Phototransformation of achiral metasurfaces into handedness-selectable transient chiral media

Significance

Plasmonic metasurfaces, periodic arrangements of noble metal nanostructures, can produce strong chiroptical effects. Chiral metasurfaces allow the utilization of chiroptical effects for controlling circularly polarized light or chirality detection of biomolecules. Switching between the enantiomeric states for chiral plasmonic nanostructures, however, is a difficult task where most attempts are either limited to slow geometric reconfiguration or fast yet incomplete chiral inversion. In this work, we simultaneously achieve fast and complete chiral inversion via all-optical construction of chiral metasurface, where the handedness can be assigned on-demand by tailored excitation of spatially inhomogeneous hot-electron populations. This work enables the utilization of dynamic chiroptical effects in a single platform with a sub-picosecond operation speed.

Abstract

Chirality is a geometric property describing the lack of mirror symmetry. This unique feature enables photonic spin-selectivity in light–matter interaction, which is of great significance in stereochemistry, drug development, quantum optics, and optical polarization control. The versatile control of optical geometry renders optical metamaterials as an effective platform for engineered chiral properties at prescribed spectral regimes. Unfortunately, geometry-imposed restrictions only allow one circular polarization state of photons to effectively interact with chiral meta-structures. This limitation motivates the idea of discovering alternative techniques for dynamically reconfiguring the chiroptical responses of metamaterials in a fast and facile manner. Here, we demonstrate an approach that enables optical, sub-picosecond conversion of achiral meta-structures to transient chiral media in the visible regime with desired handedness upon the inhomogeneous generation of plasmonic hot electrons. As a proof of concept, we utilize linearly polarized laser pulse to demonstrate near-complete conversion of spin sensitivity in an achiral meta-platform—a functionality yet achieved in a non-mechanical fashion. Owing to the generation, diffusion, and relaxation dynamics of hot electrons, the demonstrated technique for all-optical creation of chirality is inherently fast, opening new avenues for ultrafast spectro-temporal construction of chiral platforms with on-demand spin-selectivity.

The time-dependent rotation of electric fields in circularly polarized light (CPL) brings upon notable traits including the spin-angular momentum carried by CPL and the orthogonality between the left-circularly polarized light (LCP) and right-circularly polarized light (RCP). These endow the capability for utilization across various fields such as opto-spintronics, communication, displays, encryption, and quantum computing. In geometrical aspects, the helical path that the electric fields of propagating LCP and RCP follow are non-superimposable mirror images. This intrinsic property of CPL is known as “chirality.” The chiral nature of CPL gives rise to the capability of discerning the handedness of chiral molecules where the difference in the interaction between the circular polarization (CP) of CPL and chiral objects manifests itself as circular dichroism (CD) (1, 2). Indeed, discriminating chiral biomolecules is essential as different enantiomers—molecules with same molecular formulas but with opposite handedness—in medicines are influential to life, which could either be clinically beneficial or cause catastrophic health conditions (3). Therefore, an easy access to chiroptical effects is of pivotal importance due to the capability of both CPL control and chiral detection.

The microscopic length scale of chiral biomolecules when compared to the optical wavelengths, however, makes chiroptical effects inherently weak. In order to obtain strong and easily accessible chiroptical effects, two-dimensional periodic arrangements of optical nano-resonators with designed geometries have been proposed as an artificial counterpart (4–8), where numerous studies were reported for different application purposes (9–14). A common downside of such configuration, known as chiral metasurfaces, is the lack of post-fabrication tunability, in which their chiroptical functionalities are fixed upon fabrication based on their geometries and the refractive indices of the constituent materials. As chiral metasurfaces are designed to interact with one preferred spin-state of the photons specifically, this consequently requires one to prepare complete sets of enantiomorphs in order to fully exploit or manipulate both states of CPL, doubling the fabrication time and cost. Thus, the pursuit of tuning the chiroptical effect, and ultimately inverting the handedness of a chiral metasurface, has long been a companion throughout the history of chiral metasurface design (15). Various strategies for tuning the geometrical (16–18), optical (19–21), and experimental configurations (22) were introduced. However, effective geometry tuning has constraints such as intrinsically slow tuning speed or sophisticated fabrication requirements. For refractive index tuning, demonstrations are mostly done outside of the optical regime where “true” inversion is fundamentally limited due to the geometrical handedness presets of chiral metasurfaces.

