Correlative Spatial Mapping of Optoelectronic Properties in Large Area 2D MoS2 Phototransistors

2D materials‐based device performance is significantly affected by film non‐uniformity, especially for large area devices. Here, it investigates the dependence of large area 2D MoS2 phototransistor performance on film morphology through correlative mapping. Monolayer MoS2 films are quazi‐epitaxially synthesized on C‐plane sapphire (Al2O3 ) substrates by chemical vapor deposition, and the growth time and molybdenum trioxide MoO3 precursor volume are varied to obtain variations in film morphology. Raman, photoluminescence, transmittance, and photocurrent maps are generated and compared with each other to obtain a holistic understanding of large area 2D optoelectronic device performance. For example, it shows that the photoluminescence peak shift and intensity can be used to investigate strain and other defects across multiple film morphologies, giving insight into their effects on the photogenerated current in these devices. It also combines photocurrent and absorption maps to generate large area high‐resolution external quantum efficiency and internal quantum efficiency maps for the devices. This study demonstrates the benefit of correlative mapping in the understanding and advancement of large area 2D material‐based electronic and optoelectronic devices.


Introduction
2D materials are ideal nanomaterials for the development of the next generation of electronic and optoelectronic devices due to their ultrathin form factor and unique quantum properties.The family of 2D semiconductors includes the transition metal dichalcogenides (TMDCs) such as tungsten diselenide (WSe 2 ), tungsten disulfide (WS 2 ), and molybdenum disulfide (MoS 2 ) that possess desirable properties such as excellent mechanical strength, stability, and radiation hardness, [1][2][3][4] high optical DOI: 10.1002/admi.2023004557][8] 2D MoS 2 is one of the 2D TMDCs whose properties have been most widely explored.It has been an attractive material for optoelectronic and photovoltaic applications since the discovery of its indirect-to-direct bandgap transition from bulk to monolayer, taking advantage of its high specific power energy harvesting and efficient photoresponsivity attributes. [9]2D MoS 2 also has excellent chemical and mechanical stability, and the electrical property degradation of synthesized MoS 2 in ambient conditions can be prevented with encapsulation. [10]Furthermore, MoS 2 has carrier mobility reaching the tens to a few hundreds of cm 2 V −1 s −1 . [11]With a direct bandgap of ≈1.88 eV, monolayer MoS 2 can be used in optoelectronics and photovoltaics applications where semiconducting properties, high specific power, and flexibility are desired.While mechanical exfoliation of MoS 2 samples from bulk MoS 2 results in high-quality few-layer crystals, the approach is typically limited to sub 0.01 cm 2 size films.The scalability of 2D materials-based technologies depends on the synthesis of high-quality, sizeable (>1 cm 2 ) films.Quazi-epitaxial chemical vapor deposition (CVD) synthesis of MoS 2 films on C-plane sapphire (Al 2 O 3 ) is a scalable synthesis method optimized to produce wafer-scale monolayer MoS 2 films. [12]Generally, high crystallinity can be attained by seeding the substrate with promoters such as molybdenum trioxide (MoO 3 ) and sodium chloride (NaCl) for crystal nucleation at multiple points on the substrate, encouraging epitaxial growth with good domain alignment on substrates of suitable lattice such as gallium nitride (GaN), boron nitride (BN), graphene, and mica.However, C-plane Al 2 O 3 substrates provide step edges that can align with the armchair and zig-zag edges of MoS 2 , thereby further promoting domain alignment at the right sulfur (S)/MoO 3 ratio. [13]Films synthesized on two-inch c-plane Al 2 O 3 wafers show a uniform and wrinklefree monolayer film, with roughness as low as 100 pm. [12]Films with up to 99% unidirectional domain alignment and carrier mobilities of up to 102.6 cm 2 V −1 s −1 , [13,14] with consistent performance from 4 to ≈80 μm channel length, [14] have been shown using this growth method.With this technique, synthesized films can be transferred to desired substrates using various transfer techniques, [15][16][17][18] which is useful for applications that require varying substrates, such as large area flexible 2D phototransistors and 2D photovoltaics. [19,20]However, domain boundaries and multilayer islands are common features in CVD synthesized MoS 2 . [21,22]Such defects can lead to mid-gap states, accompanied by undesirable charging effects and nonradiative recombination pathways. [23,24]The domain boundaries and multilayer islands, along with local defects within single domains, [25,26] reduce film conductivity and carrier lifetime and contribute to inhomogeneity in film properties that significantly influence device performance.Several techniques such as optical transmission, photoluminescence (PL), photocurrent, and Raman spectroscopy are commonly used to characterize both 2D materials and devices, and these techniques become even more powerful when spatially mapped over critical areas.For TMDCs and their devices, photocurrent spectroscopy maps can be used to investigate characteristics in devices such as photothermoelectric effect at the material and contact interface and photovoltaic effect at a Schottky junction; [27,28] PL can be used to ascertain the thickness and electronic structure properties such as bandgap from the PL emission wavelength and the material quality from PL intensity; [29,30] optical transmission spectroscopy can give film thickness and layer dependent absorption information; [31] and Raman spectroscopy can show structural and layer characteristics. [32]A better understanding of the interdependence of CVD synthesized MoS 2 film morphology and device performance through large area correlative mapping of the structural, optical, and electronic properties of the films can lead to the improvement of device design and performance.Such mapping can also illuminate the relationship between device geometry, local defects, strain, multilayer stacking, and more and how each characteristic plays a role in the generation and collection of carriers in a phototransistor.In this work, we tuned the CVD synthesis parameters to obtain large area MoS 2 films of fully and partially coalesced flakes, flakes with bi-layer growth, uniform monolayer films, and monolayer films with bilayer growth.Additionally, layer transfer was used to examine multilayer films.We then used optical transmission, PL, and Raman spectroscopy to produce high resolution 20×20 μm spatial maps with 200 nm step size, which were subsequently correlated with photocurrent maps of large area MoS 2 -based phototransistors that were fabricated with CVD synthesized films of varying morphology.From the correlative maps, we show that regions in the film with a strong PL result in higher photogenerated current under 660 nm incident light.Additionally, we combine photocurrent and absorption maps to generate high-resolution external quantum efficiency (EQE) and internal quantum efficiency (IQE) spatial maps for the 2D MoS 2 -based devices, illustrating the impact of various growth morphologies on optoelectronic device performance.Lastly, we fabricate a back-gated phototransistor using the CVD-synthesized large area monolayer MoS 2 film and show the transfer characteristics under different light illumination power.

