A cost-effective nanoporous ultrathin film electrode

Journal of Power Sources 342 (2017) 947e955

Contents lists available at ScienceDirect

Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour

A cost-effective nanoporous ultrathin film electrode based on nanoporous gold/IrO2 composite for proton exchange membrane water electrolysis Yachao Zeng a, b, Xiaoqian Guo a, b, Zhigang Shao a, *, Hongmei Yu a, **, Wei Song a, b, Zhiqiang Wang a, b, Hongjie Zhang a, b, Baolian Yi a a b

Fuel Cell System and Engineering Laboratory, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian, Liaoning, 116023, China Graduate University of Chinese Academy of Sciences, Beijing, 100049, China

h i g h l i g h t s A nanoporous ultrathin film (NPUF) electrode is constructed for water electrolysis. The NPUF electrode is a composite ultrathin film of nanoporous gold/IrO2. The NPUF electrode is prepared from a facile thermal annealing method. The NUTF electrode is featured in its ultrahigh electrocatalyst utilization. The NUTF electrode presents a much improved single cell performance.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 15 June 2016 Received in revised form 3 January 2017 Accepted 4 January 2017 Available online 11 January 2017

A cost-effective nanoporous ultrathin film (NPUF) electrode based on nanoporous gold (NPG)/IrO2 composite has been constructed for proton exchange membrane (PEM) water electrolysis. The electrode was fabricated by integrating IrO2 nanoparticles into NPG through a facile dealloying and thermal decomposition method. The NPUF electrode is featured in its 3D interconnected nanoporosity and ultrathin thickness. The nanoporous ultrathin architecture is binder-free and beneficial for improving electrochemical active surface area, enhancing mass transport and facilitating releasing of oxygen produced during water electrolysis. Serving as anode, a single cell performance of 1.728 V (@ 2 A cm2) has been achieved by NPUF electrode with a loading of IrO2 and Au at 86.43 and 100.0 mg cm2 respectively, the electrolysis voltage is 58 mV lower than that of conventional electrode with an Ir loading an order of magnitude higher. The electrolysis voltage kept relatively constant up to 300 h (@250 mA cm2) during the course of durability test, manifesting that NPUF electrode is promising for gas evolution. © 2017 Elsevier B.V. All rights reserved.

Keywords: Proton exchange membrane water electrolysis Nanoporous gold Nanoporous ultrathin film electrode Membrane electrode assembly Ultralow iridium loading

1. Introduction Hydrogen is an important constituent of renewable energy infrastructure. With the growing capacity of renewable energy sources surpassing gigawatt, a storage system with equal magnitude is in highly demand. Water electrolysis provides solution for large scale hydrogen production, and is well suited to couple with renewable energy sources such as wind and solar. Compared with

* Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (Z. Shao), [email protected] (H. Yu). http://dx.doi.org/10.1016/j.jpowsour.2017.01.021 0378-7753/© 2017 Elsevier B.V. All rights reserved.

the conventional alkaline water electrolysis, proton exchange membrane (PEM) water electrolysis takes advantages in its higher energy efficiency, smaller installation footprint and product gas purity [1]. However, PEM water electrolyzers are rarely applied in large scale hydrogen production [2]. One factor that hinders its wide application is the high consumption of iridium in membrane electrode assembly (MEA), in which the water splitting takes place. Nonetheless iridium is one of the rarest elements in the Earth's crust [3]. To reduce the high consumption of iridium in PEM water electrolyzer, one strategy is to synthesize electrocatalyst for oxygen evolution reaction (OER) with enhanced mass specific activity [4e7], another is to construct MEA with rationally designed structure. Currently there are two MEA fabrication methods,