Meanwhile, recent studies on photoinduced symmetry breaking (23, 24) hint the possibility to alleviate the aforementioned constraints. Nano-resonators in metasurfaces made of noble metals support localized surface plasmons, which shortly after their excitation will partially decay via non-radiative energy exchange channels and produce plasmonic hot carriers (25–27). The excitation of plasmonic hot carriers initiates a chain of transient collision processes, leading to the transient modification of refractive indices of noble metals and subsequently the alteration of the optical response of the metasurface (28, 29). The extremely short timescales, down to the sub-picosecond regime, associated with the generation and relaxation of hot carriers offer ultrafast tuning of optical properties which is a matter of extensive research (30–32). Close inspections on hot-carrier dynamics suggest that hot-carrier generation efficiency is proportional to the plasmon field intensity. Therefore, the population of generated hot carriers follows the spatial profile of the plasmon field intensity defined by the geometry of employed nano-resonators (33). As such, engineered near-field profiles under optical excitation enable a degree of freedom for controlling the spatial distribution of generated plasmonic hot carriers.

In this work, we demonstrate ultrafast, complete chirality inversion in the visible regime enabled by spatially inhomogeneous hot-carrier generation within each resonator. The refractive index gradient induced by spatially varying hot-carrier population breaks the optical mirror symmetry, which instigates chirality in achiral structures.^* In this configuration, simple polarization adjustment of input light for hot-carrier generation is sufficient for assigning the chiral state of the metasurface, as desired handedness can be acquired based on the near-field profile of the resonator responsible for hot-carrier excitation. Furthermore, the relaxation of spatial asymmetry of refractive index is dictated by electron temperature spatial diffusion (34, 35), which is a recent addition to the portfolio of strategies for obtaining ultrafast transients with recovery times faster than the limits imposed by the electron–phonon decay (24, 36–38). This work introduces a unique method for the creation and versatile control of chirality in an ultrafast timescale, which achieves complete chiral inversion in the visible regime. We anticipate this technique to foster unique approaches for chirality creation and inversion and to serve as an effective tool for all-optical signal processing, near-field chirality probing, and understanding spatiotemporal dynamics of hot carriers.

Results

Principles and Design.

Fig. 1A depicts the general working principle and design requirements. Consider an achiral dual-layered metasurface comprised of an array of gold (Au) nanostripes at the bottom layer that is separated from an array of Au triangular split-ring resonators (Tr-SRR) on the top using a subwavelength dielectric film. In principle, this model platform is not capable of distinguishing between the two CP states of light under normal incidence as it possesses a mirror symmetry plane perpendicular to both the substrate and the longitudinal direction of the nanostripes. Despite the existence of the symmetry plane, a linearly polarized pump light with a polarization state, for example, parallel to the left arm of the Tr-SRR creates a dissimilar concentration of the electromagnetic field at the internal corners of the left and right bent segments (Fig. 1 A and B). As we discuss in detail below, the spatial profile of generated hot electrons inside the nanoresonator inherits the asymmetry of the pump light concentration (23), creating a geometrically symmetric Tr-SRR but with broken spatial symmetry in the transient refractive indices of the two arms. In a high-level picture, one can think of this state of the metasurface as the Tr-SRR being geometrically symmetric but with two arms made of different materials. The correspondence between the mirror asymmetry of the Tr-SRR and chirality is guaranteed as the nanostripes play a dominant role in removing the mirror symmetry plane of a standalone Tr-SRR, which the symmetry plane is oriented parallel to the substrate. Henceforth, upon judicious photoexcitation, the achiral platform can be converted to a chiral metasurface that interacts with the two CP states of a probe light differently. By mirror-flipping the linear polarization state of the pump light we optically create enantiomers with opposite handedness desired for a targeted CPL state. The temporal response of the induced chiral behavior is entangled to the dynamics of hot electrons. Initially, the pump light creates a range of high-energy non-thermal electrons within the high field regions of Au. In a few hundreds of femtoseconds, the successive emergence of electron–electron scattering establishes a quasi-thermalized electron system in Au, in which the energy distribution of electrons follows a Fermi-Dirac statistic with an elevated electron temperature. This elevation of the electron temperature leads to a fast and local modulation of the Au refractive index, where the resulting refractive index gradient induces chirality. As time passes the meta-molecule loses its chirality as the spatial electron temperature distribution homogenizes via spatial diffusion, rendering the metasurface as a transient chiral platform (Fig. 1B).

