Hydrogen Production from Glycerol Photoreforming on TiO₂/HKUST-1 Composites: Effect of Preparation Method

Abstract

Coupling metal-organic frameworks (MOFs) with inorganic semiconductors has been successfully tested in a variety of photocatalytic reactions. In this work we present the synthesis of TiO₂/HKUST-1 composites by grinding, solvothermal, and chemical methods, using different TiO₂ loadings. These composites were used as photocatalysts for hydrogen production by the photoreforming of a glycerol-water mixture under simulated solar light. Several characterization techniques were employed, including X-ray diffraction (XRD), UV-Vis diffuse reflectance spectroscopy (DRS), infrared spectroscopy (FTIR), and time-resolved microwave conductivity (TRMC). A synergetic effect was observed with all TiO₂/HKUST-1 composites (mass ratio TiO₂/MOF 1:1), which presented higher photocatalytic activity than that of individual components. These results were explained in terms of an inhibition of the charge carrier (hole-electron) recombination reaction after photoexcitation, favoring the electron transfer from TiO₂ to the MOF and creating reversible Cu¹⁺/Cu⁰ entities useful for hydrogen production.

1. Introduction

Nowadays, one of the most important necessities of society is the use of natural renewable resources to produce energy, minimizing the use of fossil fuels and reducing the associated harmful pollution produced by their combustion. On the other hand, hydrogen is considered a good candidate as a green energy carrier because it produces null pollution during its combustion, and it can be obtained from renewable sources [1,2,3]. The use of hydrogen as an energy carrier has several benefits, such as the many different storage possibilities, its ability to be converted to other energy forms with ease and to be produced from water with near-zero emissions, and its high conversion efficiency [2]. However, there are also severe limitations for the widespread use of hydrogen, for example, as a fuel for transportation. If we are planning to use hydrogen-combustion and hydrogen-fuel-cell vehicles in the future, we must first resolve outstanding issues, such as the efficient and safe storage of hydrogen, creating a fueling infrastructure, and reducing its production costs [4]. Certainly, one possibility to reduce the production cost of hydrogen is the use of green energy sources. In this sense, hydrogen production using solar energy can be categorized as: (a) thermal, (b) photovoltaic, (c) bio-photolysis, and (c) photo-electrochemical [5]. Although most of the production methods involve renewable sources, they are not well understood, and their development implies an increase in production costs and low global efficiency [2,3,4,5].

Photocatalytic hydrogen generation can be obtained mostly by two different approaches: (1) photocatalytic water splitting and (2) photocatalytic reforming of organics [6]. The first method relates to the capability of water to be reduced and oxidized by reacting with photogenerated electrons and positive holes, during semiconductor irradiation, in the presence of selected co-catalysts. The second approach is based on the ability of some organic species—namely, sacrificial agents—to donate electrons to the positive holes of the illuminated photocatalyst and be oxidized, generating proton ions, while photogenerated electrons reduce the latter to produce hydrogen in the presence of proper co-catalysts.

Glycerol is a sustainable compound that can be used for hydrogen production by photocatalytic reactions (photoreforming). Although this reaction has been studied extensively, the overall performance towards hydrogen evolution is low, and in many cases, a high photocatalytic activity is only achieved with UV-light irradiation. For this reason, the search for new materials, active and stable in the presence of sunlight, is of great interest [7,8,9].

Metal organic frameworks (MOFs) are obtained by the self-assembly of metal ions and organic ligands through the formation of covalent bonds or the presence of inter-molecular forces between them [10]. MOFs present a long-range periodic structure with good crystallinity, and MOF-based structures take some unique properties of both organic and inorganic porous materials. They exhibit several advantages such as a high surface area, tunable pore size, easy preparation, flexibility, and structural diversity [11]. CuMOF, also known as HKUST-1, copper-benzene-1,3,5-tricarboxylate (Cu-BTC), MOF-199 or Basolite^® C300, was first assembled by Chui et al. [12] through the formation of coordination bonds between trimesic acid (H₃BTC) and Cu ions [13].

MOFs have been investigated in many fields, such as sensing, drug delivery, sequestration, separation, molecular transport, electronics, bioreactors, optics, energy production, and catalysis, among others [14]. Applications in photocatalysis have been reported in the last decade, and since then, several articles and reviews have been published focusing on artificial photosynthesis (i.e., water splitting and CO₂ photoreduction) [15,16], organic photosynthesis [17], and pollutants degradation [18,19].

