Hydrogen Production via Pd-TiO₂ Photocatalytic Water Splitting under Near-UV and Visible Light: Analysis of the Reaction Mechanism

Abstract

Photocatalytic hydrogen production via water splitting using a noble metal on a TiO₂ is a technology that has developed rapidly over the past few years. Specifically, palladium doped TiO₂ irradiated with near-UV or alternatively with visible light has shown promising results. With this end in mind, strategically designed experiments were developed in the Photo-CREC Water-II (PCW-II) Reactor using a 0.25 wt.% Pd-TiO₂ under near-UV and visible light, and ethanol as an organic scavenger. Acetaldehyde, carbon monoxide, carbon dioxide, methane, ethane, ethylene, and hydrogen peroxide together with hydrogen were the main chemical species observed. A Langmuir adsorption isotherm was also established for hydrogen peroxide. On this basis, it is shown that pH variations, hydrogen peroxide formation/adsorption, and the production of various redox chemical species provide an excellent carbon element balance, as well as OH^• and H^• radicals balances. Under near-UV irradiation, 113 cm³ STP of H₂ is produced after 6 h, reaching an 99.8% elemental carbon balance and 99.2% OH^• and H^• and radical balance. It is also proven that a similar reaction network can be considered adequate for the photoreduced Pd-TiO₂ photocatalyst yielding 29 cm³ STP of H₂ with 95.4% carbon and the 97.5% OH^•–H^• radical balance closures. It is shown on this basis that a proposed “series-parallel” reaction network describes the water splitting reaction using the mesoporous Pd-TiO₂ and ethanol as organic scavenger.

1. Introduction

The global community has been working towards the production of alternative energy sources while providing sustainable lifestyles for its population [1]. Given this, hydrogen is being considered as a possible energy carrier [2]. Hydrogen can be produced from different sources such as crude oil, gas, wood, and alcohols [3]. However, among these raw materials, only hydrogen generated from water can be considered as a true environmentally friendly energy carrier. When hydrogen is produced from water and following combustion, it releases zero CO₂ [4,5].

However, not only water is needed to produce hydrogen. A light source is also vital [4]. Photons may split the water molecule with the help of a semiconductor generating electron–hole pairs [6]. Photoexcited electron–hole pairs can be separated efficiently, using organic scavenger agents that reduce the electron–hole pair recombination, acting as electron donors [7,8,9,10,11].

To date, titanium dioxide (TiO₂) has been the most common photocatalyst used to produce hydrogen, due to its stability, resistance to corrosion, cleanliness (no pollutants), availability in nature, and low cost [12,13]. However, TiO₂ is limited by its wide band gap. As an alternative, TiO₂-based photocatalysts can be synthesized and modified by adding noble metals to narrow the band gap for better sunlight utilization. For example, a TiO₂ semiconductor material can be synthesized and modified using palladium doping [14]. By utilizing this noble metal, one can increase the efficiency of the hydrogen formation, creating additional photocatalyst sites where hydrogen and intermediates are formed [15,16].

Our research group has investigated the reaction performance of platinum on TiO₂ under near-UV light [17]. By-products such as CO₂, ethane, acetaldehyde, and hydrogen peroxide were identified as the result of reduction-oxidation reactions. Given the significant Pt cost, a less expensive Pd doped mesoporous TiO₂ photocatalyst is currently being developed, with ethanol being used as a sacrificial agent [18]. Experiments are developed in the Photo-CREC Water-II Reactor (PCW-II Reactor). On this basis, a reaction network as reported here, for both near-UV and visible light irradiation sources, is postulated. Validation of this reaction network is accomplished employing (a) elemental carbon balances for reactants and products species and (b) H^• and OH^• balances including hydrogen produced, hydrogen peroxide formed and adsorbed, oxidized, and reduced carbon containing species, and pH.

2. Proposed Reaction Mechanism

Regarding the photocatalytic reaction, it can be hypothesized that different by-products are formed because of photo-redox reactions, as observed in Figure 1. Palladium creates holes that react with the organic scavenger ethanol, forming by-products. In the gas phase, in addition to hydrogen, the detected by-products include methane, ethane, ethylene, acetaldehyde, CO, and CO₂. In the liquid phase, ethanol and hydrogen peroxide were also identified.

Rusinque et al. [14,15] reported hydrogen production runs (165 experiments) using different photocatalysts concentrations (0.15, 0.30, 0.50, and 1.00 g L⁻¹) as well as Pd loadings (0.25, 0.50, 1.00, 2.50 and 5.00 wt.% Pd) on TiO₂. Mechanistic considerations reported in the present manuscript are established for a 0.15 g L⁻¹ photocatalyst concentration, with 0.25 wt.% Pd-TiO₂ loading, which was found to be the optimum Pd loading for hydrogen production. Regarding the 0.25 wt.% Pd-TiO₂, it was observed to display a reduced band gap of 2.51 eV, which leads to hydrogen being produced under visible light as well. The absorbed radiation was evaluated via macroscopic irradiation energy balances [18].

As a result of the hydrogen formation reactions using 0.25 wt.% Pd-TiO₂, the ethanol OH· radical in the organic scavenger is consumed, and the following can be postulated:

(a)

Hydrogen production proceeds via a ‘‘series–parallel’’ redox reaction network.