Fig. 1.

Concept schematics and sample configurations. (A) Visual sketch of the photoinduced transient chirality. The asymmetric near-field profile, sensitive to the polarization state of the pump light, transforms the achiral metasurface into a handedness-selectable chiral object which interacts differently to LCP and RCP probe pulses. (B) Conceptual schematic of the transient optical response and the microscopic origin of the transient chirality. The temporal behavior undergoes several phases where the description of the dominating hot-carrier dynamics is illustrated. (C) Design of the dual-layered achiral metasurface characterized in this work where the Tr-SRR and the nanostripe are separated by an 80-nm-thick spin-on-dielectric. The dimensions are given as p = 350 nm, d = 80 nm, and t_au = 40 nm. (D) Scanning electron microscopy image of the sample with further geometrical design specifications: s = 125 nm, w = 85 nm, l = 240 nm, h = 60 nm, and side angle of the Tr-SRR equals to 60°. Note that the polarization angle is defined as positive along clockwise rotation. The scale bar equals to 200 nm. (E) Measured linear optical transmission spectra under normal incidence for the two CPs with an Inset showing the magnified region where CD is apparent.

Fig. 1C shows details of the sample fabricated on glass substrate (Materials and Methods), where the two layers of nanostripes and Tr-SRRs are separated by 80-nm-thick spin-on-glass. Nanostripes were chosen as the consisting units of the lower layer as the spatial invariance, in the direction perpendicular to the mirror axis of the Tr-SRR, suppresses misalignment-induced intrinsic chirality occurring from fabrication. Both layers of plasmonic nanostructures are apparent in a scanning electron microscopy image of the fabricated structure given in Fig. 1D. As graphically described, we choose the mirror axis as the reference for the polarization angle of the incident light. The measured static optical response of the fabricated sample in Fig. 1E indeed shows minimal discrepancies in the transmission spectrum between LCP and RCP illuminations, which couldn’t be completely removed due to fabrication imperfection. Here, we attribute the main cause of chirality to stem from the geometric mirror asymmetry between the two arms of the Tr-SRR (SI Appendix, Note 1). The transmission dip originating from the plasmonic resonance of Tr-SRR is labeled as λ_CP (702 nm). To characterize the spectral and temporal responses of the device, the transient CD (ΔCD) of the device is defined as:

$$\mathrm{\Delta }\text{CD} (\lambda ,t) = \mathrm{\Delta }{\mathrm{T}}_{\mathrm{RCP}} (\lambda ,t)-\mathrm{\Delta }{\mathrm{T}}_{\mathrm{LCP}} (\lambda ,t),$$ [1]

where ΔT stands for the change in the optical transmission of the metasurface triggered by hot-carrier excitation. In the following, we show experimental proof of the envisioned concept by performing a comprehensive set of transient optical pump-probe experiments.

Hot Electron–Induced Chirality.

We start the characterization by employing a mirror-symmetric linearly polarized pump light with a center wavelength of 880 nm to excite hot electrons in Au Tr-SRRs. We then use broadband LCP and RCP weak probe pulses to monitor the temporal evolution of the chiroptical response upon hot-electron generation. The measured transient optical response of the sample is expressed in ΔOD ($\lambda ,t$) = −log₁₀ [T($\lambda ,t$)/T₀($\lambda$)]. Here, T₀($\lambda$) is the static transmission of the probe light through the sample and T($\lambda ,t$) is the transient transmission at a specific time delay with respect to the arrival time of the pump light. As the simulated electric-field distribution in Fig. 2A shows, when we set the pump polarization parallel to the longitudinal axis of the gold stripes, a symmetric field enhancement profile within the Tr-SRR is formed. The corresponding transient ΔOD maps for LCP and RCP probes are illustrated in Fig. 2B, which generally look similar. The large ΔOD values around λ_CP stems from the spectral redshift and broadening of the plasmonic mode owing to the modification of Au refractive index by generated hot electrons (37). Fig. 2C shows the measured ΔCD ($\lambda ,t$) map under a mirror-symmetric linear pump polarization. Although in principle this pump polarization should not induce any discernible CD, the ΔCD ($\lambda ,t$) map exhibits a transient feature in the vicinity of the plasmonic mode. This behavior is attributed to the intrinsic CD of the device (Fig. 2C, red spectrum), which originates from fabrication imperfections. As such, the generation of hot electrons spectrally shifts the resonance mode, which subsequently shifts the spectrum of the measured CD (Fig. 2C, blue spectrum). As plotted in the rightmost graph, spectral features of ΔCD and intrinsic CD are located within a spectral vicinity.

