Semiconductor‐Metal Hybrid Nanoparticle‐Based Hydrogels: Efficient Photocatalysts for Hydrogen Evolution Reaction

In semiconductor‐metal hybrid nanoparticles, excited charge carriers can be separated efficiently by transferring the electron to the metal, because the Fermi level is located within the bandgap of the semiconductor. Besides charge carrier separation, the catalytically active surface of the metal enables the use of these charge carriers for further reactions. Due to limited colloidal stability, the application of nanoparticles in solution is challenging. To circumvent these difficulties, the destabilization can be used to build monolithic 3D (non‐ordered) gel‐like structures with retained high surface area and an ensured diffusion within the network. Here, the resulting nanoparticle‐based hydrogels of CdSe/CdS/Pt nanoparticles show photocatalytic hydrogen production rates up to 58 (mmol(H2))/(g∙h). Due to the self‐supporting network structure, colloidal stability is unnecessary, and the applicability is improved. By simply mixing semiconductor and semiconductor–metal hybrid nanoparticles before gelation, the synthesis of the gels allows the reduction of the metal content, which further tunes the photocatalyst.


Introduction
3][4] Especially in photocatalysis, the semiconductor can efficiently DOI: 10.1002/admi.202301076absorb light.Excited charge carriers can be separated by transferring the electron to the metal so that the material can provide the charge carriers for further reactions. [5,6]dSe/CdS nanorods (NRs) with CdSe as a spherical core in a rodshaped CdS shell are known to exhibit a high absorption coefficient and are able to separate photoexcited charges due to their quasi-type-II band structure.[7,8] Adding a metal domain on the tip of the CdSe/CdS nanorod by means of epitaxial growth in solution can further separate the charges.[5,6] Charge carrier lifetimes up to microseconds have been reported in these hybrid nanoparticles.[9] Synthetically, various metals like gold, nickel, silver, cobalt, or platinum can be grown on the semiconductor.[10][11][12][13] Especially, nickel and platinum decorated CdSe/CdS nanoparticles showed high efficiencies in the hydrogen evolution reaction.[8,10,[14][15][16] To optimize the applicability of such or other nanoparticles, colloidally stable dispersions can be destabilized to build up a self-supporting, highly porous 3D (non-ordered) network with good surface accessibility.[9,[17][18][19] With this approach, limits in applications due to colloidal instability can be overcome.The selfsupporting structure enables the diffusion of various media like polar or less polar solvents inside the pore system, making the gel structures highly versatile.
In the first study, we could show that the gelation by means of ligand oxidation improved the photocatalytic activity of CdSe/CdS-based hydrogels compared to the nanoparticle dispersion.Removing ligands makes the surface better accessible, and the controlled destabilization maintains a high surface area.Furthermore, the ligand removal and network formation improved charge carrier properties for photocatalytic applications. [20]n this work, platinum has been grown on CdSe/CdS nanorods, as it is well-established in the literature.The resulting hybrid nanostructures were employed as building blocks for making macroscopic self-supporting and porous networks, so-called hydrogels.Here, the hydrogen evolution reaction was used to explore the photocatalytic advantages of this new material.Not only hydrogels made of hybrid nanoparticle dispersions were investigated, but also mixtures of semiconductor and semiconductor-metal hybrid nanoparticle dispersions by which the amount of noble metal can be reduced.Figure 1 shows the pathway from CdSe/CdS semiconductor NRs to platinum-decorated CdSe/CdS/Pt nanoparticle-based hydrogels schematically.

Results and Discussion
The synthesis of CdSe/CdS nanorods is well-established in nanocrystal research and has been used in our group for years. [9,17,20]Resulting particles have good optical properties and are colloidally stable for months.Thus, all following experiments have been performed with the same batch of NRs to ensure comparability.Still, metal growth on a semiconductor surface can be more challenging due to the larger lattice mismatches between CdS and platinum. [21]An appropriate solvent, the right reaction temperature, precursor, and surfactants can ensure the heterogeneous growth on the semiconductor instead of homogeneous nucleation of the metal itself.However, this study has not found reaction conditions to completely inhibit homogeneous nucleation.Nevertheless, selective precipitation is needed to separate the platinum nanoparticles from the hybrid nanoparticle dispersion.
