Chip-Based Chiral 3D Metamaterial with Functional Core-Shell Architecture for Femtomolar Biodetection

Advanced sensing tools capable to detect extremely low concentrations of circulating biomarkers can open unexplored routes towards early diagnostics of diseases and their progression monitoring. Plasmonic sensors are an emerging technology enabling different optical effects that can be used as molecular tracking solutions. Here we demonstrate the sensing capabilities of a chip-based metamaterial, combining the 3D chiral geometry with an optically functional core-shell architecture. The sensor can be easily handled and exhibits reliability and stability during the whole functionalization and analytical procedure thanks to the on-chip format. The system shows a linear shift of circular dichroism spectrum upon interaction with different concentrations of TAR DNA-binding protein TDP-43, a clinically relevant biomarker for neurodegenerative disease screening. The measurements were performed in spiked solution as well as in human serum, with concentrations from 1pM down to 10fM, a range not accessible with commonly used immunological assays and that can thus open new perspectives for disease knowledge and early diagnostics.


Introduction
In biomedical research and clinical practice there is a huge demand of high performance sensors for advanced diagnostics and real-time monitoring of disease evolution 1,2 , outperforming and simplifying standard approaches, such as real-time polymerase chain reaction (RT-PCR) and the enzyme-linked immunosorbent assay (ELISA). Continuous technological improvements are required to increase sensitivity, speci city, and accuracy towards the challenging detection of small-size molecules or ultralow concentration target analytes. In the last years, arti cially engineered optical devices exploiting plasmonic properties, emerged as good candidates to be implemented as low-cost, miniaturized and multiplexing biosensors [3][4][5][6][7][8] . Plasmonic biosensors are mostly based on the surface plasmon resonance (SPR) along metal/dielectric interfaces, and on the localized surface plasmon resonance (LSPR) occurring in nanostructures. Both detection mechanisms are strongly sensitive to the refractive index changes of the surrounding medium, within their respective plasmon decay lengths. Generally, SPR sensors exhibit higher sensitivity than LSPR, but require more complex excitation optics (such as prism or grating coupling) and extended smooth surfaces.
In order to improve the sensor performances, one strategy consists in the material and shape engineering of plasmonic nanostructures [9][10][11] . In particular, the addition of anisotropy to plasmonic nanoobjects could enable novel degrees of freedom and polarization dependence in the optical response and, consequently, new biosensing concepts. One example are chiral plasmonic nanostructures 12,13 , characterized by the absence of the mirror symmetry, which exhibit handedness and behave differently when interacting with circularly polarized light (CPL). In this respect, the circular dichroism (CD) spectroscopy, de ned as the differential measurement of optical absorption in chiral enantiomers upon interaction with left and right-handed CPL 14,15 , has a huge potential in the biosensing eld because of its differential nature, free from background noise, and for the presence of several spectral features, other than the typically broad plasmonic resonances. Indeed, recent studies reported the modelling and fabrication of chiral plasmonic metamaterial with various morphologies mostly for enantiomeric detection, based on the plasmonic driven ampli cation of natural occurring CD signals [16][17][18][19][20][21] . The employment of chiral plasmonic elements to detect changes in the surrounding medium induced by target analyte presence is mainly limited to few works, employing chiral nanoparticles dispersed in solution [22][23][24][25] . However, in view of developing novel biosensing schemes of practical employment towards portable point of care devices, a chip-based approach with sensing nanostructures, organized with a long-range order and integrable in photonic circuits, rather than dispersed in solution, would signi cantly impact the spread of such a technology 26 . It will offer great advantages in terms of repeatability, integration with micro uidics components and stability against different solvent and salt concentration, that can instead induce precipitation if colloidal nanoparticles are used. Furthermore, the chip-based format is particularly suitable for analytical procedures, allowing easy washing steps that are essential for real application with complex biological samples, rather than simpler protein solutions 27,28 .
