Robust Photogeneration of H2 in Water Using Semiconductor Nanocrystals and a Nickel Catalyst

Science  07 Dec 2012:
Vol. 338, Issue 6112, pp. 1321-1324
DOI: 10.1126/science.1227775


Homogeneous systems for light-driven reduction of protons to H2 typically suffer from short lifetimes because of decomposition of the light-absorbing molecule. We report a robust and highly active system for solar hydrogen generation in water that uses CdSe nanocrystals capped with dihydrolipoic acid (DHLA) as the light absorber and a soluble Ni2+-DHLA catalyst for proton reduction with ascorbic acid as an electron donor at pH = 4.5, which gives >600,000 turnovers. Under appropriate conditions, the precious-metal–free system has undiminished activity for at least 360 hours under illumination at 520 nanometers and achieves quantum yields in water of over 36%.

Molecular hydrogen (H2) is a clean-burning fuel that can be produced from protons (H+) in the reductive half-reaction of artificial photosynthesis systems (1, 2). One of the most prominent strategies for light-driven proton reduction features a multicomponent solution with a light-absorbing molecule (chromophore) that transfers electrons to a catalyst that reduces protons (3, 4). However, these solution systems often use nonaqueous solvents and always have short lifetimes from decomposition of the chromophore over a period of hours (5). This difficulty has led to more-complicated architectures that separate the sites of light absorption and proton reduction (2).

Semiconductor nanocrystals (NCs) are promising alternative chromophores for light-driven proton reduction (6, 7). Compared with traditional organic or organometallic chromophores, NCs have superior photostability, larger absorption cross-sections over a broad spectral range, orders of magnitude longer excited state lifetimes, electronic states and associated optical properties that vary with NC size, and the capacity to deliver multiple electrons with minimal structural perturbations (6, 7). Heterostructures combining NCs with traditional precious-metal nanoparticle proton-reduction catalysts, or with iron hydrogenases, have produced efficient proton-reduction catalysis in solution (810). However, small-molecule catalysts in conjunction with NCs have given only modest H2 production (11, 12).

We report here a system that provides light-driven H2 production with exceptional longevity, maintaining its high activity with no decrease for over 2 weeks using water as solvent. The system uses no precious metals and is based on light absorption and photoinduced electron transfer from semiconductor nanocrystals that are photochemically stable. Under optimal conditions, the system generates over 600,000 turnovers of H2 (with respect to catalyst) without deterioration of activity and thus has promise for incorporation into full artificial photosynthesis (AP) systems.

Hydrophobic CdSe NCs with diameters varying from 2.5 to 5.5 nm [defined on the basis of the wavelength of the peak in their lowest energy excitonic absorption feature as NC(520) and NC(620) respectively, fig. S1] were synthesized by variations of literature methods (13, 14). These NCs were subsequently rendered water soluble by capping with dihydrolipoic acid (DHLA, Fig. 1) (14). Photochemical experiments were performed in a custom-built 16-sample apparatus with excitation at 520 nm and a measurement uncertainty of 7.0% in the amount of H2 produced (based on multiple-run experiments). Each 40-ml sample vessel contained 5.0 ml of solution, a cap with a sensor to allow real-time monitoring of head space pressure, and a port for quantitative analysis of H2 production.

Fig. 1

Cartoon that illustrates the relevant energies for H2 production. dHA indicates dehydroascorbic acid. Potentials are shown versus that of an NHE at pH = 4.5.

In a complete artificial photosynthesis system, two molecules of water are photochemically split into H2 (reductive side) and O2 (oxidative side) (1, 2). Electrons from the oxidative side of the system are shuttled to the reductive side where they reduce protons to hydrogen. However, because of the complexity in building and optimizing such a complete artificial photosynthesis system, when studying the reductive half-reaction it is common for a sacrificial electron donor to be used to optimize catalysts for hydrogen production. Here, ascorbic acid (AA, 0.1 to 1.0 M) was used as the sacrificial electron donor, because reduction of protons by ascorbate is thermodynamically unfavorable (E) = –0.41 V] under these conditions (15) and therefore light energy is needed to bring about H2 production (Fig. 1). Thus, this experiment provides a proof of principle regarding photochemical proton reduction.

