Tandem Si Micropillar Array Photocathodes with Conformal Copper Oxide and a Protection Layer by Pulsed Laser Deposition.

This work demonstrates the influence of high-quality protection layers on Si-Cu2O micropillar arrays created by pulsed laser deposition (PLD), with the goal to overcome photodegradation and achieve long-term operation during photoelectrochemical (PEC) water splitting. Sequentially, we assessed planar and micropillar device designs with various design parameters and their influence on PEC hydrogen evolution reaction. On the planar device substrates, a Cu2O film thickness of 600 nm and a Cu2O/CuO heterojunction layer with a 5:1 thickness ratio between Cu2O to CuO were found to be optimal. The planar Si/Cu2O/CuO heterostructure showed a higher PV performance (Jsc = 20 mA/cm2) as compared to the planar Si/Cu2O device, but micropillar devices did not show this improvement. Multifunctional overlayers of ZnO (25 nm) and TiO2 (100 nm) were employed by PLD on Si/Cu2O planar and micropillar arrays to provide a hole-selective passivation layer that acts against photocorrosion. A micropillar Si/ITO-Au/Cu2O/ZnO/TiO2/Pt stack was compared to a planar device. Under optimized conditions, the Si/Cu2O photocathode with Pt as a HER catalyst displayed a photocurrent of 7.5 mA cm-2 at 0 V vs RHE and an onset potential of 0.85 V vs RHE, with a stable operation for 75 h.

Introduction

Photoelectrochemical water splitting is an attractive way to produce clean hydrogen fuel by economic and convenient utilization of solar energy.[1−4] The concept of photoelectrochemical (PEC) tandem device configurations that combine dual semiconducting photoabsorbers carries great potential for robust, inexpensive, and efficient unassisted solar water splitting.[5−8] In such a solar-to-fuel (S2F) device, sunlight is absorbed by a pair of semiconductor photoelectrodes to split water into hydrogen and oxygen.[9,10] Commonly, a bias voltage is needed from an external energy source for effective overall water splitting.[11] However, a practical S2F device should be a stand-alone cell with no need for an external voltage supply.[12] To gain sufficient photovoltage, many researchers have used a combination of two or more semiconductors in a tandem configuration.[13−15]

Doscher et al. and Cheng et al. have reported that the maximum achievable solar-to-fuel efficiencies for the production of H2 are achieved using a pair of light absorbers with bandgaps of approximately 1.8 and 1.0 eV.[16,17] Copper oxides (Cu2O and CuO) are interesting top absorber photocathode materials,[18−21] with narrow bandgap energies of 2.1 and 1.6 eV, which allow Cu2O and CuO to absorb a wide range of the solar spectrum.[22,23] The combination of Cu2O with Si, which has a bandgap of 1.1 eV, is therefore a logical choice for a tandem configuration. However, copper oxides present a mismatch between light absorption and carrier diffusion length, and, as a result, their photovoltage is limited, which restricts their use as effective photocathodes for unassisted water splitting.[18,24]

To solve the photovoltage problem, an approach is to build a photovoltaic (PV)-photoelectrochemical (PEC) tandem configuration by combining a PEC device and series-connect it with a buried-junction PV cell.[25−28] Recent development in PV applications inspired the fabrication of 3D-structured Si micropillar arrays.[29−31] A micropillar-based Si/Cu2O tandem device design offers many advantages including: (i) the series addition of the photovoltage in both components of the tandem to split water without external bias,[32,33] (ii) orthogonal absorption of visible light and photogenerated charge carrier separation and transport,[34,35] (iii) increased surface area for catalytic reactions, (iv) decreased usage of photoactive material relative to planar device designs, depending on pillar pitch and diameter,[36] and (v) the opportunity to spatially decouple light absorption and catalytic activity.[29]

