Direct Observation of Semimetal Contact Induced Charge Doping and Strain Effect in CVD‐Grown Monolayer MoS2 Transistors

Two‐dimensional Materials (2DMs) offer significant promise for advancing device miniaturization and extending Moore's law. Despite the challenges posed by high contact resistance in transistors, recent discoveries highlight semimetals as an effective approach for achieving ohmic contact with near‐quantum‐limit contact resistance. The energy band hybridization between semimetal and MoS2 is found to create degenerate states and heavily doped contact, which is proposed as the underlying mechanism responsible for reducing contact resistance. However, a quantitative and comprehensive characterization of the semimetal‐MoS2 interface is lacking, leaving the physical interactions elusive. This study reveals that semimetals induce n‐type doping and tensile strain in monolayer MoS2 grown using CVD, which serve as the contact resistance and mobility boosters. Among the semimetals investigated, including Bismuth (Bi), Antimony (Sb), and their alloy, Bi results in the highest electron doping of 2 × 1013 cm−2 and a 0.5% tensile strain, leading to reduced contact resistance and enhanced mobility. First‐principles calculations and spectroscopy measurements unveil the impact of electron doping and strain in MoS2, and the thermal effects are subsequently explored. This research underscores the potential of semimetals in boosting device performance and lays the foundation for reducing contact resistance in transistors made from 2D materials.


Introduction
The International Roadmap for Devices and Systems (IRDS) identifies 2D semiconductors, notably transition-metal dichalcogenides (TMDCs) such as molybdenum disulfide (MoS 2 ), as potential channel materials that will extend the DOI: 10.1002/aelm.202300820transistor scaling roadmap. [1]Scaling below 1 nm node as well as high density back-end-of-line (BEOL) integration are two areas in which 2D semiconductors have demonstrated compelling advantages. [2,3]Contact resistance (R sd ) is one of the essential factors in achieving high performance transistors.A high R sd hampers carrier transport through the semiconducting channel and reduces the on-state drain current (I d ).
6][7] Various methodologies have been developed to mitigate these issues and achieve ohmic contacts.10][11] To address MIGS, a buffer layer (e.g., metallic, semiconducting, and insulating) could be inserted between the 2DMs and the contact metal, which can attenuate metal wave function penetration. [12,13]However, the method might introduce a van der Waals gap, potentially causing a tunneling gap that limits carrier injection efficiency.Heavily doping is a commonly employed method to improve efficiency by reducing the Schottky barrier width.This allows electrons to tunnel the Schottky barrier, which outweighs thermionic emission.Considering the conventionally ion implantation or thermal diffusion is technically challenging for 2DMs, mild doping strategies such as surface charge transfer doping and phase changes have been proposed to metallicize metal-contacted TMDs, facilitating charge carrier injection and forming an ohmic contact. [14,15][18][19][20][21] First, semimetal offers the advantage of suppressing the MIGS due to the nearzero density of states (DOS) at the Fermi level.In earlier work, the use of Bismuth (Bi) results in an ohmic contact with a contact resistance as low as 105 Ω·μm. [16]Considering its low melting point of 274.5 °C, the Antimony (Sb) contact is subsequently studied which achieves a R sd of 660 Ω μm and an improved thermal stability up to 400 °C. [18]Second, semimetal is predicted to induce metal-semiconductor (M-S) energy band hybridization, leading to degenerate doping of the 2D channel beneath the semimetal contact. [17]Recently, using Sb (01 12) contact achieved a nearquantum-limit R sd of 42 Ω·μm. [17]n in-depth understanding of the interaction between 2D materials and semimetals is crucial for developing effective strategies to create ohmic contacts.However, despite the potential of such a method, the quantification of strain and charge doping at the semimetal-MoS 2 interface has yet to be thoroughly characterized.Moreover, given the semimetal has a relatively low melting point, the temperature-dependent effects of semimetal contact on 2D materials play an important role in determining transistor performance, yet these effects are rarely reported.The inherent low dimensionality of 2D materials presents challenges in characterizing their interface properties.[24][25][26][27] However, the variation in experimental conditions makes findings from previous reports difficult to compare.
