Type-II heterostructures of α-V₂O₅ nanowires interfaced with cadmium chalcogenide quantum dots: Programmable energetic offsets, ultrafast charge transfer, and photocatalytic hydrogen evolution

Nanoscale semiconductor heterostructures are intriguing material architectures for light harvesting, excited-state charge transfer, and solar energy conversion.^1–6 Interfacial energetics within heterostructures dictate the thermodynamic favorability of excited-state charge-transfer processes that can ultimately lead to the generation of electrical power or the storage of energy in chemical bonds. Type-II energetic offsets, in which the conduction- and valence-band edges of one semiconducting component lie at higher energies than the corresponding band edges of the other component, are desirable.^2,7,8 This staggered bandgap alignment renders the separation of photogenerated electrons and holes thermodynamically favorable following the photoexcitation of either constituent semiconductor.

Colloidal quantum dots (QDs) are prime candidates for incorporation into heterostructures. Their high oscillator strengths and broad excitonic absorption bands enable the efficient harvesting of light, while their size-dependent bandgaps engender tunability of interfacial energetic offsets and charge-transfer driving forces.^9–12 The rich surface chemistry of QDs provides a route to tether them to other material components and enables control over the interfacial distance and electronic coupling within resulting heterostructures.^2,13–17 QDs have been interfaced with metal oxide semiconductors and with other quantum dots to yield various type-II heterostructures that undergo efficient excited-state charge separation.^2,18–25

We report here on a new class of type-II heterostructures consisting of cadmium chalcogenide QDs (CdE where E = S, Se, or Te) and α-V₂O₅ nanowires (NWs). We previously reported the synthesis and characterization of α-V₂O₅ NWs via hydrothermal reduction of bulk V₂O₅ to V₃O₇·H₂O followed by oxidation in air.^26,27 On the basis of angle-resolved X-ray absorption near-edge structure spectroscopy measurements,²⁸ the conduction band edge of V₂O₅ primarily comprises V 3d_xy states that are “split off” from the remainder of the V 3d (which are further separated into t_2g and e_g manifolds as a result of hybridization with O 2p states); these states are lower in energy than the conduction band edge of CdE QDs and are thus expected to be amenable to accept electrons from photoexcited QDs.^26,29 Analogously, hard X-ray photoemission and resonant inelastic X-ray scattering measurements in conjunction with density functional theory (DFT) calculations corroborate that consistent with differences in electronegativity, the primarily O 2p states that constitute the valence band edge of V₂O₅ are at lower energies as compared to chalcogenide 3p, 4p, and 5p-derived valence band edges of CdE QDs.^26,29 We thus hypothesized that heterostructures consisting of CdE QDs and V₂O₅ NWs should exhibit type-II energetics resulting in interfacial charge separation following the photoexcitation of electron-hole pairs within either component. Notably, the approximately 2.4-eV bandgap of V₂O₅³⁰ is lower than those of metal oxides such as TiO₂ (3.2 eV) and SnO₂ (3.5 eV), which have been incorporated into heterostructures with QDs; therefore, V₂O₅ not only should serve as an electron-accepting substrate but also absorbs light within a significant region of the visible spectrum (as also demonstrated by its bright orange coloration). Moreover, the quasi-one-dimensional morphology of the V₂O₅ NWs allows for polaronic transport of conduction-band electrons following interfacial charge separation, enabling the dark oxidation and reduction processes that underpin redox photocatalysis to compete with deleterious electron-hole recombination.

In this article, we report on the synthesis and characterization of V₂O₅/CdE NW/QD heterostructures, prepared via successive ionic layer adsorption and reaction (SILAR) and linker-assisted assembly (LAA), the characterization of their photoinduced charge-transfer reactivity using transient absorption spectroscopy, and their performance in the photocatalytic reduction of protons to hydrogen. For heterostructures prepared by either SILAR or LAA, photogenerated electrons and holes were separated on subpicosecond-to-picosecond time scales to yield charge-separated states that persisted for microseconds. The heterostructures vastly outperformed the corresponding isolated QDs and NWs in the reduction of aqueous protons to hydrogen, indicating that ultrafast charge separation could indeed be exploited in redox photocatalysis. The V₂O₅/CdE heterostructures are thus versatile new material constructs for light harvesting, charge separation, and the photocatalytic production of solar fuels; polymorphism of V₂O₅ and compositional alloying of both components provide for a substantial design space for tuning of interfacial energy offsets.

Density functional theory (DFT) calculations were performed for bulk phases of α-V₂O₅, CdS, and CdSe by using the WIEN2k software package, which solves the Kohn-Sham equations using a full potential and linearized-augmented planewaves with local orbitals (LAPW+lo).^31,32 The generalized gradient approximation of Perdew, Burke, and Ernzerhof (GGA-PBE) was used for the correlation and exchange potentials for the layered structure α-V₂O₅, whereas the electron–electron correlation GGA+U was used for both CdSe and CdS unit cells. The cutoff between core and valence states was set as −6.0 Ry for three materials. The plane-wave cutoff parameter RKMAX was chosen to be 6.5 and 7 for α-V₂O₅ and CdE, respectively.

Reagents and solvents were obtained from the following sources and used as received: (1) Alfa Aesar [cadmium sulfate octahydrate (3CdSO₄·8H₂O), cadmium chloride hemipentahydrate (CdCl₂·5/2H₂O), cadmium nitrate tetrahydrate (Cd(NO₃)₂·4H₂O) selenium dioxide (SeO₂), sodium sulfide nonahydrate (Na₂S·9H₂O), sodium telluride (Na₂Te), sodium borohydride (NaBH₄), nickel(ii) nitrate hexahydrate ([Ni(NO₃)₂]·6H₂O), and 3-mercaptopropionic acid (3-MPA)]; (2) Sigma Aldrich [selenium powder, tellurium powder, bulk vanadium pentoxide (V₂O₅), l-cysteine, and lactic acid]; (3) Fisher Scientific [anhydrous sodium sulfite (Na₂SO₃), sodium thiosulfate (Na₂S₂O₃), sodium hydroxide, methanol, and hydrochloric acid]; (4) Decon (ethanol); and (5) J. T. Baker (oxalic acid).

Cysteinate-capped CdSe QDs, hereafter referred to as cys-CdSe QDs, were synthesized following the procedure reported by Nevins et al.³³ A selenide precursor solution was prepared by dissolving selenium powder (0.17 g, 2.2 mmol) and Na₂SO₃ (0.80 g, 6.4 mmol) in deionized (DI) H₂O (42 ml) in a round-bottom flask. The reaction mixture was heated to reflux and stirred until selenium dissolved. The cadmium precursor was prepared by dissolving 3CdSO₄·8H₂O (0.57 g, 0.74 mmol) and l-cysteine (1.03 g, 8.48 mmol) in DI H₂O (53 ml). The pH of the solution containing the cadmium precursor was adjusted to approximately 12 with NaOH pellets. The cadmium precursor solution was heated to 80 °C, and then 23 ml of the selenide precursor solution was added via hot injection. The resulting mixture was stirred at 80 °C for approximately 2 h and then cooled to room temperature and stored until further use.