To consider Si and Cu2O as effective photocathode materials in the tandem device, however, surface passivation is needed to prevent photodegradation.[19,37] Han et al. presented a simple method to prepare a Cu2O/CuO heterostructure to protect the Cu2O photocathode from photocorrosion in the PEC water splitting reaction.[38] The as-prepared Cu2O/CuO composite showed improved stability for the hydrogen evolution reaction by increasing the coverage of CuO on the Cu2O film. However, CuO has limited stability in aqueous electrolytes as suggested by the Pourbaix diagram.[39] Many studies have reported CuO as a photocathode for the PEC water reduction reaction, although it is uncertain how much of the photocurrent leads to H2 evolution.[40−42]

In other studies, attempts to address this issue and to achieve long stability were focused on modifying the semiconductor surface with various n-type oxides (hole-selective layers) and cocatalyst layers like n-Ga2O3,[43] AZO,[18] CdS,[39] NiO,[44] TiO2,[45] ITO,[46] Pt,[47] NiSi,[30] etc. Many researchers investigated semiconductor metal oxide materials, such as ZnO,[48] WO3,[49] Cu(O,[50,51] SnO(,[52] and V2O5,[53] either as nanoparticles or as continuous films, to create sensors (e.g., for humidity, ultraviolet, and biological analytes), and their sensing properties were found to be dependent on size, shape, defects, and dopants.[54] Furthermore, the effect of various dopants (Mg, Al, Ga, Fe, Ni)[55−60] in the electron-selective oxide layer to increase the donor density to effectively prevent any resultant charge recombination additionally improves the photovoltage.[61,62] Various deposition systems and techniques (i.e., atomic layer deposition (ALD),[63] pulse laser deposition,[64] chemical vapor deposition,[65] spray pyrolysis,[66] sputtering,[67] anodic oxidation,[68] thermal oxidation,[69] and electrochemical deposition[70]) have been employed for passivating semiconductor photoabsorbers, each with their own advantages and limitation.[71−73] Pulse laser deposition (PLD) is an emerging technique for industrialized use,[74] and can achieve large-area uniform coating on 3D structures. As a result of passivation, the semiconductor can be protected from the harsh acidic/basic electrolyte environment often needed for long-term operation.

Here, we demonstrate the use of pulsed laser deposition (PLD), by which ZnO and TiO2 overlayers can be grown directly and homogeneously on Si/Cu2O microstructures with an accurate film thickness and composition. Recently, we showed that a tandem of Si micropillars with copper oxide can provide efficient water-splitting photocathodes.[75] We here report the fabrication of tandem Si/Cu2O micropillar array devices with passivation layers of ZnO and TiO2 with a precise thickness deposited by PLD. This study exhibits that the PLD approach can be effective and scalable for designing efficient and stable silicon-based tandem photocathodes for solar water splitting. In addition, we explore design aspects for these micropillar-based Si/Cu2O PEC tandem devices. Specifically, we investigate the influence of geometrical variations to a micropillar array PEC device design on light absorption and onset potential (Voc). An ITO-Au (85 nm) transparent conductive film is applied for coupling the top photoabsorber on the engineered light-trapping micropillars. In addition, we fabricated a Cu2O/CuO heterojunction on Si planar and micropillar arrays to investigate the effect of the Cu2O/CuO composite in PV and PEC performances. A micropillar Si/ITO-Au/Cu2O/ZnO/TiO2/Pt stack is compared to a planar device. Furthermore, we evaluate the roles of the n-ZnO (25 nm) and TiO2 (100 nm) thin films as hole-selective and passivation layers.

Results and Discussion

Micropillar Array Fabrication and Interlayer Deposition

The substrates with vertically ordered micropillar arrays with radial pn-junctions (p-type base and n+-type emitter) were fabricated using deep-reactive ion etching (DRIE) (Figure S1). The light absorption ability in the micropillar structure is considerably improved not only by the increased surface area of the photoelectrode but also by the manifold light scattering within the micropillar structure (Scheme ),[31,76] which is a tandem device that contains two photoabsorbers. The top cell presents a large band gap material that absorbs high energy photons, while the bottom cell has a small band gap material that harvests the low energy photons. Many device and material aspects need to be tuned carefully to achieve high efficiency, such as the bandgaps and material thicknesses, but also reflection and parasitic absorption. Furthermore, the transparent conductive oxide (TCO) layer on the substrate can achieve an effective charge separation for a relatively thick film. In general, TCO-semiconductor Schottky contacts can lead to high recombination rates and therefore usually produce low Voc values. However, when a PV junction is decorated with a TCO layer, high Voc values can still be obtained.[77]