Fig. 2.

Transient optical response and induced chirality. Ultrafast pump-probe spectroscopy was conducted under specified pump conditions: pump fluence of 3.4 mJ/cm² for symmetric excitation and 4.5 mJ/cm² for asymmetric excitation, pulse duration of 89 fs, and a center wavelength of 880 nm with varying polarization angles. The pump fluence was adjusted to account for the polarization-dependent absorption. (A) Numerically simulated internal electric field enhancement profile of the Tr-SRR under a +90° polarized pump displaying mirror symmetry, where the cross section is taken at the top of the Tr-SRR. (B) Differential transmission response under RCP and LCP probe (Center). (C) Measured 2D map of transient CD as a function of wavelength and delay time (Top Left), and a 1D cross-sectional view captured at a fixed time delay indicated by a dashed line on the 2D map (Bottom Left) which is then plotted together with the intrinsic CD profile (Right). (D and E) Field profile and ΔOD maps under +50° pump polarization selectively exciting the left arm of the Tr-SRR. (F) Measured transient chirality response. We decompose ΔCD into components of intrinsic chirality modulation (ΔiCD) and photoinduced chirality due to left arm excitation (ΔpCD_L). (G and H) Results repeatedly acquired under −50° pump polarization selectively exciting the right arm of the Tr-SRR. (I) ΔCD measurements where near-perfect symmetric response between ΔpCD_R and ΔpCD_L are observed unlike the overall ΔCD values which are summated responses of ΔiCD and ΔpCD.

Next, we employ a linear pump polarization state that enables asymmetric concentration of the light field within the Tr-SRR, which breaks the mirror symmetry of the refractive index profile of the nanostructure upon the generation of hot electrons. As Fig. 2D exhibits, while the spatial profile of field enhancement overlaps with the mirror-symmetric pump, a +50° input polarization of the 880 nm pump primarily concentrates the electric field around the left bend of the Tr-SRR. At incident wavelengths near λ_CP, RCP experiences larger field concentration at the left arm compared to LCP (SI Appendix, Note 2), which therefore undergoes more pronounced modulation in light transmission through the device (Fig. 2E). As Fig. 2F depicts, under the +50° pump polarization, the measured CD is notably improved compared to the case of 90° pump polarization, indicating that the generation of hot electrons indeed impacts the sensitivity of the device to the CP state of the probe beam. To understand the nature of the measured ΔCD (Fig. 2F), we introduce two transient processes that contribute to the modulation of the chiroptical response of the device: i) the spectral modification of the plasmonic resonance originating from the hot-carrier-induced Kerr optical nonlinearity of gold (28, 36, 37); ii) the transient spatial inhomogeneity of the refractive index, which spatiotemporally varies with the energy evolution of hot electrons. The importance of the first process becomes evident once we remember that the device has a static chiral behavior around the resonance because of the nonideality in the fabrication of the device. As such, the modulation of the plasmonic resonance is expected to contribute to the change of the plasmonic response and subsequently CD. We refer to this intrinsic geometry-dependent term as ΔiCD, where “i” represents “intrinsic.” The second and more important term is indeed the outcome of the asymmetry of the refractive index change, creating a transient sensitivity to the CP state of the probe light as elaborated previously. We refer to this term as photoinduced chirality (ΔpCD) of the device or “photochirality” in short. To distinguish the contribution of these terms, we first emulate the Kerr-induced geometric chirality modulation by employing a mirror-symmetric pump polarization orientation (90°) with adequately adjusted input power. Subsequently, subtracting the obtained transient from the ΔCD signal leaves out only with the ΔpCD component (Fig. 2 F, Right). Note that the temporal nature of the Kerr-induced ΔiCD process and the polarization-independent field enhancement locations allow this approach to accurately decouple the resulting transients of the two aforementioned processes (SI Appendix, Note 3). In Fig. 2 G–I, corresponding plots under a mirror-flipped pump polarization at a −50° direction are presented. As clearly seen, the ΔCD map exhibits an inversion of the sign indicating that the transmission of the LCP probe through the device is modulated more strongly. Indeed, a comparison between the ΔpCD spectra (Fig. 2 I, Right) shows a nearly perfect inversion of the transient chiroptical response of the device, demonstrating the optical control of the device chirality via a linearly polarized pump light.