(b)

Water splits, forming intermediate OH^• and H^• radicals, with H^• reacting further and yielding molecular hydrogen, as shown in Equations (1) and (2).

(c)

Ethanol, as an OH· organic scavenger, is consumed via different reaction pathways to form various oxidation by-products, such as acetaldehyde, CO, and CO_2, as described with Equations (3)–(7).

(d)

Ethanol and ethanol by-products are reduced via the H· radicals present, yielding methane, ethane, and ethylene, as reported with Equations (8)–(10).

2.1. Step 1: Hydrogen Production Pathway

Mechanism for hydrogen production

$$\mathrm{hv} \stackrel{\mathrm{Pd}/\mathrm{Ti}{O}_2}{\to } {\mathrm{h}}^{+}+ {\mathrm{e}}^{-}$$

Mechanism for by-products.

2.2. Step 2: Ethanol Derived By-Products Formation

2.2.1. Oxidation Reactions

(a)

Acetaldehyde

$${\mathrm{C}}_2{\mathrm{H}}_5\mathrm{OH}+{\mathrm{OH}}^{•} \stackrel{\mathrm{Pd}/{\mathrm{TiO}}_2}{\to }{\mathrm{C}}_2{\mathrm{H}}_5{\mathrm{O}}^{-}+{\mathrm{H}}_2\mathrm{O}$$

(3)

$${\mathrm{C}}_2{\mathrm{H}}_5{\mathrm{O}}^{-}+\mathrm{OH} \stackrel{\mathrm{Pd}/{\mathrm{TiO}}_2}{\to }{ \mathrm{C}}_2{\mathrm{H}}_4\mathrm{O}+{\mathrm{H}}_2\mathrm{O}$$

(4)

The addition of Equations (3) and (4) yields the following overall equation:

$${\mathrm{C}}_2{\mathrm{H}}_5\mathrm{OH}+2{\mathrm{OH}}^{•}\stackrel{\mathrm{Pd}/{\mathrm{TiO}}_2}{\to }{ \mathrm{C}}_2{\mathrm{H}}_4\mathrm{O}+2{\mathrm{H}}_2\mathrm{O}$$

(5)

(b)

Carbon Dioxide

$$2{\mathrm{C}}_2{\mathrm{H}}_5\mathrm{OH}+6{\mathrm{OH}}^{•}\stackrel{\mathrm{Pd}/{\mathrm{TiO}}_2}{\to } 4{\mathrm{CO}}_2+9{\mathrm{H}}_2$$

(6)

(c)

Carbon Monoxide

$${\mathrm{C}}_2{\mathrm{H}}_5\mathrm{OH}+8{\mathrm{OH}}^{•}\stackrel{\mathrm{Pd}/{\mathrm{TiO}}_2}{\to } 2\mathrm{CO}+7{\mathrm{H}}_2\mathrm{O}$$

(7)

2.2.2. Reduction Reactions

(d)

Methane

$${\mathrm{C}}_2{\mathrm{H}}_5\mathrm{OH}+4{\mathrm{H}}^{•}\stackrel{\mathrm{Pd}/{\mathrm{TiO}}_2}{\to } 2{\mathrm{CH}}_4+{\mathrm{H}}_2\mathrm{O}$$

(8)

(e)

Ethane

$${\mathrm{C}}_2{\mathrm{H}}_5\mathrm{OH}+{\mathrm{H}}^{•}\stackrel{\mathrm{Pd}/{\mathrm{TiO}}_2}{\to }{ \mathrm{C}}_2{\mathrm{H}}_6+\mathrm{OH}$$

(9)

(f)

Ethylene

$${\mathrm{C}}_2{\mathrm{H}}_5\mathrm{OH}\stackrel{\mathrm{Pd}/{\mathrm{TiO}}_2}{\to }{\mathrm{C}}_2{\mathrm{H}}_4+{\mathrm{H}}_2\mathrm{O}$$

(10)

Additionally, hydrogen peroxide is produced due to the recombination of some of the OH radicals present:

$${\mathrm{OH}}^{•}+{\mathrm{OH}}^{•}\stackrel{\mathrm{Pd}/{\mathrm{TiO}}_2}{\leftrightarrow }{\mathrm{H}}_2{\mathrm{O}}_2$$

(11)

In summary, highly valuable products from the redox reactions are generated when using the scavenger ethanol. However, hydrogen and other hydrocarbon products are formed with a very small ethanol consumption.

2.3. Step 3: Ethanol Photoregeneration

During the 165 runs developed, a consistently small overall ethanol consumption was observed. This can be explained given that palladium is one of the strongest C-C coupling catalysts and can also form C₂H₅OH via CO photoreduction during the water splitting reaction as follows [19]:

• CO molecules are strongly adsorbed onto a Pd-TiO₂ surface until a second CO is available for C-C coupling.
• Due to the reduced band gap of the photocatalyst (2.51 eV), electrons jump from the valence band to the conduction band and are trapped by the palladium.
• The photogenerated electrons are used to activate and reduce the CO, which lead to ethanol formation via hydrogenation.

