999
(Invited) Thermodynamic Modeling of W-C-O-H-S System for Controlled Growth of WS2 Atomic Layers by True CVD

Tuesday, 30 May 2017: 10:00
Norwich (Hilton New Orleans Riverside)
S. Dhar, A. Lalithambika, K. Kumar V, and S. Raghavan (Indian Institute of Science)
Objective: Layered TMDs such as MX2 (M = Mo, W; X = S, Se) are candidate materials for future electronic and optoelectronic devices. For integration of TMD-based devices into existing microelectronics technology, a CVD technique that allows controlled deposition of uniform layers over large areas is required. Further, to achieve layered heterostructures and doping, it is essential to develop a “true” CVD method in which all the precursors are vaporised outside the reactor, and enter the reactor with specific flows controlled by MFC’s. CVD growth of TMDs (WS2 growth is demonstrated here) is dictated by variation in process parameters like total reactor pressure P, growth temperature T and gas flow rate. Variation of these parameters over a wide range may lead to contamination in the form of carbon, carbides, sulfur, other sulfides, oxides and metal in different combinations. Uncontrolled compositional disturbances are extremely undesirable for electronic devices as they result in inconsistent background. Thermodynamic equilibrium calculations become essential at this point to select optimum operating conditions for contamination-free growth. Having identified the optimum growth window, gas phase supersaturation needs to be controlled such that layer controlled pure WS2 can be synthesized. The results obtained from thermodynamic analysis may then be used to design CVD processes, assuming that equilibrium prevails, an assumption valid when the rate of deposition is low. Computations on equilibrium compositions of the solid and gaseous species formed under various CVD conditions lead to the construction of “CVD phase stability diagrams”, to identify appropriate process window for growth of contamination-free WS2. To the best of our knowledge, there is no previous report for such modelling on this system. The objective is to construct this CVD diagram and compare it with experimental observations for the W-C-O-H-S system for controlled growth of WS2 layers.

New Results: Equilibrium thermodynamic modeling of CVD, using W(CO)6 and H2S as precursors in Ar, Ar/H2 and H2 ambient, was investigated by minimizing the total Gibbs free energy in the W-C-O-H-S system. The ranges of CVD parameters used for the present calculation were: T=100-1100°C, P=10-900 Torr, and flow rate ratio W(CO)6:H2S:(Ar and/or H2) =1:(2-50):(100-1000). The ability of these phase diagrams in predicting growth outcomes is corroborated by the Raman spectroscopy data, XRD and XPS from samples deposited over a range of process parameters.

A comprehensive CVD phase stability diagram for formation of solid phases in Ar ambient is demonstrated in Fig 1(a). It predicts the stability window for pure WS2 and indicates co-deposition of WS3 and C at lower T. Quantitative carbon contamination of WS2 films for wide variation in T and P is depicted in a contour plot in Fig. 1(b). Increasing carbon contamination of WS2 at lower T and higher P is validated by Raman spectra depicted in Fig. 1(c). This makes Ar unsuitable as carrier gas for growth, as higher P is required to achieve the reduction in supersaturation that is critical to obtain uniform layers of desired thickness.

Modeling predicts, however, that the deposition of carbon can be precluded by increasing the partial pressure of hydrogen, i.e., by using a mixture of Ar and H2 as the carrier gas, as shown in Fig 1(d). In Ar/H2 and in H2 ambient, the stability domain of carbon narrows as P is increased, meeting the criterion for reduction in supersaturation. Pure H2 as a carrier gas is shown to facilitate the largest contamination free process window. CVD phase stability diagram in H2 ambient, predicting the ranges of CVD conditions for growth of contamination-free WS2, is illustrated in Fig. 1(e). WC and W deposit in various combinations for process parameters beyond this range. Raman spectroscopic data [Fig. 1(f)] showing carbon-free WS2 for wide ranges of P and T validates the phase diagrams in H2 ambient. Given the multivariable nature of the problem, excellent agreement between theoretical and experimental results was observed.

Conclusion: A generic approach suitable to all TMDs, involving a combination of thermodynamic modeling and kinetic control for desired growth, is explored for the system W-C-O-H-S. Thermodynamic equilibrium calculations serve as a first step to determine process parameter windows for the desired end product. The ability to predict the completely different outcomes on using Ar versus Ar + H2 mixture as the carrier gas highlights the importance of such thermodynamic modeling. Thus, theoretical analysis of the CVD process can be employed to synthesize desired films in a predictive manner. This reduces the effort and resources needed for process development, and aims to bring a systemic approach to hitherto ad-hoc process.