Cysteinate-capped CdS QDs, hereafter referred to as cys-CdS QDs, were synthesized by modification of the synthesis of cys-CdSe QDs.^33,34 A sulfide precursor solution was prepared by dissolving Na₂S₂O₃ (1.17 g, 7.41 mmol) in DI H₂O (25 ml). The mixture was heated to reflux and stirred to dissolve the solid. The cadmium precursor solution was prepared by dissolving 3CdSO₄·8H₂O (0.87 g, 3.4 mmol) and l-cysteine (2.05 g, 16.9 mmol) in DI H₂O (42 ml). The pH of the solution containing the cadmium precursor was adjusted to approximately 12 with NaOH pellets. The cadmium precursor was heated to 80 °C, and then 9 ml of sulfide precursor solution was added via hot injection. The resulting mixture was stirred and kept at 80 °C for approximately 3 h and then cooled to room temperature and stored until further use.

α-V₂O₅ NWs were synthesized via hydrothermal reduction of bulk V₂O₅ as reported previously.²⁷ Hydrothermal reaction of V₂O₅ and oxalic acid yielded V₃O₇ NWs, which were then oxidized in air at 300 °C in a muffle furnace to yield α-V₂O₅ NWs.

SILAR-derived α-V₂O₅/CdE heterostructures were assembled following the procedure reported by Pelcher et al.³⁵ A 100-mM cadmium precursor solution was prepared by dissolving Cd(NO₃)₂·4H₂O in ethanol (15 ml). A 50-mM selenide precursor was prepared by dissolving solid-phase powders of SeO₂ and NaBH₄, with a 1:1 molar ratio of Na:Se of in ethanol (30 ml), followed by degassing and purging with Ar. The mixed solution was stirred at room temperature for 2 h. A 50-mM sulfide precursor solution was prepared in a similar way by dissolving Na₂S·9H₂O in ethanol (30 ml). A 50-mM telluride precursor was prepared by dissolving Na₂Te in ethanol (30 ml). In the first step of SILAR deposition, dispersions of α-V₂O₅ NWs (50 mg) in ethanol (15 ml) were combined with the 100-mM cadmium precursor solution (15 ml) in a 1:1 ratio by volume, decreasing the concentration of Cd²⁺ to 50 mM. The resulting mixture was stirred for 30 s. NWs were then removed via centrifugation and washed with ethanol. NWs were then dispersed into a given chalcogenide precursor solution. The resulting mixture was stirred, centrifuged, and washed with ethanol. This series of steps represents one SILAR cycle. NW/QD heterostructures were prepared via variable numbers of SILAR deposition cycles. SILAR-derived heterostructures were prepared via 3 deposition cycles unless otherwise mentioned.

LAA-derived α-V₂O₅/CdSe and α-V₂O₅/CdS heterostructures were assembled following the procedure reported by Pelcher et al.³⁵ Stock aqueous dispersions of CdE QDs were washed 1–3 times to remove unreacted precursors. In a given washing step, QDs were flocculated by adding MeOH to the aqueous dispersion in a 3:1 ratio by volume, collecting the QDs by centrifugation, discarding the supernatant, and redispersing the QDs into DI H₂O to the original volume. Stock dispersions of α-V₂O₅ NWs were prepared by adding 10 mg of NWs to 1 ml of DI H₂O while sonicating. Aqueous dispersions of CdE QDs (0.8 ml for CdS and 0.4 ml for CdSe) were added to dispersions of NWs (2 ml) with constant stirring. The pH of the final mixture was adjusted to approximately 5 by addition of dilute HCl. Mixed dispersions were stirred for 1 h. Heterostructures were collected via centrifugation and washed once with DI H₂O to remove unattached QDs. The resulting V₂O₅/CdE heterostructures were dried and stored as a solid until further use. Films of SILAR- and LAA-derived heterostructures, which were used for transient absorption measurements, were prepared by spray-coating dispersions of heterostructures in ethanol onto glass substrates.

Scanning electron microscopy (SEM) images and energy dispersive X-ray spectroscopy (EDS) spectra were acquired with a Hitachi SU-70 instrument equipped with an X-ray detector. Transmission electron microscopy (TEM) images and selected area electron diffraction (SAED) patterns were obtained using a JEOL-2010 instrument operated at 200 kV.

UV/vis absorbance spectra were obtained with an Agilent 8453 diode array spectrophotometer, and reflectance spectra were acquired with a Labsphere RSA-HP-53 accessory. Raman spectra were acquired using a Jobin-Yvon Horiba Labram HR instrument coupled to an Olympus BX41 microscope. Samples were excited at 514.5 nm with an Ar-ion laser. The hole and slit widths were 500 and 150 µm, respectively; spectra were acquired with a resolution of 3 cm⁻¹ using a grating with 1800 lines/mm. Samples were prepared by placing solid powders onto glass microscope slides.

Soft and hard x-ray photoelectron spectroscopy (XPS, HAXPES) was employed to determine the valence band alignment of α-V₂O₅ and the QDs. XPS measurements were processed via a Phi VersaProbe 5000 system with a monochromated Al Kα source (1.5 keV). The mounting and handling process of the samples was performed in a glovebox, and then the samples were transported to the UHV chamber in inert atmosphere. HAXPES measurements were performed at the surface and interface structural analysis beamline (beamline I09) of the Diamond Light Source (DLS), Ltd., UK, using a photon energy of ∼6.0 keV. The HAXPES spectra are energy aligned to the Fermi level of a gold foil reference in electrical contact with samples. The gold reference scans were measured before and after each spectrum to reduce further energy alignment shift from beam drift. The measured valence band offsets were employed to align the DFT band structures of the α-V₂O₅ and QDs on a common energy scale.

Transient absorption experiments were performed using a Ti:sapphire amplified laser system (SpectraPhysics Spitfire Pro, 1 kHz repetition rate) in a standard pump-probe geometry, as described previously.³⁶ Briefly, the wavelength-tunable pump pulse (100 fs, 1 kHz) was generated with an optical parametric amplifier (Topas-C, LightConversion). The femtosecond white-light probe pulse was obtained via supercontinuum generation in a sapphire disk and was delayed mechanically. The nanosecond probe pulse was created by a second supercontinuum laser and was delayed electronically. The probe light was split into signal and reference beams, which were detected with fiber-coupled silicon (visible) diode arrays on a shot-by-shot basis.

