The target molecular structures of Irn (n = 1–5) are shown in Scheme 1. Despite having a basic molecular formula of [(C^∧N)₂Ir(N^∧N)]Cl, these structures were divided into three groups according to their ligands. The first group (Ir1) was introduced as the reference group without any -CHO group. Ir2 and Ir3 with a -CHO group in their C^∧N ligands comprised the second group. The third group consisted of Ir4 and Ir5, with one -CHO group each in their C^∧N ligands and two -NH₂ groups each in their N^∧N ligands.

Comparisons were performed between these three groups to determine clues for evaluating their Hcy sensing performances. The -CHO group has been reported to be sensitive and is known to form a six-membered ring with Hcy; the -NH₂ group is able to facilitate this cyclization reaction (Rusin et al., 2004). The single-crystal structures of Ir1, Ir2, Ir3, Ir4, and Ir5 are shown in Figure 1. Two C^∧N and one N^∧N ligands form a typical octahedral coordination field for the central Ir(III) ion, which is consistent with literature (Zhang et al., 2009). There are no observed signs of coordination between the Ir(III) and Cl⁻ ions, indicating that the Cl⁻ ions act only as counterions.

We first compared the absorption and emission spectra of the Irn compounds (1 μM) in DMF (Figure 2). Their key spectroscopic parameters are listed in Table 1. Owing to their similar molecular structures, the Irn compounds show similar absorption bands, and the absorption spectra are composed of intense absorption bands ranging from 250 nm to 335 nm and moderate ones ranging from 335 nm to 515 nm (Figure 2A). The intense absorption bands have been reported as ligand π→π* transitions, while the moderate absorption bands belong to metal-to-ligand charge transfer (MLCT) transitions (Liu et al., 2017). The introduction of the electron-withdrawing group -CHO tends to move the absorption edge (λ_edg) of Ir2/Ir3 (515 nm) toward longer wavelengths (red shift) than that of Ir1 (490 nm). Similarly, a larger wavelength redshift of 90 nm was observed for Ir4/Ir5 (592 nm) by the introduction of both -CHO and electron-donating -NH₂ groups.

Owing to the aggregation-induced quenching effect (ϕ values <0.04%) in the solid state (Zhang et al., 2009), all Irn compounds show little or no emissions (Table 1). However, after being dispersed in solution, their quantum yields (>7%) are enhanced by two orders of magnitude. Each emission spectrum is also composed of a major band and a shoulder band, which is a characteristic of MLCT-based Ir(III) emission (Liu et al., 2017). With the introduction of -CHO or -NH₂ and -CHO groups, the Irn (n = 2/3/4/5) emissions show redshifts compared to that of Ir1 (559 nm) (Table 1). It should be noted that the full-width at half maximum (FWHM) value of Ir5 (30 nm) is less than those of the other Irn compounds (ranging from 44 to 57 nm), indicating higher color purity (Fan et al., 2023). We next measured the decay lifetimes of the Irn (n = 1–5) compounds (Table 1), all of whom presented biexponential decay modes. The long-lived component of the decay lifetime is attributed to the decay process of the MLCT excited state, and the short-lived component is attributed to that of the ligand π→π* excited state, consistent with the findings in literature (Zhang et al., 2009). The radiative transition rate constant of the excited state (k_r) and non-radiative transition rate constant of the excited state (k_nr) of Irn were calculated using Eqs. (1, 2) and listed in Table 1. Compared with Ir1, the decrease in the electron density of Ir2/3 caused by introduction of the electron-withdrawing group -CHO significantly improves the value of k_nr and reduces the value of k_r, consistent with the short lifetime (2.73–3.06 s) and low ϕ (7.4%–7.5%) (Xia et al., 2023). Similarly, the electron-donating -NH₂ group further improves the emissive probability since it increases the electron density of the excited state. ϕ = k r k r + k n r (1) τ = 1 k r + k n r (2)

The absorption and emission spectra of Irn were recorded in the presence of Hcy (5 equiv.) (Figures 2B, D). The absorption spectrum of Ir1 remains unchanged, with no detectable spectral shift or new bands after adding Hcy, indicating that the molecular structure of Ir1 is preserved. As for Ir2 and Ir3 with the -CHO group in their ligands, slight emission redshifts (∼10 nm) accompanied by emission quenching (from 7.5%–6.4% and 7.4%–6.1%, respectively) are observed after adding Hcy. In addition, an additional shoulder band peaking at 554 nm was observed. For Ir4 and Ir5, the absorption spectra blue-shifted from ∼592 nm to ∼498 nm in the presence of Hcy. Correspondingly, their emission peaks also blueshifted from 602 to 503 nm and 603 to 509 nm, with the corresponding emission yields increasing slightly from 11.7%–13.3% and 11.2%–13.1%, respectively. The spectral changes of Ir2/Ir3/Ir4/Ir5 suggest interactions between their -CHO groups and Hcy, which make them effective sensing probes for Hcy.

