Defect Mitigation of Solution-Processed 2D WSe2 Nanoflakes for

Dec 15, 2017 - (27-29) While promising, these strategies have not been shown suitable for improving the performance of LPE TMD-based solar energy conv...
4 downloads 9 Views 5MB Size
Letter Cite This: Nano Lett. XXXX, XXX, XXX−XXX

pubs.acs.org/NanoLett

Defect Mitigation of Solution-Processed 2D WSe2 Nanoflakes for Solar-to-Hydrogen Conversion Xiaoyun Yu, Néstor Guijarro, Melissa Johnson, and Kevin Sivula* Laboratory for Molecular Engineering of Optoelectronic Nanomaterials, École Polytechnique Fédérale de Lausanne (EPFL), Station 6, 1015 Lausanne, Switzerland S Supporting Information *

ABSTRACT: Few-atomic-layer nanoflakes of liquid-phase exfoliated semiconducting transition metal dichalcogenides (TMDs) hold promise for large-area, high-performance, low-cost solar energy conversion, but their performance is limited by recombination at defect sites. Herein, we examine the role of defects on the performance of WSe2 thin film photocathodes for solar H2 production by applying two separate treatments, a pre-exfoliation annealing and a post-deposition surfactant attachment, designed to target intraflake and edge defects, respectively. Analysis by TEM, XRD, XPS, photoluminescence, and impedance spectroscopy are used to characterize the effects of the treatments and photoelectrochemical (PEC) measurements using an optimized Pt−Cu cocatalyst (found to offer improved robustness compared to Pt) are used to quantify the performance of photocathodes (ca. 11 nm thick) consisting of 100−1000 nm nanoflakes. Surfactant treatment results in an increased photocurrent attributed to edge site passivation. The pre-annealing treatment alone, while clearly altering the crystallinity of pre-exfoliated powders, does not significantly affect the photocurrent. However, applying both defect treatments affords a considerable improvement that represents a new benchmark for the performance of solutionprocessed WSe2: solar photocurrents for H2 evolution up to 4.0 mA cm−2 and internal quantum efficiency over 60% (740 nm illumination). These results also show that charge recombination at flake edges dominates performance in bare TMD nanoflakes, but when the edge defects are passivated, internal defects become important and can be reduced by pre-annealing. KEYWORDS: Transition metal dichalcogenides, liquid-phase exfoliation, photoelectrochemical hydrogen production, photocathode, defect passivation

T

of LPE TMDs and the resulting large and highly variable concentration of defects,17 which act as recombination sites for photogenerated charges. Defects in 2D TMDs occur obviously due to dangling bonds on the edges of exfoliated flakes, but they can also occur due to non- stoichiometry in the “bulk” of the flake. In general, defects have long been known to be an issue in TMDs,18,19 and approaches have been developed to address defects on single crystalline or polycrystalline bulk materials.16,19−21 Recent experimental and computational results focusing on 2D TMDs have indicated that chalcogenide vacancies, which create deep trap states, are the most abundant defects given their relatively low formation energy.22−24 To address chalcogen vacancies, atomic healing approaches have been reported to restore exposed vacancies25,26 or they can be passivated by chemical functionalization.27−29 While promising, these strategies have not been shown suitable for improving the performance of LPE TMD-based solar energy conversion

he unique layered crystal structure of the semiconducting transition metal dichalcogenides (TMDs, e.g. MoS2 and WSe2), which enables exfoliation into 2D mono- or few-layered nanometer-to-micron-sized flakes with charge-neutral basal plane surfaces and tunable optoelectronic properties, has recently attracted intense attention for application in novel optoelectronic devices.1,2 Of particular interest are liquid-phase exfoliation (LPE) techniques,3−5 which afford solvent dispersions of semiconducting 2D TMD nanoflakes and give promise to the inexpensive roll-to-roll fabrication of highperformance solution-processed devices such as transistors,6 sensors,7 batteries,8 and solar cells.9 Indeed, the favorable band gap energy and intrinsic robustness of LPE TMDs make them ideal candidates for large-area inexpensive photovoltaics and other promising solar applications such as the direct conversion of light energy to hydrogen fuel via photoelectrochemical water spitting.10,11 While LPE TMDs have been successfully applied in high-performance solution-processed solar devices as interfacial hole- or electron-transport layers,12,13 to date the performance of LPE TMDs as a primary light harvesting material has been far below the expectations set by single crystal TMDs.14−16 This is due to the relatively small flake size © XXXX American Chemical Society

