自身氧化之二硫化鉬的電性傳輸行為與記憶體特性

Abstract

本實驗元件以機械剝離法、電子束微影、熱蒸鍍、高溫熱退火等製程，將二硫化鉬製作成場效電晶體元件，並使用探針系統量測其電性表現。在室溫下二硫化鉬電晶體為n-type表現與電極間為良好的歐姆接觸，接著將二硫化鉬元件放置在通有O3氣體環境下，在經氧化製程後二硫化鉬將在表面自身氧化成MoOx (x≤3)，並藉由調控不同的氧化參數，可控制元件的表現從n-type變成ambipolar或p-type，元件的電流開關比可高達104倍，產生此現象的原因與MoOx具有高功函數的特性，造成電洞注入有關。氧化後，在電極與二硫化鉬接面產生蕭基位障，使其成為背對背蕭基二極體 (back to back Schottky diodes)，並且能與背對背蕭基二極體理論擬合，亦可以藉由汲極-源極電壓與背向閘極電壓來調控位障大小。 進一步分析二硫化鉬氧化前後的傳輸機制，分為三種不同特性來說明，元件為n-type行為的傳輸機制：80 K到180 K區間符合二維變程跳躍理論，180 K到240 K區間符合熱活化傳輸。元件為ambipolar行為的電子傳輸機制：80 K到180 K區間符合二維變程跳躍理論，180 K到240 K區間符合熱活化傳輸，而電洞傳輸機制：280 K到220 K區間符合二維變程跳躍理論，220 K以下由tunneling機制主導，元件為p-type的行為的傳輸機制：200 K到120 K區間符合二維變程跳躍理論。 另外發現二硫化鉬元件氧化後呈p-type表現時，可經由寫入電壓來改變元件的狀態，使其具有記憶體特性。除此之外，控制寫入的電壓條件，可觀察出多種可分辨的電流狀態，並且保持穩定的電流。若將此元件做為記憶體應用，除了能長時間儲存狀態外，元件的耐用度亦為相當重要的指標，經過循環迴圈測試後，電流具有相當好的穩定性與判別性。

Mechanically exfoliated molybdenum disulfide (MoS2) flakes are dispersed on silicon substrates capped with a 300-nm thick silicon dioxide layer. The standard electron beam lithography and thermal evaporation were used to pattern Au leads on the MoS2 flakes. The as-patterned MoS2 field effect transistors (FETs) were then annealed in a high vacuum to reduce the contact resistance. The MoS2 FETs demonstrates n-type semiconductor and possess good ohmic contact behaviors. The MoS2 FETs were exposed to ozone gas for self-limiting oxidation and the surface layer was converted to MoOx (x≤3). By controlling the ozone treatment, the FET devices changed from its natively n-type to an either ambipolar or a p-type semiconductor. The on-off ratio of transistors is up to 104. The variation of electron or hole dominated transport is owing to the change of the work function after oxidation. The oxidized MoOx has higher work function and gives efficient hole injections. Current-voltage (Ids -Vds) curves showed back-to-back Schottky diodes, which implies that the Schottky junction is formed between Au electrodes and MoOx. The Schottky barrier was inspected, showing a variation with bias and gating voltages. In the following discussion, we categorized oxidized MoS2 FETs into n-type, ambipolar, and p-type FETs. Electron transport of n-type FETs (MoS2) is well described by the theory of two-dimensional variable range hopping at temperatures in the range between 80 and 180 K. At higher temperatures from 180 to 240 K, it can be fitted using thermally activated transport. Nevertheless, electron transport of ambipolar FETs (mildly oxidized MoS2) can be separated into two regions of either electron or hole doping. The electron doping regime is well described by the theory of two-dimensional variable range hopping at temperatures from 80 to 180 K whereas, at higher temperatures from 180 to 240 K, it is better described by thermally activated transport. The hole doping regime is well described by the theory of two-dimensional variable range hopping from 220 to 280 K. At temperatures lower than 220 K, electron transport is dominated by tunneling. The hole transport of p-type FETs is well described by the theory of two-dimensional variable range hopping at temperatures from 120 to 200 K. On the other hand, we discovered a memory effect on p-type FETs. The state of the p-type FETs can be changed by a writing voltage. If the writing current are precisely controlled, the devices can be operated at multiple states and the states can be stored for a long time. After a simple reliability test, we observe reproducible and stable current states in the memory devices.