Thus, the following reaction mechanism towards ethanol photoregeneration can be postulated as [20]:

(a)

C-C coupling involves electron transfer, with this leading to the formation of the *C₂O₂-intermediate.

(b)

Once the *C₂O₂-intermediate is generated, hydrogenation, electron transfer takes place, with the *C₂O₂H yielding to ethanol.

$$\begin{gathered} 2 \mathrm{CO} \stackrel{{\mathrm{e}}^{-}}{\to }{ \mathrm{C}}_2{\mathrm{OO}}^{-} \stackrel{{\mathrm{H}}^{+}+{\mathrm{e}}^{-}}{\to } { \mathrm{C}}_2{\mathrm{O}}_2\mathrm{H} \stackrel{{\mathrm{H}}^{+}+{\mathrm{e}}^{-}}{\to } {\mathrm{C}}_2\mathrm{O}+{\mathrm{H}}_2\mathrm{O}\stackrel{{\mathrm{H}}^{+}+{\mathrm{e}}^{-}}{\to } {\mathrm{C}}_2\mathrm{OH} \stackrel{{\mathrm{H}}^{+}+{\mathrm{e}}^{-}}{\to }{ \mathrm{C}}_2\mathrm{HOH} \\ { \mathrm{C}}_2\mathrm{HOH} \stackrel{{\mathrm{H}}^{+}+{\mathrm{e}}^{-}}{\to }{ \mathrm{C}}_2{\mathrm{H}}_2\mathrm{OH}\stackrel{{\mathrm{H}}^{+}+{\mathrm{e}}^{-}}{\to }{ \mathrm{C}}_2{\mathrm{H}}_3\mathrm{OH} \stackrel{{\mathrm{H}}^{+}+{\mathrm{e}}^{-}}{\to }{ \mathrm{C}}_2{\mathrm{H}}_4\mathrm{OH} \stackrel{{\mathrm{H}}^{+}+{\mathrm{e}}^{-}}{\to }{ \mathrm{C}}_2{\mathrm{H}}_5\mathrm{OH} \end{gathered}$$

(12)

Thus, various reaction steps in Equation (12), may lead altogether to ethanol synthesis as:

$$2\mathrm{CO}+8{\mathrm{H}}^{•}\stackrel{\mathrm{Pd}/{\mathrm{TiO}}_2}{\to }{ \mathrm{C}}_2{\mathrm{H}}_5\mathrm{OH}+{\mathrm{H}}_2\mathrm{O}$$

(13)

In summary, ethanol consumption and ethanol formation steps may coexist during hydrogen formation. It is the goal of the present study to clarify the relative importance of the above-described reaction steps, as shown in Equations (1)–(13).

3. Results and Discussion

3.1. Photocatalytic Hydrogen Production under Near-UV Light and Visible Light

The palladium-doped TiO₂ photocatalyst of the present study was specially developed to enhance hydrogen production while compared to a previously studied undoped mesoporous TiO₂ [14]. Figure 2 reports a cumulative hydrogen of 5055 μmoles, equivalent to a volume of 113 cm³ STP (standard temperature and pressure) after six hours of near-UV irradiation, when using a 0.25 wt.% Pd on TiO₂. This photocatalyst (designated as Pd-TiO₂-nUV) yields at reaction conditions of 0.15 g L⁻¹, 2.0 v/v% of ethanol, and initial pH = 4 ± 0.05. The 5055 μmoles of H₂ produced using a 0.25 wt.% Pd-TiO₂-UV, after 6 h of reaction, are very favorable compared to the 1927.8 μmoles obtained when using undoped TiO₂ and the 696.7 moles of H₂ acquired with using the commercial DP-25 TiO₂ This hydrogen volume is equivalent to almost 300% the hydrogen volume obtained with the undoped mesoporous TiO₂ [14].

Figure 2 shows that the Pd-TiO₂-nUV photocatalyst displays a stable linear trend, with a consistent zero-order kinetics, with photocatalytic activity remaining unchanged during the 6 h of near-UV irradiation. This photocatalytic activity for hydrogen formation takes place via an “in series-parallel” reaction mechanism, proposed via Equations (1)–(13). Furthermore, the resulting quantum yield expressed as the moles of H^• over the moles of photons absorbed is 13.7%, as reported recently [14].

The same Pd-TiO₂ photocatalyst designated as Pd-TiO₂-VIS was evaluated further in the PCW-II reactor as follows: (a) photoreduction phase: near-UV light radiation was used during the first run hour, (b) visible light phase: visible light irradiation was employed during the remaining 5 h of the 6-h run. The developed Pd-TiO₂-VIS photocatalytic runs were conducted under the following conditions: 0.15 g/L of catalyst concentration, 2.0 v/v% of ethanol, and initial pH = 4 ± 0.05.

Regarding the Pd-TiO₂-VIS photocatalyst, Figure 3 shows that during the first near-UV light photoreduction hour, the Pd-TiO₂ photocatalyst yielded 979 μmoles of hydrogen. This was followed by an extra 314 μmoles of hydrogen produced during the 1–6 h period, under visible light. This yielded 1292 μmoles of hydrogen in total, with this being equivalent to a 29 cm³ STP. One should note that the 314 μmoles of extra hydrogen formed between hour 1 to hour 6 compares very favorably with the 46 μmoles of hydrogen produced exclusively with 5 h of visible light, using an undoped TiO₂ photocatalyst and the 141.2 μmoles of H₂ acquired with the commercial DP-25 TiO₂. Thus, this re-presents, altogether, a 76% hydrogen production increase, using the Pd-TiO₂-VIS in the PCW-II Reactor.

