Understanding of chiral site-dependent enantioselective identification on a plasmon-free semiconductor based SERS substrate

Chiral differentiation is an important topic in diverse fields ranging from pharmaceutics to chiral synthesis. The improvement of sensitivity and the elucidation of the mechanism of chiral recognition are still the two main challenges. Herein, a plasmon-free semiconductive surface-enhanced Raman spectroscopy (SERS) substrate with sensitive chiral recognition ability is proposed for the discrimination of enantiomers. A homochiral environment is constructed by typical π–π stacking between l-tryptophan (l-Trp) and phenyl rings on well-aligned TiO2 nanotubes (TiO2 NTs). Using 3,4-dihydroxyphenylalanine (DOPA) enantiomers as the targets and the chelating interaction of Fe3+–DOPA for the onsite growth of Prussian blue (PB), the enantioselectivity difference between l-DOPA and d-DOPA on the homochiral substrate can be directly monitored from PB signals in the Raman-silent region. By combining the experimental results with molecular dynamic (MD) simulations, it is found that satisfactory enantioselective identification not only requires a homochiral surface but also largely depends on the chiral center environment-differentiated hydrogen-bond formation availability.


Introduction
Chirality, one of the most important characteristics of chiral materials, plays an important role in life sciences. Most biologically active compounds (amino acids, sugars, peptides, proteins, etc.) and modern drugs are chiral. [1][2][3][4] Although enantiomers have similar physicochemical properties, they exhibit completely different physiological effects in terms of biological activity, toxicity, and pharmacological actions. 5 A characteristic example of enantiomers is 3,4-dihydroxyphenylalanine (DOPA). L-DOPA is a common drug used for the treatment of Parkinson's disease. However, the D-enantiomer exhibits neurotoxicity. 6 Therefore, the identication and quantication of enantiomeric forms of chiral molecules are important in chemistry, biology, and pharmaceutical sciences. In the past decade, much effort has been devoted to the study of enantiomeric discrimination, and a variety of chiral sensing techniques have been developed, including vibrational circular dichroism (VCD), 7 highperformance liquid chromatography (HPLC), 8 and electrochemical assay. 9 Although these technologies have helped to make great progress in chiral discrimination, the construction of sensitive and low-cost enantiomer recognition detection platforms and elucidation of the mechanism of enantioselective recognition are still two great challenges in the design and development of highly effective homochiral drugs.
In general, the conventional chiral recognition methods rely on the inherent optical activity of chiral molecules, in which circularly polarized light is a prerequisite for distinguishing two enantiomers. 10,11 The other types of recognition methods need to choose a chiral molecule as the enantioselector. The identication of an enantiomer is achieved based on the difference in the stereostructure of a chiral selector and each of the enantiomers. [12][13][14][15] The recognition mechanism commonly follows the three-point interaction principle, i.e., only one enantiomer can simultaneously form three interactions with the selector while the other cannot. 16,17 Therefore, the study of the binding sites of enantiomeric molecules and selectors can help us to better understand the recognition mechanism of chiral molecules and construct an enantioselective platform for the identication of enantiomers.
Surface-enhanced Raman spectroscopy (SERS) is a nondestructive but sensitive analytical technique that features unique molecular structural ngerprint information. In principle, the SERS effect is mainly attributed to two mechanisms: (i) strong electromagnetic mechanism (EM) induced by surface plasmon resonance (SPR) and (ii) chemical mechanism (CM) induced by dipole-dipole interactions or charge-transfer resonances between the SERS substrates and probe molecules. [18][19][20][21][22][23][24] Traditionally, owing to the excellent SPR effect, noble metals (Au, Ag, and Cu) have been widely used as substrate materials for SERS with a detection limit as low as 10 À10 M. On the other hand, because of a large number of structural defects, modulated surface sites, and layer-number-dependent bandgap, twodimensional (2D) semiconductor nanosheets (TiO 2 , MoS 2 , ZnO, and WO 3 ) have been discovered as alternatives to noblemetal-based SERS substrates. [25][26][27][28][29] Nevertheless, besides the poor stability of ultrathin materials, the reproducibility of preparing a uniform distribution on carrier surfaces is another challenge for the practical applications of these nanosheet-like semiconductors in SERS.
Recently, Weidinger' group has conrmed that in addition to sheet-like semiconductor materials, TiO 2 nanotubes (TiO 2 NTs) also exhibit good SERS performance. 30 The three-dimensional (3D) periodically ordered nanostructures acted as photonic lattices that reect light of certain frequencies, thus enabling a noticeable enhancement in the Raman signal. 31 Owing to their unique biocompatibility, surface groups, chemical stability, and surface morphology, TiO 2 NTs are promising SERS substrates. In this study, utilizing the affinity interaction between phosphonic acid (PA) and the TiO 2 surface, as well as the well-known p-p stacking interaction between the phenyl rings on TiO 2 NTs and L-tryptophan (L-Trp), we developed a simple but effective strategy to construct homochiral SERS substrates for enantioselective recognition. Considering a proof-of-concept application, L/D-DOPA was selected as the analyte of interest, and a background-free Raman reporter Prussian blue (PB) was generated onsite by reacting Fe 3+ -DOPA chelated complexes with [Fe(CN) 6 ] 4À (Fig. 1A). On the basis of the nitrile vibration peak of PB at 2158 cm À1 in the Raman-silent region, 32 the enantioselective recognition between L/D-DOPA and the homochiral SERS substrate can be easily determined from the peak intensity (Fig. 1B). By combining the Raman results with quantum mechanical calculation, the chiral centerenvironment-differentiated hydrogen bond formation availability was simulated to better understand the recognition mechanism. Furthermore, to conrm the role of the chiral center environment, a homochiral environment based on direct amide bonding between the chiral center and SERS substrate was also constructed and demonstrated with a negligible difference in chiral discrimination. This study provides an intelligent and sensitive strategy for the selective recognition of chiral molecules on a noble-metal-free SERS substrate.