The correlation between the dynamics of anisotropic hot carrier distribution and photochirality can be further investigated by analyzing their temporal characteristics. As a starting point of reference, we numerically simulate the excitation and spatial diffusion of hot electrons within the Tr-SRR via the extended two-temperature model (eTTM, see Materials and Methods for details). The computational results are presented in Fig. 3A, describing the delay-time-dependent spatiotemporal profiles of the electron temperature (T_e). Here, we assume the same pump condition as in Fig. 2 D–F. At first glance, we notice that the global maximum of T_e (color-coded as bright yellow) occurs at the terminus of the left arm, which doesn’t coincide with the light field concentration locations. This exemplifies the spatial diffusion of T_e being a rapid process compared to local heat generation and relaxation. We then track the temporal evolution of the electron temperature at the two terminuses. The right plot of Fig. 3A describes the time evolution of T_e at the monitored locations, where correspondences between notable temporal features of T_e and their spatial profiles are indicated with geometric markers. In line with our previous description, the rapid spatial diffusion of electron heat results in a faster relaxation time in the ΔT_e curve (blue) compared to T_e (red and green curves). The comparison between the time evolution of electron temperature and experimentally measured optical transients denotes the major role of spatially anisotropic electron energy distributions on the creation of photochirality and its faster decay time compared to the differential transmission responses. As shown in the left plot of Fig. 3B, the sampled ΔCD response and the ΔT responses for RCP and LCP probes from Fig. 2 E and F at 684 nm show a fair resemblance with the right-side plot in Fig. 3A. The decay time constants for the ΔT response for RCP and LCP probes, and the ΔCD response were then obtained for quantitative analysis. The time constants were obtained by fitting the experimental values to ex-Gaussian functions where we see a near-identical fit between the experimental and fitted values (SI Appendix, Note 4). Within the ΔCD curves which already has a shorter lifetime than the ΔT response, further decomposing ΔCD into ΔiCD and ΔpCD (Right panel of Fig. 3B), we see a fast and slow component where the fast component corresponds to the transiently induced chirality. ΔpCD quickly approaches to an infinitesimal value as ΔiCD approaches the ΔCD value, making ΔpCD a fast component with sub-picosecond response time. ΔiCD on the other hand has a lifetime similar to the ΔT response times, indicating that they share similar modulation mechanisms. In essence, the dominance of the nonlocal thermal relaxation channel over local thermal response suggests opportunities for expediting the temporal response time of photochirality through rational design of nano-resonator geometry.

Fig. 3.

Temporal characteristics of the photoinduced chirality. (A) Numerically calculated spatial maps of the T_e at different time stamps (Left). The time evolution of the electron temperature at the two terminuses and their difference are plotted (Right), where notable temporal features and its correspondence to the spatial maps are indicated with basic geometric markers (triangle: peak of input pump light, circle: maximum temperature difference, square: maximum temperature at the right terminus, star: temperature homogenization). The gray-colored curve shows the temporal Gaussian profile of the input pump pulse. (B) Transient curves and the decay time constants of ΔT_LCP, ΔT_RCP, and the ΔCD response observed at λ = 684 nm under +50° pump excitation (Left). The ΔCD curve is then decomposed to temporal profiles of ΔiCD and ΔpCD, where the decay timescales provide insights into the primary contributing factor among different hot-electron-related mechanisms for each feature (Right). The transient curves are sampled from Fig. 2 E and F, and the measurement conditions as in Fig. 2 D–F were imported to the numerical simulations.

Functionalities of Hot Electron-Induced Photochirality.