One can note that the Pd-TiO₂-VIS photocatalyst displays, as shown in Figure 3, and after the first hour of near-UV irradiation, a linear cumulative hydrogen formation. This is consistent with a zero-order kinetics and unchanged photocatalytic activity. Thus, the Pd-TiO₂ photocatalyst shows a positive performance for hydrogen production, likely diminishing electron–hole pair recombination, and consequently contributing to higher hydrogen yields under both near-UV and visible light irradiation.

3.2. By-Products Formation

Regarding the carbon containing products formed, the improved TiO₂ photocatalyst with Pd led to (a) photoreduction of H radicals to produce methane and ethane, and (b) photooxidation of OH radicals to form CO, CO₂, and acetaldehyde. Thus, both the reduced and oxidized generated carbon containing species, as described in Figure 4 and Figure 5, can be used to support the mechanism given by Equations (3)–(10).

Carbon balances can be used to validate the mechanistic steps leading to carbon containing species formation from photocatalytic hydrogen formation. Carbon balances must include both ethanol and formed products (methane, ethane, ethylene, carbon dioxide, carbon monoxide, and acetaldehyde). Figure 6 reports a 99.8% typical element carbon balance closure for experiments developed using 0.25 wt.% Pd-TiO₂-nUV.

Figure 6 considers (a) 4.10 × 10⁶ μmoles of carbon in ethanol, at the beginning of the run, and (b) 4.09 × 10⁶ μmoles of carbon, accounting for the remaining moles of ethanol and all by-products formed after 6 h of near-UV irradiation. Note that the by-products represent only 0.06% of the total carbon present, at the end of the irradiation period, accounting for 2400 μmoles. Thus, one can establish that the photocatalytic experiments take place under close to constant ethanol scavenger concentration. This minor overall ethanol consumption is attributed to ethanol photoregeneration, as shown in Equations (12) and (13). Appendix A provides additional details about the elemental carbon balances.

Figure 7 further reports a similar elemental carbon balance calculation, including all carbon containing species, using the 0.25 wt.% Pd-TiO₂-VIS photocatalyst. From the 3.92 × 10⁶ μmoles of carbon contained in ethanol, 3.74 × 10⁶ μmoles of carbon were detected in products after 6 h. This provided a 95.4% carbon balance closure and 2688 μmoles of carbon containing in products or 0.07% of the total carbon. This shows once again that under the conditions studied, ethanol, while being important in acting as an OH^• scavenger, remains at “quasi-constant” concentration during the entire run.

3.3. H^• and OH^• Radical Group Balance

While considering the various reaction steps described via Equations (1)–(13), it is important to clarify the role of the OH^• and H^• radicals. In this regard, the concentration of the OH^• and H^• radicals measured during a run can be stated to be the consequence of a net balance between their formation and consumption, with this leading to various oxidation and reduction products.

More specifically, the formation of OH^• radicals is the result of OH^- ion and h+ site interactions, as described in Equation (1). H^• radicals are generated, as the outcome of a H⁺ ion accepting an electron, as shown with Equation (2). Thus, if the mechanism proposed here is sound, there must be a balance of OH^• and H^• formed. To thoroughly test this assumption, the following can be considered:

• The H^• radicals formed can be calculated via the accounting of the experimentally obtained hydrogen, as postulated in Equation (2), and via the hydrogen consumption required by the synthesis of various reduced products (methane, ethane), as given by Equations (8) and (9).
• The OH^• can be quantified by considering the OH^• radicals consumed, according to their stochiometric requirements from several oxidation reactions, as given by Equations (3)–(6) and (11). Thus, the OH^• consumption should account for acetaldehyde, carbon monoxide, carbon dioxide, and hydrogen peroxide species.

On this basis,

Table 1. Net μMoles of H^• formed and OH^• consumed following 6 h of irradiation using the Pd-TiO₂-nUV and Pd-TiO₂-VIS.

	μmoles of H• Formed Equations (2), (8) and (9) (a)	μmoles of OH• Consumed Equations (3)–(6) (b)
Pd-TiO2-nUV.	10,191.5	2169.6
Pd-TiO2-VIS	2620.3	2342.4

Note: (a) The μmoles of OH^• radicals are calculated based on oxidized carbon products (CO₂, acetaldehyde), (b) the μmoles of H^• radicals are calculated on the basis of H₂ and reduced carbon species (methane, ethane).

reports the calculated total moles of H^• and OH^• formed, during the water splitting photocatalytic reaction, under both near-UV and visible light, accounting exclusively for Equations (2)–(10).