α-V₂O₅/CdSe heterostructures (50 mg) were dispersed into aqueous solutions of lactic acid (20%), Ni(NO₃)₂ (0.1M), and 3-MPA (0.2M). A complex of nickel(ii) and 3-MPA, hereafter referred to as Ni-{3-MPA}, served as the reduction cocatalyst.³⁷ In control experiments, dispersions containing the cocatalyst and either bare V₂O₅ NWs (50 mg) or cys-CdSe QDs (4 ml) were used. Dispersions were transferred to a sealed 100-ml Pyrex flask, at ambient temperature and atmospheric pressure, and then deaerated with Ar for 30 min. Stirred samples were illuminated with a 100 W Xe arc lamp (Oriel 133 Photomax) output through a filter transmitting 400–720 nm light; the focused irradiance on the flask was 2 W cm⁻². After 1 h of illumination, a 3-ml aliquot of gas was removed from the headspace above the dispersion and analyzed using a gas chromatograph (PerkinElmer Clarus 580) with a thermal conductivity detector and Ar carrier gas.

In LAA, colloidal cys-CdSe or cys-CdS QDs were interfaced with α-V₂O₅ NWs by reacting in mixed aqueous dispersions.³⁵ (We attempted to prepare α-V₂O₅/CdTe heterostructures via LAA; however, the achievable loading of CdTe QDs on NWs remained low for this tethering strategy, and this approach was no longer pursued.) Synthesizing QDs before incorporating them into heterostructures enables control over their size and properties. Cysteinate adheres to CdE QDs through the thiolate,³⁸ whereas protonated amines adsorb to the negatively charged, hydroxylated surface of V₂O₅.^35,39,40 We thus infer that the attachment of QDs to NWs was driven by interactions of protonated amines of QD-adsorbed cysteinates with the surface of V₂O₅.³⁵ In SILAR, ethanol dispersions of V₂O₅ NWs were dispersed sequentially into solutions containing cadmium(ii) ions and the appropriate chalcogenide, resulting in the deposition of CdS, CdSe, or CdTe onto the NWs.³⁵ 

The products of LAA and SILAR were characterized by TEM, EDS, SAED, and Raman spectroscopy. TEM images revealed that the surface morphology of bare single-crystalline V₂O₅ NWs was altered significantly following LAA and SILAR (Fig. 1 and Fig. S1 in the supplementary material). In TEM images of α-V₂O₅/CdSe heterostructures, discrete and agglomerated CdSe QDs are clearly discernible as higher electron density regions on the otherwise smooth surfaces of NWs (Fig. 1). SILAR-derived α-V₂O₅/CdS and α-V₂O₅/CdTe heterostructures exhibited high loadings of QDs and significantly roughened surfaces relative to bare NWs (Fig. S1). SILAR-derived heterostructures typically exhibited rougher surfaces and higher loadings of QDs on V₂O₅ NWs than LAA-derived heterostructures.

EDS confirmed the presence of Cd and S, Se, or Te on V₂O₅ NWs (Fig. S2 in the supplementary material). Average amounts of Cd and E, relative to the V₂O₅ formula unit, were calculated from five measurements on each sample, yielding apparent molecular formulas of V₂O₅/Cd_(0.09±0.04)Se_(0.08±0.06) and V₂O₅/Cd_(0.06±0.07)S_(0.07±0.08) for LAA-derived heterostructures and V₂O₅/Cd_(0.41±0.25)Se_(0.42±0.25), V₂O₅/Cd_(0.48±0.10)S_(0.72±0.14) and V₂O₅/Cd_(0.35±0.14)Te_(1.57±0.83) for SILAR-derived heterostructures. The higher loading of QDs on V₂O₅ NWs in SILAR-derived heterostructures relative to LAA-derived heterostructures is consistent with TEM analysis. SAED analysis (Fig. S1) established that the SILAR process deposited wurtzite-phase CdE QDs onto the surfaces of the V₂O₅ NWs.

The Raman spectrum of bare NWs [Fig. 1(h)] exhibited phonon modes of α-V₂O₅ including the external displacement of (VO₅) units below 200 cm⁻¹, rocking and bending modes in the range from 200 to 500 cm⁻¹, and V–O bond-stretching modes above 500 cm⁻¹ (including the prominent vanadyl stretch near 1000 cm⁻¹), consistent with previous measurements.²⁷ Heterostructures exhibited these characteristic Raman bands of V₂O₅ as well as features attributable to CdE QDs. Spectra of LAA- and SILAR-derived V₂O₅/CdSe heterostructures contain prominent bands at 205 cm⁻¹ and 409 cm⁻¹, corresponding to the longitudinal optical (LO) mode and second-order LO (2LO) mode of bulk wurtzite CdSe, respectively [Fig. 1(d)].^41,42 The 2LO mode of CdSe overlaps substantially with a Raman band of α-V₂O₅ at 404 cm⁻¹. α-V₂O₅/CdS heterostructures exhibited intense, sharp Raman bands at 300 and 600 cm⁻¹, corresponding to the LO and 2LO modes, respectively, of bulk wurtzite CdS (Fig. S3 in the supplementary material).^43,44 Similarly, Raman spectra of α-V₂O₅/CdTe heterostructures contained bands at 163 cm⁻¹ and 327 cm⁻¹, corresponding to the LO and 2LO modes, respectively, of bulk wurtzite CdTe (Fig. S3).^45,46 Taken together, the TEM images, EDS data, SAED patterns, and Raman spectra reveal that LAA and SILAR yielded a range of α-V₂O₅/CdE heterostructures.

Normalized absorbance spectra of colloidal CdE QDs [Fig. 2(a)] highlight the substantial variation of bandgap and light-harvesting properties with the chalcogenide. Diffuse reflectance UV/vis spectra of α-V₂O₅ NWs and LAA- and SILAR-derived heterostructures are presented in Fig. 2(b) and in Fig. S4 in the supplementary material. α-V₂O₅ NWs exhibit a bandgap absorption onset of approximately 580 nm (optical bandgap = 2.13 eV) with a steep rise at shorter wavelengths, consistent with previously reported measurements.³⁶ Absorption spectra of LAA-derived α-V₂O₅/CdSe heterostructures and LAA- and SILAR-derived α-V₂O₅/CdS heterostructures are perturbed only minimally relative to the spectrum of α-V₂O₅ because the absorption onsets of cys-CdSe QDs, cys-CdS QDs, and SILAR-deposited CdS are at shorter wavelengths than that of α-V₂O₅ NWs. In contrast, the SILAR-derived α-V₂O₅/CdSe and α-V₂O₅/CdTe heterostructures exhibit broad absorption bands extending to longer wavelengths than the absorption onset of bare α-V₂O₅; thus, SILAR deposition yields QDs on V₂O₅ with larger sizes and broader size distributions than the corresponding colloidal QDs deposited via LAA.

