(C-H)_iO_2i

This was modelled using a 151 atom cluster in the +1 charge state, CSi₇₉H₆₉O₂ (Figure 8.6). The presence of the (C-H) bond gives the defect C_1h symmetry rather than C_2v. However, more recent ENDOR symmetry analysis [10] has shown that NL10(H), the EPR/ENDOR signal associated with H-related shallow thermal donors (see Section 8.7) was misassigned using EPR to C_2v symmetry. Using ENDOR they have shown it to be triclinic, with H lying along [110] perpendicular to the plane containing O_i, in agreement with this model.

  

Figure 8.6: The (C-H)_i-O_2i `C_2v' defect core in the +1 charge state. Numbered atoms are referred to in the text. Symmetry with H included is C_1h (see text). Black dots mark the ideal lattice sites. This is less stable than the assymetric structure in Figure 8.7. The cross-hatched H atom lies perpendicular to the defect plane.

The O forms assymetric bonds of 1.66 and 1.62 Å, the shorter bond with the more positive Si atom shared with the (CH)_i. The Si-O-Si bond angle is 158$^\circ$. C forms three strong Si bonds, the top two of 1.818 Å and the bond with the core Si of 1.905 Å. The C-H unit has a short C-H bond of 1.115 Å, and C moves out of plane by 0.159 Å. However, unlike N_iO_2i, this defect was not itself a shallow donor, the level instead lying closer to mid-gap. Changing the charge state of the defect to neutral or -1 through the addition of 1 or 2 electrons respectively, had little effect on the structure, and the level remained deep in the gap. In this case the empty dangling bond on the Si was either singly (neutral) or doubly (-1 charged) occupied. The C-H bond dilated (1.138 Å when neutral, 1.145 Å when -1), and the distance between the C and the core Si increased (1.965 Å when neutral, 2.001 Å when -1). The O atoms were also pushed back into their bond centres (Si-O-Si bond angles of 173$^\circ$), and all of these shifts are consistent with increased charge in the defect core. These results show that, by itself, the (CH)_iO_2i defect in this configuration is not a single shallow donor. We deliberately breached the defect symmetry after relaxation but in all cases it returned to the symmetric structures described above with the same energies.

  

Figure 8.7: The (C-H)_i-O_2i defect with O_2i neighbouring (CH)_i, in the neutral charge state. Numbered atoms are referred to in the text. This structure is 1.36 eV more stable than its equivalent with one O atom on either side of the (CH)_i unit. Black dots mark the ideal lattice sites.

We next examined the stability of this defect and found it to be less stable than the alternative structure with both oxygen atoms on the same side of the C (similar to its unhydrogenated cousin, C_i-O_2i, responsible for the `P-line' PL absorption). This lop-sided defect structure is 1.36 eV more stable than the symmetric one in the neutral charge state (1.33 and 1.76 eV in the +1 and -1 charge states respectively). This also has a deep donor level rather than a shallow one. The structure is shown in Figure 8.7. We again examined the three possible charge states of this defect, +1, 0, or -1, corresponding to different populations of the Si p-type dangling bond in the core of the defect. In the +1 charge state the core Si stayed close to the $\langle$001$\rangle$ axis, but in the neutral and -1 charge state it bows away from the neighbouring oxygen atom, allowing it to localise the filled p-type orbital as far from the oxygen lone-pairs as possible. In this respect its behaviour mimics that of N_iO_i, and presumably the exact analogy, (CH)_iO_i would also behave in a similar way.

The lack of shallow donor properties for (CH)_iO_2i in the C_2v configuration is initially surprising, given that it is iso-electronic with the N_iO_2i shallow thermal donor. However the (CH) unit is less electronegative than N_i and so the core Si atom is less polarised. This in turn provides less driving force to draw the oxygen atoms inwards and indirectly compress the gap level. In addition the back-bonds of (CH)_i are not as dilated as those of N_i, and so there is less strain energy to be gained from placing the oxygen atoms next to the (CH)_i; this is why the oxygen atoms prefer to sit on one side where they gain energy through quadrapole interactions through their shared lattice Si.

However the C_1h defect could still act as a STD core, and would become stable with the addition of further oxygen atoms, similar to N_i-O_4i. This would also help to compress the defect core. We therefore relaxed (C-H)_i-O_4i, and found that it possessed a single shallow donor level. The Kohn-Sham eigenvalues are shown in Figure 8.10, showing the donor level. (CH)_iO_4i is examined further, below in Section 8.8.

In conclusion, it appears that C_iHO_4i is a stable shallow donor defect, possesses C_1h symmetry and contains a C-H bond lying along [011]. There is very little spin-polarised density residing on C or H. The defect core, C_iHO_2i is more stable in a C₁ configuration similar to C_iO_2i. In n-type material the (CH)_iO_2i defect would be either neutral or -1 charged.