Scheme 1

Schematic Illustration of the Light Absorption Mechanism and Charge (Electron–Hole) Separation of Cu2O Film on Conductive (a) Flat and (b) Micropillar Array Substrates

Upon following the fabrication and DRIE process of the micropillars (Figure S1), a few samples were fractured and side views were taken by scanning electron microscopy (SEM) (Figure ). The base and tip diameters of these micropillars within the arrays are approximately identical, emphasizing the highly anisotropic nature of the DRIE process. A slight tapering occurred at about ∼20% from the top of the pillars. This effect was more pronounced for the 40 μm pillars (12 min etching), where the diameter decreased approximately 100–200 nm. In addition, the SEM images (Figure ) show the entire side surface of the micropillars to be scalloped due to the cyclic steps in the etching process, involving gas pulses of SF6 (etching step) and C4F8 (passivation step), providing scallops with an overall height of ∼340 nm, for each etching cycle.

Figure 1

SEM images of bare Si micropillars after the DRIE process, with different heights and pitches: (a) 10 μm, 8 μm; (b) 20 μm, 8 μm; (c) 30 μm, 8 μm; and (d) 40 μm, 10 μm, respectively.

The first step to an efficient Si–Cu2O tandem device is to find the ideal interlayer for the Cu2O deposition on top of the Si micropillar array. Such a layer needs to be a transparent so that sunlight can pass through it and can reach the underlying Si photoabsorber. In addition, the interlayer must be conductive to provide an ohmic contact for the growth of Cu2O film and allow charge carrier exchange between the Si and Cu2O. Here, we investigated different interlayers such as gold (Au), copper (Cu), molybdenum (Mo), platinum (Pt), indium tin oxide (ITO), and ITO/Au deposited using a magnetron sputtering system.

The growth of the Cu2O film on different interlayers was characterized by scanning electron microscopy (SEM) as illustrated in Figure . By using Cu, Au, and Pt interlayers, Cu2O growth was compact and homogeneous on the Si substrate. Yet, due to the low conductivity of the Mo interlayer, Cu2O grew into large cuboctahedral crystals without covering the Si surface in a conformal manner. However, none of these metallic interlayers is suited for the tandem device due to their low optical transparency. Thus, a thin layer of the conducting ITO (80 nm) with a thin layer of Au (nominally 5 nm) was deposited to enhance the conductivity without compromising in transparency. The short sputtering time, which leads to small Au nanoparticles instead of a dense homogeneous film, plays an important role in the electrodeposition of Cu2O, leading to specifically oriented crystal growth on the substrate.[78] Furthermore, the combined ITO-Au nanoparticle film (i.e., the ITO/Au layer) is sufficiently transparent and conductive to deposit Cu2O effectively.[79,80]

Figure 2

Top-view SEM images of the electrodeposited Cu2O on a thin interlayer of (a) Cu, (b) Pt, (c) Au, and (d) Mo.

Optimization of Cu2O Film Thickness

The second step in the overall fabrication process is the deposition of Cu2O on the conductive Si/ITO-Au planar and micropillar array substrates. Conformal, compact, and semitransparent Cu2O films were prepared by chronoamperometric (CA) electrodeposition. To obtain a high-performance Cu2O film, we deposited the material with different thicknesses (250–1300 nm) by changing the electrodeposition time (200–1300 s) at a constant cathodic potential (−0.8 V), while the temperature of the bath was maintained at 50 °C. It is important to notice that we did not elongate the deposition process for micropillar arrays with different lengths (10–40 μm) and for different pitches (8–14 μm), see Figure , because the highest growth rate is reached at the applied potential and temperature of the bath solution.[81−83]

Figure 3

SEM images of pn-Si/ITO-Au/Cu2O micropillar arrays with different heights and pitches: (a) 10 μm, 12 μm; (b) 20 μm, 14 μm; (c) 30 μm, 10 μm; (d) 40 μm, 12 μm; (e) 20 μm, 8 μm; (f) 20 μm, 10 μm; (g) 20 μm, 12 μm; and (h) 20 μm, 14 μm, respectively.