Among the Irn compounds, Ir4 and Ir5 show the most sensitive changes in emission intensities and wavelength shifts upon Hcy addition (∼100 nm) (Table 1). As mentioned above, the absorption spectrum (Figure 2A), emission spectrum (Figure 2B), and decay process of Ir4/5 compounds are consistent with the MLCT mechanism. To better understand the photophysical properties, the frontier molecular orbitals (FMOs) of the Ir4/5 complexes were calculated using time-dependent density functional theory (TD-DFT) and listed in Table 2. Orbital analyses revealed that for the Ir4/5 complexes, the highest occupied molecular orbitals (HOMOs) are composed of the iridium center and phenyl parts of the cyclometalated ligands, whereas the lowest unoccupied molecular orbitals (LUMOs) are located mainly on the bipyridine derivative ligands. The HOMO→LUMO transitions could contribute to the [dπ(Ir) → π_N^∧N ^*]³MLCT, with some mixing of the ³[π_C^∧N → π_N^∧N ^*] ligand-to-ligand charge transfer (³LLCT). For Ir4/5-Hcy, the HOMOs are located on the iridium center and one newly generated six-membered thiazolidine ring, while the LUMOs reside on the bipyridine derivative ligands, indicating that their emissions are derived mainly from [π_C^∧N → π_N^∧N ^*]³LLCT and [dπ(Ir) → π_N^∧N ^*]³MLCT. Therefore, the electronic transitions of these Ir(III) complexes have an MLCT character, and the formed thiazolidine ring has better electron-donating ability than the aldehyde group, which could increase the MLCT transition energy, leading to a blue shift.

To investigate the sensing mechanism, a nuclear magnetic resonance (NMR) titration was carried out on Ir4 (the results of Ir5 are similar to those of Ir4). As the Hcy concentration increases from 0 to 2 equiv., the chemical shift signal from the H of the -CHO group (9.94 ppm) decreases significantly (Supplementary Figure S1). Meanwhile, chemical shifts are seen for the -NH- (5.24 ppm) and -N-CH-S- (5.11 ppm) groups that become stronger (Supplementary Figure S1). These results indicate that the cyclization between the -CHO group and Hcy is consistent with previous reports (Gao et al., 2017; Wang et al., 2018). The reactions of Ir4/5 on Hcy were further supported by positive-ion electrospray ionization high-resolution mass spectra (ESI-MS) (Zhang et al., 2019) (Supplementary Figure S2). There are strong signals of 978.2 and 1001.2 for the complex [Ir4+2Hcy-PF₆]⁺ and complex [Ir5+2Hcy-PF₆]⁺, respectively. Relatively weak signals corresponding to [Ir4+Hcy-PF₆]⁺ and [Ir5+Hcy-PF₆]⁺ are also observed at 861.2 and 884.2, respectively. The biform and monoform adducts observed in the ESI-MS indicate that the reactions are stepwise processes.

A spectroscopic analysis of Ir4/5@EBCOF with increasing Hcy concentration was performed to evaluate the Hcy sensing performance. As shown in Figure 4A, in the absence of Hcy, characteristic Ir(III) emissions are observed for Ir4/5@EBCOF: 604 nm (major peak) and 632 nm (shoulder peak) for Ir4@EBCOF; 604 nm (major peak) and 640 nm (shoulder peak) for Ir5@EBCOF; these are accompanied by ∼2 nm wavelength redshifts compared to the bare Ir4/5 molecules (Figure 2C, peaking at 602 nm for Ir4 and 603 nm for Ir5). The reasons for this may be explained by the stabilization or solvation effects (Yang et al., 2011).