Since the colloidal synthesis is performed in organic media, studying hydrogen production in water requires a phase transfer.In literature, a ligand exchange using mercaptopropionic acid (MPA) is often applied to stabilize the particles in water. [3,5]In our work, there were some difficulties in stabilizing the particles in water using MPA.Alternatively, using a longer aliphatic chain, namely mercaptoundecanoic acid (MUA), we could stabilize the particles in water in our experiments following the report of Kodanek et al. [22] Nevertheless, colloidal stability over days is challenging as the particles aggregated in aqueous dispersions within a few hours to days.Gelation of the nanoparticles can prevent dense agglomerates and form a material that is stable for months or more.
In previous works, we showed that low amounts of hydrogen peroxide can be used to destabilize an aqueous nanoparticle solution. [9,23]By oxidizing thiol ligands, the controlled formation of a 3D (non-ordered) porous network can be triggered. [24,25]In this work, destabilizing CdSe/CdS/Pt hybrid nanoparticles with MUA on the surface using H 2 O 2 was not possible.To stabilize MUA capped particles in water, a slightly basic pH is needed to deprotonate the acid group.However, in alkaline conditions, platinum can catalyze the decomposition of H 2 O 2 by forming oxygen and water.This prevents the thiol ligands from oxidation and thus, inhibits the destabilization of the colloidal solution.In contrast, in acidic conditions, platinum catalyzes the decomposition of H 2 O 2 as well but forms hydroxy radicals as an intermediate. [26]These radicals themselves can oxidize thiol ligands and destabilize the nanoparticle dispersion.To stabilize the CdSe/CdS/Pt hybrid NPs in acidic conditions, a phase transfer and ligand exchange with 2-(dimethylammonium)ethanethiol (DMAET) has been applied. [22]Having the hybrid NRs in acidic aqueous solution, the controlled destabilization using H 2 O 2 via the OH-radicals works as desired.However, from a certain size of platinum domains, the phase transfer using DMAET leads to partial detachment of platinum domains.These then-separated platinum nanoparticles cannot be transferred to the aqueous solution using DMAET (Figure S1, Supporting Information).This detachment was not quantitively reproducible.Distinguishing if the size of the platinum domains or the detachment had an influence on the photocatalytic results was not possible.Thus, the investigation of the optimal reaction time was not possible.Additionally, platinum domains on every nanorod are not crucial in our case, because in the hydrogel, we consider all nanorods.The hydrogel then has a certain platinum amount, which we have as a parameter.For results where the platinum amount was crucial, we used the same batch of hybrid nanoparticles in aqueous dispersion for the hydrogel formation to ensure comparability.With this, the detachment is not expected to have an influence on the main message of this work.Figure 2A-C shows TEM images of the hybrid nanoparticles in aqueous solution and nanoparticlebased gels made of this aqueous nanoparticle dispersion (more TEM images in Figure S2, Supporting Information).Here, it can be seen that not every semiconductor nanorod has a metal tip in the aqueous dispersion.STEM images of the resulting hydrogels clearly show the metal domains in the network.These TEM images show the 3D structure with a 2D projection.SEM images can visualize the three dimensionality of these structures.The microscopic structure of these porous gels is always very similar as further demonstrated in several publications in our group. [9,17]gure 3. A) Hydrogen production rates of hydrogels depending on the platinum domain growth reaction time.B) Hydrogen production rates of hydrogels made of mixtures of CdSe/CdS nanorod and CdSe/CdS/Pt nanorod dispersions (6 min reaction time) with relative amount of 0%, 25%, 50%, 75%, 100% hybrid nanorod dispersions.C) Hydrogen production rates, relative to the platinum amount, of hydrogels made of mixtures of CdSe/CdS nanorods and CdSe/CdS/Pt nanorods with relative amount of 25%, 50%, 75%, and 100% of hybrid nanoparticle dispersions.The corresponding platinum weight% in the catalysts for 25%, 50%, 75%, 100% is 0.64%, 1.29%, 1.93%, and 2.57%, respectively.
