Low Temperature Heating of Silver‐Mediated Exfoliation of MoS2

The need for high‐quality large‐scale monolayers of layered materials pushes the development of scalable gold‐mediated exfoliations. Gold proves to be a suitable adhesive for exfoliation of several 2D materials. However, the extension to other noble metals remains underwhelming as gold outperforms all previously studied metals by a large margin. This is attributed to compromised stability against oxidation and surface contamination of less noble metals, leading to nonideal interfaces for exfoliation. The closest competitor to gold is silver, where gold still leads by a factor 100 regarding exfoliated layer size. In this work, a silver‐mediated exfoliation process performing on par with gold is presented. The combination of freshly cleaved silver surfaces with a low‐temperature annealing is found to be crucial. The exfoliation yield shows a dependence with annealing temperature, leading to loss in exfoliation performance for higher temperature. Raman studies indicate inhomogeneous strain for the MoS2/Ag interface at these temperatures, which hints at the competing factors of thermal activation versus oxidation of silver. Finally, a transfer process is implemented to promote silver to a fully functional exfoliation substrate. Ultimately, heating up exfoliations tips the strict balance between interfacial interactions and surface contaminations toward robust high monolayer yield exfoliation as demonstrated for silver.

The need for high-quality large-scale monolayers of layered materials pushes the development of scalable gold-mediated exfoliations. Gold proves to be a suitable adhesive for exfoliation of several 2D materials. However, the extension to other noble metals remains underwhelming as gold outperforms all previously studied metals by a large margin. This is attributed to compromised stability against oxidation and surface contamination of less noble metals, leading to nonideal interfaces for exfoliation. The closest competitor to gold is silver, where gold still leads by a factor 100 regarding exfoliated layer size. In this work, a silver-mediated exfoliation process performing on par with gold is presented. The combination of freshly cleaved silver surfaces with a low-temperature annealing is found to be crucial. The exfoliation yield shows a dependence with annealing temperature, leading to loss in exfoliation performance for higher temperature. Raman studies indicate inhomogeneous strain for the MoS 2 /Ag interface at these temperatures, which hints at the competing factors of thermal activation versus oxidation of silver. Finally, a transfer process is implemented to promote silver to a fully functional exfoliation substrate. Ultimately, heating up exfoliations tips the strict balance between interfacial interactions and surface contaminations toward robust high monolayer yield exfoliation as demonstrated for silver.

Introduction
To showcase the true potential of 2D materials, mechanical exfoliation has been key to obtain high quality single layers to uncover the pristine and smooth template/metal interface, revealing a fresh exfoliation substrate on demand. [3,8] Unlike most gold-mediated exfoliation processes, in our previously reported process a thermal annealing step was key to activate the exfoliation. [8] This unique characteristic motivated the extension to silver following the discussion of Velický et al., [12] by heating up the love affair between MoS 2 and silver to rival gold. In this work, a silver-mediated exfoliation of MoS 2 is presented. Fresh silver surfaces were obtained by a templatestrip and exhibited thermally activated exfoliation of MoS 2 with near unity yield. The process was benchmarked against the gold analogue and showed unprecedented performance on par with gold. The temperature dependence of the exfoliation yield was studied and revealed a peak performance ≈150 °C, whereas higher temperature led to a loss in performance, a feature absent for gold. This loss suggests a competing interaction to the thermal activation, likely the oxidation of silver. In this light, Raman studies suggested a more inhomogeneous strain at the MoS 2 /Ag interface for higher temperature. X-ray photoelectron spectroscopy (XPS) indicates that at employed temperatures, silver oxidation is effectively suppressed when the metal surface is protected by MoS 2 . Lastly, a polymer-based transfer process using polystyrene is implemented to promote silver to a complete exfoliation substrate, with the ability to relocate exfoliated single layers onto target substrates.

