Hydrogen Produced from Hydrohalic Acid Solutions by a Two-Electron Mixed-Valence Photocatalyst

See allHide authors and affiliations

Science  31 Aug 2001:
Vol. 293, Issue 5535, pp. 1639-1641
DOI: 10.1126/science.1062965


Energy conversion cycles are aimed at driving unfavorable, small-molecule activation reactions with a photon harnessed by a transition metal complex. A challenge that has occupied researchers for several decades is to create molecular photocatalysts to promote the production of hydrogen from homogeneous solution. We now report the use of a two-electron mixed-valence dirhodium compound to photocatalyze the reduction of hydrohalic acid to hydrogen. In this cycle, photons break two RhII–X bonds of a LRh0–RhIIX2 core in the presence of a halogen trap to regenerate the active LRh0–Rh0 catalyst, which reacts with hydrohalic acid to produce hydrogen.

Light-driven energy conversion schemes were suggested as an alternative energy source to expendable fossil fuel reserves nearly a century ago (1). The target fuel of schemes developed since this proposal has been H2, generated from protic solutions. The most successful approaches to date have used a heterogeneous catalyst to couple the 1eequivalency provided by a photoexcited transition metal sensitizer [e.g., Ru(II) polypyridine complexes] to the 2eequivalency required for H2 production (2–5). Photoexcited electrons in the conduction bands of semiconducting electrodes and nanoparticles are especially effective in promoting the reduction of H+ to H2 (6,7). Utilization of the attendant hole in the valence band to generate a complementary oxidation product completes the photocatalytic cycle. Sutin and co-workers (8) showed that photocatalytic H2 production in homogeneous solution could also be achieved when the heterogeneous mediator is replaced by a molecule capable of mediating the 2e reduction of 2 H+ to H2.

Hydrogen production in the absence of mediators has also been realized, but only under stoichiometric conditions. The irradiation of acidic solutions containing reduced metal ions such as Ce3+(9), Cr2+ (10), and Fe2+(11) produces H2 and the 1eoxidized metal cation. In these schemes, excitation of a charge-transfer–to–solvent (CTTS) absorption band effectively produces solvated electrons that are trapped by protons to produce H radicals, which combine to produce H2. The seminal work of Gray and co-workers on the photochemistry of dirhodium(I) isocyanides in hydrohalic acid (HX) solutions (12, 13) established the viability of using a molecular excited state to promote H2 production in homogeneous solution. Irradiation of the binuclear RhI complex, [Rh2(bridge)4]2+ (bridge = 1,3-diisocyanopropane), in aqueous HX solutions resulted in the stoichiometric production of 1 equivalent (equiv) of H2 and 1 equiv of [Rh2(bridge)4Cl2]2+. The difficulty associated with the subsequent activation of the RhII–X bonds of the [Rh2-(bridge)4Cl2]2+photoproduct prevented the regeneration of the initial photoreagent, precluding catalytic turnover.

In view of the challenges posed by M–X bond formation to designing HX-splitting cycles, we sought to develop excited states that can promote M–X photoactivation to regenerate active hydrogen-producing species. By introducing the same dσ* excited state within the electronic structure of LRh0–Rh0L, LRh0–RhIIX2, and X2RhII–RhIIX2 cores (X = Cl, Br; L = CO, PR3, CNR) spanned by three bis(difluorophosphino) methylamine [dfpma = MeN(PF2)2] ligands (14,15), we could interconvert among the series members by photoeliminating halogen in 2e steps. We obtained the mixed-valence LRh0–RhIIX2 species, Rh2(dfpma)3X2(L), quantitatively when solutions of Rh2(dfpma)3X4containing excess L were photolyzed (300 nm ≤ λexc ≤ 480 nm) in the presence of a halogen-atom trap such as tetrahydrofuran (THF), dihydroanthracene, or 2,3-dimethylbutadiene. Irradiation with excitation wavelengths (300 nm ≤ λexc ≤ 400 nm) coincident with the absorption manifold of the LRh0–RhIIX2 photoproduct resulted in a second 2e elimination reaction to give the LRh0–Rh0L dimer, Rh2(dfpma)3L2, again in quantitative yield. In the overall transformation, the two-electron mixed-valence LRh0–RhIIX2 compound sustains the multielectron photoreactivity of the system by coupling the 2e M–X chemistry of the individual Rh centers. We now show that M–X photoactivation from this two-electron mixed-valence platform enables the photocatalytic production of H2 from homogeneous solutions of HX.