The transient chirality offers unique features that can be taken advantage of, one of which is optical rotation. While ideally an incident light linearly polarized along the mirror axis of the sample shouldn’t experience any polarization conversion, the input light can be converted to an elliptical form once transient chirality is present. As depicted in Fig. 4A, the long axis of the output light undergoes different degrees of rotations for varying wavelengths, known as optical rotatory dispersion (ORD). CD and ORD are Kramers–Kronig-related quantities each rising from the difference in the imaginary and real parts of the circular-polarization-dependent refractive indices. We perform transient Stokes measurements in order to characterize the ultrafast polarization rotation with a vertically polarized probe light (Materials and Methods), where the pump conditions were kept same as those in the ΔCD measurements. We also decompose ΔORD signals into ΔiORD and ΔpORD components as we did for the ΔCD signals since the Kramers–Kronig relation is a linear operation. The measured 2D maps of ΔpORD are shown in Fig. 4B as a function of probe wavelength and time delay. The spectral profile of ΔpCD and ΔpORD are plotted together in Fig. 4C at time delays when the amplitudes are maximized. The qualitative spectral features (peak-like and bisignate) and locations satisfy a typical Kramers–Kronig pair, suggesting a minimal impact of cross-polarization conversion from the C1-symmetric structure (SI Appendix, Note 5). Similar to ΔCD measurements, input pumps with mirror-inverted polarizations (−50° and +50°) result in induced ultrafast optical activities reaching up to a maximum rotation of ~10°/λ, but with opposite rotational directions. The measurement signifies the capability of a single metasurface to rotate linearly polarized light in opposite directions with continuous tunability enabled by simple pump power or polarization adjustments.

Fig. 4.

Transient ORD. (A) Conceptual schematic of photoinduced ORD (ΔpORD). (B) 2D maps of measured photoinduced optical rotation (Δθ) under +50° and −50° pump excitation acquired via Stokes polarimetry. (C) Comparison of the spectral lineshapes between maximum ΔpCD and maximum ΔpORD showing qualitative agreement as a Kramers-Kronig pair. (D) Sign-flipped responses of ΔpORD for pump excitations with mirror-inverted polarization orientations.

As another example, the deep correlation between near-field profiles and transient chirality brings connection between the far-field chiral response and near-field chiroptics—an area of study that offers many promising applications such as enantiomer detection and separation, chiral light source, and chiral crystal growth (14, 39–42). Although mirror-symmetric nanostructures possess achiral far-field response, chiral near-field response can still be observed in the form of circular dichroic hot-spot distributions (43) and optical-chirality-enhancement profiles (44, 45). Optical chirality is a measure of the twist in the near-field which can be computed for electromagnetic fields with complex electric E field and magnetic B field as (46, 47)

$$C=-\frac{\omega {\epsilon }_0}{2}\text{Im}({\mathbf{E}}^{\ast }·\mathbf{B}),$$ [2]

where ${\epsilon }_0$ is the free space electric permittivity, $\omega$ is the angular frequency of incident light and E^* is the complex conjugate of the electric E field. The Left panel of Fig. 5 shows simulated near-field chiroptical responses including field enhancements and optical-chirality enhancements ($\widehat{C}=C/|C_0|$, where $C_0$ is the optical chirality of the incident free space CPL) of our sample under LCP $\left(\widehat{C}=-1\right)$ and RCP $\left(\widehat{C}=+1\right)$ illumination. The microscopic origin of the lack of far-field chirality despite the presence of chiral near-fields is their collective effect from different regions of the nanostructure canceling each other out (46). Thus, the inability to probe near-field chirality with far-field optics requires alternative methods to observe near-field chirality (43–45). Here, ultrafast spectroscopy can serve as another approach to monitor the near-field chiral response of achiral plasmonic nanostructures adhering to specific geometric conditions, as their hot spot distributions under LCP and RCP illumination suffice the requirement for hot-electron-induced chirality. The resulting far-field chiroptical responses under same pump conditions but with LCP and RCP polarization states are given in the Right panel of Fig. 5, where we again observe photochirality with opposite handedness. Therefore, we show that near-field chirality can be indeed probed with far-field optical measurement techniques.

Fig. 5.

Bridging near-field and far-field chiroptics. Numerically simulated enhancement profiles of optical chirality and electric field magnitudes of the achiral dual-layered metasurface under LCP and RCP illumination at 880 nm pump wavelength with a fluence of 4.5 mJ/cm², showing circular dichroic spatial profiles (Left). Emergent far-field chirality due to hot-electron generation, where the transient chiral responses induced by LCP and RCP pump display opposite signs in CD, allows the chiral near-field distribution to be differentiated through far-field optics. Spectral cuts of the transient photochirality maps are also shown, where the corresponding delay time is indicated as dashed lines.

