Reduced self-heating in Si/SiGe field-effect transistors on thin virtual substrates prepared by low-energy plasma-enhanced chemical vapor deposition

Abstract

High hole mobility in Si 0.17 Ge 0.83 channel metal-oxide-semiconductor field-effect transistors grown by plasma-enhanced chemical vapor deposition

Thin relaxed SiGe virtual substrates grown by low-energy plasma-enhanced chemical vapor deposition

“…Details are presented elsewhere [11,23]. The obtained transit frequencies f T and maximum oscillation frequencies f max are shown in Table 3.…”

“…Some results indicating the promise of the new thin buffer design are published in Refs. [11,23]. Here we focus on the structure of the buffers themselves.…”

The impact of self-heating and SiGe strain-relaxed buffer thickness on the analog performance of strained Si nMOSFETs

“…Thicknesses below 100 nm were achieved for SiGe SRBs with Ge percentages as high as 70% [13]. 500 nm thick Si 0.56 Ge 0.44 SRBs prepared by low energy plasma enhanced chemical vapour deposition (PECVD) with acceptable defect density magnitudes were reported but were used in fabricating modulation doped field effect transistors (MODFETs) [20]. Olsen et al used 200 nm SiGe SRBs to demonstrate that some of the performance enhancement lost to self-heating at high terminal voltages in thick SRB devices can be recovered by scaling the buffer thickness [10].…”

“…Conventional graded buffers have thicknesses that are normally on the order of one or several micrometers thick however literature shows that progress has been made on the reduction of buffer thicknesses. Some studies on thin SiGe SRB technology focus on materials growth and characterization [12][13][14] whereas others focus on self-heating in devices [10,11,[15][16][17][18][19][20][21]. The introduction of point defects during the growth of the SRB has been demonstrated as a method of fabricating thin SiGe SRBs [13].…”

High mobility holes in a strained Ge quantum well grown on a thin and relaxed Si0.4Ge0.6/LT-Si0.4Ge0.6/Si(001) virtual substrate

“…4 This could lead to overheating of the material and result in degradation of nanoscale device performance, as demonstrated in an n-channel strained Si MODFET on a Si 0.56 Ge 0.44 buffer. 5 Therefore, a significant reduction in the SiGe buffer layer thickness is indeed required. Furthermore, thin SiGe buffers will have advantages for integration with other devices on a single chip.…”