948

Y. Zeng et al. / Journal of Power Sources 342 (2017) 947e955

including the gas diffusion layer electrode (GDE) method and catalyst-coated membrane (CCM) method. It is generally recognized that MEA prepared with CCM method exhibits better performance than those prepared with GDE method [1,2,8,9]. CCM method could possibly enhance the interfacial structure between membrane and catalyst layer, leading to improved catalyst utilization and smaller interfacial resistance [10]. To perform CCM method, a mixture of catalyst, ionomer and solvent is sonicated to form a homogeneous ink, then the catalyst ink is coated on substrates through decal [11], blade coating [12], spraying [13], or brushing [14] process. The conventional CCM fabrication methods take disadvantages in high noble metal consumption and low electrocatalyst utilization. The IrO2 loading in the current commercial solid polymer water electrolyser was 2 mg/cm2 or more on their anodes [15]. The obtained CCMs are often randomized in ionomer, porosity and electrocatalyst. Such random configuration results in insufficient utilization of electrocatalyst. Besides, the introduction of ionomer in catalyst layer would block the active sites, reduce the porosity and electronic conductivity as well, leading to a decrease in electrocatalyst utilization. It has been found that applied voltage accelerates the degradation of ionomer, which will eventually leads to a disintegration of the catalyst layer [16]. The impact of MEA structure on single cell performance has been intensively studied in the field of proton exchange membrane fuel cell. Mathematical model [17,18] and experimental work [19e22] have revealed that ordered MEA with rationally designed catalyst layer structure would enhance the single cell performance and reduce the noble metal consumption. Unlike the situation in PEM fuel cell, the application of ordered MEA in PEM water electrolysis has been rare [23,24]. One successful application of ordered MEA, which is denoted as Nanostructured Thin Film electrode (NSTF), has been achieved by Debe et al. in PEM water electrolysis [25]. NSTF electrodes are characterized in their ultrathin thickness, welldefined porosity and free of Nafion ionomer. The advantages of ultra-thin catalyst layer are obvious: the straight pore and ultrathin thickness facilitate the transport of reactant and product. Another distinguishing feature is that NSTF electrode doesn't employ Nafion ionomer as proton conductor. Hence, the cost and durability associate with Nafion ionomer addition in catalyst layer will be eliminated. However, the fabrication procedure of NSTF is complex and unthrifty. Firstly, the electronic insulating catalyst support for NSTF electrode was prepared by annealing the sputtered organic film (e.g. PR149) under ultrahigh vacuum [26], then noble metal with high loading was deposited onto the nanostructured support to form a continuous conducting film. The noble metal loading of the obtained NSTF electrode for PEM water electrolysis was >2 mg cm2 on their anode [25]. To meet the demand for constructing an electrode for PEM water electrolysis with high performance and cost efficiency, an ordered MEA based on ultrathin nanoporous gold (NPG)/IrO2 composite film has been designed for PEM water electrolysis. Nanoporous gold (NPG) has relatively high surface area, high porosity, excellent electronic conductivity and ultra-thin thickness [27]. By electrochemical [28] or chemical [29] removal of Ag from AgAu alloy, the remaining gold undergoes a self-organization process forming a 3D bicontinuous network with interconnected ligaments. By extending dealloy time or thermal coarsening temperature [29,30], the structure of NPG can be easily tuned. The unique structure of NPG is perfect for PEM water electrolysis. Firstly, the 3D biocontinuous pore and ultra-thin thickness reduce the mass transport resistance of water penetration and oxygen separation. Secondly, 3D bicontinuous network with interconnected ligaments makes an absence of ionomer to be possible. Thirdly, the interconnected ligaments form a perfect electric conductor. Though the first recorded experiment on water