Specifically, in solar-driven hydrogen evolution with the presence of a sacrificial electron donor, e.g., alcohols, most of the MOFs cannot be used as a stable and efficient photocatalyst for this application individually [20]. Certain modifications of the pristine MOF, including the decoration of the organic linker or metal center, combination with semiconductors, metal nanoparticles loading, decoration with reduced graphene oxide, sensitization, pyrolyzation, and incorporation with other functional materials, have been tested to increase their activity and stability under visible light [19,21].

Hybrid nanocomposites of semiconductors with MOFs have attracted increased attention because they improve charge transfer mechanisms with a lower charge recombination and more efficient light harvesting [22]. Hybrid nanocomposites based on TiO₂ and HKUST-1 are exciting materials which could show synergic effects enhancing photocatalytic activity under visible light. Only a few investigations have reported the synthesis, structure, and properties (i.e., as photocatalysts in hydrogen production) of TiO₂/HKUST-1 nanocomposites [23,24,25,26]. Particularly, it has been reported that in these nanocomposites, HKUST-1 is transformed to Cu-Cu₂O nanoparticles after calcination at 400 °C, presenting better rates of hydrogen production in comparison with Cu deposited on TiO₂ by conventional methods [24]. There are contradictory results concerning the stability of HKUST-1. For example, when this MOF was used in aqueous media, it decomposed after 24 h of reaction [27]. However, TiO₂/HKUST-1 composites synthesized using ionic liquids as solvents showed high activity and stability during photo-oxidation/photoreduction reactions [28].

In this context, it would be very useful to know the role played by TiO₂-HKUST interactions on the activity and stability in glycerol photoreforming. Therefore, TiO₂/HKUST-1 composites were synthesized by employing three methods: the first composite was prepared by grinding the commercial reagents, Aeroxide^® TiO₂ P25 (Evonik, P25), and HKUST-1 (Basolite^® C300). These composites were designated as TiO₂ P25/com-HKUST-1. The second one was formed by TiO₂ prepared by a solvothermal route in the presence of the commercial HKUST-1 (TiO₂-ST/com-HKUST-1), and the third composite was prepared by synthesizing HKUST-1 by a chemical route in the presence of TiO₂ P25 (TiO₂ P25/syn-HKUST-1). Furthermore, the aim of the present work was focused on the effect of the preparation method of TiO₂/HKUST-1 composites, as well as the mass ratio TiO₂:MOF employed, on their photocatalytic properties for hydrogen production, using glycerol as a sacrificial agent.

2. Results

2.1. Photocatalytic Hydrogen Evolution

Due to the lack of studies regarding the effect of the optimal amount of TiO₂ that can be deposited on the HKUST-1, Figure 1 shows the photocatalytic hydrogen evolution rates as a function of TiO₂ content. All the experiments were conducted under similar operating conditions, and the H₂ production rate was estimated after 8 h of irradiation time. As can be observed, the results demonstrate a synergic photocatalytic activity between HKUST-1 and TiO₂, and the best performance corresponds to the composites with 50 wt % TiO₂. In the case of the catalyst prepared by grinding (50TiO₂ P25/com-HKUST-1) the production rate was 2.9 mmol × g⁻¹ × h⁻¹, 2.4 mmol × g⁻¹ × h⁻¹ for the 50TiO₂-ST/com-HKUST-1, and 4.5 mmol × g⁻¹ × h⁻¹ for the 50TiO₂ P25/syn-HKUST-1. Note that the photoactivity of the synthesized HKUST-1 and commercial HKUST-1 was insignificant, and as a comparison, the production rate shown by TiO₂ P25 was 1.1 mmol × g⁻¹ × h⁻¹.

