Structural Features Dictate the Photoelectrochemical Activities of Two-Dimensional MoSe2 and WSe2 Nanostructures

The exfoliation of layered materials into two-dimensional (2D) semiconductors creates new structural domains, for example, basal planes, defect-rich in-planes, and edge sites. These surface species affect the photoelectrochemical (PEC) performance, which in turn determines their applicability in solar energy conversion technologies. In this study, a custom-designed microdroplet cell-based spatially resolved PEC approach was employed to identify the structural parts and to measure the PEC activity of the mechanically exfoliated MoSe2 and WSe2 nanosheets for bulk, few-layer, and monolayer specimens. The PEC performance decreased with the decreasing thickness of nanoflakes, and the relative PEC activity (photo/total current) reduced by introducing more defects to the 2D flakes: 1–3% loss was found for in-plane defects and 30–40% for edge defects. While edge sites act as charge carrier recombination centers, their electrocatalytic activity is higher than that of the basal planes. The comparison of PEC activity of micromechanically and liquid phase exfoliated bulk and few-layer MoSe2 and WSe2 flakes further confirmed that the PEC performance of 2D flakes decreases with an increasing number of edge sites.


Mechanical exfoliation
The SiO 2 /Si substrates were degreased by consecutive sonication in acetone and IPA for  minutes and cleaned using oxygen plasma (Harrick, PDC-32G). The crystals were repeatedly cleaved using scotch tape (Scotch Magic) to obtain a flat and pristine surface, which was pressed onto clean SiO 2 /Si wafers. This was baked on a hot plate (CAT Scientific, MCS78) at 100 °C for 3-5 minutes. A fresh flake surface was exposed by peeling off scotch tape right after taking off the sample from hot plate. Suitable flakes were selected using optical microscopy and the wafer was immobilized on a microscope slide.

Liquid phase exfoliation
The LPE flakes were prepared using a bath sonicator operating at 37 kHz and 100% power for 12 h while chilled to maintain a stable temperature at 25 °C with a recirculating cooler system (J. P. Selecta, Digiterm TFT). The initial concentration was 1 mg mL −1 in water:IPA 3:1 mixture 1 . Subsequently, the dispersions were centrifuged at 200 g to remove any non-exfoliated material, the supernatant was then collected by pipetting and the centrifugation steps repeated to ensure a narrow distribution of flake lateral size and thickness. This sediment phase was called as bulk dispersion. For this cascade centrifugation process, 2 we collected and processed the supernatant for all cases. We started with a higher speed (1000g) and the supernatant was collected again, while the sediment was kept as well. Afterward, we centrifuged the supernatant, and sediment dispersions again using 2000g. Subsequently, this supernatant was kept, named as few-layer dispersion. Every centrifugation step was applied for 30 min at 15 °C. The resultant MoSe 2 and WSe 2 dispersions were stable in the water:IPA solution for several months with no detectable sedimentation.

Characterization of samples
Atomic force micrograph and height profile from cross section (along the marked dashed black line in Figure S3a) of the MoSe 2 flakes and defects are presented in Figure S3a-b. In the case of few-layer MoSe 2 sample, we found a B 1 2g (at 357.5 cm −1 ) inactive out-of-plane Raman mode 3 , which helps for identification ( Figure S3c). The ratio of the intensities of A 1g band (240.5-243 cm −1 ) and Si band (520 cm −1 ) are shown in Figure 3d. The optical micrograph in Figure S4a shows examples for mono-, few-layer and bulk WSe 2 flakes. For bulk WSe 2 , two Raman signals were found at 248.5 cm −1 and 256.0 cm −1 ( Figure   S4b), predicting both the E 1 2g and the A 1g modes. 3,4 In the case of few-layer WSe 2 the E 1 2g and the normally inactive B 2g modes depicted. The position of E 1 2g mode changed with the number of layers, shifting to 251.5 cm −1 for monolayer WSe 2 flake. The AFM micrograph and height profile from cross section (along the marked dashed line in Figure S4d) of the WSe 2 flakes are presented in Figure S4c-d. S8

PEC activity study
The obtained photocurrent values were plotted as a function of sample thickness (bulk, fewlayer, monolayer) and defect density for both MoSe 2 and WSe 2 ( Figure S6a). It shows a decreasing trend upon decreasing the layer thickness, according the light absorption within layers 5 . The LSVs of mono-, few-layer, and bulk WSe 2 samples presented on Figure S6b. Figure S7 shows the optical micrographs ( Fig. S7a-d), the selected few-layer and bulk flakes region used as selected examples for the IPCE to APCE calculation. The additional AFM micrograph (Fig. S7b), and height profiles (Fig. S7c-e) are presented to demonstrate the exact layer thicknesses.
The PEC performance as a function of sample thickness and different structural domains were studied. The optical micrographs, the LSV curves and the total/photocurrents are presented ( Figure S8-11), demonstrating the PEC behaviors of in-plane defects ( Figure S8), of basalplanes vs. edge sites in the case of bulk sample ( Figure S9) and for few- (Figure S10), and monolayers ( Figure S11). S14

Analysis of electron transfer
The selected bulk flake for the IMPS analysis is shown by Figure S12, with the optical (Fig.   S12a) and AFM micrographs (Fig. S12b), and height profile (Fig. S12c). Additionally, the Figure S13 shows the reduction/oxidation of [Ru(NH 3 ) 6 ] 3+/2+ in 6M LiCl applying cyclic voltammetry at MoSe 2 surface varying the sample thickness, and the defect density.   Figure S14 shows the TEM images of MoSe 2 dispersions, as well as a statistical analyses to find the most probable size values. The log-normal distribution is characteristic of a random multiplicative process, e.g., ball milling, showing that exfoliation follows a linear fragmentation model, i.e., a process where the fragmentation is only driven by an external source, such as ultrasonic waves, therefore the statistical analyses are fitted using log-normal distributions. [6][7][8] The two samples, prepared by using different g-forces, have two different lateral size values.

Morphological and optical characterization of LPE flakes and modified electrodes
In particular, the 2000g (few-layer specimen) centrifugation processes give 35.5 ± 0.5 nm for lateral size. While, 265.4 ± 0.7 nm was found for lateral size of bulk MoSe 2 dispersion. The SEM images on Figure S15a