Results and discussion
Preparation and characterization of the plasmon-free SERS substrate The application of TiO 2 NTs as a SERS substrate for chiral molecule detection is shown in Fig. 1. TiO 2 NTs were grown by the electrochemical anodization of Ti foil. 33 As shown in the scanning electron microscopy (SEM) images, the well-aligned TiO 2 NTs directly grown on a Ti substrate are composed of numerous nanotubes with good uniformity and a tube diameter of $115 nm ( Fig. 2A). The self-standing NTs are scalable and have good repeatability, thus solving the challenges when using semiconductor nanosheets as SERS substrates. The atomic force microscopy (AFM) image also conrms that the TiO 2 NTs are highly ordered with a uniform distribution of pores over a long range (Fig. 2B). As further characterized by transmission electron microscopy (TEM), the tube wall of TiO 2 NTs is $41 nm, and a lattice spacing of 0.35 nm can be ascribed to the (101) plane of the anatase TiO 2 structure (Fig. 2C). The asformed TiO 2 NTs have a higher crystallinity as indexed in the selected-area electron diffraction (SAED) image.
Generally, in addition to the chemical enhancement effect caused by the formation of charge-transfer complexes between the adsorbate and surface, the enhanced localized electric elds resulting from the specic optical properties of the nanotubular geometry are identied as the dominant factor for Raman signal amplication. 30,31 To achieve a satisfactory Raman signal on TiO 2 NT based SERS substrates, the tube length was optimized using methylene blue (MB) as a Raman probe. As the tube length is mainly determined by the anodization time ( Fig. S1 †), the SERS performance of TiO 2 NTs prepared at different anodization time periods was studied ( Fig. S2 †). Compared with TiO 2 nanoparticle based membranes, the NTs show stronger SERS signals. Owing to the periodic geometric morphology of TiO 2 NTs, multiple laser scattering among the periodic voids improved the light-matter interaction and provided much more opportunities for the occurrence of Raman scattering. Furthermore, the enrichment effect of the nanotubular geometry structure with a higher specic surface area can also be attributed to the increased SERS signal. The SERS signals increase with the length of TiO 2 NTs. However, with further increase of the tube length, the SERS signals start to decrease. This is because a part of the scattered light was trapped in the nanotubes, resulting in scattering loss. 34 In the following study, $8.87 mm TiO 2 NTs were used as the SERS substrate (inset in Fig. 2A). The enhancement factor (EF) of TiO 2 NTs was determined to be 2.7 Â 10 4 based on the MB SERS signal on TiO 2 NTs and a silicon wafer (Fig. S3 †). The signal uniformity was investigated by the acquisition of Raman spectra at thirty random points (Fig. 2D). As plotted in Fig. 2E, the signal intensity of the Raman peak at 1040 and 1628 cm À1 exhibits satisfactory uniformity. Furthermore, the reproducibility was also investigated by comparing the Raman signals on ten TiO 2 NT samples that were prepared using the same process ( Fig. S4 †). As shown in Fig. 2F, the relative standard difference (RSD) for these samples is only 4.35%. These results show good reliability and application potential of the TiO 2 NT based SERS substrate in molecular sensing.