electrolysis was carried out with gold electrode in 1789 [32,33], Au has not been viewed as candidate material for water electrolysis for its high onset potential towards OER [34]. Another concern is the slow electrochemical dissolution of Au under the hash operating condition in PEM water electrolysis [35]. In this work, a novel cost-efficient 3D OER electrode based on NPG/IrO2 composite ultrathin film electrode has been constructed. The structure of nanoporous gold/IrO2 composite ultrathin film based MEA has been illustrated in Scheme 1. Different from the NSTF electrode, the NPUF electrode features in (1) all pores are 3D interconnected, (2) NPG serves as network to guarantee the integrity of electrode, (3) IrO2 nanoparticles embedded into the NPG forming a composite ultrathin film. Benefiting from the ultrathin thickness and 3D interconnected nanoporosity, the generated gas can be easily separated from the active site, thus a high electrocatalyst utilization is achieved. The novel NPUF electrode is promising for PEM water electrolysis. 2. Experimental 2.1. Chemicals All chemicals were used as received without further purification. 12-carat Au foil (Ag/Au, 1:1 ratio by weight) was purchased from Sepp Leaf Products (New York). H2IrCl6$6H2O was obtained from KunMing Borui Co. Nafion® 212 membrane and Nafion® ionomer solution (5 wt%) were purchased from DuPont. Pt/C (70 wt%) and iridium black (HiSPEC™ 160000) were purchased from Johnson Matthey. Natural mica plates (24 mm 39 mm) were purchased from Meida Mica Company (Shenzhen, China). 2.2. Preparation of NPG The NPG film was prepared by chemical de-alloying. The detail procedure is as follows: firstly, 12 K Au foil was tailored to 2.2 cm 2.2 cm, the tailored foil was then expanded on the surface of deionized water. Secondly, the Au foil was transferred onto the surface of 11 mol L1 nitrite acid solution, the dealloying time was 2 h, and the temperature was set to be 25 C. When the dealloying was terminated, NPG was transferred onto the surface of de-ionized water, the de-ionized water was refreshed for 3e4 times to remove the residual nitrite acid in NPG. Then a mica plate was employed to withdraw the NPG from de-ionized water surface. The NPG together with mica was then dried at 25 C in air. 2.3. Preparation of NPG/IrO2 composite ultrathin film The NPG/IrO2 composite film were prepared by thermal decomposition method. Firstly, the NPG thin films supported on micas were immersed into 0.1 M H2IrCl6 2-propanol solution for 3 min at 25 C and dried in air; then calcination was carried out in a ceramic tube furnace at desired temperature for 30 min in air and cooled down to room temperature. To investigate the effect of calcination temperature on water electrolysis, the temperature was set to be 450 C, 500 C and 600 C. 2.4. Preparation of MEA and single cell assembly The CCMs were prepared via decal method. For anode catalyst layer, the decal was performed at 140 C, 2.0 MPa for 2 min. For cathode catalyst layer, the decal was performed at 140 C, 0.5 MPa for 2 min. After calcination, the NPG/IrO2 composite thin films as anodes were transferred from mica plates onto Nafion® 212 membrane. Mica plates were peeled off carefully. The cathode catalyst layers were fabricated by air brushing the catalyst ink onto


949

Scheme 1. An illustration of the MEA construction based on NPG/IrO2 composite ultrathin film electrode.

PTFE sheets and then decaled onto Nafion® 212 membrane. The Pt/ C loadings were set to be 1.0 mg cm2. For comparison, Ir black was chosen as anode catalyst for conventional MEA. Ir black, Nafion ionomer and iso-propanol was sonicated in a mass ratio of 3:1:50 to form a uniform catalyst ink, then catalyst ink was air brushed onto PTFE sheet, the Ir black-based catalyst layer was decaled onto Nafion membrane under the identical condition to that of NPUF-1. The Ir black loading was set to be 0.873 mg cm2. Pt-coated porous titanium plate was employed as anode gas diffusion layer (GDL), and carbon paper (Toray®, TGP-H-60) with microporous layer was employed as cathode GDL. The CCM and cathode GDL were assembled by hot pressing at 140 C and 0.05 MPa for 3 min. The active surface area for both cathode and anode is 4.0 cm2. For simplicity, MEAs prepared by assembling composite films calcinated at temperatures of 450 C, 500 C and 600 C were denoted as NPUF-1, NPUF-2 and NPUF-3. Then MEAs were sandwiched between two Ptecoated end plates with parallel flow fields, and silicon gaskets were positioned between MEA and each of the Ptcoated end plates.

2.5. Characterizations The surface and cross section morphologies of the electrodes were investigated by scanning electron microscopy (SEM, JEOL JSM-7800F). The cross sections of the electrodes were obtained by quenching the sample in liquid nitrogen. High resolution TEM (HRTEM) was carried out with JEM 2100 (JEOL). High angle annular dark field scanning transmission microscopy (HAADF-STEM) and mapping were carried out with a Tecnai G2 F30 microscope (FEI). To reveal the detail structural configuration of NPUF-1, ultramicrotomy was performed to obtain its cross section. The crystal structure was characterized by X-ray diffraction using an X-ray diffractometer (PANalytical EMPYREAN) operating at 40 kV and 200 mA with Cu Ka (l ¼ 1.5405 Å) as a radiation source. X-ray photoelectron spectroscopy (XPS) analysis was performed to reveal the surface properties of NPUF-1. The XPS data were obtained with an ESCALab250 Xi electron spectrometer using 300 W Al Ka radiation. The

