Quantum Dot-Catalyzed Photoreductive Removal of Sulfonyl-Based Protecting Groups

Abstract

This communication describes the use of CuInS₂/ZnS quantum dots (QDs) as photocatalysts for the reductive deprotection of aryl sulfonyl-protected phenols. For a series of aryl sulfonates with electron-withdrawing substituents, the rate of deprotection for the corresponding phenyl aryl sulfonates increases with decreasing electrochemical potential for the two electron transfers within the catalytic cycle. The rate of deprotection for a substrate that contains a carboxylic acid, a known QD-binding group, is accelerated by more than a factor of ten from that expected from the electrochemical potential for the transformation, a result that suggests that formation of metastable electron donor-acceptor complexes provides a significant kinetic advantage. This deprotection method does not perturb the common NHBoc or toluenesulfonyl protecting groups and, as demonstrated with an estrone substrate, does not perturb proximate ketones, which are generally vulnerable to many chemical reduction methods used for this class of reactions.

Graphical Abstract

Quantum dots photocatalyze the reductive deprotection of phenyl aryl sulfonates. The rate of deprotection is enhanced for substrates that bind to the surface of the quantum dots.

Sulfonyls are viable protecting groups for phenols, because, unlike alkylsulfonates, phenylsulfonates are stable to nucleophilic attack. Sulfonyl protection strategies for phenols are perhaps not as widely used as they could be, however, because the deprotection requires harsh conditions. Electrochemical cleavage of the S-O bond of a sulfonyl protected phenol requires reduction with a potential of at least −1.4 V vs. SCE,^[1] and thermal deprotection reactions employ strong acids (H₂SO₄) or strong reducing agents (Na, Li or Mg) that are incompatible with substituents or other protecting groups on the phenol.^[2] Given that these deprotection reactions have been accomplished via electrochemical deprotection^{[1a, 3]} and direct photolysis,^[4] and that visible-spectrum photons have energies ranging from 3.10 eV (400 nm) to 1.77 eV (700 nm), the deprotection of sulfonated phenols is a good candidate reaction for a photoredox approach.

In the photoreductive deprotection, a photosensitizer absorbs two successive photons, donates two electrons in sequence to liberate a phenol and the corresponding sulfinate anion from the protected species, and donates two holes to a sacrificial terminal reductant. Unlike direct photolysis, photoredox catalysis does not require that a substrate be directly photoexcited, so UV light-driven undesirable side reactions are minimized.^[5] For more than 30 years,^[6] photoredox strategies using organic dyes and ruthenium- and iridium-based complexes as photocatalysts have enabled deprotection of, for example, para-methoxybenzyl protected alcohols^[7] and amides,^[8] 4-methoxybenzyl protected primary amines and alcohols,^[9] and N-tosyl amides.^{[6a, 10]} This strategy has not been demonstrated for sulfonyl protected phenols, however, possibly because the only commercially available inorganic complexes with excited-state oxidation potentials greater than −1.4 V vs. SCE are fac-Ir(ppy)₃, fac-Ir(dF-ppy)_3, and [Ir(ppy)₂(dtbbpy)]⁺, all of which have peak absorption at wavelengths shorter than 400 nm.^[11] Photons with energy above 400 nm often directly photoexcite the substrate (in addition to the photosensitizer) and thereby enable parasitic reactions. Recently, colloidal semiconductor quantum dots have been investigated as visible-light-absorbing photoredox catalysts for organic transformations.^[12]

This paper describes the photocatalytic reductive deprotection of a series of aryl sulfonyl-protected phenols using copper indium sulfide/zinc sulfide core/shell quantum dots (CuInS₂/ZnS QDs) as light harvesters and photo-redox catalysts. CuInS₂/ZnS QDs have high excited-state oxidation potentials, >−1.9 V vs. SCE, and excited-state lifetimes of ~30–500 ns.^[13] While, as mentioned above, these QDs are not unique in their ability to catalyze this class of reaction, we highlight in this report two advantages of using CuInS₂/ZnS QDs for photocatalyzed deprotections: (i) their size can be tuned to absorb photons as low-energy as 730 nm (we use 532-nm illumination in this study),^{[13b, 14]} whereas the Ir compounds require blue or UV light, and (ii) while photoredox reactions catalyzed by molecules require diffusion-limited collisions between substrates and excited-state catalysts, we can design QDs to reversibly associate with suitably functionalized or charged substrates.^[15] Formation of metastable complexes of photocatalyst and substrate is beneficial for increasing the yield of photoredox reactions and extending the scope to include endergonic reactions, because this association increases the probability of electron transfer to the substrate or intermediate upon photoexcitation of the catalyst.^[16]

