IChO 2019 · Theory problem 2

Hydrogen production by water-splitting

Data:

Compound \(\ce{H2(g)}\) \(\ce{H2O(l)}\) \(\ce{H2O(g)}\) \(\ce{O2(g)}\)
\(\Delta_\mathrm{f}H^\circ\) (\(\mathrm{kJ\,mol^{-1}}\)) 0 −285.8 −241.8 0
\(S_\mathrm{m}^\circ\) (\(\mathrm{J\,mol^{-1}\,K^{-1}}\)) 130.6 69.9 188.7 205.2

Molecular hydrogen (\(\ce{H2}\)) can be used as an alternative to carbon dioxide-emitting fuels. Hence, lowering the cost and the environmental impact of its production is a major challenge. In this field, water-splitting is a promising candidate technology.

2.1 Write down the balanced equation of liquid water splitting reaction using a stoichiometric coefficient of 1 for water.

2.2 Using only the provided thermodynamic data, this justify numerically whether reaction is thermodynamically favorable at 298 K.

Water splitting can be performed electrochemically using two electrodes in an acidic water bath, connected by a generator (Fig. 1). Gas bubbles are formed at both electrodes.

Figure 1

Fig. 1. Water-splitting electrochemical cell.

2.3 Write down the balanced net electrochemical half reactions occurring at each electrode.

2.4 Using only the provided thermodynamic data (or question 2), derive the condition on the applied voltage \(\Delta E_\mathrm{applied}\) between electrodes, compared to value \(\Delta E_\mathrm{th}\) (to determine), for the process to be thermodynamically favorable at 298 K, when all reactants and products are in their standard state. Tick the right condition and give the numerical value with 3 decimal places.

Experimentally, a higher voltage is needed to observe water splitting. For a given \(\ce{Pt}\) cathode, the minimum voltage necessary to observe water splitting, \(\Delta E_\mathrm{min}\), depends on the nature of the anode, as displayed in the table below:

Anode \(\Delta E_\mathrm{min}\) (V)
\(\ce{IrO_x}\) 1.6
\(\ce{NiO_x}\) 1.7
\(\ce{CoO_x}\) 1.7
\(\ce{Fe2O3}\) 1.9

The difference between \(\Delta E_\mathrm{min}\) and \(\Delta E_\mathrm{th}\) is responsible for losses in the device.

2.5 Give the expression of the device power efficiency \(\eta_\mathrm{elec}\) (fraction of the power used for water splitting) as a function of \(\Delta E_\mathrm{th}\) and \(\Delta E_\mathrm{min}\). Assuming an identical current value \(I\), calculate the water electrolysis power efficiency when a \(\ce{Pt}\) cathode and a \(\ce{Fe2O3}\) anode are used. Give the most efficient anode.

An alternative to water electrolysis is direct photocatalytic water-splitting. It uses a semiconductor that can be activated by absorbing light.

Figure 2

Fig. 2: Activation condition and equivalent electrode potentials of different semiconductors. Dashed lines correspond to water oxidation and reduction potentials. SHE = Standard Hydrogen Electrode

Figure 3

Fig. 3 – Left axis: Spectral distribution of the solar photon flux \(\Phi\). The photon flux is the number of photons per unit area per unit time arriving on the semiconductor. Right axis and dashed line: cumulative photon flux (i.e. fraction of the photon flux with smaller wavelength).

2.6 Estimate the fraction of the solar photon flux that can activate the following semiconductors: \(\ce{TiO2}\), \(\ce{CdS}\), \(\ce{Si}\). State explicitly the equations and units used for the computation.

The activation of the semi-conductor results in a modification of the surface potentials, so that it can be seen as two electrodes of different potentials.

2.7 Using the data in Fig 2, choose the semiconductor(s) in the following list that, once activated, can play both roles of anode and cathode for water-splitting reaction.

2.8 Give the semiconductor that, used as both cathode and anode, is expected to be the most efficient for water splitting upon a given solar shining.