In the present work, we demonstrate the possibility to exploit a chip-based approach for chiral sensing technology through compact helix metamaterial arrays, engineered with a functional metallic core/dielectric shell composition. As dielectric shell we have used the same polymer-mediated surface functionalization layer, needed for high-yield and speci c biorecognition of any type of biomolecule. Such a composite structure, modifying the metamaterial far-eld and near-eld response, leads to a more e cient interaction and detection of biomolecules, as compared to other bioconjugation schemes. The device is used to detect a neurodegenerative related biomarker, for which speci city and detection up to the femtomolar range are demonstrated in controlled solutions as well as in complex human uids.
The scheme of our biosensor is shown in gure 1. The engineered building block, grown by Focused Ion Beam Induced Deposition (FIBID) 29 , consists of a periodic array of chiral core-shell nanostructures, where the single element is a free-standing metallic helix, that, given the 3D nature, can have the additional advantage of increasing the available binding surface exposed to the analyte 30 . The chiral metamaterial is then prepared for biochemical functionalization through the conformal coverage with an ultrathin dielectric polymeric shell, the Poly-o-phenylediamine (P-oPD). Such an approach is chemically simple, cost-effective and time-saving, while ensuring stability, effective coverage and selectivity of the sensing nanostructures especially for analysis of complex biological matrix 31 . The resulting core-shell architecture, with respect to dielectric function pro le, represents the key element to achieve high sensitivity, because, on one side, it offers an ideal surface coverage for high stability molecular immobilization onto the 3D nanohelices, while, on the other side, it allows to enhance the near-and fareld optical response of the nanohelices.
After antibody covalent binding, we applied our engineered sensing device against variable concentration of the transactive response (TAR) DNA-binding protein 43 (TDP-43), a distinctive protein of amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD) 32,33 . Currently, the diagnosis of these neurodegenerative diseases is still clinically-based and no biomarkers have yet been routinely incorporated into the clinical practice or clinical trials. TDP-43 is the main component of the pathological inclusions found in the cytoplasm of neurons and glial cells of the majority of ALS and Tau-negative FTLD cases. As a consequences TDP-43 has been largely proposed and studied as a potential biomarker for ALS and FTLD. Increased level of TDP-43 protein was found in cerebrospinal uid (CSF) and plasma of patients with ALS and FTLD 34 using mass spectrometry 35 ,Western Blot and ELISA analysis 36,37 . However, the extremely low concentration of TDP-43 in bio uids represents a major limitation for its use as diagnostic biomarkers and requires the implementation of more challenging detection methods 38 .
Recently, an ELISA test detected TDP-43 in CSF at concentration below 0.49 ng/mL 34 and a detection limit of 0.5ng/mL 39 has been achieved in serum using an electrochemical sensor. However, quanti cation of TDP-43 in CSF and plasma results highly variable across studies 38,40 and the large part of patients and healthy subjects show TDP-43 levels below the detection limit of the typical immunoassays 36,37 . In this respect, more sensitive assays are required to measure low TDP-43 concentrations present in body uids for a more accurate validation of this biomarker and even for early diagnosis of ALS and FTLD.
Our sensor demonstrated to detect TDP-43 concentrations down to 10fM (corresponding to 0.43pg/mL). Moreover, according to speci city control experiments, our device scheme results robust against nonspeci c background noise in both dry and liquid environments, providing a fast and real-time detection even in complex body uids like human serum.

Results
Fabrication and characterization. The biosensing device active area consists of a compact and periodic array of right-handed platinum nanohelices ( Figure 2a). The geometrical and structural parameters have been engineered in our former work 41 to achieve optimized chiro-optical effects in the visible spectral range. More details on fabrication process are reported in the Methods section.
Afterwards, in order to build the e cient sensing core-shell scaffolding, we have exposed the nanohelices to oPD for polymerization by means of a cyclic voltammetry (CV) process (see Methods for further details) for the conformal deposition of a thin and compact dielectric polymer shell [42][43][44] . Such a functionalization strategy presents many structural advantages. First, the used monomer, the opheniledyamine (oPD), is characterized by a self-limiting polymeric growth under anodic oxidation in aqueous solution affording the formation of an ultrathin coating layer (few nanometers), where the aminic groups can be exploited to bind biomolecules. The self-limiting deposition is controlled by the polymer concentration, the buffer composition and the buffer pH in the CV process 45,46 .