Enantioselective recognition and Raman sensing
To achieve a satisfactory sensitivity and selectivity in enantioselective identication, the homochiral environment and signal amplication are the two crucial issues. Therefore, TiO 2 NT based chiral recognition systems contain two parts: enantioselective identication and signal generation (Fig. 1A). As a proofof-concept for enantiomer recognition, DOPA enantiomers were used as the targets to understand the chiral sensing ability of the as-proposed SERS substrate and elucidate the mechanism in enantioselective identication. Fig. 3A shows the step-by-step modication details for constructing a homochiral environment and L/D-DOPA recognition. The aromatic amino acid L-Trp was used as the chiral host in this study to prepare homochiral substrates. Based on plenty of surface hydroxyl groups (Ti-OH) on the as-prepared TiO 2 NTs, the PA monolayer can be easily self-assembled on a TiO 2 surface via the well-known affinity interaction between PA and Ti-OH. 35 In the next step, L-Trp was anchored onto the tube wall by the p-p stacking interactions of the phenyl rings of L-Trp and PA, thus constructing the homochiral environment. X-ray photoelectron spectra (XPS) showed that the resulting sample (L-Trp/PA/TiO 2 NTs) is composed of Ti, O, P, and N elements without other detectable impurities (Fig. S5 †). The appearance of P 2p signals conrms the successful PA anchoring (Fig. 3B). In addition, the characteristic peaks of N 1s at 399.98 eV (Fig. 3C) can be attributed to the amino group and N atom on the pyridine ring, indicating the existence of L-Trp. Furthermore, the energy-dispersive X-ray spectroscopy (EDS) mapping images also showed the welldispersed P and N elements on the channel wall (Fig. S6 †). In the Raman spectrum ( Fig. S7 †), the characteristic peak at 1000 cm À1 (bCCC, benzene ring) further indicates the successful modication of PA. 36 To identify the homochiral character of the resulting SERS substrate, the circular dichroism (CD) spectra of TiO 2 NTs aer each modication step were obtained (Fig. 3D). An adsorption peak at 294 nm appeared in the CD spectrum of L-Trp/PA/TiO 2 NTs that can be ascribed to L-Trp (Fig. S8 †), indicating the homochirality of the L-Trp/PA/TiO 2 NT based SERS substrate.
Because of the small Raman cross section of chiral molecules, it is difficult to directly use the characteristic Raman peaks of L/D-DOPA for the sensitive recognition difference (Fig. S9 †). To solve this problem, PB, a Raman-silent region probe, was generated onsite via the reaction induced by DOPA recognition. As shown in Fig. 1A, the well-known chelation interaction between Fe 3+ and phenolic hydroxyl groups on DOPA was rst used to introduce Fe 3+ ions. 37,38 The formed Fe 3+ -catechol coordination complexes then reacted with [Fe(CN) 6 ] 4À to grow PB nanocrystals onsite. The C^N vibration in PB resulted in a strong and sharp characteristic peak at 2158 cm À1 (Fig. 1B). Therefore, the enantioselective difference of the homochiral SERS substrate for L/D-DOPA recognition could be achieved by simply comparing the characteristic peak of PB. Considering the self-polymerization of DOPA in a basic solution, the chiral recognition was performed under an acidic aqueous solution (pH 5.5). 39 To obtain a strong PB signal for the sensitive qualication of DOPA, the reaction conditions such as chiral recognition time (Fig. S10 †), Fe 3+ concentration ( Fig. S11 †), and chelation time (Fig. S12 †) were optimized. In this study, the optimal experimental conditions of chiral recognition time, Fe 3+ concentration, and Fe 3+ chelation time used in the subsequent experiments are 60 min, 0.25 mM, and 30 min, respectively. Considering that Fe 3+ ions would be adsorbed on the negatively charged surfaces (Fig. S13 †), we also investigated the PB generation possibility on TiO 2 NTs, PA/TiO 2 NTs, and L-Trp/PA/TiO 2 NTs (Fig. S14 †). Apparently, very weak PB signals were detected on these samples, indicating negligible interference from the electrostatically adsorbed Fe 3+ ions on DOPA sensing. Fig. 3E and F show the Raman spectra of PB for recognizing different concentrations of L-DOPA and D-DOPA, respectively. The Raman intensity increased with increasing L/D-DOPA concentration. More importantly, the as-proposed sensing system showed good enantioselectivity with stronger Raman signals appearing for D-DOPA. The EDS elemental mapping results indicate the formation of PB by the well-dispersed distribution of Fe and N elements (Fig. S15 †). In addition, the X-ray diffraction (XRD) patterns (Fig. S16A †) and UV/visible diffuse reectance absorbance spectra (Fig. S16B †) also provide evidence for the existence of PB nanoparticles. Fig. 3G shows the corresponding relationship between the L/D-DOPA concentration and the intensity of the Raman peak at 2158 cm À1 . A good linear relationship in the L/D-DOPA concentration was observed from 10 À6 to 10 À12 M. The limit of detection (LOD) based on a signal-to-noise ratio of 3 (S/N ¼ 3) was estimated to be as low as 1.7 Â 10 À13 M for L-DOPA and 1.3 Â 10 À13 M for D-DOPA. More importantly, D-DOPA always induced a higher Raman intensity than L-DOPA at the same sample concentration. The detection sensitivity obtained in this study has obvious advantages over the recently reported colorimetric and electrochemical methods (Table S1 †). From the Raman intensity, the enantioselectivity coefficient (dened as: I 2158 (D-DOPA)/I 2158 (L-DOPA) to quantify the chiral distinction) was determined to be 7.6 at 10 À12 M and 1.4 even at a high concentration of 10 À6 M. Such a difference in SERS intensity can be explained by the discrepancy in the interaction between L-Trp and L/D-DOPA enantiomers: more D-DOPA molecules were captured on the homochiral L-Trp/PA/TiO 2 NTs. The difference in the PB generation amount on L-Trp/PA/TiO 2 NTs can be distinguished by SEM characterization (Fig. S17 †). Apparently, the nanotube diameter became smaller aer the chiral recognition of D-DOPA followed by PB formation (Fig. S17E and F, † 68.2 nm) than that of L-DOPA (Fig. S17C and D, † 75.8 nm).