MoS 2 Flake and Film Synthesis and Morphology
The MoS 2 films in this work were synthesized by tube furnace CVD via the reaction of sulfur (S) with molybdenum oxide (MoO 3 ) precursors at 750 °C.C-plane sapphire substrates were used for quasi-epitaxial growth, where a 3-on-2 superstructure of MoS 2 (lattice constant 3.212 Å) and sapphire (lattice constant 4.814 Å) is expected. [14]Ultrahigh-purity argon was used as the carrier gas at 180 sccm flow rate.The sulfur precursor was heated to 120 °C when the furnace reached 750 °C, and the sulfur vapor was passed over the substrate for different sulfurization times to obtain different film morphologies.More details on the synthesis can be found in the Experimental Section.The films synthesized in this work were 1 cm 2 to 2.25 cm 2 .The film size limitation was due to the 1-inch diameter quartz tube used for the synthesis process.An example of a domain is shown in Figure 1g (dashed white circle), where the edges of a single triangular flake that is partially coalesced with adjacent flakes can be easily identified by the difference in optical contrast.There is a gap between the lower edge of the domain and the adjacent domain, whereas the other two edges are partially merged with the adjacent MoS 2 flakes.In the CVD synthesis process, an increased or decreased Mo vapor will typically lead to rapid or slow MoS 2 crystal formation, respectively.This can be achieved by increasing or decreasing the chamber temperature, carrier gas flow rate, or Mo precursor supply.The growth mechanism follows the Stranski-Krastanov "layer-plus-island" growth, where "islands" are formed as vertical growth on top of monolayer flakes, [34] resulting in fewlayer MoS 2 regions.High Mo supply results in small domains and early island-like growth as shown in Figure 1a-f, where 3 and 5 mg of MoO 3 were used.Interestingly, for a low Mo supply, domains are generally larger, but island-like growth can still occur even before the domains coalesce due to the increased nucleation sites on top of monolayer MoS 2 crystals or at domain boundaries as shown in Figure 1l.High performance, large area 2D materialbased devices require large domains and continuous layer uniformity over the device active area.In pursuit of this goal, the Mo supply and growth time were swept to obtain the right synthesis parameters for large flakes with unidirectional alignment over large area that subsequently coalesce into a uniform monolayer film as shown in Figure 1i.For 0.5 mg of MoO 3 , mainly flake sizes of ≈10 μm or larger were obtained for 8 min of sulfurization time with no island formation.However, smaller flake sizes in the range of 1 to 5 μm with increased island formation arise for 3 and 5 mg of MoO 3 for the same 8 min, and the island density further increased with sulfurization time.Flakes >10 μm were obtained for 0.5 and 1 mg of MoO 3 .Islands formed for both 0.5 mg of MoO 3 with 12 min sulfurization and 3 mg of MoO 3 with 8 min of sulfurization without the domains being well coalesced.However, a well coalesced, large domain MoS 2 film with no islands was obtained for 1 mg of MoO 3 and 12 min of sulfurization.This film morphology is reproducible over a large area (Figure S1, Supporting Information) and is repeatable in successive and ongoing growth attempts.Although lower MoO 3 flux results in larger flake sizes and less island formation, the right combination of MoO 3 flux and sulfurization time is needed to synthesize a uniform large area epitaxial growth of MoS 2 film on C-plane Al 2 O 3 substrates.

