Half-filled orbital and unconventional geometry of a common dopant in Si(001)

The determining factor of the bulk properties of doped Si is the column rather than the row in the periodic table from which the dopants are drawn. It is unknown whether the basic properties of dopants at surfaces and interfaces, steadily growing in importance as microelectronic devices shrink, are also solely governed by their column of origin. The common light impurity P replaces individual Si atoms and maintains the integrity of the dimer superstructure of the Si(001) surface, but loses its valence electrons to surface states. Here we report that isolated heavy dopants are entirely different: Bi atoms form pairs with Si vacancies, retain their electrons, and have highly localized, half-filled orbitals.

Figure 1

Structure of Bi-Si vacancy complex in the Si(001) surface. (a),(b) Bias-dependent topographic images of Bi donor in the Si(001) surface at sample-bias voltages of V${}_s$ $=$ $+$2 V (empty state) and $-$2 V (filled state), respectively. The tunneling current was I${}_t$ $\sim$ 100 pA. Bi sites are indicated by markers. The color scale is in units of pm. The horizontal scale is indicated in (a) by the spacing between Si-Si dimer rows ($\sim$0.77 nm). (c) Schematic structure of Bi-Si vacancy complex in the Si(001) surface (top-, side-view). The neutral Bi atom has a half-filled $p$ orbital perpendicular to the Si dimer row, and at the Si vacancy site the second-layer Si atoms bond across the vacancy as shown by a dotted line.

Figure 2

(a) Typical topographic image of Si(001):Bi at ${\text{V}}_S=-1$ V. Scan size is 40 nm $\times$ 40 nm. Bi donors are indicated by arrows. (b) Flashing effect on the density of Bi dopants in STM image. The flashing was performed at 1100 ${}^{\circ }$C for 10 sec. The sample was exposed to air after 20th flashing intentionally, and flashed again in UHV condition (blue square). For comparison, the horizontal dashed line represents the Bi concentration estimated from the bulk density ($\sim 3\times {10}^{11}$ cm${}^{-2}$).

Figure 3

Comparison between STM/STS and spin-polarized DFT. (a) STM images (top) and simulated STM (bottom) of the Bi donor (plus sign). STM images were taken at exactly the same location; simulated images used isosurface values of 0.1 electron/nm${}^3$. (b) dI/dV maps (top) and simulated DOS maps around Bi-Si vacancy complex (bottom). Feedback set-point was $\sim$100 pA, ${\text{V}}_S=-2$ V. The spacing between Si-Si dimer rows ($\sim$0.77 nm) is shown in all images. The units of color scale in the dI/dV map and DOS map are pS and electron/Å${}^3$.

Figure 4

Schematic structures of arrangements of Bi and vacancies on Si(001). “1DV” is the well-known single Si dimer vacancy, and “V” is a single Si vacancy.

Figure 5

Comparison between STS and both non-spin-polarized and spin-polarized DFT simulated density of states (DOS) of Bi-Si vacancy complex. (a) Differential conductance dI/dV at Bi donor and Si dimer far from Bi. Characteristic peaks are indicated by arrows and numbers. (b) Comparison of peak bias-voltages between dI/dV and spin-polarized calculated DOS. (c) DOS of Bi atom calculated by non-spin-polarized DFT. (d) Simulated DOS of Si far from Bi by non-spin-polarized DFT. (e) Spin-polarized DOS of Bi atom embedded in the supercell shown in Fig. 1c. The calculated exchange energy is $\sim$0.2 eV. Peaks are indicated with the same colors and numbers as for dI/dV. (f) Spin-polarized DOS of Si far away from Bi.

Figure 6

Contours of constant spin density for broken-symmetry AFM configurations (left column) and FM configurations (right column) of Bi-V pairs on Si(001), computed using hybrid DFT. The dimer rows run from left to right; red and blue isosurfaces show positive and negative spin densities, respectively. (a),(b): spacing of four dimers. (c),(d): spacing of six dimers. (e),(f): spacing of eight dimers.

Figure 7

Band structure of bulk silicon, Si(001), and Bi-V feature along $\Gamma$-X (100) direction in the conventional cubic cell.

Figure 8

Simulated integrated DOS images and DOS of Bi-Si and P-Si hetero-dimers in Si(001), with the Bi and P atoms replacing the upper atom of a Si dimer in c(4$\times$ 2) reconstruction. (a) Left: Schematic structure of Bi-Si hetero-dimer. The Bi atom has a neutral configuration. Right: Simulated integrated DOS images at Vs $=$ −2 V, $+$2 V. Bi sites are shown by markers. (b) Spin-polarized DOS of Bi and Si atoms in the Bi-Si hetero-dimer structure shown in (a). (c) Left: Schematic structure of P-Si hetero-dimer. Right: Simulated integrated DOS images at Vs $=$ -2 V, $+$2 V, without the additional electron discussed in Ref. 16. The P sites are marked by crosses ‘+’. Both images are for isosurface values of 0.1 electron/nm${}^3$. (d) Spin-polarized DOS of P and Si atoms in the P-Si hetero-dimer structure. All integrated DOS images are for isosurface values of 0.1 electron/Å${}^3$.