10B - 11B isotope shifts have been reported recently for the vibrational frequencies of hydrogen (H) and its isotope deuterium (D) in the H-B complex in silicon. The D-10 B - D11 B shift was found to be anomalously large. We show that this effect finds a natural explanation in a phenomenon called “Fermi resonance”, arising from a weak anharmonic coupling between the second harmonic of the transverse B vibration and the longitudinal D vibration. We first present a simple classical explanation of the effect in terms of a “parametric oscillator”, or a child pumping a swing. We then outline a simple quantum mechanical treatment that provides a satisfactory quantitative explanation of the results. Our calculations also predict infrared absorption at the boron second harmonic frequencies. These are observed for both 10B and 11B with intensities and polarization as predicted, providing direct confirmation of the interpretation. The Pankove Si-H-B model, therefore, remains intact.

We present for the first time an optically detected magnetic resonance (ODMR) study of a hydrogen-related defect in silicon. The defect is present in hydrogenated boron-doped silicon single crystals, after room-temperature electron-irradiation. A spin-triplet (S=1) is shown to be the electronic state responsible for the observed ODMR spectrum. An angular dependence study of the ODMR spectrum reveals a C2v defect symmetry. The defect model is discussed in terms of a di-hydrogen-vacancy complex. The role of this defect as an efficient recombination channel (presumably non-radiative) for the non-equilibrium free carriers is also demonstrated.

We investigate static and dynamic properties of hydrogen in crystalline and amorphous silicon by means of ab initio molecular dynamics simulations. In the crystalline case we focus mainly on the diffusion process of an isolated positively charged hydrogen impurity at high temperature. In the amorphous case we analyze the local order and the dynamical properties corresponding to an atomic hydrogen concentration of ~ 11%, typical of a device quality material. In both crystalline and amorphous cases, our results are in good agreement with available experimental data and give unique insight into the microscopic details of hydrogen incorporation in silicon.

Infrared reflectance spectroscopy is employed to obtain hydrogen depth profiles in boron-doped silicon, hydrogenated under various plasma conditions. From the obtained profiles, H-diffusion coefficients are calculated for different temperatures and dopant concentrations. Our results are interpreted by assuming that diffusion in the bulk is limited by trapping at the acceptor sites. A binding energy of 0.6 eV is deduced for B-H pairs. We also analyze the influence of a bias applied to the sample on the hydrogenation process. This sample bias can favor or completely hinder the diffusion of hydrogen into the silicon bulk. Also, a surface oxide layer can drastically inhibit the hydrogen in-diffusion.

A model is proposed to describe hydrogen motion in silicon near 150°C. This model leads to a consistent view of H° behaviour in low doped n and p-type Si, with a diffusivity in agreement with the high temperature data. On the other hand, a systematic variation of DH+ with the boron concentration forces us to conclude that some still unknown interactions take place and contribute to hydrogen trapping in highly doped p-Si.

The study of isolated hydrogen in semiconductors is difficult because of the tendency for hydrogen to form complexes. An alternative to direct studies of hydrogen is the study of tnuonium in semiconductors. In such experiments the muon is essentially an isotope of hydrogen with th the mass of the proton. Twenty isolated muonium centers have been observed in tetrahedrally-coordinated crystals ranging from Si to the cuprous halides. Muon level-crossing resonance has provided detailed information on neutral interstitial muonium located at a bond center in Si (the analog of the hydrogen center seen by EPR) and very near a bond center in CaP and GaAs. This paper reviews these measurements and discusses the structures and their metastabilities. Comparison is made to EPR for Si and to theoretical studies. Recent results on the metastable and stable muonium centers in CuCl are discussed.

First-principles spin-density-functional calculations are used to evaluate hyperfine and superhyperfine parameters for hydrogen and muonium at various sites in the Si lattice. The results can be directly compared with values from muon-spin-rotation experiments, leading to an unambiguous identification of “anomalous muonium” with the bond-center site. The agreement found in this case instills confidence in the general use of spin-density-functional calculations for predicting hyperfine parameters of defects.