Correlative Mapping
To illustrate the value of correlative mapping for 2D materials and devices, 2D MoS 2 films of varying morphology were grown on c-plane sapphire, and interdigitated phototransistor contacts were fabricated on top to form a phototransistor and enable optoelectronic characterization.The transparent sapphire substrate enables measurement of optical transmission alongside other measurements.An optical image of a phototransistor with a monolayer MoS 2 film is shown in Figure 2a and the interdigitated contacts are shown from the 0.15 mm 2 phototransistor active area.Fabrication of a device up to ≈1 cm 2 active area is possible due to uniform film growth across the entire substrate.Films of similar area and uniformity were used as the active material for photovoltaics in our previous work. [20]Details on the device fabrication are discussed in the Experimental Section.
The transmission maps and photocurrent maps were produced during the same scan to account for any fluctuations in the light source and ambient conditions during the measurement, giving the correlation consistency needed to accurately calculate the EQE and IQE.The schematic of this optical table setup is shown in Figure 2b.The PL and Raman spatial maps were also performed simultaneously using an inverted Raman imaging confocal microscope.More details on the correlative map measurements are discussed in the Experimental Section.Due to scattering of photons within the substrate, and other nonlocal current spreading effects, photocurrent map resolution is broader than the 0.5 μm width incident laser spot.Because of this, incident light in locations with no MoS 2 still results in some photocurrent from carriers generated at different locations of the device; therefore, the spatial resolution is better in the absorption, PL, and Raman maps than it is for photocurrent.Additionally, PL peak wavelength was mapped as well, as this revealed domain boundaries better than the PL intensity since PL intensity is more sensitive to nonuniformity in film quality within domains.Finally, we map the spectral spacing between the Raman A 1g and E 1 2g peaks, which provides additional insight into the number of MoS 2 layers at each point in space. [35]

Flake Sample Mapping
A film grown using 3 mg MoO 3 and 8 min of sulfurization time, as shown in Figure 1d, was selected for the flake sample correlative maps.The devices fabricated on 2D MoS 2 with non-coalesced flake morphology have domains with mostly unidirectional orientation within the channel, and the contacts are aligned so that one of the flake edges in the device channel almost run parallel to the contact edge, as shown in Figure S2, (Supporting Information).A channel length of 4 μm was chosen for the flake sample, where the triangular flake sizes were roughly estimated to be within 5 to 10 μm to ensure that most of a single flake spans the entire channel length.The flakes within the channel are partially coalesced, resulting in multiple film features in the channel active area.Some of these features are shown in the optical image Figure 3a, which has three metal contact fingers of width 4 μm running horizontally through the image.Location 1 shows a region where the monolayer domains are well coalesced and locations 1, 2, and 4 show flakes that span the entire length of the channel.Smaller flakes with bilayer island growth can be seen in locations 3 and 4, highlighted by the difference in contrast, where the more reflective, lighter regions show multilayer MoS 2 .The absorption spatial map Figure 3c complements the optical image, where ≈5% and ≈9% of the 660 nm incident light is absorbed in the monolayer and bilayer MoS 2 , respectively.Furthermore, as depicted in Figure 3i, the bilayer regions show a standard increase in A 1g to E 1 2g Raman peak separation, from the monolayer Δ = ≈19.2cm −1 to Δ = ≈21.2cm −1 for bilayer.Additionally, increased Raman signal intensity (Figure 3h) from thicker material, and reduced PL intensity (Figure 3g) from the MoS 2 transition from direct (monolayer) to indirect (bilayer) bandgap, both indicate bilayer MoS 2 crystals in the same regions of this sample.Some flake edges are either fully or partially electrically isolated from one or both contact edges, which restricts charge carrier transport across the channel thus reducing the effective channel active area.One example of this is shown in location 3, where the low photocurrent in this region in Figure 3b can be partly explained by the discontinuity in the film across the channel, including non-coalesced small flakes that are isolated from the contacts.The locations with the best photocurrent are 1, 2, and 5.These locations show regions of strong PL within the flakes and near the contact edges (Figure 3g) and a blue shift in the PL peak (Figure 3f) within these better coalesced domains.The strong PL indicates high quality and purity in the MoS 2 crystal in these regions, resulting in longer carrier non-radiative recombination lifetime, and the blue-shifted PL peak position distribution within these domains infers less defects, tensile strain relaxation, and a higher carrier conductivity favorable for larger photogenerated current in the device.While location four had the largest PL red shift, and the domain spans the entire channel length, it is accompanied by a low PL intensity which explains the relatively low photocurrent when comparing to locations 1, 2, and 5.With a S-D voltage bias of 1 V, the maximum EQE of 0.254% (Figure 3d), and IQE of 5.821% (Figure 3e), were at location 5, which is also the largest area of strong uniform PL.The IQE is about one order of magnitude larger than the EQE.