Table 1 reports, in this regard, a significant imbalance between the moles of H^• and the moles of OH^• radicals consumed, with only 21.3% of the moles of OH^• radicals contributing to the formation of by-products under near-UV light. On the other hand, under visible light, 89.4% of the total moles of OH^• radicals led to carbon containing oxidation by-products. Thus, under both near-UV and visible light, the proposed redox mechanism, as postulated via Equations (2)–(10), for the total moles of H^• and OH^•, is deficient in the moles of OH^• radicals consumed. Thus, further adjustments of the moles of OH^• radicals reacted are required, as considered in Section 3.3.1 and Section 3.3.2 of this manuscript.

3.3.1. Further Establishing of the Total OH^• Formed during Photocatalytic Hydrogen Production

Regarding hydrogen peroxide species under near UV, they are considered as the net result of the rate of OH^• dimerization, as shown in Equation (11), and the rate of H₂O₂ decomposition as explained later via Equation (14). To account for the H₂O₂ formation during the photocatalytic hydrogen production, liquid samples were periodically analyzed using a colorimetric method. As reported in Figure 8, during 6 h of near-UV irradiation, the hydrogen peroxide concentration consistently increased, with a maximum of 188.4 μmoles of H₂O₂ being obtained.

On the other hand, Figure 8 also shows that OH^• dimerization plays an important role, as described via Equation (11), under visible light irradiation. In this case, the reaction pathway involves H₂O₂, which is formed during the first hour of near-UV irradiation, with modest additional H₂O₂ formation during the five following hours of visible light.

Table 2. Cumulative H^• formed/consumed and OH^• consumed-1.

	μmoles of H• Formed as Shown in Equations (2), (8) and (9)	Cumulative μmoles of OH• Consumed Forming H2O2 as Shown in Equation (11) (Liquid Phase)	Cumulative μmoles of OH• Consumed as Shown in Equations (3)–(7) and (11)
Pd-TiO2-nUV.	10,191.5	188.4	2359
Pd-TiO2-VIS	2620.3	89.1	2431

reports the cumulative OH^• consumption that leads to H₂O₂ formation. Hydrogen peroxide is detected in the liquid phase using both Pd-TiO₂-nUV and Pd-TiO₂-VIS. It can be observed that the OH^• consumption due to H₂O₂ generation only modifies the cumulative moles of OH^• by 1.8% and 3.4% of the total amount, respectively.

Furthermore, Pd-TiO₂ may also adsorb chemical species including hydrogen peroxide [21]. This adsorption may affect the cumulative amount of OH^• moles consumed. Thus, to evaluate this effect, adsorption measurements lasting 60 min under dark conditions were effected [22].

Figure 9a describes the obtained Langmuir chemisorption isotherm (Q_e = Q_e,max KCe/1 + KC_e), showing the H₂O₂ adsorption equilibrium concentration. Through Langmuir equation linearization (Figure 9b), the H₂O₂ adsorption parameters were calculated, for 0.25 wt.% Pd-TiO₂, as shown in

Table 3. Adsorption constants for hydrogen peroxide and ethanol.

Adsorption Constants
	K	Qe,max
Hydrogen Peroxide	0.93 mg−1 L	11.1 mg−1 gcat

.

Table 2 reports both the adsorption constant, K, and the maximum adsorption capacity, Q_e,max, for a hydrogen peroxide adsorption isotherm. The obtained Q_e,max differ from the one reported by Sahel [23], who found a Q_e,max = 7.48 mg⁻¹ L value for a undoped TiO₂ photocatalyst. This value is lower than the 11.1 mg⁻¹ g_cat maximum adsorption capacity reported in our study. The higher Q_e,max of the present study, can be justified given the higher surface area of the palladium photocatalyst (131 m² g⁻¹), with pore sizes in the 16–20 nm range. The surface area is almost three times larger than the one reported by Sahel, where the TiO₂ surface area was only 50 m² g⁻¹ [23].

Thus, it can be established that there is an extra 45% of hydrogen peroxide formed and adsorbed on the photocatalyst. On this basis, the μmoles of OH^• consumed during the runs can be revised, as shown in

Table 4. Cumulative H^• formed/consumed and OH^• consumed-2.

	Cumulative μmoles of H• Formed Calculated with Equations (2), (8) and (9)	Cumulative μmoles of OH• Forming H2O2 (Adsorbed)	Cumulative μmoles of OH• Consumed Using Equations (3)–(7) and (11), and OH• Adsorbed
Pd-TiO2-nUV	10,191	84.4	2444
Pd-TiO2-VIS	2620	40.1	2472

.

Table 4 data also show that the addition of the adsorbed H₂O₂ species accounts for 0.82% for Pd-TiO₂-nUV and 1.52% for Pd-TiO₂-VIS, in the context of H^• and OH^• mole balance only. Thus, there is still a significant deficiency of calculated OH^• radicals consumed and H^• radicals produced that must be accounted for.

3.3.2. pH Influence on the Photocatalytic Reaction

In the water splitting reaction for hydrogen production, an important factor that should be considered is the pH of the solution. This is the case given its influence on the process efficiency [24]. The redox reactions take place via hydrogen formation due to the combination of excited electrons and H⁺ protons adsorbed on the photocatalyst. It was proven in this respect that hydrogen production is favored in acidic conditions, due to availability of dissolved H⁺ ions [25].

