plasmons on the CD spectrum without the need for correcting for filling fraction effects. Figure 3g,h shows the spectra of (R)-Tol-BINAP, (S)-Tol-BINAP, and the racemic mix of (R)-Tol-BINAP and (S)-Tol-BINAP on random arrays of 70 nm Al NPs with the corresponding controls recorded on the Ag NP array benchmark. The CD spectral intensities of the racemic mix of (R)- and (S)-Tol-BINAP are zero for the Al and the Ag NP arrays, confirming that the NP arrays themselves show no structural chirality. The reproducibility of the Tol-BINAP CD measurements was tested experimentally by repeating the experimental cycle of spin-coating of Tol-BINAP on an Al NP array, CD and extinction measurements, and subsequent removal of Tol-BINAP by washing with toluene, acetone, and isopropyl alcohol before drying in a nitrogen stream, three times. The acquired spectra show a high degree of reproducibility with average variations of the CD and extinction signal of less than 0.312 mdeg and 0.00467 a.u. between measurements (Figure S11). The Al NP array generated measurably higher signal intensities at 194 nm and 237 nm compared to the Ag NP array for both (R)- and (S)-Tol-BINAP. In Figure 3g, the average CD amplitudes of both enantiomers increased by 30.5% at 194 nm (see insets in Figure 3g) and by 15.4% at 237 nm. The percentage increase in average CD amplitude is determined by comparing the CD signals of the test molecules on Al NPs and Ag NPs, using the following formula:

$$\frac{{\text{CD}}_{\text{amplitude}}^{\text{Al NPs}}-{\text{CD}}_{\text{amplitude}}^{\text{Ag NPs}}}{{\text{CD}}_{\text{amplitude}}^{\text{Ag NPs}}}\times 100(\%)$$ (3)

${\mathrm{C}\mathrm{D}}_{\text{amplitude}}^{\text{Al NPs}}$ and ${\mathrm{C}\mathrm{D}}_{\text{amplitude}}^{\text{Ag NPs}}$ represent the CD amplitudes of the test molecules on Al and Ag NPs, respectively.

In the next step, we annealed the Tol-BINAP films on the metal NP arrays to increase the molecular packing,^46,47 and promote the integration of the nanostructures into the analyte film for improved CD signal enhancement.¹⁴ While annealing changed the appearance of the molecular film on a bare substrate, the presence of the NPs in molecular thin films prevented this transformation and no new CD signals were observed. The annealing did not affect the overall shape of the extinction and CD spectra of Tol-BINAP but induced small spectral shifts (Figure S12). After annealing, the CD feature with a peak at 237 nm for the off-resonance Ag NP array is, for instance, red-shifted on the Al NP array in Figure 3h by 4 nm or 2 nm for (R)-Tol-BINAP and (S)-Tol-BINAP, respectively. These shifts are first indications of some interactions between molecular absorption bands and Al NP plasmons. The CD feature of the (R)- and (S)-Tol-BINAP films at 194 nm, which overlaps most strongly with the quadrupole resonance of the Al NP array, experiences an average peak intensity increase of 154% for the Al NP array relative to the Ag NP array (see insets in Figure 3h). The prominent CD feature at 237 nm that still overlaps with the Al quadrupole mode and also with the high energy tail of the Al dipole mode increases in intensity by an average of 42.0%. The CD signal enhancements reported above are CD signal amplitude averages for the (R)- / (S)-Tol-BINAP measurements as we observed some differences in the calculated enhancement factors depending on the handedness of the Tol-BINAP film at different wavelengths. Features in the CD spectrum at wavelengths > 270 nm also exhibit a noticeable increase in intensity and changes in spectral shape on the Al array, but the effect is less obvious than for the better-defined CD features at shorter wavelengths in the spectral range that contains the molecular absorption bands.