Because the ITO-Au interlayer is conducting, and the system is mass transport-limited during the electrodeposition, a conformal layer of Cu2O is deposited on the micropillars. It was observed that Cu2O forms columnar crystals in a uniform manner on Si/ITO-Au planar substrates (Figure S2) and at the bottom side of the Si/ITO-Au micropillar array substrates (Figure S3) as well as in the sidewall scalloped area. This uniformity occurred despite the fact that the ITO/Au interlayer was thinner on the scalloped sides of the micropillars when compared to the bottom side of the Si substrate, which was caused by a shadowing effect by the scallops during the interlayer deposition by the directional magnetron sputtering process.

The SEM images and X-ray diffraction (XRD) characterization revealed that the prepared films have columnar triangular cubes or truncated cubic crystal growth with (111) orientation for shorter and longer deposition times, respectively (Figure ). Paracchino et al. reported that the columnar film with cubes exposing the triangular (111) faces parallel to the substrate showed a higher photocurrent.[19] The crystal structure of the deposited layers was confirmed by X-ray diffraction (XRD). Figure shows the XRD pattern of Cu2O showing (110), (111), and (200) peaks at 30°, 36.9°, and 42.7°, respectively, on p-Si/ITO/Au micropillar array. For the ITO/Au interlayer, an ITO peak was found at 31° and for the Au (200) peak was found at 38.6°. For the silicon substrate, two peaks were observed, (200) and (400) at 33.3° and 61.3°, respectively. The intensity of the ITO peak diminished with increasing Cu2O film thickness, as well as the Au peak, which was completely covered in a broad peak of the Cu2O with higher intensity. Finally, a Cu2O (220) peak was found in close proximity with the p-Si (400) peak at 61.3°, upon thicker Cu2O deposition, which occurs from the difference in the surface energy of the cubic crystal facets.

Figure 4

XRD patterns for (a) the Si Substrate, (b) Si/ITO-Au interlayer, (c) Si/ITO-Au/Cu2O (250 nm thick film), and (d) Si/ITO-Au/Cu2O (1000 nm thick film).

Preparation of Cu2O/CuO Heterojunction

After the electrodeposition, the Cu2O film was subsequently converted into a Cu2O/CuO heterojunction film by a thermal annealing process at 350 °C in air. To optimize the film thickness ratio between Cu2O and CuO, we followed two different strategies. First, electrodeposited Cu2O samples were prepared with different film thicknesses by variation of the deposition time on the Si/ITO-Au substrate, followed by annealing all samples at a constant annealing time of 1 h at 400 °C. Second, samples were prepared with the same initial Cu2O thickness, followed by annealing the samples at different annealing times (0.5–3 h) at 400 °C.

SEM images were taken to provide the thicknesses of both copper oxides on the Si/ITO-Au substrates (Figure ). The morphology of the sample changed after thermal oxidation in air, from a continuous film before thermal oxidation to a bilayer-structured film after thermal oxidation. This observation implies that the outer layer of the Cu2O film was transformed into CuO, and a Cu2O/CuO heterojunction was formed. When samples with different initial thicknesses of Cu2O were oxidized in air using the same annealing time (Figure a–c), the thickness of the newly formed CuO film on top of Cu2O layer was the same. In the second case, when the thermal oxidation process time was increased from 0.5 to 3 h (Figure d–f), the thickness of the outer CuO layer increased. By consuming the Cu2O layer underneath the CuO, the thickness of the inner Cu2O layer was reduced. This trend is clearly indicated in Figure a. Diao et al. reported that the whole Cu2O layer can be oxidized to CuO, leading to a continuous porous film.[21] These results confirm that the Cu2O to CuO thickness ratio can be controlled accurately by varying the deposition time of the initial Cu2O film thickness and the subsequent thermal oxidation time.

