membranes Article

Polymer Electrolyte Membranes for Water Photo-Electrolysis Antonino S. Aricò 1, *, Mariarita Girolamo 1 , Stefania Siracusano 1 , David Sebastian 1 , Vincenzo Baglio 1 and Michael Schuster 2 1

2

*

CNR-ITAE Institute for Advanced Energy Technologies “N. Giordano”, Via Salita S. Lucia sopra Contesse 5, 98126 Messina, Italy; [email protected] (M.G.); [email protected] (S.S.); [email protected] (D.S.); [email protected] (V.B.) FUMATECH BWT GmbH Gesellschaft für Funktionelle Membranen und Anlagentechnologie mbH, Carl-Benz-Strasse 4, Bietigheim-Bissingen, D-74321 Baden-Württemberg, Germany; [email protected] Correspondence: [email protected]; Tel.: +39-090-624-237

Academic Editor: Tongwen Xu Received: 22 December 2016; Accepted: 25 April 2017; Published: 29 April 2017

Abstract: Water-fed photo-electrolysis cells equipped with perfluorosulfonic acid (Nafion® 115) and quaternary ammonium-based (Fumatech® FAA3) ion exchange membranes as separator for hydrogen and oxygen evolution reactions were investigated. Protonic or anionic ionomer dispersions were deposited on the electrodes to extend the interface with the electrolyte. The photo-anode consisted of a large band-gap Ti-oxide semiconductor. The effect of membrane characteristics on the photo-electrochemical conversion of solar energy was investigated for photo-voltage-driven electrolysis cells. Photo-electrolysis cells were also studied for operation under electrical bias-assisted mode. The pH of the membrane/ionomer had a paramount effect on the photo-electrolytic conversion. The anionic membrane showed enhanced performance compared to the Nafion® -based cell when just TiO2 anatase was used as photo-anode. This was associated with better oxygen evolution kinetics in alkaline conditions compared to acidic environment. However, oxygen evolution kinetics in acidic conditions were significantly enhanced by using a Ti sub-oxide as surface promoter in order to facilitate the adsorption of OH species as precursors of oxygen evolution. However, the same surface promoter appeared to inhibit oxygen evolution in an alkaline environment probably as a consequence of the strong adsorption of OH species on the surface under such conditions. These results show that a proper combination of photo-anode and polymer electrolyte membrane is essential to maximize photo-electrolytic conversion. Keywords: anion exchange polymer electrolyte membrane; proton exchange polymer electrolyte membrane; water splitting; photo-electrolysis; TiO2 ; Ti-suboxides

1. Introduction The continuous increase of energy consumption and the associated increase of greenhouse gas emissions causing global warming have produced significant concerns. It is widely recognised that there is an urgent need to increase the level of utilisation renewable sources in order to address the environmental issues. Hydrogen is an alternative fuel and, according to its high energy density and clean combustion, can represent a suitable energy vector mediating between renewable sources and sustainable mobility [1–4]. The energy associated with solar radiation striking the earth is several orders of magnitude the global energy consumption. An efficient and wide-scale use of sunlight can occur through a decentralised conversion into hydrogen that can be stored and used as fuel in several applications including highly efficient fuel cell devices for stationary generation, transportation and portable power [4]. Membranes 2017, 7, 25; doi:10.3390/membranes7020025

www.mdpi.com/journal/membranes

Membranes 2017, 7, 25

2 of 16

The direct conversion of solar energy to hydrogen represents a long-term objective for the next generation energy system based on renewable sources. In this context, hydrogen generation from photo-electrochemical water splitting using solar energy is of primary interest. Significant progress has been recently achieved for the photo-electrolytic hydrogen production through the development of advanced water splitting semiconductors. However, this progress remains at the level of laboratory curiosity since relevant aspects dealing with practical development of the photo-electrolysis cell devices, in particular the separator system, have not been addressed with the same efforts. Without a proper development of the technology around the water splitting semiconductors, these systems can hardly became economically competitive with respect to current technologies involving the extraction of fossil fuels and their internal combustion [1–3]. As discussed above, most of the efforts made so far have been addressed to the development of new photo-catalytic materials with better efficiencies but often operating under voltage bias-assisted mode. However, to reach the level of commercial viability for the water photo-electrolysis technology, the efforts should also cover the other cell components and provide characterisation in a real device. According to several literature reports [3], it appears that water splitting semiconductors characterised by improved efficiency are often consisting of unstable materials. Some interesting semiconductors showing good electronic properties and proper capability of harvesting the visible fraction of solar radiation have been assessed just in half-cells in the presence of corrosive liquid electrolytes. The latter can produce a chemical attack, thus diminishing the endurance potentialities of the novel materials, whereas solid polymer electrolytes usually stabilise the electrolysis materials against corrosion [5]. It is widely recognised that, in order to further advance the photo-electrolysis technology and favour market diffusion, a parallel development of reliable membrane separators is required. Moreover, the utilisation of appropriate ionomer dispersion at the electrode—electrolyte interface [5,6], providing suitable pH for the photo-electrolysis environment to maximise solar-to-hydrogen-efficiency and offering good stability while being cost-effective, is also relevant. The membrane separator is an important component of electrolysis and photo-electrolysis systems thus avoids recombination of the formed gas products. It provides a pathway for ion conduction, and, together with the ionomer dispersion used in the electrodes, establishes the operating pH of the photo-electrolytic device when this is fed with pure water as largely preferred from a technological viewpoint [3,7]. The membrane and ionomer dispersions are thus part of the advanced device engineering required to maximise the overall performance of the photo-electrolysis system and to prevent corrosion especially in the case of semiconductors with relatively low energy gap [3,7]. A simple configuration for a photo-electrolysis cell consists of a semiconductor photo-anode for the oxygen evolution with relatively large energy gap and a Pt-based counter electrode for hydrogen evolution [8,9]. The photo-anode is illuminated, whereas the Pt cathode operates in the dark. The energy gap of a photo-anode must be larger than that of the energy required for splitting the water molecule into hydrogen and oxygen gases (1.23 eV) and the electronic levels (valence band and conduction band edges) should properly match with the electronic levels associated with the redox couples (half-cell reactions) involved in water splitting [10]. Accordingly, most of the photo-electrolysis cells use large gap semiconductors e.g., TiO2 , thus absorbing mainly a small UV portion of the solar spectrum [3,7]. This approach does not allow achieving high levels of efficiency and has limited practical interest. However, large band gap TiO2 (3.2 eV) has been widely investigated as a model system and can be still used to provide a baseline of comparison when other cell components are developed e.g., alternative cathodes and membranes.‘The research trend is to use semiconductors with a smaller energy gap in order to harvest a good fraction of the visible region while keeping the band gap above 1.23 eV [3,7]. Other approaches deal with combining the photo-electrolysis cell with a Dye Sensitized Solar Cells (DSSC) [11]. In the latter case, the visible light reaches the DSSC cell and is converted into electricity that can be utilised to help water splitting in the photo-electrolysis cell using the UV fraction of sunlight [11].

Membranes 2017, 7, 25

3 of 16

TiO2 anode semiconductors are characterised by suitable electronic band structure for water splitting in combination with platinum cathodes in the photo-electrolysis cell [3,7,10]. However, the energy gap of 3.2 eV in TiO2 limits significantly the level of efficiency in photo-electrochemical systems. Despite this aspect, TiO2 represents a model system for photo-electrolysis device being quite stable in all operating conditions and highly photoactive in the specific operating region of the solar spectrum. The wide energy gap also allows the occurrence of a spontaneous photo-voltage [11]. Besides the energy implications, e.g., no need of external electric bias, this aspect can provide insights into the capability of the membrane to separate the formed gas (i.e., direct recombination causing occurrence of mixed potentials is avoided). It is well known that the photo-electrocatalytic properties of titanium dioxide depend on the operating environment, which is similar to most of the water splitting semiconductors [7]. The oxygen evolution process involves a charge transfer between photo-generated carriers at the surface of the TiO2 photo-anode and the adsorbed species on the electrode surface, e.g., OH species, which depend very much on the electrolyte properties [3]. The semiconductor/electrolyte interface plays an important role in determining the efficiency and stability of the water photo-electrolysis process [3,7]. This work is aiming at comparing the behaviour of commercial perfluorosulfonic acid (Nafion® 115) and quaternary ammonium-based anionic (Fumatech® FAA3) ion exchange membranes in photo-electrolysis cells fed with pure water. The membrane has the primary role of acting as separator of the hydrogen and oxygen [5,6] evolved at the photo-anode and cathode surface, respectively, and to provide continuity for the transfer of ions (protons or hydroxides) between the two cell compartments. Protonic and alkaline ionomer dispersions (micelles) with the same chemistry of the membranes were used at the cathode and photo-anode to extend the interface between electrodes and electrolyte; these ionomer dispersions contribute together with the membrane to characterize the cell environment in terms of pH and process characteristics (adsorption of active species, ionic conduction-type, ionic conductivity, etc.). High surface area n-type commercial TiO2 powder [12] was used as model photo-anode to allow a fair comparison of the membrane characteristics. Modification of the photo-anode surface with a Ti-suboxide layer [13] was also carried out to derive information on the influence of the electrolyte environment on different surface characteristics of the photo-anode for the water splitting process. 2. Results and Discussion 2.1. Photo-Electrolysis Cell Based on Ion Exchange Membrane Separator A sketch of the photo-electrolysis cell fed with pure deaerated water is represented in Figure 1 together with the energy diagram related to the water splitting process for this device. The cell consists of a glass coated with fluorine doped tin oxide SnO2 :F forming a transparent conductive oxide (TCO) as support of the TiO2 photo-anode (Figure 1, top). The latter is coated with the ionomer dispersion (a solution containing the solid polymer electrolyte micelles, which is deposited on the electrode surface and successively dried to favour micelles adsorption on the electrode layer) to extend the reaction region at the interface and to provide a path for ion percolation. The polymer electrolyte membrane separates the photo-anode from the Pt counter electrode (Figure 1, top). To allow water feed and gas escape from the cell, a segmented sub-gasket is used between the membrane and the electrodes, whereas water is injected through a hole drilled on the backside of the counter electrode. The segmented sub-gasket used between the membrane and the photo-electrode is made of the same ionically conductive polymer as the membrane. This assures the ionic percolation. The free space between each electrode and membrane is filled with water and hydrophilic ion conductive ionomer to favour ionic percolation. The counter electrode is formed by Pt black particles deposited on a TCO.