Hydrohalic acid reacted at the Rh2 0,0-(dfpma)3(L) core to produce the two-electron mixed-valence species Rh2 0,II(dfpma)3-X2(L) (1) and H2 upon removal of a single axial ligand from the coordinatively saturated Rh2 0,0(dfpma)3LL′ (2) complex (Scheme 1). The vacant coordination site can be generated photochemically when one axial ligand is CO. We prepared the asymmetric complex, Rh2 0,0(dfpma)3(PPh3)(CO) (16), which has one photoinert axial ligand, PPh3, and one photolabile axial ligand, CO. Whereas CH2Cl2 solutions of Rh2 0,0(dfpma)3(PPh3)(CO) did not react with HX (X = Cl or Br), a rapid reaction was observed when the CO was removed by ultraviolet-visible (UV-Vis) irradiation (excitation wavelength λexc ≥ 338 nm) (17). The spectral changes for the HCl reaction are displayed in Fig. 1. The final absorption profile is identical to that of Rh2 0,II(dfpma)3Cl2(PPh3), which was obtained in quantitative yield. Experiments in which a Toepler pump was used to collect H2 show the formation of 0.87 ± 0.14 equiv of H2 gas per 1.0 equiv of metal complex (18, 19), establishing the overall stoichiometry of the reaction to beEmbedded Image Embedded Image Embedded Image(1)Wavelength selection of the excitation light reveals that H2 is produced in a stepwise reaction sequence. Irradiation of Rh2 0,0-(dfpma)3(CO)(PPh3) in a 0.1 M HCl/THF solution with λexc = 335 nm led to changes in the UV-Vis absorption spectrum (Fig. 2A). Disappearance of the Rh2 0,0(dfpma)3(PPh3)- (CO) absorption features was accompanied by the growth of an absorption band centered at 580 nm. The reaction proceeds smoothly with isosbestic points maintained at 296 and 338 nm. After 90 min of irradiation, no further changes were observed in the 580-nm absorption band; 0.35 ± 0.07 equiv of H2 was produced during this photoconversion. The 580-nm absorption profile did not change with continued irradiation, nor did its intensity change for solutions stored in the dark. Although the 580-nm intermediate has not been isolated and characterized, its spectral profile is a signature of a tetranuclear Rh cluster with mixed-valence character (20–23). As observed by Gray and co-workers in their studies of [Rh2 bridge)4]2+(20), a tetranuclear Rh species can result from the bimolecular reaction of two hydrido-halo dirhodium cores to produce 1 equiv of H2 and a tetranuclear species (0.5 equiv of H2 per equiv of Rh2 complex). When the excitation wavelength is changed to frequencies within the spectral envelope of the 580-nm band, the absorbance decreases with the concomitant appearance of the two-electron mixed-valence dihalide, Rh2 0,II(dfpma)3Cl2(PPh3) (Fig. 2B). Consistent with the overall stoichiometry of Eq. 1, Toepler pumping revealed that the remaining 0.38 ± 0.07 equiv of H2 was produced upon conversion to the final LRh0–RhIIX2 photoproduct.

Figure 1

Changes in the electronic absorption spectrum during photolysis. UV-Vis photolysis (λexc ≥ 338 nm) of Rh2(dfpma)3(CO)(PPh3) in 0.1 M HCl/CH2Cl2 solution at 20°C. The sample was irradiated for a total of 25 min. Arrows denote the evolution of the absorption spectrum with time.

Figure 2

Changes in the electronic absorption spectrum during photolysis. (A) Monochromatic UV photolysis (λexc = 335 nm) of Rh2(dfpma)3(CO)(PPh3) in 0.1 M HCl/THF solution at 20°C. The sample was irradiated for a total of 90 min. (B) Continued visible photolysis of the solution of (A) with long-pass filtered excitation light (λexc ≥ 338 nm). Arrows denote the evolution of the absorption spectrum with time.