Second, the C a nity to the polymer ensures a homogeneous immobilization of molecules on the helix surface, consisting of an alloy of Pt crystalline nanograins uniformly embedded into an amorphous carbon matrix 41 .
The shell morphology obtained on a Pt-nanohelix, grown on-purpose on a copper grid before and after P-oPD coating, has been studied by Scanning Transmission Electron Microscope (STEM) in Z-contrast mode, as shown in gure 2b, c, d. Figure 2b, obtained by combining both dark (DF) and bright eld (BF) acquisition modes, evidences the conformal coverage of the low index polymeric shell around the helix core. High-magni cation STEM images of the helix, acquired in dark eld mode, before and after the deposition ( gure 2c and 2d), show a very high uniformity of the polymeric thin shell throughout the whole helix structures beyond the shining helix core edges (red arrows), with average thickness of 12±2 nm. The random dark/bright contrast visible into the helix core is related to the complex composition of the structure, with platinum grains appearing dark because of the high Z, and carbon matrix, with lower atomic number, appearing bright.
Optical properties of the 3D metal/dielectric core-shell nano-helices. The circular dichroism spectra of both the core and the core-shell systems are calculated by the experimentally measured (see Methods for measurement details) intensity of left and right-handed circularly polarized transmitted light (T LCP and T RCP , respectively), according to: The CD spectrum of the bare helix-based sensor exhibits two opposite dichroic bands (D1 and D2 centered at λ M =500nm and λ m =840nm, respectively as shown in gure 3a, blue line) due to hybridization among the chiral dipoles of the helix arms 47,48 , with a zero dichroism point (ZDP) at λ ZDP =600nm. After the shell deposition, a spectral redshift of the CD bands occurs because the refractive index around the nanostructure increases ( gure 3a red line). The unpatterned substrate remains barely affected by such a thin dielectric layer deposition ( gure S1 in the Supporting).
Core-shell nanosystems combining materials with different sign permittivities (i.e., metals with dielectrics) are expected to exhibit different effects on absorption and scattering at their resonance frequency, and to modify the electric eld distribution around the nanostructures [49][50][51] . This depends on the relative core-shell size and on the interplay between nanostructure shape and size and inspecting wavelength. In our speci c case, a decay length of 128nm is expected for the plasmonic eld 52 at the CD peak of 500nm through the shell, by considering the optical dispersion of the Pt-based core 41 and of the P-oPD shell (Supporting gure S2). Therefore, we can assume that the measured shell thickness of 12nm still supports the plasmon propagation out to the nanosystem surface.
Moreover, numerical analysis of the near eld distribution for both, bare and core-shell nanosystems, at the resonance peaks for the two incident CPL components (λ LCP and λ RCP ), ( gure 3 b, c) show that the electric eld hot spots, close to the metal/air interface in the bare helix, move towards the shell/air interface in the core-shell system. In such positions the eld intensity is also increased, thus suggesting a stronger interaction with surrounding biomolecules.
This bene cial contribution to system sensing capabilities is also evidenced by the experimental scattering spectra of RCP and LCP light, measured for the bare single helix, and after the deposition of the outer shell (Figures 3d-e). Along with the slight spectral redshift, in line with what observed from the CD spectra, the core-shell architecture, when incident CPL matches the structure handedness, induces an enhancement of the scattering intensity, by a factor of 1.1 with respect to the bare core. This enhancement further increases by a factor of 3 with the opposite incident handedness. The results are in agreement with the numerical simulations of the scattering intensity in the supporting gure S3. These two interrelated effects, the enhanced electric eld and the increased far eld scattering, could be attributed to the energy transfer between polarization charges of the dielectric shell and free electrons in the plasmonic core 53,54 . As seen later, with respect to biosensing application, the resulting interaction of polymer-coated helices with biomolecules can be more e cient, if compared with other monolayer-thin bioconjugation schemes.
Sensor characterization. A rst assessment of the core-shell system sensitivity was performed by measuring the CD in a known refractive index environment, that is glycerol-water mixtures varying concentration from 0 to 20% (corresponding to a refractive index range between 1.333 and 1.358) 55 , as shown in gure 4a, b, c.