Universality and selectivity investigation
To verify the universality of this SERS-sensing strategy, we attempted to apply p-p stacking for constructing chiral environments using other chiral aromatic molecules. For this purpose, L-phenylalanine (L-Phe) was used to replace L-Trp. The adsorption peak at 221 nm in the CD spectrum of L-Phe/ PA/TiO 2 NTs conrmed the homochiral properties of the resulting SERS substrate (Fig. S18 †). Under the optimized conditions, L-Phe/PA/TiO 2 NTs showed a remarkable enantioselectivity for L/D-DOPA with a distinct SERS signal of PB (Fig. S19 †). The selectivity of the chiral sensing system was also evaluated by the identication of other enantiomers, involving  (Fig. S20 †). The obvious PB signal at 2158 cm À1 was only observed for L/D-DOPA recognition (Fig. S21 †), which can be ascribed to the special chelation interaction between Fe 3+ and phenolic hydroxyl groups on DOPA. These results indicate that the asproposed SERS-based chiral sensing system obtained using chiral aromatic molecules to construct homochiral environments has satisfactory enantioselectivity, versatility, and reliability.

Mechanism of enantioselective identication
To elucidate the underlying mechanism, we conducted molecular dynamics (MD) simulations for the recognition of DOPA enantiomers on the chiral site of L-Trp. It was suggested that the hydrogen bond formed by-NH 3 + /-COO À interaction would be the strongest interaction in this chiral recognition system. 40 The distance between -NH 3 + and -COO À was monitored to determine the chiral recognition difference of the enantiomers on the homochiral L-Trp/PA/TiO 2 NTs, which serves as an indicator for the stability of H bonds. We calculated the distances between (i) the -NH 3 + group on the chiral site for L-Trp (denoted as L-Trp (N)) and the -COO À group of DOPA (denoted as DOPA (C)); (ii) the -COO À groups of L-Trp (denoted as L-Trp (C)) and -NH 3 + of DOPA (denoted as DOPA (N)). For L-DOPA, as shown in Fig. 4A, the distance between L-Trp (N) and DOPA (C) is larger than that between L-Trp (C) and DOPA (N) (within 4Å) during the simulation, suggesting that the -NH 3 + groups of L-DOPA are available for binding onto the homochiral substrate via the formation of H bonds compared with the -COO À groups of L-DOPA. Fig. 4B shows that the distance between L-Trp (N) and D-DOPA (C) as well as the distance between L-Trp (C) and D-DOPA (N) are both in a short range (within 4Å). These results indicate that two types of H bonds are favorable for D-DOPA recognition. Because of the energetically favorable stereoselectivity, the -NH 3 + and -COO À groups in both molecules are paired with each other. As shown in Fig. 4C, only one H bond is expected to be formed between L-Trp and L-DOPA. For comparison, two H bonds would be formed between L-Trp and D-DOPA (Fig. 4D). In this case, the two H bonds largely strengthen the interactions between L-Trp and D-DOPA and thus more D-DOPA is recognized on the homochiral L-Trp/PA/TiO 2 NTs. As a result, a larger number of Fe 3+ ions were anchored on the substrate in the subsequent step, and through the PB generation step, the enantioselective identication can be directly determined from the Raman signal of PB. Stereoselectivity is also crucial for chiral binding sites in the enantiomer recognition event. To demonstrate the key role of stereoselectivity, L-Trp was linked with TiO 2 NTs via an amide bond by utilizing the -COOH group on the chiral carbon of L-Trp and -NH 2 groups on the TiO 2 surface. As shown in Fig. 5A, the -NH 2 group was rst modied on TiO 2 NTs by the condensation reaction between P-OH groups in o-phosphorylethanolamine (O-phos) and Ti-OH groups. 