It is very important to highlight that during the reaction, there was a change in color in the photocatalysts from light blue (original composite color) to reddish brown (spent composite, see Figure 2b), which is indicative that the Cu²⁺ originally present in the HKUST-1 was partially reduced towards Cu¹⁺ or Cu⁰ [28,29]. This observation suggests that HKUST-1 assembled with Cu ions and benzene 1,3,5-tricarboxylate ligands (Cu-BTC) is an unstable material when irradiated in an aqueous medium, and functions as a precursor of Cu reduced species interacting with TiO₂, as co-catalysts in the production of hydrogen. Note that the highest amounts of CO₂ and CH₄ were obtained with the 50TiO₂ P25/syn-HKUST-1 composite, which comes from the photocatalytic oxidation of an aqueous solution of glycerol. Indeed, it has been proposed that a secondary alcohol photoreforming can produce methane via β-hydride elimination, which could explain the origin of produced methane [30]. Figure 1d compares the amount of hydrogen produced as a function of the irradiation time for the prepared 50TiO₂/HKUST-1 composites. Hydrogen production followed almost the same trend with the three photocatalysts. However, a higher photoactivity was observed with the 50TiO₂ P25/syn-HKUST-1 composite.

It is important to point out that only the 50TiO₂ P25/syn-HKUST-1 composite presented long-term activity, and it was evaluated in five cycles, under simulated solar light. Figure 2a shows the H₂ production rate reached during 8 h of reaction time in each cycle. It can be seen from the second cycle that the composite shows a reduction on the production rate, and in the fifth cycle the observed reduction was ca. 50% of the production rate observed in the first run. An explanation for this unfavorable behavior can be given in terms of a partial reduction of Cu²⁺ contained in the original HKUST-1 by the photogenerated electrons in the TiO₂ conduction band, which was clearly demonstrated by the color change of the original HKUST-1 from light blue (original composite) to reddish brown (spent composite), as shown in Figure 2b. Furthermore, the zone attributed to the d-d spin allowed the transition of the Cu²⁺ between 500–800 nm (discussed later), which was modified to a reddish-brown color, as is characteristic of Cu reduced species in the HKUST-1 structure [31]. Interestingly, after each reaction cycle and subsequent washing and purging of the reaction cell, the solid returned to the original light blue color of the composite. This means that the HKUST-1 structure was not completely destroyed, otherwise it would be forming the Cu¹⁺-Cu²⁺ MOF meta-stable phase with photocatalytic activity to reduce protons to hydrogen. This behavior was previously reported in other applications of HKUST-1 [28,29,31,32] however, this is the first time that it has been observed in the photocatalytic hydrogen evolution reaction, which requires a systematic and thorough study.

2.2. Characterization

Figure 3 shows the Fourier-transform infrared (FT-IR) spectra of com-HKUST-1, syn-HKUST-1, and TiO₂/HKUST-1 composites. Clearly, the com-HKUST-1 and syn-HKUST-1 spectra are quite similar to those reported in previous works [33,34,35], which indicates that the method employed for the synthesis of syn-HKUST-1 was effective. In these spectra, several signals appeared in the range from 1300 to 1500 cm⁻¹ and from 1500 to 1700 cm⁻¹, which are associated with the interactions between the carboxylate anion—in symmetric and asymmetric modes respectively—with the metal ion [34,35]. The signals indicated at 1110, 765, and 740 cm⁻¹ are associated with the C-H vibration modes in the aromatic ring [34]. The band at 1060 cm⁻¹ is attributed to the presence of copper coordinated N,N-dimethylformamide (DMF) molecules [33], and the band centered at 507 cm⁻¹ is assigned to the Cu-O stretching mode [35]. On the other hand, all the bands mentioned previously appeared in the spectrum of the 50TiO₂ P25/com-HKUST-1 composite, indicating a weak interaction between TiO₂ P25 and com-HKUST-1. In the case of the 50TiO₂-ST/com-HKUST-1 and 50TiO₂ P25/syn-HKUST-1, the bands between 1300 and 1700 cm⁻¹ were less defined, and the signals at 507, 740, and 765 cm⁻¹ were replaced by a broad band (500–900 cm⁻¹) in the case of TiO₂-ST/com-HKUST-1 and two bands (500–700 and 765–830 cm⁻¹) in the spectrum of 50TiO₂ P25/syn-HKUST-1. These last results could indicate that there is a chemical interaction between TiO₂ and HKUST-1 when either component is obtained by a chemical route.