base pressure was about 1 107 Pa. The binding energies were referenced to the C1s line at 284.8 eV from adventitious carbon. For NPUF, the IrO2 and Au loadings were determined by dissolving the H2IrCl6-dipped NPG without calcination in aqua regia and analyzed by inductively coupled plasma optical emission spectroscopy (ICPOES). The single cell performance was tested on a home-made test stand [36]. I-V curves were recorded at 80 C and ambient pressure in galvanic static mode. Pure water was fed to the anode at a flow rate of 10 mL min1. The durability test was conducted at 250 mA cm2 and 80 C. To get better understanding of the single cell performance evaluation, electrochemical impedance spectroscopy (EIS) was performed prior and post to the durability test. EIS data was recorded at a constant potential of 1.45 V by applying an ac amplitude of 10 mV over the ac frequency range from 1 Hz to 10 kHz. The EIS was carried out on Solartron 1287 Electrochemical Interface in conjunction with a Solartron 1260 Frequency Respond Analyzer. After single cell evaluation, the Pt-coated porous Ti plate was replaced with carbon paper (Toray®, TGP-H-60) in single cell assembly. The anode was purged with pre-humidified N2 and the cathode was purged with pre-humidified H2. The cell temperature was 30 C. The electrochemical characterization was performed on an electrochemical analyzer (CHI 600C, CH Instruments, USA). The scan rates were set to be 2, 5, 10, 20, 30, 40, 50, 100, 200, 300, 400 and 500 mV/s, the potential range was (0.40 V, 1.40 V) vs. RHE.

3. Results and discussion 3.1. Synthesis and characterization The structure of bare NPG is presented in Fig. 1. From Fig. 1a and c, a 3D biocontinous skeleton network with open porosity was achieved post to a fully dealloying process, and such unique structure merits retains in a form of membrane (Fig. 1b). The characteristic pore diameter of NPG is ca. 15.7 nm (Fig. S1), and its average thickness is ca. 125 nm. The pore diameter and the

950


Fig. 1. (a) in panel and (b) cross sectional SEM image of NPG. (c) TEM image of NPG skeleton. (d) HRTEM image of ligament of NPG.

thickness of NPG were derived from an image processing method [37]. The porosity of the fully dealloyed NPG is estimated to be 64.7%, this value was calculated according to the weight ratio of AuAg alloy. The high-resolution TEM (HRTEM) image (Fig. 1d) indicates that the ligament of NPG is ca. 20 nm, the lattice fringe is 2.36 Å. By IUPAC definition, the as-prepared material is mesoporous. The formation process of NPG has been qualitatively depicted by Erlebacher et al. [38]. Firstly, Ag atoms selectively dissolves away at the interface of alloy and electrolyte. Secondly, the Ag atoms that are laterally coordinated to the dissolved atom have fewer bonds, which are more susceptible to dissolute compared to the Ag atoms further in the bulk. The gold atoms in the topmost terrace do not dissolve, while they agglomerate into clusters. And the etch front propagated into the bulk of the alloy, eventually a thorough dissolution of Ag is achieved leaving an Au skeleton with open nanoporous structure. Zinchenko et al. [39] performed a Monte Carlo simulation to study the formation process of NPG, and their work confirmed Erlebacher's hypothesis. As the dealloying process terminated, the NPG was loaded onto a natural mica plate. The employment of natural mica to transfer NPG was firstly reported by R. Zeis [40]. NPG with a thickness of hundreds nanometers is extremely fragile, but it expands easily on the water surface by the surface extension strength without the danger of crush, this inspired us that substrate with hydrophilic property can also employed to stabilize NPG. In this work, 2propanol with relatively low boiling point (82.45 C) and viscosity (2.4 mPa s @ 20 C) has been chosen as solvent. When H2IrCl6 2propanol solution is dipped on NPG, the solution will instantly penetrate into the nanopore space under the driving of capillary pressure. As the solvent evaporates, H2IrCl6 is confined into the nanopore, then calcination is carried out in tube furnace. As the calcination terminated, NPUF electrode was prepared by directly decaling the NPG/IrO2 composite ultrathin film onto Nafion membrane. Fig. 2 aec are the HRTEM images of NPUF-1. A more detail catalyst layer structure information is presented in Fig. S2. Compared with the bare NPG, the structure of NPG was reserved except for that its mesopore was filled with IrO2 nanoparticles. From Fig. 2a, the IrO2 nanoparticles packed loosely in NPG, the