Figure 3

The plot of total H2 collected as a function of time for the UV-Vis white-light irradiation (λexc ≥ 338 nm) of Rh2(dfpma)3Cl2(PPh3) in 0.1 M HCl/THF solution at 20°C.

Scheme 1

Because the LRh0–Rh0 species can be regenerated by photolysis of Rh2 0,II-(dfpma)3X2(PPh3) in the presence of halogen-atom traps (14, 15), the reaction of Eq. 1 can be turned over and H2 can be photocatalytically generated. A 50-ml THF solution containing Rh2 0,II(dfpma)3Cl2(PPh3) (0.95 μmol) and 0.1 M HCl was irradiated with λexc≥ 338 nm, after which the photolysis was halted and H2 was collected by Toepler pumping (24). Figure 3 displays the plot of total H2 production versus time; an initial rate of 27 turnovers per hour was observed during the initial 3 hours of irradiation. The decrease in rate of H2 production with continued photolysis was accompanied by a steady decrease in the Rh2 0,II(dfpma)3Cl2(PPh3) absorbance, indicating the decomposition of the two-electron mixed-valence dimer. Preliminary investigations indicate that this decomposition product is a Rh monomer. When the photoreaction was performed with monochromatic excitation at 335 nm, the foregoing 580-nm intermediate was observed. Under these conditions, subsequent irradiation into the 580-nm absorption band was needed to turn the reaction over.

The above results are consistent with the catalytic cycle shown in Scheme 2. Per Toepler pump experiments, HX reacts with a coordinatively unsaturated LRh0–Rh0 core to produce H2 (0.5 equiv per Rh2 complex) and the blue intermediate, which we tentatively ascribe to a tetranuclear Rh species. This intermediate species is photochemically stable in the absence of visible-light irradiation, but a 580-nm photon prompts reaction to give 2 equiv of the LRh0–RhIIX2 mixed-valence complex with the generation of an additional equivalent of H2. The near-UV irradiation of the two-electron mixed-valence photoproduct in the presence of a halogen-atom trap regenerates LRh0–Rh0, which is available to react with HX to turn the cycle over.

Scheme 2

We believe that the two-electron mixed-valence character of the dihalide species is crucial to the overall reactivity. Reductive elimination from Rh2 cores is facile when a Rh–Rh bond is preserved (25, 26). InScheme 2, the LRh0–Rh0L species contains a metal-metal bond that is characteristic of the pairing of electrons within the dz2σ orbital of a d9–d9 core (27). As we have established computationally and spectroscopically (14), the same Rh–Rh bond interaction arises from the pairing of electrons in the dz2 orbitals of a d7–d9 core of a LRh0–RhIIX2 complex. Consequently, a Rh–Rh bond is maintained upon X elimination, a situation that does not arise for valence-symmetric binuclear Rh cores, including that of [Rh2(bridge)4Cl2]2+. This distinguishing trait of the two-electron mixed-valence core may be crucial to the smooth conversion of Rh2 0,II(dfpma)3X2(L) to Rh2 0,0(dfpma)3L, thus enabling turnover to be achieved.

Scheme 2 constitutes a photocatalytic HX-splitting cycle that is driven by halogen-atom trapping. An authentic energy-storing cycle demands the isolation of X2 in the absence of a halogen-trapping reagent. The key to achieving this objective is intimately tied to the M–X photoelimination step. That the activation of the Rh–X bond is determinant to overall photocatalytic activity ofScheme 2 is established by the rate of H2 production, which is commensurate with the quantum efficiency for the stoichiometric conversion of Rh20,II(dfpma)3Cl2(L) to Rh20,0(dfpma)3L2p = 6 × 10−3(15)]. This result highlights that an increased understanding of M–X photoactivation processes is imperative to the development of an energy conversion photocatalyst at the molecular level.

  • * To whom correspondence should be addressed. E-mail: nocera{at}


View Abstract

Navigate This Article