The optical properties of HKUST-1 and the TiO₂/Cu MOFs composites were investigated by UV-Vis diffuse reflectance spectroscopy (DRS). As can be seen in Figure 4a, the as-prepared HKUST-1 sample and the commercial HKUST-1 showed a similar spectrum, exhibiting two characteristic absorption bands centered at 300 and 700 nm, similar values to those reported in the literature [36]. Note that one shoulder can also be detected at 375 nm. The first band located in the UV region is assigned to π-π* transitions of the ligands and the band in the visible zone is attributed to the d-d spin and allowed transition of the Cu²⁺ [37]. The shoulder at 375 nm is ascribed to the ligand-to-metal charge transfer (LMCT), and the additional broad absorption band between 500 and 800 nm is assigned to the d-d spin and allowed transition of the Cu²⁺ (d⁹) ions [37].

Figure 4b corresponds to the UV-Vis DRS spectra of the TiO₂/HKUST-1 composites. In general, all composite photocatalysts showed similar absorption behavior to the pristine HKUST-1. However, it is worth noting a slight change of their absorption edge to the UV zone (350–400 nm), compared to those of commercial and synthesized HKUST-1 (Figure 4a), which can be related to the TiO₂-HKUST-1 interaction. On the other hand, the slight differences in the 50TiO₂ ST/com-HKUST-1 composite spectrum (e.g., a lower absorption in the visible region) could be related to a shielding effect by TiO₂, partially inhibiting the visible light absorption of the Cu²⁺ ions in the HKUST-1 structure, because TiO₂, in this particular composite, was grown in intimate contact with the commercial HKUST-1.

Figure 5a presents the X-ray diffraction (XRD) pattern of syn-HKUST-1, which is quite similar to the pattern of com-HKUST-1, showing the main reflections peaks at 11.6°, 13.4°, 17.4° and 19° [38,39]. These results prove that the HKUST-1 structure was successfully obtained using our described preparation method. XRD patterns of 50TiO₂ P25/com-HKUST-1, 50TiO₂ ST/com-HKUST-1, and 50TiO₂ P25/syn-HKUST-1 composites are shown in Figure 5b. These three samples displayed the same reflections described earlier, indicating that the HKUST-1 structure was preserved despite the preparation method used. Note that a higher crystallinity is observed in samples 50TiO₂ P25/com-HKUST-1 and 50TiO₂ P25/syn-HKUST-1 in comparison with sample 50TiO₂ ST/com-HKUST-1, which means that a poor crystallization of TiO₂ occurred due to the low synthesis temperature (i.e., 100 °C) compared to that reported in the literature, above 150 °C under solvothermal process [40].

Figure 6a compares the time-resolved microwave conductivity (TRMC) profiles of TiO₂ P25, syn-HKUST-1, and com-HKUST-1, obtained under a wavelength excitation of 355 nm. The highest signal was exhibited by TiO₂ P25, which also presented a long-time decay. It is worth noting that syn-HKUST-1 showed a TRMC signal because it behaves like a semiconductor material; however, it decays faster than TiO₂ P25, revealing a short lifetime of photogenerated electrons. Surprisingly, com-HKUST-1, which presented very similar structure and light absorption (see Figure 4a) to those of syn-HKUST-1, displayed a much lower TRMC signal, revealing a great difficulty in executing the charge separation after irradiation with UV light of 355 nm.

On the other hand, by analyzing the TRMC signals of the composites in Figure 6b, the sample prepared by grinding (sample 4) displayed a decay profile quite similar than that obtained with TiO₂ P25, indicating that the charge carrier dynamics are mainly due to the TiO₂ P25 contribution. Unexpectedly, the 50TiO₂ ST/com-HKUST-1 (sample 5) presented a TRMC signal with an I_max value slightly smaller than that of the grinding composite (sample 4), clearly showing a charge carrier separation, but with a short time decay. The decay signal abruptly becomes highly noisy after 70 ns, denoting a charge carrier recombination or electron transfer from TiO₂ to HKUST-1. The 50TiO₂ ST/com-HKUST-1 (sample 6) did not display a clear TRMC signal, similar to that shown by the com-HKUST-1(Figure 6a), which can be connected with the XRD results, meaning that TiO₂ was poorly crystallized.