During the photocatalytic water splitting reaction using the Pd-TiO₂, there is a significant change of pH with irradiation time, as shown in Figure 10. At the beginning of each experiment, the pH of the water–ethanol solution was set at 4.0 ± 0.005 [26]. Upon completion of the photocatalytic reaction, the pH increased to 5.89 ± 0.005 under near-UV light and to 4.60 ± 0.005 under visible light. Thus, in the case of the Pd-TiO₂-nUV, it was noticed that a pH of 6 and the TiO₂ isoelectric point were reached [27].

The reported pH variation might be attributed to an electron exchange between the photocatalyst and the split water molecules. An electron can be donated by the photocatalyst surface, in order for the active sites on the Pd⁰ to yield HO^• and OH⁻ ion free radicals [28]. Furthermore, it can be hypothesized that a fraction of the pH change can occur due to H₂O₂ decomposition. One can consider that for both the Pd-TiO₂-nUV and Pd-TiO₂-VIS, the near-UV irradiation leads to the following chain of reactions:

$${\mathrm{Pd}}^{2+}+2{\mathrm{H}}_2{\mathrm{O}}_2\stackrel{nUV}{\to }{\mathrm{Pd}}^0+2{\mathrm{H}}^{+}+2{\mathrm{HO}}_2^{•}$$

(14)

$${\mathrm{HO}}_2^{•}\leftrightarrow {\mathrm{H}}^{+}+{\mathrm{HO}}_2^{-}$$

(15)

$$2{\mathrm{HO}}_2^{-} \to 2{\mathrm{HO}}^{-}+{\mathrm{O}}_2$$

(16)

One should note that in this postulated pathway, the photocatalyst accepts electrons, with the Pd²⁺ ion sites yielding a HO₂^• radical, as described in Equation (14). This HO₂^• radical gives OH⁻ ions via Equations (15) and (16) [28].

Thus, the total moles of OH^• radicals consumed can be revised further, accounting for the change in pH.

Table 5. Cumulative H^• formed/consumed and cumulative μmoles OH^• consumed.

	Cumulative μmoles of H• Formed via Equation (2), (8) and (9)	Cumulative μmoles of OH• Produced via pH Change	Cumulative μmoles of OH• Consumed Via Equations (3), (6), and (11) Based on H2O2 Adsorbed and pHChange	Percentual H• OH• Balance Closure (%)
Pd-TiO2-UV	10,191	7662.5	10,106.3	99.2
Pd-TiO2-VIS	2620	81.9	2553.5	97.5

and Figure 11a and Figure 12a show that the accounting of the cumulative moles of OH^• consumed via a pH change provides a 97–99% balance of the moles of H^• formed and the moles OH^• consumed while using Pd-TiO₂-nUV and Pd-TiO₂-VIS.

Figure 11b and Figure 12b also show that these consistent balance of H^• and OH^• μmoles were also observed at various other irradiation times, providing significant strength and validation to the photocatalytic reaction involving both Pd-TiO₂-UV and Pd-TiO₂-VIS. Appendix B provides additional details of the H^• and OH^• mole balance calculations.

Thus, and given the reported results and the various described considerations, the following mechanistic steps can be considered for water splitting:

(a)

H₂ is a main product from the photocatalytic water splitting reaction using ethanol as an organic scavenger and a 0.25 wt.% Pd-TiO₂-nUV or alternatively, a 0.25 wt.% Pd-TiO₂-VIS.

(b)

Formed photoreduction species (methane, ethane) and photooxidation species (CO₂, acetaldehyde) are all important carbon containing by-products.

(c)

Hydrogen peroxide, present in the liquid phase, is formed and adsorbed on the photocatalyst during water splitting.

(d)

OH⁻ species in the water solution progressively increase with irradiation time, with this leading to a pH increase.

4. Experimental Methodology

4.1. Photocatalyst Synthesis

The photocatalyst was synthesized by using the evaporation induced self-assembly (EISA) Method. The semiconductor was prepared using a polymeric template (pluronic F-127 (PEO₁₀₆PPO₇₀PEO₁₀₆), a titanium-based compound (titanium IV isopropoxide), a noble metal (PdCl₂) precursor, ethanol USP (C₂H₅OH), hydrochloric acid (HCl, 37% purity), and anhydrous citric acid. All the reagents were purchased from Sigma Aldrich Co (Oakville, ON, Canada).

Following the sol–gel methodology, micelles and a mesoporous structure were formed. First, 400 mL of ethanol, 33 g of hydrochloric acid and 20 g of Pluronic F-127 were mixed together. Then, 6.30 g of citric acid were added in 20 mL of water, for posterior addition to the initial suspension. Then, 28.5 g of titanium (IV) isopropoxide was dissolved in ethanol and added dropwise to the mixture. Finally, palladium (II) chloride was incorporated. The sol–gel suspension was stirred for 24 h. The removal of the template occurred during the calcination step, at 500 °C, to subsequently create a mesoporous structure with a homogenous dispersion of the palladium nanoparticles [14]. More detailed information about the synthesis of the photocatalyst was presented in Rusinque et al. [15].

4.2. Photocatalyst Characterization

The photocatalyst was characterized using physico-chemical techniques such as nitrogen physisorption (BET), hydrogen chemisorption, X-Ray diffraction (XRD), temperature programmed reduction (TPR), and UV-vis spectroscopy.