In a typical experiment, production of hydrogen occurred upon irradiation (λ = 520 nm) of a solution formed from nickel(II) nitrate and NCs in water. Under appropriate conditions [10.0 μM Ni(NO3)2, 0.5 μM NC(570), and 1.0 M AA], the system continues to produce H2 at a constant rate for over 360 hours (Fig. 2A). A control experiment without added Ni2+ yielded no substantial H2 production (Fig. 2A). The unusual longevity is attributed to the use of NCs as the photosensitizer, because other systems using transition metal catalysis and small-molecule photosensitizers (organic dyes or Ru, Ir, Rh, or Re coordination compounds) cease activity in under 50 hours because of bleaching of the dye (1622). Other recent reports have described NC photosensitizers in systems for H2 production, but the activity and longevity were not comparable to those described here (11, 12).

Fig. 2

(A) H2 production over time from irradiation of an aqueous solution of Ni(NO3)2, NC(540), and AA. (B) Photoreductive H2 production with different initial concentrations of AA in a system containing 20.0 μM Ni(NO3)2 and 1.0 μM NC(570). The marks on the right axis indicate the theoretical maximum of H2 production on the basis of the amount of AA added. Hydrogen photogeneration experiments used a LED source (λ = 520 nm, 13 mW/cm2) at 15°C and 1 atm initial pressure of N2:CH4 (79:21 mole %) with CH4 as an internal standard for H2 quantification by gas chromatography analysis.

By using different concentrations of system components that were chosen to maximize catalyst activity [1.0 μM Ni(NO3)2, 5.0 μM NC(540), and 0.8 M AA, pH = 4.5 in water], this same system achieves a turnover number over 600,000 moles of H2 per mole of catalyst after 110 hours and an initial turnover frequency of 7000 moles of H2 per mole of catalyst per hour upon irradiation with 520-nm light (fig. S2). The slight decrease in H2 production rate in fig. S2 results from depletion of AA as the electron donor and not from photochemical instability, as evidenced by the increasing lifetime of H2 production with higher starting concentrations of AA (Fig. 2B). Even higher activity is obtained under the same irradiation conditions if the solvent is changed to 1:1 ethanol (EtOH):H2O (fig. S3). The initial rate of H2 production is saturated above [AA] = 0.3 M (Fig. 2B) and slows over time only upon depletion of the electron donor AA. Consistent with this interpretation, subsequent addition of AA restarts H2 production (fig. S4).

We hypothesize that the catalytic system functions through light absorption by the CdSe nanocrystal, electron transfer to the catalyst, and then proton reduction by the catalyst. The oxidation of AA, which fills the photogenerated hole on the NC, leads to dehydroascorbic acid, 2e, and 2H+. Ascorbic acid thus serves as both a potential source of hydrogen and as a buffer because of the production of the conjugate base, helping to maintain the acidic pH even as protons are reduced to hydrogen. The absorption wavelength maximum of the first excitonic state can be controlled by NC size, which correlates with the reduction potential of the excited state (23). In our photocatalytic H2 production system, reducing NC size leads to an increase in the activity (Fig. 3A), which we attribute to an increase in NC reducing power. Conversely, there is no formation of H2 with NC(620), presumably because the reduction potential for NC(620) is not reducing enough for catalyst activity (12). Because the NC absorption edge is to the blue of the light-emitting diode (LED) spectral emission profile, the system with NC(520) produces less H2 than an identical one with NC(540).

Fig. 3

(A) H2 production from irradiation of aqueous solutions using different sizes of CdSe NCs (4.0 μM) labeled by the peak in their first excitonic absorption, 4.0 μM Ni(NO3)2, and 0.5 M AA. (B) After 24 hours, an active solution was filtered to separate the NCs from the solution and the nickel(II) catalyst. After the separation, each individual component was inactive but regained activity when the other was added.