Thus, to visualize the 3D (non-ordered) structure and high porosity of the gel structures, we refer to publications of our group with SEM images of gold decorated CdSe/CdS nanorod gels after supercritical drying. [12]he optical properties change from nanoparticles (NP) in organic dispersion over NP in aqueous dispersion to NP-based hydrogels.When a metal is added to the semiconductor, a broad absorption of the metal is observed (Figure 2D).The photoluminescence quantum yield (PLQY) decreases (Table S1, Supporting Information) but the spectral position of the PL does not change (Figure 2D; Figure S3, Supporting Information).Major changes were found regarding PL lifetimes (Table S1, Supporting Information) and the radiative and non-radiative recombination rates which were calculated using the PLQYs and PL lifetimes according to Equations ( 1)-( 4) and are shown in Figure 2E.
Both the phase transfer (investigated using pure semiconductor NRs) and the metal growth lead to an increase in the nonradiative recombination rates (k NR ) and a decrease in the radiative recombination (k R ) rates.The changes due to the phase transfer can be explained by an increased amount of hole trap states.Thiol ligands are known to efficiently trap holes on the surface.For radiative recombination, the hole has to be located in the CdSe core. [5]Thus, hole trapping increases the probability of non-radiative recombination while the probability of radiative recombination is decreased.The metal growth leads to electron transfer from the semiconductor (SC) to the metal domain, which competes with the radiative recombination and introduces an additional non-radiative recombination pathway.The phase transfer of NPs after metal growth combines the addition of hole trap states and the electron transfer to the metal.Colloidal destabilization and gel formation of pure semiconductor NRs leads to slightly reduced PLQY compared to the colloidal NP counterparts but the PL lifetime increases as it has been shown and explained in previous publications. [9,20]Due to the crystal contact between the nanoparticles, electrons can be delocalized beyond one nanoparticle building block.Additionally, oxidizing the thiol ligands minimizes the amount of hole trap states and thus improves hole transfer to the CdSe core.This leads to a decrease in radiative and non-radiative recombination rates.Platinum-decorated NR-based hydrogels show lower PLQY and PL lifetime.In our case, this leads to increased non-radiative recombination rates and decreased radiative recombination rates compared to the platinum-decorated NRs in an aqueous solution.These changes indicated a more efficient electron transfer to the platinum domains in the nanoparticle network in the hydrogel compared to the colloidal ensemble.As shown in the transmission electron microscopy (TEM) images, not all SC nanorods have contact with a platinum domain in the aqueous solution.Therefore, some SC NRs would show no changed optical properties.However, in the hydrogel electrons from SC NR without direct contact with platinum can reach a platinum domain by traveling through the SC network (Figure 4C).This was shown in a previous publication with gold nanoparticles. [23]o investigate the photocatalytic properties of CdSe/CdS/Pt NR-based hydrogels, photocatalytic measurements have been performed in 1 mm thin glass cuvettes (Figure S4, Supporting Information).This ensures the excitation of the whole material under illumination with a 300 W xenon lamp equipped with an AM 1.5G filter at an intensity of 1 sun (100 mW cm −2 ).Since the gel structures are self-supporting porous materials with high surface area, diffusion of the hole scavenger and the reactants throughout the catalytic measurements within the gels is possible.Therefore, all measurements were carried out without stirring.The photocatalytic mechanism is shown in Figure S10 (Supporting Information).Excited electrons are transferred to the lower Fermi level of the platinum domain where the hydrogen production is catalyzed on the platinum surface.Ideally, photoexcited holes are transferred to the CdSe core due to the higher valance band of CdSe compared to CdS.A hole scavenger then reduces the hole to reach charge neutrality. [16]he benefit of using NP-based gel structures is that the problem of uncontrolled aggregation of particle colloids in the hole scavenger solution during the photocatalytic measurement can be avoided (Figure S5, Supporting Information).Photoexcited charge carriers cannot only be used for the catalytic reaction but also decompose the stabilizing ligands leading to agglomeration.