The TiO₂ doped with 0.25 wt.% palladium displayed a surface area of 131 m² g⁻¹, with pore sizes in the 16–20 nm range. When characterizing the photocatalyst with hydrogen pulse chemisorption, the fraction of dispersed Pd on the photocatalyst was 75%. The average size of the crystallites, which was determined using the Scherrer equation and the XRD peak broadening, was calculated to be 9 nm. The Pd reduction temperature was found to be above temperatures of 350 °C, which was an indication of strong Pd-TiO₂ interactions. When using the 0.25 wt.% Pd-TiO₂ photocatalyst, and following the Tauc plot methodology, the semiconductor yielded a 2.51 eV band gap, a smaller band gap than the 3.2 eV obtained from undoped TiO₂. A more detailed characterization of the photocatalyst was presented in Rusinque et al. [14].

4.3. Photo-CREC Water-II Reactor

The Photo-CREC Water-II Reactor is a novel unit engineered to uniformly irradiate the surface area of a photocatalyst. This is a slurry batch piece of equipment that ensures that no internal and/or external diffusion transport phenomena takes place.

The PCW-II unit is equipped with a storage tank where the photocatalyst, water, and organic scavenger are loaded and are kept under agitation. This tank has two ports for liquid and gas phase sampling. Figure 13 describes the main components of PCW-II Reactor.

This reactor was designed and built with materials to provide the following: (a) uniform catalyst distribution, (b) a high surface/volume reactor ratio, (c) negligible catalyst fouling, (d) a high near-UV and visible light transmittance of 97%, (e) a well-mixed suspension, and (f) chemical and pH resistance.

4.4. Photocatalytic Experiments

The photocatalyst was evaluated using the Photo-CREC Water-II reactor with a BLB near-UV lamp or alternatively with a fluorescent visible light lamp. The hydrogen storage/mixing tank was loaded with 6000 mL of water and 0.15 g L⁻¹ of 0.25 wt.% Pd-TiO₂ photocatalyst. Ethanol was used as sacrificial agent, and the pH was adjusted at the beginning of the reaction to 4 ± 0.05 using H₂SO₄ [2 M].

Prior to the initiation of the experiment, 0.15 g/L of the 0.25 wt.% Pd-TiO₂ photocatalyst was added to the solution with the following steps being considered: (a) 0.9 g in total of 0.25 wt.% Pd-TiO₂ was mixed with 100 mL of water and subjected to sonication for a 10 min period, to ensure good particle distribution and avoid possible agglomeration; (b) once a thorough dispersion of the 0.25 wt.% Pd-TiO₂ particles was achieved, the photocatalyst–water solution was added to the 6 L of water contained in the Photo-CREC Water-II Reactor, (c) Following this, the pump and lamp were turned on for 30 min prior to the reaction, allowing lamp stabilization and better photocatalyst dispersion in the liquid solution. (d) Finally, for 10 min, argon was used as an inert gas for oxygen removal, from the gas phase in the hydrogen storage tank.

Gas and liquid samples were taken every hour, for 6 h of continuous irradiation. For the experiments under visible light, an initial photoreduction step with near-UV light was considered. Before the reaction began, the photocatalyst was photoirradiated for one hour, with near-UV light, to achieve the further reduction of the catalyst. This approach was reported by Rusinque et al. [15].

4.5. Analytical Techniques

The gas phase was analyzed with a Shimadzu GC2010 Gas Chromatograph Inc (Mandel, Guelph, ON, Canada) using argon (Praxair 99.999%) as a carrier gas. This unit was equipped with a HayeSepD 100/120 mesh packed column (9.1 m × 2 mm × 2 μm nominal SS) (Sigma Aldrich, Oakville, ON, Canada) used for the separation of hydrogen from air. Additional details and information regarding the GC analysis of the Pd-TiO₂ photocatalyst are provided in Appendix C.

For the liquid phase, the Shimadzu HPLC Model UFLC (ultra-fast liquid chromatography) System was utilized using 0.1% H₃PO₄ as a mobile phase. This unit contains a Supelcogel C-610H 30 cm × 7.8 mm ID column. This quantitative analysis was performed by employing the RID (Refractive Index Detector) 10A due to the polar nature of ethanol. This HPLC separated ethanol from water for further quantification.

A colorimetric method was employed for the quantification of H₂O₂ at low concentrations (0–10 mg L⁻¹ approximately). In the colorimetric method, iodide and N-dimethyl-p-phenylenediamine (DPD) were used to detect H₂O₂ during the photocatalytic reaction. The collected sample was mixed with ammonium molybdate that decomposes H₂O₂ in solution and with KI that oxidizes iodide to iodine [29]. Iodine posteriorly oxidizes the DPD compound, generating a pink color. The DPD compound was then measured using a spectrophotometer Spectronic 200+, Thermo Spectronic (Thermo Fischer, Mississauga, ON, Canada), which provides a 340 to 950 nm wavelength range and a nominal spectral bandwidth of 20 nm.

The hydrogen peroxide concentration was estimated using a linear calibration for 530 nm, considering the absorption spectra of the sample. All the reagents used for hydrogen peroxide detection were purchased from Hach^® (London, ON, Canada). A commercial H₂O₂ technical-grade solution (30% w/w of H₂O₂) was supplied by BioShop Canada (Burlington, ON, Canada).