The evolution of \(\ce{H2}\) and \(\ce{O2}\) when a semiconductor is irradiated by simulated solar light at \(T=25\,^\circ\mathrm{C}\) at \(p_\mathrm{atm}\) was recently studied. Using an incident power light of \(P=1.0\,\mathrm{kW\,m^{-2}}\) and a photoelectrode with a \(S=16\,\mathrm{mm^2}\) surface, the production of \(V=0.37\,\mathrm{cm^3}\) of \(\ce{H2(g)}\) was measured after \(\Delta t=1\) hour of reaction.

2.9 Calculate the power efficiency \(\eta_\mathrm{direct}\) of the conversion.

Two modes of converting solar energy to hydrogen can thus be compared: direct photocatalysis, and indirect photo-electrolysis combining a photovoltaic panel with an electrolyzer. The efficiency of photovoltaic panels on the market is around \(\eta_\mathrm{panels}=20\%\).

2.10 Compare the power efficiencies of the two modes, \(\eta_\mathrm{direct}\) and \(\eta_\mathrm{indirect}\), using \(\ce{Fe2O3}\) and \(\ce{Pt}\) electrodes for the electrolysis.

Official English text of the problem.

Answers

Molecular and ionic species used below are water, SMILES O; dihydrogen, SMILES [H][H]; dioxygen, SMILES O=O; and the hydrogen ion (proton), SMILES [H+]. These SMILES do not specify physical phase, aqueous solvation, or the triplet spin state of dioxygen. The electrode and semiconductor materials are extended solids, for which a molecular SMILES does not specify the crystal structure.

  • 2.1 \[ \ce{H2O(l) -> H2(g) + 1/2 O2(g)} \]

  • 2.2 \(\Delta_\mathrm{r}G^\circ(298\,\mathrm K)=+237.1\,\mathrm{kJ\,mol^{-1}}\). The reaction is not thermodynamically favorable under standard conditions.

  • 2.3 Electrode (1) is the negative cathode: \[ \ce{2H+(aq) + 2e- -> H2(g)} \] Electrode (2) is the positive anode: \[ \ce{2H2O(l) -> O2(g) + 4H+(aq) + 4e-} \]

  • 2.4 \[ \boxed{\Delta E_\mathrm{applied}>\Delta E_\mathrm{th}, \qquad \Delta E_\mathrm{th}=1.229\,\mathrm V} \] Equality is the reversible limit, with no net thermodynamic driving force.

  • 2.5 \[ \boxed{\eta_\mathrm{elec}= \frac{\Delta E_\mathrm{th}}{\Delta E_\mathrm{min}}} \] For \(\ce{Pt}/\ce{Fe2O3}\), \(\eta_\mathrm{elec}\approx64.7\%\). The most efficient anode is iridium oxide, \(\ce{IrO_x}\), giving approximately \(76.8\%\).

  • 2.6

Semiconductor Maximum activating wavelength Activating fraction of photon flux
Titanium dioxide, \(\ce{TiO2}\) \(413\,\mathrm{nm}\) approximately \(4\%\)
Cadmium sulfide, \(\ce{CdS}\) \(517\,\mathrm{nm}\) approximately \(12\%\)
Silicon, \(\ce{Si}\) \(1127\,\mathrm{nm}\) approximately \(65\%\)
  • 2.7 Among all the materials in Fig. 2, those capable of both half-reactions are \[ \boxed{\ce{ZrO2},\ \ce{ZnS},\ \ce{KTaO3},\ \ce{ZnO},\ \ce{SrTiO3},\ \ce{TiO2},\ \ce{SiC},\ \ce{CdS}}. \]

  • 2.8 Cadmium sulfide, \(\ce{CdS}\).

  • 2.9 \[ \boxed{\eta_\mathrm{direct}\approx6.2\%} \] using \(p_\mathrm{atm}=1\,\mathrm{atm}\).

  • 2.10 \[ \boxed{\eta_\mathrm{indirect} =\eta_\mathrm{panels}\eta_\mathrm{elec} \approx12.9\%} \] The indirect route is approximately 2.1 times as efficient as the measured direct route.

Solution

2.1 — Overall water-splitting reaction

Liquid water forms gaseous hydrogen and oxygen. With the coefficient of water fixed at one, atom balance gives

\[ \boxed{\ce{H2O(l) -> H2(g) + 1/2 O2(g)}}. \]

All molar reaction quantities below refer to this equation: one mole of liquid water consumed and one mole of hydrogen produced.