In this work, we quantitatively investigate the strain and charge doping in CVD-grown monolayer MoS 2 as induced by semimetal contact, including Bi, Sb, and various stochastic alloys.All semimetals were deposited under the same condition using an ultra-high vacuum evaporation system, and underwent characterization through AFM, XPS, and XRD techniques.Levering on Raman and photoluminescence (PL) spectroscopy ɛ-n map, we discovered that semimetals induce n-type doping and tensile strain in MoS 2 , with Bi yielding the highest electron doping concentration of 2 × 10 13 cm −2 and tensile strain of 0.5%.Density functional theory (DFT) calculations further confirm the electron transfer from semimetal to MoS 2 .The impact of the thermal budget was subsequently investigated through spectroscopy and the performance of MoS 2 transistors.After being annealed at 200 °C, the device exhibits the highest mobility and lowest contact resistance.The antimony demonstrates the best thermal stability, which is confirmed by the consistent threshold voltage throughout the whole annealing process up to 400 °C.

Results and Discussion
First, we perform first-principles calculations to gain an in-depth understanding of the electrical properties of Bismuth (Bi)-and Antimony (Sb)-MoS 2 heterojunctions (see the Supporting Information for details).We construct a model consisting of Bi or Sb, with their (0001) surface in contact with a monolayer of MoS 2 , respectively.It is well known that contact resistance is highly dependent on the interfacial bonding states and electron transfer.
We first analyze the wavefunction of the conduction band maximum (CBM), and the calculated isosurfaces of wavefunctions and electron densities are shown in Figure 1a,b.It can be seen the major wavefunction of CBM is contributed by Bi atoms and a minor part from Mo atoms in Bi-MoS 2 shown in Figure 1a.Compared to the result of Sb-MoS 2 shown in Figure S2 (Supporting Information), the wavefunction of CBM in Bi-MoS 2 is more delocalized than that of CBM in Sb-MoS 2 , resulting in a lower contact resistance.Figure 1b shows that the bonding states are more than the antibonding state, revealing a stable interfacial structure and the existence of electron transfer between semimetals and MoS 2 .Then we calculate the partial density of states (PDOS) and the total density of states (TDOS) of these two types of heterojunctions (Figure 1c).It can be seen that Bi and Sb have a predominant contribution near the Fermi level, resulting a good contact conductivity.In comparison, Bi is better than Sb for enhancing the contact conductivity since Bi has a stronger PDOS than Sb near the Fermi level.We further analyze the electron transfer between semimetals and MoS 2 , and the results are shown in Figure 1d.It was found that each Bi can donate 0.01988 e while each Sb can donate 0.0150 e.These results are consistent with the calculated PDOS, in which, Bi has more electron density near the Fermi level.Then we change the atomic ratio of Sb to Bi from 125:25, 100:50, and 75:75.The simulated electron transfer between the Bi-Sb alloys and MoS 2 is 0.0144, 0.0146, and 0.0150 e per atom.Our results show that stoichiometric Bi and Sb results in the highest electron transfer among three atomic ratios.