Discussion

A unique strategy for creating chirality with desired handedness in achiral nanostructures was introduced. The spatial asymmetry of hot-electron-induced complex refractive index distribution originating from the near-field distribution of the control light is introduced as the strategy to optically break the mirror symmetry within the achiral plasmonic nanostructures. Simple tuning of the excitation polarization can induce chirality with near-perfectly-invertible handedness, making the technique a powerful tool for achieving dynamic control over the chiroptical response. Transient chirality especially shows superior switching speed compared to conventional hot-electron-based all-optical switches employing differential transmission responses, since electron heat diffusion occurs at a faster rate compared to electron–phonon relaxation. Overall, the nonlocal nature of transiently induced chirality makes it possible to further enhance the magnitude of photochirality, boost the switching speed, or adjust the operation wavelength with ease via judicious geometrical design. Ultrafast optical rotation and near-field chirality probing were explored, showcasing the method as an alternative addition to conventional techniques utilized for high-speed optical rotation and near-field probing. We envision our method to spur an additional branch of designs for chirality switching and benefit both in technological and academic aspects including ultrafast polarization control, active chiroptics, and hot-carrier photophysics.

Materials and Methods

Device Fabrication.

The double-layered plasmonic structures were fabricated on a glass substrate (Corning, Eagle XG glass) via two runs of aligned e-beam lithography, where each run follows a three-step fabrication process: i) a standard electron beam lithography process (Elionix ELS G-100) to define the nanopatterns, ii) electron beam evaporation of 2 nm/40 nm Titanium (Ti)/Gold (Au) metal using Denton Explorer, and iii) a lift-off process in acetone to resolve the plasmonic structures. The nanostripes and the alignment marks were formed during the first run, which was then followed by spin-coating a transparent dielectric IC1-200 (Futurrex Inc.). In order to ensure the desired spacing (40 nm) between the first and second layers, the spin-on-glass was then etched down to a thickness of 80 nm via reactive ion etching (Unaxis RIE). Subsequently, the Tr-SRR structures were fabricated on top of the spin-on-glass through the second run of e-beam lithography (SI Appendix, Note 6).

Static Optical Characterization.

A collimated broadband light from a fiber-coupled tungsten halogen lamp (B&W Tek BPS 120, spectral range 250 to 2,600 nm) was used to characterize the broadband optical response of the device. A linear polarizer and an achromatic quarter waveplate were used to create LCP and RCP. The transmission spectra of the device at normal incidence were acquired by a homemade spectroscopy system (SI Appendix, Note 7). The light was focused onto the device with a lens and then the transmitted light was collected through an objective lens (Mitutoyo, 20× plan Apo NIR infinity-corrected). The collected light was delivered to the spectrograph system (Princeton Instrument Acton SP 2300i and PIXIS 400B camera), where the transmission of the device was normalized to the transmission through air.

Ultrafast Optical Characterization.

The ultrafast pump–probe spectroscopy setup was operated through a regenerative amplified Ti:sapphire femtosecond laser system (Spectra-Physics Solstice Spitfire ACE, 2.5 kHz repetition rate, 89.3 fs pulse width, 800 nm wavelength, and a pulse energy of 1.6 mJ per pulse), and the data collection was conducted using a Helios spectrometer (Ultrafast Systems Inc.). An 80:20 beam splitter divides the 800 nm fundamental light from the laser system into two beams. The low-intensity part of the fundamental light passes through another 80:20 beam splitter, and the power level gets reduced again by 80%. The laser beam then gets focused on a 2-mm-thick sapphire window to generate a white light continuum (WLC) as the probe signal. Prior to WLC generation, the low-intensity beam first passes through a motorized delay stage in order to control the delay between the pump and probe signal, which has a minimum resolution of 30 fs. The probe beam was then focused onto the sample using a parabolic mirror, and the transmitted probe light was collected by an optical fiber coupled into a visible spectrometer. The high-intensity part of the fundamental light is used to operate an optical parametric amplifier (Ultrafast Systems Inc. Apollo). The fundamental light is partially converted into two near-IR pulses, signal, and idler, and the combination between the three beams via different nonlinear processes produces pump pulses with desired wavelengths. The pump pulse is also focused on the sample through a parabolic mirror, where the pump is incident with a 10^∘ offset from the probe light. An instrument response function value of 150 fs typically returns good numerical fittings for the transient kinetics. Chirp corrections and coherent artifact removal of the obtained transient maps were carried out via post processing.

Numerical Modeling.

The electric field distributions within the Tr-SRR were calculated through a commercially available software package (COMSOL 6.0, wave optics module). The refractive index of Au was retrieved from existing literature (48). More simulations on the static transmission response of the dual-layered metasurface can be found in SI Appendix, Note 1.