2.2 — Thermodynamic favorability at 298 K

The standard reaction enthalpy is

\[ \begin{aligned} \Delta_\mathrm{r}H^\circ &=\Delta_\mathrm{f}H^\circ(\ce{H2(g)}) +\frac12\Delta_\mathrm{f}H^\circ(\ce{O2(g)}) -\Delta_\mathrm{f}H^\circ(\ce{H2O(l)})\\ &=0+\frac12(0)-(-285.8)\\ &=+285.8\,\mathrm{kJ\,mol^{-1}}. \end{aligned} \]

The standard reaction entropy is

\[ \begin{aligned} \Delta_\mathrm{r}S^\circ &=130.6+\frac12(205.2)-69.9\\ &=+163.3\,\mathrm{J\,mol^{-1}\,K^{-1}}\\ &=+0.1633\,\mathrm{kJ\,mol^{-1}\,K^{-1}}. \end{aligned} \]

Therefore,

\[ \begin{aligned} \Delta_\mathrm{r}G^\circ &=\Delta_\mathrm{r}H^\circ-T\Delta_\mathrm{r}S^\circ\\ &=285.8-298(0.1633)\\ &=\boxed{+237.1366\,\mathrm{kJ\,mol^{-1}}}. \end{aligned} \]

Since \(\Delta_\mathrm{r}G^\circ>0\), standard-state water splitting is thermodynamically unfavorable at \(298\,\mathrm K\). Energy must be supplied to drive it.

2.3 — Electrode half-reactions

In Fig. 1, electrode (1) is connected to the generator’s negative terminal. It receives electrons and is the cathode, where hydrogen ions are reduced:

\[ \boxed{\ce{2H+(aq) + 2e- -> H2(g)}}. \]

Electrode (2) is connected to the positive terminal. It is the anode, where water is oxidized:

\[ \boxed{\ce{2H2O(l) -> O2(g) + 4H+(aq) + 4e-}}. \]

Multiplying the cathode reaction by two and adding cancels both electrons and hydrogen ions:

\[ \ce{2H2O(l) -> 2H2(g) + O2(g)}. \]

Dividing by two reproduces the equation in 2.1.

2.4 — Thermodynamic threshold voltage

Producing one mole of hydrogen transfers two moles of electrons, corresponding to charge

\[ q=2F, \]

where \(F=96485\,\mathrm{C\,mol^{-1}}\).

The reversible electrical energy required per mole is therefore

\[ 2F\Delta E_\mathrm{th}=\Delta_\mathrm{r}G^\circ. \]

Using \(1\,\mathrm V=1\,\mathrm{J\,C^{-1}}\),

\[ \begin{aligned} \Delta E_\mathrm{th} &=\frac{237136.6\,\mathrm{J\,mol^{-1}}} {2(96485\,\mathrm{C\,mol^{-1}})}\\ &=1.22888\,\mathrm V\\ &=\boxed{1.229\,\mathrm V}. \end{aligned} \]

With an applied voltage supplying electrical work, the net free-energy change per mole is

\[ \Delta G_\mathrm{net} =\Delta_\mathrm{r}G^\circ-2F\Delta E_\mathrm{applied}. \]

For a thermodynamic driving force in the water-splitting direction,

\[ \Delta G_\mathrm{net}<0, \]

so the condition to select is

\[ \boxed{\Delta E_\mathrm{applied}>\Delta E_\mathrm{th}}. \]

At equality, electrolysis is at its reversible threshold.

