基于離子插層的可編程電阻已成為模擬深度學習應用的潛在下一代技術。質子，作為最小的離子，是一個非常有前途的候選者，可以使器件具有高調制速度、低能耗和增強的耐久性。

      美國麻省理工學院-IBM Watson人工智能實驗室李巨教授團隊發(fā)表在NANO LETTER的一篇題目為“基于磷硅酸鹽玻璃電解質的CMOS兼容質子可編程電阻用于模擬深度學習”的文章（https://pubs.acs.org/doi/full/10.1021/acs.nanolett.1c01614），報告了第一個后端CMOS兼容的非易失性質子可編程電阻，該電阻通過整合磷硅酸鹽玻璃（PSG）作為質子固體電解質層而實現(xiàn)。PSG是一種出色的固體電解質材料，它同時顯示出優(yōu)異的質子傳導和電子絕緣特性。此外，它是傳統(tǒng)硅制造中的一種知名材料，能夠實現(xiàn)精確的沉積控制和可擴展性。規(guī)?；娜虘B(tài)三端設備在對稱性、保持性、持久性和能源效率方面顯示出理想的調制特性。因此，基于磷硅酸鹽玻璃的質子可編程電阻代表了實現(xiàn)單片CMOS集成的納米級模擬交叉棒處理器的有希望的候選者。

      這篇開創(chuàng)性的研究論文中關于CMOS兼容質子可編程電阻的電學表征工作均基于韓國NEXTRON公司變溫微探針系統(tǒng)MPS-PT通過和美國Keysight公司的B1500半導體參數(shù)分析儀連用進行測試（All electrical measurements were performed using a micro-probe station enclosed chamber (MPS-PT) manufactured by NEXTRON, Korea. In experiments performed in forming gas(FG, 3% H2 in N2) all 4 DC probes where first connected to the pads of the device under test(1 to Source, 1 to Drain and 2 to Gate terminals). The chamber was then purged by flowing in FG for 60 s with both gas inlet and outlet open, followed by another 60 s with outlet shut to create positive pressure inside the chamber.
Three of the probes (1 Source, 1 Drain and 1st Gate) were then connected to the Source Measurement Units (SMUs) of a Keysight B1500 Semiconductor Analyzer, while the fourth probe (2nd Gate) was connected to the Pulse Generation Unit (PGU) of the same instrument. The experiment sequence and data acquisition were controlled via an in-house developed MATLAB suite.）