Figure 1 summarizes our proposed photocatalytic cycle for the 2-e⁻ deprotection of phenyl aryl sulfonates. CuInS₂/ZnS QDs, synthesized using an adapted literature procedure^[13b] and solubilized in DMSO-d₆ with a 3-mercapto-1-propanol ligand shell (details in the Supporting Information, “SI”), are photoexcited by a 532-nm photon to form QD*. QD* then transfers one electron to the substrate and one hole to the sacrificial reductant, triethylamine (TEA), to return to its ground state. The S-O bond in the singly reduced substrate breaks to form a phenoxide ion, which is protonated to form phenol, and a sulfonyl radical. The QD is then excited a second time to form QD*, whereupon QD* transfers a hole to TEA and an electron to the sulfonyl radical to form a sulfinate anion.

Figure 1.

Reaction conditions and proposed catalytic cycle for the photocatalyzed 2-e⁻ deprotection of aryl sulfonates by CuInS₂/ZnS QDs. Ar₁ and Ar₂ are aryl groups on the protected alcohol and the protecting group, respectively, see Table 1. Sacrificial triethylammonium (TEA⁺) decomposes into diethyl amine and acetaldehyde, see Figure S1 of the SI.

We synthesized a series of phenyl aryl sulfonates, Table 1, using adapted literature procedures (see the SI),^[17] and measured the potential necessary to liberate phenol from each phenyl aryl sulfonate substrate from the potential corresponding to the peak of the irreversible cathodic wave in its cyclic voltammogram (CV), E_p,c, Table 1 and the SI. The reductive wave at E_p,c is irreversible for every substrate in the presence of our sacrificial reductant, TEA, and there are no other reductive waves at potentials lower than the reduction potential of the QDs, E_red,QD, of ~−1.9 V vs. SCE (see the SI),^[13a] so we conclude that the wave includes both electrons within the catalytic cycle in Figure 1.

Table 1.

CuInS₂/ZnS QD-Catalyzed Deprotection of Phenyl Aryl Sulfonates.

Substrate	Ar1	Ar2	Ep,c[S/S2-] (V vs. SCE)a	% Conversionb	% Yieldb
1			−2.20	0	0
2			−2.20	0	0
3			−2.09	0	0
4			−2.02	0	0
5			−2.00	0	0
6			−1.81	100	92
7			−1.72	57	57
8			−1.72	78	78
9			−1.55	100	91
10			−1.48	87	83
11			−1.44	100	93
12			−1.50	69c,d	54c,d
13			−1.49	43c,d	43c,d
14			−1.48	100c	93c
15			−1.46	100c	85e
16			−1.45	98c	0c

^aThe potentials at which we observed peak cathodic current (E_p,c) were determined from CVs of 10 mM of the protected substrate under N₂ with 100 mM tetrabutylammonium hexafluorophosphate, and 1 M TEA in DMSO, using a glassy carbon working electrode, a Ag wire reference electrode, and a Pt wire counter electrode. The potentials are referenced to SCE using a decamethylferrocene internal standard. The notation “E_p,c[S/S^2-]” indicates that two electrons are transferred during the wave, one to the substrate and one to the liberated sulfonyl radical after S-O bond cleavage.

^bThe conversions and yields listed are NMR yields for samples of 10 μM CuInS₂/ZnS QDs, 0.001 M substrate, and 0.1 M TEA in DMSO-d₆ illuminated with a 4.5-mW, 532-nm laser diode for 48 hours.

^cYield and conversion determined after 24 hours.

^dThe reaction stopped progressing after 24 hours even though starting material was still present.

^eIsolated yield for sample of 0.22 mM CuInS₂/ZnS QDs, 0.022 M substrate, 2.2 M TEA in DMSO-d₆ illuminated with a 4.5-mW 532-nm laser diode for 24 hours.

Table 1 lists the ¹H NMR-determined yields of phenol product and % conversions of substrates 1 – 11 for the QD-photocatalyzed deprotection reactions, after 48 hours of illumination with a 4.5-mW, 532-nm laser diode. All NMR spectra are in the SI. For substrates where E_p,c > E_red,QD (substrates 1–5), the yield of the deprotection is zero. For substrates where E_p,c < E_red,QD (substrates 6–11), the yield generally increases as E_p,c decreases, with the exception of substrate 6, discussed below. The deprotection of 6-11 proceeds cleanly to produce phenol, as indicated by the close agreement between % conversion and % yield.

