Pseudoaromaticity-driven, transition metal detection by squaraine-derived enol phosphonium ylide chemodosimeters

Abstract

The addition of PⁿBu₃ to o-substituted dianiline squaraine dyes leads to bench stable ylides. Exposure to a metal analyte in solution, results in P^III abstraction and rapid disruption of the ylide conjugation to promote reversion back to the squaraine dye giving an immediate turn-on colorimetric response. The stoichiometric sensitivity and accessibility of these chemodosimeters constitute effective organic dyes for trace transition metal detection.

---

Dianiline squaraines are colorful organic dyes utilized in non-invasive bio-imaging,¹ photovoltaics,² and ion-sensing.³ They are characterized by two para-substituted electron donating anilines in a 1,3-orientation across an electrophilic, cyclobutadione core, resulting in a donor–acceptor–donor resonance stabilization effect.³ However, the electrophilicity of the squaraine core renders it susceptible to nucleophiles present in biological systems, and is a frequent obstacle in fluorescence imaging.⁴ This has led to a number of elegant macromolecular protection strategies to minimize the addition of exogenous nucleophiles and quenching of the squaraine dye.⁵ While these approaches have proven effective at protecting the squaraine core and rendering the composite complex stable to biologically relevant environments, few studies have sought to exploit its inherent electrophilicity in the templated design of new chromophoric sensing agents.⁶

We recently sought to exploit the inherent electrophilicity of squaraine dyes in the design of new chemodosimeters wherein the addition of phosphine leads to stable, zwitterionic complex that serves as a positive response “turn on” sensor for the detection of specific trace analytes in solution (Fig. 1a).^6a The phosphine-squaraine adducts 2 were effective as colorimetric sensors for transition metals such as Rh^I, Pd^II, Ir^I, and Au^I.^6a However, one limitation of our initial design was that the adducts underwent only 1% reversion to the parent dye 1 despite an excess of the analyte present, thereby limiting the sensitivity of this class of indicators. Hence, we sought an alternative squaraine-based design that would ultimately result in a more dramatic positive response to the transition metal analyte. Herein, we introduce the successful identification of a rather unusual, highly conjugated phosphonium-based platform that is exceptionally stable and chemoselective for the detection of trace transition metals with a substantially more pronounced propensity to undergo complete reversion to the parent squaraine dye.

Fig. 1 Addition of phosphorus (III) to dianiline squaraines. (a) Colorimetric sensors for trace metals. (b) Reversion to squaraine dyes via Brønsted acid. (c) Reversion to squaraine dyes through metal-induced process.

Recently, we reported the dearomative skeletal reorganization of phosphine-squaraine adduct 2a, derived from the addition of ⁿBu₃P to the corresponding dianiline squaraine bearing a hydroxy group vicinal to the four-membered ring, to benzofuranone 3a (Fig. 1b).⁷ Employing squaraines with an N-acyl group in place of the o-OH provided oxindole 3b wherein the enhanced acidity of the N-acyl proton was critical to accessing the rearranged product.^7,8 A notable feature of these frameworks is the exceptionally stable enol and phosphorus ylide motifs afforded by the high degree of extended conjugation. Over the course of our study, we observed that while unreactive to various nucleophilic and electrophilic reagents, adducts 3 underwent reversion to the parent squaraine 4 in the presence of a Brønsted acid, presumably driven by the pseudo-rearomatization of the squaraine core. This observation led us to speculate on the propensity of phosphine adducts 3 to undergo complete and rapid reversion to the parent squaraine in the presence of an analyte capable of extracting PⁿBu₃ leading to more sensitive and efficient chemodosimeters for latent metal detection than our previous design (Fig. 1c).⁷ Herein, we report the development of a thermally stable, stoichiometric colorimetric chemodosimeter for trace metal recognition.