Valence band offsets at NW/QD interfaces were determined from XPS and HAXPES measurements supported by DFT calculations. We first consider α-V₂O₅/CdSe heterostructures. DFT calculations and the HAXPES spectra are presented in Fig. 3(a). A valence band offset of 2.1 eV was determined for the α-V₂O₅/CdSe heterostructure. To interpret the convoluted HAXPES spectra of the α-V₂O₅/CdSe heterostructure, DFT calculations were performed for α-V₂O₅ and CdSe bulk structures. The valence band of α-V₂O₅ is derived mainly from O 2p states, whereas the conduction band is derived predominantly from V 3d states and has a split-off band derived from V 3d_xy.^28,47 The GGA+U calculation with U = 8 eV on Cd was used for calculating the density of states of CdSe to account for the localized nature of the Cd 4d semicore level.⁴⁸ The top of the valence band of CdSe is derived mainly from Se 4p, whereas the conduction band edge originates from mixed Cd and Se s orbitals.⁴⁹ The valence band edge energy (E_v) of α-V₂O₅ is lower than that of CdSe consistent with the more ionic nature of this compound derived from the greater electronegativity of oxygen. In this manner, the shaded differences in the bare and CdSe QD-coated α-V₂O₅ reflect the DFT of the two materials shifted by the observed valence band offset. We note that the DFT-aligned density of states agrees well with the measured occupied states from HAXPES, especially the Cd 4d semicore level and valence band edge (i.e., the shaded regions in different spectra).

DFT calculations and XPS spectra relevant to α-V₂O₅/CdS heterostructures are presented in Fig. S5 in the supplementary material. The band edge character of CdS is similar to CdSe except for a larger gap due to the substitution of Se 4s, p states for S 3s and p;⁵⁰ in addition, it shares the semicore Cd 4d which is useful for energy-alignment purposes. The value of E_v of α-V₂O₅ is lower than that of CdS QDs. The lower valence band maximum of CdS as compared to CdSe can be rationalized on the basis of Fajan’s formalisms given the greater ionicity of this compound stemming from the higher electronegativity of sulfur. The energetic offset between valence band edges at the V₂O₅/CdS interface was determined to be 1.3 eV by comparison of the XPS spectra of bare α-V₂O₅ NWs and SILAR-derived α-V₂O₅/CdS heterostructures. The XPS spectra of the valence band region were energy-aligned to the semicore Cd 4d peak at 11 eV, as previously reported for Pb_0.33V₂O₅/CdS heterostructures.⁵¹ 

Conduction band-edge energies (E_c) were estimated by adding optical bandgaps, determined from absorption onsets in the diffuse reflectance spectrum of α-V₂O₅ NWs [Fig. 2(b)] and the transmission-based absorption spectra of aqueous dispersions of cys-CdE QDs [Fig. 2(a)], to the values of E_v determined from XPS and HAXPES. We used the absorption spectra of dispersed QDs in this analysis given their well-defined first-excitonic absorption bands. Excitonic absorption onsets of SILAR-deposited QDs were red-shifted relative to those of colloidal QDs (Fig. S4 and Fig. 2); thus, interfacial conduction band-edge offsets varied slightly with the preparation method. (For CdTe, we estimated E_v from the measured E_v of CdSe QDs and the reported difference in E_v values of bulk CdSe and bulk CdTe).⁵² On the basis of these considerations, we predict type-II energetic offsets for heterostructures of α-V₂O₅ NWs and CdS, CdSe, and CdTe QDs, in which both the conduction and valence band edges of CdE QDs are higher in energy than the corresponding band edge of α-V₂O₅, as shown in Fig. 3(b).

In order to align the determined band offsets with the hydrogen- and oxygen-evolving redox potentials, we plotted the energies with respect the vacuum level in Fig. 3(b). We used the ionization potentials of α-V₂O₅ and CdSe²⁹ and the fact that the standard hydrogen electrode (SHE) potential is 4.5 eV below the vacuum level.⁵³ SHE indicates the H⁺/H₂ redox couple for an idealized solution with pH 7, and thus the reversible hydrogen electrode (RHE) should be considered in the experiment due to the inclusion of lactic acid in the electrolyte.⁵⁴ Nevertheless, the SHE can provide insight into the observed electrochemistry. The valence band-edge offsets of 1.3–2.3 eV, for CdS and CdSe, respectively, render the transfer of photogenerated holes from NWs to QDs thermodynamically favorable, whereas conduction band-edge offsets of approximately 2 eV provide a driving force for the transfer of excited electrons from QDs to NWs. Thus, charge separation is predicted to be thermodynamically favorable following the excitation of either component of any of our heterostructures. Photogenerated electrons transferred from the QD to the NW are likely to be spatially localized in the V 3d split-off conduction band state of the α-V₂O₅.

We characterized excited-state deactivation and charge transfer using transient absorption spectroscopy. In the following discussion, we focus first on α-V₂O₅/CdSe heterostructures and their isolated components. The transient absorbance (TA) spectrum of dispersed cys-CdSe QDs in aqueous solution, acquired on the nanosecond time scale [Fig. 4(a) and Fig. S6(a) in the supplementary material], exhibits a first-excitonic bleach centered at 510 nm, consistent with filling of the 1S(e) state of CdSe QDs, and a broad induced absorption feature from 550 to 900 nm, arising from the intraband excitation of photogenerated holes.^55,56 Decay traces were extracted from TA data matrices by averaging ΔA values at probe wavelengths within the bleach and induced absorption. Decay traces on nanosecond-to-microsecond time scales were fit to multiexponential decay kinetics,

where t is the time after pulsed excitation, τ_i is the ith lifetime, A_i is the amplitude of population decaying with τ_i, and IRF is the instrument response function estimated as a Gaussian. Goodness of fit was evaluated by plots of residual (data minus fit) as a function of wavelength and by values of χ²; an additional exponential component was added to a given fit when it lowered χ² and discernibly flattened the plot of residual vs wavelength. Amplitude-weighted average lifetimes (⟨τ⟩) were calculated as follows:⁵⁷ 

For dispersed cys-CdSe QDs, the bleach at 510 nm recovered and the induced absorption at 650 nm decayed with triexponential kinetics with values of ⟨τ⟩ of (260 ± 10) ns and (440 ± 40) ns, respectively [Fig. S6(b)].