35 The graing of the O-phos monolayer was conrmed by XPS analysis from the appearance of P 2p signals (Fig. 5B). Then, L-Trp was introduced by forming an amide bond through a classic N-(3-dimethylaminopropyl)-Nethylcarbodiimide hydrochloride (EDC)/N-hydroxysulfosuccinimide (NHS) coupling reaction, 41 in which the -COOH group on a chiral site is used for forming an amide bond with -NH 2 groups on the tube wall. The successful covering of L-Trp on O-phos/TiO 2 NTs leads to the decrease of P 2p signals (Fig. 5B) as well as the increase of N 1s signals (Fig. 5C). The homochiral feature was conrmed from CD spectra with the appearance of an adsorption peak at 294 nm (Fig. 5D). Notably, the PB signals obtained from L-DOPA and D-DOPA recognition on the L-Trp/O-phos/TiO 2 NT substrate show almost the same intensities (Fig. 5E-G), indicating that the homochiral substrate has a similar recognition ability for the targeted enantiomers. Such a phenomenon can be attributed to the steric hindrance from the indole group on L-Trp. Because of the steric hindrance from the indole group, DOPA recognition is originated from the H bonds formed between the -NHon pyrrole and L/D-DOPA. As it is difficult to reach the chiral site of L-Trp, DOPA enantiomers exhibit the same recognition ability on L-Trp/O-phos/TiO 2 NTs. These results indicate that even though the substrate is homochiral, stereoselectivity, especially the steric hindrance at a chiral site, is an important factor that should be considered for designing an enantioselective material.

Preparation of TiO 2 NTs and TiO 2 nanoparticle based membrane
The TiO 2 NTs were grown on Ti substrates (15 mm Â 15 mm Â 0.1 mm) by electrochemical anodization. The Ti substrates were rst ultrasonically cleaned in isopropanol, ethanol, and deionized water in sequence and dried under a N 2 gas ow. Anodization was carried out in an ethylene glycol/lactic acid/water/ NH 4 F electrolyte containing 0.1 M NH 4 F at 120 V for 4 min. Ti substrates and platinum foil served as the working electrode and counter electrode, respectively. A compact TiO 2 nanoparticle based membrane was grown by the anodization of the Ti substrate in 1 M H 2 SO 4 at 20 V for 20 min. The prepared samples were annealed at 450 C for 2 h in air at a heating rate of 3 C min À1 .

Synthesis of L-Trp/PA/TiO 2 NTs
The as-formed TiO 2 NTs were rst functionalized with PA by immersing the sample in a 10 mM PA ethanol solution at 4 C for 12 h, followed by rinsing with ethanol three times. Then, chiral recognition molecules (L-Trp) were introduced onto the prepared PA/TiO 2 NTs through p-p stacking interaction. The substrates were soaked in a 1 mM L-Trp aqueous solution at room temperature for 12 h, producing L-Trp/PA/TiO 2 NTs.