Membranes 2017, 7, 25 Membranes 2017, 7, 25

4 of 16 4 of 15

TCO

TiO2

Gases

Hydrophilic Membrane Pt

Iphotocurrent

hν

Hydrophilic Ionomer + H2O

Vphotovoltage

TCO

Water injection

-2 -1

e-

0

eH2O/H2

1 h

O2/H2O

2

Potential Energy

Potential / V vs. RHE

Segmented subgasket for sealing and gas release

3

h+

of of thethe photo-electrolysis cell cell (top)(top) and energy diagram for thefor water process Figure 1. 1. Sketch Sketch photo-electrolysis and energy diagram thesplitting water splitting process under shortthe circuit; the segmented ionically conductive polymer usedgas to favour gas proton escape under short circuit; segmented ionically conductive polymer used to favour escape and and proton percolation shown theFermi right;levels the Fermi levelsinare in the semiconductor and percolation is shown onisthe right;onthe are equal theequal semiconductor and electrolyte electrolyte (bottom). (bottom).

The principle principle of The of operation operation of of the the photo-electrolysis photo-electrolysis cell cell is is also also represented represented in in Figure Figure 11 (bottom). (bottom). Solar irradiation leads to a generation of electron-hole pairs when the wavelength energy is larger Solar irradiation leads to a generation of electron-hole pairs when the wavelength energy is larger than thanenergy the energy gap the semiconductor Eg). These are separated by electric the electric of the gap of theofsemiconductor (hν >(hν Eg ).> These pairspairs are separated by the fieldfield of the the space charge region [3,7]. The holes reaching the semiconductor-electrolyte interface are space charge region [3,7]. The holes reaching the semiconductor-electrolyte interface are consumed by consumed the oxygen evolution, whereas electrons, which are forced towards the semiconductor the oxygenby evolution, whereas electrons, which are forced towards the semiconductor back contact, back contact, are transferred, through the external electric circuit, to the Pt cathode the are transferred, through the external electric circuit, to the Pt cathode where the hydrogenwhere evolution hydrogen evolution occurs in the dark. occurs in the dark. 2.2. Physico-Chemical Characterization of the Membrane ® 115) and quaternary ammonium-based anionic Commercial perfluorosulfonic acid (Nafion® quaternary ammonium-based ® ® (Fumatech, Fumion FumionFumasep Fumasep FAA-3-20) FAA-3-20) exchange membranes purified use (a (Fumatech, ionion exchange membranes werewere purified beforebefore use (a picture picture the as received membranes shown Figure 2). of the asofreceived membranes is shownisin Figurein2). The main properties present work areare presented in properties of ofthe thetwo twodifferent differentmembranes membranesused usedininthe the present work presented Table 1. 1. The in Table Thenature natureofofpolymer polymerchemistry, chemistry, equivalent equivalent weight weight (EW), (EW), membrane membrane thickness, ion conductivity are reported. It is clearly observed that the perfluorosulfonic membrane is characterised by better ionic conductivity than the anionic membrane, despite the fact that equivalent weight is larger for Nafion®® . Of course, the proton diffusivity is much higher than the diffusivity of hydroxide ® backbone enhances dissociation of anions; moreover, moreover,thethe presence of fluorine the ® Nafion presence of fluorine in thein Nafion backbone enhances dissociation of hydrogen hydrogen of thesulfonic terminal sulfonic groups the and sidethe chain and separation. the charge separation. Charge ions of theions terminal groups in the side in chain charge Charge separation separation is also in the anionic membrane tocharacter the ionic character of the bond terminal between is also relevant inrelevant the anionic membrane due to thedue ionic of the bond between terminal ammonium groups andcompensate OH−. To compensate the lower the conductivity, the anionic ammonium groups and OH− . To for the lowerfor conductivity, anionic membrane was ® ® membrane was selected significantly thinner the Nafionthis membrane; this wasof also light of the selected significantly thinner than the Nafionthan membrane; was also in light theinfact that fact thatmembranes the anionic are membranes are usually characterised lower gas cross-over [14]. anionic usually characterised by lower gasby cross-over [14].

Membranes 2017, 7, 25

5 of 16

Membranes 2017, 7, 25

5 of 15

(a)

(b)

® 115 (a) and Fumion® FAA3-20 ionomer membranes (b) prior Figure 2. Photographs Photographs of Nafion ® 115 Figure 2. of Nafion (a) and Fumion® FAA3-20 ionomer membranes (b) prior purification. purification.

Table 1. Ex situ characterisation data of the Fumatech® anionic FAA3-20 membrane compared to Table 1. Ex situ characterisation data of the Fumatech® anionic FAA3-20 membrane compared to Nafion® 115. Nafion® 115. Membrane Membrane Acronym Acronym

Unit Unit

® FAA3-20 Fumasep Fumasep® FAA3-20

® Nafion Nafion®-115 -115

Fumion Fumion ionomer ionomer Polyaromate Polyaromate Anion exchange Anion exchangematerial material Quaternary ammonium Quaternary ammoniumgroup group 555 555 1.8 1.8 20 20 6 (Cl (Cl−−) ) 30 (OH (OH−−) )

Nafion® Nafion ionomer ionomer Perfluorosulfonic acid acid Perfluorosulfonic Protonexchange exchange material material Proton Sulfonicacid acid group group Sulfonic 1100 1100 0.9 0.9 125 125

®®

Tradename name Trade

––

Polymer type type Polymer Conductivity type type Conductivity Terminalionic ionic groups groups Terminal EW(theoretical) (theoretical) EW IEC (exp.) (exp.) IEC Thickness (dry) Thickness

–– –– –– g/mol g/mol mmol/g mmol/g μm µm

◦C ConductivityininH H22O O at T == 25 Conductivity 25°C

mS cm mS cm−1 −1

®

62(H (H++) 62

2.3. 2.3. Photo-Electrode Photo-Electrode Characterisation Characterisation According According to to the the X-ray X-ray diffraction diffraction analysis analysis (Figure (Figure 3), 3), the the TiO TiO22 Degussa Degussa P90 P90 powder powder (Frankfurt, (Frankfurt, Germany) used as photo-anode precursor mainly showed the presence of an anatase phase. Germany) used as photo-anode precursor mainly showed the presence of an anatase phase. The The main main characteristic peak at 25.4° 2θ was assigned to the (101) Miller index of anatase (JCPDS schedule: 21◦ characteristic peak at 25.4 2θ was assigned to the (101) Miller index of anatase (JCPDS schedule: 1272) [12,15]. There is a small amount of evidence of the rutile structure (JCPDS schedule: 21-1276) 21-1272) [12,15]. There is a small amount of evidence of the rutile structure (JCPDS schedule: 21-1276) with peaks at at 27 27° (110) and and 37 37° the occurrence occurrence of of small small percentages percentages of of brookite brookite (JCPDS (JCPDS ◦ (110) ◦ (101) with peaks (101) 2θ; 2θ; the schedule: 16-617) is not excluded [12]. According to the broadening of the X-ray diffraction peaks, schedule: 16-617) is not excluded [12]. According to the broadening of the X-ray diffraction peaks, the the mean crystallite size for this sample, related to the anatase structure, was about 15 nm as determined mean crystallite size for this sample, related to the anatase structure, was about 15 nm as determined from theScherrer Scherrerequation. equation. Chemical reduction ofTiO the phase TiO2 phase to the ◦ C1050 from the Chemical reduction of the at 1050at gives°C risegives to therise formation 2 formation of a mixture of sub-stoichiometric Ti-oxides [13]. The occurrence of these processes of a mixture of sub-stoichiometric Ti-oxides [13]. The occurrence of these processes was confirmedwas by confirmed by XRD analysis (Figure 3) showing typical peaks of Ti nO2n−1 (Magneli phase). In particular, XRD analysis (Figure 3) showing typical peaks of Tin O2n−1 (Magneli phase). In particular, the XRD the XRD of patterns of the Ti-suboxide 3) the indicates the occurrence Ti9O1718-1405) (JCPDS:structure 18-1405) patterns the Ti-suboxide (Figure 3)(Figure indicates occurrence of the Ti9 Oof17the (JCPDS: structure at 24°–24.5°. The Ti 6 O 11 phase (JCPDS: 18-1401) was also present, whereas the Ti 4 O 7 at 24◦ –24.5◦ . The Ti6 O11 phase (JCPDS: 18-1401) was also present, whereas the Ti4 O7 (JCPDS: (JCPDS: 18-1402) 18-1402) was essentially present in traces. Thus, there was a high degree of sub-stoichiometry in the was essentially present in traces. Thus, there was a high degree of sub-stoichiometry in the Ti-oxide Ti-oxide reduced at high temperature. No evidence of peaks of Anatase or Rutile was observed the reduced at high temperature. No evidence of peaks of Anatase or Rutile was observed in thein XRD XRD pattern the reduced Ti-oxide. the assigned reflections the Magneli pattern of theof reduced Ti-oxide. All ofAll theof assigned reflections belongbelong to the to Magneli phase.phase.