Monolayer Film Mapping
Higher efficiency is expected for a 2D material film-based device with well-coalesced domains.A film grown using 1 mg MoO 3 and 12 min of sulfurization time, as shown in Figure 1i, was selected for the monolayer film sample correlative maps.The correlative maps for the uniform monolayer MoS 2 film-based phototransistor, with a maximum EQE of 0.374% and IQE of 10.15%, are shown in Figure 4d,e.The optical image (Figure 4a) shows uniform MoS 2 over the 20 μm x 20 μm (400 μm 2 ) device area.Large area monolayer uniformity is confirmed by the consistent ≈5% absorption of 660 nm incident light (Figure 4c), Raman intensity (Figure 4h), and monolayer A 1g to E 1 2g Raman peak separation of Δ = ≈19.2cm −1 (Figure 4i), all mapped within the channels of the phototransistor.These maps show that both electrically isolated and multilayer regions do not exist in the film.This indicates that less film morphology defects are present to significantly affect the device performance as compared to the flake sample device.The photocurrent of >0.07 nA was generated at all locations within the channel (Figure 4b) compared to the flake sample with less film coverage (Figure 3b), where <10% of the device area produce a photocurrent >0.07 nA.The maximum photocurrent in the film sample was 0.12 nA near the contact edges.Photothermoelectric effect is known to contribute to the total collected current, [28] and can be seen near some of the contact edges in the photocurrent map.While the domain boundaries cannot be easily identified in the optical image or the photocurrent, absorption, or Raman maps, the domain boundaries can clearly be seen in the map of PL peak position (Figure 4f).Where the domains coalesce, there is a slight blue shift of the PL peak.The boundaries affect device efficiency through undesirable carrier recombination.The largest photocurrent and IQE are mapped between the top and middle contacts.This upper channel contains a larger domain that correlates with uniform PL peak position; the upper channel also has less well-defined domain boundaries parallel to the contacts as compared to the lower channel, which also favors enhanced photocurrent.Even when the light was incident on the device contacts, photocurrent >0.04 nA was collected from the photogenerated carriers resulting from the scattered light, indicating relatively good film quality.

Film with Bilayer Islands Mapping
[38] To show this, a device was fabricated using a continuous monolayer MoS 2 film with bilayer island growth in the active area.The A 1g to E 1 2g Raman peak difference in Figure 5i shows the bilayer regions in the film even at a small sub-micron scale, with better resolution than any of the other mapping techniques.One of the bilayer islands between the contacts is shown at location 1 in the optical image in Figure 5a.The ≈5% monolayer absorption and ≈9% bilayer absorption of the 660 nm incident light clearly shows the number of layers in Figure 5c and correlate with the optical image.Additionally, the bilayer Raman shift of Δ = ≈21.2cm −1 , increased Raman intensity, and low PL all point to the bilayer islands at the same positions on the film, as seen in the respective spatial maps in Figure 5.The standard red shift in PL peak position from as-grown monolayer to bilayer is shown in Figure 5f, where the bilayer peak positions () are >673.7 nm (bandgap above 1.84 eV).The monolayer film PL peak is blue shifted in parts of the film near location 2, with a corresponding bandgap of ≈1.85 eV.This blue shift again corresponds to strong PL in this location, indicative of higher electronic quality and relatively higher photocurrent than in nearby regions with red-shifted and reduced PL.Although the device has a fully coalesced monolayer film, the film generates a non-uniform photocurrent as shown in Figure 5b.A likely explanation for the regions of enhanced photocurrent is increased absorption at the bilayer regions, clearly seen in the top and middle channels.Of the bilayer regions, the most efficient light to electricity conversion is at location 1, where the bilayer connects with the Ti contact.The IQE at location 1 is the highest of the bilayer regions (>3.4%) compared to the other bilayer IQE regions (<3.34%) and is higher than parts of the monolayer film IQE.This indicates that high-efficiency devices can be fabricated by increasing the number MoS 2 film layers.The lower channel has the most uniform photocurrent over the mapped area.This corresponds to the region with strong PL that is most widely distributed across the mapped area, as compared to the other channels, indicating a region of monolayer film with lower defect density.