Figure 5

SEM images of Si/ITO-Au/Cu2O/CuO films prepared by electrodeposition of Cu2O at different deposition times of (a) 350 s, (b) 700 s, and (c) 1000 s, followed by thermal oxidation for 1 h at 400 °C, or by electrodeposition of Cu2O for 700 s, followed by thermal oxidation for (d) 0.5 h, (e) 1 h, and (f) 3 h at 400 °C.

Figure 6

(a) Copper oxide (Cu2O and CuO) film thicknesses as a function of annealing time. (b) XRD patterns of Si/ITO-Au/Cu2O/CuO photocathodes after thermal oxidation at 400 °C for different annealing times. (c) XRD patterns of Si/ITO-Au/Cu2O/CuO photocathodes at different electrodeposition times after annealing for 30 min at 400 °C.

The XRD patterns of the Cu2O/CuO heterostructure film on Si/ITO-Au substrates, in which the diffraction peaks of both cubic Cu2O and monoclinic CuO appear (Figure b,c), provide direct evidence for the formation of the Cu2O/CuO bilayer heterostructure. The emerging XRD diffraction peaks at 31.1°, 36.5°, and 39.2° could be indexed to the (110), (002), and (200) planes of CuO, respectively. In addition, from the XRD plots at increasing annealing times (Figure b), the intensity of the CuO [002] peaks increases, while the Cu2O [111] peak intensities decrease. In contrast, when the initial Cu2O film thickness increased by the subsequent oxidation time is kept constant (Figure c), the spectrum hardly changes.

Photovoltaic (JV) Measurements on Si Micropillar Arrays with a Radial pn-Junction

Micropillar array dimensions (height and pitch) have an influence on important functional parameters such as surface area, light absorption, and charge separation, which finally regulate the device efficiency. In addition, the relationships between the fabrication process and the characteristics of device performance are tough to predict. For instance, it is known that with increasing micropillar height, the total junction area in the solar cell increases, and the reflectivity of sunlight probably decreases.[35,76] The effect of micropillar dimensions (i.e., height and pitch) with integrating radial pn-junction was here investigated by examining the resulting photoelectrical performance.

Figure a–d shows the JV characteristics of samples with different pillar heights and pitch from 10 to 40 μm and 8 to 14 μm, respectively. Samples with pillars performed better in both current density (JSC) and open circuit voltage (VOC) as compared to a flat sample, which indicates that the pillar structures significantly improve the properties such as local surface area and light absorption properties in solar cells. The values of the fill factor (FF) and efficiency (η), along with their respective JSC and VOC, are shown in Table S1. A minor increase of VOC is observed, from 450 mV for the flat sample to ≥550 mV for micropillars.

Figure 7

JV measurements of (a–d) pn-Si micropillar array substrates without Cu2O, for varying pillar heights and pitches as indicated, and (e) comparison between planar and micropillar arrays with varying pillar heights and at a 12 μm pitch.

The FF determines the maximum power point of a solar cell and can be calculated from JSC and VOC with eq . Using the fill factor (FF), short-circuit current density (JSC), and open-circuit voltage (VOC), the efficiency can be calculated using eq .where Jmp is the current density, Vmp is the voltage at the maximum power point, and Pin is the input power, which is 100 mW/cm2 (AM 1.5).

When viewing the dependence on pillar height (Figure e), JSC values increase from the flat substrate up to 20 μm height, but drop again for ≥30 μm pillar heights. The latter decrease is due to charge carrier recombination by an increasing number of defect states with increasing micropillar height.[76,84] Most likely, these defects have been introduced by the DRIE method used to make the pillars. The working principle of the applied Bosch process, which is an alternating process of etching (SF6) and passivation (C4F8) steps, has led to the scallop structures,[85] which further increases the number of defect states. We conclude that for these pillar arrays, optimal performance is achieved with a 4 μm diameter, 12 μm pitch, and pillar height of 20 μm.