Membranes 2017, 2017, 7, 7, 25 25 Membranes Membranes 2017, 7, 25

of 16 15 66 of 6 of 15 1200 1200

/ a.u. Counts / a.u. Counts

1000 1000

TiO2 P90 TiO2 P90

800 800 600 600 400 400

Ti-suboxide (TiO2-x) Ti-suboxide (TiO2-x)

200 200 0 0 20 20

30 30

40 50 60 40 Two Theta 50 Degrees 60 2θ ( ) Two Theta Degrees

70 70

80 80

2θ ( )

Figure 3. X-ray diffraction patterns of a bare P90 TiO2 and Ti-sub-oxide powders used for the Figure 3.3.X-ray patterns of a bare TiO Ti-sub-oxide powders used for theused photoanode. X-raydiffraction diffraction patterns of a P90 bare P90 TiO 2 and Ti-sub-oxide powders for the 2 and photoanode. photoanode.

of P90 P90 was was around around 90 90m m22/g /g according to the The Brunauer–Emmett–Teller (BET) surface area of 2/g according to the The Brunauer–Emmett–Teller (BET) surface area of P90 was around 90 m supplier, whereas the measured BET surface area of the in-house prepared Ti sub-oxide was26 26m m22/g. /g. supplier, whereas the measured BET surface area of the in-house prepared Ti sub-oxide was 2/g. supplier, whereas the measured BET surface area of the in-house prepared Ti sub-oxide was 26 m ◦ This relevant difference in surface area is essentially related to the sub-oxide reduction at 1050 °C, C, This relevant difference in surface area is morphology essentially related toofthe atwas 1050also °C, which promotes particle sintering [13]. The morphology thesub-oxide Ti-oxidereduction materials which promotes particle sintering [13]. The of the Ti-oxide materials was also investigated which particle sintering [13]. The The morphology of the Ti-oxide materials wassmall also investigated by TEM analysis (Figure 4a,b). P90 Ti-oxide showed spherical particles with by TEMpromotes analysis (Figure 4a,b). The P90 Ti-oxide showed spherical particles with relatively investigated by TEM analysis (Figure 4a,b). The P90 Ti-oxide showed spherical particles with relatively small particle (around nm), whereas Ti-suboxide, due to treatment, the high particle size (around 15–20size nm), whereas15–20 the Ti-suboxide, due the to the high temperature relatively small particle size (around 15–20 nm), whereas the Ti-suboxide, due to 25–50 the high temperature treatment, showed largewith particle agglomerates with primary particles of about nm showed large particle agglomerates primary particles of about 25–50 nm in size. However, the temperature treatment, showed large particle agglomerates with primary particles of about 25–50 nm in size. However, the high Ti-suboxide highwhich surface which is possibly associated with Ti-suboxide showed surface showed roughness, is roughness, possibly associated with a large fraction of in size. fraction However, Ti-suboxide a large surface defects. showed high surface roughness, which is possibly associated with surface defects. of the a large fraction of surface defects.

TiO P90 TiO22 P90

Ti-suboxide (TiO2-x) Ti-suboxide (TiO2-x ) (a) (a)

(b) (b)

Figure 4. Transmission electron micrographs of a bare P90 TiO2 (a) and Ti-sub-oxide powders (b) used Figure 4. Transmission electron micrographs of a bare P90 TiO2 (a) and Ti-sub-oxide powders (b) used Figure Transmission electron micrographs of a bare P90 TiO2 (a) and Ti-sub-oxide powders (b) used for the the 4. photoanode. for photoanode. for the photoanode.

Two types of photoelectrodes were investigated. A bare TiO2 layer was coated onto the TCO Two ofofphotoelectrodes were investigated. A bare coated theonto TCO substrate 2 layer Twotypes types photoelectrodes were investigated. AaTiO bare TiOset 2was layer was onto coated TCO substrate and used in the first experiments (Figure 5a). In second of experiments, the TiOthe 2 photoand used in the first experiments (Figure 5a). In a second set of experiments, the TiO photo-anode layer 2 cross-section substrate andwas usedcoated in the first experiments (Figuresub-oxide 5a). In a second of experiments, the TiO2 photoanode layer with a thin titanium layer set (Figure 5b). The of was coated with a thin titanium sub-oxide layer (Figure 5b). Thelayer cross-section of photoanodes prepared of in anode layer was coated with a thin titanium sub-oxide (Figure 5b). The cross-section photoanodes prepared in a similar way was investigated by scanning electron microscopy. The a similar way was investigated by scanning electron microscopy. The thickness of the TiO layer ranged 2 photoanodes prepared in a similar way was investigated by scanning electron microscopy. The thickness of the TiO2 layer ranged between two and four microns, whereas the thickness of the Tibetween two andTiO four2 layer microns, whereas the thickness the microns, Ti-suboxide layer was threeofmicrons. thickness of the ranged between two andoffour whereas the about thickness the Tisuboxide layer was about three microns. suboxide layer was about three microns.

Membranes 2017, 7, 25 Membranes 2017, 7, 25

7 of 16 7 of 15

a TiO2

b TiOx

TiO2

SnO2:F

Glass

Glass SnO2:F

Figure 5. Scanning electron micrographs of a bare P90 TiO22 photoanode layer deposited onto a TCO a bilayer consisting of Ti sub-oxide coated onto the TiO (b). 2Inlayer these(b). micrographs, substrate (a) (a)and and a bilayer consisting of Ti sub-oxide coated onto2 layer the TiO In these the semiconductor layers appear layers on stepped surfaces due tosurfaces the cutting micrographs, the semiconductor appear on stepped due procedure. to the cutting procedure.