Detection of L/D-DOPA
The as-prepared L-Trp/PA/TiO 2 NTs were rst immersed in an aqueous solution of L/D-DOPA (pH 5.5) at room temperature for 60 min and then lightly cleaned with deionized water three times and dried under a N 2 stream. These samples were then incubated in FeCl 3 aqueous solution for 30 min to anchor Fe 3+ ions, and then gently washed with deionized water to remove the electrostatically adsorbed Fe 3+ ions. Aer that, PB was generated by soaking the samples in a 0.25 mM K 4 [Fe(CN) 6 ] aqueous solution at room temperature for 30 min. The resultant samples were washed with deionized water three times and dried with N 2 gas.

Raman measurement
MB ethanol solutions were used as the analyte. Then, 200 mL of MB solution was added dropwise and spread on a substrate (50 mm Â 50 mm scale) and dried in the dark. Raman spectra were recorded using a 638 nm laser (10% power) equipped with a 50Â long-distance objective lens and a 1 mm spot size. The data acquisition time was kept as 10 s; the confocal hole size was 500 mm; the slit aperture size was 100 mm. The spectrometer was calibrated using the Raman spectra of silicon wafer at 520.7 cm À1 . Raman spectra were collected from ve different places, and then the average signal intensity was determined.

EF measurement
As a SERS sample, 4 mL of 2 Â 10 À5 M MB solution was added dropwise and spread on the TiO 2 NT substrate (50 mm Â 50 mm scale) and dried at room temperature. As a non-SERS sample, 20 mL of 2 Â 10 À3 M MB ethanol solution was added dropwise and spread on a bare silicon wafer (50 mm Â 50 mm scale) and dried at room temperature. The Raman spectra were recorded using a 638 nm wavelength laser (10% power) with a 50Â objective, and the data acquisition time was 10 s.

MD simulations
The initial structures of L-Trp and DOPA were constructed using GaussView 6 (ref. 42) and optimized at the B3LYP/6-31G(d) level of theory using the Gaussian 16 package. 43 Possible binding modes between L-Trp and L/D-DOPA were predicted using AutoDock Vina 1.1.2. 44 For each dimer, nine docked conformations were selected for the simulations. Trajectories containing the following two conditions were selected for subsequent analysis: (1) the initial structure contained at least one NH 3 + -COO À interaction, and (2) the dimer did not dissociate in 1 ns simulations. A general AMBER force eld (GAFF) 45 was applied to the molecules, and the AM1 bcc charges were calculated with the optimized structures of L-Trp and L/D-DOPA. L-Trp and L/D-DOPA were dissolved in a periodic cubic TIP3P 46 water box with the distance between the solute and the box boundary no less than 22.5Å. All the input les containing parameters and congurations were generated with the LEaP module in AmberTools 19. 47 The whole systems were optimized for 1000 steps and then heated up to 298.15 K in 100 ps using Langevin dynamics 48 with a collision frequency of 1 ps À1 . The nonbonded cutoff was set to 8Å. The pressure was regulated at 1 atm using Berendsen's scheme. Congurations were collected at 298.15 K with a 1 ps interval from each 1 ns simulation. All the simulations were conducted using AMBER 18. 47

Conclusions
In summary, we constructed a chiral sensing platform with a high enantioselectivity on a plasmon-free TiO 2 NT based SERS substrate. The homochiral environment was constructed by utilizing typical p-p stacking between L-Trp and PA-modied TiO 2 NTs. The formed L-Trp/PA/TiO 2 NTs exhibited very good stereoselectivity for D-DOPA over L-DOPA. Beneting from the chelation binding between DOPA and Fe 3+ ions, a simple signal output strategy was thus achieved by forming a Raman reporter PB onsite via the reaction of Fe 3+ -DOPA and [Fe(CN) 6 ] 4À , providing a sensitive and selective approach for L/D-DOPA recognition and quantication. The mechanism for identication of the difference on the homochiral substrate was well studied by conducting MD simulations and experiments; the results indicate that both enantioselectivity and stereoselectivity are crucial factors for chiral sensing.

Data availability
The data supporting the ndings of this study are available within the article and in the ESI. †

Conflicts of interest
There are no conicts to declare.