Hydrogen in silicon

Successive burial of hydrogen-rich surface layers leads to the formation of the a-Si H material. The large amount of hydrogen at the surface is to saturate the surface dangling bonds. The much lower hydrogen content in the bulk is due to the solubility limit of hydrogen in silicon. 

As was pointed out in previous chapters, the role of hydrogen in silicon is to passivate all the Si dangling bonds. It is the realization that there is a Si dangling bond near every acceptor that led us to search for the possibility that H might neutralize the acceptor. 

To study solution and diffusion of hydrogen in silicon, one ought properly to be able to measure the concentration of all hydrogen species and hydrogen-containing complexes separately as functions of position and... 

Fig. 11. High-temperature behavior of the solubility s of hydrogen in silicon and of its diffusion coefficient D. Squares are from the measurements of Van Wieringen and Warmholtz (1956) on H2, fitted by the full lines (80) and (81), respectively. Crosses are measurements of s using 3H2, by Ichimiya and Furuichi (1968), fitted by the dotted line(82). 

To make contact with atomic theories of the binding of interstitial hydrogen in silicon, and to extrapolate the solubility to lower temperatures, some thermodynamic analysis of these data is needed a convenient procedure is that of Johnson, etal. (1986). As we have seen in Section II. l,Eqs. (2) et seq., the equilibrium concentration of any interstitial species is determined by the concentration of possible sites for this species, the vibrational partition function for each occupied site, and the difference between the chemical potential p, of the hydrogen and the ground state energy E0 on this type of site. In equilibrium with external H2 gas, /x is accurately known from thermochemical tables for the latter. A convenient source is the... 

The final question we shall consider here has to do with the extrapolation of the solubility of hydrogen in silicon to lower temperatures. Extrapolation of a high-temperature Arrhenius line, e.g., from Fig. 11, would at best give an estimate of the equilibrium concentration of H°, or perhaps of all monatomic species, in intrinsic material the concentration of H2 complexes would not be properly allowed for, nor would the effects of Fermi-level shifts. Obviously the temperature dependence of the total dissolved hydrogen concentration in equilibrium with, say, H2 gas at one atmosphere, will depend on a number of parameters whose values are not yet adequately known the binding energy AE2 of two H° into H2 in the crystal, the locations of the hydrogen donor and acceptor levels eD, eA, respectively, etc. However, the uncertainties in such quantities are not so... 

Fig. 16. Panorama of values in the literature for diffusion coefficients of hydrogen in silicon and for other diffusion-related descriptors. Black symbols represent what can plausibly be argued to be diffusion coefficients of a single species or of a mixture of species appropriate to intrinsic conditions. Other points are effective diffusion coefficients dependent on doping and hydrogenation conditions polygons represent values inferred from passivation profiles [i.e., similar to the Dapp = L2/t of Eq. (95) and the ensuing discussion] pluses and crosses represent other quantities that have been called diffusion coefficients. The full line is a rough estimation for H+, drawn assuming the top points to refer mainly to this species otherwise the line should be higher at this end. The dashed line is drawn parallel a factor 2 lower to illustrate a plausible order of magnitude of the difference between 2H and H.

Fig. 1. Ligand-field model for the electronic structure of substitutional hydrogen in silicon in terms of the interactions between the vacancy orbitals and the atomic-hydrogen orbitals [Although the a state is shown as being not entirely passivated (still below the bottom of the conduction-band edge), it could in fact be in the conduction band, but with a host-like state pushed down slightly into the band gap.] (Reprinted with permission from the American Physical Society, DeLeo, G.G., Fowler, W.B., Watkins, G.D. (1984). Phys. Rev. B 29, 1819.)...

The close correspondence between the properties of Mu in Si as determined by /u,SR and pLCR and those for the AA9 center produced by implanting hydrogen in silicon shows that Mu in silicon and the AA9 center are isostructural and in fact almost identical. They are neutral isolated bond-centered interstitials. Numerous theoretical studies support this conclusion. The observation of such similar centers for muonium and hydrogen supports the generalization that hydrogen analogs of many of the muonium centers exist. Of course, this assumes that the effects of the larger zero-point vibration of the muon relative to the proton do not make a major contribution to structural differences. The p-SR experiments, reinforced by theory, demonstrate that another structure also exists for muonium in silicon, called normal muonium or Mu. This structure is metastable and almost certainly is isolated neutral muonium at a tetrahedral interstitial site. 

In many cases, the deposited material can retain some of the original chemical constituents, such as hydrogen in silicon from the deposition from silane, or chlorine in tungsten from the deposition from WC1 . This can be beneficial or detrimental. For example, the retention of hydrogen in silicon allows the deposition of amorphous silicon, a-Si H, which is used in solar cells, but the retention of chlorine in tungsten is detrimental to subsequent fusion welding of the tungsten. 

---