For the efficient hydrogen production reaction and to prevent material degradation, a hole scavenger was utilized, fulfilling a complete redox reaction. [27]Full water splitting would be an ideal redox reaction but is not possible with CdSe/CdS/Pt NPs as photocatalysts.The hydrogen evolution reaction (HER) can be catalyzed efficiently by the platinum [3,28] whereas the oxygen evolution reaction (OER) cannot take place on the CdS surface. [29]herefore, the choice of the right hole scavenger is crucial for high efficiencies as well as to prevent photo corrosion of the semiconducting materials. [16]Efficient charge carrier separation and long charge carrier lifetimes improve the catalytic activity but also give the possibility for charge accumulation, which increases the risk of photo corrosion.Here, 0.1 m sodium sulfite/sodium sulfide (Na 2 SO 3 /Na 2 S) solution, isopropanol (IPA) at pH 7, and isopropanol at pH 14 have been investigated.0.1 m (Na 2 SO 3 /Na 2 S) solution showed good hole scavenging properties in previous projects, whereas isopropanol at pH 7 and 14 are known in literature to be efficient hole scavengers.The electrochemical potential of OH-is pH dependent.At pH 14, OH-can be oxidized by holes from CdSe forming hydroxy radicals.These highly reactive radicals can act as redox shuttles to oxidize the alcohol in the hole-scavenging solution. [30]First the influence of the different hole scavenger solutions on the hydrogen production rate has been investigated.The 10 V% IPA solution at pH 14 was the most efficient while the 0.1 m (Na 2 SO 3 /Na 2 S) solution was the least efficient (Figure S6, Supporting Information).Because of these results, all further experiments were carried out using 10 V% IPA in water at pH 14 as hole scavenger solution.
Six different reaction times have been investigated to get insights into the influence of the reaction time of the platinum growth on the photocatalytic properties of the resulting hydrogels.In general, the longer the reaction time of the platinum growth is, the larger the platinum domains and the lower the probability of finding nanorods without a platinum domain.Performing the ligand exchange and phase transfer, however, leads to partial detachment or dissolution of the platinum domains as shortly discussed above.Hybrid NRs with small platinum domains can be phase transferred without any observed detachment while hybrid NRs with larger platinum domains showed detachment of the platinum domains (Figure S1, Supporting Information).This being said, investigating the influence of the domain size was not feasible since the size and the amount of platinum could not be controlled to the anticipated degree during the phase transfer.Nevertheless, the results of the photocatalytic measurements of hydrogels with hybrid NRs from different reaction times showed interesting results (Figure 3A).From two to three minutes of reaction time, a significant increase in the hydrogen production rate has been observed.For these small domain sizes (up to 3 min reaction time), no detachment during the phase transfer has been observed which makes the comparison of one, two, and three minutes adequate.From four minutes on, detachment during the phase transfer has been observed.Interestingly, the detachment after five minutes of reaction time was significantly more than after 4 min (Figure S1, Supporting Information) affecting the hydrogen production rate negatively.Regardless the hydrogels synthesized from NRs, that underwent six minutes of platinum domain growth showed the highest hydrogen production rates so that further experiments have been performed using particles with 6-min reaction time of platinum growth.
A benefit of nanoparticle-based gel structures is the ability to combine different materials by simple mixing and cogelation. [23,31]This way, it is possible to achieve crystal contact between the different materials and circumvent highly com-plex chemical synthesis routes.Here, this can be used to reduce the noble metal amount by simply mixing hybrid NRs and pure semiconductor NRs before gelation.The influence of the amount of platinum on the photocatalytic activity has been investigated by using certain ratios of the two components.Reducing the amount of the noble metal without a drastic decrease in the catalytic activity would be desirable due to cost and abundance of rear earth metals.Hydrogels with 0% (pure semiconductor), 25%, 50%, 75%, and 100% of the hybrid NRs were synthesized.The platinum weight% in the overall catalysts for 25%, 50%, 75%, and 100% of hybrid nanorods is 0.64%, 1.29%, 1.93%, and 2.57%, respectively.TEM images show the reduced number of platinum domains when comparing 25% and 100% as expected (Figure S7, Supporting Information).Optical properties also depend on the ratio of the hybrid NRs.With increasing platinum amounts within the hydrogel, the PLQY decreases and the PL lifetime shortens (Table S1, Supporting Information).As discussed above, combining the PLQY and PL lifetime can give information on the radiative and non-radiative recombination rates.For 0%-to 100%-hydrogels the non-radiative recombination rate increases with increasing amount of platinum.In contrast, the radiative recombination rates are very similar for all mixtures and the pure semiconducting gel, as illustrated in Figure 4A.Furthermore, the amount of absorbed light increases with increasing amount of platinum.
