Characterization of photonic devices by secondary electron potential contrast

The Mechanistic Determination of Doping Contrast from Fermi Level Pinned Surfaces in the Scanning Electron Microscope Using Energy-Filtered Imaging and Calculated Potential Distributions

“…For the freshly cleaved specimen, p was brighter than n under standard imaging conditions (Fig. 1b), as expected (Oatley & Everhart, 1957; Perovic et al, 1995, 1998; Venables et al, 1998; Elliott et al, 2000; Sealy et al, 2000; Elliott, 2001; Elliott et al, 2002; Lin & Lee, 2003; El-Gomati et al, 2004; Buzzo et al, 2006, 2007; Chee et al, 2008 a , 2008 b ; Chee et al, 2010, 2011; Liu & Lee, 2011 a , 2011 b ; Chung et al, 2011; Farrell et al, 2013; Xu et al, 2013; Druckmüllerová et al, 2014; Chee, 2016 a ; Alugubelli et al, 2019; Chee, 2019, 2020). Moreover, surface treatment using fresh (degraded) NH 4 F (aq.)…”

“…For HF-based treated specimens, conventional wisdom ascribed the enhanced contrast to an attendant patch field increase for the reason that the hydrogen atoms saturate the electrically active dangling bonds and remove a cohort of surface states within the fundamental bulk band gap, thereby unpinning the Fermi level and reducing surface band-bending (Elliott et al, 2000;Sealy et al, 2000;Elliott, 2001;Müllerová et al, 2002;Lin & Lee, 2003). This classical depiction of Fermi level depinning relating to a high structural perfection of the surface was partly because of the widely held belief that the doping contrast is not conceivable without patch fields (Oatley & Everhart, 1957;Elliott et al, 2000;Sealy et al, 2000;Elliott, 2001;Elliott et al, 2002;Lin & Lee, 2003;Buzzo et al, 2006Buzzo et al, , 2007Liu & Lee, 2011a, 2011bChung et al, 2011), even despite energy filtering (at least within the linear regime of the detector response function (DRF)) (Kazemian et al, 2006(Kazemian et al, , 2007. An increase in the surface band-bending and surface state density was therefore effectively ruled out a priori.…”

“…Doped layers as thin as 1 nm can be imaged (Elliott et al, 2002; Chee et al, 2008 b ), and doping contrast is proportionate to dopant concentration from sub-10 14 through 10 20 cm −3 , at a quantification accuracy of at least ± 3% (Chee, 2016 a ). Analogous to physical and failure analysis (PFA) of integrated circuits or critical dimension measurements, dopant profiling has been demonstrated in AlGaAs/GaAs-based vertical-cavity surface-emitting lasers (Buzzo et al, 2006), InGaAs/InP multi-quantum well heterostructures (Buzzo et al, 2006), AlGaAs/GaAs or InGaN/GaN light-emitting diodes (Chung et al, 2011; Farrell et al, 2013), laser-doped Si solar cells (Xu et al, 2013), static random access memory chips (Liu & Lee, 2011 a ), Si laterally double-diffused p -channel metal-oxide–semiconductor field-effect transistor (MOSFET)-based current sources (Liu & Lee, 2011 a ), SiC junction field-effect transistors and power MOSFETs (Buzzo et al, 2007), high-power GaN p – i – n diode structures (Alugubelli et al, 2019), etc . Notwithstanding its practical application to a vast gamut of (opto)electronic or power device architectures, the rich tapestry of results is heretofore largely qualitative while harboring major sensitivity bottlenecks, and any existing quantification, albeit also stymied by sizable errors, is no more than for p -type material.…”