A comparison of the Tol-BINAP CD signals obtained with Al NP arrays and Tol-BINAP film without any metal NP substrate (Figure S13) requires correction for the different number of molecules in the sampled area, which can be achieved by normalizing the signals by the effective number of Tol-BINAP molecules. This analysis (see Supporting Information) yields an ensemble-averaged CD signal enhancement factor of 1.2 for the entire molecular film on Al NPs before annealing. This modest increase is at least partly due to an ensemble averaging over areas in the film with and without enhancement. Near fields decay rapidly with distance and if one assumes that only the molecules within approximately 3 nm of the Al₂O₃-encapsulated Al NPs contribute to the CD enhancement, the enhancement factor increases to 7.6 before annealing. A direct comparison of the CD signals measured after annealing on films with and without NPs was not possible, as molecular packing developed differently on the structured substrate compared to the non-structured bare one,^48,49 resulting in different CD spectrum shapes. We estimated the enhancement in this case based on the ratio of the differences in CD signal at 194 nm for films on Al and Ag NPs before and after annealing. This ratio was determined as 1.9, which together with the geometric factor outlined above translates into an ensemble-averaged enhancement factor of 2.3 and of 14 for the assumption of the entire molecular film and a 3 nm thin active layer around the NPs, respectively.

Modeling the Microscopic Origin of the Detected CD.

The absorption cross section, i.e., the optical absorption rate normalized by the intensity $I_0$ of the incident light at the frequency $\omega$, is given as ${\sigma }_{\mathrm{a}\mathrm{b}\mathrm{s}}={\sigma }_{\mathrm{a}\mathrm{b}\mathrm{s},\mathrm{E}}+{\sigma }_{\mathrm{a}\mathrm{b}\mathrm{s},\mathrm{H}}+{\sigma }_{\mathrm{a}\mathrm{b}\mathrm{s},\mathrm{C}}$, with:⁸

$${\sigma }_{\mathrm{a}\mathrm{b}\mathrm{s},\mathrm{E}}\equiv \frac{\omega }{2I_0}{\int }_V\left[\mathrm{I}\mathrm{m}\left(\epsilon \right)|\tilde{\mathbf{E}}{|}^2\right]dV$$ (4)

$${\sigma }_{\text{abs},\text{H}}\equiv \frac{\omega }{2I_0}{\int }_V\left[\text{Im}(\mu )|\tilde{H}{|}^2\right]dV$$ (5)

$${\sigma }_{\text{abs},\text{C}}\equiv \frac{\omega }{2I_0}{\int }_V\left[2\text{Im}(\kappa )\text{Im}\left(\tilde{E}\cdot {\tilde{H}}^{\ast }\right)\right]dV$$ (6)

Here, $\epsilon$ is the electric permittivity, $\mu$ the magnetic permeability and $\kappa$ the chirality parameter of a given medium. $V$ denotes the integration volumes of the chiral film or NP. Because of the negligible optical magnetism in natural materials, the magnetic part of absorption is ${\sigma }_{\mathrm{a}\mathrm{b}\mathrm{s},\mathrm{H}}=0$. Although two CPLs illuminate the chiral film-coated NP structures at the same intensity, the near-field amplitudes (${\tilde{\mathbf{E}}}^{+},{\tilde{\mathbf{E}}}^{-}$) for the two opposite CPL illuminations can be different because of the asymmetric perturbation to the near-field by the chiral film. Consequently, the difference in the absorption cross section, namely circular dichroic absorption, of the chiral film and NPs under illumination with left or right-CPL (${\tilde{\mathbf{E}}}^{+},{\tilde{\mathbf{E}}}^{-}$) has contributions from the electrically induced CD in both NP and molecules:

$$\mathrm{\Delta }{\sigma }_{\mathrm{a}\mathrm{b}\mathrm{s},\mathrm{E}}={\sigma }_{a\mathrm{b}\mathrm{s},\mathrm{E}}\left({\tilde{\mathbf{E}}}^{+}\right)-{\sigma }_{\mathrm{a}\mathrm{b}\mathrm{s},\mathrm{E}}\left({\tilde{\mathbf{E}}}^{-}\right)$$ (7)

as well from the inherent CD in the molecules:

$$\mathrm{\Delta }{\sigma }_{\mathrm{a}\mathrm{b}\mathrm{s},\mathrm{C}}={\sigma }_{\mathrm{a}\mathrm{b}\mathrm{s},\mathrm{C}}\left({\tilde{\mathbf{E}}}^{+},{\tilde{\mathbf{H}}}^{+}\right)-{\sigma }_{\mathrm{a}\mathrm{b}\mathrm{s},\mathrm{C}}\left({\tilde{\mathbf{E}}}^{-},{\tilde{\mathbf{H}}}^{-}\right)$$ (8)