Measuring the photocatalytic hydrogen production using 10 V% isopropanol at pH 14 as the hole scavenger, gave the following results: From 0% to 75% hybrid NRs the hydrogen production rate increases with increasing amount of platinum in the hydrogel.However, the hydrogen production rate then decreases slightly having a hydrogel based on 100% hybrid NRs (Figure 3B; Figure S8, Supporting Information).Combining these results with the trend of the non-radiative recombination rates indicates that the electron transfer to the metal is not the limiting process.More probable the bottleneck for the hydrogen production can be the diffusion within the gel structure or the catalytic reaction itself since it is a two-electron process.Diffusion in the network takes place but having these high hydrogen production rates without stirring it would be reasonable to reach diffusion limits.Reducing the amount of hybrid nanorods and still reaching higher hydrogen production rates is a promising result.As we showed in previous works, charge carriers can be transported through gel structures.In pure semiconducting CdSe/CdS gels, the hole is transferred to the CdSe due to the quasi-type-II band structure and the electron can be delocalized beyond one nanoparticle building block. [9,20]In semiconductormetal hybrid gel structures, electrons can travel over several NRs to one metal domain as shown by Rosebrock et al. for gold domains in CdSe/CdS NR based gels. [23]Here, mixing pure semiconductor NR with hybrid NR might show similar effects.Electrons from more than one semiconductor NR reach a platinum domain which could lead to slight electron accumulation in the platinum domain as it is known for the combination of semiconductors and metals on the nanoscale (Figure 4C,D). [32,33]aving multiple electrons in the same platinum domain would be beneficial for the hydrogen production since it is a twoelectron process.Further reducing the platinum amount by 50% and 25% leads to similar or only slightly lower hydrogen production rates compared to 100% hybrid NR-based hydrogels, respectively.
Considering that platinum is about 1000 times more expensive than cadmium, the amount of platinum in the catalyst is crucial.To shed light on the hydrogen production rate per platinum, the amount of platinum has been measured using SEM EDX (Table S2, Supporting Information).Figure 3C shows the rates per gram of platinum and shows that in this regard the most efficient mixture is 50% of the hybrid NRs.Even with 25% the hydrogen production rate is only slightly lower.This further indicates the benefits of the hydrogel formation of hybrid nanoparticles.In the hydrogels, the amount of platinum can be controlled easily.Due to the network formation, also with small amounts of platinum the electron transfer from the semiconductor NRs to the metal domains remains possible.This is the crucial difference to colloidal dispersions.When mixing semiconductor and hybrid nanorods in a dispersion, an electron transfer from a semiconductor NR to a metal domain of a hybrid NR is not feasible.

Conclusion
Semiconductor-metal hybrid nanoparticle colloids with CdSe/CdS nanorods as semiconductor and platinum domains as metal can be phase transferred and destabilized in a controlled manner to build up nanoparticle-based hydrogels.The phase transfer of these hybrid nanoparticles from organic to aqueous media leads to partial detachment of metal domains.Metal growth and hydrogel formation increase non-radiative recombination rates and decrease radiative recombination rates.Photocatalytic hydrogen production rates increase significantly by adding platinum to the semiconductor.Furthermore, by mixing semiconductor nanorods with hybrid nanorods, the platinum amount can be reduced easily while even increasing the hydrogen production rate by lowering the platinum amount.
Four milliliters of 0.5 m DMAET solution and 2 mL of water (pH 4.8 with HCl) were added.The mixture was shaken overnight.The chloroform phase was removed using a pasteur pipette.The phase with the particles was centrifuged for 3 min (≈4000 rcf).Particles had been redispersed in 1 mL of water (pH 4.8 with HCl).Washing was possible with a centrifuge filter (100.000MWCO).
Assembly to Hydrogels and Transfer to Aerogels: To induce the destabilization of ligand-stabilized nanoparticles through the oxidation of the ligands using hydrogen peroxide (H 2 O 2 ), the aqueous nanoparticle solution, containing cadmium (Cd) at a concentration of 3.6 grams per liter as determined by Atomic Absorption Spectroscopy (AAS), was combined with an H 2 O 2 solution (0.35 wt%, 93.75 microliters per milliliter of nanoparticle solution).This mixture was placed in either an Eppendorf tube or photocatalytic glass cells, followed by vigorous shaking.Subsequently, the solution was left undisturbed in darkness overnight.
