源極與汲極工程對磷化銦通道金氧半場效電晶體特性影響之研究

Abstract

在本篇論文中，第一部份我們製作了二氧化鋯/三氧化二鋁/磷化銦基板之金氧半電容。並加入氧氣電漿處理比較氧氣電漿對於二氧化鋯/三氧化二鋁/磷化銦基板金氧半電容特性的影響，並且針對不同溫度的氮氣環境後沉積退火來對退火溫度條件做最佳化。並且，我們還搭配高頻法來萃取金氧半電容的介面缺陷密度，對於未加入氧氣電漿處理的二氧化鋯/三氧化二鋁/磷化銦金氧半電容，聚積區域的頻率分散現象非常嚴重，利用高頻法萃取出來的介面缺陷密度最小值大約(3×10E12 eV-1cm-2)；而對於加入氧氣電漿處理的二氧化鋯/三氧化二鋁/磷化銦金氧半電容，聚積區域的頻率分散現象有很明顯的改善。但是介面缺陷密度最小值上升至(5×10E12 eV-1cm-2)，接著以XPS分析，我們推測In2O3的生成會使介面品質劣化導致缺陷密度上升，而經過氧氣電漿處理的金氧半電容在In2O3的比例上有些微增加的趨勢，造成介面缺陷密度些微上升。 接著，我們以矽摻雜於P型磷化銦基板並以氮化鈦金屬做為活化退火之覆蓋層，利用快速熱退火750度15秒完成PN接面二極體，在不同氮化鈦覆蓋厚度條件下，其中以覆蓋氮化鈦2000Å的PN接面二極體擁有最佳電性，包含理想因子(1.74)及接面串聯電阻(208Ω)，接著我們利用傳輸線模型來萃取未覆蓋氮化鈦及覆蓋2000Å氮化鈦這兩條件下的特徵接觸阻值及片電阻，發現活化退火過程以2000Å氮化鈦覆蓋，其特徵接觸阻值相較未覆蓋氮化鈦的條件有明顯的降低( )，而片電阻對於兩種條件並無明顯差異(74.27Ω/□)，表示矽摻雜的活化程度是接近的。雖然XRD分析並沒有明顯觀察到氮化鈦厚度與分析結果之相關性。而由XPS縱深分析可以發現當氮化鈦厚度增加，氮元素往磷化銦基板擴散的程度越為明顯，所以可初步認為接面串聯電阻的降低與氮元素的擴散相關。接著，利用前述的PN接面二極體製作流程，我們成功地製作出磷化銦金氧半場效電晶體，而我們使用二氧化鋯/三氧化二鋁作為元件的閘極氧化層，源極和汲極區域使用矽摻雜並以氮化鈦作為覆蓋層。此元件具備電流開關比(6.39×105)及次臨界擺幅(238 mV/dec)。我們比較了不同氮化鈦覆蓋厚度的元件電性，得到在使用2000Å氮化鈦的條件下，在驅動電流方面相較500Å氮化鈦覆蓋高了8.3倍。 接著，為了降低等效氧化層厚度，我們改變二氧化鋯/三氧化二鋁的厚度，同樣地，可成功製作出磷化銦金氧半場效電晶體，得到電流開關比(1.84×10E4)和次臨界擺幅(200 mV/dec)。接著我們比較改變等效氧化層厚度所製作的兩種元件特性，由於等效氧化層厚度的降低，使得最高驅動電流上升3.25倍，但是由於等效氧化層厚度的降低，也會使閘極漏電流增加，而降低電流開關比。 最後，為了降低源極和汲極區域電阻，我們以快速熱退火250度60秒製作鎳磷化銦合金，此鎳磷化銦蕭特基接面具備理想因子(1.42)和電流開關比(6.37×10E4)，電子蕭特基能障大約(0.53eV)，而金氧半場效電晶體擁有高電流開關比(1.29×10E5)和不錯的次臨界擺幅(152 mV/dec)，最後我們比較矽摻雜磷化銦及鎳磷化銦合金的片電阻，使用鎳磷化銦合金得到相對低的片電阻，而從金氧半場效電晶體同樣萃取到相對低的源極/汲極區域電阻。

In the thesis, we fabricated the ZrO2/Al2O3/InP MOS capacitors and compared the characteristics of ZrO2/Al2O3/InP MOSCAPs w/ or w/o O2 plasma treatments. Besides, we compared the characteristics of various PDA treatment temperatures which were completed in N2 ambient. Moreover, the Dit was extracted by terman method. For ZrO2/Al2O3/InP MOSCAPs w/o O2 plasma treatment, the frequency dispersion in accumulation region was severely degraded, and the Dit minimum was about 3×10E12 eV-1cm-2 at PDA 300℃; for the ZrO2/Al2O3/InP MOSCAPs w/ O2 plasma treatment, the frequency dispersion in accumulation region was obviously improved. However, the Dit minimum was about 5×10E12 eV-1cm-2. From the XPS analysis, we speculated that the interface quality was degraded due to the In2O3 state generation. For the MOSCAPs w/ O2 plasma treatment, the In2O3 ratio was increased slightly and led to higher Dit. Following, the p-type InP was selectively implanted with Si dose of 2×10E14 cm-2 at 80keV; then TiN with various thicknesses was deposited as encapsulation layer, and the PN junction was completed by RTA at 750℃ for 15s. Among the TiN thicknesses conditions, the PN junction with 2000Å TiN encapsulation exhibited the best characteristics, including the lowest ideality factor of 1.74 and series resistance of 208Ω. After that, the TLM stuctures were fabricated to extract the specific contact resistances and sheet resistances, the condition with TiN 2000Å encapsulation exhibited the specific contact resistances of 3.49×10E-6 Ω∙cm2, which was considerably lower than the condition without TiN encapsulation. On the other hand, the sheet resistances were almost equal values due to the same implantation conditions. From the XRD analysis, we could not relate the TiN thicknesses to the XRD results. From the XPS depth profiles, the N atoms diffused into InP substrate with TiN encapsulation, so we speculated that the contact resistance was improved due to the N atoms diffusion. Next, the InP MOSFETs were fabricated using ZrO2/Al2O3 as gate oxide; source and drain were selectively implanted with Si dose and deposited TiN as encapsulations; the dopant activation was completed by RTA at 750℃. The device exhibited Ion/Ioff ratio of 6.39×105 and subthreshold swing of 238mV/dec. We compared the devices with various TiN encapsulation thicknesses, and the on current with TiN 2000Å was 8.3 times higher than the on current with TiN 500Å. Subsequently, in order to scale down equivalent oxide thickness, the ZrO2/Al2O3 gate oxide thickness was adjusted. The device exhibited Ion/Ioff ratio of 1.84×10E4 and subthreshold swing of 200mV/dec. The driving current was 3.25 times higher but the off-state current was also increased due to the reduction of EOT. Finally, in order to reduce the RSD, the Ni-InP/InP Schottky junction was completed with RTA 250 ℃ for 60s. The Schottky junction exhibited Ion/Ioff ratio of 6.37×10E4 and ideality factor of 1.42. The device exhibited Ion/Ioff ratio of 1.29×10E5 and subthreshold swing of 152 mV/dec. Moreover, the Ni-InP MOSFETs exhibited lower RSD than N+-InP MOSFETs due to the lower sheet resistance of Ni-InP.