High, Low, and Zero Field Spin Dependent Recombination in 4H SiC Metal Oxide Semiconductor Field Effect and Bipolar Junction Transistors

Silicon carbide metal oxide semiconductor field effect transistors (MOSFETs) have great promise in high power and high temperature applications.  Among the many polytypes, 4H is the most promising for MOSFET applications. Unfortunately, the effective channel mobility in these devices remains disappointingly low and the devices frequently exhibit as yet poorly understood bias temperature instabilities. The underlying causes of the low effective channel mobility and the bias temperature instabilities are as yet not fully understood but clearly involve the presence of electrically active point defects very close to the silicon carbide-silicon dioxide interface. 

Electron paramagnetic resonance (EPR) has unrivaled sensitivity and analytical power in the identification of point defects in semiconductors and insulators. However, the detection limit of conventional EPR in solids is about ten billion defects under near ideal circumstances. Thus, conventional EPR lacks the sensitivity to detect these defects in a transistor even remotely approaching technological relevance. The electrically detected magnetic resonance (EDMR) technique called spin dependent recombination (SDR) provides a sensitivity at least ten million times higher than that of conventional EPR. SDR/EDMR can also provide information about the physical location and energy levels of defects, with the additional advantage that it is readily made sensitive to only those defects which actually affect the performance of the device under observation.  

In this study, we utilize SDR/EDMR measurements carried out over a wide range of magnetic fields,  frequencies, and temperatures utilizing multiple electronic measurement platforms (spin dependent charge pumping, spin dependent gated diode recombination current measurements, and bipolar amplification effect spin dependent recombination measurements) to identify defects on both sides of the silicon carbide-silicon dioxide interface. Our measurements identify multiple defect centers, some in the silicon dioxide near the interface, others in the silicon carbide near the interface. The silicon dioxide centers include simple oxygen vacancy centers and hydrogen complexed oxygen vacancy centers. The silicon carbide centers include silicon vacancies and, as yet only partially identified hydrogen complexed defects.  Of particular interest are defect centers which are either created or activated by bias temperature stressing.

We show that the presence of one near interface silicon carbide defect is strongly correlated with the interface trap density in as processed devices, the silicon vacancy. We also show that the amlitudes of several SDR/EDMR spectra, those associated with oxygen vacancy centers in the oxide and at least one hydrogen complexed center are correlated with bias temperature instability response.

We believe our work will be useful to SiC device process designers in particular. We also believe that the work will be of interest to a broader audience, including   those involved in research dealing with other wide band gap semiconductors  because the novel low to near zero field SDR measurements we utilize can likely be readily adapted to devices based upon  virtually any  wide band gap material.