After the formation of the hydrogel, a total of ten sequential washing steps were carried out over the course of three days using Millipore water.Each washing step involved the removal of 3 mL of the supernatant located above the hydrogel, accomplished using a pipette.Following the removal, 3 mL of fresh Millipore water were meticulously added using a pipette, with the aim of eliminating any residual destabilizing agent and other associated by-products.The procedure is described in more detail here. [9]hotocatalysis Measurements: The photocatalysis experiments were conducted within Duran glass cells, as illustrated in Figure S4 (Supporting Information).The hydrogels were synthesized inside these cells, following the previously described procedure.Prior to exposure to a 300 W xenon lamp (QuantumDesign LSH302) fitted with an AM 1.5 G filter (Quan-tumDesign solar simulator LS0308), the hydrogels underwent five washes with the designated hole scavenger solution.For the assessment of hydrogen production, 100 μL aliquots were withdrawn using a gas syringe and subsequently analyzed using a GC (Agilent GC 8860) equipped with an Agilent HP-PLOT Mole sieve column measuring 30 m × 0.320 mm × 25 μm, coupled with a thermal conductivity detector (TCD).The hydrogen production rate was computed using the following formula: H 2 production rate = n H 2 m Cd t (6) here, A represents the area under the hydrogen or nitrogen peak in the GC chromatogram, and the GC Factor is the ratio of the thermal conductivity detector's sensitivity for hydrogen to that for nitrogen, which was determined through calibration experiments. [20]ptical Characterization: Absorption spectra of the nanoparticle solutions and hydrogels were measured in a central position using an Agilent CARY 5000, which was equipped with a Diffuse Reflectance Accessory.PLQY were measured with a Quanta- integrating sphere coupled to a Horiba Dual-FL spectrofluorometer.Photoluminescence emission and lifetime measurements were conducted in an Edinburgh Instruments FLS 1000 spectrofluorometer with an Edinburgh Instruments EPL 445 pulsed laser (445.1 nm) for excitation of time-resolved measurements.Measurements of nanoparticle solutions, as well as hydrogels, were performed in Hellma Analytic quartz cuvettes.
Structural Characterization: TEM measurements were performed with an FEI Tecnai G2 F20 TMP (operated at 200 kV).Samples were prepared on copper grids which were covered with a carbon layer (Quantifoil).For preparation hydrogels and colloidal nanocrystal solutions were dropped on the grid.After drying the grids were ready for the measurement in the TEM.

Figure 1 .
Figure 1.Schematic pathway from CdSe/CdS semiconductor nanorods to metal-decorated CdSe/CdS/Pt nanoparticle-based hydrogels, which are used for the photocatalytic hydrogen production.

Figure 2 .
Figure 2. A) TEM measurement of CdSe/CdS/Pt nanoparticles in aqueous solution.Spots with darker contrast show platinum domains B,C) STEM measurements of a CdSe/CdS/Pt hydrogel.Brighter spots show platinum domains.D) Absorption and emission spectra of CdSe/CdS and CdSe/CdS/Pt hydrogels.E) Radiative and non-radiative recombination rates of CdSe/CdS and CdSe/CdS/Pt colloids and hydrogels.Calculated using the PLQY and PL lifetimes using Equations (1)-(4).

Figure 4 .
Figure 4. A) Radiative and non-radiative recombination rates of colloids and hydrogels.Including mixtures of CdSe/CdS nanorods and CdSe/CdS/Pt nanorods with relative amounts of 0%, 25%, 50%, 75%, and 100% of hybrid nanorods.B) PLQY and average PL lifetimes of colloids and hydrogels.Including mixtures of CdSe/CdS nanorods and CdSe/CdS/Pt nanorods with relative amounts of 0%, 25%, 50%, 75%, and 100% of hybrid nanorods.C,D) Schematic illustration of electron transfer to a metal domain.C) low amount of metal domains.D) higher amount of metal domains.(C) and (D)do not represent the actual platinum amount in the synthesized hydrogels.These schemes shall rather illustrate the difference between higher and lower platinum content.