Film with Transferred Second Monolayer Mapping
From the monolayer with as-grown bilayer island device performance, we know that increased absorption can lead to better phototransistor responsivity; therefore, multilayer uniform MoS 2 films can lead to higher efficiency devices.Several groups have explored the direct synthesis of multilayer TMDC films; [39][40][41][42] however, it is still challenging to grow large area uniform multilayer films by CVD.Our approach to increase absorption is to use layer transfer to stack a uniform monolayer film onto a device fabricated on an as-grown monolayer film, resulting in a two monolayer MoS 2 film phototransistor.Two films grown using 1 mg MoO 3 and 12 min of sulfurization time, as shown in Figure 1i, were selected for the device.More details on the transfer and device fabrication can be found in the Experimental Section.The optical image for a section of one device is shown in  The correlative maps are shown in Figure 6.The absorption map agrees with the optical image, where uniform ≈5% and ≈9% absorption was obtained for monolayer and two monolayer regions, respectively.The Raman intensity map shows uniformity in the transferred film over the mapped area, including at the Ti contacts since the film was transferred on top.Like the as-grown films with bilayer islands, the intensity increased for two stacked monolayers; however, the monolayer A 1g to E 1 2g Raman peak separation Δ = ≈19.2cm −1 was maintained across both the single and double monolayer regions, as seen in in Figure 6i, indicating that there was no structural transition from monolayer to bilayer MoS 2 .A comparison of the Raman spectra for the as-grown and stacked monolayer films is shown in Figure S3, (Supporting Information).
Since the MoS 2 CVD synthesis on C-plane Al 2 O 3 is quaziepitaxial, [14] and the lattices are not perfectly matched, some lattice strain is expected.The PL peak position map (Figure 6f) shows a blue shift in the transferred monolayer film PL, unlike the standard red shift for bilayer MoS 2 , as seen in Figure 5f.The map shows an average increase in the bandgap of ≈0.011 eV and an even further bandgap increase of ≈0.02 eV at some regions on the Ti contact.We attribute this shift to strain relaxation, along with film and contact interface effects.Unlike in conventional bilayer MoS 2 films, where PL decreases due to a transition to an indirect bandgap, [32] the PL intensity is roughly tripled in the stacked monolayer region relative to the single monolayer region, as seen in Figure 6g.A PL decrease approaching as much as 14% in some locations is seen for the parts of the transferred film that are directly on the Ti contacts due to PL quenching.A comparison of the PL spectra for the stacked monolayer and for a single film pre and post transfer is shown in Figure S5 (Supporting Information).The average photocurrent in the channels for the monolayer transferred film is ≈0.09 nA, while the stacked monolayer regions have an average photocurrent of ≈0.11 nA, with photocurrent approaching 0.15 nA in several locations.The thicker ab-sorber created by stacking a second film produces more photocurrent than the monolayer region, with a maximum EQE of 0.54%, which is more than the maximum as-grown bilayer EQE of 0.25% in Figure 5d.An IQE ranging from 3.7% to 5.1% is shown in Figure 6e for the stacked layers, which like the as-grown bilayer, is lower than the monolayer IQE, which is 6.3% on average in Figure 6e.
While each of these 2D MoS 2 film morphologies shows different features, consistent trends appear for the mapped properties.Tracking PL peak shifts is one of the less obvious trends.For the as-grown MoS 2 films, compressive strain in the monolayer crystal lattice due to the quasi-epitaxial growth on sapphire substrates affects the film quality and the efficiency of the devices (see Figure S5, Supporting Information).Across multiple morphologies with intentionally grown lower quality films, a stronger PL correlates with a blue shift in the PL peak.We attributed this shift to localized strain relaxation in regions with reduced defect density.This infers that parts of the monolayer film and flakes with lower crystalline order have a more difficult time relaxing into unstrained configurations from the as-grown position.The higher-quality regions with blue-shifted PL are associated with enhanced photocurrent, especially when adjacent to other relatively low defect regions.A different trend is seen for the highquality well-coalesced monolayer as-grown films, where PL peak position is relatively uniform, and a small blue shift is observed only at the grain boundaries.