2.5 — Electrolyzer power efficiency

At current \(I\), the electrical input power at the observed minimum voltage is

\[ P_\mathrm{input}=I\Delta E_\mathrm{min}. \]

The useful power associated with the Gibbs-energy increase of the products is

\[ P_\mathrm{useful}=I\Delta E_\mathrm{th}. \]

Thus,

\[ \boxed{\eta_\mathrm{elec} =\frac{I\Delta E_\mathrm{th}}{I\Delta E_\mathrm{min}} =\frac{\Delta E_\mathrm{th}}{\Delta E_\mathrm{min}}}. \]

For a platinum cathode and iron(III) oxide anode,

\[ \eta_\mathrm{elec} =\frac{1.22888}{1.9} =0.6468 \approx\boxed{64.7\%}. \]

The most efficient anode has the lowest \(\Delta E_\mathrm{min}\). The table gives this as iridium oxide, \(\ce{IrO_x}\):

\[ \eta_\mathrm{elec}(\ce{IrO_x}) =\frac{1.22888}{1.6} =0.7680 \approx\boxed{76.8\%}. \]

2.6 — Fraction of sunlight capable of activation

A photon activates the semiconductor if its energy reaches the minimum activation energy \(\varepsilon\):

\[ E_\gamma=h\nu=\frac{hc}{\lambda}\geq\varepsilon. \]

Hence the activating wavelengths satisfy

\[ \boxed{\lambda\leq\lambda_\mathrm{max}=\frac{hc}{\varepsilon}}. \]

Using

\[ h=6.626\times10^{-34}\,\mathrm{J\,s},\qquad c=2.998\times10^8\,\mathrm{m\,s^{-1}}, \]

and

\[ 1\,\mathrm{eV}=1.602\times10^{-19}\,\mathrm J, \]

the convenient numerical form is

\[ \boxed{\lambda_\mathrm{max}\,[\mathrm{nm}] =\frac{1239.84}{\varepsilon\,[\mathrm{eV}]}}. \]

From Fig. 2:

\[ \begin{aligned} \ce{TiO2}:&\quad \varepsilon=3.0\,\mathrm{eV}, &\lambda_\mathrm{max}&=\frac{1239.84}{3.0}=413\,\mathrm{nm},\\ \ce{CdS}:&\quad \varepsilon=2.4\,\mathrm{eV}, &\lambda_\mathrm{max}&=\frac{1239.84}{2.4}=517\,\mathrm{nm},\\ \ce{Si}:&\quad \varepsilon=1.1\,\mathrm{eV}, &\lambda_\mathrm{max}&=\frac{1239.84}{1.1}=1127\,\mathrm{nm}. \end{aligned} \]

If \(\Phi(\lambda)\) is the spectral photon flux in \(\mathrm{photons\,m^{-2}\,s^{-1}\,nm^{-1}}\), the activating fraction is

\[ f= \frac{\displaystyle\int_0^{\lambda_\mathrm{max}}\Phi(\lambda)\,d\lambda} {\displaystyle\int_0^\infty\Phi(\lambda)\,d\lambda}. \]

Both integrals have units \(\mathrm{photons\,m^{-2}\,s^{-1}}\), so \(f\) is dimensionless. This fraction is given directly by the dashed cumulative curve in Fig. 3.

Reading that curve at the three threshold wavelengths gives approximately

\[ \boxed{ f_{\ce{TiO2}}\approx4\%,\qquad f_{\ce{CdS}}\approx12\%,\qquad f_{\ce{Si}}\approx65\%. } \]

These are fractions of photon flux, rather than fractions of incident radiant power.

2.7 — Semiconductors capable of both half-reactions

For hydrogen evolution, the equivalent cathode must have a potential more negative than the hydrogen-ion/hydrogen couple:

\[ E_\mathrm{cathode} <E^\circ_{\ce{H+}/\ce{H2}} =0\,\mathrm V\quad\text{vs SHE}. \]

For oxygen evolution, the equivalent anode must have a potential more positive than the oxygen/water couple:

\[ E_\mathrm{anode} >E^\circ_{\ce{O2}/\ce{H2O}} \approx+1.229\,\mathrm V\quad\text{vs SHE}. \]

Thus, the semiconductor’s potential interval must span both dashed lines in Fig. 2. Because the numerical potential increases downward in the figure, its upper endpoint must lie above the hydrogen line and its lower endpoint below the oxygen line.