Note that ${\tilde{\mathbf{E}}}^{+}={\tilde{\mathbf{E}}}^{-}$ for a bare chiral film without nanostructure, and Eq. (7) reduces to the well-known formula in ref 9. In addition to the circular difference in absorption described by Eqs. (7) and (8), the chirally perturbed nanostructures can also have a circular difference in scattering. The scattering cross section is defined by the surface-integration, ${\sigma }_{\mathrm{s}\mathrm{c}\mathrm{a}}=∯\mathbf{S}\left(\tilde{\mathbf{E}},\tilde{\mathbf{H}}\right)\cdot d\mathbf{a}/I_0$, where $\mathbf{S}$ and $d\mathbf{a}$ are the Poynting vector and the area vector, respectively. The integration surface is an arbitrary surface enclosing the whole nanostructure. Then, the circular dichroic scattering is defined by

$$\mathrm{\Delta }{\sigma }_{\mathrm{s}\mathrm{c}\mathrm{a}}={\sigma }_{\mathrm{s}\mathrm{c}\mathrm{a}}\left({\tilde{\mathbf{E}}}^{+},{\tilde{\mathbf{H}}}^{+}\right)-{\sigma }_{\mathrm{s}\mathrm{c}\mathrm{a}}\left({\tilde{\mathbf{E}}}^{-},{\tilde{\mathbf{H}}}^{-}\right)$$ (9)

What we measure in experiments is a sum of absorption and scattering, i.e. extinction. Therefore, the experimentally observed CD is the sum of Eqs. (7)–(9).

We performed electromagnetic simulations for a 3 nm thin Tol-BINAP film on a 3 nm native oxide layer on top of Al or Ag nanocylinders with a diameter of 70 nm and a height of 65 nm. A schematic representation of the simulation geometry is shown in Figure 4a. The permittivity (Figure 4b) and chirality parameter (Figure 4c) of the Tol-BINAP films were extracted from ellipsometry and CD spectroscopy measurements, while the published permittivity (ε) data for Al and Ag were used as input parameters.^50,51 These values were incorporated into electromagnetic simulations to evaluate Eqs. (7)–(9) for a model that captures key aspects of the investigated NP-supported Tol-BINAP films.

Figure 4.

(a) Schematic of the Al/Ag nanocylinder geometry used in COMSOL simulations to evaluate the contributions from induced and inherent absorptive CD, as well as scattering CD. Each metal nanocylinder has a diameter of 70 nm and height of 65 nm, is covered by a 3 nm thin oxide layer and supported on a glass substrate. A 3 nm thin chiral molecular film is placed on top of the metal oxide layer. (b) Imaginary part of the permittivity (ε) and (c) Imaginary part of the chirality parameter (κ) of the molecular film. Simulated extinction spectra of (d) Al and (e) Ag nanocylinders, with quadrupole and dipole modes labeled as Q_S (Simulated Quadrupole) and D_S (Simulated Dipole), respectively. Simulated spectra of total absorptive CD (Δσ_abs), scattering CD (Δσ_sca), and extinction CD (Δσ_ext) for (R)-enantiomers on (f) Al and (g) Ag nanocylinders. Dashed lines indicate peak wavelengths of the imaginary parts of both ε and κ of Tol-BINAP films.