Results and Discussion
The thermally activated exfoliation with template-strip silver substrates is shown in Figure 1a.
The template-stripped silver substrates are prepared by evaporating a 200 nm silver layer onto a polished silicon wafer. Glass slides for solid support are glued onto the silver-covered wafer surface using UV-curable epoxy. Fresh and smooth silver surfaces are now available on demand by releasing the silver off the silicon wafer, by using a razor blade. The silver as well as the MoS 2 parent crystal are stripped shortly before exfoliation to maintain both surfaces to be as fresh as possible. Pressing the crystal onto the silver surface followed by some heating leads to exfoliation MoS 2 on silver. At room temperature no discernible exfoliation occurs, as already observed for gold. [8] The annealing at 150 °C for 1 min is sufficient to activate exfoliation. Before peeling off the crystal to reveal the exfoliated single layers, a brief precooling step is performed. [16] The resulting large-scale MoS 2 single layer is directly visible in the optical microscopy image reported in Figure 1b (and zoomed area in Figure 1c). Some multilayers are visible, yet the single layer (1L) areas easily outweigh them with 23 mm 2 total (≈78% single layer yield). A closer look in Figure 1c reveals a continuous single layer without damage, with a small bilayer (2L) nearby. Atomic force microscopy (AFM) is used to validate the layered nature at the 1L-2L transition and is reported in Figure 1d. The measured step height of ≈0.65 nm is in good agreement with the expected MoS 2 interlayer distance. [17] The 1L-MoS 2 /Ag step remained elusive due to the abrupt roughness change ( Figure S1, Supporting Information). Overall, the large-scale exfoliation shows results unprecedented for silver, qualifying it as a sustainable and low-cost alternative to gold. It is important to note that this exfoliation works in air despite the previously suggested oxidation of silver being a limiting factor. [12] The process seems to hit the sweet spot between boosted interfacial interactions and oxidation of silver enabled by the annealing step. Furthermore, the brief duration and low temperature of the annealing step should avoid any material degradation due to defect generations, which are known to occur at temperatures above 200 °C for prolonged time. [18,19] The large-scale exfoliation was reproduceable as well as the near unity single layer yield ( Figure S2, Supporting Information). This motivated the extension of this process to copper, a metal with even worse oxidation resistance.

www.advmatinterfaces.de
However, for copper no exfoliation was observed ( Figure S3, Supporting Information).
The silver-mediated exfoliation was benchmarked against gold at several temperatures, as presented in the statistics in Figure 2. The single layer yield (Figure 2a) for silver and gold exhibits a temperature dependence with no appreciable continuous single layers exfoliated below 150 °C ( Figure S4, Supporting Information). Silver exhibited peak performance at 150 °C and loss in yield for increasing temperature up to 200 °C. For gold, the decrease in performance with increasing temperature is less pronounced. This is consistent with the higher oxidation resistance for gold, [12] indicating that the loss for silver could be oxidation related. This behaviour is also reflected in the maximum achieved 1L-area (Figure 2b, with multilayer areas in Figure S5, Supporting Information), which again peak at 150 °C for silver, while for gold no significant change is observed between 150 and 200 °C. Furthermore, the statistics clearly show that silver is capable to perform on the same level as gold with the added heat in this procedure. When compared to previous reports, where silver exfoliation size was 1/100 that of gold, [12] it becomes evident that the presented procedure dramatically increases the exfoliation performance of silver. The otherwise strict process conditions are relaxed upon annealing, potentially due to surface reconstructions of silver at the MoS 2 /Ag interface. The restructuring of silver or gold at the MoS 2 interface due to annealing can be rationalized by the hopping barrier (E hop ) experienced by metal adatoms on MoS 2 , which was previously reported for gold (39 meV) and silver (49 meV). [20] Increasing thermal energy from room temperature (26 meV) to 150 or 200 °C increases thermal energy to ≈36 and 41 meV, respectively, which enables both metals to restructure accordingly at the MoS 2 interface during exfoliation. This allows to bypass otherwise strict cleanliness requirements by reforming parts of the metal/MoS 2 interface, as indicated in Figure 1d exhibiting regions with intimate contact of silver with MoS 2 and holes in-between.
To elucidate the role of interfacial strain, Raman studies were conducted (Figure 3a). The peak position shift of the E 2 1g mode relative to SiO 2 indicates tensile straining of MoS 2 on gold and silver. [12,21] The shape of this mode hints at a strain distribution, where for silver the more asymmetric and broadened E 2 1g mode suggests a wider strain distribution compared to gold, consistent with previous work. [12] The shift of the E 2 1g mode was used to estimate the strain, [21] using E 2 1g peak positions obtained from Voigt fits (see Figure 3a). For MoS 2 exfoliated on gold, a strain of ≈0.7% is observed, where a single component almost suffices to fit the E 2 1g mode. For MoS 2 on silver exfoliated at 150 °C, two components are necessary, and the high-strain feature ≈ 2% strain seems even more pronounced at 200 °C. This indicates an increase in the strain inhomogeneity in MoS 2 ( Figure S6, Supporting Information), consistent with increased oxidation of silver. Another factor for increased strain  . Raman and photoelectron spectroscopy study of the MoS 2metal interface. a) Raman spectroscopy for strain analysis of single layer MoS 2 on gold, silver, and SiO 2 . The strain is derived from the E 1 2g peak position relative to the position on SiO 2 . [12,21] Strain values of the E 1 2g modes on different metals are indicated. Voigt curve fits are used to determine peak positions and are indicated as grey lines. For silver, two peaks are visible, which is even more apparent for 200 °C. b) X-ray photoelectron spectroscopy survey of the MoS 2 /Ag interface, annealed silver surface and pristine template-stripped silver (Ag RT). The increase of the O 1s peak indicates oxidation for annealed silver in air, which is suppressed at the MoS 2 /Ag interface.