We began our evaluation of ylides 3 as colorimetric chemodosimeters by assessing the extent of squaraine 4 formation in the presence of a transition metal complex.^6a Construction of ylides 3a and 3b was accomplished following our previous report by treatment of o-substituted squaraines 4a and 4b, independently, with PⁿBu₃ to rapidly (ca. ∼5 min) provide the desired heterocyclic frameworks in high yield (Scheme 1).⁷ Our initial set of experiments toward evaluating the efficacy of ylides 3a and 3b as chemodosimeters consisted of assessing the extent to which squaraine dyes 4a and 4b underwent regeneration upon treatment with [Rh(COD)Cl]₂, [Ir(COD)Cl]₂, Pd(OAc)₂, Au(PPh)₃Cl and Rh(PPh₃)₃Cl. Using ¹H NMR to track the conversion of 3a to 4a, exposure to 1 equiv. of [Rh(COD)Cl]₂, [Ir(COD)Cl]₂, and Pd(OAc)₂, independently, resulted in complete conversion to dye 4a (Fig. 2). While the addition of Au(PPh₃)Cl and Rh(PPh₃)₃Cl to 3a resulted in significant dye regeneration, and were substantially more responsive than our original phosphine squaraine adducts 2, the 9 : 1 and 5 : 1 ratios of 4a:3a, respectively, indicated a metal selective response. Similarly, the addition of [Rh(COD)Cl]₂, [Ir(COD)Cl]₂, and Pd(OAc)₂ to N-acyl oxindole 3b resulted in complete conversion to squaraine 4b, while exposure to Au(PPh₃)Cl and Rh(PPh₃)₃Cl to 3b led to less dye formation, providing 4b : 3b in 4 : 1 and 2 : 1 ratios, respectively (Fig. 3).

Scheme 1 Chemodosimeter 3 construction.

Fig. 2 Partial ¹H NMR spectra (500 MHz, CDCl₃) that illustrates conversion of 3a (0.02 M) to 4a upon addition of ML_n (0.02 M). The ratio of 4a/3a was determined by ¹H NMR.

Fig. 3 Partial ¹H NMR spectra (500 MHz, CDCl₃) that illustrates conversion of 3b (0.02 M) to 4b upon addition of ML_n (0.02 M). The ratio of 4b/3b was determined by ¹H NMR.

The disparate responses observed with [Rh(COD)Cl]₂, [Ir(COD)Cl]₂, and Pd(OAc)₂ relative to Au(PPh₃)Cl and Rh(PPh₃)₃Cl for ylides 3a/b are likely due to the coordination capabilities of these complexes. For example, upon abstraction of phosphine from ylide 3, the relatively electron rich Wilkinson's catalyst adopts an unfavorable square pyramidal geometry and the resulting phosphine-ligated Au^I complex a trigonal planar geometry.^6a In contrast, the olefinic ligated Rh^I and Ir^I complexes, and electrophilic Pd^II, may more readily accommodate additional phosphine.^6a Given that no reaction was observed when PPh₃ was added to 4a/b, we suspect that those instances of diminished dye regeneration is unlikely due to reformation of 3a/b by the reincorporation of liberated PPh₃.

Due to frequent purification challenges, detection and removal of residual transition metals can pose a hinderance, especially at low concentrations.^9–11 Employing UV-Vis spectroscopy to monitor the regeneration of dyes 4, we tested the capability of adducts 3 to act as colorimetric chemodosimeters for metals at lower concentrations (ca. ∼5 μM). The addition of 0.1 equiv. (i.e., 0.269 ppm) of [Rh(COD)Cl]₂ to adduct 3a resulted in formation of squaraine 4a at 2.01 μM (Fig. 4a). Increasing the concentration of [Rh(COD)Cl]₂ led to an increase in absorbance at 655 nm corresponding to formation of 4a. However, the isotherm of [Rh(COD)Cl]₂ at 1 equiv. (i.e., 2.69 ppm) plateaus with 5.41 μM free squaraine 4a, indicating saturation of the metal center (Fig. 4b). To determine whether commonly used transition metals would result in similar UV-Vis spectroscopic responses, we examined the regeneration of 4a from 3a and 1 equiv. of Au(PPh₃)Cl, HgCl₂, PtCl₂, AgNO₃, or Pd(OAc)₂. We found that Au(PPh₃)Cl plateaued to give the highest concentration of 4a at 4.45 μM, and AgNO₃ and Pd(OAc)₂ the least in 3.24 μM and 2.97 μM, respectively.

Fig. 4 (a) Appearance of 4a absorption maxima band (λ_max = 655 nm) upon titration of 3a (5.46 μM in DMSO; λ_max = 324 nm) with increasing amount of [Rh(COD)Cl]₂. (b) Titration plots showing regeneration of 4a when separate solutions of 3a (ca. 5 μM in DMSO) were titrated with PtCl₂, AgNO₃, Pd(OAc)₂, HgCl₂, Au(PPh₃)Cl and [Rh(COD)Cl]₂.