The simulated absorption, scattering, and extinction cross-sections for Al and Ag nanocylinders between 180 - 500 nm are shown in Figure 4d,e. The extinction spectrum for Al contains plasmon resonances at $Q_S^{70}=245$ nm for the quadrupole mode and $D_S^{70}=310$ nm for the dipole mode. Both scattering and absorption contribute to the extinction at the quadrupole resonance, but at the dipole response absorption has further reduced and extinction is primarily caused by scattering. The Ag nanocylinders lack plasmonic resonances in the wavelength range below 350 nm, and although localized plasmons emerge at longer wavelengths, absorption dominates over scattering across the entire spectral range. The simulated extinction spectra are overall in good agreement with the experimental spectra in Figure 3e,f. The different relative contributions from scattering and absorption for Al and Ag nanocylinders have direct implications for the CD signal (Figure 4f,g). In the case of the Al nanocylinders, the CD signal is dominated by differences in scattering, whereas for the Ag nanocylinder, the differences in absorption provide the largest contribution. This finding is in general agreement with the precious studies that found high scattering to absorption ratios for Al NP plasmon resonances in the UV.⁵² The intrinsic predominance of the scattering response and the increased sensitivity to chiral field at the plasmon resonance contribute to the observed induced chiral scattering response in the case of Al. We conclude that in addition to the question of on- vs. off-resonance, the nature of the resonances and specifically the relative contributions from scattering vs. absorption also affect the CD measured for a molecular film on a plasmonic NP substrate.

The contribution from absorption can be further decomposed into induced and inherent components (Figure 5a,b). For both Al and Ag nanocylinders the contributions from induced absorptive CD in the metal and inherent absorptive CD in the molecular film have opposite sign and compensate each other to a significant degree. An intuitive physical picture of this compensation is given by a depolarization effect in which chiral near-fields of the molecular film induce fields of opposite handedness in the metal. Reflection of a chiral field at the substrate can also reverse its handedness and contribute to the observed effect. For both Al and Ag, the magnitude of the inherent CD exceeds that of the induced CD in the metal. In the case of the Ag nanocylinders, the interplay of inherent and induced CD determines the contribution from absorption to the CD. In contrast, for the Al nanocylinders the electric field associated with the plasmons in the investigated wavelength range acts back on the molecular film to induce an additional, albeit weaker, induced CD contribution in the molecular film.

Figure 5.

(a, b) Simulated absorptive CD spectrum of (R)-Tol-BINAP (Δσ_abs, black) and its components: induced CD in the metal (Δσ_abs,E(metal), red), induced CD in the molecule (Δσ_abs,E(mol), blue), and inherent molecular CD (Δσ_abs,C(mol), green) for (a) Al nanocylinder and (b) Ag nanocylinder.

Differences in the wavelength dependent sign and magnitude of the individual CD contributions can account for some of the experimentally observed reshaping of the CD spectra. There is, however, a noticeable difference in the total extinction-based CD spectra for Al (scattering dominates) and Ag (absorption dominates) nanocylinders. While the predicted CD for Al nanocylinders is in good agreement with the experimental spectrum, the correspondence for Ag is poor. We attribute this discrepancy to the fact that the CD enhancement provided by the Ag nanostructures is weak so that the detected CD in the experiment is determined by the CD “background” of the surrounding molecular film.

Cumulative contributions from inherent and induced molecular absorption, metal-induced absorption, and scattering determine the total CD for a plasmonic NP embedded in the chiral matrix. The sign and magnitude of these contributions depend, however, on the type of the plasmon resonance, the dispersion of the chirality parameter, and the spectral overlap between plasmon and molecular absorption among other factors. In the case of the investigated Al nanostructures the simulations reveal that the contribution from scattering CD is responsible for the total increase in CD. Although the UV-resonant nanostructures also provide induced and inherent absorptive CD, these effects have opposite signs, which reduces their net contributions to the detected CD. Overall, this analysis highlights asymmetric scattering by resonant, achiral plasmonic nanostructures as the key factor in the difference of the CD observed for Al and Ag NPs embedded in chiral molecular films.

CONCLUSIONS

This work has demonstrated that large area Al NP arrays fabricated by HCL support size-tunable plasmonic resonances throughout the UV. The extinction spectra of the fabricated 70 nm Al NP arrays have confirmed that the Al NP arrays provide plasmon resonances across the 180 nm to 500 nm range (UV-Vis), which covers the absorption window (UV) of nucleic acids, proteins, and many pharmaceutical compounds. Arrays of 70 nm Al NPs achieved an average CD signal of annealed Tol-BINAP films that exceeded that of Ag NP arrays without plasmonic resonances in the UV by 154% at 194 nm. Comparison of the CD signals collected on Al NP arrays and on quartz substrate without metal structures indicate ensemble-averaged signal enhancement factors of 2.3 after annealing. For the assumption that only molecules in the direct vicinity of the NPs experience a signal enhancement, the signal enhancement factor increases to 14.