Next, we deposit semimetals on MoS 2 and proceed with material characterizations.The monolayer MoS 2 films were directly grown on a sapphire substrate by the Chemical Vapor Evaporation (CVD) method.Then, 20 nm semimetals (Bi, Sb, and alloy) were subsequently deposited onto MoS 2 separately for spectroscopy characterizations using an ultra-high vacuum (<10 −9 torr) molecular beam epitaxy (MBE) system.The growth of different semimetal types was carried out at room temperature, each with the same growth duration of 10 min.Ultra-high vacuum and room temperature deposition have been proven effective in preserving the crystalline structure of 2DMs and minimizing the interfacial defects, [8] which can reduce contact resistance.][30][31] The X-ray photoelectron spectroscopy (XPS) spectra of Bi 4f and Sb 3d core levels in Figure S6 (Supporting Information) confirm the existence of Bi, Sb, and their alloys.This study examined two different Bi-Sb alloys with atomic ratios of 22% and 42%.Recently, both the Sb (0001) and Sb (01 12) surfaces are explored as semimetal contact to MoS 2 transistors. [17]Figure S7 (Supporting Information) shows the X-ray diffraction system (XRD) diffractograms of our Sb thin films grown on MoS 2 at room temperature.31] To quantify the strain and electron doping properties of the semimetal-MoS 2 interface, we then conducted Raman and PL spectroscopy characterizations.MoS 2 exhibits two characteristic Raman phonon modes, one out-of-plane A 1g and one in-plane E 1 2g vibration mode.The Raman peak at ≈415 cm −1 is attributed to the Sapphire substrate. [32]When compared to the baseline condition (MoS 2 without contacts), both the A 1g and E 1 2g vibration modes of the semimetal-MoS 2 interface show redshift, indicating n-type doping and tensile strain effect (Figure 2a). [23,24]Two Raman vibration modes of Bi-contacted monolayer MoS 2 exhibit the maximum redshift, which can be translated into a higher electron concentration and tensile strain.25][26] The Raman-derived ɛ-n map is depicted in Figure 2b.Compared with suspended MoS 2 (E 1 2g mode at 385 cm −1 and A 1g mode at 405 cm −1 ), the sapphire substrate induces a tensile strain effect on MoS 2 films, causing a redshift of the E 1 2g mode while having a negligible effect on the A 1g mode. [24]Within this work, the Raman modes of blank MoS 2 grown on a sapphire substrate exhibit an E 1 2g mode at 381 cm −1 and an A 1g mode at 405.5 cm −1 .The Raman peaks of the blank MoS 2 are used as the reference point, representing that there is no additional strain and charge doping induced by semimetals.By calculating the values of Raman shift values relative to this reference point, the distributions of doping concentration and strain effect of the semimetal on MoS 2 are shown.25][26] Leveraging the computational and experimental derived parameters, a new coordinate system of strain (ɛ) and electron concentration (n) is constructed, overlapped with the Raman peak  A - E coordinate system.Here, Δn > 0 indicates a higher electron concentration and Δɛ > 0 signifies the presence of tensile strain.Further details of calculation are shown in the Section S6 (Supporting Information).The scatter plots reveal that Bi-MoS 2 exhibits the most considerable tensile strain (0.5%) and highest electron density (≈1-2 × 10 13 cm −2 ).Conversely, the Sb-MoS 2 displays the least amount of strain and doping.The alloy, on the other hand, falls in between Bi and Sb.
The observation of PL is consistent with the trend found in Raman.Figure 2c compares four representative PL spectra of different semimetals on monolayer MoS 2 .The curve peaked at 680 nm corresponds to the emission caused by the recombination of neutral excitons (A 0 ), and the curve peaked 20-30 nm higher is attributed to trions (A − ).The Bi-MoS 2 shows the most apparent redshift and the most severe quenching of PL.The effect of Sb on PL peak shift and intensity is minimal.The redshift in the PL peak in semimetal-MoS 2 suggests that the trions are more populated than neutral excitons, a common sign of more n-type doping. [33]Figure 2d shows that PL is dependent on strain and charge doping concentration, wherein the horizontal axis indicates the A 0 shift and the vertical axis indicates the PL intensity ratio (I − /I tot ), respectively.The x-axis uses the relationship between A 0 and strain, showing a redshift of 110 meV redshift per % biaxial tensile strain. [27]On the y-axis, the n is correlated to the PL intensity ratio using a mass-action model and a threelevel model. [24,27,34,35]Here, we use the PL intensity ratio and the A 0 peak position of pristine MoS 2 on a sapphire substrate as a reference point, denoted by Δn = 0 and Δɛ = 0.More details of the calculation are shown in the Section S7 (Supporting Information).It can be seen that the PL intensity ratio increases and PL peak shifts toward a lower energy with increasing Bi concentration.In other words, Bi has a more significant electron doping and tensile strain effect on MoS 2 than Sb and Bi-Sb alloy.There is some discrepancy in the maximum doping levels determined from Raman and PL, which might arise from the inhomogeneity of monolayer MoS 2 films and the different theories behind the ɛ-n map derivation.