An assessment of oxindole 3b at 5.03 μM with [Rh(COD)Cl]₂ (0.1 equiv.) resulted in formation of squaraine 4b at 1.08 μM, and increasing the concentration of [Rh(COD)Cl]₂ led to an increase in absorbance at 685 nm (Fig. 5a). A systematic titration provided an isotherm that plateaus upon the addition of 1 equiv. of metal and generation of squaraine 4b at 4.40 μM (Fig. 5b). The titration of 3b with Au(PPh₃)Cl, HgCl₂, PtCl₂, AgNO₃, and Pd(OAc)₂ also led to regeneration of 4b, but unlike ylide 3a, the amount of dye regenerated proved less distinguishable across the analytes evaluated. In contrast, the addition of Cu(OAc)₂, Pb(NO₃)₂ and Rh(PPh₃)₃Cl to squaraine–phosphine ylide complexes 3 led to less 4.¹² The formation of squaraines 4a/b upon solvation, even in the absence of a metal analyte, suggests a dynamic equilibrium between 3 and 4.

Fig. 5 (a) Appearance of 4b absorption maxima band (λ_max = 685 nm) upon titration of 3b (5.03 μM in DMSO; λ_max = 316 nm) with increasing amount of [Rh(COD)Cl]₂. (b) Titration plots showing regeneration of 4b when separate solutions of 3b (ca. 5 μM in DMSO) were titrated with PtCl₂, AgNO₃, Pd(OAc)₂, HgCl₂, Au(PPh₃)Cl and [Rh(COD)Cl]₂.

In general, the electronic and ligation structure of the metal appears to influence the extent of squaraine 4 regeneration. Electron-deficient, unsaturated metals, and those complexes bearing labile olefinic ligands, resulted in near quantitative reversion to squaraine dyes 4. In contrast, more electron-rich metals, such as Rh(PPh₃)₃Cl, resulted in lower concentrations of squaraine dye 3.^6a Our results indicate that the metal and ortho-heteroatom that distinguishes between chemodosimeters 4a/b influences the extent of dye regeneration. For example, while 60% conversion of benzofuranone 3a to squaraine dye 4a upon addition of Pd(OAc)₂ (1 equiv.) was observed, oxindole 3b gave 89% of 4b. It is conceivable that a weaker affinity of oxindole 4b for PⁿBu₃ is a driving force for these disparate responses.

Phosphonium ylides 3 exhibited a greater sensitivity to transition metal analytes than our previous dosimeter 2. For example, while 3b underwent 89% reversion to squaraine 4b in the presence of 1 equiv. of Pd(OAc)₂, less than 0.4% of 1a was obtained from squaraine–phosphine adduct 2a (Fig. 6). Similarly, the addition of Au(PPh₃)Cl (1 equiv.) resulted in 78% conversion to 4b from 3b, while 1a was obtained in only 0.6% from 2a.^6a The ability of 3 to undergo near complete reversion may be attributed to key architectural distinctions between these two chemodosimeter designs. Dosimeter 3 is a 1,2-dipolar compound consisting of a conjugated phosphorus ylide that exhibits less double bond character than a standard ylide.⁷ In contrast, the squaraine–phosphine adduct 2 is 1,3-dipolar with a kite-quadrilateral architecture that is not in conjugation with phosphorus.^6a Upon phosphine dissociation, recapitulation of the original squaraine dye is presumably driven by rearomatization and the instability of a transient, intermediate carbene.^7,13 A primary distinction between squaraines 1 and 4 is the presence of an ortho-aryl heteroatom bearing a relatively acidic proton.⁷ A penultimate proton transfer step is critical to driving the equilibrium between the “turned off” ylide 3 and “turned on” dye 4, which is absent in the corresponding conversion of squaraine 1 and adduct 2.^6a Attractive features of this class of ylide sensors include their robust bench stability, enabling later use,⁷ ability to detect metals at low concentrations, and detection of transition metal complexes not sufficiently thiophilic. While sensitivity is comparable to analogous squaraine-thiol based dosimeters, selectivity in detecting metal analytes is reduced.^6c,d

Fig. 6 Performance of chemodosimeters 3b (ca. 5 μM in DMSO) and 2 (1.5 mM in DMSO), indicating conversion to 4b and 1, respectively, with increasing Pd(OAc)₂ and Au(PPh₃)Cl.