loose structure may derive from the gas releasing from the thermal decomposition of H2IrCl6. Such configuration is vital for the transport of reactant and product, which will be discussed later. Fig. 2b indicates that the skeletons of NPG were covered with a thin layer of IrO2. The IrO2 thin film is a composite of amorphous and crystalline IrO2 nanoparticles. The lattice fringe of the crystallized IrO2 is 2.59 Å, which is slightly large than that of NPG. From Fig. 2c, the thickness of the resulted catalyst layer is ca. 200 nm. Though without addition of ionomer on the interface between Nafion membrane and catalyst layer, the as-prepared electrodes presents an excellent dimensional stability even under the HRTEM test condition of ultra-high vacuum. The cross-sectional images of NPUF-1 and conventional electrode are presented in Fig. S4. The catalyst layer thickness for conventional electrode is estimated to be 1.2 mm. It can be seen that the iridium black agglomerated into large spheres, the diameter of the spheres is up to 500 nm. According to the analysis result of inductively coupled plasma optical emission spectroscopy (ICP-OES), the IrO2 and Au loading for nanostructured catalyst layer are 86.43 and 100 mg cm2 respectively. While the loading of Ir black in conventional MEA is 873 mg cm2. The surface properties of NPUF-1 were investigated by X-ray photoelectron spectroscopy (XPS). The results are presented in Fig. 2dee. Two binding states of Ir are identified as 4f7/2 at 61.6 eV and 4f5/2 at 64.6 eV, which are a bit of lower to those reported IrO2 values of 61.9 and 64.8 eV. Two broader feature at 62.7 and 66.05 eV are observed, which are higher than [Ir4þ] 4f7/2 at 62.5 eV and [Ir4þ] 4f5/2 at 65.9 eV. The O 1s signal peak appears at 532.06 eV, which is higher than the standard IrO2 at 530.5 eV [41]. Two strong peaks at 84.7 and 88.4 eV are from Au 4f7/2 and 4f5/2. For the pure Au, the peaks of 4f7/2 and 4f5/2 locate at 84.0 eV and 87.94 eV, the peak shifts to higher values demonstrates the transfer of electronic charge from Au to Ir oxides [42]. The higher binding energy of Ir 4f and O 1s signals indicates a higher oxidation state of Ir in the NPUF1. The oxygen peak indicates that the oxygen signal of the surface hydroxides but not the lattice oxygen. It is well known that Au is the most electronegative transition metal, it could serve as electron sink to facilitate the transfer of electrons from water electrooxidation, the electronegativity is further enhanced due to the


951

Fig. 2. (a, b) HRTEM images of NPUF-1. (c) the cross sectional HRTEM image of NPUF-1 electrode. XPS spectra of NPUF-1in the d) Ir 4f, e) Au 4f and f) O 1s.

high oxidation state of Ir, which favors the water splitting reaction. The XRD patterns for the NPUF electrodes and also NPG are presented in Fig. 3. Five distinct peaks corresponding to the (111), (200), (220), (311) and (222) lattices panels of Au crystals are observed in all the samples, which are identical to their standard values. In the case of the NPUF electrodes, with the progressively increase of calcination temperature, broad reflections at 2q ¼ 27.57, 34.63 , and 53.9 gradually appeared, the signals correspond with the (110), (101), and (211) lattices panels of IrO2 crystals. The gradual appearance of characteristic diffraction peaks of IrO2 indicates that a phase transition of IrO2 from amorphous into crystalline structure. According to the Scherrer equation, the average crystallite size of IrO2 is estimated to be 2.72 nm (NPUF-1), 6.2 nm (NPUF-2) and 9.45 nm (NPUF-3). Calcination above 450 C results in crystalline IrO2 nanoparticles. 3.2. Electrochemical characterization The electrochemical surface information of the prepared electrodes were investigated by cyclic voltammetry (CV). The

Fig. 3. XRD patterns of NPG and NPUF electrodes annealed at 450 C, 500 C and 600 C with standard reference patterns listing below.