Back-Gated Phototransistor Analysis
As an example of another useful device type from these materials, the electrical characteristics of a back-gated phototransistor from the MoS 2 monolayer film, grown using 1 mg MoO 3 and 12 min of sulfurization time as seen in Figure 1i, are shown in Figure 7. Titanium contacts in a grid pattern are fabricated on a SiO 2 (300 nm) on Si substrate, and the MoS 2 film is transferred on top of these contacts, as shown schematically in Figure 7a.The transistor response was modulated with different incident light power at an excitation wavelength of 670 nm.With a source-drain voltage bias of V SD = 2 V, light-enhanced source-drain current I SD is seen at incident optical power as low as 8 nW, with further enhancement up to 25.2 μW, where the I SD reached 9 μA, which is 3 orders of magnitude larger than the dark current as shown in Figure 7b.Back-gate tunability was also obtained for the phototransistor, as shown in Figure 7c, where increased I SD is shown for higher gate voltages at a source-drain bias V SD = 2 V.The gate bias V G at which the transistor switches from the OFF to ON state is reduced as the incident light power is increased.At gate bias V G = 55 V, a photocurrent of 60.4 μA, was obtained at incident power of 8 nW.The photocurrent and photoresponsivity as a function of incident light power are shown in Figure 7d.The expected decrease in responsivity as the incident optical power and the corresponding photocurrent is increased due to increased electrontrap saturation, photo-excited carrier recombination, and Auger recombination.

Conclusion
In summary, we demonstrated the synthesis of large area, highquality MoS 2 films with different film morphologies and domain sizes up to 30 μm by quazi-epitaxial growth on Al 2 O 3 substrates using the CVD process.The work gives insight into the film growth mechanism under various conditions, which can be used to optimize film quality toward high quality, large area, and highly uniform films for use in efficient devices.Furthermore, we used high-resolution correlative spatial mapping to reveal the connection between the optical, electronic, and structural properties of these 2D MoS 2 morphologies in the context of their use in a phototransistor device.The PL peak position mapping revealed domain boundaries more clearly in the monolayer film, while the A 1g to E 1 2g Raman peak difference showed the bilayer regions, with better resolution than any of the other mapping techniques employed in mapping the multilayer as-grown films.Correlating the PL maps with photocurrent maps, we showed that strain relaxation led to stronger PL and enhanced photocurrent.Furthermore, by stacking two monolayer films, we showed that a maximum EQE of 0.29%, more than the maximum asgrown bilayer EQE, was achieved.This analysis can be employed in the development and characterization of a range of 2D-based materials and devices.This approach will aid the ongoing pursuit of high-efficiency, lightweight, flexible, and wearable electronics, optoelectronics, and photovoltaics from 2D materials.