PV Performance of Copper Oxide-Covered Micropillar Arrays

In a tandem device, the overall photovoltage is generated by the absorption of two photons instead of one. Each of these photons creates a pair of charge carriers (e– and h+) in the total photoabsorber stack. First, it is important to find the proper thickness of the top photoabsorber so that part of the sunlight can pass through it and illuminate the bottom photoabsorber to get the total photovoltage. Second, we fabricated Si micropillar substrates with varying heights, pitches, and copper oxide thicknesses to find the maximum current density output resulting from these aspects. A Cu2O film thickness of ∼650 nm appears most favorable with respect to the resulting short circuit current density (Jsc), measured in an aqueous PEC setup (Figure S6b), as shown in Figure a. It is well-known that a thicker layer limits the charge separation efficiency and a thinner layer does not absorb all incoming light. Si micropillars covered with optimized Cu2O film thickness of ∼650 nm were fabricated with varying pillar heights, from a planar substrate to 40 μm height, and the highest Jsc was produced at 20 μm height (Figure b). Finally, the micropillar pitches were varied between 8 and 14 μm, and 12 μm pitch gave the highest Jsc; see Figure c.

Figure 8

Plots of the current density as a function of different parameters of the Si/ITO-Au/Cu(O PEC tandem samples with (a–c) Cu2O film and (d–f) Cu2O/CuO heterostructure film as top photoabsorbers: (a,d) varying the Cu2O film and Cu2O/CuO film thickness on planar substrates, (b,e) varying the micropillar height, for micropillar samples covered with a Cu2O film (650 nm) or Cu2O (630 nm)/CuO (120 nm) heterostructure film (with 5:1 thickness ratio between Cu2O/CuO) at a pitch of 12 μm, and (c,f) varying the micropillar pitch, for micropillar samples covered with a Cu2O film (650 nm) or Cu2O (630 nm)/CuO (120 nm) heterostructure film (with 5:1 thickness ratio between Cu2O/CuO) at a pillar height of 20 μm.

In case of Si/Cu2O/CuO heterostructure fabrication, we deposited 650 nm thick Cu2O on a planar substrate followed by thermal oxidation process at 400 °C while varying the annealing time as shown in Figure a. As a result, Cu2O/CuO heterojunction layer with a different thickness ratio between Cu2O and CuO film was formed on planar Si substrates. The highest Jsc output was observed for the Cu2O (630 nm)/CuO (120 nm) heterostructure film with thickness ratio of 5:1 on planar Si substrate, respectively (Figure d). This is considered as an optimum Cu2O/CuO heterostructure film for flat samples, and this was applied on Si micropillar substrates with varying micropillars height and pitch. When the optimized layer thickness of the Cu2O/CuO heterostructure was placed on Si micropillars with varying heights and pitches, surprisingly no clear trends in the output current density were observed with respect to these parameters (Figure e and f). The observed behavior could possibly be due to parasitic resistances and optical losses, where the former occurs presumably between the copper oxides and between Cu2O and silicon. In a comparison between the Cu2O and Cu2O/CuO heterojunction systems, the former all perform better at the optimized pillar parameters (Figure ), while only at the low-performance sides of these parameters are comparable performances observed. Therefore, we conclude that Si micropillar arrays with a pitch of 12 μm, a height of 20 μm, and a (nonoxidized) Cu2O layer thickness of 650 nm are optimal, and we fixed these for further study.

The JV curves of planar and micropillar Si substrates were measured without and with different overlayers on top (Figure ; JV-measurement setup shown in Figure S6a), by measuring dry and contacting only the Si cell on top and bottom. After the ITO layer (∼80 nm) was deposited, the current density increased for both the planar and the micropillar Si solar cells due to its antireflection property. After a Cu2O film of ∼650 nm thickness was electrodeposited over the Si/ITO-Au substrates, both the VOC as well as the JSC values dropped significantly, due to the parasitic light absorption by Cu2O and the consequently reduced light intensity on the Si. For a planar solar cell, the JSC decreased by 7 mA/cm2 due to the Cu2O layer, while for a Si micropillar solar cell the JSC decreased with 22 mA/cm2, making the output practically comparable. This indicates that the Cu2O layer absorbs more sunlight when fabricated over the micropillar array.