To investigate the thermal stability of the semimetal and the influence of annealing temperature on the contact, the samples were then subjected to a rapid thermal annealing in an N 2 atmosphere, ranging from 100 to 400 °C, for a duration of 20 s.Considering a low melting point of Bi (≈271.4°C), the samples were annealed at a maximum temperature of 200 °C.The peak positions of the A 1g and E 1 2g modes are plotted as a function of temperature, as depicted in Figure 3a,b.Notably, the redshift of both Raman peaks of Bi-MoS 2 indicates that the thermal treatment further enhances electron doping density and strain.The thermal coefficient  was subsequently extracted from the slope following the Grüneisen model. [36]It is worth noting that the thermal coefficients of Bi-MoS 2 show an order of magnitude higher than pristine MoS 2 , which indicates reduced thermal stability.The statistical analysis in Figure 3c,d reveals that Sb and the alloys induce lower doping levels and weaken the strain effect while increasing thermal stability.Note that when the temperature exceeds 300 °C, both the Raman mode of Sb and the alloys change to blue shift, probably due to the disintegration of the semimetal's structure at higher temperatures.Even though Sb has a higher melting point, mixing with metals of even higher melting point may hold promise for improved thermal stability.The origin of the strain effect in MoS 2 layers probably arises from the thermal expansion coefficient (TEC) mismatch between the layers and the semimetal contacts. [24]The in-plane TEC of monolayer MoS 2 , Bi, and Sb is 7.6 ± 0.9 × 10 −6 /K, [37] 13.4 × 10 −6 /K, [38] and 11 × 10 −6 /K, [38] respectively.The larger TEC of Bi induces a greater level of tensile strain on MoS 2 than Sb, which is consistent with the experimental results.Moreover, alloys with a higher Bi content of 42% consequently exhibit a more significant thermal mismatch, leading to more pronounced tensile strain, as evident in Figure 3d.
To gain deeper insights into the impact of semimetals on MoS 2 , back-gated field-effect transistors (FETs) with varying semimetal contacts were fabricated.All the devices were subjected to the same thermal annealing condition ranging from 100 to 300 °C. Figure 4a shows the transfer curves of the Bi-MoS 2 FET at different annealing temperature, showcasing optimal performance after annealing at 100 and 200 °C.When the annealing temperature reaches 300 °C, the device current decreases as the semimetal's metaling point is exceeded.Figure 4b displays the output curves following an anneal at 200 °C.It can be seen that the contact between the metal electrode and MoS 2 is greatly improved and the ohmic contact is achieved.Figure 4c compares three transfer curves after being annealed at 200 °C, showing the achievement of high on-current density and a substantial on/off ratio in all semimetal-contacted MoS 2 transistors.
For a comprehensive analysis, we employed a universal extraction technique to extract and compare intrinsic mobility (μ o ), contact resistance (R sd ), and threshold voltage (V T ) for all devices (see Section S8, Supporting Information for details). [39]Initially, a H(Y) function was calculated from the experimental results throughout the entire temperature range.Good linear fittings were attained in Figure 4d, wherein the mobility dependency factor  was determined from the slope.Subsequently, the μ o and V T were obtained from the slope and x-intercept of Figure 4e, respectively.Finally, R sd was calculated using the drain current equation.Extracted parameters are compared in Figure 4f,h.Notably, all devices achieve their peak current density and mobility after being annealed at 200 °C (Figure 4f), arising from a higher concentration of electron doping and reduction of the effective mass (m*) due to tensile strains. [40]This observation is consistent with the trend identified through spectroscopy characterization.The thermal annealing also reduces the contact resistance, despite a tendency to degrade after 300 °C treatment (Figure 4g).Thus, the trade-off between doping/strain and thermal stability should be thoughtfully weighed.While Sb shows the best thermal stability, as further confirmed by the stable threshold voltage shift (Figure 4h), exploring alloys using metals with higher melting points could present a promising avenue for even better thermal stability.The semimetals have demonstrated substantial potential in minimizing contact resistance approaching the quantum limit and facilitating ongoing device scaling.Further refinements, including improvements in material growth technique, device thermal treatment, and optimization of contact alloy ratios, are essential for continued performance enhancement.

