IN Brief:
- Hydrogen passivation suppresses electrically active dangling-bond defects in silicon, but energetic carriers can eventually break the Si-H bond.
- First-principles modelling shows that temporary occupation of an antibonding state can launch hydrogen along a dissociative quantum trajectory.
- The framework reproduces experimentally observed threshold, isotope, and temperature behaviour, providing a more physical basis for reliability modelling.
Researchers led by University of California, Santa Barbara, with TU Wien and Samsung Research America, have developed a first-principles framework explaining how an individual energetic electron can trigger silicon-hydrogen bond dissociation. The mechanism provides an atomistic description of a process involved in hot-carrier degradation, one of the long-term reliability problems affecting silicon devices.
Hydrogen is widely used during semiconductor processing to passivate silicon dangling bonds, particularly around silicon/oxide interfaces. Unpassivated bonds create electrically active states inside the semiconductor bandgap, so attaching hydrogen reduces the defect population and improves device behaviour. Reliability deteriorates when that hydrogen eventually detaches and the electrically active defect reappears.
The new framework concentrates on how energy is transferred from an energetic carrier to the Si-H bond. A purely thermal or classical collision picture struggles to account for several experimental observations, including the strong isotope effect, the weak temperature dependence of the process, and the characteristic electron-energy threshold associated with dissociation.
The researchers used first-principles density-functional calculations and a partitioning method to identify the bonding and antibonding electronic states associated with the Si-H system. When an energetic electron temporarily occupies the antibonding state, the potential-energy surface experienced by the hydrogen nucleus becomes strongly repulsive.
Hydrogen then begins to move away from its equilibrium position on a femtosecond timescale. The electronic excitation itself is transient, but part of the nuclear wave packet can continue along a dissociative path after the electron has left the antibonding state and the system relaxes towards its electronic ground state.
That quantum description changes the way bond failure is represented. Hydrogen is not treated simply as a classical particle that breaks free once it has crossed a fixed distance with sufficient thermal energy. Instead, its position is represented by a wave packet, and dissociation becomes a probability associated with part of that wave packet propagating far enough from the original bond.
The model reproduces several measurements that have proved difficult to reconcile using simpler descriptions. It accounts for a dissociation threshold around 7V electron injection while retaining a finite probability below that nominal threshold, and it reproduces the substantially different behaviour of hydrogen and its heavier isotope, deuterium.
Experiments have shown that replacing hydrogen with deuterium can reduce the bond-breaking rate by roughly two orders of magnitude. The electronic structure is essentially unchanged by the isotope substitution, but the nuclear mass doubles. A model that explicitly treats nuclear quantum dynamics can therefore explain the difference without inventing a different electronic degradation mechanism.
The framework also reproduces the observed temperature independence of the underlying dissociation event. That supports the conclusion that the process is not simply local heating of the bond by repeated energetic carriers. One sufficiently energetic electronic excitation can initiate the atomic motion that produces a permanent defect.
For semiconductor reliability work, the attraction of a first-principles mechanism is predictive capability. Traditional hot-carrier lifetime models are often fitted to measurements from a particular technology and then extrapolated across voltage, temperature, and operating time. That remains useful, but the confidence of the extrapolation can weaken as materials, dimensions, and electric fields change.
A model built from electronic states and atomic motion offers a route to evaluating which chemical bonds or interface structures are most vulnerable before a complete set of long-duration reliability data exists. It will not replace device-level qualification because real transistors contain distributions of defects, fields, stresses, and local environments that a single-bond calculation does not capture.
The authors applied the framework to a representative oxide-stress scenario and found that it could explain observed degradation behaviour. They also point to wider relevance in other semiconductor systems where carrier-induced bond breaking contributes to reliability loss, including some light-emitting and power-electronic materials.
The underlying Physical Review B paper was published on 11 February 2026, and UCSB publicised the research on 14 April. It is therefore being retained in this run because of the Stage 2 editorial override, not presented as newly published August research. The physics itself remains useful: a reliability mechanism accumulated over years of device operation can begin with an electronic state occupied for only a few femtoseconds.