To elucidate the microscopic origin of the observed CD enhancement, we performed electromagnetic simulations, and they reveal wavelength-dependent contributions from inherent and induced absorptive CD in both molecular film and metal NP as well as from scattering CD. The individual components are, however, not all additive due to differences in their sign. The performed simulations suggest that in the case of Al nanocylinders the experimental CD enhancement is dominated by scattering contributions, whereas the contributions from induced and inherent absorptive CD are reduced as they have similar magnitude but opposite sign. Ag nanocylinders lack plasmon resonances in the UV and therefore have much weaker scattering contributions to the total CD spectrum, resulting in weaker overall CD signals. Our analysis highlights the role of the increased scattering contribution as a direct consequence of the Al plasmon resonance in enhancing the total CD signal in the UV.

A better understanding of the relationship between nanoantenna design and wavelength-dependent CD contributions holds potential for further optimizing CD enhancements beyond what was experimentally observed for the random Al NP arrays in this work. In general, rational strategies for enhancing molecular CD in the UV through Al nanostructures that are easy to fabricate in a scalable format and compatible with widely available CD spectrometers (no microscope required) are desirable for applications of CD in sensing and structural characterization. In this study, the analysis was limited to spin-coated Tol-BINAP as probe molecules, but the thin oxide layer around the Al NPs provides opportunities for chemical functionalization that can be utilized to tether specific molecules of interest to the UV-resonant plasmonic nanostructures and thus selectively enrich the analytes in a region with optimal conditions for CD induction to facilitate their spectroscopic characterization.

METHODS

Sample Fabrication.

Al NPs were produced by hole-mask colloidal lithography (HCL) following previously published protocols⁵³ with some modifications. PS nanobead solutions with different NaCl concentrations were tested and optimized to maximize the surface density of Ag or Al NP monomers and a thicker PMMA layer of 1 μm on quartz with a 10 nm thin Al (instead of Cr or Au) mask were used to achieve a better lift-off process. Quartz cover slips were cut into chips of approximately 2.54 by 0.85 cm² using a dicer saw. Subsequently the chips were cleaned in 3:1 H₂SO₄/H₂O₂ piranha solution for 30 minutes at 150 °C. Then, the substrates were rinsed several times with deionized water (DI) and dried under a stream of nitrogen. The chips were stored for future use and cleaned with acetone, isopropyl alcohol, and DI water right before use, and heated to above 100 °C to make sure that the substrate is completely dried. Once the chips had cooled down, a film of poly(methyl methacrylate) (PMMA) film was spin-coated at 2000 rpm on the substrate, and then soft-baked on a hot plate at 180 °C for 1.5 minutes. The series of coating and baking processes was repeated until a total film thickness of 1 μm was reached. The last soft-baked step was for 3 minutes to ensure curing of the PMMA layer. PMMA films were treated with RIE oxygen etching at 50 W for 10 s and incubated with 0.2% poly(diallyldimethylammonium chloride) (PDDA) for 1 minute. The positively charged substrate was then immersed in solutions containing 40, 60, or 100 nm diameter negatively charged PS nanobeads (0.1% w/v) for 1 minute (solutions with NaCl) and 10 minutes (solutions without NaCl). PS nanobead solutions containing different concentrations of NaCl, ranging between 0-5 mM, were used. Subsequently, the PS nanobead arrays were washed with DI water and dried in an air stream. Samples were placed in the center of the sample holder of an electron-beam deposition system (Angstrom) and a 10 nm thin Al layer was deposited. After PS nanobeads were removed by tape stripping, random hole arrays in the PMMA film were created by RIE oxygen etching at 50 W for 60 minutes. After the RIE etching, the samples needed to be handled with care as the masks were easily damaged. The substrates were transferred into an Angstrom thin film deposition system to deposit 80 nm of Al or Ag. The deposition of thicker Al films was a key to achieve higher energy resonances in the UV. The base pressure was kept below 3 × 10⁻⁷ Torr, and the deposition rate was 2 Å/s. The lift-off of the mask was carried out by immersing the samples into acetone for 30 minutes and subsequent sonication in acetone for 5 minutes at the lowest power. The samples were washed with acetone and isopropyl alcohol, then dried under a stream of nitrogen.