The TA spectrum of α-V₂O₅ NWs, obtained as the average of spectra acquired at delay times of 1–10 ns [Figs. 4(b) and 4(d) and Fig. S7(a) in the supplementary material], exhibits an excitonic bleach centered at 430 nm, a weaker bleach centered at 500 nm, and a broad induced absorption from 510 to 900 nm with maximum at approximately 550 nm. The 500-nm bleach has been ascribed to the filling of the lower-energy split-off conduction band of α-V₂O₅ NWs.^28,47 The broad induced absorption feature is similar to the steady-state absorption of oxidized α-V₂O₅ NWs, which we assigned previously to the intraband excitation of holes to states deeper within the VB.³⁶ Decays of the 430-nm bleach [Fig. 4(e)] and the 550-nm absorption [Fig. S8(a) in the supplementary material] were modeled accurately by triexponential kinetics, yielding values of ⟨τ⟩ of (0.5 ± 0.3) µs and (1.4 ± 0.2) µs, respectively. The multiexponential decay kinetics of TA features of α-V₂O₅ NWs have been attributed to the presence of distributions of electron- and hole-trap states in the NWs.^36,58

TA spectra of α-V₂O₅ NWs acquired at delay times of 1–100 ps [Fig. 5(a) and Fig. S9(a) in the supplementary material] differ greatly from nanosecond-time scale spectra. First, the 430-nm excitonic bleach of α-V₂O₅ NWs is poorly resolved in picosecond-time scale spectra. Second, spectra acquired at delay times of several picoseconds contain a broad and intense negative-ΔA feature, from approximately 475 to 800 nm, which decays within several nanoseconds to the long-lived TA spectrum of photoexcited α-V₂O₅ NWs. The broad and short-lived negative-ΔA feature extends to wavelengths much longer than the ground-state absorption onset of α-V₂O₅ NWs (480 nm), indicating that the feature is not a bleach of the ground-state absorption. After approximately 500 ps, the 510-nm bleach and the long-wavelength absorption do not evolve further as expected given that the excited state of α-V₂O₅ NWs decays on the microsecond time scale (Fig. S7). To isolate the short-lived negative-ΔA feature, the 2.5-ns TA spectrum of α-V₂O₅ NWs was subtracted from TA spectra acquired at shorter delay times; thus-calculated spectra are hereafter referred to as ΔΔA spectra. The profile of the ΔΔA spectrum of α-V₂O₅ NWs [Fig. 5(b)] corresponds closely to the steady-state emission spectrum of the NWs [Fig. S10(a) in the supplementary material], suggesting that the initial short-lived negative-ΔA feature arose from emission. Thus, at early time scales, the TA spectrum of α-V₂O₅ NWs can be regarded as a linear combination of the TA spectrum of excited NWs and a contribution from emission.

To aid in our understanding of the short-lived emission (negative-ΔA feature) of α-V₂O₅ NWs, we acquired TA spectra of γ-V₂O₅ NWs and ζ-V₂O₅ NWs [Figs. S10(b) and S10(c) in the supplementary material]. TA spectra of γ- and ζ-V₂O₅ NWs contain induced absorption bands and ground-state bleaches but not the long-wavelength negative-ΔA feature of α-V₂O₅ NWs. Whereas α-V₂O₅ exhibits a split-off conduction band and γ- and ζ-V₂O₅ do not,^26,28,59 we infer that the observed short-lived emission from α-V₂O₅ involves the recombination of electrons in the split-off conduction band with valence-band holes. Notably, the valence band edge, split-off V 3d_xy band, and the conduction band yield a three-level system that is able to mediate population inversion, potentially amplifying emission. The emission from α-V₂O₅ NWs decayed with ⟨τ⟩ of 10–50 ps [Fig. S11(a) in the supplementary material]. Whereas the emission is much shorter-lived than the ground-state bleach, we speculate that trapping of electrons or holes quenched emission.

Nanosecond-time scale TA spectra of α-V₂O₅/CdSe heterostructures [Figs. 4(c) and 4(d) and Figs. S7(b) and S7(c)] differ greatly from those of the isolated NWs and QDs. The spectrum of SILAR-derived heterostructures consists of a broad induced absorption from 470 to 900 nm and a weak bleach at wavelengths below 470 nm. Notably absent from the TA spectrum are features attributable to photoexcited CdSe QDs or α-V₂O₅ NWs. (It is not possible to prepare isolated SILAR-derived QDs in the absence of a substrate, precluding the independent acquisition of their TA spectrum; nonetheless, because the excitonic absorption onset of SILAR-deposited CdSe QDs is red-shifted relative to that of dispersed cys-CdSe QDs, we would expect a red-shifted excitonic bleach.) The TA spectrum of a charge-separated state formed via interfacial charge transfer, with an electron in α-V₂O₅ NWs and a hole in CdSe QDs [Fig. 3(b)], should exhibit the spectral features of reduced NWs and oxidized QDs. We previously reported the ΔA spectrum of electrochemically reduced α-V₂O₅ NWs, which exhibits a well-resolved bleach centered at approximately 420 nm.³⁶ The ΔA spectrum of oxidized CdSe QDs has been reported to exhibit a broad induced absorption feature extending from the visible into the near-IR, which has been assigned to the excitation of holes.^56,60–63 Therefore, the measured TA spectrum of SILAR-derived α-V₂O₅/CdSe heterostructures [Figs. 4(c) and 4(d)] indeed exhibits features associated with electrons localized in V₂O₅ and holes localized in CdSe. That the spectrum exhibits these features, but none attributable to excited states of α-V₂O₅ NWs or CdSe QDs, provides compelling evidence for rapid and efficient interfacial charge separation. The TA spectrum in Fig. 4(c) was acquired after exciting heterostructures at 360 nm, where both V₂O₅ and CdSe absorb strongly (Fig. 2 and Fig. S4); thus, photoexcitation of either component was likely followed by interfacial charge separation, as expected and desired for these type-II interfaces (Fig. 3). TA decay traces were extracted from the induced absorption feature of SILAR-derived heterostructures (Fig. S8b). Decay traces at 650 nm followed triexponential kinetics with ⟨τ⟩ of (1.3 ± 0.6) μs.

The TA spectrum of LAA-derived α-V₂O₅/CdSe heterostructures [Fig. 4(d)] exhibits features similar to the spectra of both the isolated NWs and the SILAR-derived heterostructures. A bleach at wavelengths less than 520 nm is nearly superimposable with that of isolated α-V₂O₅ NWs. A broad absorption from 520 to 900 nm is similar to that of SILAR-derived α-V₂O₅/CdSe heterostructures and provides evidence for photogenerated holes localized in QDs. The TA spectrum of LAA-derived heterostructures thus has contributions from the charge-separated state, with electrons in V₂O₅ and holes in CdSe, and the excited state of α-V₂O₅ NWs that did not participate in charge transfer. The lower loading of hole-accepting QDs per NW for LAA-derived heterostructures, relative to SILAR-derived heterostructures, apparently decreased the yield of excited-state hole transfer from NWs to QDs. Indeed, the 2.5-ns TA spectrum of LAA-derived heterostructures was modeled accurately as a linear combination of the 2.5-ns TA spectra of SILAR-derived heterostructures (as a signature of the charge-separated state) and isolated α-V₂O₅ NWs (Fig. S12 in the supplementary material). The close alignment between the measured TA spectrum and the fit supports our interpretation of the TA spectrum of LAA-derived heterostructures as arising from charge-separated state and residual photoexcited α-V₂O₅ NWs. Whereas LAA-derived heterostructures have fewer QDs per NW, relative to SILAR-derived heterostructures (Fig. 1), it is not surprising that their TA spectrum contains features attributable to photoexcited V₂O₅.