2.4. Photo-Electrolysis Behaviour The influence influenceof of membrane and ionomer characteristics on photo-splitting the water photo-splitting thethe membrane and ionomer characteristics on the water performance performance wasininvestigated in a photo-electrolysis single cell. As discussed above, the photowas investigated a photo-electrolysis single cell. As discussed above, the photo-electrolysis device electrolysis device consisted of the an TiO assembly between the TiO 2 photo-anode, the membrane consisted of an assembly between photo-anode, the membrane and platinum as cathode. In and this 2 platinum as cathode. In this system, the hydrogen evolution occurs at the Pt cathode that was used system, the hydrogen evolution occurs at the Pt cathode that was used as working/sense electrode, as working/sense electrode, whereas, the photo-anode, where O2 evolution occurs, was Since used the as whereas, the photo-anode, where O2 evolution occurs, was used as counter/reference electrode. counter/reference electrode. Since aimof ofthe theion work was to membrane compare the of thephoto-splitting, ion exchange aim of the work was to compare thethe effect exchange oneffect the water membrane on the water photo-splitting, Titania-based photo-anode wasininitially used without any a Titania-based photo-anode was initiallya used without any modification terms of doping, surface modification terms of doping, etc., surface with reaction promoters, etc., as itefficiencies. would be treatment withinreaction promoters, as it treatment would be required to achieve useful conversion ◦ C of theefficiencies. required to achieve useful conversion Only thermal sintering at 450 of the photoOnly a thermal sintering at 450 photo-anode wasacarried out to achieve good°Cadhesion of the anode was carried to achieve adhesion of the photo-electrode film in to the the electrode substratelayer. and photo-electrode filmout to the substrategood and proper continuity of the TiO2 network proper continuity ofan theun-doped TiO2 network in the electrode layer. It is pointed thatjust an un-doped TiO2 It is pointed out that TiO2 according to its wide energy gap canout absorb a small fraction according its wideinenergy can absorb just athus small fractionlow of the solar spectrum in the of the solartospectrum the UVgap region. This system generates photocurrents; however, at UV the region. Thisthe system thus generates lowfor photocurrents; however, at thephoto-voltage. same time, the high energy same time, high energy gap allows achieving large spontaneous gap allows for achieving largepresented spontaneous photo-voltage. According to the sketch in Figure 1, being the TiO2 semiconductor (SC) brought in According to the sketch presented in Figure being the 2 semiconductor (SC) brought in contact with the water film that is embedded into the1,ionomer andTiO membrane electrolyte, a depletion of contact thecarriers water film that in is TiO embedded into the ionomer and membrane electrolyte, a depletion majoritywith charge (n-type ) across the interface occurs due to the difference in the chemical 2 of majorityofcharge TiO 2) across interface occurs due to in thethe difference in the potentials the twocarriers phases(n-type [15,16]. in This results in athe band bending phenomena semiconductor chemical potentialsThe of illumination the two phases [15,16]. This results in a band bending phenomena in the near the interface. of the semiconductor–electrolyte interface causes a modification semiconductor the interface. The illumination of the semiconductor–electrolyte causes of the electrodenear potential as a consequence of the generation of electron–hole pairs,interface which are thus aseparated modification of the electrode potential as a consequence of the generation of electron–hole pairs, by the effect of the electric field across the depletion region [15,16]. At the open circuit (OCP), which are thus separated the effect ofofthe electric field across depletion [15,16]. At the the illumination causes an by accumulation photo-generated chargethe carriers acrossregion the depletion region open acircuit (OCP), the illumination causesproducing an accumulation of photo-generated carriers with corresponding upward band bending a spontaneous photo-voltagecharge [16]. The band across thecan depletion regionbywith a corresponding upward band bending a spontaneous bending be increased polarising the electrode-electrolyte interfaceproducing (Figure 1 is showing the photo-voltage [16]. band bending can beillumination). increased by polarising the electrode-electrolyte band bending in TiO2The at the short circuit under interface (Figure 1the is showing the band bending in TiOof 2 at theelectrode-polymer short circuit under illumination). Accordingly, photo-electrochemical behaviour the electrolyte membrane Accordingly, the photo-electrochemical behaviour of the electrode-polymer electrolyte junction is characterised by a specific photocurrent-photo-voltage (I–V) profile that is very similar membrane junctioncharacteristic is characterised by a specific (I–V) profile very to a photodiode (Figures 6–8). photocurrent-photo-voltage The spontaneous photo-voltage drives that the iswater similar to a photodiode characteristic (Figures 6–8). The spontaneous photo-voltage drives the water photo-splitting process up to the short circuit (short circuit photocurrent, Isc). No hydrogen evolution photo-splitting process the short circuit (short photo-voltage circuit photocurrent, Isc). No evolution occurs at the OCP, andup thetoobserved open circuit represents an hydrogen electromotive force occurs the OCP, and the observed circuitthe photo-voltage electromotive for for the at water photo-splitting process;open whereas, spontaneousrepresents hydrogenan production is, inforce general, the water photo-splitting process; whereas, the spontaneous hydrogen production is, in general, maximised at the short circuit (cell potential equal to zero in Figures 6–8). The potential region between maximised at Isc the isshort circuit (cell to zero in Figures 6–8).∆G, Thecorresponding potential region the OCP and characterised bypotential negative equal variation of the free energy, to between the OCP and Iscevolution is characterised by negative variation of the free energy, ∆G,potential corresponding a spontaneous hydrogen and concomitant production electricity (positive region to a spontaneous hydrogen evolution and concomitant production of electricity (positive potential region when the cathode is used as working/sense electrode) [15]. Under such conditions, the Fermi

Membranes 2017, 7, 25

8 of 16

Membranes 2017, 7, 25

8 of 15

when the cathode is used as working/sense electrode) [15]. Under such conditions, the Fermi level in the semiconductor is higherisinhigher potential energy than the Fermi level in thelevel electrolyte. This difference level in the semiconductor in potential energy than the Fermi in the electrolyte. This is the spontaneous photo-voltage [15]. The production of electricity is null at the Isc where the Fermi difference is the spontaneous photo-voltage [15]. The production of electricity is null at the Isc where levels in thelevels semiconductor and the electrolyte are the same. The rangepolarisation between therange OCP the Fermi in the semiconductor and the electrolyte arepolarisation the same. The and Isc is also called photo-voltage driven region and the photo-electrolysis cell operating under between the OCP and Isc is also called photo-voltage driven region and the photo-electrolysissuch cell conditions is assisted only by the solar radiation. If an external bias is applied the band banding in the operating under such conditions is assisted only by the solar radiation. If an external bias is applied semiconductor increases and the photocurrent can persist or may increase further. In this case the cell the band banding in the semiconductor increases and the photocurrent can persist or may increase voltage cathode, as working electrode, and itsas associated leveland in the further. is Innegative this casebecause the cell the voltage is negative because the cathode, working Fermi electrode, its electrolyte is higher in terms of potential energy with respect to the Fermi level in the photo-anode. associated Fermi level in the electrolyte is higher in terms of potential energy with respect to the Above the reversible potential (1.23 V) for the water splitting,potential the endothermic electrolysis occurs Fermi level in the photo-anode. Above reversible (1.23 V) water for water splitting, the also in the dark. The potential region between the Isc and the reversible potential is the bias-assisted endothermic water electrolysis occurs also in the dark. The potential region between the Isc and the photo-electrolysis region most ofphoto-electrolysis the photo-electrolysis semiconductors by small reversible potential is thewhere bias-assisted region where most ofcharacterised the photo-electrolysis band gap or un-appropriate band levels usually operate. semiconductors characterised by small band gap or un-appropriate band levels usually operate. 0

Current density / mA cm-2

Open circuit photo-voltage

0.73 V 0.91 V

Nafion 115 perfluorosulfonic membrane

-0.05

-0.1

FAA3-20 anionic membrane -0.15

Endothermic process

Spontaneous photo-voltage driven cell

External bias driven cell

-0.2 -1.5

-1

-0.5

0

0.5

1

Cell Voltage / V

Figure 6. Photo-electrolysis polarisation curves under AM1.5 illumination for Nafion® 115 and Figure 6. Photo-electrolysis polarisation curves under AM1.5 illumination for Nafion® 115 and ® FAA3-20 membranes based cells consisting of a bare P90 TiO2 photoanode and Pt cathode. Fumion® Fumion FAA3-20 membranes based cells consisting of a bare P90 TiO2 photoanode and Pt cathode.

The I–V characteristics of the photo-electrolysis cells under investigation are shown in Figures The I–V characteristics of the photo-electrolysis cells under investigation are shown in Figures 6–8. 6–8. The I–V response is reported under illumination only; the dark current was quite low and similar The I–V response is reported under illumination only; the dark current was quite low and similar in all in all cases. As reported elsewhere [14], for a large band gap n-type TiO2 semiconductor, the cases. As reported elsewhere [14], for a large band gap n-type TiO2 semiconductor, the equilibrium equilibrium concentration of holes is very low. Accordingly, the anodic dark current that derives concentration of holes is very low. Accordingly, the anodic dark current that derives from the diffusion from the diffusion of holes towards the surface is also low. of holes towards the surface is also low. The photo-electrolytic characteristics for the bare TiO2 photo-anode combined with proton The photo-electrolytic characteristics for the bare TiO2 photo-anode combined with proton ® 115 and anion ® FAA3-20 exchange Nafion® exchange Fumion® polymer electrolyte membranes are exchange Nafion 115 and anion exchange Fumion−2 FAA3-20 polymer electrolyte membranes are shown in Figure 6. Upon illumination (100 mW cm ), the spontaneous photo-voltage registered for −2 ), the spontaneous photo-voltage registered for the shown in Figure 6. Upon illumination (100 mW cm ® -based the Nafion cell is about 0.73 V, where it reaches 0.91 V with the alkaline membrane. More ® -based cell is about 0.73 V, where it reaches 0.91 V with the alkaline membrane. More relevant Nafion relevant is the increase of photocurrent with the alkaline system in the overall potential region. The is the circuit increase of photocurrent with with the alkaline system in the overall potential region. The short short photocurrent recorded the anionic membrane is about twice that observed with circuit photocurrent recorded with the anionic membrane is about twice that observed with Nafion® . Nafion® . The higher slope of the photo-electrolytic characteristic recorded with the anionic membrane The higher slope of the photo-electrolytic characteristic recorded thecorresponding anionic membrane the in the photo-voltage-driven region is indicative of a better fill with factor to a in lower photo-voltage-driven region isphenomena indicative of a better fill factor corresponding to a lower occurrence of occurrence of recombination [15–22]. Thus, the minority charge carriers (holes for the nrecombination phenomena [15–22]. Thus, the minority charge carriers (holes for the n-type TiO ) are type TiO2) are more easily transferred to the electrolyte under alkaline conditions. This aspect2 may more easilytotransferred to the electrolyte under alkaline conditions. This aspect may related to be related the larger concentration of OH species under alkaline conditions that arebe involved in the the larger concentration of OH species under alkaline conditions that are involved in the intermediate intermediate steps of the oxygen evolution process [15,16]. A large concentration of these species at steps of the oxygen process on [15,16]. A large concentration of these the species at the the interface favoursevolution their adsorption the photo-anode surface favouring capture of interface holes by favours their adsorption on the photo-anode surface favouring the capture of holes by the adsorbed the adsorbed OH species with respect to the recombination with photo-generated electrons inside the OH species with respect to the recombination with photo-generated electrons inside the semiconductor. semiconductor. The large spontaneous photo-voltage recorded for the present electrolytic cells is associated with a relevant upper band bending occurring at the TiO2 semiconductor upon illumination. This produces the onset of photocurrents at potentials quite negative with respect to the equilibrium potential of water splitting, which is the same in acid and alkaline electrolytes, i.e., 1.23 V at 25 °C.