As simulated solar light is being used in the photocatalytic evaluation, it is interesting to see the TRMC signals of the materials under 410 nm excitation. As can be seen in Figure 6c, TiO₂ P25, com-HKUST-1 and syn-HKUST-1MOFs presented similar behaviors to those presented under 355 nm excitation, but all the samples showed a lower I_max value. The TiO₂ P25 signal can be attributed to the presence of rutile, which has a bandgap of 3.0 eV, making possible the generation of electron-hole pairs under visible-light irradiation. All the composites shown in Figure 6d exhibited a small TRMC signal, which means that they have the capacity to generate electron-hole pairs under 410 nm excitation.

3. Discussion

The above results suggest that the photocatalytic performance of TiO₂ is improved by the incorporation of HKUST-1, forming a semiconductor-MOF composite regardless the preparation method. Nonetheless, the integration of HKUST-1 with TiO₂ P25 (TiO₂ P25/syn-HKUST-1) by a chemical method showed greater photocatalytic activity and stability compared with to grinding (TiO₂ P25/com-HKUST-1) or the TiO₂-ST/com-HKUST-1. At first glance, the greater photocatalytic activity shown by the composites in comparison with TiO₂ P25 or HKUST-1 is explained by the synergy between the semiconductor and the MOF, inhibiting electron-hole recombination. This cooperative behavior implied the partial reduction of the Cu²⁺ contained in the MOF, forming Cu¹⁺ species which absorb visible light and could contribute to the proton reduction, as shown in Figure 7. The creation of reversible Cu¹⁺/Cu²⁺ entities in the composite was attributed to the electron transfer from P25 to HKUST-1 generating in situ species, i.e., Cu¹⁺-Cu²⁺ MOF, giving rise to improved hydrogen production.

Furthermore, it was found that the higher photoactivity and stability shown by the TiO₂ P25/syn-HKUST-1 can also be related to a strong interaction between the two components, which was not seen with the other two composites prepared by grinding or mixing poor crystallized TiO₂ with commercial HKUST-1.

4. Materials and Methods

4.1. Materials

Copper (II) acetate monohydrate (Cu(OAc)₂H₂O), trimesic acid (H₃BTC), N,N-dimethylformamide (DMF), Triethylamine (Et₃N), and Titanium (IV) isopropoxide were purchased from Sigma-Aldrich (St. Louis, MO, USA), Ethanol (EtOH) and ammonium hydroxide 30% were purchased from Panreac Chemicals (Chicago, IL, USA), and deionized water (H₂O), Aeroxide^® TiO₂ P25 (Degussa), and Basolite^® C300 (Sigma-Aldrich) were used as reference materials without further purification.

4.2. Preparation Methods

4.2.1. Grinding (TiO₂ P25/com-HKUST-1)

The composites were prepared by grinding the TiO₂ P25 and commercial HKUST-1 powders by hand in an agate mortar until a homogeneous light blue color was obtained. Five composites were prepared with a TiO₂ content of 25, 50 and 75 wt %.

4.2.2. TiO₂ Solvothermal Deposition on Commercial HKUST-1 (TiO₂-ST/com-HKUST-1)

TiO₂ was prepared by mixing 1.14 mL of titanium (IV) isopropoxide with 100 mL of anhydrous ethanol and sonicating this for 5 min. Then, under vigorous magnetic stirring, concentrated nitric acid (70% v/v) was added drop by drop to get a pH around 1. The solution was diluted with 10 mL of distilled water, and 8 mL of ammonium hydroxide was added as a precipitating agent. Subsequently, a given amount of commercial HKUST-1 was added to the suspension under vigorous stirring. The resultant suspension was placed in a homemade PTFE-lined autoclave and sealed hermetically, and then introduced into a convective furnace (Fisher Scientific, Pittsburgh PA, USA) at 100 °C for 24 h. The composite formed was recovered and washed five times with a mixture of DMF/EtOH/H₂O (molar ratio of 1:1:1) to eliminate the residues of any organic compound. Then, the product was dried at 50 °C for 5 h. The solid was ground in an agate mortar and sieved using a US 80 mesh to get a homogeneous particle size. Three composites were prepared by this method with a nominal TiO₂ content of 25, 50, and 75 wt %.