The refractive index sensitivity, calculated from the linear ts as S=∆λ/∆n (where ∆λ represents the wavelength shift and ∆n the change of the refractive index of the glycerol-water solution), is reported for both λ M (squares) and ZDP (rhombs) in gure 4d. ZDP exhibits a higher sensitivity (S=766nmRIU -1 ) and a larger gure of merit (FOM =S/δλ) up to 1276 RIU -1 with respect to λ M (S=316nmRIU -1 and FOM= 235RIU -1 ) con rming the tracking e ciency of the zero dichroism point for molecular sensing 25,56 . A comparison with state of art performances for LSPR-based systems 31 demonstrates the large potentiality for the proposed biosensing approach.
The system was then tested for biomolecular recognition (see Methods for details) after antibody immobilization with gluthereldeyde (acting as crosslinker agent between the aminic groups of the polymeric shell and the antibody) and after the incubation of TDP-43 antigen. Analyte measurements at different concentration were then performed in dry medium (see Supplementary S4 for details) [57][58][59] .
We have measured the difference of the CD spectral features (Figure 5a) acquired after the antibody layer incubation and after the TDP-43 binding at different molar concentrations, ranging from 1pM down to 10fM, therefore beyond the interval accessible for this analyte through typical immunoassays 38,39 . ZDP ( gure 5b) con rms a larger CD redshift as compared to the maximum ( )and the minimum ( )CD features ( gure S5). The linear trend of the spectral shifts as a function of the molar concentration is noticeable in gure 5c, with a concentration sensitivity of 27nm/pM.
Speci city and imperturbability. The real optical and biochemical effectiveness of our functionalization method has been tested by comparing the TDP-43 detection results with the standard approach based on the self-assembly monolayer of thiols, commonly used for metallic surfaces, including platinum 60,61 . In the thiols-based functionalization experiment (see methods for details), the detected ZDP redshift for an analyte concentration of 1pM was 5nm ( gure S6), ve-times lower than the value obtained with the core-shell architecture (27nm, Figure 5). This result comes from the combination of the core-shell optical properties discussed above with the fact that, while thiols layer binds only to the Pt grains in the complex helix material alloy leaving the carbon surface fraction uncovered, the polymeric shell conformally coats the entire helix surface. This allows to maximize the total speci c binding sites available for the target analyte and to amplify the detected signal.
In addition, the sensor speci city has been tested performing two control experiments. In the rst case, we have used our nanohelix-based sensor with immobilized antibodies for TDP-43 on the polymeric shell, to reveal the non-speci c detection of Tau protein, a biomarker related to Alzheimer Disease (AD) and Parkinson Disease (PD) neurodegenerative diseases 62 . The sensor has been incubated with a solution of Tau protein at 500 fM molar concentration. As shown in gure 6a that compares the chiroptical measurements before and after the antigen deposition, the multiple CD spectral features show no distinct variation on addition of non-target Tau protein, because no speci c binding events occur, demonstrating that the nano sensor is selective and speci c to target molecules. Moreover, the possibility offered by our chiral sensor to use the CD spectrum introduces the additional property of the unperturbability to signals coming outside the active area. Indeed, considering that the thin layer of p-oPD is also deposited on the substrate, it creates binding sites for antibody-antigen pairs, generating the same transmission offset for both the circularly polarized transmitted lights, that the CD can delete because it is a differential signal. These results point out the excellent and stable performances and the reliability of the device.

Discussion
In summary, we engineered an ordered array of three-dimensional core-shell chiral nanostructures to be considered as novel chip-based optical biosensing concept. We analytically and experimentally demonstrated the sensing capabilities of our system owed to the combination of a metallic core of 3D chiral nano-helices conformally coated by a thin layer of dielectric polymer.
The high sensing performances are achieved thanks to the combination of many factors, namely: (i) the third dimension, which ensures a large binding area; (ii) the intrinsic chiral shape, which makes the sensor stable against the background interference and, consequently, suitable to work on optically dense environments like body uids; (iii) and the polymer-mediated functionalization, which provides additional advantages. On one hand, it creates a homogeneous and large surface for biorecognition. On the other hand, from an optical standpoint the metal-dielectric core-shell system enhances the scattering signal due to the interaction between polarization charges of the dielectric shell and free electrons of the plasmonic core. First of all, the engineering of such device provides refractive index sensitivity related to CD spectral features of about 800nm/RIU, which is among the highest achieved with LSPR-based sensors.