Membranes 2017, 7, 25

9 of 16

The large spontaneous photo-voltage recorded for the present electrolytic cells is associated with a relevant upper band bending occurring at the TiO2 semiconductor upon illumination. This produces the onset of photocurrents at potentials quite negative with respect to the equilibrium potential of water splitting, which is the same in acid and alkaline electrolytes, i.e., 1.23 V at 25 ◦ C. Thus, the energy required for water splitting is provided by the illumination and the generated high energy holes (Figure 1 bottom) can oxidise water. Accordingly, water splitting can occur under conditions more favourably than at the equilibrium, and electricity is also generated together with hydrogen at positive cell potentials. The protonic or anionic membranes do not affect this mechanism but they influence the charge transfer at the photo-anode–electrolyte interface that must compete with the recombination processes inside the semiconductor. Besides the above-discussed mechanism, another phenomenon may occur in the photo-electrolysis device that can explain the large spontaneous photo-voltages observed here. However, specific efforts were adopted to avoid exposure to air of the water filled to the cell, since the cell itself is not gastight (the cell is designed to favour gas escape), and it is not excluded that some air infiltration may have occurred during this procedure and when the cell was operated. This would mean that part of the large spontaneous photo-voltage observed here could also derive from a shift of the Fermi level of the Pt electrode to positive values associated with the adsorption of oxygen species on Pt. In principle, the photo-voltage is caused by a shift of the Fermi level of TiO2 upon illumination from its rest potential in the dark, but the Fermi level positioning in the dark is determined by a specific interaction between the semiconductor surface states and the electrolyte. In our case, the electrolyte is composed of pure water in combination with a solid polymer electrolyte. The thermodynamic basis for hydrogen production relies on the fact that the conduction band edge of TiO2 is higher in energy or negative in potential with respect to RHE under the present experimental conditions (Figure 1 bottom). On the other hand, the measured cell voltage under illumination results from the difference between the cathode potential and the anode potential. The anode potential is affected by the interaction between the semiconductor surface states and the electrolyte whereas the cathode potential may also be affected by the adsorption of traces of oxygen gas molecules on the Pt surface. The oxygen evolution process occurring at the TiO2 surface is widely considered to be the rate determining step for the photo-electrolysis configuration under study in both acidic and alkaline systems [17–22]. Light harvesting produces the intrinsic ionization of the titania semiconductor, which leads to the formation of holes in the valence band and electrons in the conduction band provided that the energy of photons (hν) is larger than the energy gap: 2hν → 2e− + 2h+ ; 2h+ + H2 O →

1 O2 + 2H+ ; 2H+ + 2e− → H2 2

(1)

The electric field at the electrode–electrolyte interface avoids recombination of the photo-generated charge carriers allowing the light-induced electron-hole pairs to produce the splitting of water into oxygen and hydrogen at the photo-anode and photocathode–electrolyte interface. Besides influencing the charge transfer at the interface, a different type of electrolyte can favour the formation of electronic levels at the surface causing a pinning of the Fermi level [23]. Accordingly, the difference in the observed photo-voltages for acidic and alkaline environment is associated with these two effects i.e., surface states and charge transfer kinetics. To get more insights on the surface effects and the influence of the different electrolytes on the adsorption properties of OH species, the TiO2 photo-anode was coated with a thin Titanium sub-oxide layer. The presence of the sub-stoichiometric TiO2−x species on the photo-anode surface promotes significantly the adsorption of OH species. These suboxides have been identified as promoters for the oxygen evolution in acidic environment [13]. Figure 7 shows a small enhancement of photo-voltage for the Nafion® based photo-electrolysis cell when TiO2−x is coated on the TiO2 photo-anode and a significant increase of photocurrent in the entire range, especially at potentials close to the reversible potential (−1.23 V). This provides a clear indication that OH adsorption is a rate determining step

Membranes 2017, 7, 25 Membranes 2017, 7, 25

10 of 15 10 of 16

environment does not reach that of bare TiO2 in alkaline environment both in terms of OCP and photocurrent. in the photo-electrolytic water splitting in acidic environment being this process promoted by the

Current density / mA cm-2 Current density / mA cm-2

presence of sub-stoichiometric oxides. TiO2−x surface states favour the adsorption of the oxygen 0 Nafionthe 115 membrane 0.73 of V a wide coverage of OH species species on the surface to saturate defective sites. The presence Open circuit on the electrode surface favours the hole transfer tophoto-voltage the electrolyte while avoiding the recombination 0.78 V -0.05 phenomena [22–25]. The positive effect is observed Bare TiO2in terms of both photo-voltage and photocurrent. Membranes 2017, 7, 25 10 of 15 The adsorption of OH species possibly shifts the flat-band potential producing an increase of the photo-voltage. In any thethat performance of the TiO2 in acidic doesand not environment does notcase, reach of bare TiO 2 inTiO alkaline environment both environment in terms of OCP 2−x -coated -0.1 TiO -coated TiO reach that of bare TiO2 in alkaline environment both in terms of OCP and photocurrent. photocurrent. 2-x 2 0 -0.15

Nafion 115 membrane 0.73 V Spontaneous photo-voltage Open circuit Endothermic driven cell External bias driven cell photo-voltage process 0.78 V

-0.05 -0.2 -1.5

Bare TiO2 -1

-0.5

0

0.5

1

Cell Voltage / V -0.1

TiO2-x -coated TiO2 Figure 7. Photo-electrolysis polarisation curves under AM1.5 illumination for Nafion® 115 membrane based cells equipped with a bare P90 TiO2 photoanode and a TiO2−x sub-oxided coated TiO2 -0.15 photoanode. Spontaneous photo-voltage Endothermic driven cell

External bias driven cell

process

Current Density / mA cm-2Current Density / mA cm-2

Curiously, when the TiO2−x-coated TiO2 photoanode is used in the presence of the anionic system -0.2 (Figure 8), a strong decrease of performance both in terms of photo-voltage and photocurrent is -1.5 -1 -0.5 0 0.5 1 observed. It appears that the large concentration ofVoltage OH species in the alkaline environment causes a Cell /V strong adsorption on the Ti sub-oxide surface, and this hinders the desorption of oxygen species to Figure 7. 7. Photo-electrolysis polarisation curves curves under under AM1.5 AM1.5 illumination illuminationfor forNafion Nafion®® 115 115 membrane membrane Photo-electrolysis polarisation give Figure rise to oxygen gas evolution. It is well known that in electro-catalytic reactions, the binding based cells equipped with a bare P90 TiO 2 photoanode and a TiO2−x sub-oxided coated TiO2 based cells equipped with a bare P90 TiO and a TiO2surface coated TiOneither 2 photoanode −x sub-oxided 2 photoanode. energy between the adsorbed species and the electrocatalyst should not be strong nor photoanode. weak but intermediate to maximise the reaction rate [7,13]. Otherwise, the rate of adsorption or Curiously, whenrate the TiO2−x -coatedifTiO photoanode is used in or thestrong, presence of the anionic system desorption becomes the22bond strength weak respectively [7,13]. Curiously, when the determining TiO2−x-coated TiO photoanode is is used in the presence of the anionic system (Figure 8), a strong decrease of performance both in terms of photo-voltage and photocurrent is The8),presence widely distributed electronic levels sub-stoichiometric oxide surface (Figure a strongofdecrease of performance both in termsinofthe photo-voltage and photocurrent is observed. It appears that the large concentration of OH species in the alkaline environment causes causes recombination and trapping effects for the photo-generated carriers if the oxygen species can observed. It appears that the large concentration of OH species in the alkaline environment causes a anot strong adsorption on the Ti sub-oxide surface, andphoto-generated this hinders the desorption of determining oxygen species to beadsorption easily desorbed the capture holes, thusof strong on the Tiafter sub-oxide surface, of and this hinders the desorption oxygen specieslow to give rise to oxygen gas evolution. It is well known that in electro-catalytic reactions, the binding energy photocurrent. give rise to oxygen gas evolution. It is well known that in electro-catalytic reactions, the binding between the adsorbed species and thehow electrocatalyst should not be neither strong the norchoice weak The present results clearly show theelectrocatalyst type ofsurface electrolyte canshould drastically influence energy between the adsorbed species and the surface not be neither strong nor but intermediate to maximise the reaction rate [7,13]. Otherwise, the rate of adsorption or desorption of surface in photo-electrocatalysis and rate the performance achievable withofdifferent photoweak but promoters intermediate to maximise the reaction [7,13]. Otherwise, the rate adsorption or becomes rate determining if the bond strength is weak or strong, respectively [7,13]. electrode systems. desorption becomes rate determining if the bond strength is weak or strong, respectively [7,13]. The presence of widely distributed electronic levels in the sub-stoichiometric oxide surface 0 FAA3-20 effects membranefor theOpen causes recombination and trapping photo-generated carriers if the oxygen species can circuit 0.83 V photo-voltage not be easily desorbed after the capture of photo-generated holes, thus determining low 0.91 V -0.05 photocurrent. Bare TiO2 The present results clearly show how the type of electrolyte can drastically influence the choice of surface promoters in photo-electrocatalysis and the performance achievable with different photo-0.1 electrode systems. 0 -0.15