To assess the longevity of the charge-separated state in LAA-derived heterostructures, we generated a TA decay trace at 850 nm [Fig. 4(e)], within the broad induced absorption feature of the heterostructures but beyond that of the isolated α-V₂O₅ NWs [Fig. 4(d)]. The 850-nm absorption decayed triexponentially with ⟨τ⟩ of (5.0 ± 1.1) μs. The charge-separated state for LAA-derived heterostructures was thus several-fold longer-lived than that of SILAR-derived heterostructures. The presence of cysteinate at the NW/QD interface may have slowed charge recombination.

We acquired picosecond-time scale transient absorption data to learn how fast electrons and holes were separated across NW/QD interfaces. Full TA data matrices are presented as color maps in Fig. S9, and spectra are presented in Fig. 5. TA spectra of SILAR-derived α-V₂O₅/CdSe heterostructures [Fig. 5(c)] differed significantly from those of isolated α-V₂O₅ NWs [Fig. 5(a)]. Spectra acquired at delay times of 1–5 ps exhibit a bleach from 450 to 550 nm and an absorption from 550 to 800 nm. These features decay within a nanosecond to the long-lived TA spectrum of SILAR-derived α-V₂O₅/CdSe heterostructures, assigned previously to the charge-separated state. TA spectra acquired at delay times from 3 ps to 2.5 ns contain a well-resolved isosbestic point at 605 nm with ΔA of approximately 1 m OD. TA spectra of photoexcited colloidal QDs or other chromophores necessarily contain isosbestic points, with ΔA equal to zero, at wavelengths at which the excited and ground states have identical molar absorption coefficients. However, the nonzero value of ΔA at the 605-nm isosbestic point in the TA spectra of SILAR-derived α-V₂O₅/CdSe heterostructures indicates that a third component, other than the ground and excited states, is present. This third component was formed within 3 ps and did not decay to any measurable extent within 2.5 ns.

Corresponding ΔΔA spectra [Fig. 5(d)], calculated by subtracting the TA spectrum acquired at 2.5 ns from earlier-time scale spectra, consist of a well-resolved bleach from 450 to 605 nm, an absorption from 605 to 800 nm, and an isosbestic point at 605 nm with ΔΔA of 0. (The ΔΔA spectrum extracted at 1 ps was red-shifted relative to the spectra at longer delay times). The ΔΔA spectra of SILAR-derived α-V₂O₅/CdSe heterostructures are similar to the TA spectrum of dispersed cys-CdSe QDs [Fig. S6(a)]. We thus assign the ΔΔA spectra of SILAR-derived α-V₂O₅/CdSe heterostructures to the excited state of SILAR-deposited CdSe QDs. (Equivalently, TA spectra of SILAR-derived α-V₂O₅/CdSe heterostructures have contributions from the charge-separated state and from residual excited CdSe QDs.) Notably, the value of ΔA at the 605-nm isosbestic point is unchanged in the TA spectra (or, equivalently, the isosbestic point in ΔΔA spectra remains at 0 on the y-axis), as the bleach and absorption arising from excited CdSe QDs decay. Therefore, a population of CdSe QDs within the heterostructures decays to the ground state rather than through electron transfer to V₂O₅. The loading of QDs on NWs was highest in SILAR-derived heterostructures, and a fraction of QDs was apparently not in close-enough contact with NWs to promote interfacial electron transfer.

Picosecond-time scale TA decay traces were fit to multiexponential kinetics,

We fit ΔΔA decay traces at 500 and 725 nm and extracted ⟨τ⟩ of (16.1 ± 0.9) ps and (17 ± 5) ps, respectively [Fig. S11(b)]. These values of ⟨τ⟩ are much shorter than those of dispersed cys-CdSe QDs [Fig. S6(b)], perhaps owing to the high local concentration of QDs on the NWs. Importantly, the persistence of the 605-nm isosbestic point, without any change of ΔA, from 3 ps onward, reveals that charge separation within SILAR-derived α-V₂O₅/CdSe heterostructures was complete within several picoseconds after pulsed excitation of the heterostructures.

TA spectra of SILAR-derived α-V₂O₅/CdSe heterostructures, acquired at delay times of 3 ps and longer, do not contain any spectral signature of emission from V₂O₅. In contrast, the red-shifted bleach in the 1-ps TA spectrum [Fig. 5(c)] may have arisen from emission. Indeed, the 1-ps spectrum is modeled accurately as a linear combination of spectra corresponding to emissive V₂O₅ (the ΔΔA spectrum of α-V₂O₅ NWs), excited CdSe (the ΔΔA spectrum of SILAR-derived α-V₂O₅/CdSe heterostructures at delay times greater than 3 ps), and charge-separated state (the TA spectrum of heterostructures at 2.5 ns) (Fig. S12 in the supplementary material). The close correspondence between the measured and fitted spectra suggests that α-V₂O₅ NWs were initially emissive, but that emission was quenched by rapid charge transfer. In summary, in SILAR-derived α-V₂O₅/CdSe heterostructures, charge separation occurred within approximately 3 ps to yield a charge-separated state that persisted for microseconds.

TA spectra of LAA-derived α-V₂O₅/CdSe heterostructures acquired at delay times of 1–10 ps [Fig. 5(e)] consist of a broad negative-ΔA feature throughout the visible and thus resemble the spectrum of isolated V₂O₅ NWs [Fig. 5(a)]. Spectra of LAA-derived heterostructures evolve with time to yield a spectrum similar to that measured on nanosecond time scales, except that the 430-nm bleach is less well-resolved. Picosecond-time scale ΔΔA spectra of LAA-derived α-V₂O₅/CdSe heterostructures, generated by subtracting the 2.5-ns TA spectrum from a given picosecond-time scale TA spectrum [Fig. 5(f)], exhibit the spectral signature of emission from V₂O₅ NWs [Fig. 5(b)]. This feature for LAA-derived heterostructures persists for approximately 500 ps [Fig. 5(f)], indicating that holes are transferred much more slowly in LAA-derived heterostructures than in SILAR-derived heterostructures. This difference may have arisen from the increased distance and poorer electronic coupling between CdSe QDs and V₂O₅ NWs in LAA-derived heterostructures due to the presence of cysteinate as a molecular linker. The lower loading of hole-accepting QDs per NW in the LAA-derived heterostructures may also have decreased the rate of hole transfer. The ΔΔA spectra of LAA-derived heterostructures did not exhibit a bleach attributable to photoexcited CdSe QDs [Fig. 5(f)]. The bleach in TA spectra of cadmium chalcogenide QDs is proportional to the population of excited electrons.^61,64 Thus, the absence of the bleach in ΔΔA spectra of LAA-derived heterostructures suggests that electrons were transferred from CdSe QDs to V₂O₅ NWs within the instrument response.