electrochemical surface areas (ECSAs) of the prepared electrodes are quantified by determining the mean value of the anodic and cathodic charge of the CVs and normalized to the electrode active surface area and Ir loadings at a scan rate of 50 mV s1 [23,43,44]. In Fig. 4a, a typical pseudo capacitive behaviour is observed for all electrodes which is mainly attributed to the reversible oxidation and reduction on the IrO2 surface. For NPUF electrodes, the ECSAs decreased along with the calcination temperature. Though the ECSA of NPUF-3 is the smallest among NPUF electrodes, its value is still nearly 3 times higher than that of conventional MEA. The much improved specific ECSA of NPUF electrodes manifests a better dispersion of IrO2 nanoparticles on the skeletons of NPG. The decay of ECSAs was mainly attributed to IrO2 crystallization. As indicated in Fig. 3, with the temperature increases, a phase transition of IrO2 from amorphous into crystalline structure occurs, leading to a less active sites exposure. Fig. S5 presents the CVs of the prepared electrodes at different scan rates obtained at 30 C. Fig. 4b presents the voltammetric charges (Q) of the prepared electrodes obtained at different scan rates. It can be seen that the charge decreased with increasing scan rate, especially in the lower scan rate region. It has been proposed that the dependence of Q on the scan rate related to two processes. One is related to the slow proton diffusion within the oxide catalyst film, which was the rate determine step of the charging/discharging process [45]. Another process is related to a fast charging/discharging of the electrical double layer at the electrode-electrolyte interface [46]. Fierro et al. [47] investigated the charging/discharging process of an IrO2 film deposited on a p-Si substrate at different scan rates and temperatures and obtained the activation energy (Ea) for the charging/discharging process. It has been found that the Ea strongly depended on the scan rate. At low scan rate, the Ea for the charging/discharging process has a value of 2.4 kJ mol1 [47], which is related to the slow proton diffusion in IrO2 film. At high scan rate the Ea is close to zero [47], which is related to a fast charging/discharging of the electrical double layer at the electrode-electrolyte interface. By increasing the scan rate, the slow charging process related to proton diffusion was gradually excluded and only the charging of the double layer persisted due to its scan rate dependence. As proposed by Ardizzone et al. [45], the voltammetric charges obtained at different scan rates, namely the “outer” voltammetric charges (QO) and the “inner” voltammetric charges (QT), give two kinds of ECSA information of the metal oxide.

952


Fig. 4. (a) Characteristic CVs recorded at 30 C with a scan rate of 50 mV s1 (b) charge as function of scan rates for the NPUF electrode and conventional MEA. (c) a comparison of specific ECSAs of NPUF electrode and conventional MEA. (d) a comparison of Qo/QT for the NPUF electrode and conventional MEA.

The former (QO) corresponds to the easily accessible active surface area which can be derived as the scan rate extrapolated to infinity, while the latter (QT) corresponds to the “total” surface area and can be obtained as the scan rate approaching to “zero” [48]. Therefore the charge ratio of QO/QT is associated with the proton conduction and porosity of the electrode [23,49]. Trasatti et al. [45] proposed an experimental method to derive QO and QT, however the method is not always valid. A more feasible method is to choose the voltammetric charge at scan rate of 2 mV/s as QO and the voltammetric charge at scan rate of 500 mV/s as QT. From Fig. 4b, the voltammetric charges of NPUF electrodes take advantages over conventional MEA at all scan rates, indicating that more specific accessible area and total area of NPUF electrodes than that of conventional MEA. For NPUF electrodes, the voltammetric charges decrease as the electrode calcination increase. The result is within accordance with the ECSA characterization. This is because the cracks, pores, crevices, and grain boundaries of the metal oxide film contributed to the proton diffusion and active sites. As the calcination temperature increases, IrO2 nanoparticles experienced a phase change from amorphous to crystallization. NPUF electrode prepared at higher calcination temperature has fewer surface defects, hence has a lower voltammetric charge. From Fig. 4b, NPUF-1 has the largest value in both QO and QT, indicating that NPUF-1 has more accessible area and faster proton diffusion rate. For the conventional MEA, however, the QO and QT fell below. Though there is no proton exchange ionomer presents in the NPUF electrodes, the values of QO which is related to proton diffusion are much higher than that of conventional MEA with Nafion ionomer serving as the proton conductor, indicating that the proton diffusion beneath the subsurface of IrO2 film in NPUF electrodes is sufficiently fast. For the ultrathin thickness and ultralow electrocatalyst loading for NPUF electrode, it is difficult to perform the porosimetry. Even the data derived from physical characterization with effort cannot yet fully

