Anomalous band alignment change of SiO2/4H–SiC (0001) and (000–1) MOS capacitors induced by NO-POA and its possible origin

Abstract

For SiO2/4H–SiC (0001) and (000–1) n-type metal-oxide-semiconductor capacitors, the relationship between flatband voltage and the thickness of oxide was investigated after NO post-oxidation annealing to evaluate the expected flatband voltage (VFB) without a fixed charge effect. After removal of the fixed charge effect, there was an anomalous negative shift of VFB on (0001) 4H–SiC, which would be attributed to the result of dipole layer formation at the interface. The effects of the dipoles were investigated from the perspective of the SiO2/4H–SiC band alignment shift. We could find the correlation between the shift of VFB and that of the band alignment between SiO2 and the 4H–SiC interface; we concluded that stable Si–N bonds at the interface induce a dipole layer, and this is one of the reasons for the unexpected shift reported for VFB or threshold voltage of metal-oxide-semiconductor field-effect transistors, as well as the fixed charge effects. A significant difference in the band alignment on (0001) and (000–1) was also clarified, which would be one of the reasons for the disagreement of VFB on those faces.

Nitrogen‐Induced Changes in the Electronic and Structural Properties of 4H‐SiC (0001)/SiO2 Interfaces

“…[55] The bandgap energies for 4H-SiC (E SiC g ¼ 3.2 eV) and SiO 2 (E SiO 2 g ¼ 8.3 eV) are determined by choosing the optimum mixing parameter (0.25 for SiC and 0.35 for SiO 2 [52] ) in the HSE calculations. The resulting band offsets for our systems show values comparable to previous experimental results, [20,56] but no definite change under nitridation (Table 1). The most stable case (model 4) exhibits a slight increase in ΔE C , while model 3 presents reduced barrier heights in closer agreement with experiments.…”

“…The most stable case (model 4) exhibits a slight increase in ΔE C , while model 3 presents reduced barrier heights in closer agreement with experiments. [20,56] To assess the correlation between the composition and geometric structure of interfaces, we analyze strain fluctuations in both the SiC and the SiO 2 slabs. For the ith Si atom, we compute the average bond strain as follows.…”

“…The nitrogen reference is chosen because nitridation is produced through N 2 or NO during POA processes. [1,2,20,21] Additionally, other reference chemical potential choices (e.g., N and O) that could impact our results are discussed in the Supplementary Material. [41]…”

“…[1,2] Easily grown interfaces with the SiC native oxide, SiO 2 , can present a high density of interface states that largely degrade the performance of power devices by reducing carrier mobility and changing the threshold voltage. [3][4][5][6] The improved carrier mobilities after post-oxidation annealing (POA) of 4H-SiC/SiO 2 interfaces in N-rich conditions (N 2 , NO, or N 2 O) [1,2,[7][8][9][10][11][12][13][14][15][16][17][18][19][20][21] or combined oxidation/annealing [22,23] have led to the adoption of the nitridation process in the production of SiC power electronics applications. The underlying mechanism of this process, however, is still elusive despite the considerable efforts made to date in modeling these interfaces, [24][25][26][27][28][29] with suggested mechanisms ranging from the healing of dangling bonds to the removal of interstitial C atoms.…”

“…More recent models, incorporating nitrogen amounts associated with device processes, have been limited to ultra-thin SiO 2 layers. [33] Yet, theoretical atom-level descriptions are needed to establish a correlation between nitridation and physical properties of these interfaces as reported in recent experiments [20,40] and envision novel mechanisms to systematically improve properties arising from chemical treatments.…”