In summary, the α-V₂O₅/CdSe heterostructures prepared by both SILAR and LAA undergo subpicosecond electron transfer following excitation of CdSe QDs to yield long-lived charge-separated states. Photoexcitation of α-V₂O₅ NWs within SILAR-derived heterostructures is followed by the transfer of holes to CdSe QDs within several picoseconds. Interfacial hole transfer is slower in LAA-derived heterostructures, occurring over several hundred picoseconds. Likewise, interfacial charge recombination occurs more rapidly within SILAR-derived heterostructures than LAA-derived heterostructures. These differences in the rates of interfacial charge transfer probably arose from the properties of V₂O₅/CdSe interfaces. The presence of cysteinate as a molecular linker within LAA-derived heterostructures should increase the distance and decrease the electronic coupling between NWs and QDs, which apparently gave rise to measurable differences in the time scales of charge separation and recombination. The lower loading of QDs on NWs within LAA-derived heterostructures may also have slowed NW-to-QD hole transfer relative to SILAR-derived heterostructures. These differences, notwithstanding the efficient and prolonged separation of photogenerated electrons and holes, render both SILAR- and LAA-derived α-V₂O₅/CdSe heterostructures intriguing for redox photocatalysis. The viability of this approach is presented below.

Nanosecond-time scale TA spectra of SILAR-derived α-V₂O₅/CdS and α-V₂O₅/CdTe heterostructures are similar to those of α-V₂O₅/CdSe heterostructures and vary only minimally with identity of the QDs [Fig. 6(a)]. TA spectra of all SILAR-derived heterostructures exhibit a weak bleach below approximately 500 nm and a broad absorption extending from 500 to 900 nm. The absorption feature red-shifted slightly from CdS to CdSe to CdTe, but spectra were otherwise nearly identical. We therefore assign the nanosecond-time scale TA spectra of SILAR-derived α-V₂O₅/CdS and α-V₂O₅/CdTe heterostructures to charge-separated state. Charge-separated states for all SILAR-derived heterostructures persisted for microseconds and were substantially longer-lived than the bleach of photoexcited V₂O₅ NWs [Fig. 6(c)]. Decay traces within the absorptive signals were fit to biexponential or triexponential kinetics [Figs. S8(c) and S8(d)], and values of ⟨τ⟩ were on the order of 10⁻⁶ s (Table S1 in the supplementary material).

Nanosecond TA spectra of LAA-derived α-V₂O₅/CdS heterostructures contain features of both excited V₂O₅ NWs and the charge-separated state [Fig. 6(b)]. Decay traces were extracted at various wavelengths [Fig. S8(e)]; values of ⟨τ⟩ are tabulated in Table S2 in the supplementary material. Charges recombined 2 to 3 times more slowly in LAA-derived α-V₂O₅/CdS heterostructures than in the corresponding SILAR-derived heterostructures, which we attribute to the presence of cysteinate between QDs and NWs.

Picosecond-time scale TA spectra of SILAR- and LAA-derived α-V₂O₅/CdS and α-V₂O₅/CdTe heterostructures were acquired to evaluate charge-transfer dynamics (Fig. 7). TA spectra of SILAR-derived α-V₂O₅/CdTe heterostructures at the shortest delay times consist of a broad bleach; the spectrum evolves with time to that of the charge-separated state with a weak bleach centered at 470 nm and a broad absorption. At delay times of 3 ps and longer, spectra exhibit an isosbestic point at 760 nm with ΔA of approximately 0.8 mOD; the rapid formation of this isosbestic point with nonzero ΔA suggests that charge separation was complete within several picoseconds [Fig. 7(a)]. Corresponding ΔΔA spectra [Fig. 7(b)] at delay times of 3 ps and longer exhibit a broad excitonic bleach centered at 600 nm. By analogy with our interpretation of the TA spectra of α-V₂O₅/CdSe heterostructures, we attribute the bleach to excited states of CdTe QDs that decayed independently and did not participate in charge separation. The TA spectrum of SILAR-derived α-V₂O₅/CdTe heterostructures, acquired at 1 ps, was modeled accurately as a linear combination of the emission-derived feature in TA spectra of α-V₂O₅ NWs at short time scales (ΔΔA spectrum of α-V₂O₅ NWs), the TA spectrum of charge-separated state (2.5-ns TA spectrum of SILAR-derived α-V₂O₅/CdTe heterostructures), and the TA spectrum of residual CdTe QDs (3-ps ΔΔA spectrum of SILAR-derived α-V₂O₅/CdTe heterostructures) (Fig. S14 in the supplementary material). The quality of this fit suggests that emission contributed to the TA spectrum at the earliest measurable time scales and that charge separation indeed occurred within 3 ps.

Picosecond-time scale TA spectra of SILAR- and LAA-derived α-V₂O₅/CdS heterostructures are interesting [Figs. 7(c) and 7(d)]. At the earliest delay times, spectra for both samples exhibit a narrow short-wavelength bleach (centered at 430 nm for LAA- and 500 nm for SILAR-derived heterostructures) attributable to the excitonic bleach of QDs and a broad, longer-wavelength negative-ΔA feature corresponding to emission from α-V₂O₅ NWs. The excitonic CdS bleach for LAA-derived heterostructures is narrower and blue-shifted relative to SILAR-derived heterostructures, consistent with the smaller and narrower size distributions of the presynthesized QDs used in LAA.

The TA spectrum of SILAR-derived α-V₂O₅/CdS heterostructures evolved within 500 ps to that of the charge-separated state but, at shorter delay times, differed markedly from the spectra of SILAR-derived α-V₂O₅/CdSe and α-V₂O₅/CdTe heterostructures. First, for α-V₂O₅/CdS heterostructures, emission from α-V₂O₅ NWs persisted for several hundred picoseconds. Second, the spectra contain no isosbestic point. These differences indicate that interfacial charge separation occurred more slowly in SILAR-derived α-V₂O₅/CdS heterostructures than in α-V₂O₅/CdSe and α-V₂O₅/CdTe heterostructures, which may be attributable to the lower driving force for the transfer of holes from V₂O₅ to CdS (Fig. 3). We fit TA decay traces within the bleach of CdS QDs and extracted ⟨τ⟩ of (48 ± 3) ps (Fig. S15 in the supplementary material), which is vastly shorter than the approximately 160-ns excited-state liftetime of dispersed cys-CdS QDs. The accelerated bleach decay provides evidence that electrons were transferred from photoexcited CdS to V₂O₅. Both emission from α-V₂O₅ NWs and the charge-separated state give rise to broad spectral features; therefore, deconvolution of electron-transfer and hole-transfer kinetics is not trivial. Nonetheless, the persistence of the excitonic bleach of CdS QDs and emission from α-V₂O₅ NWs reveal that charge separation occurred more slowly for SILAR-derived α-V₂O₅/CdS heterostructures than for the corresponding CdSe- and CdTe-containing heterostructures.