Figure 9

JV characteristics of (a) planar and (b) micropillar Si substrates covered with different overlayers; a bare substrate, with ITO, with Cu2O and Cu2O/CuO films over the ITO-Au layer.

Interestingly, when a Cu2O (630 nm)/CuO (120 nm) heterojunction layer was deposited on the planar and micropillar substrates, the drop of the VOC and JSC values was nearly similar as compared to the Cu2O overlayer, but the fill factor (FF) decreased considerably (Figure ). The low slope close to VOC in the JV curves shows that the device behavior is dominated by a series resistance, which is attributed to a poor conductance between the top (Cu2O/CuO) and bottom (Si) photoabsorbers. We found voids formed between the Si surface and the Cu2O/CuO layer after thermal oxidation of the preceding Si/Cu2O device. Interestingly, this void formation was more extensive in case of the micropillar arrays (Figure S4b), potentially because of the larger surface area and/or the presence of different curvatures in the micropillar samples. Apparently these voids explain the increased series resistance observed in Figure . Possibly, the thermal annealing process, which was carried out to prepare the Cu2O/CuO heterojunction, led to cracks at the Cu2O–Si interface (see Figure S4).

Pulsed Laser Deposition of Protection Layers

For a high-performance Cu2O-based PEC electrode, the quality of the buried p–n junction, the protection layer, and the cocatalyst are the most important parameters. The p–n junction must be connecting and uniform, the protection layer must be conformal and pinhole-free, and the catalyst nanoparticle islands must be robust and adhered strongly on the electrode surface. A protection layer is needed to prevent direct contact of the Cu2O film with the electrolyte to avoid photocorrosion, and thus to maximize the performance of the device. Here, we explored the use of pulsed laser deposition (PLD) as a method to apply a conformal and high-quality protection overlayer on micropillar array structures.[86]

Previously, Luo et al. reported a high photocurrent density by fabricating a Cu2O nanowire-structured photoelectrode.[87] Yet, the challenging part is to increase the photovoltage for unassisted water-splitting devices. The photovoltage is dependent on the difference between the quasi-Fermi level of the electrons in the n-type oxide layer and the holes in Cu2O under illumination in a heterojunction device. Therefore, choosing a suitable n-type oxide layer is important for producing a high photovoltage.[43] We explored the use of zinc oxide (n-ZnO), an n-type material, combined with titanium oxide (TiO2) as a protective layer to improve the stability and performance of a Cu2O-based photocathode.

HR-SEM images of the electrodes after protection layer deposition of 20 nm of ZnO and 100 nm of TiO2 are shown in Figure a,b on both planar and micropillar samples. These images clearly show the homogeneous coating of the complete stack of layers on both substrates. The difference in contrast of each layer in images shows the quality and conformity of deposition achieved by the PLD technique. The band energy-level alignment diagrams of p-Cu2O with n-ZnO and n-TiO2 are shown in Figure S5. A conduction band offset between the Cu2O and ZnO layer is about ∼1 eV, which improves the separation of the quasi-Fermi levels in the two oxides under illumination to 0.45 eV in agreement with the 0.45 V positive shift of the onset potential, and hence the buildup of an extra photovoltage for hydrogen generation.[88−90]

Figure 10

HR-SEM cross-section images of PLD-coated ZnO (20 nm) and TiO2 (100 nm) on top of the Si/ITO-Au/Cu2O (a) planar and (b) micropillar devices. False-colored on each figures indicates conformal layers (ITO/Au layer, yellow; Cu2O film, light red; n-ZnO, blue; and TiO2, green).