Organic electron acceptors were also used as indicators for the reducing power of the CdSe NCs. When a 3.8 μM solution of NC(530) in 1:1 EtOH:H2O was irradiated in the presence of methyl viologen dication (MV2+) under N2 for 5 min, a color change from orange to blue indicated formation of reduced viologen MV+•. A similar result was obtained by using a diquat acceptor DQ2+ [N,N'-(1,3-propylene)-5,5′-dimethylbipyridine], as indicated by the pink color of the reduced DQ+• (fig. S5). Although the precise potential for each NC was not determined, the result for DQ2+ suggests that the reducing ability of NC(530) corresponds to a potential more negative than –0.7 V versus a normal hydrogen electrode (NHE), and thus the excited state is sufficiently reducing for H+ reduction. These measurements agree with published cyclic voltammetric studies that indicate a reduction potential more negative than –1 V for CdSe NCs of this size (23).

The catalytic mechanism was evaluated by varying the concentrations of system components. When the Ni2+ concentration was varied, the rate of H2 production reached a maximum at 20 μM Ni2+, whereas when the concentration of NC(520) was varied, the rate leveled off above 4.0 μM NC (fig. S6). These results suggest that at a Ni2+ concentration of 20 μM or greater, the rate becomes limited by NC light absorption, whereas at a NC concentration of 4.0 μM, the system is limited by the H2-forming reaction at Ni2+. Similarly, the rate of H2 production depends linearly on light intensity up to about 13 mW/cm−2 (fig. S7). Overall, it appears that photon absorption by NCs and catalysis by Ni2+ have similar enough rates that each can be rate-limiting under different conditions.

Quantum yields for H2 generation were determined for the system at [Ni2+] = 20 μM (where the rate of H2 evolution is controlled by NC concentration) with NCs of different sizes. In water, the quantum yield ϕ based on two photons per H2 evolved is 36 ± 10% at [NC(520)] = 1 to 2 μM, decreasing to 20 ± 2% at [NC(520)] = 4.0 μM (supplementary materials); similarly, ϕ(Η2) is 35 ± 4% at [NC(540)] = 4.0 μM. The quantum yield reaches 59 ± 8% when using a 1:1 EtOH/H2O mixture as the solvent (table S2). Complete conversion of all available light energy is presumably prevented by electron-hole recombination and/or fast nonradiative electronic relaxation.

We next sought to determine whether the active catalyst was on the NC surface or in solution during hydrogen photogeneration. After irradiation of a NC(540)-based H2 generating system for 24 hours, the NCs were separated from the solution by centrifugation and filtration, and each component was examined separately for its H2-generating activity with added AA. Neither the NCs nor the solution was found to have any substantial activity for photochemical H2 generation. The chemical composition of the NCs and solution were each examined by atomic absorption spectroscopy (AAS), showing that >97% of the Ni and <3% of the Cd remained in solution whereas >97% of Cd and <3% of the Ni remained in the precipitated NCs (table S3). Additionally, transmission electron microscopy images of the separated NCs showed no substantial change in NC size, and energy dispersive x-ray analysis showed no evidence of colloidal Ni deposited on the NC surface (figs. S8 and S9). Addition of Ni2+ and AA to the NCs restored activity for H2 production upon resuspension; likewise, when fresh NC(540) and AA were added to the Ni-containing solution, we observed activities that were similar to those during the initial irradiation (Fig. 3B). The results indicate that the active catalyst is a Ni species generated in solution and that the NCs maintain their ability to act as the photosensitizer during the catalytic process.