The behavior of hydrogen in crystalline silicon (c-Si) containing regions of compressive or tensile stress is important for understanding the solute’s interaction with dislocations, grain boundaries, and crack tips. A series of first-principles total-energy calculations probing the stable site for hydrogen as a function of its charge state, the Fermi level position, and the crystalline lattice constant has been performed. We find that the stable site for hydrogen depends critically on both pressure and on the hydrogen charge state. Furthermore, hydrogen is predicted to undergo a transition from an interstitial site to the bond-center site as a function of pressure.

Hydrogenation of metal/thin oxide/p-type silicon and metal/n-type silicon diodes has been studied using high frequency capacitance profiling. In situ observations of acceptor and donor passivation were made while H ions were implanted through thin gate metallizations. Direct measurement of ion transits at a variety of electric fields establish that a unique mobility can be assigned to positive H ions, and modeling of low and high field data in both n and p-type samples is consistent with the notion that the positive charge state is occupied ~1/10 of the time. The time dependence of hydrogen penetration for both p and n-type diodes indicates that hydrogen is, in addition to being trapped at unpassivated shallow donors or acceptors, becoming immobilized at other sites in silicon. The density of these secondary trapping sites correlates well with the shallow dopant population suggesting that additional hydrogen atoms may become trapped near already-passivated dopant atoms.

Hydrogen segregation at the interface between an epitaxial Al film and a Si (111) substrate is studied using the 1H(15N, αγ)12C nuclear resonant reaction. Hydrogen depth profiles show that H atoms diffuse through the 1600 Å thick Al layer during 500 eV H implantation and are trapped at the Al/Si interface. The total amount of interface H is about 2 × 1015 /cm2 after a 1.4 × 1018 H/cm2 implantation, and the H atoms are narrowly distributed in the direction normal to the interface. During an isothermal anneal at 360 K, the amount of interface H decreases exponentially with annealing time; and during ramp annealing from 110 to 500 K, an abrupt release of the interface H is observed at temperature around 380 K. The release rates in both cases are controlled by a first order thermally activated de-trapping process with a binding energy of 0.86 eV/atom.

Annealing of hydrogenated p-type silicon with a reverse bias applied to a Schottky diode allows us to precisely determine the dissociation frequency vA of shallow acceptor-hydrogen pairs (AH with A = B, Al, Ga, and In). The temperature dependent values of vA satisfy the relation vA = voAexp (-EA/kT), with voB = 2.8 . 1014 s-1, voAl = 3.1 . 1013 s-1, VoGa = 6.9 . 1013 s-1, and voIn = 8.4 · 1013 s-1. The dissociation energies EA depend only weakly on the acceptors: EB = (1.28±0.03)eV, EAl = (1.44±0.02) eV, EGa = (1.40±0.03) eV, and EIn = (1.42±0.05) eV. The dissociation frequency of BH pairs shifts to a lower value when H is replaced by the deuterium isotope.

A silicon epitaxial structure containing spatially confined arrays of misfit dislocations has been used in order to investigate the interaction between hydrogen and individual extended defects. Hydrogen was introduced using a Kaufman plasma ion beam source. A characteristic Si-H peak at 2100 cm-1 was obtained using multiple internal reflection infrared spectrophotometry. Microdefects such as gas bubbles and {111} planar defects were found near the surface, as well as at the misfit dislocation interfaces, using transmission electron microscopy. A heavily damaged region was obtained on the top Si surface and an extended area SEM/EBIC contrast was obtained due to a surface electrical field.

The time evolution of dopant passivation in p and n-type silicon Schottky and MIS barriers has been investigated for low-energy H ions implanted directly into the silicon through a 400-Å front-electrode metallization. Knowledge of the dependence of the near-surface hydrogen concentration on time and experimental parameters is crucial to the analysis of these experiments. Hydrogenation effects are observed to vanish at ion energies below 800 eV, which suggests that the front electrode, rather than being a source of H, will behave as a sink for H diffusing in the silicon. Accordingly, we show that a steady-state H concentration proportional to the ion-deposition flux and deposition depth is established in a time interval less than a second near the electrode. Although some of the mobile H appears to be in a positively charged state, we calculate that the bias voltage applied to samples has only a small influence on the near-surface H concentration. An extensive study of the effects of sample temperature, beam flux, and ion energy supports these concepts and has revealed no evidence of ion-damage induced H trapping. The possible importance of H.H recombination is suggested by data obtained at the higher H concentrations.