Conclusion
Bismuth (Bi), antimony (Sb), and their alloys were deposited onto MoS 2 using an ultra-high vacuum evaporation system, and the effects of semimetal-induced doping and strain on MoS 2 films were investigated through spectroscope and electrical charac-terizations.The results reveal that electron doping and tensile strain effects can be introduced to MoS 2 by semimetals, with these effects being more pronounced when subjected to thermal annealing at temperatures not exceeding 200 °C.Among the semimetals investigated, Bi demonstrated the most significant impact concerning both electron doping and tensile strain, while Sb showcased the most consistent and stable properties.The presence of Bi within the alloy compositions led to a more pronounced strain effect, primarily attributed to its higher thermal expansion coefficient (TEC).The beneficial effects of semimetals in terms of reducing contact resistivity and enhancing mobility in MoS 2 are underscored by these findings.Concurrently, the critical necessity for precise alloy ratio design and optimized thermal treatments to further enhance transistor performance is emphasized.

Figure 1 .
Figure 1.DFT calculations of semimetal-MoS 2 interface.a) The calculated isosurfaces of wavefunctions of the conduction band maximum (CBM) of Bi on MoS 2 b) The calculated isosurfaces of electron density difference between Bi and MoS 2 .The red and blue color represents the electron accumulation and depletion, respectively.c) The calculated partial density of states (PDOS) and the total density of states (TDOS) of Bi-MoS 2 and Sb-MoS 2 heterojunctions.The Fermi level is set at zero as indicated by grey dash line.d) The calculated charge transfer between Bi, Sb, and Bi-Sb alloy with MoS 2 based on the Bader charge calculation.The negative value indicates the electron lose (donor).

Figure 2 .
Figure 2. Raman and PL characterization of strain and charge doping effect.a) Raman spectra of blanked MoS 2 and three types of semimetal-MoS 2 films.The red shift of A 1g and E 1 2g mode suggests higher electron doping and tensile strain.b) A Raman-derived strain-charge doping (ɛ-n) map.The peak position of blanked MoS 2 film without any contact is used as reference point.c) PL spectra of blanked MoS 2 and three types of semimetal-MoS 2 films.The redshift of PL peak position and the quenching of PL intensity indicates the trions are more populated than neutral excitons, a common sign of more n-type doping.d) PL-derived ɛ-n map.

Figure 3 .
Figure 3. Thermal budge effect on semimetal-MoS 2 heterojunctions.Temperature-dependent Raman spectra peak positions for a) E 1 2g and b) A 1g mode of Bi-MoS 2 heterojunctions.The peak positions obtained from Lorentzian fittings for A 1g and E 1 2g modes versus temperature were fitted using the Grüneisen model, (T) =  0 + T, where  0 is the peak position at 0 K and  is the first-order temperature coefficient.A linear function was employed to fit the curve and the slopes indicate the thermal coefficients .c) Strain-doping relationship of Bi-and Sb attached MoS 2 as a function of temperature.d) Strain-doping relationship of alloy attached MoS 2 as a function of temperature, the alloy ratio of Bi is 42% and 22%.

Figure 4 .
Figure 4. Electrical characterization of semimetal contacted MoS 2 transistors.a) Transfer curves of Bi contacted MoS 2 transistor at different annealing temperatures.b) Output curve of the Bi contacted MoS 2 transistor with channel length of 1 μm.The linear behavior at low drain voltage indicates ohmic contact.c) Transfer curves comparing Bi-, Sb-, and alloy contacted MoS 2 transistor with channel length of 1 μm at elevated annealing temperatures.d) Linear regression of the H(Y) and Y functions to V G for Bi-contacted MoS 2 at different annealing temperatures.e) Extracted intrinsic mobility μ 0 and contact resistance R sd of the device described in (d).f-h) Extracted μ 0 , R sd , and V T of Bi, Sb, and Alloy contacted MoS 2 transistor at different annealing temperature using the extraction method described in (d) and (e).