TiO2-x -coated TiO2 FAA3-20 membrane Open circuit Endothermic External bias photo-voltage driven cell process

Spontaneous photo-voltage 0.83 V driven cell 0.91 V

-0.05 -0.2 -1.5

-1

Bare TiO2 -0.5

0

0.5

1

Cell Voltage / V -0.1

Figure 8. 8. Photo-electrolysis Photo-electrolysis polarisation polarisation curves curves under under AM1.5 AM1.5 illumination illumination for for Fumion Fumion®® FAA3-20 FAA3-20 Figure membrane based cells equipped with a bare P90 TiO 2 photoanode and a TiO2−x sub-oxided coated TiO -coated TiO 2-x a bare P90 2TiO2 photoanode and a TiO2−x sub-oxided coated membrane based cells equipped with TiO2 photoanode. photoanode. -0.15 TiO Spontaneous photo-voltage 2 Endothermic -0.2 -1.5

driven cell

External bias driven cell

process

-1

-0.5

0

0.5

1

Cell Voltage / V

Figure 8. Photo-electrolysis polarisation curves under AM1.5 illumination for Fumion® FAA3-20

Membranes 2017, 7, 25

11 of 16

The presence of widely distributed electronic levels in the sub-stoichiometric oxide surface causes recombination and trapping effects for the photo-generated carriers if the oxygen species can not be easily desorbed after the capture of photo-generated holes, thus determining low photocurrent. The present results clearly show how the type of electrolyte can drastically influence the choice of2017, surface Membranes 7, 25 promoters in photo-electrocatalysis and the performance achievable with different 11 of 15 photo-electrode systems. Ac-impedance analysis (Figure (Figure9)9)was was carried to understand if there was some relevant Ac-impedance analysis carried outout to understand if there was some relevant ohmic ohmic limitation associated with the different membranes used in the photo-electrolysis cells. The limitation associated with the different membranes used in the photo-electrolysis cells. The series series resistance thefrequency high frequency intercept the Nyquist plot observed the various systems resistance or theor high intercept in the in Nyquist plot observed for thefor various systems under 2. Thus, this should not produce 2 . Thus, under illumination was generally lower 15 Ohm illumination was generally lower thanthan 15 Ohm cmcm this should not produce any relevant relevant potential potential loss loss since since the the photocurrent photocurrent measured measured in in the the photo-electrolysis photo-electrolysisdevice deviceisisrelatively relativelysmall. small. However, However, polarization polarization resistance resistance associated associated with with both both the thecharge chargetransfer transferat atthe theelectrode–electrolyte electrode–electrolyte interface interface and and the the charge charge transport transport in in the the space space charge charge region region appears appears dominant dominant since since no no complete complete semicircle is observed. The impedance profile in the Nyquist plot is mainly capacitive and semicircle is observed. The impedance profile in the Nyquist plot is mainly capacitive andthe the high high ® ® frequency frequency intercept intercept shifts shifts to to larger larger values values for for the alkaline membrane. The The Nafion based based cell cell shows shows slightly slightly lower lower series series resistance resistance due due to to the the better better conductivity conductivity even even if if the Nafion®® 115 115 membrane membrane thickness thickness is ismuch muchlarger largerthan thanthe theFAA3-20 FAA3-20 membrane. membrane. Most Most of ofthe thecontribution contributionto tothe theseries seriesresistance resistance isis generally generallydue due to to the the TCO TCO substrate substrate even even ifif the the membrane, membrane, interface interfaceand and cell cell design designcan can also also have have an impact on the ohmic properties of the cell. an impact on the ohmic properties of the cell.

Figure 9. AC impedance spectra under AM1.5 illumination at open circuit potential for Nafion®® 115 Figure 9. AC impedance spectra under AM1.5 illumination at open circuit potential for Nafion 115 and Fumion®®FAA3-20 membrane based cells equipped with a bare P90 TiO2 photo-anode. and Fumion FAA3-20 membrane based cells equipped with a bare P90 TiO2 photo-anode.

Current density / mA cm-2

The photo-electrolysis devices were also investigated in terms of time studies (Figure 10). For photo-electrolysis werethe also investigated in terms of time (Figure 10). For these The studies, only the cellsdevices providing best performances during the studies initial experiments, i.e.,these the studies, only the cells providing the best performances during the initial experiments, i.e., the bare bare TiO2 in combination with the anionic membrane and the Ti sub-oxide coated-TiO2 combined TiO2 the in combination with the anionic membrane and the Ti sub-oxidecurves coated-TiO the 2 combined with protonic membrane, were selected. Chrono-amperomentric (Figure 10) werewith carried protonic membrane, were selected. Chrono-amperomentric curves (Figure 10) were carried out for out for both anionic and protonic devices for 4 h at the short circuit (0 V). both anionic and protonic devices for 4 h at the short circuit (0 V). A gradual decrease0 of photocurrent with time was observed for both configurations even if the Nafion® -based system appeared more stable (Figure 10). After the time study, photoelectrochemical polarisations were repeated in the dark and under illumination (Figure 11). The open circuit photovoltage -0.05 Nafion 115 perfluorosulfonic membrane decreased by about 170 and 140 mV in the case of the anionic and protonic cells, respectively, with respect to the initial measurements. However, the photocurrent decay was about 25% and 11% for the -0.1 respectively, compared to the initial values. The curves registered in the dark anionic and protonic cells, showed a positive voltage of about 130 mV and 97 mV for the protonic and anionic cells, respectively (Figure 11). However, the dark current was negligible in the photovoltage-driven region, increasing -0.15

FAA3-20 anionic membrane

-0.2

0

4000

8000

12000 12,000

16000 16,000

Time / s ®

Figure 9. AC impedance spectra under AM1.5 illumination at open circuit potential for Nafion® 115 and Fumion® FAA3-20 membrane based cells equipped with a bare P90 TiO2 photo-anode.

The photo-electrolysis devices were also investigated in terms of time studies (Figure 10). For 12 of 16 these studies, only the cells providing the best performances during the initial experiments, i.e., the bare TiO2 in combination with the anionic membrane and the Ti sub-oxide coated-TiO2 combined with theclose protonic were selected. Chrono-amperomentric curves (Figure 10) were carried rapidly to themembrane, reversible potential. This increase of dark current in the bias-assisted region close out forreversible both anionic and protonic devices forfor 4 hthe at anionic the short circuitcompared (0 V). to the potential was more evident system to the protonic system. Membranes 2017, 7, 25

Membranes 2017, 7, 25

Current density / mA cm-2

0

-0.05

Nafion 115 perfluorosulfonic membrane

12 of 15 -0.1

Current density / mA cm-2

A gradual decrease of photocurrent with time was observed for both configurations even if the Nafion® -based system appeared more stable (Figure 10).0After the time study, photoelectrochemical -0.15 dark current a polarisations were repeated in the dark and under illumination (Figure 11). The open circuit FAA3-20 anionic membrane -0.05 photovoltage decreased by about 170 and 140 mV in the case of the anionic and protonic cells, -0.2 to the initial measurements. However, the photocurrent decay was about respectively, with respect under 0 8000 12000 4000 16,000 12,000 25% and 11% for the anionic and protonic cells, respectively, compared to16000 the initial values. The -0.1 illumination curves registered in the dark showed a positiveTime voltage of about 130 mV and 97 mV for the protonic /s and anionic cells, respectively (Figure 11). However, in the -0.15 the dark current was negligible ® 115 and ® Figure10. 10.Chrono-amperometric Chrono-amperometric curves under AM1.5 illumination 0Nafion V for® Nafion Figure curves under AM1.5 illumination at 0 V at for 115 and Fumion photovoltage-driven region, increasing rapidly close to the reversible potential. This increase of dark 115 perfluorosulfonic membrane ® FAA3-20 membrane-based cells equipped with a TiONafion Fumion 2−x sub-oxide coated TiO 2 and bare P90 FAA3-20 cells equipped with a TiO2−x sub-oxide coated TiO2evident and barefor P90 TiO 2 current in themembrane-based bias-assisted region close to the reversible potential was more the anionic -0.2 TiO2 photo-anodes, respectively. photo-anodes, system compared respectively. to the protonic system. -1.5 -1 -0.5 0 0.5 1 Cell Voltage / V dark current

Current density / mA cm-2

Current density / mA cm-2

0

a

-0.05

under illumination

-0.1

-0.15

Nafion 115 perfluorosulfonic membrane -0.2 -1.5

-1

-0.5

0

Current density / mA cm-2

Cell Voltage / V

0.5

1

0

dark current

b

-0.05

under illumination

-0.1 -0.15

FAA3-20 anionic membrane -0.2 -1.5

-1

-0.5

0

0.5

1

Cell Voltage / V

0

Figure 11. polarisation curves carried out out afterafter 4 h operation at the at short dark current Figure 11. Photo-electrolysis Photo-electrolysis polarisation carried 4 h operation the circuit short bcurves ® 115 (a) and ® FAA3-20 ® under AM1.5 illumination and in the dark for Nafion Fumion (b) membrane® -0.05 under AM1.5 illumination and in the dark for Nafion 115 (a) and Fumion FAA3-20 (b) circuit based cells equipped with a TiO 2−x sub-oxided coated TiO2 photoanode and a bare P90 TiO2 membrane-based cells equipped with a TiO2−x sub-oxided coated TiO2 photoanode and a bare P90 under photoanode, respectively. -0.1photoanode, TiO respectively. illumination 2