Hydrogen passivation using a Kaufmann ion source at 400°C has been carried out on FZ and CZ silicon dislocated by four-point bending at high temperatures. The results differ from those reported for dislocations passivated in silicon sheet grown by the EFG technique. A model for hydrogen diffusion and trapping is presented to argue that the differences observed are not produced by hydrogen transport effects in the bulk.

A review of the occurence, nature, structure and stability of the H complexes in III-V compounds is presented, mainly based on electrical and spectroscopical data. The importance of the problem of formation and dissociation of the complexes is pointed out and the use of H complexes in actual III-V devices is described.

It is confirmed that Sn donors in GaAs are passivated by exposure to a hydrogen plasma. The Sn-H complexes give rise to vibrational absorption bands at 1327.8 cm-1 and 967.7 cm-1 that are assigned to H-stretching and H-wagging modes respectively. A study of the thermal stability of the Sn-H complexes shows that they dissociate for annealing temperatures above ~150°C. The properties of the Sn-H complexes are compared to those of other donor-H complexes. Our results suggest a configuration for the complex with H at the antibonding site adjacent to the Sn.

The existing microscopic models of acceptor passivation in p-type hydrogenated GaAs are reviewed in light of new experimental results concerning the relative thermodynamic stability of the passivating complexes. In particular, the present model for neutralization of Group II acceptors, Be, Mg and Zn, on Ga sites is shown to be inadequate to account for the observed trends, which imply existence of a strong interaction between the hydrogen and acceptor. It is proposed that a direct acceptor-hydrogen bond is formed due to attractive Coulomb interaction between the ionized species. The relative stability of the pair complex can be then explained based on electronegativity of the acceptor species. Passivation at intermediate pair separations up to about twice the Bohr radius of the nearest acceptor, is also discussed.

The results of a low temperature (5K) photoluminescence study of hydrogenation of GaAs on InP grown by metal organic chemical vapor deposition are presented. An emission band at ~ 1.4 eV originating from the GaAs/InP interracial region shows a 30 fold increase in intensity relative to the GaAs band edge emission after exposure to hydrogen plasma for 30 min at 250°C. This improvement in intensity is attributed to hydrogen passivation of defects at the heterointerface caused by the large (≈4%) lattice mismatch between GaAs and InP. The passivation effect recovers on annealing the hydrogenated sample at 350°C. Excitation dependence of the ~1.4 eV band suggests that the interfacial region consists of a compositionally graded layer. Further, this band shifts to higher energy on annealing the sample in the temperature range 150-450°C with the hydrogenated sample exhibiting a larger shift than the untreated sample. It is suggested that the annealing induced peak shift arises due to intermixing of the compositionally graded interface and that the degree of intermixing is greater in the hydrogenated sample compared to the untreated sample.

Reactive ion etching of highly Zn-doped InGaAs with hydrogen-containing gases such as CHF3/H2, CH4/H2 and CH4/He leads to a drastical reduction of the concentration of electrically active acceptors in a near-surface layer. Ti/Pt/Au contacts applied on as-treated p++-InGaAs layers show non-ohmic I/V-characteristics due to this passivation effect. Annealing at moderate temperatures (330°C, 20 minutes) was found to restore full electrical activity.

Various dielectric and metallic films were examined as H-permeable surface protection layers on InP during H2 or D2 plasma exposure for passivation of acceptors in the InP. Plasma deposited SiNx, SiO2, and a-Si(H) films ranging in thickness from 85 to 225 Å were used to protect p-InP during D2 plasma exposure at 250°C. Optimum protective layer thicknesses were determined by a trade-off between the effectiveness of the layer to prevent P loss from the wafer surface and the ability to diffuse atomic H or D at a rate greater than or equal to that in the underlying InP. SIMS and capacitance-voltage depth profiling were used to determine the extent of D in-diffusion and acceptor passivation respectively. Sputter deposited W and e-beam evaporated Ti films ~100 Å thick were also evaluated. The W coated sample yielded similar results to those with dielectric films in that acceptors in p-InP were passivated to a similar depth for the same plasma exposure. The 100 Å Ti film, however, did not allow the D to diffuse into the InP substrate. It is surmised that the Ti film trapped the D, thus preventing diffusion into the substrate.