Then the sensor has been tested by evaluating the shift in due to the presence of TDP-43, whose identi cation at low concentration in body bio uids can help in the ght against neurodegenerative diseases. In a low concentration range, between 1pM and 10fM, the sensor shows a linear behavior with a CD shift of 27nm for TDP-43 picomoles.
Our platform also demonstrated a high speci city toward the target biomarker with very low signal due to aspeci c interaction of bioreceptor with other molecules such as Tau protein that is a biomarker related to different neurodegenerative disease: it would be of great importance from a clinical point of view since it would allow to distinguish between ASL and other neurodegenerative diseases.
Further improvement in sensitivity can be envisioned thanks to the size-, geometry-and materialdependent engineering of the sensor, leading to even more pronounced bisignated circular dichroism, with steeper stop bands, for larger and easier-to-detect spectral shifts. Thanks to the chip-based approach our device could be integrated with micro uidic tools, including for example sample preparation module, to separate human serum from a single drop of blood, to achieve a fully automated platform. In addition being a miniaturized photonic component, it can be embedded within a portable point of care system, with a simple integrated optical read out, considering that it works in transmittance under normal incidence conditions, thus not requiring complex excitation geometries 26 . Moreover, the demonstration of femtomolar detection of a protein such as TDP-43 and the very low aspeci c signal due to the interference of complex matrices of biological uids impose our analytical approach as an ideal candidate for the detection of blood-circulating protein biomarkers, especially in cancer diagnostics or infectious disease where speci c analysis are often complicated both by the low concentrations and by strong interference from serum/blood molecules 63,64 . This goal can be reached just by changing the recognition element and choosing an appropriate receptor able to bind to the target biomarker.
In addition, the intrinsic chirality of the nanohelices can open perspectives in enantiomeric detection, of critical importance for chemical and pharmaceutical applications. Furthermore, the nding that the coreshell architecture, used for biochemical functionalization, can enhance the near-and far-eld optical properties opens wide perspectives to explore novel nanophotonics schemes, not only for sensing but also for fundamental light-matter interactions.

Experimental Section
Sample Fabrication: A series of arrays of 3D Platinum-based Nano-helices single loop with lateral period (LP) 500nm, vertical period (VP) ranging from 450 to 550nm, external diameter (ED) 300nm and wire radius (WR) 90-120nm, were realized on ITO-on-glass substrate with Focused Ion Beam Induced Deposition technique employing a Carl Zeiss Auriga40 Crossbeam FIB/SEM system. This system allows the realization of 3D structures together with a gas injection system that contains the source of trimethylmethylcyclopentadienyl-platinum(IV) precursor. The gas is injected in the chamber and the ion beam (with parameters 1pA, 30KeV acceleration voltage and 10nm step size) dissociates the gas molecules to obtain, locally, a controlled and uniform growth of the nano helices array, sized 10x10μm. The vacuum chamber was kept within a pressure range from 8.80x10 -7 mbar to 9x10 -6 mbar during the deposition time. To perform the STEM characterization a single helix was grown on-purpose on a copper grid with the same growth conditions of the array.