4.5.1. Determination of H₂O₂ Concentrations

To determine the amount of H₂O₂, 0.15 mL of KI solution (20%) and 0.15 mL of Mo(VI) solution (ammonium molybdate in sulfuric acid) were placed in a 10-mL sample. The volumetric flask was capped and shaken for proper mixing. After 6 min of reaction time, one pillow of DPD (bag of N, N-dimethyl-p-phenylenediamine), with a total of 25-mL of chlorine powder, was added to the prepared sample cell. A pink color developed, indicating the presence of H₂O₂. Subsequently, the sample was transferred to a quartz cuvette, and the absorbance was measured by a spectrophotometer (Spectronic 200+, Thermo Spectronic). The absorbance was obtained at 530 nm, in terms of total chlorine concentration ([Cl₂]), and according to a calibration curve shown in Figure 14. This methodology allowed the quantification of H₂O₂ in the sample. Deionized water was used as blank.

The colorimetric and permanganometry methods were compared to determine the best approach for hydrogen peroxide identification, specifically for the present study. The permanganometric titration showed a standard deviation between 9–16%, whereas the colorimetric method displayed a standard deviation in the 1–3% range, for 0.1 to 1.3 mg L⁻¹ H₂O₂ concentrations. Thus, it was proven that the colorimetric methodology provides more reliable results when measuring hydrogen peroxide concentrations.

Additional tests were performed to determine the accuracy of both methods (colorimetric and permanganometric), in the presence of an alcohol. In this case, ethanol was used, given that it is employed as a scavenger in the photocatalytic reaction. It was observed that the permanganometric method yielded a 43% standard deviation, while only a ±1% standard deviation was observed using the colorimetry method in the presence of ethanol. The lower reliability of the permanganometric titration was assigned to the ethanol scavenger interference. Thus, it was proven that the colorimetric methodology provides, in the present study, more reliable results when measuring hydrogen peroxide concentrations.

4.5.2. Effect of the pH on the Photocatalytic Reaction

The pH was measured with a digital pH meter Thermo Scientific Orion Star, with an accuracy of ±0.05. The pH was monitored in the slurry every hour, to determine its effect during the photocatalytic reaction.

4.6. Adsorption of Hydrogen Peroxide and Ethanol

The adsorption of hydrogen peroxide was carried out in the Photo-CREC Water-II Reactor at 25 ± 1 °C. Working conditions for the reactor were equivalent to the ones in the photocatalysis tests. First, the reactor was loaded with 6 L of water at certain reagent concentrations (0 to 1.3 ppm-H₂O₂). Following this, 0.15 g/L of the TiO₂ catalyst was added to the solution. The liquid slurry was recirculated for one hour to reach adsorption equilibrium. During this period, a liquid sample was taken every 10 min, and the H₂O₂ concentration in the liquid at equilibrium (Ce) was measured, using the colorimetric method described in Section 4.5.1 [22]. Based on the experimental data, the maximum adsorption capacity was given by the following relation:

$$Q_e=\frac{Q_{e,max}{\mathrm{KC}}_{\mathrm{e}}}{\left(1+{\mathrm{KC}}_{\mathrm{e}}\right)}$$

(17)

where Q_e is the H₂O₂ equilibrium adsorbent-phase concentration; C_e is the H₂O₂ equilibrium concentration in the liquid (mg L⁻¹); Q_e,max is the H₂O₂ maximum adsorption capacity (mg g⁻¹); and K is the adsorption constant [30].

5. Conclusions

Photocatalysis is a promising method for hydrogen production. This method involves the use of a semiconductor material that generates electron–hole pairs. Photoexcited electron–hole pairs are separated using sacrificial agents such as ethanol, which allow the formation of hydrogen and reduce electron–hole pair recombination. The mesoporous TiO₂ doped with a palladium semiconductor is a valuable photocatalyst for hydrogen production, while using ethanol as an organic scavenger. The synthesized 0.25 wt.% Pd-TiO₂ photocatalyst proved to be suitable for hydrogen production, in the Photo-CREC Water Reactor-II, under both near-UV and visible light, reaching a hydrogen volume of 113 and 29 cm³ STP, respectively, after 6 h of irradiation. The 0.25 wt.% Pd-TiO₂ photocatalyst displayed an in ‘‘series–parallel’’ reaction network, via water splitting, with hydrogen, methane, ethane, ethylene, acetaldehyde, carbon monoxide, and carbon dioxide products being formed. The accounting of all carbon containing species allowed one to establish a good carbon closure, with a slightly reduced ethanol concentration. OH^• and H^• radicals were consumed/reduced to form carbon containing by-products. OH^• radicals formed additional species such as H₂O₂, which was adsorbed by the photocatalyst, as well as dissociated, leading to pH changes throughout the reaction. The moles of OH^• and H^• radicals were evaluated, involving various redox species. This allowed one to confirm for both Pd-TiO₂-nUV and Pd-TiO₂-VIS, that the photocatalytic water splitting reaction proceeded via the equimolar formation of OH^• and H^• radicals.