The spatiotemporal energy evolution of hot electrons was computed through the following coupled partial differential equations (35):

$$\begin{array}{c}\frac{\partial N(\overrightarrow{r},t)}{\partial t}=\frac{1}{C_e\left(T_e\right)}\nabla ·\left[{\kappa }_e\left(T_e,T_l\right)\nabla N\right]-\left(a+b\right)N+Q(\overrightarrow{r})·A(t),\hfill \end{array}$$ [3]

$$C_e\left(T_e\right)\frac{\partial T_e\left(\overrightarrow{r},t\right)}{\partial t}=\nabla ·\left[{\kappa }_e\left(T_e,T_l\right)\nabla T_e\right]-G_{e-ph}(T_e-T_l)+aN,$$ [4]

$$C_l\frac{\partial T_l\left(\overrightarrow{r},t\right)}{\partial t}={\kappa }_l{\nabla }^2{T}_l+G_{e-ph}(T_e-T_l)+bN,$$ [5]

where $N(\overrightarrow{r},t)$, $T_e(\overrightarrow{r},t)$, and $T_l(\overrightarrow{r},t)$ describe the energy density stored in the non-thermal portion of hot-electrons (J m⁻³), the effective electron temperature, and the lattice temperature, respectively. The thermal conduction rate of the thermal and nonthermal electrons are assumed to be equal. Detailed descriptions of other parameters and their values incorporated in the eTTM are provided in SI Appendix, Note 8. The absorbed power density by an Au nanostructure under a pump angular frequency of ${\omega }_{\mathit{pump}}$ can be expressed as:

$$Q\left(\overrightarrow{r}\right)=\frac{1}{2}{\epsilon }_0{\epsilon }_{\mathit{Au}}^{\mathit{"}}{\omega }_{\mathit{pump}}\langle |{E\left(\overrightarrow{r},t\right)|}^2\rangle ,$$ [6]

which can also be simulated through the numerical software. To replicate the experimental condition through numerical modeling, a linearly polarized plane wave excitation configuration with 880 nm wavelength and an incident intensity of $\frac{2\sqrt{\mathrm{l}\mathrm{n}2}F}{\sqrt{\pi }\mathrm{\Delta }t}$ was assumed. Here, $F$ stands for the incident pump fluence and $\mathrm{\Delta }t$ stands for the pulse width. Once the peak absorbed power density is calculated, the peak absorption is scaled by a temporal gaussian profile:

$$A\left(t\right)=\exp \left[-4ln2{\left(\frac{t}{\text{Δ}t}\right)}^2\right],$$ [7]

as Gaussian pulses are employed as pump light. Thus, we can simulate the time-dependent absorbed power density for our experimental setup. The computed time-dependent energy input is then imported to the PDE Interfaces module as the energy source in order to solve Eq. 3. Due to electron-temperature-dependent heat capacity and thermal conductivity of electrons, Eq. 3 is solved simultaneously with Eqs. 4 and 5, which are computed through the Heat Transfer module available in COMSOL.

Optical Rotation Measurements.

The ultrafast optical rotation of light was characterized by installing a liquid crystal (LC) phase retarder (Thorlabs) and a linear polarizer between the sample and the detector. The incident probe light was vertically polarized, which corresponds to 0° based on our notation. We measure the intensity of the transmitted light through the linear polarizer under four different conditions: (β = 0°, 15°, 30°, 45°; α = 0°), where β and α are defined as the direction of the fast axis of the LC and the transmission axis of the linear polarizer, respectively. The induced phase difference between electric field components that are parallel and perpendicular to the fast axis of the phase retarder is notated as δ. The δ values for our LC phase retarder are highly dispersive, which the values as a function of wavelength are given in SI Appendix, Note 9. The Stokes parameters (I, M, C, and S) were retrieved from the following equation (49):

$$\begin{array}{c}{I}_T\left(\alpha ,\beta ,\delta \right)=\frac{1}{2}[I+\left(M \text{cos} 2\beta +C \sin 2\beta \right)\text{cos} 2\left(\alpha -\beta \right)\hfill \\ +\left\{\left(C \text{cos} 2\beta -M \text{sin} 2\beta \right)\text{cos} \delta +S \text{sin} \delta \right\}\text{sin} 2\left(\alpha -\beta \right)].\end{array}$$ [8]

And the polarization rotation can be computed through the following relation:

$$\theta =\frac{1}{2}{\text{tan}}^{-1}\frac{C}{M}.$$ [9]

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.