Optical characterization: Transmission spectra were recorded with an optical microscope Zeiss Axioscope A1 with a home-made confocal system coupled to an imaging spectrometer. Light from a tungsten lamp is focused on the sample with a condenser with NA<0.1 and it is collected using a 40x objective lens with NA<0.95. Then, the light transmitted through the sample is guided through a system made by three lenses that reconstruct, collimate and refocus the real space image. The selected real image is reconstructed and directed to a CCD camera (Hamamatsu Orca R2) coupled with a 200 mm spectrometer for measurements in the visible spectral range and to an InGaAs detector (Princeton Instruments, OMA V InGaAs linear array) coupled with a 300 mm spectrometer in the near-infrared region of the spectrum. Adjustable square slits were used to select the array area. The circularly polarized light has been produced using a combination of a linear polarizer and a quarter-wave plate. For measurements in visible spectral range a linear polarizer and a superachromatic waveplate (Carl Zeiss, 400-800nm) have been used while for measurements in NIR region a linear polarizer (Thorlabs, LPVis100-MP 550nm -1.5µm) coupled to an achromatic quarter wave plate (Newport, Achromatic Waveplate 700-1000nm) have been used. All transmission measurements were normalized to the optical response of the substrate. The SEM and STEM images: SEM and STEM characterization was performed by means of a Merlin Zeiss microscope operating in scanning mode on helix array and single helix fabricated on a TEM grid copper by combining dark-eld and bright eld. In order to obtain Z contrast sensitiveness from the images, STEM was con gured in high-angle annular dark-eld mode.
Numerical simulation: Finite element method (FEM) simulation was developed through Comsol Multiphysics 5.4, by exploiting wave optics module and carrying out a frequency domain study of electromagnetic waves. All the physical dimensions were evaluated by realizing a geometry consisting of the Pt/C helix, the oPD shell, the air medium around the structure and a perfectly matched layer. Wavelength sweeping was performed in a range between 400 nm and 1000 nm with a step of 10nm, by exciting the structures with circularly polarized light through a background eld directed with zenith distance of p/3, then collecting far eld signal along zenith, in a solid angle of 0.7\π steradians, Regeneration. After optical measurements, the sensors can be regenerated through UV-ozone exposure followed by ethanol rinsing, leading to a weakening and complete remotion of the polymer chain with the linked antibody-antigen pair and returning to original CD spectral features. Scheme of the sensing device and functionalization protocol. The core-shell architecture arises from covering the fabricated helix array with the P-oPD insulating polymer. Then, the antibody is immobilized onto the shell after crosslinking by gluthereldeyde. Finally, the target analyte, deposited on the sample surface, is recognized by the speci c antibody sites in dry environment. Figure 2 a. SEM image of nano-helix array based active area; the geometrical parameters showed in the inset are lateral period (LP) 500nm, vertical period (VP) ranging from 450 to 550nm, external diameter (ED) 310nm and wire diameter (WD) 120nm. The scale bar is 1um while the scale bar of the inset is 100nm. b. DF+BF High Angle Annular Magni cation STEM of a core-shell Pt/P-oPD helix grown on a copper grid highlighting the thin and conformal nature of the P-oPD outer shell. c, d. STEM magni cation of a nanohelix section before (c) and after (d) the P-oPD coating. The red arrows indicate the oPD shell thickness of (12±2) nm.

Figure 3
a. CD spectra of nano-helices array before and after shell coating. b, c. Electric eld distribution pro le for both bare and core-shell single nano-helix calculated at the maximum resonance peak for RCP and LCP, respectively, showing a near eld enhancement after shell coating. The insets show the cross section electric eld intensity distribution in marked positions. d, e. RCP (d) and LCP (e) scattering spectra measured for a single helix before and after the polymeric shell coating. The curves highlight a strong enhancement of the far eld scattering for the core-shell architecture with respect to the core case, by a factor of 1.1 for RCP and 3 for LCP. 3D metal/dielectric core-shell nano-helices array as refractive index sensor. a. Normalized CD spectra of the core-shell nano-helices immersed in a glycerol-water solution at different molar concentrations. b, c.
High-magni cation CD spectra for λM (indicated with the orange circle) and ZDP (indicated with purple circle). In particular, 1/|CD| is calculated for ZDP in order to evaluate the FWHM values. d. Relationship between λM (black square symbols) and ZDP (black rhombus symbols), and the refractive index. The standard deviation (below 0.4nm) retrieved for the data points falls within the size of the symbols.

Figure 5
Surface-sensing detection of TDP-43. a. Normalized CD spectra acquired after the only deposition of the antibody layer (spectral reference) and after different concentrations of TDP-43; b. the high-magni cation around the ZDP region allows to evaluate the spectral shift of the crossing point between antibody and the antigen at different molar concentrations; c. Linear t of the spectral shift for λM, λm and ZDP respectively. The size of data points represents the error bars. Figure 6