The use of different Ni2+ salts [Ni(NO3)2, NiCl2, and Ni(acetate)2] produced a similar level of H2 production activity, suggesting that the actual catalyst is generated in situ. Because DHLA binds to Ni2+ with binding constants near or above 1010 (24) and nickel thiolates have previously been reported as catalysts for light-driven hydrogen production (25), we focused on the hypothesis that the catalyst is a nickel complex chelated by DHLA. Maintaining a solution of NC(520) at pH 4.5 under N2 in the absence of light for 5 hours and centrifuging to precipitate the NCs gave a solution with 8 to 14 molecules of DHLA per NC, which had apparently dissociated from the NC. Due to the presence of free DHLA the formation of a Ni2+-DHLA complex is both possible and favorable in the catalytic solutions. Adding up to 100 equivalents of excess DHLA gave similar activity toward H2 production (fig. S10), but addition of EDTA (which prevents formation of a Ni2+-DHLA complex) eliminated activity (fig. S11). Addition of colloidal Ni0 (4 nm in diameter) in place of the nickel(II) salt gave no significant amount of H2 under the standard catalytic conditions (fig. S11). Synthesizing the Ni2+-DHLA catalyst ex situ and adding it to the NC solution (in place of Ni2+ salts) gives the same overall activity for H2 production. All together, these experiments are consistent with a soluble nickel(II)-DHLA species being catalytically active. Lastly, electrochemical studies on independently prepared 1:1 Ni2+:DHLA solutions (0.2 μM in 1:1 EtOH:H2O) showed a cathodic feature at –0.9 V versus NHE that appears only upon addition of acid (fig. S12), indicating that Ni-DHLA can reduce protons catalytically and at a potential comparable to that produced by the excited NCs.

Although we have not yet determined the structure of the catalytically active nickel species, spectroscopic studies on Ni2+-DHLA help to understand the predominant forms of nickel in solution. The ultraviolet-visible spectrum of a 1:5 mixture of Ni2+ (50 μM) and DHLA (250 μM) at pH = 4.5 is very similar to that generated with a mixture of Ni2+ (50 μM) and 1,3-propanedithiol, suggesting that Ni2+ coordinates via the S donors of DHLA (fig. S13). The absorption maxima from a 1:1 Ni:DHLA solution follow Beer’s Law between 5.0 and 500 μM, suggesting that nickel speciation does not change over the Ni2+ and DHLA concentrations used in the catalytic experiments (fig. S14). A Job plot of Ni2+ and DHLA in this concentration regime has a maximum at a metal:DHLA ratio of roughly 1:1, suggesting that the predominant complex has one DHLA per Ni2+ (fig. S14). X-ray crystal structures of nickel complexes with related dithiols (including 1,3-propanedithiol) have shown multimetallic structures containing square-planar nickel(II) centers bridged by thiolates, with stoichiometries such as 3:4, 4:4, 6:6, and 6:7 (2628). Because nickel-thiolate species are labile in solution, many nickel species are accessible under the reaction conditions, and detailed mechanistic studies will be necessary to identify the one(s) responsible for proton reduction in this system.

A light-driven system for the photogeneration of hydrogen that consists of simple components containing only Earth-abundant elements could have a substantial impact on the sustainable production of chemical fuels. Further, the robustness of the system may be generalizable to other nanoparticle systems, such as type II NCs and dot-in-rod NCs (6, 10), which are better engineered for charge separation. When considering the efficiency of this system in a real-world context, further improvements could be made by adding light-harvesting components that absorb more of the solar spectrum, because with NC(540) only about 25% of the available solar flux is absorbed. Nonetheless, this particular NC-DHLA-Ni system exhibits high activity for proton reduction and impressive durability, which suggests that it could also serve as a valuable component in complete artificial photosynthetic water splitting systems for light-to-chemical energy conversion.

Supplementary Materials

Materials and Methods

Figs. S1 to S16

Tables S1 to S3

References (2936)

References and Notes

  1. Materials and methods are available in the online supplementary materials.
  2. Acknowledgments: Z.H. led the effort for hydrogen generation and catalysis, and F.Q. for nanocrystal synthesis and characterization. This work was supported by the Office of Basic Energy Sciences, U.S. Department of Energy, grant DE-FG02-09ER16121 and DE-SC0002106. The authors acknowledge the University of Rochester Medical Center and the Department of Environmental Medicine for the analytical support and R. M. Gelein for the cadmium and nickel AAS measurements.
View Abstract


Navigate This Article