As the Hcy concentration increases from 0 to 110 μM, the emission intensities of Ir4@EBCOF at 504 nm and Ir5@EBCOF at 509 nm continue to increase until saturation. Meanwhile, the emission intensity of Ir4/5@EBCOF at 604 nm decreases correspondingly, with the residual emission still having an intensity as high as 1000 counts (Figures 4B–D). This may be attributed to the contribution from the bare EBCOF (627 nm, Figure 4A). To avoid this emission interference, green emission (504 nm for Ir4@EBCOF and 509 nm for Ir5@EBCOF) was selected for evaluating the sensing sensitivity based on Stern–Volmer analysis (Lei et al., 2006). The emission of Ir5 (509 nm) was also analyzed to explore the effect of the EBCOF on Hcy sensing performance, as given by Eq. (3): I I 0 = C + K s v H c y (3) where I and I₀ indicate steady emission intensities at 504 nm or 509 nm with and without the addition of Hcy, respectively; C is a constant, K_sv is the Stern–Volmer coefficient also known as the sensitivity, and [Hcy] is the concentration of Hcy. As shown in the fitting plots of Figure 4E, linear calibration curves are observed for both Ir4/5@EBCOF and Ir5 samples, with their working range, fitting parameter, LOD (3σ/N), and limit of quantification (LOQ, 10σ/N) (Magnusson and Örnemark, 2014) values summarized in Table 3. The LOD and sensitivity values are determined as 0.44 μM and 0.5641 μM^-1 for Ir4@EBCOF, and 1.18 μM and 0.2127 μM^-1 for Ir5, respectively. In the comparison, Ir5@EBCOF shows the best sensing performance, with LOD and sensitivity values of 0.23 μM and 1.0106 μM^-1, respectively. The much better detection performance of Ir4/5@EBCOF compared to that of Ir5 implies good dispersion of EBCOF.

It should be noted that the opposing changes between the green and red emissions of Ir4/5@EBCOF can be viewed based on color. In the absence of Hcy, the CIE color coordinates of Ir4@EBCOF (0.64, 0.35) and Ir5@EBCOF (0.64, 0.35) fall within the red region. For a Hcy concentration of 110 μM, the CIE color coordinates of Ir4@EBCOF (0.32, 0.60) and Ir5@EBCOF (0.33, 0.61) move toward the green region (Figure 4F). Based on these characteristics, Ir4/5@EBCOF are also expected to be useful as visual sensors for Hcy.

We next investigated the dynamic sensing performances of the probes by monitoring their emission intensity for the optimal Ir4/5@EBCOF upon gradual addition of Hcy (Figure 5A). The emission intensity was enhanced rapidly in the first 50 s and then remained smooth until saturation. The response time, defined as the time to reach 95% of the maximum emission intensity, is about 56 s for Ir4@EBCOF and 88 s for Ir5@EBCOF. During this interval, the emission color changes from red to green (Figure 5B).

A practical sensing platform always entails an inevitable situation where the analyte is dispersed in a complex environment full of competing species and interferents. According to literature (Xiong et al., 2010; Liu et al., 2012; Wang et al., 2021), most chemical sensors struggle to distinguish between Hcy and Cys molecules owing to their similar molecular structures, which means that Cys is an important interferent in Hcy sensing. However, when reacting with Cys, Ir4/5@EBCOF shows different dynamic sensing performance than Hcy. After adding Cys, the emission intensity of Ir4/5@EBCOF remained nearly constant over the initial 600 s, and the reaction time is over 25 min, which is much longer than that of Hcy (Supplementary Figure S4). When the sample was treated in an ultrasonic bath for 10 min, Ir4/5@EBCOF completed their sensing process for Hcy and reached the maximum emission intensity, and the presence of excess Cys did not affect the detection of Hcy (Supplementary Figure S5), indicating good resistance to interference at the appropriate reaction time. With a sonication time of 10 min, the sensing selectivity of Ir4/5@EBCOF was evaluated in the presence of Na⁺, K⁺, Ca²⁺, Fe²⁺, Fe³⁺, Hg²⁺, CO₃ ^2-, PO₄ ^3-, Cl⁻, glucose, urea, Cys, GSH, aspartic acid (Asp), methionine (Met), glycine (Gly), and thioglycolic acid (Tga) (Figures 5C–F). As seen in the figure, although the emissions of Ir4/5@EBCOF at 604 nm exist in the presence of all the interferents, the emissions at 504 nm/509 nm appear only in the presence of Hcy. These results clearly indicate the good selectivity of the proposed probe to Hcy.

Based on the encouraging sensing performance, we carried out practical experiments on human serum samples to detect Hcy. For the control experiments, PBS solutions with different Hcy concentrations were adopted, and the recovery values were calculated by recording the emission spectra of Ir4/5@EBCOF in different solutions. As shown in Table 3, satisfactory recovery (96.06%–105.83%) values indicate the good sensing performances of the probes. When Hcy concentrations are lower than 40 μM, positive sensing errors are observed. As the Hcy concentration increases further, negative errors are observed. It is assumed that for higher Hcy concentrations, some of the Hcy molecules are adsorbed and trapped by the EBCOF matrix, leading to the negative sensing errors.