Solution Preparation and Molecular Film Deposition.

20 mg of (R)-Tol-BINAP or (S)-Tol-BINAP were dissolved in toluene, and a racemic solution was made by mixing identical volumes of the two chiral solutions. Thin films were spin-coated at 2000 rpm for 40 s. This gave the film thickness of 106 nm on NP arrays and the film thickness of 104 nm on quartz substrate without nanostructures. Film annealing was achieved by placing samples on a hotplate at 150 °C for 15 min. After each spin coating and measurement cycle, the samples were cleaned by sonication in toluene and acetone for 5 minutes each, rinsed with acetone and isopropyl alcohol, and dried under a stream of nitrogen. By this cleaning procedure, the nanostructure substrates were used multiple times for different experiments.

Sample Characterization and CD Measurements.

Electron microscopy was performed using a field emission scanning electron microscope (FE-SEM) Zeiss Gemini 560. Heights were measured by the Cypher atomic force microscopy (AFM) from Asylum Research. CD and extinction spectra were collected using a Chirascan CD spectrometer (Applied Photophysics). The samples were inserted into cuvettes as holders and mounted vertically to the beam path. The beam area on the samples (i.e. the measured area) was decided by the wide window in the beam mask for the cuvette holder. To eliminate substrate and cuvette effects, the background CD and extinction spectra were acquired with only the quartz substrate in the cuvette prior to measuring random NP array samples. All CD spectra were background corrected. A Veeco Dektak 8 profilometer was used to determine the thickness of the chiral films.

Electromagnetic Simulations.

Numerical simulations were performed using COMSOL Multiphysics. The permittivity ($\epsilon$) and the chirality parameter ($\kappa$) of the chiral molecules were obtained from ellipsometry data and the CD spectroscopy measurements of a bare chiral molecular film on a quartz substrate, respectively. The permittivity values of Al and Ag were taken from the tabulated data in Refs. 50 and 51. The refractive indices of the oxide and glass were chosen as 1.62 and 1.5, respectively. We excited chiral molecules adsorbed on $\mathrm{A}\mathrm{l}/\mathrm{A}\mathrm{g}$ nanocylinders using circularly polarized light with an electric field, ${\tilde{\mathbf{E}}}_0^{\pm }=E_0(\stackrel{\mathbf{ˆ}}{\mathbf{x}}\pm i\stackrel{\mathbf{ˆ}}{\mathbf{y}})$. The subscripts $\pm$ represent left and right circular polarization, respectively. The background electric field (${\tilde{\mathbf{E}}}_b$) was derived from incident, reflected, and transmitted ${\tilde{\mathbf{E}}}_0^{\pm }$ at the air/glass substrate interface using Fresnel equation. The total electromagnetic field was determined by the summation of background (${\tilde{\mathbf{E}}}_b$) and scattered (${\tilde{\mathbf{E}}}_s$) field, $\tilde{\mathbf{E}}={\tilde{\mathbf{E}}}_b+{\tilde{\mathbf{E}}}_s$. To calculate the optical response of the chiral molecule on $\mathrm{A}\mathrm{l}/\mathrm{A}\mathrm{g}$ nanocylinder, we numerically solved Maxwell’s equation with constitutive relation for chiral molecule as $\tilde{\mathbf{D}}=\epsilon \tilde{\mathbf{E}}+i\kappa \tilde{\mathbf{H}}$ and $\tilde{\mathbf{B}}=\mu \tilde{\mathbf{H}}-i\kappa \tilde{\mathbf{E}}$.

Supplementary Material

The Supporting Information is available free of charge at XXX (website link).

Details on the fabrication process, optical measurements, and the calculation of CD enhancement factors discussed in the main text.

Data Availability Statement

CD data is available at 10.5281/zenodo.10215342. All other raw data and images are accessible from the corresponding authors upon reasonable request.