We would expect charge separation to occur even more slowly for LAA-derived α-V₂O₅/CdS heterostructures due to the presence of cysteinate at the QD/NW interface. However, due to the overlap of the excitonic bleach of CdS QDs, emission from α-V₂O₅ NWs, and the signal from charge-separated state, we were unable to quantify charge-transfer dynamics.

We endeavored to exploit the photoinduced charge-transfer reactivity of the NW/QD heterostructures in redox photocatalysis. We focus here on α-V₂O₅/CdSe heterostructures, which underwent rapid charge separation to yield electrons in V₂O₅ and holes in CdSe. We envisioned a mechanism in which electrons in NWs would reduce solvated protons rather than recombining with photogenerated holes. Reaction mixtures consisting of α-V₂O₅/CdSe heterostructures, lactic acid (as a source of protons and a sacrificial electron donor), and Ni-{3-MPA} (as a reduction cocatalyst³⁷) were illuminated with white light, and H₂ was quantified by gas chromatography. Control experiments were performed with dispersions containing either α-V₂O₅ NWs or cys-CdSe QDs, the isolated components of heterostructures, with lactic acid and Ni-{3-MPA}. Hydrogen was evolved from dispersions containing LAA-derived α-V₂O₅/CdSe heterostructures at a rate of 6.1 ± 0.2 µmol h⁻¹, whereas hydrogen evolution was negligible (<0.05 µmol h⁻¹) from dispersions containing SILAR-derived α-V₂O₅/CdSe heterostructures (Fig. 8). Colloidal cys-CdSe QDs evolved hydrogen 20-fold more slowly (0.26 ± 0.03 µmol h⁻¹) than the LAA-derived heterostructures, and dispersions containing α-V₂O₅ NWs exhibited no measurable hydrogen evolution (Fig. 8). These results are consistent with a mechanism in which photogenerated electrons in α-V₂O₅ NWs are transferred to Ni-{3-MPA}, which subsequently reduces protons to hydrogen and in which photogenerated holes in cys-CdSe QDs oxidize lactic acid. The vastly increased rate at which hydrogen was produced by LAA-derived heterostructures, relative to isolated NWs or QDs, reveals that the separation of photogenerated electrons and holes across the NW/QD interface, which increases their lifetimes 10-to-20-fold, can indeed enable reduction and oxidation reactions to compete with electron-hole recombination. The much slower evolution of hydrogen from SILAR-derived heterostructures suggests that the more rapid electron-hole recombination at linker-free interfaces can limit the rate and yield of redox catalysis. Our α-V₂O₅/CdE heterostructures are intriguing candidates for an array of applications in redox photocatalysis.

We interfaced α-V₂O₅ NWs with CdE QDs to yield a range of new heterostructures. SILAR yielded the highest loadings of QDs on NWs but with limited control over their size and energetics. In contrast, LAA yielded lower loadings of QDs but afforded more precise control over the size and properties of presynthesized CdE QDs. The two synthetic methods are thus complementary in terms of their ability to structure interfaces between the two semiconducting components. HAXPES and XPS, together with electronic absorption spectroscopy, revealed that α-V₂O₅/CdS, α-V₂O₅/CdSe, and α-V₂O₅/CdTe heterostructures all exhibited Type-II interfacial energetics. The offsets between valence band edges, which determined the driving force for the transfer of photogenerated holes from α-V₂O₅ NWs to CdE QDs, ranged from 1.3 eV (for CdS) to 2.3 eV (for CdTe). Offsets between conduction band edges, which drove QD-to-NW electron transfer, were 1.5 eV–2 eV.

Transient absorption spectroscopy revealed that all heterostructures underwent photoinduced charge separation following pulsed excitation at wavelengths absorbed by both α-V₂O₅ NWs and CdE QDs. For SILAR-derived α-V₂O₅/CdSe and α-V₂O₅/CdTe heterostructures, charge separation was complete within 3 ps. Charges were separated more slowly within SILAR-derived α-V₂O₅/CdS heterostructures, probably owing to the decreased driving force for hole transfer from V₂O₅ to CdS. Similarly, excited-state hole transfer occurred more slowly in LAA-derived α-V₂O₅/CdSe heterostructures than in corresponding SILAR-derived heterostructures, probably due to the presence of cysteinate as a molecular linker. Lifetimes of charge-separated states ranged from approximately 1–5 µs and were longer for LAA-derived heterostructures than SILAR-derived heterostructures. The subtle dependence of the dynamics of charge separation and recombination on the composition and interconnectivity of heterostructures suggests an intriguing potential to control both light-harvesting properties and charge-transfer reactivity.

Whereas all of the heterostructures underwent ultrafast charge transfer to yield long-lived charge-separated states, we were eager to evaluate their performance as photocatalysts. Proof-of-concept experiments revealed that LAA-derived α-V₂O₅/CdSe heterostructures vastly outperformed isolated NWs and QDs in the reduction of aqueous protons to H₂.

In summary, α-V₂O₅/CdE heterostructures can be prepared with programmable compositions, interconnectivity between components, photophysical properties, and interfacial energetics. The heterostructures reported in this article undergo ultrafast charge separation to yield long-lived charge-separated states. Photoinduced charge transfer can be exploited in reductive hydrogen evolution. We therefore conclude that the heterostructures are intriguing candidates for a range of applications in light-harvesting, excited-state charge transfer, and photocatalysis. Our ongoing efforts involve exploring this potential.

See the supplementary material for Figs. S1–S15 and Tables S1 and S2. Additional TEM images, spectra (EDS, Raman, electronic absorption and emission, XPS, and TA) and fits, TA decay traces and fits, and spectral fitting parameters for QDs, NWs, and heterostructures.

L.F.J.P., D.F.W., and S.B. acknowledge support from the National Science Foundation under Grant Nos. DMREF-1627583, DMREF-1626967, and DMREF-1627197. Transient absorption measurements were performed using resources of the Center for Functional Nanomaterials, which is a U.S. DOE Office of Science Facility, at Brookhaven National Laboratory under Contract No. DE-SC0012704. We acknowledge Diamond Light Source for time on Beamline I09 under Proposal No. SI22148 SI22148 and thank Dr. Tien-Lin Lee (Diamond) and Dr. Matt Wahila (Binghamton) for assistance with the HAXPES measurements. We thank Professor David Lacy and Karthika Kadassery (Buffalo) for their guidance in the use of gas chromatography to quantify evolved hydrogen.