We confirmed that both electrons required to accomplish the sulfonate deprotection reaction originate from the QD (as opposed to TEA, which can act as a hydride donor), by performing the deprotection of 11 using 1000 equiv 5,10-dihydro-5,10-dimethylphenazine (DMPZ) as a sacrificial reductant instead of 1000 equiv TEA.^[18] The yield of the reaction (after illumination with 532-nm light for 2 hours) is 20% lower with DMPZ than with TEA, probably because DMPZ is larger and has a lower probability of extracting the hole from the QD, but the deprotection clearly does not require a hydride donor to occur. This result substantiates the mechanism we proposed in Figure 1.

Figure 2A shows the kinetics of the deprotection reactions for substrates 6–11. The QDs remain stable during photocatalysis; we do not observe evidence of aggregation or etching over the course of these reactions via optical spectroscopy (see the SI). We determined the initial rate of each reaction from fits to these kinetic traces in the linear regime and plotted these initial rates versus E_p,c in Figure 2B. The initial rate of the deprotection reaction generally increases as E_p,c decreases. We also observe a positive correlation between the initial rate of the reaction and the level at which % yield ([phenol]) plateaus (see the SI). We suspect the plateau is caused by competition for catalytic sites between substrates and liberated sulfinate anions (i.e. given ample time, sulfinate ions will bind to QD surface and block catalytic sites).

Figure 2.

(A) The concentration of phenol product in reaction mixtures with 10 μM CuInS₂/ZnS QDs, 0.001 M substrate (labeled as in Table 1), and 0.1 M TEA in DMSO-d₆ as function of time of illumination with a 4.5-mW, 532-nm laser diode. The potentials listed in the legend correspond to the measured E_p,c[S/S^2-] value for each substrate. The lines are the fits used to acquire the initial rates of the reactions. (B) The rate constants, as determined by the method of initial rates, for the deprotection reactions as a function of the E_p,c[S/S^2-] of the substrates, listed in Table 1.

As is clear from inspection of Figure 2B, the initial rate of deprotection of 6 is enhanced by more than a factor of ten from that expected from its E_p,c value. A direct comparison of the rate constant for deprotection of 6 with that of the corresponding ester 9, which has a lower reduction potential than 6 (by 0.26 V) leads to the tentative conclusion that the ability of 6 to associate with unoccupied binding sites on the surface of the QD through its carboxylate group provides a kinetic advantage in this reaction. None of the other substrates have substituents known to bind to the surfaces of semiconductor QDs, which typically occurs through the coordination of surface metal ions (by, e.g., COO⁻, S⁻) or dative binding of surface metal ions (by, e.g., NH₂).^[19] Importantly, carboxylate groups such as the one on 6 is known, at least for CdSe QDs, to be reversible binders,^[15] so that once the deprotection occurs, the product will not permanently block its catalytic site on the QD surface.

To support our proposal that increasing a substrate’s affinity for the QD surface increases the initial rate of 2-e⁻ deprotection of that substrate, we compared the performances of photocatalysts CuInS₂/ZnS and fac-Ir(ppy)₃ for the deprotection of substrates 6 and 11, Figure 3. Because fac-Ir(ppy)₃ does not absorb 532 nm light, we used a 5-mW 405-nm laser as an excitation source; we adjusted the catalyst loading to achieve the same optical density at 405 nm for the two photocatalysts (1 mol% CuInS₂ QDs; 9.7 mol% Ir(ppy)₃). The photoredox steps for deprotection of 11 should be diffusion controlled for both Ir(ppy)₃ and the QDs, since 11 does not have a group with affinity for the QD surface. Ir(ppy)₃ generates a factor of six more phenol from 11 than does the QD photocatalyst within the first 15 minutes of illumination. This result is expected, at least at short reaction times, because Ir(ppy)₃’s excited-state lifetime (~2 μs)^[20] is ~50× longer than that of the QDs (35 ns).^[13a] A longer excited state lifetime increases the probability of a charge transfer reaction before the photocatalyst returns to its ground state. The longer lifetime of Ir(ppy)₃ makes the results of the reaction of substrate 6, which does have a carboxylate binding group for the QD surface but has no mechanism for generating a complex with Ir(ppy)₃, even more notable: the QDs generate a factor of 6.7 more phenol from 6 than does Ir(ppy)₃ within the first 15 minutes of illumination. This comparison not only supports our proposal that meta-stable association of the substrate and the QD provides a kinetic advantage for this 2-e⁻ reaction, but also that this kinetic advantage more than compensates for its shorter excited-state lifetime when compared to Ir(ppy)₃.