Experimental Section
Synthesis of MoS 2 Films: MoS 2 films of up to 1 cm 2 were grown using an MTI OTF-1200X-II dual zone split tube furnace.A third low-temperature zone was introduced by wrapping the quartz tube with a Grainger SLR series silicone heating blanket.ACS reagent ≥99.5% molybdenum (VI) oxide (MoO 3 ) powder and 99.98% trace metals basis sulfur (S) powder from Sigma-Aldrich were used as precursors for the CVD synthesis.A 1-inch diameter quartz tube was placed in the three-zone furnace where zones 1, 2, and 3 were the upstream (heating blanket), middle, and downstream zones, respectively.A C-plane sapphire substrate was placed on top of an alumina boat, and the MoO 3 powder was positioned in the boat 1.5 cm upstream from the center of the substrate.The amount of sulfur was a critical parameter in the formation of oxy-particles of MoO 3 (or MoO 2 ) throughout the films. [43]For the range of MoO 3 precursors used, 230 mg of sulfur was used and resulted in no oxy-phases in the films, as shown in the Raman spectra in Figure S6 (Supporting Information).The boat with the MoO 3 powder and sapphire substrate was centered in zone 2, and sulfur powder was placed in an alumina boat in zone 1, 30 cm upstream from the MoO 3 powder.The tube was initially vacuumed down to a pressure of 15 mTorr, then ultrahigh-purity argon (Ar) was used as the carrier gas at a flow rate of 180 sccm, with the tube pressure maintained at 2.4 Torr.The temperatures of the second and third zones of the furnace were set to 750 °C with a 50 min ramp time.At the end of the ramp, zone 1 was heated up to 120 °C, and zones 2 and 3 were held at 750 °C throughout the duration of the growth.The growth time for the purposes of optimization and repeatability (see Figure 1) begins when zone 1 reaches 750 °C.More growth details and schematics are reported in the previous work. [44]he MoO 3 precursor supply of 0.5, 1.0, 3.0, and 5.0 mg were used for sulfurization times of 8, 10, and 12 min.The PL results for these films are shown in Figurez S7 (Supporting Information).Uniform monolayer films, monolayer films with bilayer growth, films of fully and partially coalesced flakes, and flakes with bi-layer growth were chosen for the MoS 2 phototransistors.To assess the stability of these 2D films, it measures PL on a uniform monolayer film, grown using 1 mg MoO 3 and 12 min of sulfurization time as seen in Figure 1i, directly after growth and six months after growth (stored in a nitrogen environment at room temperature), showing no notable degradation (see Figure S8, Supporting Information).
Device Fabrication and Film Transfer: To fabricate the phototransistors, 200 nm of 495 A4 PMMA resist, followed by 400 nm of 950 C4 PMMA resist, both from Kayaku Advanced Materials, were spin coated on the as-grown MoS 2 on Al 2 O 3 substrates.A thin chromium layer was used for charge dissipation.The contacts were patterned using a RAITH VOY-AGER 100 electron beam lithography (EBL) tool, and 50 nm Ti was deposited at a rate of 0.5 Å s −1 using an Angstrom Engineering Nexdep electron beam evaporation tool.The complete steps were shown schematically in Figure S9a (Supporting Information).To fabricate the monolayer plus stacked monolayer phototransistor, a second EBL step and development was done to expose part of the as-grown MoS 2 film in the device active area.Chromium etchant from Sigma-Aldrich was used to etch away the exposed MoS 2 and the remaining resist was dissolved with acetone creating a clean monolayer film to no film step edge.After a solvent (acetone and IPA) clean, the surface-energy-assisted film transfer process, [16] was used to transfer a second monolayer MoS 2 film from an Al 2 O 3 substrate onto the phototransistor.The fabrication steps for this second device type were added shown schematically in Figure S9b (Supporting Information).The device was then annealed at 200 °C for 2 h.
Analysis of Film Strain: Since the MoS 2 CVD synthesis on C-plane Al 2 O 3 results in strain in the MoS 2 lattice, the strain was expected to relax when the film was transferred onto other substrates or device contacts.The A 1g to E 1 2g Raman peak separation did not clearly indicate any difference in strain for the as-grown MoS 2 on sapphire, MoS 2 transferred onto the titanium contacts, MoS 2 transferred onto sapphire, and MoS 2 transferred onto as-grown MoS 2 , as shown in Figures S4a and S4b (Supporting Information).However, strain relaxation from as-grown to transferred films is evident in the difference in Raman intensity ratio I(A 1g )/I(E 1 2g ) as shown in Figure S4c (Supporting Information); these trends indicate that the as-grown films are compressively strained. [45]The as-grown MoS 2 compressive strain was more relaxed when transferred onto the titanium contact than onto the sapphire substrate.
Producing the Spatial Maps: The Raman and PL spectroscopy measurements and maps were generated using a WITec alpha300 Ri inverted Raman imaging confocal microscope and the analysis was carried out using WITec Suite FIVE data acquisition and analysis software.The Raman system's 532 nm excitation laser line was used to make the measurements in an ambient air environment, and the laser power was kept at 1.5 mW to avoid thermal effects in the film during scans.The emission collection objective was a 100 x Nikon TU Plan Fluor EPI with a numerical aperture of 0.9.A ≈300 nm spot size and stage step size of 200 nm was used for the area scans with an integration time of 0.02 s for PL and 1 s for Raman.The PL measurements were taken using a 300 g mm −1 grating with a 750 nm blaze wavelength.The Raman measurements were done using a 600 g mm −1 grating with a 500 nm blaze wavelength.
The photocurrent and absorption maps were obtained with a single scan where photocurrent and transmission were measured simultaneously.An MLS203-2 fast XY scanning stage from Thorlabs was used to raster a ≈0.3 μm beam width, 660 nm laser over a 20×20 μm scan area, with a 0.2 um step size.The light source was a supercontinuum laser, coupled to a laser line tunable filter, both from Fianium/NKT Photonics.For photocurrent maps, the MoS 2 phototransistor contacts were probed using XYZ 300 micro positioners from Quater Research & Development.A source-drain bias of 1 V was applied and the transistor photogenerated current was measured using a Keithley 2450 sourcemeter; the dark current under 1 V bias was subtracted.The optical transmission through the film was measured using a Thorlabs FDS1010-CAL calibrated photodiode (CPD), mounted 1 mm below the sample, for both the 100% reference (Al 2 O 3 substrate) and sample (MoS 2 + Al 2 O 3 substrate) transmission measurements.All the components in the measurement setup were integrated with a LabVIEW virtual interface for data collection.The absorption was calculated from the transmittance and reflectance.[48][49] This calculation cannot be applied at regions within the mapped area where the laser spot was the device contacts, where a reflectance of ≈100% and transmittance of 0% was expected.The average monolayer absorption was used in the two-stacked monolayer devices at the regions with MoS 2 monolayer film directly on top of the Ti contacts.
Statistical Analysis: Synthesis of MoS 2 films: Growth parameters for all 12 film morphologies shown in Fig. 1 were used to show reproducibility over at least three statistically independent samples.Each sample's optical, electronic, and structural properties were characterized at nine locations (the center of each square of a 3×3 grid defined over the entire growth substrate) to assess film quality by transmission, photoluminescence, and Raman spectroscopy.
For the photocurrent measurements, a dark measurement was taken for baseline correction to account for detector noise and was subtracted from the final measurement as discussed in the experimental section on the spatial map production.A dark measurement was also taken and subtracted from the measurement obtained for the transmission maps.
Spatial Mapping: Spatial maps were produced over a 20 μm × 20 μm (400 μm 2 ) area with a 200 nm step size to obtain 10000 sample points per map.Hyperspectral PL and Raman spectroscopy were used to obtain 10000 spectra for all PL peak position, PL peak intensity, Raman peak intensity, and Raman A 1g to E 1 2g peak separation.Raman A 1g to E 1 2g peak positions, separation, and intensity ratio I(A 1g )/I(E 1 2g ) are represented with mean ± standard deviation for nine measurements.The one sample t test was used for statistical analysis of the Raman peak intensity ratio I(A 1g )/I(E 1 2g ) to show that the MoS 2 transferred on as-grown MoS 2 , MoS 2 transferred on sapphire, and MoS 2 transferred on titanium samples were significantly different from the as-grown MoS 2 prior to transfer, in order to assess strain relaxation.Significance was defined as p < 0.001.Origin 2021b software by OriginLab Corporation was used for statistical analysis.