Materials in Fig. 2 Hydrogen evolution possible? Oxygen evolution possible? Both possible?
\(\ce{ZrO2}\), \(\ce{ZnS}\), \(\ce{KTaO3}\), \(\ce{ZnO}\), \(\ce{SrTiO3}\), \(\ce{TiO2}\), \(\ce{SiC}\), \(\ce{CdS}\) Yes Yes Yes
\(\ce{WO3}\), \(\ce{Fe2O3}\), \(\ce{MoS2}\) No: equivalent cathode potential is too positive Yes No
\(\ce{CdSe}\), \(\ce{Si}\) Yes No: equivalent anode potential is insufficiently positive No

The complete qualifying set is therefore zirconium dioxide, zinc sulfide, potassium tantalate, zinc oxide, strontium titanate, titanium dioxide, silicon carbide, and cadmium sulfide.

2.8 — Most efficient semiconductor under the stated model

The semiconductor must first satisfy both potential conditions in 2.7. Among the qualifying materials, cadmium sulfide has the smallest activation energy:

\[ \varepsilon_{\ce{CdS}}=2.4\,\mathrm{eV}. \]

It therefore has the longest activation cutoff wavelength and can use the largest fraction of the incident solar photon flux among those materials. Under the comparison implied by the problem,

\[ \boxed{\text{the expected most efficient semiconductor is }\ce{CdS}.} \]

Although silicon absorbs a larger fraction of solar photons, its equivalent anode potential does not permit water oxidation, so it cannot serve both roles.

2.9 — Direct solar-to-hydrogen efficiency

For consistency with the efficiency definition in 2.5, the useful chemical energy is the Gibbs-energy increase, \(\Delta_\mathrm{r}G^\circ\), per mole of hydrogen produced.

Treat the hydrogen as an ideal gas. Using

\[ p_\mathrm{atm}=101325\,\mathrm{Pa},\qquad T=298.15\,\mathrm K,\qquad R=8.314\,\mathrm{J\,mol^{-1}\,K^{-1}}, \]

and converting the volume,

\[ V=0.37\,\mathrm{cm^3}=3.7\times10^{-7}\,\mathrm{m^3}, \]

gives

\[ \begin{aligned} n(\ce{H2}) &=\frac{p_\mathrm{atm}V}{RT}\\ &=\frac{(101325)(3.7\times10^{-7})} {(8.314)(298.15)}\\ &=1.51\times10^{-5}\,\mathrm{mol}. \end{aligned} \]

The useful chemical energy produced is

\[ \begin{aligned} E_\mathrm{chemical} &=n(\ce{H2})\Delta_\mathrm{r}G^\circ\\ &=(1.512\times10^{-5})(237136.6)\\ &\approx3.59\,\mathrm J. \end{aligned} \]

The incident irradiance, area, and exposure time are

\[ P=1.0\times10^3\,\mathrm{W\,m^{-2}},\qquad S=16\times10^{-6}\,\mathrm{m^2},\qquad \Delta t=3600\,\mathrm s. \]

Therefore,

\[ E_\mathrm{incident}=PS\Delta t =(1.0\times10^3)(16\times10^{-6})(3600) =57.6\,\mathrm J. \]

The ratio of average powers equals the ratio of energies over the same interval:

\[ \begin{aligned} \eta_\mathrm{direct} &=\frac{n(\ce{H2})\Delta_\mathrm{r}G^\circ}{PS\Delta t}\\ &=\frac{3.59}{57.6}\\ &=0.0623. \end{aligned} \]

Thus,

\[ \boxed{\eta_\mathrm{direct}\approx6.2\%}. \]

2.10 — Comparison with photovoltaic-powered electrolysis

The indirect route consists of two successive conversions: sunlight to electricity, then electricity to chemical energy. Its overall efficiency is the product:

\[ \eta_\mathrm{indirect} =\eta_\mathrm{panels}\eta_\mathrm{elec}. \]

For the stated photovoltaic panels and platinum/iron(III) oxide electrolyzer,

\[ \begin{aligned} \eta_\mathrm{indirect} &=0.20\left(\frac{1.22888}{1.9}\right)\\ &=0.1294\\ &\approx\boxed{12.9\%}. \end{aligned} \]

Consequently,

\[ \frac{\eta_\mathrm{indirect}}{\eta_\mathrm{direct}} =\frac{0.1294}{0.0623} \approx2.1. \]

Under the supplied conditions, the indirect route is approximately twice as efficient as the measured direct photocatalytic route.

The solution was checked against the official solution.