Figure 3.

The yield of phenol product after the photocatalyzed deprotections of substrate 6 (black squares) and substrate 11 (blue circles) using 1 mol % CuInS₂ QDs (filled symbols) and 9.7 mol % fac-Ir(ppy)₃ (open symbols) upon illumination from a 5-mW, 405-nm laser. 6 has a carboxylic acid substituent and 11 has no substituent with reported affinity for the QD surface.

To demonstrate the orthogonality of this photocatalyzed deprotection method, we synthesized substrates that contain two protecting groups (compounds 12, 13, 14), one 4-acetylphenyl sulfonate group and another protecting group that, thermodynamically, should not be redox-active with respect to the photoexcited QD. The synthetic procedures for these substrates are in the SI. Table 1 shows that, after 24 hours of illumination of the QD-substrate mixtures with 532-nm light, the % conversion matches the % yield for deprotection of 13 and 14; we do not observe any perturbation of the tert-butyloxycarbonyl (Boc) or toluenesulfonyl (Ts) protecting groups on those substrates. Although QDs successfully liberate phenol from 12, the % conversion of starting material is 15 percentage points greater than the % yield of desired product. We suspect the benzyl ether groups on the substrate and the desired, deprotected product may partially decompose when subjected to these reductive conditions, see SI for more details. Additionally, as discussed above, we suspect that the % conversion of substrates 12 and 13 is limited by binding of the sulfinate anion product to the surface of the QD, which poisons the catalyst.

Common reducing agents for aryl sulfonyl deprotections, including magnesium, SmI₂, LiAlH₄, and sodium amalgam^[2] can also cause unwanted reductions of carbonyl groups. To demonstrate the functional group tolerance of the photocatalytic deprotection route, we attempted photo-deprotections of 15, derived from estrone, and 16, derived from vanillin, which contain carbonyl groups that would be reduced in the presence of any of the chemical reducing agents listed above. Table 1 shows that, upon illumination with 532-nm light for 24 h, we achieve selective cleavage of 15 to liberate estrone without perturbation of the ketone. In contrast, we observe no formation of vanillin from 16 even though 98% of starting material has disappeared. Given the disappearance of the sharp aldehyde peak after illumination (see the SI), we suspect the QD reduces the aldehyde to form a ketyl radical, which then couples to another radical to form C–C products.^[12b]

In summary, photo-redox catalysis is a simple and effective route to the reductive deprotection of aryl sulfonyl-protected phenols using green (532-nm) light and CuInS₂/ZnS QDs as photocatalysts, as long as the potential needed for the required two electron transfers, E_p,c, is less than the measured reduction potential of the QD (~−1.9 V vs. SCE). The initial rate of deprotection increases as E_p,c becomes more positive. We find that, if a substrate contains a QD-binding group, here a carboxylic acid, the rate of deprotection is greatly accelerated from that expected from the E_p,c value of the substrate, likely because the substrate can associate with a surface of a QD, and the rates of the two electron transfer reactions are not diffusion-controlled. The kinetic advantage provided by reversible association of QD and substrate more than compensates for the shorter excited-state lifetime of CuInS₂/ZnS QDs when we compared their performance to that of fac-Ir(ppy)₃. The QD-catalyzed photodeprotection method is selective for the removal of sulfonyl protecting groups with electron-withdrawing substituents in the presence of Boc and tosyl groups, and does not perturb proximate ketones, which are generally vulnerable to many chemical reduction methods used for this class of reactions. The orthogonality and selectivity of this method makes it potentially attractive for accomplishing deprotection steps within multi-step syntheses of complex molecules. Furthermore, it might be useful in light-controlled syntheses of polymers that proceed through successive protection and deprotection reactions.^[21] Finally, photocatalyzed deprotection reactions may be a route to unique ligand shell structures on QDs that cannot be accessed by ligand exchange; such structures could be useful for a wide variety of chemical sensing and biological tagging applications.^[22]