Figure 1 .
Figure 1.a-l) Optical images of CVD synthesized MoS 2 on C-plane sapphire substrates.As-grown MoS 2 flake and film morphology results from varying the amount of MoO 3 powder precursor (y-axis) and sulfurization time (x-axis).The scalebars are 20 μm.

Figure 2 .
Figure 2. a) Optical image of a large area phototransistor fabricated on as-grown 2D MoS 2 .b) The optical microscope-based setup for simultaneous spatial mapping of transmission and photocurrent.

Figure 3 .
Figure 3. Spatial maps of MoS 2 flakes with phototransistor Ti contacts on top, all correlated to the same region: a) the optical image, b) photocurrent, c) absorption, d) EQE, e) IQE, f) PL peak wavelength, g) PL intensity, h) Raman A 1g peak intensity, and i) Raman A 1g to E 1 2g peak separation.

Figure 4 .
Figure 4. Spatial maps of a monolayer MoS 2 film with phototransistor Ti contacts on top, all correlated to the same region: a) the optical image, b) photocurrent, c) absorption, d) EQE, e) IQE, f) PL peak wavelength, g) PL intensity, h) Raman A 1g peak intensity, and i) Raman A 1g to E 1 2g peak separation.

Figure 5 .
Figure 5. Spatial maps of a monolayer MoS 2 films with as-grown bilayer islands and phototransistor Ti contacts on top, all correlated to the same region: a) the optical image, b) photocurrent, c) absorption, d) EQE, e) IQE, f) PL peak wavelength, g) PL intensity, h) Raman A 1g peak intensity, and i) Raman A 1g to E 1 2g peak separation.

Figure 6a ,
Figure 6a, where the difference in contrast clearly shows the contacts, monolayer MoS 2 region noted by 1, and the region with a second stacked monolayer MoS 2 film on top noted by 2. The lighter region to the left of the image has two MoS 2 layers (asgrown and transferred), and the darker region to the right has only the transferred monolayer film; the metal contacts are on top of the as-grown layer and underneath the transferred layer.The correlative maps are shown in Figure6.The absorption map agrees with the optical image, where uniform ≈5% and ≈9% absorption was obtained for monolayer and two monolayer regions, respectively.The Raman intensity map shows uniformity in the transferred film over the mapped area, including at the Ti contacts since the film was transferred on top.Like the as-grown films with bilayer islands, the intensity increased for two stacked

Figure 6 .
Figure 6.Spatial maps of (left) monolayer MoS 2 film with Ti contacts on top and a transferred monolayer MoS 2 film stacked on top of that and (right) just a monolayer MoS 2 film transferred onto Ti contacts, showing correlation between each: a) the optical image, b) photocurrent, c) absorption, d) EQE, e) IQE, f) PL peak wavelength, g) PL intensity, h) Raman A 1g peak intensity, and i) Raman A 1g to E 1 2g peak separation.

Figure 7 .
Figure 7. Optoelectronic characterization of a monolayer MoS 2 phototransistor with a back gate.a) Drain current (I D ) as a function of source-drain voltage (V DS ) for an excitation wavelength of 670 nm at different illumination power and gate bias V G = 0 V. b) Schematic of a phototransistor with monolayer MoS 2 film transferred on top of Ti contacts.c) Drain current as a function of gate voltage under different illumination power and source-drain bias V SD = 2 V. d) The dependence of photocurrent and responsivity on incident optical power is shown.