Photoelectrochemical Performance of the Photocathodes

To assess the PEC performance of the devices, JE measurements of Cu2O/ZnO/TiO2/Pt on the Si/ITO-Au planar and micropillar array photocathodes were performed with the same Cu2O film thickness of ∼650 nm (see Figure ). As a proof of concept, Pt was used as the HER catalyst, and it was deposited electrochemically under dark condition using an aqueous solution of 1 mM H2PtCl6 for 15 min (for planar) and 19 min (for micropillars) with a thickness of nominally 3 nm using a procedure reported previously.[18] Pt is used as a catalyst because (i) it has been well studied, and (ii) it is purely as a reference case for testing the activity of the PEC cell without having to worry about the catalytic performance. PEC measurements were performed under illumination in 0.5 M Na2SO4 with 0.1 M sodium phosphate buffer solution (pH 5) as electrolyte.

Figure 11

(a) PEC performance in the dark (dashed lines) and under illumination (solid lines) from simulated 1 sun illumination for planar and micropillar Cu2O/ZnO/TiO2/Pt photocathodes. (b) Current density–time behavior of pn-Si/ITO-Au/Cu2O/ZnO/TiO2/Pt photocathodes measured at 0 V vs RHE. Here, two “dark” periods were incorporated of each 30 min. (c) IPCE measurements of both photocathode devices at 0 V vs RHE under monochromatic illumination (solid lines) and the integrated photocurrent over the solar spectrum (dotted lines). (d) Hydrogen gas produced and calculated versus time for the micropillar device.

The Cu2O film on the planar substrate only delivered a photocurrent density of 5 mA cm–2 at 0 V vs RHE, which is lower than that on the micropillar substrate (7.5 mA cm–2) (see Figure a). In addition, as compared to the planar device, the electrode with the micropillar array showed a slightly higher anodic shift of the onset potential of +0.85 V versus RHE. It suggests that a large proportion of photogenerated carriers effectively separate in the micropillar arrays. A stability test for planar and micropillar devices was performed for 75 h (Figure b). The devices showed an excellent stability without degradation over time, indicating the PLD method is effective for conformal and pit-free layer deposition of ZnO and TiO2.

To better understand the PEC performance of these photocathodes, we measured the incident-photon-to-current efficiency (IPCE) of both devices and used it to calculate the integrated photocurrent under AM 1.5 G (100 mW cm–2) solar irradiation (Figure c). The overall current densities correspond well with the values of the photocurrent plotted in Figure a. The IPCE clearly shows a broad plateau response through a wider spectral range for the micropillar array compared to the planar device. The production of H2 gas was confirmed by the formation of bubbles evolving, and measurements by gas chromatography for a micropillar photocathode biased at 0 V vs RHE with a near-ideal faradaic efficiency (Figure d; small deviations attributed to measurement inaccuracies, gas dissolution, and possible parasitic processes).

Conclusions

In summary, we have developed a highly photostable tandem Si/Cu2O micropillar photocathode with passivating ZnO and TiO2 overlayers directly on the micropillar array devices by using the pulsed laser deposition method. These conformal protection films grown with precise thickness formed on both planar and microstructure devices. The Cu2O photocathode with a thickness of ∼650 nm combined with Si micropillar arrays (20 μm height and 12 μm pitch) delivered an optimal photocurrent of 7.5 mA/cm2 at 0 V vs RHE, a photovoltage of 0.85 V, and stability beyond 75 h using an n-ZnO hole-selective layer, a TiO2 protection layer, and a Pt HER catalyst. In case of Cu2O/CuO heterostructure, an improved performance was gained for planar samples but not for micropillared samples. The high onset potential achieved in Si/Cu2O micropillar sample is attributed to the radial buried pn-junction PV cell as the bottom photoabsorber. The ZnO layer forms a promising heterojunction with Cu2O for a good separation of the photogenerated charge carriers in the film. As a result, high photostability is achieved by the protective layers that prevent photocorrosion of the PV–PEC tandem device. Our study presents a successful case for the use of PLD to achieve conformal coating of protective layers on high-aspect ratio micropillar arrays, which offers a new path to stable PEC water splitting devices as well as to high-efficiency solar cells.