下閘極氧化鋅鎂薄膜電晶體之電穩定性研究

Abstract

本論文主要探討兩種退火條件下氧化鋅鎂下閘極薄膜電晶體(thin-film transistor)在不同溫度下之閘極偏壓穩定性。由文獻中得知，在氧化鋅中添加鎂可以減少氧空缺的產生，進而增加材料的電穩定性。另一方面，銦在地球的存量相對稀少，因此本論文針對無銦的氧化鋅鎂薄膜電晶體進行研究。 X光繞射(XRD)分析氧化鋅鎂的單層膜發現200°C退火的薄膜只有些許的峰值偏移，而350°C退火的薄膜有較大的峰值偏移，並且也有較小的半高全寬(full width at half-maximum)，表示在350°C的退火條件下有較多鎂取代鋅與較佳的結晶性。 接著於變溫量測中，200°C退火的電晶體隨溫度升高有比較不一致的電性變化，而350°C退火的電晶體則相當一致。在常溫正偏壓穩定性測試中，兩者皆沒有明顯的次臨界擺幅(subthreshold swing)變化，顯示臨界電壓(threshold voltage)的偏移機制為電荷捕陷(charge trapping)，而350°C退火的電晶體有比較少的臨界電壓偏移，顯示350°C退火的電晶體穩定性較佳。於變溫的正偏壓測試中，隨著環境溫度升高以及偏壓時間增加，200°C退火的電晶體轉換特性曲線(transfer characteristic curve)出現駝峰(hump)現象，此一現象在較高溫的350°C退火的薄膜電晶體則沒有發現。而此一駝峰現象在外加閘極偏壓移徐後，會完全回復。因此被認為是由閘極偏壓所造成之介穩態現象。這一個電晶體轉換特性曲線次臨界區域的駝峰現象，機制尚不完全明朗。一般認為和閘極偏壓導致的介穩態(meta-stable)缺陷有關。在較高溫進行閘極偏壓測試時，所造成的介穩態中性氧空缺可能激發成正一價或正二價的氧空缺，所釋放的電子在通道層形成漏電通道(leakage path)，而造成電晶體的早期導通狀態，因此在電晶體轉換特性曲線之次臨界區域造成駝峰現象。 在負偏壓下並沒有駝峰的現象出現，然而，於高溫負偏壓測試時，200°C退火的臨界電壓變化有折返(turn-around)現象，主要由缺陷產生(defect creation)機制與電荷捕陷互相競爭造成。相較上述的直流偏壓，交流偏壓測試有較小的臨界電壓偏移，但是隨著偏壓訊號的頻率增加，臨界電壓偏移量也隨著增加，此一現象可能是由於被捕陷的載子回復較慢導致。 總體而言，350°C退火的氧化鋅鎂薄膜電晶體在各方面的穩定性測試中，皆相較於200°C退火的電晶體來的穩定，這可能是由於較高溫退火電晶體的通道層結晶中有較多鎂的取代鋅以及較好的氧化鋅鎂結晶品質所導致。

This thesis reports the experimental studies on the gate-bias temperature stability of inverted staggered bottom-gate Mg0.05Zn0.95O thin-film transistors (TFT). According to literatures, the addition of Mg into ZnO related materials can reduce the oxygen vacancies due to higher ionic character of Mg-O than Zn-O bonds. In addition, indium is rare in earth. Therefore, we chose indium free MgZnO TFTs as our research target. ZnO films crystallize easily, even when grown at room temperature. With the application of post-deposition annealing at 200°C, the (002) peak increased slightly, indicating the substitution of a small portion of Mg for Zn in the ZnO crystals. This substitution reduced the lattice constant of wurtzite ZnO, caused by the slightly smaller ionic radius of Mg2+ than that of Zn2+. Moreover, as the annealing temperature increased to 350°C, grains grew and the crystallinity of Mg0.05Zn0.95O improved, as denoted by a decrease of full width at half maximum (FWHM) of (002) peak. Furthermore, (002) the peak shifted even higher, indicating the substitution of more Zn by Mg in the ZnO crystals. In the positive gate-bias stability test at room temperature, the subthreshold swing was nearly unchanged for the devices annealed at two different annealing conditions, revealing that the main mechanism for the threshold voltage (Vth) shift was charge trapping. The 350°C-annealed TFT showed less Vth shift, indicating better device stability. As the positive gate-bias stress applied to TFTs at elevated temperatures, humps occurred in the subthreshold region of the transfer curves in the 200°C-annealed TFT, and became severe as temperature and stressing time increases. The hump phenomenon was much less significant in 350°C annealed TFTs; merely a degradation of SS was observed at 80°C, the highest testing temperature in this study. The hump disappeared shortly after removing the positive gate-bias, suggesting that this phenomenon was meta-stable and was resulted from gate-bias induced electric field. This hump phenomenon might have been due to the creation of meta-stable oxygen vacancies in which the neutral vacancies were thermally excited into ionized states and released electrons into the active layer to form a leakage path, when TFTs were subjected to gate-bias stressing at elevated temperatures. The humps were not identified in the transfer curves when TFTs were subjected to negative bias stress. Instead, a turn-around of Vth shift occurred in the 200°C-annealed TFT. It was attributed to the competing mechanisms of the defect creation and the charge trapping. In the AC bias stress, Vth shift was less severe to the DC bias stress. Nevertheless, the Vth shift increased with the increasing frequency of AC bias stress. It may come from the slow recovery of trapped charges. In conclusion, the 350°C-annealed TFT showed a better bias temperature stability. The more substitution of Zn by Mg and better crystallinity help improve the stability of MgZnO TFTs.