www.advmatinterfaces.de
inhomogeneity could be related to the distinct growth morphologies of gold and silver on MoS 2 . Gold has been reported to cluster into ordered nanostructures, where silver showed more random clusters. [20,22] This would explain the presence of a high strain component on Au (Figure 3a) due to cluster formation. Consequently, larger strain broadening is expected for the more random clustering at the MoS 2 /Ag interface.
The finding that gold strains MoS 2 more homogenously is consistent with previous reports, where it was attributed to the oxide-free gold surface. [12] This might explain the loss in exfoliation performance for higher temperature due to oxidation of silver. XPS was used to study the MoS 2 /Ag interface in more detail and the survey scans are shown in Figure 3b. Freshly evaporated silver does not display a notable peak corresponding to the presence of oxygen, i.e., silver oxidation. For silver annealed at 200 °C in air, pronounced oxidation is indicated by the increased O 1s peak. This is not the case for the sample with MoS 2 on silver. This indicates that at 200 °C the oxidation of silver is strongly suppressed once the metallic surface is passivated by the MoS 2 .
With large scale single layers on silver, the next step is to transfer these onto arbitrary substrates. Hence, a polystyrenebased transfer was adopted [23] and implemented (Figure 4a). All features remained unchanged, as seen in Figure 4b,c, indicating a deterministic transfer process. AFM revealed a 1L-MoS 2 /SiO 2 step height of ≈0.9 nm, consistent with previous work. [8,24] The photoluminescence spectrum (Figure 4d) highlights the characteristic direct bandgap emission for an intact single layer. [25] Notably, the previously strained MoS 2 on silver shows a narrow and symmetric E 2 1g mode on SiO 2 after transfer (Figure 4e), indicating effective strain release. [26] With the transfer implemented, silver serves as a fully functioning exfoliation substrate.

Conclusion
In conclusion, heating up the silver-mediated exfoliation dramatically increases the exfoliation performance by tipping the strict balance of process conditions, i.e., interfacial interactions and surface contaminations, toward a high single layer yield exfoliation process. The combination of fresh templatestripped silver and the low temperature annealing enables unprecedented performance on par with gold. A temperature dependence of the exfoliation yield is observed, suggesting a thermally activated process limited by oxidation for higher temperatures. Thereby, this work highlights the importance of temperature as a key parameter in metal-mediated exfoliation to enhance process robustness and performance.
Metal-Assisted Exfoliation: MoS 2 (2D semiconductors, synthetic MoS 2 crystal) was cleaved with heat-resistant Kapton tape and pressed onto the freshly template-stripped metal substrate. Then the annealing step followed on a hotplate in ambient conditions at the temperatures described in the main text for 60 s, followed by a cool down by removing the substrate from the hotplate and waiting for ≈15 s before peeling the tape.
Polystyrene-Assisted Transfer: MoS 2 on Ag was transferred with polystyrene adapting a reported process. [23] Polystyrene was spin coated (Sigma-Aldrich, average M w ≈ 280.000, 90 mg mL −1 in toluol, 3000 rpm 60 s) onto the MoS 2 /Ag substrate and annealed at 80 °C for 10 min. The sample was left on the KI/I 2 (Sigma-Aldrich) metal-etchant until the polystyrene foil floated on top (≈12 h). The foil was fished out with a clean wafer piece and transferred into a beaker with deionized water to clean off etchant residues. The foil was transferred into a beaker with  [24] e) Photoluminescence spectrum of transferred 1L-MoS 2 . [25] f) Raman spectroscopy of the transferred single layer. The A 1g -E 1 2g distance equals to 18.3 cm -1 , as expected for a single layer. [27,28] The E 1 2g peak is well defined, compared to the strain-broadened peak on silver. Raman and Photoluminescence Measurements: Raman and PL spectroscopy were performed using a confocal microscope setup (XploRA, Horiba Ltd.) with 532 nm laser excitation source and 100× objective (≈1 µm laser spot size) using a 2400 L mm −1 grating for Raman and 600 L mm −1 for PL. The measurements were conducted in ambient conditions. X-Ray Photoelectron Spectroscopy: XPS spectra were acquired on the JEOL JPS-9030 photoelectron spectrometer system. As excitation source, monochromatic Al Kα (1486 eV) was employed. The sample was grounded during the spectra acquisition.
Statistical Analysis: The data of single layer yield and maximum single area ( Figure 2) were used without further pre-processing. In this data presentation the mean value was represented by the bar height and the standard deviation (±SD) is given as error bar. The sample size n was 4 for all temperature/metal pairs but 150 °C/Ag, where n = 8. To measure the exfoliated areas, optical micrograph images were analyzed in the Gwyddion software. For silver the blue channel of the RGB image was used for the best contrast while for gold the green channel was the best. Monolayer areas were detected and measured by threshold detection as implemented in Gwyddion. For the statistical analysis, the Igor Pro software was used.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.