-0.15 Several hypotheses are formulated for the photocurrent and photo-voltage decay observed after Several hypotheses are formulated for the photocurrent and photo-voltage decay observed after the time studies. However, we have used deaerated water, since this is filled in the cell by a syringe FAA3-20 anionic membrane the time studies. However, we have used deaerated water, since this is filled in the cell by a syringe -0.2 system, water is necessarily exposed to the air. Thus, we do not exclude that some traces of dissolved -1.5 is necessarily -1 -0.5 exposed 0 system, water to0.5 the air. 1Thus, we do not exclude that some traces of dissolved oxygen may be presented in water. This may have possibly caused a depolarisation of the cathode Cell in Voltage / V This may have possibly caused a depolarisation of the cathode oxygen may be presented water. by shifting the Fermi level to higher potentials. Such effect may explain the loss of photo-voltage after by shifting the Fermi level to higher potentials. Such effect may explain the loss of photo-voltage prolonged operation since the residual oxygen at the Pt electrode is consumed by the produced after prolonged operation since the residual oxygen at the Pt electrode is consumed by the produced hydrogen. The decrease of photocurrent could be related to mass transport limitations for the hydrogen. The decrease of photocurrent could be related to mass transport limitations for the evolution evolution of the gas molecules inside the device or to some modification of the electrode–electrolyte of the gas molecules inside the device or to some modification of the electrode–electrolyte interface. interface. However, the performance decay was almost reversible in both cases since the photoHowever, the performance decay was almost reversible in both cases since the photo-electrolysis cells electrolysis cells re-gained the initial performance when the electrochemical operation under re-gained the initial performance when the electrochemical operation under illumination was stopped illumination was stopped for some hours. In this regard, we think that device design should be for some hours. In this regard, we think that device design should be improved to address these improved to address these recoverable losses. recoverable losses. The information gained from this study, using TiO2 as model system, and the selected membranes can be useful in developing photo-electrolysis cells based on smaller energy gap semiconductors than TiO2, thus providing a relevant increase of water splitting efficiency.

3. Materials and Methods A commercial TiO2 powder, Degussa P90, was used for preparing the photo-anode. The Tisuboxide (TiO2−x), used as surface promoter was prepared according to Ref. [13]. The thermal

Membranes 2017, 7, 25

13 of 16

The information gained from this study, using TiO2 as model system, and the selected membranes can be useful in developing photo-electrolysis cells based on smaller energy gap semiconductors than TiO2 , thus providing a relevant increase of water splitting efficiency. 3. Materials and Methods A commercial TiO2 powder, Degussa P90, was used for preparing the photo-anode. The Ti-suboxide (TiO2−x ), used as surface promoter was prepared according to Ref. [13]. The thermal treatment was designed to achieve a sub-oxide phase while avoiding growth of catalyst particles. The eventual occurrence of hot spots, often causing particle sintering, was controlled by an infrared camera [26]. Briefly, the ceramic powders were prepared from TiCl4 by complexation of Ti ions using EDTA (Aldrich, Milan, Italy) as chelating agent, and successive decomposition by hydrogen peroxide of the formed complex to produce an amorphous oxide. A high temperature reduction (1050 ◦ C) of the amorphous oxide was carried out by using diluted hydrogen (10% H2 in Ar). Structural characterization of the Ti-oxide powders was carried out by X-ray diffraction using a Panalytical X’Pert powder diffractometer (Philips, Eindhoven, The Netherlands) with CuKα radiation. Surface area and pore volume were investigated by BET analysis using a Micromeritics ASAP 2020 instrument (Micromeritics, Milan, Italy). Morphological characteristics of the powders were observed by Transmission electron (microscopy (TEM) measurements (FEI CM12 microscope, Eindhoven, The Netherlands). For photoelectrolysis measurements, the TiO2 photo-anode was prepared by spraying solutions of TiO2 nanoparticles dispersed in water and isopropyl alcohol onto a transparent conductive oxide, TCO, (19 Ω/sq.) substrate (SnO2 :F). In a second set of experiments, an additional layer formed by the Ti-suboxide was deposited onto the TiO2 layer. Before use, TCO glasses were treated for 5 min in an ultrasonic bath of isopropyl alcohol and rinsed with acetone. Triton X-100 was added as dispersant. The coated electrode was dried at 70 ◦ C onto a hot plate and then annealed in an oven at 450 ◦ C for 30 min in air. The active area was 4 cm2 . For the cathode, the TCO substrate was coated with a drop of H2 PtCl6 and dried at 70 ◦ C onto a hot plate. It was thermally annealed at 450 ◦ C to form a Pt black particles-based layer onto the TCO. This was used as counter electrode for hydrogen evolution. The morphology of the photo-anode was investigated by scanning electron microscopy (FEI XL30 SFEG, Eindhoven, The Netherlands). Commercial perfluorosulfonic acid (Nafion® 115) and quaternary ammonium-based anionic (Fumatech® , Fumasep® FAA3) ion exchange membranes were purified before use (a picture of the raw membranes is shown in Figure 1). Ex situ membrane characterisation was carried out according to the procedures described in Ref. [27]. Nafion® 115 (Ion Power, equivalent weight, EW, 1100) of 125 µm thickness was used as proton exchange membrane polymer electrolyte. The Nafion® membrane was purified with H2 SO4 0.1 M for 1 h, treated with water several times up to achieve neutral pH for the washing solution and successively boiled in water before use. In addition, 5 wt % Nafion® ionomer (Ion Power, 5 wt % solution) was used as dispersion. This was deposited on the formed anode, which was dried at 120 ◦ C for 10 min to evaporate the solvent. The ionomer dispersion was also deposited on the counter electrode. A Fumatech® FAA3 anionic membrane of 20 µm thickness was purified with 0.5 NaOH for 1 h to produce a high swelling of the membrane and successively with NaCl 0.5 M for 72 h at 25 ◦ C, exchanging the solution several times to replace bromide counter ions with chlorides. Thereafter, the membrane was washed in water and treated with NaOH 0.1 M at room temperature for 1 h. Finally, it was treated with water several times to achieve neutral pH for the washing solution. The Fumatech® FAA-3 dispersion was obtained by dissolving the FAA3 powder (5 wt %) in ethanol and n-propanol. The dispersion was deposited onto the photoanode surface and dried at 60 ◦ C to evaporate the solvent. The coated photoelectrode was first treated with 0.1 M NaCl and thereafter with 0.1 M NaOH. The same procedure was adopted for the counter electrode.

Membranes 2017, 7, 25

14 of 16

The ionomer-coated electrodes, TiO2 /TCO photoanode and Pt/TCO cathode were pressed together onto the membrane using an ultrathin segmented sub-gasket in between to allow for gas escape from the two cell compartments. The segmented sub-gasket was laser cut and made of the same ion conducting polymer material as the membrane. Deaerated Milli-Q® water was fed to the cell through a hole drilled on the Pt/TCO counter electrode until achieving full hydration of the cell. The electrochemical apparatus consisted of a Metrohm Autolab potentiostat/galvanostat (Utrecht, The Netherlands) equipped with a Frequency Responce Analyser (FRA). An Oriel® solar simulator (Newport, Irvine, CA, USA) providing an irradiation of 100 mW cm−2 (AM1.5) was used. The cell was illuminated through the photo-anode glass backing. Thus, the light penetrated through the glass and the transparent TCO layer before reaching the TiO2 -hydrated polymer-electrolyte interface. Polarisation measurements (20 mV s−1 ) under illumination (100 mW cm−2 ) were carried out in a wide potential range starting at the open circuit photovoltage (OCP), passing through the short circuit photocurrent (Isc) up to achieving the reversible potential for the endothermic water splitting (1.23 V). Ac-impedance measurements were carried out at OCP under single-sine mode by sweeping frequencies from 100 kHz to 10 mHz using 10 mV r.m.s excitation sinusoidal signal. 4. Conclusions Perfluorosulfonic acid (Nafion® 115) and quaternary ammonium-based anionic (Fumatech® FAA3) ion exchange membranes were investigated for photo-electrolysis applications using TiO2 photo-anode as a model system. This allowed for getting insights about the behaviour of the different membranes both under photo-voltage-driven and bias-assisted operations. The two polymer electrolyte membranes systems were used together with the corresponding ionomer dispersion and determined the occurrence of different reaction environments since pure water only was fed to the cell. In particular, the membrane/ionomer chemistry produced a different operating pH and this influenced significantly the photo-electrolytic conversion. The anionic membrane enhanced the performance of the TiO2 semiconductor compared to the Nafion® membrane as a consequence of the increased concentration of hydroxides with enhanced adsorption of OH species that promote the oxygen evolution reaction. Interestingly, this was not the case of the Ti sub-oxide coated TiO2 in alkaline environment. Such phenomenon was explained with the strong adsorption of OH species in alkaline environment to saturate surface defects in the sub-stochiometric oxide. Thus, strongly bonded surface oxygen species occurred, which were not easily desorbed. Such species could not acquire additional photo-generated minority carriers. These were very likely recombining with photo-generated electrons inside the semiconductor bulk. On the contrary, since the adsorption of OH species on stoichiometric TiO2 is very labile in the acidic environment, the increased adsorption strength at the sub-oxide surface sites played a favourable role for that oxygen evolution in the presence of Nafion. The results clearly demonstrate that the photocurrent characteristics are strongly dependent on both properties and polymer electrolyte membrane photo-anode. Membranes can be tailored to modulate electronic levels at the interface in order to achieve the best properties to match between the semiconductor band gap edges and the redox levels in the electrolyte. This may be especially useful in the case of smaller energy gap semiconductors that can harvest a larger fraction of the solar radiation than TiO2 , thus boosting water splitting efficiency. Author Contributions: Antonino S. Aricò conceived and designed the experiments, and wrote the manuscript; Mariarita Girolamo, David Sebastian, carried out the electrochemical experiments and data analysis; Stefania Siracusano prepared and characterised the Ti-suboxides; Vincenzo Baglio assisted at experimental researches and data analysis; Michael Schuster carried out anionic membrane preparation and characterization. Conflicts of Interest: The authors declare no conflict of interest.