Received: September 14, 2017 Revised: November 29, 2017

A

DOI: 10.1021/acs.nanolett.7b03948 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

conditions was employed as a starting material for our study. This as-received (AR) powder was either directly exfoliated via ultrasonication in N-methyl-2-pyrrolidone (NMP) to give the “Ex-AR” WSe2 sample (Figure 1a), or pre-annealed (PA) at 1100 °C in the presence of Se vapor before solvent exfoliation to afford a sample of “Ex-PA” WSe2 (Figure 1b). Because the solvent exfoliation conditions used are reported to not significantly perturb the TMD crystal structure, 17 we hypothesize that not only will the concentration of internal Se vacancies be reduced in the Ex-PA WSe2 sample but also that surface and edge vacancies will be present due to the exfoliation process (as schematically indicated in Figure 1). To address edge defects, we sought a surface passivation approach that could be performed via simple solution-based processing on the as-deposited WSe2 photoelectrodes and that would tolerate the substrate (F:SnO2-coated glass) and harsh photoelectrochemical operation conditions (in 1 M H2SO4) required for solar hydrogen production. Given the known interaction between silane-based surfactants and dangling surface bonds of semiconductors30,31 including TMDs,32 we employed a hexyl-trichlorosilane (HTS) treatment on photoelectrodes prepared from both Ex-AR and Ex-PA WSe2 (to give “Ex-AR-HTS” and “Ex-PA-HTS” WSe2 photoelectrodes) by simply dipping the electrodes in a solution of HTS followed by rinsing and drying. We expect that the HTS can bind to accessible edge and surface defects to passivate them but that any internal Se vacancies would remain in the Ex-AR-HTS sample as shown in Figure 1c. In contrast, the Ex-PA-HTS, which benefits from both treatments, is expected to have a reduced concentration of both active surface and internal defects (Figure 1d). Full experimental details of the material and electrode preparation are given the Methods section. To examine the effects of the different treatments on the physical properties of the WSe2 and the resulting nanoflake electrodes, first the AR and PA bulk powders were compared. Optical microscope images in Figure 2 show that the typical particle size of the AR WSe2 powder (ca. 10 μm, Figure 2a) increases drastically after the pre-annealing treatment in which crystal flakes with lateral dimensions of over 200 μm are observed (Figure 2b). The bulk WSe2 powders were further investigated by XPS (survey spectra are shown in Figure S1), and as shown in Figure 2c, the W 4f peaks of the PA WSe2 powder are shifted to lower binding energy compared to the AR powder, consistent with an increased Se content via the repair of Se vacancies.33 The Se 3d peaks show a similar shift to lower binding energy (Figure S2), and overall, the Se-to-W atomic ratio estimated by XPS is also consistent with an increased Se content in the in the PA powder (see Table S1). While photoluminescence (PL) measurements indicate very poor emission characteristics due to the indirect nature of the band gap in the bulk powders, a difference in the PL spectra between the AR and PA powders consistent with the change in particle size is observed (see Figure S3 for complete details). Despite the obvious change in the bulk powder upon preannealing, is it difficult to attribute the differences in the XPS and PL spectra solely due to a change in internal Se defect concentration as clearly the edge density (surface area) of the bulk powders has also changed significantly. Moreover, as the bulk powders require exfoliation to be processed into thin-film photoelectrodes, a comparison of the exfoliated WSe2 nanoflakes is also needed. The exfoliation, purification, and thin-film deposition (by liquid−liquid interfacial self-assembly10) of the AR and PA bulk

devices. Moreover, for few-layer TMDs that can possess defects enclosed within the flakes (between layers), these restoration or passivation approaches cannot be effective. Indeed, the fundamental effects of these “internal” defects on device performance relative to surface defects remains unexplored. In general, passivating all defects in few-layer LPE TMD nanoflakes using techniques compatible with the solution-based processing from which they derive considerable advantage is the key to advancing TMD materials in large-area highperformance solar energy conversion. Herein, we describe two strategies to repair different types of defects in LPE WSe2 nanoflakes and demonstrate their effectiveness in photocathodes for solar hydrogen production. Specifically, we passivate flake edges by selectively attaching a functionalized surfactant, while internal defects are addressed via a preannealing approach. Applying these techniques leads to further insights into the dominate recombination processes and the limitations of multiflake WSe2 photoelectrodes. Furthermore, employing both techniques together with an optimized Pt−Cu co-catalyst to drive the hydrogen evolution reaction leads to a new benchmark performance for solution-processed TMDbased devices. To investigate and mitigate the effects of both internal and edge defects in LPE WSe2, we prepared thin-film photoelectrodes of WSe2 nanoflakes with different treatments as depicted in Figure 1. Commercial WSe2 powder (