4.2.3. TiO₂ P25 Incorporation During HKUST-1 Synthesis (TiO₂ P25/syn-HKUST-1)

The synthesis route of synthesized HKUST-1 mainly followed the procedure reported by Tranchemontagne et al. [41] with some modifications, such as the integration of the TiO₂ P25 during the preparation of precursor solution. First, two solutions were prepared: one solution containing 100 mg of trimesic acid (H₃BTC) dissolved in 6 mL of a mixture of DMF/EtOH/H₂O with a molar ratio (1:1:1), and a second solution contained 200 mg of Cu(OAc)₂ × H₂O dissolved in 6 mL of the solvent DMF/EtOH/H₂O. Both solutions were mixed under magnetic stirring to get a homogeneous solution. Then, 0.2 mL of Et₃N was added drop by drop as an oxidant agent to the reaction mixture under magnetic stirring. After that, a given quantity of TiO₂ P25 (25, 50 and 75 wt %) was added to the MOFs precursor solution. This suspension was kept under magnetic stirring for 24 h, and the powder was obtained by centrifugation. The solid was washed five times with 5 mL of DMF to eliminate the residues of any organic compound, and finally, the solid was dried—at 50 °C for 5 h—milled and sieved to get a homogeneous particle size.

4.3. Characterization Techniques

All composites were characterized by several techniques. X-ray diffraction patterns were recorded on a Siemens D-5000 diffractometer (Munich, Germany), with a copper anode and Cu-Kα radiation over a 2 theta range of 10–80° using a step size of 4 °/min. FTIR and UV-Vis spectra of powder samples were respectively obtained using a Nicolet system (Nexus 470, Thermo Fisher Scientific, Waltham, MA, USA) (with KBr pellet samples) and a GBC spectrophotometer (Cintra 20, GBC Scientific, Hampshire, IL, USA), repectively.

The dynamics of the charge carriers in the photocatalysts were studied by the TRMC technique. The TRMC set-up consists of two main components: (1) a pulse light source, which has the objective to photo-excite the samples and (2) microwave source. A Gunn diode of Kα band at 30 GHz was used to generate the incident microwaves. A tunable laser in the range between 220 and 2000 nm (NT342B; EKSPLA, Vilnius, Lithuania) was used as a light source. It was equipped with an optical parametric oscillator (OPO). The laser delivered 8 ns FWMH pulses with a frequency of 10 Hz. The selected excitation wavelengths were 355 and 410 nm, with a light energy density of 747.6 μJ × cm⁻² and 2.6 mJ × cm⁻², respectively.

4.4. Photocatalytic H₂ Evolution

The composite powders were evaluated in the hydrogen production reaction using a glycerol-water solution with a volumetric ratio glycerol/water = 1:9. The photocatalytic reaction was carried out in a 25 mL glass cell. The composites (1 g/L) and the glycerol-water mixture was placed in the cell and mixed to form a homogeneous suspension, and then purged with nitrogen to eliminate all dissolved oxygen. Before irradiation, the reaction cell was maintained under stirring for one hour for adsorption/desorption equilibration and then irradiated with a solar simulator (Model 9600, 150 W; Newport Corporation, Irvine, CA, USA) for 8 h or 24 h. The gas mixture (H₂, CO₂, CH₄) produced during the reaction was analyzed in a Perkin–Elmer Gas Chromatograph (Autosystem XL, Waltham, MA, USA).

5. Conclusions

A series of TiO₂/HKUST-1 composites were successfully prepared using grinding, solvothermal, and chemical methods. All composites showed a higher photocatalytic activity than the individual components, particularly those containing a TiO₂/HKUST-1 weight ratio of 1:1. These results demonstrated the effect of the synthesis method of composites on photocatalytic activity and stability. The best performance was obtained with the composite prepared by a chemical route, i.e., the synthesis of HKUST-1 in the presence of TiO₂ P25, leading to a strong interaction between the two components. The higher photocatalytic performance of the composites, compared with TiO₂ or HKUST-1, was explained regarding a synergy between the semiconductor and the HKUST-1, inhibiting electron-hole recombination. There was experimental evidence of the reversible partial reduction of Cu²⁺ towards the Cu¹⁺-Cu⁰ entities contained in HKUST-1, which could indicate the in situ formation of highly active HKUST-1 co-catalysts, improving the photocatalytic activity.