elucidate the reaction process, noting that the active surface area or porosity involved in reaction is different from that derived from physical characterization. While an in situ porosimetry under single cell configuration is more convincible to reveal the nature of an electrode. Here the charge ratio was chosen to evaluate the porosity of the prepared electrodes. The charge ratio of QO/QT was summarized in Fig. 4d. For NPUF electrodes, as indicated from Fig. 4d, the charge ratio decreases along with calcination temperature increases. This is because higher calcination temperature leads to nanoparticle sintering, the agglomerated nanoparticles block the pore leading to a reduction in porosity. For conventional MEA, the reduction of ECSA and also the charge ratio could be attributed to the introduction of ionomer. The introduction of ionomer has two contrary impacts on the catalyst layer in PEM water electrolysis. On one hand, ionomer provides dimensional stability and proton conductivity for catalyst layer. On the other hand, the introduction of ionomer in catalyst layer reduces the porosity and electronic conductivity leading to a decrease in electrocatalyst utilization. 3.3. Single cell performance The single cell performance of the prepared electrodes is presented in Fig. 5. For PEM water electrolysis, the I-V curve can be divided into three regions: the activation polarization region, the ohmic polarization region and mass transport polarization region. The activation polarization occurs in low current density range, while mass transport polarization occurs over the entire region of IV curve, but it becomes prominent in the high current density region [50]. In the low current density range (1000 mA cm2), a noticeable divergence of I-V curves has be observed. The divergence may originate from mass transport polarization which is strongly dependent on electrode structure. During the course of water electrolysis, mass transport is coupled with electrochemical

reactions. With the increase of electrolysis current density, mass transport becomes prominent. An insufficient mass supplement will lead to an increase in diffusion overpotential at the anode and cathode, which is quantified by Nernst equation [52]. In terms of mass transport in water electrolysis, the electrode structure decides the mass transport length and effective diffusive coefficient. Middleman proposed that electrode with parallel straight porosity was ideal for gas diffusion electrode [53], however, this does not certainly mean that such electrode configuration is ultimate. Much effort is still needed in developing in situ characterization method such as soft X-ray tomography to reveal the transport phenomena of microfluid flow in gas diffusion electrode [54]. In this work, NPUF with ultrathin thickness and 3D interconnected nanoporosity was proposed in a pursuit of developing novel electrode with high performance and cost efficiency. From Fig. S4, the catalyst layer thickness of NPUF-1 is ca. 200 nm, while the catalyst layer thickness of conventional electrode is ca. 1.2 mm. From Fig. S2, the presence of IrO2 nanoparticles in NPG has reduced the porosity of NPG an order of magnitude lower. During the course of water electrolysis, twophase flow couples with electronic and ionic conductions. A reduction in catalyst layer thickness will accelerate dissipation of gases and enhance penetration of water, more importantly the ionic and electronic conducting pathway will be shortened. It has been expected that macro porosity favours the process of water penetration and gas dissipation in gas evolution electrode [55]. However, the performance of NPUF electrodes provides a more comprehensive view of transport phenomenon in gas evolution electrode. Though there is no macroporosity presents in NPUF electrodes, no severe transport polarization has been observed. This is because the transport of water in a hydrophilic substrate with nanoporosity is driven by capillary pressure. According to Young-Laplace equation, the capillary pressure is inversely

954


proportional to the pore diameter and proportional to the cosine function of contact angle (qC). Since the capillary pressure is defined as the difference between gaseous phase and liquid phase, for a hydrophilic porous substrate (qC < 90 ), the capillary pressure will be always positive. As the pore diameter reduced to a lever of nanometer, enormous capillary pressure will present on the interface of liquid/gas phase, causing an expulsion of gas out of NPUF electrode. For NPUF-1 and conventional MEA, the electrolysis voltage is 58 mV lower than that of the conventional MEA at 2.0 A cm2, noting that its IrO2 loading is only 86.43 mg cm2. For NPUF-2 and conventional MEA, although the specific intrinsic activity of NPUF2 is much lower than that of conventional MEA as indicated from the I-V curves in low current density range (