In conclusion, a new class of squaraine-based colorimetric dosimeters, obtained from the addition of PⁿBu₃ to o-OH or o-NHAc substituted dianiline squaraines, are introduced. These scaffolds are stable at rt, and exhibit enhanced sensitivity to transition metal analytes under mild conditions in comparison to our previous squaraine-derived chemodosimeters. These sensing agents underwent significant retroversion to the parent squaraines even in the presence of metal complexes exhibiting a low affinity for P^III at low concentrations. The stability and sensitivity of these colorimetric dosimeters suggest potential uses in detecting metals in various media. The development and examination of these dosimeters for analyte detection in different environmental conditions is underway and will be reported in due course.

Conflicts of interest

The authors have no conflicts to declare.

References

1. (a) S. Sreejith, et al. , ACS Nano, 2015, 9(6), 5695–5704; (b) K. M. Harmatys, et al. , Mol. Pharm., 2013, 10(11), 4263–4271; (c) E. L. Cole, et al. , Org. Biomol. Chem., 2012, 10(30), 5769–5773; (d) A. G. White, et al. , Bioconjug. Chem., 2010, 21(7), 1297–1304; (e) R. R. Avirah, et al. , Org. Biomol. Chem., 2012, 10(5), 911–920.
2. (a) J. H. Yum, et al. , J. Am. Chem. Soc., 2007, 129(34), 10320–10321; (b) T. Maeda, et al. , J. Phys. Chem. C, 2018, 122(38), 21745–21754; (c) G. Chen, et al. , J. Mater. Chem. A, 2015, 3, 14517–14534; (d) G. Chen, et al. , Appl. Phys. Lett., 2012, 101, 083904; (e) G. Chen, et al. , Phys. Chem. Chem. Phys., 2012, 14, 14661–14666.
3. (a) A. Ajayaghosh, Acc. Chem. Res., 2005, 38(6), 449–459; (b) S. Das, et al. , J. Phys. Chem., 1994, 98(37), 9291–9296; (c) U. Oguz and E. U. Akkaya, Tetrahedron Lett., 1997, 38(25), 4509–4512; (d) K. George Thomas, et al. , Chem. Commun., 1997, 597–598; (e) E. Arunkumar, et al. , J. Am. Chem. Soc., 2005, 127(9), 3156–3164; (f) D. D. Ta and V. S. Dzyuba, Chemosensors, 2021, 9, 302.
4. (a) J. V. Ros-Lis, et al. , Chem. Commun., 2002, 2248–2249; (b) J. V. Ros-Lis, et al. , J. Am. Chem. Soc., 2004, 126, 4064–4065.
5. (a) J. J. Gassensmith, et al. , Org. Lett., 2008, 10, 3343–3346; (b) J. J. Gassensmith, et al. , Chem. Commun., 2009, 6329–6338; (c) E. Arunkumar, et al. , J. Am. Chem. Soc., 2005, 127, 3288; (d) E. Arunkumar, et al. , Chem. – Eur. J., 2006, 12, 4684.
6. (a) P. E. Bacher, et al. , Chem. Commun., 2019, 55, 3286–3289; (b) J. T. R. Houk, et al. , Tetrahedron, 2008, 64, 8271–8278; (c) S. H. Hewage and V. E. Anslyn, J. Am. Chem. Soc., 2009, 131, 13099–13106; (d) J. Soto, et al. , Angew. Chem., Int. Ed., 2005, 44, 4405–4407.
7. P. E. Bacher, et al. , Org. Lett., 2021, 23(8), 2853–2857.
8. (a) W. P. Jencks, Chem. Rev., 1972, 72, 705; (b) D. A. Dougherty and E. V. Anslyn, Modern Physical Organic Chemistry, University Science, Sausalito, CA, 2006.
9. M. M. Reddy, et al. , Pharmaceut. Reg. Affairs, 2016, 5, 168.
10. S. L. Buchwald, et al. , Adv. Synth. Catal., 2006, 348, 23–39.
11. B. Kaur, et al. , Coord. Chem. Rev., 2018, 358, 13–69.
12. For additional UV-Vis data on Cu(OAc)2, Pb(NO3)2, see Fig. S9, S10, S19 and S20 in the ESI†.
13. C. Wentrup, Acc. Chem. Res., 2011, 44(6), 393–404.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3cc05268k

---