Membranes 2017, 7, 25

15 of 16

References 1.

2. 3. 4.

5.

6.

7. 8. 9. 10. 11. 12.

13.

14.

15.

16. 17.

18.

19. 20.

Aricò, A.S.; Siracusano, S.; Briguglio, N.; Baglio, V.; Di Blasi, A.; Antonucci, V. Polymer electrolyte membrane water electrolysis: Status of technologies and potential applications in combination with renewable power sources. J. Appl. Electrochem. 2013, 43, 107–118. [CrossRef] Lewis, N.S. Powering the Planet. MRS Bull 2007, 32, 808–820. [CrossRef] Peter, L.M.; Upul Wijayantha, K.G. Photoelectrochemical Water Splitting at Semiconductor Electrodes: Fundamental Problems and New Perspectives. ChemPhysChem 2014, 15, 1983–1995. [CrossRef] [PubMed] Aricò, A.S.; Di Blasi, A.; Brunaccini, G.; Sergi, F.; Dispenza, G.; Andaloro, L.; Ferraro, M.; Antonucci, V.; Asher, P.; Buche, S.; et al. High temperature operation of a solid polymer electrolyte fuel cell stack based on a new ionomer membrane. Fuel Cells 2010, 10, 1013–1023. [CrossRef] Siracusano, S.; Baglio, V.; Stassi, A.; Merlo, L.; Moukheiber, E.; Aricò, A.S. Performance analysis of short-side-chain Aquivion® perfluorosulfonic acid polymer for proton exchange membrane water electrolysis. J. Membr. Sci. 2014, 466, 1–7. [CrossRef] Siracusano, S.; Baglio, V.; Van Dijk, N.; Merlo, L.; Aricò, A.S. Enhanced Performance and Durability of Low Catalyst Loading PEM Water Electrolyser Based on a Short-Side Chain Perfluorosulfonic Ionomer. Appl. Energy 2017, 192, 477–489. [CrossRef] Tomkiewicz, M.; Fay, H. Photoelectrolysis of Water with Semiconductors. Appl. Phys. 1979, 18, 1–28. [CrossRef] Doscher, H.; Geisz, J.F.; Deutsch, T.G.; Turner, J.A. Sunlight absorption in water–Efficiency and design implications for photoelectrochemical devices. Energy Environ. Sci. 2014, 7, 2951–2962. [CrossRef] Lopes, T.; Dias, P.; Andrade, L.; Mendes, A. An innovative photoelectrochemical lab device for solar water splitting. Sol. Energy Mater. Sol. Cells 2014, 128, 399–410. [CrossRef] Yang, H.B.; Miao, J.; Hung, S.-F.; Huo, F.; Chen, H.M.; Liu, B. Stable Quantum Dot Photoelectrolysis Cell for Unassisted Visible Light Solar Water Splitting. ACS Nano 2014, 8, 10403–10413. [CrossRef] [PubMed] Yum, J.H.; Chen, P.; Grätzel, M.; Nazeeruddin, M.K. Recent Developments in Solid-State Dye-Sensitized Solar Cells. ChemSusChem 2008, 1, 699–707. [CrossRef] [PubMed] Siracusano, S.; Stassi, A.; Modica, E.; Baglio, V.; Aricò, A.S. Preparation and characterisation of Ti oxide based catalyst supports for low temperature fuel cells International. J. Hydrog. Energy 2013, 38, 11600–11608. [CrossRef] Siracusano, S.; Baglio, V.; D’Urso, C.; Antonucci, V.; Aricò, A.S. Preparation and characterization of titanium suboxides as conductive supports of IrO2 electrocatalysts for application in SPE electrolysers. Electrochim. Acta 2009, 54, 6292–6299. [CrossRef] Varcoe, J.R.; Atanassov, P.; Dekel, D.R.; Herring, A.M.; Hickner, M.A.; Kohl, P.A.; Kucernak, A.R.; Mustain, W.E.; Nijmeijer, K.; Scott, K.; et al. Anion-exchange membranes in electrochemical energy systems. Energy Environ. Sci. 2014, 7, 3135–3191. [CrossRef] Denaro, T.; Baglio, V.; Girolamo, M.; Neri, G.; Deorsola, F.; Ornelas, R.; Matteucci, F.; Antonucci, V.; Aricò, A.S. The influence of physico-chemical properties of bare titania powders obtained from various synthesis routes on their photo-electrochemical performance. Int. J. Electrochem. Sci. 2012, 7, 2254–2275. Fujishima, A.; Honda, K. Electrochemical Photolysis of Water at a Semiconductor Electrode. Nature 1972, 37–38, 238–239. [CrossRef] Denaro, T.; Baglio, V.; Girolamo, M.; Antonucci, V.; Aricò, A.S.; Matteucci, F.; Ornelas, R. Investigation of low cost carbonaceous materials for application as counter electrode in dye-sensitized solar cells. J. Appl. Electrochem. 2009, 39, 2173. [CrossRef] Hernández, S.; Gerardi, G.; Bejtka, K.; Fina, A.; Russo, N. Evaluation of the charge transfer kinetics of spin-coated BiVO4 thin films for sun-driven water photoelectrolysis. Appl. Catal. B 2016, 190, 66–74. [CrossRef] Seabold, J.A.; Zhu, K.; Neale, N.R. Efficient solar photoelectrolysis by nanoporous Mo:BiVO4 through controlled electron transport. Phys. Chem. Chem. Phys. 2014, 16, 1121–1129. [CrossRef] [PubMed] Kaouk, A.; Ruoko, T.-P.; Gonullu, Y.; Kaunisto, K.; Mettenborger, A.; Gurevich, E.; Lemmetyinen, H.; Ostendorf, A.; Mathur, S. Graphene-intercalated Fe2 O3 /TiO2 heterojunctions for efficient photoelectrolysis of water. RSC Adv. 2015, 5, 101401–101412. [CrossRef]

Membranes 2017, 7, 25

21.

22. 23.

24.

25. 26. 27.

16 of 16

Hernandez, S.; Barbero, G.; Saracco, G.; Alexe-Ionescu, A.L. Considerations on Oxygen Bubble Formation and Evolution on BiVO4 Porous Anodes Used in Water Splitting Photoelectrochemical Cells. J. Phys. Chem. C 2015, 119, 9916–9925. [CrossRef] Su, J.; Guo, L.; Bao, N.; Grimes, C.A. Nanostructured WO3 /BiVO4 Heterojunction Films for Efficient Photoelectrochemical Water Splitting. Nano Lett. 2011, 11, 1928–1933. [CrossRef] [PubMed] Bard, A.J.; Bocarsly, A.B.; Ren, F.; Fan, F.; Walton, E.G.; Wrighton, M.S. The concept of Fermi level pinning at semiconductor/liquid junctions. Consequences for energy conversion efficiency and selection of useful solution redox couples in solar devices. J. Am. Chem. Soc. 1980, 102, 3671–3677. [CrossRef] Park, H.S.; Kweon, K.E.; Ye, H.; Paek, E.; Hwang, G.S.; Bard, A.J. Factors in the Metal Doping of BiVO4 for Improved Photoelectrocatalytic Activity as Studied by Scanning Electrochemical Microscopy and First-Principles Density-Functional Calculation. J. Phys. Chem. C 2011, 115, 17870–17879. [CrossRef] Cots, A.; Cibrev, D.; Bonete, P.; Gómez, R. Hematite Nanorod Electrodes Modified with Molybdenum: Photoelectrochemical Studies. ChemElectroChem 2016. [CrossRef] Montanini, R.; Freni, F.; Rossi, G.L. Quantitative evaluation of hidden defects in cast iron components using ultrasound activated lock-in vibrothermography. Rev. Sci. Instrum. 2012, 83, 094902. [CrossRef] [PubMed] Aricò, A.S.; Sebastian, D.; Schuster, M.; Bauer, B.; D’Urso, C.; Lufrano, F.; Baglio, V. Selectivity of Direct Methanol Fuel Cell Membranes. Membranes 2015, 5, 793–809. [CrossRef] [PubMed] © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).